mirrors/Odin
1
0
Fork 0
mirror of https://github.com/odin-lang/Odin.git synced 2026-08-27 15:31:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into bill/subtype-type-info

Browse Source
This commit is contained in:
gingerBill
2025-03-28 10:10:59 +00:00
committed by GitHub
parent 5b0783b5fa e7ae7b8fd4
commit f8fb089ee3
125 changed files with 8110 additions and 1209 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
.github/workflows/nightly.yml vendored
View File

@@ -53,8 +53,8 @@ jobs:
- name: (Linux) Download LLVM
run: |
apk add --no-cache \
musl-dev llvm20-dev clang20 git mold lz4 \
libxml2-static llvm20-static zlib-static zstd-static \
musl-dev llvm18-dev clang18 git mold lz4 \
libxml2-static llvm18-static zlib-static zstd-static \
make
shell: alpine.sh --root {0}
- name: build odin

13
base/runtime/default_temp_allocator_arena.odin
View File

@@ -104,13 +104,15 @@ arena_alloc :: proc(arena: ^Arena, size, alignment: uint, loc := #caller_locatio
if size == 0 {
return
}
needed := align_forward_uint(size, alignment)
if arena.curr_block == nil || (safe_add(arena.curr_block.used, needed) or_else 0) > arena.curr_block.capacity {
prev_used := 0 if arena.curr_block == nil else arena.curr_block.used
data, err = alloc_from_memory_block(arena.curr_block, size, alignment)
if err == .Out_Of_Memory {
if arena.minimum_block_size == 0 {
arena.minimum_block_size = DEFAULT_ARENA_GROWING_MINIMUM_BLOCK_SIZE
}
needed := align_forward_uint(size, alignment)
block_size := max(needed, arena.minimum_block_size)
if arena.backing_allocator.procedure == nil {
@@ -121,10 +123,9 @@ arena_alloc :: proc(arena: ^Arena, size, alignment: uint, loc := #caller_locatio
new_block.prev = arena.curr_block
arena.curr_block = new_block
arena.total_capacity += new_block.capacity
prev_used = 0
data, err = alloc_from_memory_block(arena.curr_block, size, alignment)
}
prev_used := arena.curr_block.used
data, err = alloc_from_memory_block(arena.curr_block, size, alignment)
arena.total_used += arena.curr_block.used - prev_used
return
}

2
base/runtime/os_specific_bsd.odin
View File

@@ -9,7 +9,7 @@ foreign libc {
@(link_name="write")
_unix_write :: proc(fd: i32, buf: rawptr, size: int) -> int ---
when ODIN_OS == .NetBSD {
when ODIN_OS == .NetBSD || ODIN_OS == .OpenBSD {
@(link_name="__errno") __error :: proc() -> ^i32 ---
} else {
__error :: proc() -> ^i32 ---

8
build.bat
View File

@@ -4,12 +4,12 @@ setlocal EnableDelayedExpansion
where /Q cl.exe || (
set __VSCMD_ARG_NO_LOGO=1
for /f "tokens=*" %%i in ('"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe" -latest -requires Microsoft.VisualStudio.Workload.NativeDesktop -property installationPath') do set VS=%%i
for /f "tokens=*" %%i in ('"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe" -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -property installationPath') do set VS=%%i
if "!VS!" equ "" (
echo ERROR: Visual Studio installation not found
echo ERROR: MSVC installation not found
exit /b 1
)
call "!VS!\VC\Auxiliary\Build\vcvarsall.bat" amd64 || exit /b 1
call "!VS!\Common7\Tools\vsdevcmd.bat" -arch=x64 -host_arch=x64 || exit /b 1
)
if "%VSCMD_ARG_TGT_ARCH%" neq "x64" (
@@ -152,4 +152,4 @@ if %release_mode% EQU 0 echo: & echo Debug compiler built. Note: run "build.bat
del *.obj > NUL 2> NUL
:end_of_build
:end_of_build

13
build_odin.sh
View File

@@ -25,7 +25,8 @@ error() {
# Brew advises people not to add llvm to their $PATH, so try and use brew to find it.
if [ -z "$LLVM_CONFIG" ] && [ -n "$(command -v brew)" ]; then
if [ -n "$(command -v $(brew --prefix llvm@19)/bin/llvm-config)" ]; then LLVM_CONFIG="$(brew --prefix llvm@19)/bin/llvm-config"
if [ -n "$(command -v $(brew --prefix llvm)/bin/llvm-config)" ]; then LLVM_CONFIG="$(brew --prefix llvm)/bin/llvm-config"
elif [ -n "$(command -v $(brew --prefix llvm@19)/bin/llvm-config)" ]; then LLVM_CONFIG="$(brew --prefix llvm@19)/bin/llvm-config"
elif [ -n "$(command -v $(brew --prefix llvm@18)/bin/llvm-config)" ]; then LLVM_CONFIG="$(brew --prefix llvm@18)/bin/llvm-config"
elif [ -n "$(command -v $(brew --prefix llvm@17)/bin/llvm-config)" ]; then LLVM_CONFIG="$(brew --prefix llvm@17)/bin/llvm-config"
elif [ -n "$(command -v $(brew --prefix llvm@14)/bin/llvm-config)" ]; then LLVM_CONFIG="$(brew --prefix llvm@14)/bin/llvm-config"
@@ -34,7 +35,8 @@ fi
if [ -z "$LLVM_CONFIG" ]; then
# darwin, linux, openbsd
if [ -n "$(command -v llvm-config-19)" ]; then LLVM_CONFIG="llvm-config-19"
if [ -n "$(command -v llvm-config-20)" ]; then LLVM_CONFIG="llvm-config-20"
elif [ -n "$(command -v llvm-config-19)" ]; then LLVM_CONFIG="llvm-config-19"
elif [ -n "$(command -v llvm-config-18)" ]; then LLVM_CONFIG="llvm-config-18"
elif [ -n "$(command -v llvm-config-17)" ]; then LLVM_CONFIG="llvm-config-17"
elif [ -n "$(command -v llvm-config-14)" ]; then LLVM_CONFIG="llvm-config-14"
@@ -42,6 +44,7 @@ if [ -z "$LLVM_CONFIG" ]; then
elif [ -n "$(command -v llvm-config-12)" ]; then LLVM_CONFIG="llvm-config-12"
elif [ -n "$(command -v llvm-config-11)" ]; then LLVM_CONFIG="llvm-config-11"
# freebsd
elif [ -n "$(command -v llvm-config20)" ]; then LLVM_CONFIG="llvm-config20"
elif [ -n "$(command -v llvm-config19)" ]; then LLVM_CONFIG="llvm-config19"
elif [ -n "$(command -v llvm-config18)" ]; then LLVM_CONFIG="llvm-config18"
elif [ -n "$(command -v llvm-config17)" ]; then LLVM_CONFIG="llvm-config17"
@@ -69,15 +72,15 @@ LLVM_VERSION_MAJOR="$(echo $LLVM_VERSION | awk -F. '{print $1}')"
LLVM_VERSION_MINOR="$(echo $LLVM_VERSION | awk -F. '{print $2}')"
LLVM_VERSION_PATCH="$(echo $LLVM_VERSION | awk -F. '{print $3}')"
if [ $LLVM_VERSION_MAJOR -lt 11 ] || ([ $LLVM_VERSION_MAJOR -gt 14 ] && [ $LLVM_VERSION_MAJOR -lt 17 ]) || [ $LLVM_VERSION_MAJOR -gt 19 ]; then
error "Invalid LLVM version $LLVM_VERSION: must be 11, 12, 13, 14, 17, 18 or 19"
if [ $LLVM_VERSION_MAJOR -lt 11 ] || ([ $LLVM_VERSION_MAJOR -gt 14 ] && [ $LLVM_VERSION_MAJOR -lt 17 ]) || [ $LLVM_VERSION_MAJOR -gt 20 ]; then
error "Invalid LLVM version $LLVM_VERSION: must be 11, 12, 13, 14, 17, 18, 19 or 20"
fi
case "$OS_NAME" in
Darwin)
if [ "$OS_ARCH" = "arm64" ]; then
if [ $LLVM_VERSION_MAJOR -lt 13 ]; then
error "Invalid LLVM version $LLVM_VERSION: Darwin Arm64 requires LLVM 13, 14, 17, 18 or 19"
error "Invalid LLVM version $LLVM_VERSION: Darwin Arm64 requires LLVM 13, 14, 17, 18, 19 or 20"
fi
fi

1
core/crypto/_aes/aes.odin
View File

@@ -25,4 +25,5 @@ GHASH_BLOCK_SIZE :: 16
GHASH_TAG_SIZE :: 16
// RCON is the AES keyschedule round constants.
@(rodata)
RCON := [10]byte{0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80, 0x1B, 0x36}

9
core/crypto/_aes/ct64/ct64.odin
View File

@@ -22,8 +22,6 @@
package aes_ct64
import "base:intrinsics"
// Bitsliced AES for 64-bit general purpose (integer) registers. Each
// invocation will process up to 4 blocks at a time. This implementation
// is derived from the BearSSL ct64 code, and distributed under a 1-clause
@@ -212,11 +210,8 @@ orthogonalize :: proc "contextless" (q: ^[8]u64) {
}
@(require_results)
interleave_in :: proc "contextless" (w: []u32) -> (q0, q1: u64) #no_bounds_check {
if len(w) < 4 {
intrinsics.trap()
}
x0, x1, x2, x3 := u64(w[0]), u64(w[1]), u64(w[2]), u64(w[3])
interleave_in :: proc "contextless" (w0, w1, w2, w3: u32) -> (q0, q1: u64) #no_bounds_check {
x0, x1, x2, x3 := u64(w0), u64(w1), u64(w2), u64(w3)
x0 |= (x0 << 16)
x1 |= (x1 << 16)
x2 |= (x2 << 16)

6
core/crypto/_aes/ct64/ct64_enc.odin
View File

@@ -22,12 +22,8 @@
package aes_ct64
import "base:intrinsics"
add_round_key :: proc "contextless" (q: ^[8]u64, sk: []u64) #no_bounds_check {
if len(sk) < 8 {
intrinsics.trap()
}
ensure_contextless(len(sk) >= 8, "aes/ct64: invalid round key size")
q[0] ~= sk[0]
q[1] ~= sk[1]

61
core/crypto/_aes/ct64/ct64_keysched.odin
View File

@@ -22,7 +22,6 @@
package aes_ct64
import "base:intrinsics"
import "core:crypto/_aes"
import "core:encoding/endian"
import "core:mem"
@@ -42,7 +41,7 @@ sub_word :: proc "contextless" (x: u32) -> u32 {
}
@(private, require_results)
keysched :: proc(comp_skey: []u64, key: []byte) -> int {
keysched :: proc "contextless" (comp_skey: []u64, key: []byte) -> int {
num_rounds, key_len := 0, len(key)
switch key_len {
case _aes.KEY_SIZE_128:
@@ -52,7 +51,7 @@ keysched :: proc(comp_skey: []u64, key: []byte) -> int {
case _aes.KEY_SIZE_256:
num_rounds = _aes.ROUNDS_256
case:
panic("crypto/aes: invalid AES key size")
panic_contextless("crypto/aes: invalid AES key size")
}
skey: [60]u32 = ---
@@ -78,7 +77,7 @@ keysched :: proc(comp_skey: []u64, key: []byte) -> int {
q: [8]u64 = ---
for i, j := 0, 0; i < nkf; i, j = i + 4, j + 2 {
q[0], q[4] = interleave_in(skey[i:])
q[0], q[4] = interleave_in(skey[i], skey[i+1], skey[i+2], skey[i+3])
q[1] = q[0]
q[2] = q[0]
q[3] = q[0]
@@ -123,57 +122,3 @@ skey_expand :: proc "contextless" (skey, comp_skey: []u64, num_rounds: int) {
skey[v + 3] = (x3 << 4) - x3
}
}
orthogonalize_roundkey :: proc "contextless" (qq: []u64, key: []byte) {
if len(qq) < 8 || len(key) != 16 {
intrinsics.trap()
}
skey: [4]u32 = ---
skey[0] = endian.unchecked_get_u32le(key[0:])
skey[1] = endian.unchecked_get_u32le(key[4:])
skey[2] = endian.unchecked_get_u32le(key[8:])
skey[3] = endian.unchecked_get_u32le(key[12:])
q: [8]u64 = ---
q[0], q[4] = interleave_in(skey[:])
q[1] = q[0]
q[2] = q[0]
q[3] = q[0]
q[5] = q[4]
q[6] = q[4]
q[7] = q[4]
orthogonalize(&q)
comp_skey: [2]u64 = ---
comp_skey[0] =
(q[0] & 0x1111111111111111) |
(q[1] & 0x2222222222222222) |
(q[2] & 0x4444444444444444) |
(q[3] & 0x8888888888888888)
comp_skey[1] =
(q[4] & 0x1111111111111111) |
(q[5] & 0x2222222222222222) |
(q[6] & 0x4444444444444444) |
(q[7] & 0x8888888888888888)
for x, u in comp_skey {
x0 := x
x1, x2, x3 := x0, x0, x0
x0 &= 0x1111111111111111
x1 &= 0x2222222222222222
x2 &= 0x4444444444444444
x3 &= 0x8888888888888888
x1 >>= 1
x2 >>= 2
x3 >>= 3
qq[u * 4 + 0] = (x0 << 4) - x0
qq[u * 4 + 1] = (x1 << 4) - x1
qq[u * 4 + 2] = (x2 << 4) - x2
qq[u * 4 + 3] = (x3 << 4) - x3
}
mem.zero_explicit(&skey, size_of(skey))
mem.zero_explicit(&q, size_of(q))
mem.zero_explicit(&comp_skey, size_of(comp_skey))
}

6
core/crypto/_aes/ct64/ghash.odin
View File

@@ -22,7 +22,6 @@
package aes_ct64
import "base:intrinsics"
import "core:crypto/_aes"
import "core:encoding/endian"
@@ -64,9 +63,8 @@ rev64 :: proc "contextless" (x: u64) -> u64 {
// Note: `dst` is both an input and an output, to support easy implementation
// of GCM.
ghash :: proc "contextless" (dst, key, data: []byte) {
if len(dst) != _aes.GHASH_BLOCK_SIZE || len(key) != _aes.GHASH_BLOCK_SIZE {
intrinsics.trap()
}
ensure_contextless(len(dst) == _aes.GHASH_BLOCK_SIZE)
ensure_contextless(len(key) == _aes.GHASH_BLOCK_SIZE)
buf := data
l := len(buf)

84
core/crypto/_aes/ct64/helpers.odin
View File

@@ -1,60 +1,61 @@
package aes_ct64
import "base:intrinsics"
import "core:crypto/_aes"
import "core:encoding/endian"
load_blockx1 :: proc "contextless" (q: ^[8]u64, src: []byte) {
if len(src) != _aes.BLOCK_SIZE {
intrinsics.trap()
}
w: [4]u32 = ---
w[0] = endian.unchecked_get_u32le(src[0:])
w[1] = endian.unchecked_get_u32le(src[4:])
w[2] = endian.unchecked_get_u32le(src[8:])
w[3] = endian.unchecked_get_u32le(src[12:])
q[0], q[4] = interleave_in(w[:])
orthogonalize(q)
@(require_results)
load_interleaved :: proc "contextless" (src: []byte) -> (u64, u64) #no_bounds_check {
w0 := endian.unchecked_get_u32le(src[0:])
w1 := endian.unchecked_get_u32le(src[4:])
w2 := endian.unchecked_get_u32le(src[8:])
w3 := endian.unchecked_get_u32le(src[12:])
return interleave_in(w0, w1, w2, w3)
}
store_blockx1 :: proc "contextless" (dst: []byte, q: ^[8]u64) {
if len(dst) != _aes.BLOCK_SIZE {
intrinsics.trap()
}
orthogonalize(q)
w0, w1, w2, w3 := interleave_out(q[0], q[4])
store_interleaved :: proc "contextless" (dst: []byte, a0, a1: u64) #no_bounds_check {
w0, w1, w2, w3 := interleave_out(a0, a1)
endian.unchecked_put_u32le(dst[0:], w0)
endian.unchecked_put_u32le(dst[4:], w1)
endian.unchecked_put_u32le(dst[8:], w2)
endian.unchecked_put_u32le(dst[12:], w3)
}
@(require_results)
xor_interleaved :: #force_inline proc "contextless" (a0, a1, b0, b1: u64) -> (u64, u64) {
return a0 ~ b0, a1 ~ b1
}
@(require_results)
and_interleaved :: #force_inline proc "contextless" (a0, a1, b0, b1: u64) -> (u64, u64) {
return a0 & b0, a1 & b1
}
load_blockx1 :: proc "contextless" (q: ^[8]u64, src: []byte) {
ensure_contextless(len(src) == _aes.BLOCK_SIZE, "aes/ct64: invalid block size")
q[0], q[4] = #force_inline load_interleaved(src)
orthogonalize(q)
}
store_blockx1 :: proc "contextless" (dst: []byte, q: ^[8]u64) {
ensure_contextless(len(dst) == _aes.BLOCK_SIZE, "aes/ct64: invalid block size")
orthogonalize(q)
#force_inline store_interleaved(dst, q[0], q[4])
}
load_blocks :: proc "contextless" (q: ^[8]u64, src: [][]byte) {
if n := len(src); n > STRIDE || n == 0 {
intrinsics.trap()
}
ensure_contextless(len(src) == 0 || len(src) <= STRIDE, "aes/ct64: invalid block(s) size")
w: [4]u32 = ---
for s, i in src {
if len(s) != _aes.BLOCK_SIZE {
intrinsics.trap()
}
w[0] = endian.unchecked_get_u32le(s[0:])
w[1] = endian.unchecked_get_u32le(s[4:])
w[2] = endian.unchecked_get_u32le(s[8:])
w[3] = endian.unchecked_get_u32le(s[12:])
q[i], q[i + 4] = interleave_in(w[:])
ensure_contextless(len(s) == _aes.BLOCK_SIZE, "aes/ct64: invalid block size")
q[i], q[i + 4] = #force_inline load_interleaved(s)
}
orthogonalize(q)
}
store_blocks :: proc "contextless" (dst: [][]byte, q: ^[8]u64) {
if n := len(dst); n > STRIDE || n == 0 {
intrinsics.trap()
}
ensure_contextless(len(dst) == 0 || len(dst) <= STRIDE, "aes/ct64: invalid block(s) size")
orthogonalize(q)
for d, i in dst {
@@ -62,14 +63,7 @@ store_blocks :: proc "contextless" (dst: [][]byte, q: ^[8]u64) {
if d == nil {
break
}
if len(d) != _aes.BLOCK_SIZE {
intrinsics.trap()
}
w0, w1, w2, w3 := interleave_out(q[i], q[i + 4])
endian.unchecked_put_u32le(d[0:], w0)
endian.unchecked_put_u32le(d[4:], w1)
endian.unchecked_put_u32le(d[8:], w2)
endian.unchecked_put_u32le(d[12:], w3)
ensure_contextless(len(d) == _aes.BLOCK_SIZE, "aes/ct64: invalid block size")
#force_inline store_interleaved(d, q[i], q[i + 4])
}
}

4
core/crypto/_aes/hw_intel/ghash.odin
View File

@@ -52,7 +52,7 @@ GHASH_STRIDE_BYTES_HW :: GHASH_STRIDE_HW * _aes.GHASH_BLOCK_SIZE
// that it is right-shifted by 1 bit. The left-shift is relatively
// inexpensive, and it can be mutualised.
//
// Since SSE2 opcodes do not have facilities for shitfting full 128-bit
// Since SSE2 opcodes do not have facilities for shifting full 128-bit
// values with bit precision, we have to break down values into 64-bit
// chunks. We number chunks from 0 to 3 in left to right order.
@@ -155,7 +155,7 @@ square_f128 :: #force_inline proc "contextless" (kw: x86.__m128i) -> (x86.__m128
@(enable_target_feature = "sse2,ssse3,pclmul")
ghash :: proc "contextless" (dst, key, data: []byte) #no_bounds_check {
if len(dst) != _aes.GHASH_BLOCK_SIZE || len(key) != _aes.GHASH_BLOCK_SIZE {
intrinsics.trap()
panic_contextless("aes/ghash: invalid dst or key size")
}
// Note: BearSSL opts to copy the remainder into a zero-filled

32
core/crypto/_blake2/blake2.odin
View File

@@ -18,6 +18,8 @@ BLAKE2S_SIZE :: 32
BLAKE2B_BLOCK_SIZE :: 128
BLAKE2B_SIZE :: 64
MAX_SIZE :: 255
Blake2s_Context :: struct {
h: [8]u32,
t: [2]u32,
@@ -68,13 +70,13 @@ Blake2_Tree :: struct {
is_last_node: bool,
}
@(private)
@(private, rodata)
BLAKE2S_IV := [8]u32 {
0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a,
0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19,
}
@(private)
@(private, rodata)
BLAKE2B_IV := [8]u64 {
0x6a09e667f3bcc908, 0xbb67ae8584caa73b,
0x3c6ef372fe94f82b, 0xa54ff53a5f1d36f1,
@@ -82,16 +84,13 @@ BLAKE2B_IV := [8]u64 {
0x1f83d9abfb41bd6b, 0x5be0cd19137e2179,
}
init :: proc(ctx: ^$T, cfg: ^Blake2_Config) {
init :: proc "contextless" (ctx: ^$T, cfg: ^Blake2_Config) {
when T == Blake2s_Context {
max_size :: BLAKE2S_SIZE
} else when T == Blake2b_Context {
max_size :: BLAKE2B_SIZE
}
if cfg.size > max_size {
panic("blake2: requested output size exceeeds algorithm max")
}
ensure_contextless(cfg.size <= max_size, "blake2: requested output size exceeeds algorithm max")
// To save having to allocate a scratch buffer, use the internal
// data buffer (`ctx.x`), as it is exactly the correct size.
@@ -167,8 +166,8 @@ init :: proc(ctx: ^$T, cfg: ^Blake2_Config) {
ctx.is_initialized = true
}
update :: proc(ctx: ^$T, p: []byte) {
assert(ctx.is_initialized)
update :: proc "contextless" (ctx: ^$T, p: []byte) {
ensure_contextless(ctx.is_initialized)
p := p
when T == Blake2s_Context {
@@ -195,8 +194,8 @@ update :: proc(ctx: ^$T, p: []byte) {
ctx.nx += copy(ctx.x[ctx.nx:], p)
}
final :: proc(ctx: ^$T, hash: []byte, finalize_clone: bool = false) {
assert(ctx.is_initialized)
final :: proc "contextless" (ctx: ^$T, hash: []byte, finalize_clone: bool = false) {
ensure_contextless(ctx.is_initialized)
ctx := ctx
if finalize_clone {
@@ -206,24 +205,19 @@ final :: proc(ctx: ^$T, hash: []byte, finalize_clone: bool = false) {
}
defer(reset(ctx))
ensure_contextless(len(hash) >= int(ctx.size), "crypto/blake2: invalid destination digest size")
when T == Blake2s_Context {
if len(hash) < int(ctx.size) {
panic("crypto/blake2s: invalid destination digest size")
}
blake2s_final(ctx, hash)
} else when T == Blake2b_Context {
if len(hash) < int(ctx.size) {
panic("crypto/blake2b: invalid destination digest size")
}
blake2b_final(ctx, hash)
}
}
clone :: proc(ctx, other: ^$T) {
clone :: proc "contextless" (ctx, other: ^$T) {
ctx^ = other^
}
reset :: proc(ctx: ^$T) {
reset :: proc "contextless" (ctx: ^$T) {
if !ctx.is_initialized {
return
}

26
core/crypto/_chacha20/chacha20.odin
View File

@@ -1,6 +1,5 @@
package _chacha20
import "base:intrinsics"
import "core:encoding/endian"
import "core:math/bits"
import "core:mem"
@@ -46,9 +45,8 @@ Context :: struct {
// derivation is expected to be handled by the caller, so that the
// HChaCha call can be suitably accelerated.
init :: proc "contextless" (ctx: ^Context, key, iv: []byte, is_xchacha: bool) {
if len(key) != KEY_SIZE || len(iv) != IV_SIZE {
intrinsics.trap()
}
ensure_contextless(len(key) == KEY_SIZE, "chacha20: invalid key size")
ensure_contextless(len(iv) == IV_SIZE, "chacha20: invalid key size")
k, n := key, iv
@@ -76,12 +74,10 @@ init :: proc "contextless" (ctx: ^Context, key, iv: []byte, is_xchacha: bool) {
// seek seeks the (X)ChaCha20 stream counter to the specified block.
seek :: proc(ctx: ^Context, block_nr: u64) {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
if ctx._is_ietf_flavor {
if block_nr > MAX_CTR_IETF {
panic("crypto/chacha20: attempted to seek past maximum counter")
}
ensure(block_nr <= MAX_CTR_IETF, "crypto/chacha20: attempted to seek past maximum counter")
} else {
ctx._s[13] = u32(block_nr >> 32)
}
@@ -102,7 +98,7 @@ check_counter_limit :: proc(ctx: ^Context, nr_blocks: int) {
// Enforce the maximum consumed keystream per IV.
//
// While all modern "standard" definitions of ChaCha20 use
// the IETF 32-bit counter, for XChaCha20 most common
// the IETF 32-bit counter, for XChaCha20 historical
// implementations allow for a 64-bit counter.
//
// Honestly, the answer here is "use a MRAE primitive", but
@@ -110,14 +106,14 @@ check_counter_limit :: proc(ctx: ^Context, nr_blocks: int) {
ERR_CTR_EXHAUSTED :: "crypto/chacha20: maximum (X)ChaCha20 keystream per IV reached"
ctr_ok: bool
if ctx._is_ietf_flavor {
if u64(ctx._s[12]) + u64(nr_blocks) > MAX_CTR_IETF {
panic(ERR_CTR_EXHAUSTED)
}
ctr_ok = u64(ctx._s[12]) + u64(nr_blocks) <= MAX_CTR_IETF
} else {
ctr := (u64(ctx._s[13]) << 32) | u64(ctx._s[12])
if _, carry := bits.add_u64(ctr, u64(nr_blocks), 0); carry != 0 {
panic(ERR_CTR_EXHAUSTED)
}
_, carry := bits.add_u64(ctr, u64(nr_blocks), 0)
ctr_ok = carry == 0
}
ensure(ctr_ok, "crypto/chacha20: maximum (X)ChaCha20 keystream per IV reached")
}

38
core/crypto/_chacha20/simd128/chacha20_simd128.odin
View File

@@ -29,11 +29,24 @@ when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 {
// explicitly using simd.u8x16 shuffles.
@(private = "file")
TARGET_SIMD_FEATURES :: "sse2,ssse3"
} else when ODIN_ARCH == .riscv64 {
@(private = "file")
TARGET_SIMD_FEATURES :: "v"
} else {
@(private = "file")
TARGET_SIMD_FEATURES :: ""
}
// Some targets lack runtime feature detection, and will flat out refuse
// to load binaries that have unknown instructions. This is distinct from
// `simd.IS_EMULATED` as actually good designs support runtime feature
// detection and that constant establishes a baseline.
//
// See:
// - https://github.com/WebAssembly/design/issues/1161
@(private = "file")
TARGET_IS_DESIGNED_BY_IDIOTS :: (ODIN_ARCH == .wasm64p32 || ODIN_ARCH == .wasm32) && !intrinsics.has_target_feature("simd128")
@(private = "file")
_ROT_7L: simd.u32x4 : {7, 7, 7, 7}
@(private = "file")
@@ -205,11 +218,13 @@ _store_simd128 :: #force_inline proc "contextless" (
// is_performant returns true iff the target and current host both support
// "enough" 128-bit SIMD to make this implementation performant.
is_performant :: proc "contextless" () -> bool {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 || ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 || ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 || ODIN_ARCH == .riscv64 {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 {
req_features :: info.CPU_Features{.asimd}
} else when ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 {
req_features :: info.CPU_Features{.sse2, .ssse3}
} else when ODIN_ARCH == .riscv64 {
req_features :: info.CPU_Features{.V}
}
features, ok := info.cpu_features.?
@@ -245,8 +260,17 @@ stream_blocks :: proc(ctx: ^_chacha20.Context, dst, src: []byte, nr_blocks: int)
// 8 blocks at a time.
//
// Note: This is only worth it on Aarch64.
when ODIN_ARCH == .arm64 {
// Note:
// This uses a ton of registers so it is only worth it on targets
// that have something like 32 128-bit registers. This is currently
// all ARMv8 targets, and RISC-V Zvl128b (`V` application profile)
// targets.
//
// While our current definition of `.arm32` is 32-bit ARMv8, this
// may change in the future (ARMv7 is still relevant), and things
// like Cortex-A8/A9 does "pretend" 128-bit SIMD 64-bits at a time
// thus needs bemchmarking.
when ODIN_ARCH == .arm64 || ODIN_ARCH == .riscv64 {
for ; n >= 8; n = n - 8 {
v0, v1, v2, v3 := s0, s1, s2, s3
@@ -354,9 +378,11 @@ stream_blocks :: proc(ctx: ^_chacha20.Context, dst, src: []byte, nr_blocks: int)
// 4 blocks at a time.
//
// Note: The i386 target lacks the required number of registers
// for this to be performant, so it is skipped.
when ODIN_ARCH != .i386 {
// Note: This is skipped on several targets for various reasons.
// - i386 lacks the required number of registers
// - Generating code when runtime "hardware" SIMD support is impossible
// to detect is pointless, since this will be emulated using GP regs.
when ODIN_ARCH != .i386 && !TARGET_IS_DESIGNED_BY_IDIOTS {
for ; n >= 4; n = n - 4 {
v0, v1, v2, v3 := s0, s1, s2, s3

2
core/crypto/_chacha20/simd256/chacha20_simd256_stub.odin
View File

@@ -13,5 +13,5 @@ stream_blocks :: proc(ctx: ^_chacha20.Context, dst, src: []byte, nr_blocks: int)
}
hchacha20 :: proc "contextless" (dst, key, iv: []byte) {
intrinsics.trap()
panic_contextless("crypto/chacha20: simd256 implementation unsupported")
}

17
core/crypto/_edwards25519/edwards25519.odin
View File

@@ -11,7 +11,6 @@ See:
- https://www.hyperelliptic.org/EFD/g1p/auto-twisted-extended-1.html
*/
import "base:intrinsics"
import "core:crypto"
import field "core:crypto/_fiat/field_curve25519"
import "core:mem"
@@ -32,6 +31,7 @@ import "core:mem"
// - The group element decoding routine takes the opinionated stance of
// rejecting non-canonical encodings.
@(rodata)
FE_D := field.Tight_Field_Element {
929955233495203,
466365720129213,
@@ -39,7 +39,7 @@ FE_D := field.Tight_Field_Element {
2033849074728123,
1442794654840575,
}
@(private)
@(private, rodata)
FE_A := field.Tight_Field_Element {
2251799813685228,
2251799813685247,
@@ -47,7 +47,7 @@ FE_A := field.Tight_Field_Element {
2251799813685247,
2251799813685247,
}
@(private)
@(private, rodata)
FE_D2 := field.Tight_Field_Element {
1859910466990425,
932731440258426,
@@ -55,7 +55,7 @@ FE_D2 := field.Tight_Field_Element {
1815898335770999,
633789495995903,
}
@(private)
@(private, rodata)
GE_BASEPOINT := Group_Element {
field.Tight_Field_Element {
1738742601995546,
@@ -80,6 +80,7 @@ GE_BASEPOINT := Group_Element {
1821297809914039,
},
}
@(rodata)
GE_IDENTITY := Group_Element {
field.Tight_Field_Element{0, 0, 0, 0, 0},
field.Tight_Field_Element{1, 0, 0, 0, 0},
@@ -107,9 +108,7 @@ ge_set :: proc "contextless" (ge, a: ^Group_Element) {
@(require_results)
ge_set_bytes :: proc "contextless" (ge: ^Group_Element, b: []byte) -> bool {
if len(b) != 32 {
intrinsics.trap()
}
ensure_contextless(len(b) == 32, "edwards25519: invalid group element size")
b_ := (^[32]byte)(raw_data(b))
// Do the work in a scratch element, so that ge is unchanged on
@@ -166,9 +165,7 @@ ge_set_bytes :: proc "contextless" (ge: ^Group_Element, b: []byte) -> bool {
}
ge_bytes :: proc "contextless" (ge: ^Group_Element, dst: []byte) {
if len(dst) != 32 {
intrinsics.trap()
}
ensure_contextless(len(dst) == 32, "edwards25519: invalid group element size")
dst_ := (^[32]byte)(raw_data(dst))
// Convert the element to affine (x, y) representation.

11
core/crypto/_edwards25519/edwards25519_scalar.odin
View File

@@ -1,6 +1,5 @@
package _edwards25519
import "base:intrinsics"
import field "core:crypto/_fiat/field_scalar25519"
import "core:mem"
@@ -8,7 +7,7 @@ Scalar :: field.Montgomery_Domain_Field_Element
// WARNING: This is non-canonical and only to be used when checking if
// a group element is on the prime-order subgroup.
@(private)
@(private, rodata)
SC_ELL := field.Non_Montgomery_Domain_Field_Element {
field.ELL[0],
field.ELL[1],
@@ -25,17 +24,13 @@ sc_set_u64 :: proc "contextless" (sc: ^Scalar, i: u64) {
@(require_results)
sc_set_bytes :: proc "contextless" (sc: ^Scalar, b: []byte) -> bool {
if len(b) != 32 {
intrinsics.trap()
}
ensure_contextless(len(b) == 32, "edwards25519: invalid scalar size")
b_ := (^[32]byte)(raw_data(b))
return field.fe_from_bytes(sc, b_)
}
sc_set_bytes_rfc8032 :: proc "contextless" (sc: ^Scalar, b: []byte) {
if len(b) != 32 {
intrinsics.trap()
}
ensure_contextless(len(b) == 32, "edwards25519: invalid scalar size")
b_ := (^[32]byte)(raw_data(b))
field.fe_from_bytes_rfc8032(sc, b_)
}

3
core/crypto/_fiat/field_curve25519/field51.odin
View File

@@ -42,9 +42,12 @@ import "core:math/bits"
Loose_Field_Element :: distinct [5]u64
Tight_Field_Element :: distinct [5]u64
@(rodata)
FE_ZERO := Tight_Field_Element{0, 0, 0, 0, 0}
@(rodata)
FE_ONE := Tight_Field_Element{1, 0, 0, 0, 0}
@(rodata)
FE_SQRT_M1 := Tight_Field_Element {
1718705420411056,
234908883556509,

235
core/crypto/_fiat/field_curve448/field.odin Normal file
View File

@@ -0,0 +1,235 @@
package field_curve448
import "core:mem"
fe_relax_cast :: #force_inline proc "contextless" (
arg1: ^Tight_Field_Element,
) -> ^Loose_Field_Element {
return (^Loose_Field_Element)(arg1)
}
fe_tighten_cast :: #force_inline proc "contextless" (
arg1: ^Loose_Field_Element,
) -> ^Tight_Field_Element {
return (^Tight_Field_Element)(arg1)
}
fe_clear :: proc "contextless" (
arg1: $T,
) where T == ^Tight_Field_Element || T == ^Loose_Field_Element {
mem.zero_explicit(arg1, size_of(arg1^))
}
fe_clear_vec :: proc "contextless" (
arg1: $T,
) where T == []^Tight_Field_Element || T == []^Loose_Field_Element {
for fe in arg1 {
fe_clear(fe)
}
}
fe_carry_mul_small :: proc "contextless" (
out1: ^Tight_Field_Element,
arg1: ^Loose_Field_Element,
arg2: u64,
) {
arg2_ := Loose_Field_Element{arg2, 0, 0, 0, 0, 0, 0, 0}
fe_carry_mul(out1, arg1, &arg2_)
}
fe_carry_pow2k :: proc "contextless" (
out1: ^Tight_Field_Element,
arg1: ^Loose_Field_Element,
arg2: uint,
) {
// Special case: `arg1^(2 * 0) = 1`, though this should never happen.
if arg2 == 0 {
fe_one(out1)
return
}
fe_carry_square(out1, arg1)
for _ in 1 ..< arg2 {
fe_carry_square(out1, fe_relax_cast(out1))
}
}
fe_carry_inv :: proc "contextless" (
out1: ^Tight_Field_Element,
arg1: ^Loose_Field_Element,
) {
// Inversion computation is derived from the addition chain:
//
// _10 = 2*1
// _11 = 1 + _10
// _110 = 2*_11
// _111 = 1 + _110
// _111000 = _111 << 3
// _111111 = _111 + _111000
// x12 = _111111 << 6 + _111111
// x24 = x12 << 12 + x12
// i34 = x24 << 6
// x30 = _111111 + i34
// x48 = i34 << 18 + x24
// x96 = x48 << 48 + x48
// x192 = x96 << 96 + x96
// x222 = x192 << 30 + x30
// x223 = 2*x222 + 1
// return (x223 << 223 + x222) << 2 + 1
//
// Operations: 447 squares 13 multiplies
//
// Generated by github.com/mmcloughlin/addchain v0.4.0.
t0, t1, t2: Tight_Field_Element = ---, ---, ---
// Step 1: t0 = x^0x2
fe_carry_square(&t0, arg1)
// Step 2: t0 = x^0x3
fe_carry_mul(&t0, arg1, fe_relax_cast(&t0))
// t0.Sqr(t0)
fe_carry_square(&t0, fe_relax_cast(&t0))
// Step 4: t0 = x^0x7
fe_carry_mul(&t0, arg1, fe_relax_cast(&t0))
// Step 7: t1 = x^0x38
fe_carry_pow2k(&t1, fe_relax_cast(&t0), 3)
// Step 8: t0 = x^0x3f
fe_carry_mul(&t0, fe_relax_cast(&t0), fe_relax_cast(&t1))
// Step 14: t1 = x^0xfc0
fe_carry_pow2k(&t1, fe_relax_cast(&t0), 6)
// Step 15: t1 = x^0xfff
fe_carry_mul(&t1, fe_relax_cast(&t0), fe_relax_cast(&t1))
// Step 27: t2 = x^0xfff000
fe_carry_pow2k(&t2, fe_relax_cast(&t1), 12)
// Step 28: t1 = x^0xffffff
fe_carry_mul(&t1, fe_relax_cast(&t1), fe_relax_cast(&t2))
// Step 34: t2 = x^0x3fffffc0
fe_carry_pow2k(&t2, fe_relax_cast(&t1), 6)
// Step 35: t0 = x^0x3fffffff
fe_carry_mul(&t0, fe_relax_cast(&t0), fe_relax_cast(&t2))
// Step 53: t2 = x^0xffffff000000
fe_carry_pow2k(&t2, fe_relax_cast(&t2), 18)
// Step 54: t1 = x^0xffffffffffff
fe_carry_mul(&t1, fe_relax_cast(&t1), fe_relax_cast(&t2))
// Step 102: t2 = x^0xffffffffffff000000000000
fe_carry_pow2k(&t2, fe_relax_cast(&t1), 48)
// Step 103: t1 = x^0xffffffffffffffffffffffff
fe_carry_mul(&t1, fe_relax_cast(&t1), fe_relax_cast(&t2))
// Step 199: t2 = x^0xffffffffffffffffffffffff000000000000000000000000
fe_carry_pow2k(&t2, fe_relax_cast(&t1), 96)
// Step 200: t1 = x^0xffffffffffffffffffffffffffffffffffffffffffffffff
fe_carry_mul(&t1, fe_relax_cast(&t1), fe_relax_cast(&t2))
// Step 230: t1 = x^0x3fffffffffffffffffffffffffffffffffffffffffffffffc0000000
fe_carry_pow2k(&t1, fe_relax_cast(&t1), 30)
// Step 231: t0 = x^0x3fffffffffffffffffffffffffffffffffffffffffffffffffffffff
fe_carry_mul(&t0, fe_relax_cast(&t0), fe_relax_cast(&t1))
// Step 232: t1 = x^0x7ffffffffffffffffffffffffffffffffffffffffffffffffffffffe
fe_carry_square(&t1, fe_relax_cast(&t0))
// Step 233: t1 = x^0x7fffffffffffffffffffffffffffffffffffffffffffffffffffffff
fe_carry_mul(&t1, arg1, fe_relax_cast(&t1))
// Step 456: t1 = x^0x3fffffffffffffffffffffffffffffffffffffffffffffffffffffff80000000000000000000000000000000000000000000000000000000
fe_carry_pow2k(&t1, fe_relax_cast(&t1), 223)
// Step 457: t0 = x^0x3fffffffffffffffffffffffffffffffffffffffffffffffffffffffbfffffffffffffffffffffffffffffffffffffffffffffffffffffff
fe_carry_mul(&t0, fe_relax_cast(&t0), fe_relax_cast(&t1))
// Step 459: t0 = x^0xfffffffffffffffffffffffffffffffffffffffffffffffffffffffefffffffffffffffffffffffffffffffffffffffffffffffffffffffc
fe_carry_pow2k(&t0, fe_relax_cast(&t0), 2)
// Step 460: z = x^0xfffffffffffffffffffffffffffffffffffffffffffffffffffffffefffffffffffffffffffffffffffffffffffffffffffffffffffffffd
fe_carry_mul(out1, arg1, fe_relax_cast(&t0))
fe_clear_vec([]^Tight_Field_Element{&t0, &t1, &t2})
}
fe_zero :: proc "contextless" (out1: ^Tight_Field_Element) {
out1[0] = 0
out1[1] = 0
out1[2] = 0
out1[3] = 0
out1[4] = 0
out1[5] = 0
out1[6] = 0
out1[7] = 0
}
fe_one :: proc "contextless" (out1: ^Tight_Field_Element) {
out1[0] = 1
out1[1] = 0
out1[2] = 0
out1[3] = 0
out1[4] = 0
out1[5] = 0
out1[6] = 0
out1[7] = 0
}
fe_set :: proc "contextless" (out1, arg1: ^Tight_Field_Element) {
x1 := arg1[0]
x2 := arg1[1]
x3 := arg1[2]
x4 := arg1[3]
x5 := arg1[4]
x6 := arg1[5]
x7 := arg1[6]
x8 := arg1[7]
out1[0] = x1
out1[1] = x2
out1[2] = x3
out1[3] = x4
out1[4] = x5
out1[5] = x6
out1[6] = x7
out1[7] = x8
}
@(optimization_mode = "none")
fe_cond_swap :: #force_no_inline proc "contextless" (out1, out2: ^Tight_Field_Element, arg1: int) {
mask := (u64(arg1) * 0xffffffffffffffff)
x := (out1[0] ~ out2[0]) & mask
x1, y1 := out1[0] ~ x, out2[0] ~ x
x = (out1[1] ~ out2[1]) & mask
x2, y2 := out1[1] ~ x, out2[1] ~ x
x = (out1[2] ~ out2[2]) & mask
x3, y3 := out1[2] ~ x, out2[2] ~ x
x = (out1[3] ~ out2[3]) & mask
x4, y4 := out1[3] ~ x, out2[3] ~ x
x = (out1[4] ~ out2[4]) & mask
x5, y5 := out1[4] ~ x, out2[4] ~ x
x = (out1[5] ~ out2[5]) & mask
x6, y6 := out1[5] ~ x, out2[5] ~ x
x = (out1[6] ~ out2[6]) & mask
x7, y7 := out1[6] ~ x, out2[6] ~ x
x = (out1[7] ~ out2[7]) & mask
x8, y8 := out1[7] ~ x, out2[7] ~ x
out1[0], out2[0] = x1, y1
out1[1], out2[1] = x2, y2
out1[2], out2[2] = x3, y3
out1[3], out2[3] = x4, y4
out1[4], out2[4] = x5, y5
out1[5], out2[5] = x6, y6
out1[6], out2[6] = x7, y7
out1[7], out2[7] = x8, y8
}

1060
core/crypto/_fiat/field_curve448/field51.odin Normal file
View File

File diff suppressed because it is too large Load Diff

5
core/crypto/_fiat/field_poly1305/field.odin
View File

@@ -1,6 +1,5 @@
package field_poly1305
import "base:intrinsics"
import "core:encoding/endian"
import "core:mem"
@@ -29,9 +28,7 @@ fe_from_bytes :: #force_inline proc "contextless" (
// makes implementing the actual MAC block processing considerably
// neater.
if len(arg1) != 16 {
intrinsics.trap()
}
ensure_contextless(len(arg1) == 16, "poly1305: invalid field element size")
// While it may be unwise to do deserialization here on our
// own when fiat-crypto provides equivalent functionality,

14
core/crypto/_fiat/field_scalar25519/field.odin
View File

@@ -1,18 +1,17 @@
package field_scalar25519
import "base:intrinsics"
import "core:encoding/endian"
import "core:math/bits"
import "core:mem"
@(private)
@(private, rodata)
_TWO_168 := Montgomery_Domain_Field_Element {
0x5b8ab432eac74798,
0x38afddd6de59d5d7,
0xa2c131b399411b7c,
0x6329a7ed9ce5a30,
}
@(private)
@(private, rodata)
_TWO_336 := Montgomery_Domain_Field_Element {
0xbd3d108e2b35ecc5,
0x5c3a3718bdf9c90b,
@@ -95,9 +94,8 @@ fe_from_bytes_wide :: proc "contextless" (
@(private)
_fe_from_bytes_short :: proc "contextless" (out1: ^Montgomery_Domain_Field_Element, arg1: []byte) {
// INVARIANT: len(arg1) < 32.
if len(arg1) >= 32 {
intrinsics.trap()
}
ensure_contextless(len(arg1) < 32, "edwards25519: oversized short scalar")
tmp: [32]byte
copy(tmp[:], arg1)
@@ -106,9 +104,7 @@ _fe_from_bytes_short :: proc "contextless" (out1: ^Montgomery_Domain_Field_Eleme
}
fe_to_bytes :: proc "contextless" (out1: []byte, arg1: ^Montgomery_Domain_Field_Element) {
if len(out1) != 32 {
intrinsics.trap()
}
ensure_contextless(len(out1) == 32, "edwards25519: oversized scalar output buffer")
tmp: Non_Montgomery_Domain_Field_Element
fe_from_montgomery(&tmp, arg1)

39
core/crypto/_sha3/sha3.odin
View File

@@ -44,7 +44,7 @@ Context :: struct {
is_finalized: bool, // For SHAKE (unlimited squeeze is allowed)
}
@(private)
@(private, rodata)
keccakf_rndc := [?]u64 {
0x0000000000000001, 0x0000000000008082, 0x800000000000808a,
0x8000000080008000, 0x000000000000808b, 0x0000000080000001,
@@ -56,13 +56,13 @@ keccakf_rndc := [?]u64 {
0x8000000000008080, 0x0000000080000001, 0x8000000080008008,
}
@(private)
@(private, rodata)
keccakf_rotc := [?]int {
1, 3, 6, 10, 15, 21, 28, 36, 45, 55, 2, 14,
27, 41, 56, 8, 25, 43, 62, 18, 39, 61, 20, 44,
}
@(private)
@(private, rodata)
keccakf_piln := [?]i32 {
10, 7, 11, 17, 18, 3, 5, 16, 8, 21, 24, 4,
15, 23, 19, 13, 12, 2, 20, 14, 22, 9, 6, 1,
@@ -122,7 +122,7 @@ keccakf :: proc "contextless" (st: ^[25]u64) {
}
}
init :: proc(ctx: ^Context) {
init :: proc "contextless" (ctx: ^Context) {
for i := 0; i < 25; i += 1 {
ctx.st.q[i] = 0
}
@@ -133,9 +133,9 @@ init :: proc(ctx: ^Context) {
ctx.is_finalized = false
}
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx.is_initialized)
assert(!ctx.is_finalized)
update :: proc "contextless" (ctx: ^Context, data: []byte) {
ensure_contextless(ctx.is_initialized)
ensure_contextless(!ctx.is_finalized)
j := ctx.pt
for i := 0; i < len(data); i += 1 {
@@ -149,12 +149,9 @@ update :: proc(ctx: ^Context, data: []byte) {
ctx.pt = j
}
final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
assert(ctx.is_initialized)
if len(hash) < ctx.mdlen {
panic("crypto/sha3: invalid destination digest size")
}
final :: proc "contextless" (ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
ensure_contextless(ctx.is_initialized)
ensure_contextless(len(hash) >= ctx.mdlen, "crypto/sha3: invalid destination digest size")
ctx := ctx
if finalize_clone {
@@ -173,11 +170,11 @@ final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
}
}
clone :: proc(ctx, other: ^Context) {
clone :: proc "contextless" (ctx, other: ^Context) {
ctx^ = other^
}
reset :: proc(ctx: ^Context) {
reset :: proc "contextless" (ctx: ^Context) {
if !ctx.is_initialized {
return
}
@@ -185,9 +182,9 @@ reset :: proc(ctx: ^Context) {
mem.zero_explicit(ctx, size_of(ctx^))
}
shake_xof :: proc(ctx: ^Context) {
assert(ctx.is_initialized)
assert(!ctx.is_finalized)
shake_xof :: proc "contextless" (ctx: ^Context) {
ensure_contextless(ctx.is_initialized)
ensure_contextless(!ctx.is_finalized)
ctx.st.b[ctx.pt] ~= ctx.dsbyte
ctx.st.b[ctx.rsiz - 1] ~= 0x80
@@ -197,9 +194,9 @@ shake_xof :: proc(ctx: ^Context) {
ctx.is_finalized = true // No more absorb, unlimited squeeze.
}
shake_out :: proc(ctx: ^Context, hash: []byte) {
assert(ctx.is_initialized)
assert(ctx.is_finalized)
shake_out :: proc "contextless" (ctx: ^Context, hash: []byte) {
ensure_contextless(ctx.is_initialized)
ensure_contextless(ctx.is_finalized)
j := ctx.pt
for i := 0; i < len(hash); i += 1 {

22
core/crypto/_sha3/sp800_185.odin
View File

@@ -3,7 +3,7 @@ package _sha3
import "core:encoding/endian"
import "core:math/bits"
init_cshake :: proc(ctx: ^Context, n, s: []byte, sec_strength: int) {
init_cshake :: proc "contextless" (ctx: ^Context, n, s: []byte, sec_strength: int) {
ctx.mdlen = sec_strength / 8
// No domain separator is equivalent to vanilla SHAKE.
@@ -18,7 +18,7 @@ init_cshake :: proc(ctx: ^Context, n, s: []byte, sec_strength: int) {
bytepad(ctx, [][]byte{n, s}, rate_cshake(sec_strength))
}
final_cshake :: proc(ctx: ^Context, dst: []byte, finalize_clone: bool = false) {
final_cshake :: proc "contextless" (ctx: ^Context, dst: []byte, finalize_clone: bool = false) {
ctx := ctx
if finalize_clone {
tmp_ctx: Context
@@ -32,7 +32,7 @@ final_cshake :: proc(ctx: ^Context, dst: []byte, finalize_clone: bool = false) {
shake_out(ctx, dst)
}
rate_cshake :: #force_inline proc(sec_strength: int) -> int {
rate_cshake :: #force_inline proc "contextless" (sec_strength: int) -> int {
switch sec_strength {
case 128:
return RATE_128
@@ -40,7 +40,7 @@ rate_cshake :: #force_inline proc(sec_strength: int) -> int {
return RATE_256
}
panic("crypto/sha3: invalid security strength")
panic_contextless("crypto/sha3: invalid security strength")
}
// right_encode and left_encode are defined to support 0 <= x < 2^2040
@@ -52,10 +52,10 @@ rate_cshake :: #force_inline proc(sec_strength: int) -> int {
//
// Thus we support 0 <= x < 2^128.
@(private)
@(private, rodata)
_PAD: [RATE_128]byte // Biggest possible value of w per spec.
bytepad :: proc(ctx: ^Context, x_strings: [][]byte, w: int) {
bytepad :: proc "contextless" (ctx: ^Context, x_strings: [][]byte, w: int) {
// 1. z = left_encode(w) || X.
z_hi: u64
z_lo := left_right_encode(ctx, 0, u64(w), true)
@@ -70,9 +70,7 @@ bytepad :: proc(ctx: ^Context, x_strings: [][]byte, w: int) {
// This isn't actually possible, at least with the currently
// defined SP 800-185 routines.
if carry != 0 {
panic("crypto/sha3: bytepad input length overflow")
}
ensure_contextless(carry == 0, "crypto/sha3: bytepad input length overflow")
}
// We skip this step as we are doing a byte-oriented implementation
@@ -95,7 +93,7 @@ bytepad :: proc(ctx: ^Context, x_strings: [][]byte, w: int) {
}
}
encode_string :: #force_inline proc(ctx: ^Context, s: []byte) -> (u64, u64) {
encode_string :: #force_inline proc "contextless" (ctx: ^Context, s: []byte) -> (u64, u64) {
l := encode_byte_len(ctx, len(s), true) // left_encode
update(ctx, s)
@@ -104,13 +102,13 @@ encode_string :: #force_inline proc(ctx: ^Context, s: []byte) -> (u64, u64) {
return hi, lo
}
encode_byte_len :: #force_inline proc(ctx: ^Context, l: int, is_left: bool) -> u64 {
encode_byte_len :: #force_inline proc "contextless" (ctx: ^Context, l: int, is_left: bool) -> u64 {
hi, lo := bits.mul_u64(u64(l), 8)
return left_right_encode(ctx, hi, lo, is_left)
}
@(private)
left_right_encode :: proc(ctx: ^Context, hi, lo: u64, is_left: bool) -> u64 {
left_right_encode :: proc "contextless" (ctx: ^Context, hi, lo: u64, is_left: bool) -> u64 {
HI_OFFSET :: 1
LO_OFFSET :: HI_OFFSET + 8
RIGHT_OFFSET :: LO_OFFSET + 8

2
core/crypto/aead/aead.odin
View File

@@ -16,7 +16,7 @@ seal_oneshot :: proc(algo: Algorithm, dst, tag, key, iv, aad, plaintext: []byte,
// returning true iff the authentication was successful. If authentication
// fails, the destination buffer will be zeroed.
//
// dst and plaintext MUST alias exactly or not at all.
// dst and ciphertext MUST alias exactly or not at all.
@(require_results)
open_oneshot :: proc(algo: Algorithm, dst, key, iv, aad, ciphertext, tag: []byte, impl: Implementation = nil) -> bool {
ctx: Context

64
core/crypto/aead/low_level.odin
View File

@@ -1,8 +1,10 @@
package aead
import "core:crypto/aegis"
import "core:crypto/aes"
import "core:crypto/chacha20"
import "core:crypto/chacha20poly1305"
import "core:crypto/deoxysii"
import "core:reflect"
// Implementation is an AEAD implementation. Most callers will not need
@@ -15,7 +17,7 @@ Implementation :: union {
// MAX_TAG_SIZE is the maximum size tag that can be returned by any of the
// Algorithms supported via this package.
MAX_TAG_SIZE :: 16
MAX_TAG_SIZE :: 32
// Algorithm is the algorithm identifier associated with a given Context.
Algorithm :: enum {
@@ -25,9 +27,14 @@ Algorithm :: enum {
AES_GCM_256,
CHACHA20POLY1305,
XCHACHA20POLY1305,
AEGIS_128L,
AEGIS_128L_256, // AEGIS-128L (256-bit tag)
AEGIS_256,
AEGIS_256_256, // AEGIS-256 (256-bit tag)
DEOXYS_II_256,
}
// ALGORITM_NAMES is the Agorithm to algorithm name string.
// ALGORITM_NAMES is the Algorithm to algorithm name string.
ALGORITHM_NAMES := [Algorithm]string {
.Invalid = "Invalid",
.AES_GCM_128 = "AES-GCM-128",
@@ -35,6 +42,11 @@ ALGORITHM_NAMES := [Algorithm]string {
.AES_GCM_256 = "AES-GCM-256",
.CHACHA20POLY1305 = "chacha20poly1305",
.XCHACHA20POLY1305 = "xchacha20poly1305",
.AEGIS_128L = "AEGIS-128L",
.AEGIS_128L_256 = "AEGIS-128L-256",
.AEGIS_256 = "AEGIS-256",
.AEGIS_256_256 = "AEGIS-256-256",
.DEOXYS_II_256 = "Deoxys-II-256",
}
// TAG_SIZES is the Algorithm to tag size in bytes.
@@ -45,6 +57,11 @@ TAG_SIZES := [Algorithm]int {
.AES_GCM_256 = aes.GCM_TAG_SIZE,
.CHACHA20POLY1305 = chacha20poly1305.TAG_SIZE,
.XCHACHA20POLY1305 = chacha20poly1305.TAG_SIZE,
.AEGIS_128L = aegis.TAG_SIZE_128,
.AEGIS_128L_256 = aegis.TAG_SIZE_256,
.AEGIS_256 = aegis.TAG_SIZE_128,
.AEGIS_256_256 = aegis.TAG_SIZE_256,
.DEOXYS_II_256 = deoxysii.TAG_SIZE,
}
// KEY_SIZES is the Algorithm to key size in bytes.
@@ -55,6 +72,11 @@ KEY_SIZES := [Algorithm]int {
.AES_GCM_256 = aes.KEY_SIZE_256,
.CHACHA20POLY1305 = chacha20poly1305.KEY_SIZE,
.XCHACHA20POLY1305 = chacha20poly1305.KEY_SIZE,
.AEGIS_128L = aegis.KEY_SIZE_128L,
.AEGIS_128L_256 = aegis.KEY_SIZE_128L,
.AEGIS_256 = aegis.KEY_SIZE_256,
.AEGIS_256_256 = aegis.KEY_SIZE_256,
.DEOXYS_II_256 = deoxysii.KEY_SIZE,
}
// IV_SIZES is the Algorithm to initialization vector size in bytes.
@@ -67,6 +89,11 @@ IV_SIZES := [Algorithm]int {
.AES_GCM_256 = aes.GCM_IV_SIZE,
.CHACHA20POLY1305 = chacha20poly1305.IV_SIZE,
.XCHACHA20POLY1305 = chacha20poly1305.XIV_SIZE,
.AEGIS_128L = aegis.IV_SIZE_128L,
.AEGIS_128L_256 = aegis.IV_SIZE_128L,
.AEGIS_256 = aegis.IV_SIZE_256,
.AEGIS_256_256 = aegis.IV_SIZE_256,
.DEOXYS_II_256 = deoxysii.IV_SIZE,
}
// Context is a concrete instantiation of a specific AEAD algorithm.
@@ -75,6 +102,8 @@ Context :: struct {
_impl: union {
aes.Context_GCM,
chacha20poly1305.Context,
aegis.Context,
deoxysii.Context,
},
}
@@ -86,6 +115,11 @@ _IMPL_IDS := [Algorithm]typeid {
.AES_GCM_256 = typeid_of(aes.Context_GCM),
.CHACHA20POLY1305 = typeid_of(chacha20poly1305.Context),
.XCHACHA20POLY1305 = typeid_of(chacha20poly1305.Context),
.AEGIS_128L = typeid_of(aegis.Context),
.AEGIS_128L_256 = typeid_of(aegis.Context),
.AEGIS_256 = typeid_of(aegis.Context),
.AEGIS_256_256 = typeid_of(aegis.Context),
.DEOXYS_II_256 = typeid_of(deoxysii.Context),
}
// init initializes a Context with a specific AEAD Algorithm.
@@ -94,9 +128,7 @@ init :: proc(ctx: ^Context, algorithm: Algorithm, key: []byte, impl: Implementat
reset(ctx)
}
if len(key) != KEY_SIZES[algorithm] {
panic("crypto/aead: invalid key size")
}
ensure(len(key) == KEY_SIZES[algorithm], "crypto/aead: invalid key size")
// Directly specialize the union by setting the type ID (save a copy).
reflect.set_union_variant_typeid(
@@ -113,6 +145,12 @@ init :: proc(ctx: ^Context, algorithm: Algorithm, key: []byte, impl: Implementat
case .XCHACHA20POLY1305:
impl_ := impl != nil ? impl.(chacha20.Implementation) : chacha20.DEFAULT_IMPLEMENTATION
chacha20poly1305.init_xchacha(&ctx._impl.(chacha20poly1305.Context), key, impl_)
case .AEGIS_128L, .AEGIS_128L_256, .AEGIS_256, .AEGIS_256_256:
impl_ := impl != nil ? impl.(aes.Implementation) : aes.DEFAULT_IMPLEMENTATION
aegis.init(&ctx._impl.(aegis.Context), key, impl_)
case .DEOXYS_II_256:
impl_ := impl != nil ? impl.(aes.Implementation) : aes.DEFAULT_IMPLEMENTATION
deoxysii.init(&ctx._impl.(deoxysii.Context), key, impl_)
case .Invalid:
panic("crypto/aead: uninitialized algorithm")
case:
@@ -127,11 +165,17 @@ init :: proc(ctx: ^Context, algorithm: Algorithm, key: []byte, impl: Implementat
//
// dst and plaintext MUST alias exactly or not at all.
seal_ctx :: proc(ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) {
ensure(len(tag) == TAG_SIZES[ctx._algo], "crypto/aead: invalid tag size")
switch &impl in ctx._impl {
case aes.Context_GCM:
aes.seal_gcm(&impl, dst, tag, iv, aad, plaintext)
case chacha20poly1305.Context:
chacha20poly1305.seal(&impl, dst, tag, iv, aad, plaintext)
case aegis.Context:
aegis.seal(&impl, dst, tag, iv, aad, plaintext)
case deoxysii.Context:
deoxysii.seal(&impl, dst, tag, iv, aad, plaintext)
case:
panic("crypto/aead: uninitialized algorithm")
}
@@ -145,11 +189,17 @@ seal_ctx :: proc(ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) {
// dst and plaintext MUST alias exactly or not at all.
@(require_results)
open_ctx :: proc(ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
ensure(len(tag) == TAG_SIZES[ctx._algo], "crypto/aead: invalid tag size")
switch &impl in ctx._impl {
case aes.Context_GCM:
return aes.open_gcm(&impl, dst, iv, aad, ciphertext, tag)
case chacha20poly1305.Context:
return chacha20poly1305.open(&impl, dst, iv, aad, ciphertext, tag)
case aegis.Context:
return aegis.open(&impl, dst, iv, aad, ciphertext, tag)
case deoxysii.Context:
return deoxysii.open(&impl, dst, iv, aad, ciphertext, tag)
case:
panic("crypto/aead: uninitialized algorithm")
}
@@ -163,6 +213,10 @@ reset :: proc(ctx: ^Context) {
aes.reset_gcm(&impl)
case chacha20poly1305.Context:
chacha20poly1305.reset(&impl)
case aegis.Context:
aegis.reset(&impl)
case deoxysii.Context:
deoxysii.reset(&impl)
case:
// Calling reset repeatedly is fine.
}

213
core/crypto/aegis/aegis.odin Normal file
View File

@@ -0,0 +1,213 @@
/*
package aegis implements the AEGIS-128L and AEGIS-256 Authenticated
Encryption with Additional Data algorithms.
See:
- [[ https://www.ietf.org/archive/id/draft-irtf-cfrg-aegis-aead-12.txt ]]
*/
package aegis
import "core:bytes"
import "core:crypto"
import "core:crypto/aes"
import "core:mem"
// KEY_SIZE_128L is the AEGIS-128L key size in bytes.
KEY_SIZE_128L :: 16
// KEY_SIZE_256 is the AEGIS-256 key size in bytes.
KEY_SIZE_256 :: 32
// IV_SIZE_128L is the AEGIS-128L IV size in bytes.
IV_SIZE_128L :: 16
// IV_SIZE_256 is the AEGIS-256 IV size in bytes.
IV_SIZE_256 :: 32
// TAG_SIZE_128 is the AEGIS-128L or AEGIS-256 128-bit tag size in bytes.
TAG_SIZE_128 :: 16
// TAG_SIZE_256 is the AEGIS-128L or AEGIS-256 256-bit tag size in bytes.
TAG_SIZE_256 :: 32
@(private)
_RATE_128L :: 32
@(private)
_RATE_256 :: 16
@(private)
_RATE_MAX :: _RATE_128L
@(private, rodata)
_C0 := [16]byte{
0x00, 0x01, 0x01, 0x02, 0x03, 0x05, 0x08, 0x0d,
0x15, 0x22, 0x37, 0x59, 0x90, 0xe9, 0x79, 0x62,
}
@(private, rodata)
_C1 := [16]byte {
0xdb, 0x3d, 0x18, 0x55, 0x6d, 0xc2, 0x2f, 0xf1,
0x20, 0x11, 0x31, 0x42, 0x73, 0xb5, 0x28, 0xdd,
}
// Context is a keyed AEGIS-128L or AEGIS-256 instance.
Context :: struct {
_key: [KEY_SIZE_256]byte,
_key_len: int,
_impl: aes.Implementation,
_is_initialized: bool,
}
@(private)
_validate_common_slice_sizes :: proc (ctx: ^Context, tag, iv, aad, text: []byte) {
switch len(tag) {
case TAG_SIZE_128, TAG_SIZE_256:
case:
panic("crypto/aegis: invalid tag size")
}
iv_ok: bool
switch ctx._key_len {
case KEY_SIZE_128L:
iv_ok = len(iv) == IV_SIZE_128L
case KEY_SIZE_256:
iv_ok = len(iv) == IV_SIZE_256
}
ensure(iv_ok,"crypto/aegis: invalid IV size")
#assert(size_of(int) == 8 || size_of(int) <= 4)
// As A_MAX and P_MAX are both defined to be 2^61 - 1 bytes, and
// the maximum length of a slice is bound by `size_of(int)`, where
// `int` is register sized, there is no need to check AAD/text
// lengths.
}
// init initializes a Context with the provided key, for AEGIS-128L or AEGIS-256.
init :: proc(ctx: ^Context, key: []byte, impl := aes.DEFAULT_IMPLEMENTATION) {
switch len(key) {
case KEY_SIZE_128L, KEY_SIZE_256:
case:
panic("crypto/aegis: invalid key size")
}
copy(ctx._key[:], key)
ctx._key_len = len(key)
ctx._impl = impl
if ctx._impl == .Hardware && !is_hardware_accelerated() {
ctx._impl = .Portable
}
ctx._is_initialized = true
}
// seal encrypts the plaintext and authenticates the aad and ciphertext,
// with the provided Context and iv, stores the output in dst and tag.
//
// dst and plaintext MUST alias exactly or not at all.
seal :: proc(ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) {
ensure(ctx._is_initialized)
_validate_common_slice_sizes(ctx, tag, iv, aad, plaintext)
ensure(len(dst) == len(plaintext), "crypto/aegis: invalid destination ciphertext size")
ensure(!bytes.alias_inexactly(dst, plaintext), "crypto/aegis: dst and plaintext alias inexactly")
switch ctx._impl {
case .Hardware:
st: State_HW
defer reset_state_hw(&st)
init_hw(ctx, &st, iv)
aad_len, pt_len := len(aad), len(plaintext)
if aad_len > 0 {
absorb_hw(&st, aad)
}
if pt_len > 0 {
enc_hw(&st, dst, plaintext)
}
finalize_hw(&st, tag, aad_len, pt_len)
case .Portable:
st: State_SW
defer reset_state_sw(&st)
init_sw(ctx, &st, iv)
aad_len, pt_len := len(aad), len(plaintext)
if aad_len > 0 {
absorb_sw(&st, aad)
}
if pt_len > 0 {
enc_sw(&st, dst, plaintext)
}
finalize_sw(&st, tag, aad_len, pt_len)
case:
panic("core/crypto/aegis: not implemented")
}
}
// open authenticates the aad and ciphertext, and decrypts the ciphertext,
// with the provided Context, iv, and tag, and stores the output in dst,
// returning true iff the authentication was successful. If authentication
// fails, the destination buffer will be zeroed.
//
// dst and plaintext MUST alias exactly or not at all.
@(require_results)
open :: proc(ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
ensure(ctx._is_initialized)
_validate_common_slice_sizes(ctx, tag, iv, aad, ciphertext)
ensure(len(dst) == len(ciphertext), "crypto/aegis: invalid destination plaintext size")
ensure(!bytes.alias_inexactly(dst, ciphertext), "crypto/aegis: dst and ciphertext alias inexactly")
tmp: [TAG_SIZE_256]byte
derived_tag := tmp[:len(tag)]
aad_len, ct_len := len(aad), len(ciphertext)
switch ctx._impl {
case .Hardware:
st: State_HW
defer reset_state_hw(&st)
init_hw(ctx, &st, iv)
if aad_len > 0 {
absorb_hw(&st, aad)
}
if ct_len > 0 {
dec_hw(&st, dst, ciphertext)
}
finalize_hw(&st, derived_tag, aad_len, ct_len)
case .Portable:
st: State_SW
defer reset_state_sw(&st)
init_sw(ctx, &st, iv)
if aad_len > 0 {
absorb_sw(&st, aad)
}
if ct_len > 0 {
dec_sw(&st, dst, ciphertext)
}
finalize_sw(&st, derived_tag, aad_len, ct_len)
case:
panic("core/crypto/aegis: not implemented")
}
if crypto.compare_constant_time(tag, derived_tag) != 1 {
mem.zero_explicit(raw_data(derived_tag), len(derived_tag))
mem.zero_explicit(raw_data(dst), ct_len)
return false
}
return true
}
// reset sanitizes the Context. The Context must be
// re-initialized to be used again.
reset :: proc "contextless" (ctx: ^Context) {
mem.zero_explicit(&ctx._key, len(ctx._key))
ctx._key_len = 0
ctx._is_initialized = false
}

452
core/crypto/aegis/aegis_impl_ct64.odin Normal file
View File

@@ -0,0 +1,452 @@
package aegis
import aes "core:crypto/_aes/ct64"
import "core:encoding/endian"
import "core:mem"
// This uses the bitlsiced 64-bit general purpose register SWAR AES
// round function. The intermediate state is stored in interleaved
// but NOT orthogonalized form, as leaving things in the orthgonalized
// format would overly complicate the update implementation.
//
// Note/perf: Per Frank Denis and a review of the specification, it is
// possible to gain slightly more performance by leaving the state in
// orthogonalized form while doing initialization, finalization, and
// absorbing AAD. This implementation opts out of those optimizations
// for the sake of simplicity.
//
// The update function leverages the paralleism (4xblocks) at once.
@(private)
State_SW :: struct {
s0_0, s0_1: u64,
s1_0, s1_1: u64,
s2_0, s2_1: u64,
s3_0, s3_1: u64,
s4_0, s4_1: u64,
s5_0, s5_1: u64,
s6_0, s6_1: u64,
s7_0, s7_1: u64,
q_k, q_b: [8]u64,
rate: int,
}
@(private)
init_sw :: proc "contextless" (ctx: ^Context, st: ^State_SW, iv: []byte) {
switch ctx._key_len {
case KEY_SIZE_128L:
key_0, key_1 := aes.load_interleaved(ctx._key[:16])
iv_0, iv_1 := aes.load_interleaved(iv)
st.s0_0, st.s0_1 = aes.xor_interleaved(key_0, key_1, iv_0, iv_1)
st.s1_0, st.s1_1 = aes.load_interleaved(_C1[:])
st.s2_0, st.s2_1 = aes.load_interleaved(_C0[:])
st.s3_0, st.s3_1 = st.s1_0, st.s1_1
st.s4_0, st.s4_1 = st.s0_0, st.s0_1
st.s5_0, st.s5_1 = aes.xor_interleaved(key_0, key_1, st.s2_0, st.s2_1)
st.s6_0, st.s6_1 = aes.xor_interleaved(key_0, key_1, st.s1_0, st.s1_1)
st.s7_0, st.s7_1 = st.s5_0, st.s5_1
st.rate = _RATE_128L
for _ in 0 ..< 10 {
update_sw_128l(st, iv_0, iv_1, key_0, key_1)
}
case KEY_SIZE_256:
k0_0, k0_1 := aes.load_interleaved(ctx._key[:16])
k1_0, k1_1 := aes.load_interleaved(ctx._key[16:])
n0_0, n0_1 := aes.load_interleaved(iv[:16])
n1_0, n1_1 := aes.load_interleaved(iv[16:])
st.s0_0, st.s0_1 = aes.xor_interleaved(k0_0, k0_1, n0_0, n0_1)
st.s1_0, st.s1_1 = aes.xor_interleaved(k1_0, k1_1, n1_0, n1_1)
st.s2_0, st.s2_1 = aes.load_interleaved(_C1[:])
st.s3_0, st.s3_1 = aes.load_interleaved(_C0[:])
st.s4_0, st.s4_1 = aes.xor_interleaved(k0_0, k0_1, st.s3_0, st.s3_1)
st.s5_0, st.s5_1 = aes.xor_interleaved(k1_0, k1_1, st.s2_0, st.s2_1)
st.rate = _RATE_256
u0_0, u0_1, u1_0, u1_1 := st.s0_0, st.s0_1, st.s1_0, st.s1_1
for _ in 0 ..< 4 {
update_sw_256(st, k0_0, k0_1)
update_sw_256(st, k1_0, k1_1)
update_sw_256(st, u0_0, u0_1)
update_sw_256(st, u1_0, u1_1)
}
}
}
@(private = "file")
update_sw_128l :: proc "contextless" (st: ^State_SW, m0_0, m0_1, m1_0, m1_1: u64) {
st.q_k[0], st.q_k[4] = aes.xor_interleaved(st.s0_0, st.s0_1, m0_0, m0_1)
st.q_k[1], st.q_k[5] = st.s1_0, st.s1_1
st.q_k[2], st.q_k[6] = st.s2_0, st.s2_1
st.q_k[3], st.q_k[7] = st.s3_0, st.s3_1
aes.orthogonalize(&st.q_k)
st.q_b[0], st.q_b[4] = st.s7_0, st.s7_1
st.q_b[1], st.q_b[5] = st.s0_0, st.s0_1
st.q_b[2], st.q_b[6] = st.s1_0, st.s1_1
st.q_b[3], st.q_b[7] = st.s2_0, st.s2_1
aes.orthogonalize(&st.q_b)
aes.sub_bytes(&st.q_b)
aes.shift_rows(&st.q_b)
aes.mix_columns(&st.q_b)
aes.add_round_key(&st.q_b, st.q_k[:])
aes.orthogonalize(&st.q_b)
st.s0_0, st.s0_1 = st.q_b[0], st.q_b[4]
st.s1_0, st.s1_1 = st.q_b[1], st.q_b[5]
st.s2_0, st.s2_1 = st.q_b[2], st.q_b[6]
s3_0, s3_1 := st.q_b[3], st.q_b[7]
st.q_k[0], st.q_k[4] = aes.xor_interleaved(st.s4_0, st.s4_1, m1_0, m1_1)
st.q_k[1], st.q_k[5] = st.s5_0, st.s5_1
st.q_k[2], st.q_k[6] = st.s6_0, st.s6_1
st.q_k[3], st.q_k[7] = st.s7_0, st.s7_1
aes.orthogonalize(&st.q_k)
st.q_b[0], st.q_b[4] = st.s3_0, st.s3_1
st.q_b[1], st.q_b[5] = st.s4_0, st.s4_1
st.q_b[2], st.q_b[6] = st.s5_0, st.s5_1
st.q_b[3], st.q_b[7] = st.s6_0, st.s6_1
aes.orthogonalize(&st.q_b)
aes.sub_bytes(&st.q_b)
aes.shift_rows(&st.q_b)
aes.mix_columns(&st.q_b)
aes.add_round_key(&st.q_b, st.q_k[:])
aes.orthogonalize(&st.q_b)
st.s3_0, st.s3_1 = s3_0, s3_1
st.s4_0, st.s4_1 = st.q_b[0], st.q_b[4]
st.s5_0, st.s5_1 = st.q_b[1], st.q_b[5]
st.s6_0, st.s6_1 = st.q_b[2], st.q_b[6]
st.s7_0, st.s7_1 = st.q_b[3], st.q_b[7]
}
@(private = "file")
update_sw_256 :: proc "contextless" (st: ^State_SW, m_0, m_1: u64) {
st.q_k[0], st.q_k[4] = aes.xor_interleaved(st.s0_0, st.s0_1, m_0, m_1)
st.q_k[1], st.q_k[5] = st.s1_0, st.s1_1
st.q_k[2], st.q_k[6] = st.s2_0, st.s2_1
st.q_k[3], st.q_k[7] = st.s3_0, st.s3_1
aes.orthogonalize(&st.q_k)
st.q_b[0], st.q_b[4] = st.s5_0, st.s5_1
st.q_b[1], st.q_b[5] = st.s0_0, st.s0_1
st.q_b[2], st.q_b[6] = st.s1_0, st.s1_1
st.q_b[3], st.q_b[7] = st.s2_0, st.s2_1
aes.orthogonalize(&st.q_b)
aes.sub_bytes(&st.q_b)
aes.shift_rows(&st.q_b)
aes.mix_columns(&st.q_b)
aes.add_round_key(&st.q_b, st.q_k[:])
aes.orthogonalize(&st.q_b)
st.s0_0, st.s0_1 = st.q_b[0], st.q_b[4]
st.s1_0, st.s1_1 = st.q_b[1], st.q_b[5]
st.s2_0, st.s2_1 = st.q_b[2], st.q_b[6]
s3_0, s3_1 := st.q_b[3], st.q_b[7]
st.q_k[0], st.q_k[4] = st.s4_0, st.s4_1
st.q_k[1], st.q_k[5] = st.s5_0, st.s5_1
aes.orthogonalize(&st.q_k)
st.q_b[0], st.q_b[4] = st.s3_0, st.s3_1
st.q_b[1], st.q_b[5] = st.s4_0, st.s4_1
aes.orthogonalize(&st.q_b)
aes.sub_bytes(&st.q_b)
aes.shift_rows(&st.q_b)
aes.mix_columns(&st.q_b)
aes.add_round_key(&st.q_b, st.q_k[:])
aes.orthogonalize(&st.q_b)
st.s3_0, st.s3_1 = s3_0, s3_1
st.s4_0, st.s4_1 = st.q_b[0], st.q_b[4]
st.s5_0, st.s5_1 = st.q_b[1], st.q_b[5]
}
@(private = "file")
absorb_sw_128l :: #force_inline proc "contextless" (st: ^State_SW, ai: []byte) #no_bounds_check {
t0_0, t0_1 := aes.load_interleaved(ai[:16])
t1_0, t1_1 := aes.load_interleaved(ai[16:])
update_sw_128l(st, t0_0, t0_1, t1_0, t1_1)
}
@(private = "file")
absorb_sw_256 :: #force_inline proc "contextless" (st: ^State_SW, ai: []byte) {
m_0, m_1 := aes.load_interleaved(ai)
update_sw_256(st, m_0, m_1)
}
@(private)
absorb_sw :: proc "contextless" (st: ^State_SW, aad: []byte) #no_bounds_check {
ai, l := aad, len(aad)
switch st.rate {
case _RATE_128L:
for l >= _RATE_128L {
absorb_sw_128l(st, ai)
ai = ai[_RATE_128L:]
l -= _RATE_128L
}
case _RATE_256:
for l >= _RATE_256 {
absorb_sw_256(st, ai)
ai = ai[_RATE_256:]
l -= _RATE_256
}
}
// Pad out the remainder with `0`s till it is rate sized.
if l > 0 {
tmp: [_RATE_MAX]byte // AAD is not confidential.
copy(tmp[:], ai)
switch st.rate {
case _RATE_128L:
absorb_sw_128l(st, tmp[:])
case _RATE_256:
absorb_sw_256(st, tmp[:])
}
}
}
@(private = "file", require_results)
z_sw_128l :: proc "contextless" (st: ^State_SW) -> (u64, u64, u64, u64) {
z0_0, z0_1 := aes.and_interleaved(st.s2_0, st.s2_1, st.s3_0, st.s3_1)
z0_0, z0_1 = aes.xor_interleaved(st.s1_0, st.s1_1, z0_0, z0_1)
z0_0, z0_1 = aes.xor_interleaved(st.s6_0, st.s6_1, z0_0, z0_1)
z1_0, z1_1 := aes.and_interleaved(st.s6_0, st.s6_1, st.s7_0, st.s7_1)
z1_0, z1_1 = aes.xor_interleaved(st.s5_0, st.s5_1, z1_0, z1_1)
z1_0, z1_1 = aes.xor_interleaved(st.s2_0, st.s2_1, z1_0, z1_1)
return z0_0, z0_1, z1_0, z1_1
}
@(private = "file", require_results)
z_sw_256 :: proc "contextless" (st: ^State_SW) -> (u64, u64) {
z_0, z_1 := aes.and_interleaved(st.s2_0, st.s2_1, st.s3_0, st.s3_1)
z_0, z_1 = aes.xor_interleaved(st.s5_0, st.s5_1, z_0, z_1)
z_0, z_1 = aes.xor_interleaved(st.s4_0, st.s4_1, z_0, z_1)
return aes.xor_interleaved(st.s1_0, st.s1_1, z_0, z_1)
}
@(private = "file")
enc_sw_128l :: #force_inline proc "contextless" (st: ^State_SW, ci, xi: []byte) #no_bounds_check {
z0_0, z0_1, z1_0, z1_1 := z_sw_128l(st)
t0_0, t0_1 := aes.load_interleaved(xi[:16])
t1_0, t1_1 := aes.load_interleaved(xi[16:])
update_sw_128l(st, t0_0, t0_1, t1_0, t1_1)
out0_0, out0_1 := aes.xor_interleaved(t0_0, t0_1, z0_0, z0_1)
out1_0, out1_1 := aes.xor_interleaved(t1_0, t1_1, z1_0, z1_1)
aes.store_interleaved(ci[:16], out0_0, out0_1)
aes.store_interleaved(ci[16:], out1_0, out1_1)
}
@(private = "file")
enc_sw_256 :: #force_inline proc "contextless" (st: ^State_SW, ci, xi: []byte) #no_bounds_check {
z_0, z_1 := z_sw_256(st)
xi_0, xi_1 := aes.load_interleaved(xi)
update_sw_256(st, xi_0, xi_1)
ci_0, ci_1 := aes.xor_interleaved(xi_0, xi_1, z_0, z_1)
aes.store_interleaved(ci, ci_0, ci_1)
}
@(private)
enc_sw :: proc "contextless" (st: ^State_SW, dst, src: []byte) #no_bounds_check {
ci, xi, l := dst, src, len(src)
switch st.rate {
case _RATE_128L:
for l >= _RATE_128L {
enc_sw_128l(st, ci, xi)
ci = ci[_RATE_128L:]
xi = xi[_RATE_128L:]
l -= _RATE_128L
}
case _RATE_256:
for l >= _RATE_256 {
enc_sw_256(st, ci, xi)
ci = ci[_RATE_256:]
xi = xi[_RATE_256:]
l -= _RATE_256
}
}
// Pad out the remainder with `0`s till it is rate sized.
if l > 0 {
tmp: [_RATE_MAX]byte // Ciphertext is not confidential.
copy(tmp[:], xi)
switch st.rate {
case _RATE_128L:
enc_sw_128l(st, tmp[:], tmp[:])
case _RATE_256:
enc_sw_256(st, tmp[:], tmp[:])
}
copy(ci, tmp[:l])
}
}
@(private = "file")
dec_sw_128l :: #force_inline proc "contextless" (st: ^State_SW, xi, ci: []byte) #no_bounds_check {
z0_0, z0_1, z1_0, z1_1 := z_sw_128l(st)
t0_0, t0_1 := aes.load_interleaved(ci[:16])
t1_0, t1_1 := aes.load_interleaved(ci[16:])
out0_0, out0_1 := aes.xor_interleaved(t0_0, t0_1, z0_0, z0_1)
out1_0, out1_1 := aes.xor_interleaved(t1_0, t1_1, z1_0, z1_1)
update_sw_128l(st, out0_0, out0_1, out1_0, out1_1)
aes.store_interleaved(xi[:16], out0_0, out0_1)
aes.store_interleaved(xi[16:], out1_0, out1_1)
}
@(private = "file")
dec_sw_256 :: #force_inline proc "contextless" (st: ^State_SW, xi, ci: []byte) #no_bounds_check {
z_0, z_1 := z_sw_256(st)
ci_0, ci_1 := aes.load_interleaved(ci)
xi_0, xi_1 := aes.xor_interleaved(ci_0, ci_1, z_0, z_1)
update_sw_256(st, xi_0, xi_1)
aes.store_interleaved(xi, xi_0, xi_1)
}
@(private = "file")
dec_partial_sw_128l :: proc "contextless" (st: ^State_SW, xn, cn: []byte) #no_bounds_check {
tmp: [_RATE_128L]byte
defer mem.zero_explicit(&tmp, size_of(tmp))
z0_0, z0_1, z1_0, z1_1 := z_sw_128l(st)
copy(tmp[:], cn)
t0_0, t0_1 := aes.load_interleaved(tmp[:16])
t1_0, t1_1 := aes.load_interleaved(tmp[16:])
out0_0, out0_1 := aes.xor_interleaved(t0_0, t0_1, z0_0, z0_1)
out1_0, out1_1 := aes.xor_interleaved(t1_0, t1_1, z1_0, z1_1)
aes.store_interleaved(tmp[:16], out0_0, out0_1)
aes.store_interleaved(tmp[16:], out1_0, out1_1)
copy(xn, tmp[:])
for off := len(xn); off < _RATE_128L; off += 1 {
tmp[off] = 0
}
out0_0, out0_1 = aes.load_interleaved(tmp[:16])
out1_0, out1_1 = aes.load_interleaved(tmp[16:])
update_sw_128l(st, out0_0, out0_1, out1_0, out1_1)
}
@(private = "file")
dec_partial_sw_256 :: proc "contextless" (st: ^State_SW, xn, cn: []byte) #no_bounds_check {
tmp: [_RATE_256]byte
defer mem.zero_explicit(&tmp, size_of(tmp))
z_0, z_1 := z_sw_256(st)
copy(tmp[:], cn)
cn_0, cn_1 := aes.load_interleaved(tmp[:])
xn_0, xn_1 := aes.xor_interleaved(cn_0, cn_1, z_0, z_1)
aes.store_interleaved(tmp[:], xn_0, xn_1)
copy(xn, tmp[:])
for off := len(xn); off < _RATE_256; off += 1 {
tmp[off] = 0
}
xn_0, xn_1 = aes.load_interleaved(tmp[:])
update_sw_256(st, xn_0, xn_1)
}
@(private)
dec_sw :: proc "contextless" (st: ^State_SW, dst, src: []byte) #no_bounds_check {
xi, ci, l := dst, src, len(src)
switch st.rate {
case _RATE_128L:
for l >= _RATE_128L {
dec_sw_128l(st, xi, ci)
xi = xi[_RATE_128L:]
ci = ci[_RATE_128L:]
l -= _RATE_128L
}
case _RATE_256:
for l >= _RATE_256 {
dec_sw_256(st, xi, ci)
xi = xi[_RATE_256:]
ci = ci[_RATE_256:]
l -= _RATE_256
}
}
// Process the remainder.
if l > 0 {
switch st.rate {
case _RATE_128L:
dec_partial_sw_128l(st, xi, ci)
case _RATE_256:
dec_partial_sw_256(st, xi, ci)
}
}
}
@(private)
finalize_sw :: proc "contextless" (st: ^State_SW, tag: []byte, ad_len, msg_len: int) {
tmp: [16]byte
endian.unchecked_put_u64le(tmp[0:], u64(ad_len) * 8)
endian.unchecked_put_u64le(tmp[8:], u64(msg_len) * 8)
t_0, t_1 := aes.load_interleaved(tmp[:])
t0_0, t0_1, t1_0, t1_1: u64 = ---, ---, ---, ---
switch st.rate {
case _RATE_128L:
t_0, t_1 = aes.xor_interleaved(st.s2_0, st.s2_1, t_0, t_1)
for _ in 0 ..< 7 {
update_sw_128l(st, t_0, t_1, t_0, t_1)
}
t0_0, t0_1 = aes.xor_interleaved(st.s0_0, st.s0_1, st.s1_0, st.s1_1)
t0_0, t0_1 = aes.xor_interleaved(t0_0, t0_1, st.s2_0, st.s2_1)
t0_0, t0_1 = aes.xor_interleaved(t0_0, t0_1, st.s3_0, st.s3_1)
t1_0, t1_1 = aes.xor_interleaved(st.s4_0, st.s4_1, st.s5_0, st.s5_1)
t1_0, t1_1 = aes.xor_interleaved(t1_0, t1_1, st.s6_0, st.s6_1)
if len(tag) == TAG_SIZE_256 {
t1_0, t1_1 = aes.xor_interleaved(t1_0, t1_1, st.s7_0, st.s7_1)
}
case _RATE_256:
t_0, t_1 = aes.xor_interleaved(st.s3_0, st.s3_1, t_0, t_1)
for _ in 0 ..< 7 {
update_sw_256(st, t_0, t_1)
}
t0_0, t0_1 = aes.xor_interleaved(st.s0_0, st.s0_1, st.s1_0, st.s1_1)
t0_0, t0_1 = aes.xor_interleaved(t0_0, t0_1, st.s2_0, st.s2_1)
t1_0, t1_1 = aes.xor_interleaved(st.s3_0, st.s3_1, st.s4_0, st.s4_1)
t1_0, t1_1 = aes.xor_interleaved(t1_0, t1_1, st.s5_0, st.s5_1)
}
switch len(tag) {
case TAG_SIZE_128:
t0_0, t0_1 = aes.xor_interleaved(t0_0, t0_1, t1_0, t1_1)
aes.store_interleaved(tag, t0_0, t0_1)
case TAG_SIZE_256:
aes.store_interleaved(tag[:16], t0_0, t0_1)
aes.store_interleaved(tag[16:], t1_0, t1_1)
}
}
@(private)
reset_state_sw :: proc "contextless" (st: ^State_SW) {
mem.zero_explicit(st, size_of(st^))
}

44
core/crypto/aegis/aegis_impl_hw_gen.odin Normal file
View File

@@ -0,0 +1,44 @@
#+build !amd64
package aegis
@(private = "file")
ERR_HW_NOT_SUPPORTED :: "crypto/aegis: hardware implementation unsupported"
@(private)
State_HW :: struct {}
// is_hardware_accelerated returns true iff hardware accelerated AEGIS
// is supported.
is_hardware_accelerated :: proc "contextless" () -> bool {
return false
}
@(private)
init_hw :: proc "contextless" (ctx: ^Context, st: ^State_HW, iv: []byte) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}
@(private)
absorb_hw :: proc "contextless" (st: ^State_HW, aad: []byte) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}
@(private)
enc_hw :: proc "contextless" (st: ^State_HW, dst, src: []byte) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}
@(private)
dec_hw :: proc "contextless" (st: ^State_HW, dst, src: []byte) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}
@(private)
finalize_hw :: proc "contextless" (st: ^State_HW, tag: []byte, ad_len, msg_len: int) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}
@(private)
reset_state_hw :: proc "contextless" (st: ^State_HW) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}

389
core/crypto/aegis/aegis_impl_hw_intel.odin Normal file
View File

@@ -0,0 +1,389 @@
#+build amd64
package aegis
import "base:intrinsics"
import "core:crypto/aes"
import "core:encoding/endian"
import "core:mem"
import "core:simd/x86"
@(private)
State_HW :: struct {
s0: x86.__m128i,
s1: x86.__m128i,
s2: x86.__m128i,
s3: x86.__m128i,
s4: x86.__m128i,
s5: x86.__m128i,
s6: x86.__m128i,
s7: x86.__m128i,
rate: int,
}
// is_hardware_accelerated returns true iff hardware accelerated AEGIS
// is supported.
is_hardware_accelerated :: proc "contextless" () -> bool {
return aes.is_hardware_accelerated()
}
@(private, enable_target_feature = "sse2,aes")
init_hw :: proc "contextless" (ctx: ^Context, st: ^State_HW, iv: []byte) {
switch ctx._key_len {
case KEY_SIZE_128L:
key := intrinsics.unaligned_load((^x86.__m128i)(&ctx._key[0]))
iv := intrinsics.unaligned_load((^x86.__m128i)(raw_data(iv)))
st.s0 = x86._mm_xor_si128(key, iv)
st.s1 = intrinsics.unaligned_load((^x86.__m128i)(&_C1[0]))
st.s2 = intrinsics.unaligned_load((^x86.__m128i)(&_C0[0]))
st.s3 = st.s1
st.s4 = st.s0
st.s5 = x86._mm_xor_si128(key, st.s2) // key ^ C0
st.s6 = x86._mm_xor_si128(key, st.s1) // key ^ C1
st.s7 = st.s5
st.rate = _RATE_128L
for _ in 0 ..< 10 {
update_hw_128l(st, iv, key)
}
case KEY_SIZE_256:
k0 := intrinsics.unaligned_load((^x86.__m128i)(&ctx._key[0]))
k1 := intrinsics.unaligned_load((^x86.__m128i)(&ctx._key[16]))
n0 := intrinsics.unaligned_load((^x86.__m128i)(&iv[0]))
n1 := intrinsics.unaligned_load((^x86.__m128i)(&iv[16]))
st.s0 = x86._mm_xor_si128(k0, n0)
st.s1 = x86._mm_xor_si128(k1, n1)
st.s2 = intrinsics.unaligned_load((^x86.__m128i)(&_C1[0]))
st.s3 = intrinsics.unaligned_load((^x86.__m128i)(&_C0[0]))
st.s4 = x86._mm_xor_si128(k0, st.s3) // k0 ^ C0
st.s5 = x86._mm_xor_si128(k1, st.s2) // k1 ^ C1
st.rate = _RATE_256
u0, u1 := st.s0, st.s1
for _ in 0 ..< 4 {
update_hw_256(st, k0)
update_hw_256(st, k1)
update_hw_256(st, u0)
update_hw_256(st, u1)
}
}
}
@(private = "file", enable_target_feature = "sse2,aes")
update_hw_128l :: #force_inline proc "contextless" (st: ^State_HW, m0, m1: x86.__m128i) {
s0_ := x86._mm_aesenc_si128(st.s7, x86._mm_xor_si128(st.s0, m0))
s1_ := x86._mm_aesenc_si128(st.s0, st.s1)
s2_ := x86._mm_aesenc_si128(st.s1, st.s2)
s3_ := x86._mm_aesenc_si128(st.s2, st.s3)
s4_ := x86._mm_aesenc_si128(st.s3, x86._mm_xor_si128(st.s4, m1))
s5_ := x86._mm_aesenc_si128(st.s4, st.s5)
s6_ := x86._mm_aesenc_si128(st.s5, st.s6)
s7_ := x86._mm_aesenc_si128(st.s6, st.s7)
st.s0, st.s1, st.s2, st.s3, st.s4, st.s5, st.s6, st.s7 = s0_, s1_, s2_, s3_, s4_, s5_, s6_, s7_
}
@(private = "file", enable_target_feature = "sse2,aes")
update_hw_256 :: #force_inline proc "contextless" (st: ^State_HW, m: x86.__m128i) {
s0_ := x86._mm_aesenc_si128(st.s5, x86._mm_xor_si128(st.s0, m))
s1_ := x86._mm_aesenc_si128(st.s0, st.s1)
s2_ := x86._mm_aesenc_si128(st.s1, st.s2)
s3_ := x86._mm_aesenc_si128(st.s2, st.s3)
s4_ := x86._mm_aesenc_si128(st.s3, st.s4)
s5_ := x86._mm_aesenc_si128(st.s4, st.s5)
st.s0, st.s1, st.s2, st.s3, st.s4, st.s5 = s0_, s1_, s2_, s3_, s4_, s5_
}
@(private = "file", enable_target_feature = "sse2,aes")
absorb_hw_128l :: #force_inline proc "contextless" (st: ^State_HW, ai: []byte) {
t0 := intrinsics.unaligned_load((^x86.__m128i)(&ai[0]))
t1 := intrinsics.unaligned_load((^x86.__m128i)(&ai[16]))
update_hw_128l(st, t0, t1)
}
@(private = "file", enable_target_feature = "sse2,aes")
absorb_hw_256 :: #force_inline proc "contextless" (st: ^State_HW, ai: []byte) {
m := intrinsics.unaligned_load((^x86.__m128i)(&ai[0]))
update_hw_256(st, m)
}
@(private, enable_target_feature = "sse2,aes")
absorb_hw :: proc "contextless" (st: ^State_HW, aad: []byte) #no_bounds_check {
ai, l := aad, len(aad)
switch st.rate {
case _RATE_128L:
for l >= _RATE_128L {
absorb_hw_128l(st, ai)
ai = ai[_RATE_128L:]
l -= _RATE_128L
}
case _RATE_256:
for l >= _RATE_256 {
absorb_hw_256(st, ai)
ai = ai[_RATE_256:]
l -= _RATE_256
}
}
// Pad out the remainder with `0`s till it is rate sized.
if l > 0 {
tmp: [_RATE_MAX]byte // AAD is not confidential.
copy(tmp[:], ai)
switch st.rate {
case _RATE_128L:
absorb_hw_128l(st, tmp[:])
case _RATE_256:
absorb_hw_256(st, tmp[:])
}
}
}
@(private = "file", enable_target_feature = "sse2", require_results)
z_hw_128l :: #force_inline proc "contextless" (st: ^State_HW) -> (x86.__m128i, x86.__m128i) {
z0 := x86._mm_xor_si128(
st.s6,
x86._mm_xor_si128(
st.s1,
x86._mm_and_si128(st.s2, st.s3),
),
)
z1 := x86._mm_xor_si128(
st.s2,
x86._mm_xor_si128(
st.s5,
x86._mm_and_si128(st.s6, st.s7),
),
)
return z0, z1
}
@(private = "file", enable_target_feature = "sse2", require_results)
z_hw_256 :: #force_inline proc "contextless" (st: ^State_HW) -> x86.__m128i {
return x86._mm_xor_si128(
st.s1,
x86._mm_xor_si128(
st.s4,
x86._mm_xor_si128(
st.s5,
x86._mm_and_si128(st.s2, st.s3),
),
),
)
}
@(private = "file", enable_target_feature = "sse2,aes")
enc_hw_128l :: #force_inline proc "contextless" (st: ^State_HW, ci, xi: []byte) #no_bounds_check {
z0, z1 := z_hw_128l(st)
t0 := intrinsics.unaligned_load((^x86.__m128i)(&xi[0]))
t1 := intrinsics.unaligned_load((^x86.__m128i)(&xi[16]))
update_hw_128l(st, t0, t1)
out0 := x86._mm_xor_si128(t0, z0)
out1 := x86._mm_xor_si128(t1, z1)
intrinsics.unaligned_store((^x86.__m128i)(&ci[0]), out0)
intrinsics.unaligned_store((^x86.__m128i)(&ci[16]), out1)
}
@(private = "file", enable_target_feature = "sse2,aes")
enc_hw_256 :: #force_inline proc "contextless" (st: ^State_HW, ci, xi: []byte) #no_bounds_check {
z := z_hw_256(st)
xi_ := intrinsics.unaligned_load((^x86.__m128i)(raw_data(xi)))
update_hw_256(st, xi_)
ci_ := x86._mm_xor_si128(xi_, z)
intrinsics.unaligned_store((^x86.__m128i)(raw_data(ci)), ci_)
}
@(private, enable_target_feature = "sse2,aes")
enc_hw :: proc "contextless" (st: ^State_HW, dst, src: []byte) #no_bounds_check {
ci, xi, l := dst, src, len(src)
switch st.rate {
case _RATE_128L:
for l >= _RATE_128L {
enc_hw_128l(st, ci, xi)
ci = ci[_RATE_128L:]
xi = xi[_RATE_128L:]
l -= _RATE_128L
}
case _RATE_256:
for l >= _RATE_256 {
enc_hw_256(st, ci, xi)
ci = ci[_RATE_256:]
xi = xi[_RATE_256:]
l -= _RATE_256
}
}
// Pad out the remainder with `0`s till it is rate sized.
if l > 0 {
tmp: [_RATE_MAX]byte // Ciphertext is not confidential.
copy(tmp[:], xi)
switch st.rate {
case _RATE_128L:
enc_hw_128l(st, tmp[:], tmp[:])
case _RATE_256:
enc_hw_256(st, tmp[:], tmp[:])
}
copy(ci, tmp[:l])
}
}
@(private = "file", enable_target_feature = "sse2,aes")
dec_hw_128l :: #force_inline proc "contextless" (st: ^State_HW, xi, ci: []byte) #no_bounds_check {
z0, z1 := z_hw_128l(st)
t0 := intrinsics.unaligned_load((^x86.__m128i)(&ci[0]))
t1 := intrinsics.unaligned_load((^x86.__m128i)(&ci[16]))
out0 := x86._mm_xor_si128(t0, z0)
out1 := x86._mm_xor_si128(t1, z1)
update_hw_128l(st, out0, out1)
intrinsics.unaligned_store((^x86.__m128i)(&xi[0]), out0)
intrinsics.unaligned_store((^x86.__m128i)(&xi[16]), out1)
}
@(private = "file", enable_target_feature = "sse2,aes")
dec_hw_256 :: #force_inline proc "contextless" (st: ^State_HW, xi, ci: []byte) #no_bounds_check {
z := z_hw_256(st)
ci_ := intrinsics.unaligned_load((^x86.__m128i)(raw_data(ci)))
xi_ := x86._mm_xor_si128(ci_, z)
update_hw_256(st, xi_)
intrinsics.unaligned_store((^x86.__m128i)(raw_data(xi)), xi_)
}
@(private = "file", enable_target_feature = "sse2,aes")
dec_partial_hw_128l :: #force_inline proc "contextless" (st: ^State_HW, xn, cn: []byte) #no_bounds_check {
tmp: [_RATE_128L]byte
defer mem.zero_explicit(&tmp, size_of(tmp))
z0, z1 := z_hw_128l(st)
copy(tmp[:], cn)
t0 := intrinsics.unaligned_load((^x86.__m128i)(&tmp[0]))
t1 := intrinsics.unaligned_load((^x86.__m128i)(&tmp[16]))
out0 := x86._mm_xor_si128(t0, z0)
out1 := x86._mm_xor_si128(t1, z1)
intrinsics.unaligned_store((^x86.__m128i)(&tmp[0]), out0)
intrinsics.unaligned_store((^x86.__m128i)(&tmp[16]), out1)
copy(xn, tmp[:])
for off := len(xn); off < _RATE_128L; off += 1 {
tmp[off] = 0
}
out0 = intrinsics.unaligned_load((^x86.__m128i)(&tmp[0])) // v0
out1 = intrinsics.unaligned_load((^x86.__m128i)(&tmp[16])) // v1
update_hw_128l(st, out0, out1)
}
@(private = "file", enable_target_feature = "sse2,aes")
dec_partial_hw_256 :: #force_inline proc "contextless" (st: ^State_HW, xn, cn: []byte) #no_bounds_check {
tmp: [_RATE_256]byte
defer mem.zero_explicit(&tmp, size_of(tmp))
z := z_hw_256(st)
copy(tmp[:], cn)
cn_ := intrinsics.unaligned_load((^x86.__m128i)(&tmp[0]))
xn_ := x86._mm_xor_si128(cn_, z)
intrinsics.unaligned_store((^x86.__m128i)(&tmp[0]), xn_)
copy(xn, tmp[:])
for off := len(xn); off < _RATE_256; off += 1 {
tmp[off] = 0
}
xn_ = intrinsics.unaligned_load((^x86.__m128i)(&tmp[0]))
update_hw_256(st, xn_)
}
@(private, enable_target_feature = "sse2,aes")
dec_hw :: proc "contextless" (st: ^State_HW, dst, src: []byte) #no_bounds_check {
xi, ci, l := dst, src, len(src)
switch st.rate {
case _RATE_128L:
for l >= _RATE_128L {
dec_hw_128l(st, xi, ci)
xi = xi[_RATE_128L:]
ci = ci[_RATE_128L:]
l -= _RATE_128L
}
case _RATE_256:
for l >= _RATE_256 {
dec_hw_256(st, xi, ci)
xi = xi[_RATE_256:]
ci = ci[_RATE_256:]
l -= _RATE_256
}
}
// Process the remainder.
if l > 0 {
switch st.rate {
case _RATE_128L:
dec_partial_hw_128l(st, xi, ci)
case _RATE_256:
dec_partial_hw_256(st, xi, ci)
}
}
}
@(private, enable_target_feature = "sse2,aes")
finalize_hw :: proc "contextless" (st: ^State_HW, tag: []byte, ad_len, msg_len: int) {
tmp: [16]byte
endian.unchecked_put_u64le(tmp[0:], u64(ad_len) * 8)
endian.unchecked_put_u64le(tmp[8:], u64(msg_len) * 8)
t := intrinsics.unaligned_load((^x86.__m128i)(&tmp[0]))
t0, t1: x86.__m128i = ---, ---
switch st.rate {
case _RATE_128L:
t = x86._mm_xor_si128(st.s2, t)
for _ in 0 ..< 7 {
update_hw_128l(st, t, t)
}
t0 = x86._mm_xor_si128(st.s0, st.s1)
t0 = x86._mm_xor_si128(t0, st.s2)
t0 = x86._mm_xor_si128(t0, st.s3)
t1 = x86._mm_xor_si128(st.s4, st.s5)
t1 = x86._mm_xor_si128(t1, st.s6)
if len(tag) == TAG_SIZE_256 {
t1 = x86._mm_xor_si128(t1, st.s7)
}
case _RATE_256:
t = x86._mm_xor_si128(st.s3, t)
for _ in 0 ..< 7 {
update_hw_256(st, t)
}
t0 = x86._mm_xor_si128(st.s0, st.s1)
t0 = x86._mm_xor_si128(t0, st.s2)
t1 = x86._mm_xor_si128(st.s3, st.s4)
t1 = x86._mm_xor_si128(t1, st.s5)
}
switch len(tag) {
case TAG_SIZE_128:
t0 = x86._mm_xor_si128(t0, t1)
intrinsics.unaligned_store((^x86.__m128i)(&tag[0]), t0)
case TAG_SIZE_256:
intrinsics.unaligned_store((^x86.__m128i)(&tag[0]), t0)
intrinsics.unaligned_store((^x86.__m128i)(&tag[16]), t1)
}
}
@(private)
reset_state_hw :: proc "contextless" (st: ^State_HW) {
mem.zero_explicit(st, size_of(st^))
}

12
core/crypto/aes/aes_ctr.odin
View File

@@ -21,9 +21,7 @@ Context_CTR :: struct {
// init_ctr initializes a Context_CTR with the provided key and IV.
init_ctr :: proc(ctx: ^Context_CTR, key, iv: []byte, impl := DEFAULT_IMPLEMENTATION) {
if len(iv) != CTR_IV_SIZE {
panic("crypto/aes: invalid CTR IV size")
}
ensure(len(iv) == CTR_IV_SIZE, "crypto/aes: invalid CTR IV size")
init_impl(&ctx._impl, key, impl)
ctx._off = BLOCK_SIZE
@@ -36,16 +34,14 @@ init_ctr :: proc(ctx: ^Context_CTR, key, iv: []byte, impl := DEFAULT_IMPLEMENTAT
// keystream, and writes the resulting output to dst. dst and src MUST
// alias exactly or not at all.
xor_bytes_ctr :: proc(ctx: ^Context_CTR, dst, src: []byte) {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
src, dst := src, dst
if dst_len := len(dst); dst_len < len(src) {
src = src[:dst_len]
}
if bytes.alias_inexactly(dst, src) {
panic("crypto/aes: dst and src alias inexactly")
}
ensure(!bytes.alias_inexactly(dst, src), "crypto/aes: dst and src alias inexactly")
#no_bounds_check for remaining := len(src); remaining > 0; {
// Process multiple blocks at once
@@ -82,7 +78,7 @@ xor_bytes_ctr :: proc(ctx: ^Context_CTR, dst, src: []byte) {
// keystream_bytes_ctr fills dst with the raw AES-CTR keystream output.
keystream_bytes_ctr :: proc(ctx: ^Context_CTR, dst: []byte) {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
dst := dst
#no_bounds_check for remaining := len(dst); remaining > 0; {

16
core/crypto/aes/aes_ecb.odin
View File

@@ -19,11 +19,9 @@ init_ecb :: proc(ctx: ^Context_ECB, key: []byte, impl := DEFAULT_IMPLEMENTATION)
// encrypt_ecb encrypts the BLOCK_SIZE buffer src, and writes the result to dst.
encrypt_ecb :: proc(ctx: ^Context_ECB, dst, src: []byte) {
assert(ctx._is_initialized)
if len(dst) != BLOCK_SIZE || len(src) != BLOCK_SIZE {
panic("crypto/aes: invalid buffer size(s)")
}
ensure(ctx._is_initialized)
ensure(len(dst) == BLOCK_SIZE, "crypto/aes: invalid dst size")
ensure(len(dst) == BLOCK_SIZE, "crypto/aes: invalid src size")
switch &impl in ctx._impl {
case ct64.Context:
@@ -35,11 +33,9 @@ encrypt_ecb :: proc(ctx: ^Context_ECB, dst, src: []byte) {
// decrypt_ecb decrypts the BLOCK_SIZE buffer src, and writes the result to dst.
decrypt_ecb :: proc(ctx: ^Context_ECB, dst, src: []byte) {
assert(ctx._is_initialized)
if len(dst) != BLOCK_SIZE || len(src) != BLOCK_SIZE {
panic("crypto/aes: invalid buffer size(s)")
}
ensure(ctx._is_initialized)
ensure(len(dst) == BLOCK_SIZE, "crypto/aes: invalid dst size")
ensure(len(dst) == BLOCK_SIZE, "crypto/aes: invalid src size")
switch &impl in ctx._impl {
case ct64.Context:

36
core/crypto/aes/aes_gcm.odin
View File

@@ -36,15 +36,11 @@ init_gcm :: proc(ctx: ^Context_GCM, key: []byte, impl := DEFAULT_IMPLEMENTATION)
//
// dst and plaintext MUST alias exactly or not at all.
seal_gcm :: proc(ctx: ^Context_GCM, dst, tag, iv, aad, plaintext: []byte) {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
gcm_validate_common_slice_sizes(tag, iv, aad, plaintext)
if len(dst) != len(plaintext) {
panic("crypto/aes: invalid destination ciphertext size")
}
if bytes.alias_inexactly(dst, plaintext) {
panic("crypto/aes: dst and plaintext alias inexactly")
}
ensure(len(dst) == len(plaintext), "crypto/aes: invalid destination ciphertext size")
ensure(!bytes.alias_inexactly(dst, plaintext), "crypto/aes: dst and plaintext alias inexactly")
if impl, is_hw := ctx._impl.(Context_Impl_Hardware); is_hw {
gcm_seal_hw(&impl, dst, tag, iv, aad, plaintext)
@@ -76,15 +72,11 @@ seal_gcm :: proc(ctx: ^Context_GCM, dst, tag, iv, aad, plaintext: []byte) {
// dst and plaintext MUST alias exactly or not at all.
@(require_results)
open_gcm :: proc(ctx: ^Context_GCM, dst, iv, aad, ciphertext, tag: []byte) -> bool {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
gcm_validate_common_slice_sizes(tag, iv, aad, ciphertext)
if len(dst) != len(ciphertext) {
panic("crypto/aes: invalid destination plaintext size")
}
if bytes.alias_inexactly(dst, ciphertext) {
panic("crypto/aes: dst and ciphertext alias inexactly")
}
ensure(len(dst) == len(ciphertext), "crypto/aes: invalid destination plaintext size")
ensure(!bytes.alias_inexactly(dst, ciphertext), "crypto/aes: dst and ciphertext alias inexactly")
if impl, is_hw := ctx._impl.(Context_Impl_Hardware); is_hw {
return gcm_open_hw(&impl, dst, iv, aad, ciphertext, tag)
@@ -122,21 +114,13 @@ reset_gcm :: proc "contextless" (ctx: ^Context_GCM) {
@(private = "file")
gcm_validate_common_slice_sizes :: proc(tag, iv, aad, text: []byte) {
if len(tag) != GCM_TAG_SIZE {
panic("crypto/aes: invalid GCM tag size")
}
ensure(len(tag) == GCM_TAG_SIZE, "crypto/aes: invalid GCM tag size")
// The specification supports IVs in the range [1, 2^64) bits.
if l := len(iv); l == 0 || u64(l) >= GCM_IV_SIZE_MAX {
panic("crypto/aes: invalid GCM IV size")
}
ensure(len(iv) == 0 || u64(len(iv)) <= GCM_IV_SIZE_MAX, "crypto/aes: invalid GCM IV size")
if aad_len := u64(len(aad)); aad_len > GCM_A_MAX {
panic("crypto/aes: oversized GCM aad")
}
if text_len := u64(len(text)); text_len > GCM_P_MAX {
panic("crypto/aes: oversized GCM src data")
}
ensure(u64(len(aad)) <= GCM_A_MAX, "crypto/aes: oversized GCM aad")
ensure(u64(len(text)) <= GCM_P_MAX, "crypto/aes: oversized GCM data")
}
@(private = "file")

2
core/crypto/aes/aes_gcm_hw_intel.odin
View File

@@ -235,7 +235,7 @@ gctr_hw :: proc(
// BUG: Sticking this in gctr_hw (like the other implementations) crashes
// the compiler.
//
// src/check_expr.cpp(7892): Assertion Failure: `c->curr_proc_decl->entity`
// src/check_expr.cpp(8104): Assertion Failure: `c->curr_proc_decl->entity`
@(private = "file", enable_target_feature = "sse4.1")
hw_inc_ctr32 :: #force_inline proc "contextless" (src: ^x86.__m128i, ctr: u32) -> (x86.__m128i, u32) {
ret := x86._mm_insert_epi32(src^, i32(intrinsics.byte_swap(ctr)), 3)

8
core/crypto/blake2b/blake2b.odin
View File

@@ -18,7 +18,7 @@ package blake2b
import "../_blake2"
// DIGEST_SIZE is the BLAKE2b digest size in bytes.
DIGEST_SIZE :: 64
DIGEST_SIZE :: _blake2.BLAKE2B_SIZE
// BLOCK_SIZE is the BLAKE2b block size in bytes.
BLOCK_SIZE :: _blake2.BLAKE2B_BLOCK_SIZE
@@ -27,9 +27,11 @@ BLOCK_SIZE :: _blake2.BLAKE2B_BLOCK_SIZE
Context :: _blake2.Blake2b_Context
// init initializes a Context with the default BLAKE2b config.
init :: proc(ctx: ^Context) {
init :: proc(ctx: ^Context, digest_size := DIGEST_SIZE) {
ensure(digest_size <= _blake2.MAX_SIZE, "crypto/blake2b: invalid digest size")
cfg: _blake2.Blake2_Config
cfg.size = _blake2.BLAKE2B_SIZE
cfg.size = u8(digest_size)
_blake2.init(ctx, &cfg)
}

8
core/crypto/blake2s/blake2s.odin
View File

@@ -18,7 +18,7 @@ package blake2s
import "../_blake2"
// DIGEST_SIZE is the BLAKE2s digest size in bytes.
DIGEST_SIZE :: 32
DIGEST_SIZE :: _blake2.BLAKE2S_SIZE
// BLOCK_SIZE is the BLAKE2s block size in bytes.
BLOCK_SIZE :: _blake2.BLAKE2S_BLOCK_SIZE
@@ -27,9 +27,11 @@ BLOCK_SIZE :: _blake2.BLAKE2S_BLOCK_SIZE
Context :: _blake2.Blake2s_Context
// init initializes a Context with the default BLAKE2s config.
init :: proc(ctx: ^Context) {
init :: proc(ctx: ^Context, digest_size := DIGEST_SIZE) {
ensure(digest_size <= _blake2.MAX_SIZE, "crypto/blake2s: invalid digest size")
cfg: _blake2.Blake2_Config
cfg.size = _blake2.BLAKE2S_SIZE
cfg.size = u8(digest_size)
_blake2.init(ctx, &cfg)
}

16
core/crypto/chacha20/chacha20.odin
View File

@@ -27,12 +27,8 @@ Context :: struct {
// init inititializes a Context for ChaCha20 or XChaCha20 with the provided
// key and iv.
init :: proc(ctx: ^Context, key, iv: []byte, impl := DEFAULT_IMPLEMENTATION) {
if len(key) != KEY_SIZE {
panic("crypto/chacha20: invalid (X)ChaCha20 key size")
}
if l := len(iv); l != IV_SIZE && l != XIV_SIZE {
panic("crypto/chacha20: invalid (X)ChaCha20 IV size")
}
ensure(len(key) == KEY_SIZE, "crypto/chacha20: invalid (X)ChaCha20 key size")
ensure(len(iv) == IV_SIZE || len(iv) == XIV_SIZE, "crypto/chacha20: invalid (X)ChaCha20 IV size")
k, n := key, iv
@@ -67,16 +63,14 @@ seek :: proc(ctx: ^Context, block_nr: u64) {
// keystream, and writes the resulting output to dst. Dst and src MUST
// alias exactly or not at all.
xor_bytes :: proc(ctx: ^Context, dst, src: []byte) {
assert(ctx._state._is_initialized)
ensure(ctx._state._is_initialized)
src, dst := src, dst
if dst_len := len(dst); dst_len < len(src) {
src = src[:dst_len]
}
if bytes.alias_inexactly(dst, src) {
panic("crypto/chacha20: dst and src alias inexactly")
}
ensure(!bytes.alias_inexactly(dst, src), "crypto/chacha20: dst and src alias inexactly")
st := &ctx._state
#no_bounds_check for remaining := len(src); remaining > 0; {
@@ -114,7 +108,7 @@ xor_bytes :: proc(ctx: ^Context, dst, src: []byte) {
// keystream_bytes fills dst with the raw (X)ChaCha20 keystream output.
keystream_bytes :: proc(ctx: ^Context, dst: []byte) {
assert(ctx._state._is_initialized)
ensure(ctx._state._is_initialized)
dst, st := dst, &ctx._state
#no_bounds_check for remaining := len(dst); remaining > 0; {

30
core/crypto/chacha20poly1305/chacha20poly1305.odin
View File

@@ -29,13 +29,9 @@ _P_MAX :: 64 * 0xffffffff // 64 * (2^32-1)
@(private)
_validate_common_slice_sizes :: proc (tag, iv, aad, text: []byte, is_xchacha: bool) {
if len(tag) != TAG_SIZE {
panic("crypto/chacha20poly1305: invalid destination tag size")
}
expected_iv_len := is_xchacha ? XIV_SIZE : IV_SIZE
if len(iv) != expected_iv_len {
panic("crypto/chacha20poly1305: invalid IV size")
}
ensure(len(tag) == TAG_SIZE, "crypto/chacha20poly1305: invalid destination tag size")
ensure(len(iv) == expected_iv_len, "crypto/chacha20poly1305: invalid IV size")
#assert(size_of(int) == 8 || size_of(int) <= 4)
when size_of(int) == 8 {
@@ -45,13 +41,11 @@ _validate_common_slice_sizes :: proc (tag, iv, aad, text: []byte, is_xchacha: bo
// A_MAX is limited by size_of(int), so there is no need to
// enforce it. P_MAX only needs to be checked on 64-bit targets,
// for reasons that should be obvious.
if text_len := len(text); text_len > _P_MAX {
panic("crypto/chacha20poly1305: oversized src data")
}
ensure(len(text) <= _P_MAX, "crypto/chacha20poly1305: oversized src data")
}
}
@(private)
@(private, rodata)
_PAD: [16]byte
@(private)
@@ -71,9 +65,7 @@ Context :: struct {
// init initializes a Context with the provided key, for AEAD_CHACHA20_POLY1305.
init :: proc(ctx: ^Context, key: []byte, impl := chacha20.DEFAULT_IMPLEMENTATION) {
if len(key) != KEY_SIZE {
panic("crypto/chacha20poly1305: invalid key size")
}
ensure(len(key) == KEY_SIZE, "crypto/chacha20poly1305: invalid key size")
copy(ctx._key[:], key)
ctx._impl = impl
@@ -96,11 +88,11 @@ init_xchacha :: proc(ctx: ^Context, key: []byte, impl := chacha20.DEFAULT_IMPLEM
//
// dst and plaintext MUST alias exactly or not at all.
seal :: proc(ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) {
ensure(ctx._is_initialized)
ciphertext := dst
_validate_common_slice_sizes(tag, iv, aad, plaintext, ctx._is_xchacha)
if len(ciphertext) != len(plaintext) {
panic("crypto/chacha20poly1305: invalid destination ciphertext size")
}
ensure(len(ciphertext) == len(plaintext), "crypto/chacha20poly1305: invalid destination ciphertext size")
stream_ctx: chacha20.Context = ---
chacha20.init(&stream_ctx, ctx._key[:],iv, ctx._impl)
@@ -151,11 +143,11 @@ seal :: proc(ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) {
// dst and plaintext MUST alias exactly or not at all.
@(require_results)
open :: proc(ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
ensure(ctx._is_initialized)
plaintext := dst
_validate_common_slice_sizes(tag, iv, aad, ciphertext, ctx._is_xchacha)
if len(ciphertext) != len(plaintext) {
panic("crypto/chacha20poly1305: invalid destination plaintext size")
}
ensure(len(ciphertext) == len(plaintext), "crypto/chacha20poly1305: invalid destination plaintext size")
// Note: Unlike encrypt, this can fail early, so use defer for
// sanitization rather than assuming control flow reaches certain

280
core/crypto/deoxysii/deoxysii.odin Normal file
View File

@@ -0,0 +1,280 @@
/*
package deoxysii implements the Deoxys-II-256 Authenticated Encryption
with Additional Data algorithm.
- [[ https://sites.google.com/view/deoxyscipher ]]
- [[ https://thomaspeyrin.github.io/web/assets/docs/papers/Jean-etal-JoC2021.pdf ]]
*/
package deoxysii
import "base:intrinsics"
import "core:bytes"
import "core:crypto/aes"
import "core:mem"
import "core:simd"
// KEY_SIZE is the Deoxys-II-256 key size in bytes.
KEY_SIZE :: 32
// IV_SIZE iss the Deoxys-II-256 IV size in bytes.
IV_SIZE :: 15 // 120-bits
// TAG_SIZE is the Deoxys-II-256 tag size in bytes.
TAG_SIZE :: 16
@(private)
PREFIX_AD_BLOCK :: 0b0010
@(private)
PREFIX_AD_FINAL :: 0b0110
@(private)
PREFIX_MSG_BLOCK :: 0b0000
@(private)
PREFIX_MSG_FINAL :: 0b0100
@(private)
PREFIX_TAG :: 0b0001
@(private)
PREFIX_SHIFT :: 4
@(private)
BC_ROUNDS :: 16
@(private)
BLOCK_SIZE :: aes.BLOCK_SIZE
@(private = "file")
_LFSR2_MASK :: simd.u8x16{
0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01, 0x01,
}
@(private = "file")
_LFSR3_MASK :: simd.u8x16{
0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80,
}
@(private = "file")
_LFSR_SH1 :: _LFSR2_MASK
@(private = "file")
_LFSR_SH5 :: simd.u8x16{
0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05, 0x05,
}
@(private = "file")
_LFSR_SH7 :: simd.u8x16{
0x07, 0x07, 0x07, 0x07, 0x07, 0x07, 0x07, 0x07,
0x07, 0x07, 0x07, 0x07, 0x07, 0x07, 0x07, 0x07,
}
@(private = "file", rodata)
_RCONS := []byte {
0x2f, 0x5e, 0xbc, 0x63, 0xc6, 0x97, 0x35, 0x6a,
0xd4, 0xb3, 0x7d, 0xfa, 0xef, 0xc5, 0x91, 0x39,
0x72,
}
// Context is a keyed Deoxys-II-256 instance.
Context :: struct {
_subkeys: [BC_ROUNDS+1][16]byte,
_impl: aes.Implementation,
_is_initialized: bool,
}
@(private)
_validate_common_slice_sizes :: proc (ctx: ^Context, tag, iv, aad, text: []byte) {
ensure(len(tag) == TAG_SIZE, "crypto/deoxysii: invalid tag size")
ensure(len(iv) == IV_SIZE, "crypto/deoxysii: invalid IV size")
#assert(size_of(int) == 8 || size_of(int) <= 4)
// For the nonce-misuse resistant mode, the total size of the
// associated data and the total size of the message do not exceed
// `16 * 2^max_l * 2^max_m bytes`, thus 2^128 bytes for all variants
// of Deoxys-II. Moreover, the maximum number of messages that can
// be handled for a same key is 2^max_m, that is 2^64 for all variants
// of Deoxys.
}
// init initializes a Context with the provided key.
init :: proc(ctx: ^Context, key: []byte, impl := aes.DEFAULT_IMPLEMENTATION) {
ensure(len(key) == KEY_SIZE, "crypto/deoxysii: invalid key size")
ctx._impl = impl
if ctx._impl == .Hardware && !is_hardware_accelerated() {
ctx._impl = .Portable
}
derive_ks(ctx, key)
ctx._is_initialized = true
}
// seal encrypts the plaintext and authenticates the aad and ciphertext,
// with the provided Context and iv, stores the output in dst and tag.
//
// dst and plaintext MUST alias exactly or not at all.
seal :: proc(ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) {
ensure(ctx._is_initialized)
_validate_common_slice_sizes(ctx, tag, iv, aad, plaintext)
ensure(len(dst) == len(plaintext), "crypto/deoxysii: invalid destination ciphertext size")
ensure(!bytes.alias_inexactly(dst, plaintext), "crypto/deoxysii: dst and plaintext alias inexactly")
switch ctx._impl {
case .Hardware:
e_hw(ctx, dst, tag, iv, aad, plaintext)
case .Portable:
e_ref(ctx, dst, tag, iv, aad, plaintext)
}
}
// open authenticates the aad and ciphertext, and decrypts the ciphertext,
// with the provided Context, iv, and tag, and stores the output in dst,
// returning true iff the authentication was successful. If authentication
// fails, the destination buffer will be zeroed.
//
// dst and plaintext MUST alias exactly or not at all.
@(require_results)
open :: proc(ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
ensure(ctx._is_initialized)
_validate_common_slice_sizes(ctx, tag, iv, aad, ciphertext)
ensure(len(dst) == len(ciphertext), "crypto/deoxysii: invalid destination plaintext size")
ensure(!bytes.alias_inexactly(dst, ciphertext), "crypto/deoxysii: dst and ciphertext alias inexactly")
ok: bool
switch ctx._impl {
case .Hardware:
ok = d_hw(ctx, dst, iv, aad, ciphertext, tag)
case .Portable:
ok = d_ref(ctx, dst, iv, aad, ciphertext, tag)
}
if !ok {
mem.zero_explicit(raw_data(dst), len(ciphertext))
}
return ok
}
// reset sanitizes the Context. The Context must be
// re-initialized to be used again.
reset :: proc "contextless" (ctx: ^Context) {
mem.zero_explicit(&ctx._subkeys, len(ctx._subkeys))
ctx._is_initialized = false
}
@(private = "file")
derive_ks :: proc "contextless" (ctx: ^Context, key: []byte) {
// Derive the constant component of each subtweakkey.
//
// The key schedule is as thus:
//
// STK_i = TK1_i ^ TK2_i ^ TK3_i ^ RC_i
//
// TK1_i = h(TK1_(i-1))
// TK2_i = h(LFSR2(TK2_(i-1)))
// TK3_i = h(LFSR3(TK2_(i-1)))
//
// where:
//
// KT = K || T
// W3 = KT[:16]
// W2 = KT[16:32]
// W1 = KT[32:]
//
// TK1_0 = W1
// TK2_0 = W2
// TK3_0 = W3
//
// As `K` is fixed per Context, the XORs of `TK3_0 .. TK3_n`,
// `TK2_0 .. TK2_n` and RC_i can be precomputed in advance like
// thus:
//
// subkey_i = TK3_i ^ TK2_i ^ RC_i
//
// When it is time to actually call Deoxys-BC-384, it is then
// a simple matter of deriving each round subtweakkey via:
//
// TK1_0 = T (Tweak)
// STK_0 = subkey_0 ^ TK1_0
// STK_i = subkey_i (precomputed) ^ H(TK1_(i-1))
//
// We opt to use SIMD here and for the subtweakkey deriviation
// as `H()` is typically a single vector instruction.
tk2 := intrinsics.unaligned_load((^simd.u8x16)(raw_data(key[16:])))
tk3 := intrinsics.unaligned_load((^simd.u8x16)(raw_data(key)))
// subkey_0 does not apply LFSR2/3 or H.
intrinsics.unaligned_store(
(^simd.u8x16)(&ctx._subkeys[0]),
simd.bit_xor(
tk2,
simd.bit_xor(
tk3,
rcon(0),
),
),
)
// Precompute k_1 .. k_16.
for i in 1 ..< BC_ROUNDS+1 {
tk2 = h(lfsr2(tk2))
tk3 = h(lfsr3(tk3))
intrinsics.unaligned_store(
(^simd.u8x16)(&ctx._subkeys[i]),
simd.bit_xor(
tk2,
simd.bit_xor(
tk3,
rcon(i),
),
),
)
}
}
@(private = "file")
lfsr2 :: #force_inline proc "contextless" (tk: simd.u8x16) -> simd.u8x16 {
// LFSR2 is a application of the following LFSR to each byte of input.
// (x7||x6||x5||x4||x3||x2||x1||x0) -> (x6||x5||x4||x3||x2||x1||x0||x7 ^ x5)
return simd.bit_or(
simd.shl(tk, _LFSR_SH1),
simd.bit_and(
simd.bit_xor(
simd.shr(tk, _LFSR_SH7), // x7
simd.shr(tk, _LFSR_SH5), // x5
),
_LFSR2_MASK,
),
)
}
@(private = "file")
lfsr3 :: #force_inline proc "contextless" (tk: simd.u8x16) -> simd.u8x16 {
// LFSR3 is a application of the following LFSR to each byte of input.
// (x7||x6||x5||x4||x3||x2||x1||x0) -> (x0 ^ x6||x7||x6||x5||x4||x3||x2||x1)
return simd.bit_or(
simd.shr(tk, _LFSR_SH1),
simd.bit_and(
simd.bit_xor(
simd.shl(tk, _LFSR_SH7), // x0
simd.shl(tk, _LFSR_SH1), // x6
),
_LFSR3_MASK,
),
)
}
@(private)
h :: #force_inline proc "contextless" (tk: simd.u8x16) -> simd.u8x16 {
return simd.swizzle(
tk,
0x01, 0x06, 0x0b, 0x0c, 0x05, 0x0a, 0x0f, 0x00,
0x09, 0x0e, 0x03, 0x04, 0x0d, 0x02, 0x07, 0x08,
)
}
@(private = "file")
rcon :: #force_inline proc "contextless" (rd: int) -> simd.u8x16 #no_bounds_check {
rc := _RCONS[rd]
return simd.u8x16{
1, 2, 4, 8,
rc, rc, rc, rc,
0, 0, 0, 0,
0, 0, 0, 0,
}
}

399
core/crypto/deoxysii/deoxysii_impl_ct64.odin Normal file
View File

@@ -0,0 +1,399 @@
package deoxysii
import "base:intrinsics"
import "core:crypto"
import aes "core:crypto/_aes/ct64"
import "core:encoding/endian"
import "core:mem"
import "core:simd"
// This uses the bitlsiced 64-bit general purpose register SWAR AES
// round function. The encryption pass skips orthogonalizing the
// AES round function input as it is aways going to be the leading 0
// padded IV, and doing a 64-byte copy is faster.
@(private = "file")
TWEAK_SIZE :: 16
@(private = "file")
State_SW :: struct {
ctx: ^Context,
q_stk, q_b: [8]u64,
}
@(private = "file")
auth_tweak :: #force_inline proc "contextless" (
dst: ^[TWEAK_SIZE]byte,
prefix: byte,
block_nr: int,
) {
endian.unchecked_put_u64be(dst[8:], u64(block_nr))
endian.unchecked_put_u64le(dst[0:], u64(prefix) << PREFIX_SHIFT) // dst[0] = prefix << PREFIX_SHIFT
}
@(private = "file")
enc_tweak :: #force_inline proc "contextless" (
dst: ^[TWEAK_SIZE]byte,
tag: ^[TAG_SIZE]byte,
block_nr: int,
) {
tmp: [8]byte
endian.unchecked_put_u64be(tmp[:], u64(block_nr))
copy(dst[:], tag[:])
dst[0] |= 0x80
for i in 0 ..< 8 {
dst[i+8] ~= tmp[i]
}
}
@(private = "file")
enc_plaintext :: #force_inline proc "contextless" (
dst: ^[8]u64,
iv: []byte,
) {
tmp: [BLOCK_SIZE]byte = ---
tmp[0] = 0
copy(tmp[1:], iv[:])
q_0, q_1 := aes.load_interleaved(tmp[:])
for i in 0 ..< 4 {
dst[i], dst[i+4] = q_0, q_1
}
aes.orthogonalize(dst)
}
@(private = "file")
bc_x4 :: proc "contextless" (
ctx: ^Context,
dst: []byte,
tweaks: ^[4][TWEAK_SIZE]byte,
q_stk: ^[8]u64,
q_b: ^[8]u64, // Orthogonalized
n: int,
) {
tk1s: [4]simd.u8x16
for j in 0 ..< n {
tk1s[j] = intrinsics.unaligned_load((^simd.u8x16)(&tweaks[j]))
}
// Deoxys-BC-384
for i in 0 ..= BC_ROUNDS {
// Derive the round's subtweakkey
sk := intrinsics.unaligned_load((^simd.u8x16)(&ctx._subkeys[i]))
for j in 0 ..< n {
if i != 0 {
tk1s[j] = h(tk1s[j])
}
intrinsics.unaligned_store(
(^simd.u8x16)(raw_data(dst)),
simd.bit_xor(sk, tk1s[j]),
)
q_stk[j], q_stk[j+4] = aes.load_interleaved(dst[:])
}
aes.orthogonalize(q_stk)
if i != 0 {
aes.sub_bytes(q_b)
aes.shift_rows(q_b)
aes.mix_columns(q_b)
}
aes.add_round_key(q_b, q_stk[:])
}
aes.orthogonalize(q_b)
for i in 0 ..< n {
aes.store_interleaved(dst[i*BLOCK_SIZE:], q_b[i], q_b[i+4])
}
}
@(private = "file", require_results)
bc_absorb :: proc "contextless" (
st: ^State_SW,
dst: []byte,
src: []byte,
tweak_prefix: byte,
stk_block_nr: int,
) -> int {
tweaks: [4][TWEAK_SIZE]byte = ---
tmp: [BLOCK_SIZE*4]byte = ---
src, stk_block_nr := src, stk_block_nr
dst_ := intrinsics.unaligned_load((^simd.u8x16)(raw_data(dst)))
nr_blocks := len(src) / BLOCK_SIZE
for nr_blocks > 0 {
// Derive the tweak(s), orthogonalize the plaintext
n := min(nr_blocks, 4)
for i in 0 ..< n {
auth_tweak(&tweaks[i], tweak_prefix, stk_block_nr + i)
st.q_b[i], st.q_b[i + 4] = aes.load_interleaved(src)
src = src[BLOCK_SIZE:]
}
aes.orthogonalize(&st.q_b)
// Deoxys-BC-384
bc_x4(st.ctx, tmp[:], &tweaks, &st.q_stk, &st.q_b, n)
// XOR in the existing Auth/tag
for i in 0 ..< n {
dst_ = simd.bit_xor(
dst_,
intrinsics.unaligned_load((^simd.u8x16)(raw_data(tmp[i*BLOCK_SIZE:]))),
)
}
stk_block_nr += n
nr_blocks -= n
}
intrinsics.unaligned_store((^simd.u8x16)(raw_data(dst)), dst_)
mem.zero_explicit(&tweaks, size_of(tweaks))
mem.zero_explicit(&tmp, size_of(tmp))
return stk_block_nr
}
@(private = "file")
bc_final :: proc "contextless" (
st: ^State_SW,
dst: []byte,
iv: []byte,
) {
tweaks: [4][TWEAK_SIZE]byte = ---
tweaks[0][0] = PREFIX_TAG << PREFIX_SHIFT
copy(tweaks[0][1:], iv)
st.q_b[0], st.q_b[4] = aes.load_interleaved(dst)
aes.orthogonalize(&st.q_b)
bc_x4(st.ctx, dst, &tweaks, &st.q_stk, &st.q_b, 1)
}
@(private = "file", require_results)
bc_encrypt :: proc "contextless" (
st: ^State_SW,
dst: []byte,
src: []byte,
q_n: ^[8]u64, // Orthogonalized
tweak_tag: ^[TAG_SIZE]byte,
stk_block_nr: int,
) -> int {
tweaks: [4][TWEAK_SIZE]byte = ---
tmp: [BLOCK_SIZE*4]byte = ---
dst, src, stk_block_nr := dst, src, stk_block_nr
nr_blocks := len(src) / BLOCK_SIZE
for nr_blocks > 0 {
// Derive the tweak(s)
n := min(nr_blocks, 4)
for i in 0 ..< n {
enc_tweak(&tweaks[i], tweak_tag, stk_block_nr + i)
}
st.q_b = q_n^ // The plaintext is always `0^8 || N`
// Deoxys-BC-384
bc_x4(st.ctx, tmp[:], &tweaks, &st.q_stk, &st.q_b, n)
// XOR the ciphertext
for i in 0 ..< n {
intrinsics.unaligned_store(
(^simd.u8x16)(raw_data(dst[i*BLOCK_SIZE:])),
simd.bit_xor(
intrinsics.unaligned_load((^simd.u8x16)(raw_data(src[i*BLOCK_SIZE:]))),
intrinsics.unaligned_load((^simd.u8x16)(raw_data(tmp[i*BLOCK_SIZE:]))),
),
)
}
dst, src = dst[n*BLOCK_SIZE:], src[n*BLOCK_SIZE:]
stk_block_nr += n
nr_blocks -= n
}
mem.zero_explicit(&tweaks, size_of(tweaks))
mem.zero_explicit(&tmp, size_of(tmp))
return stk_block_nr
}
@(private)
e_ref :: proc "contextless" (ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) #no_bounds_check {
st: State_SW = ---
st.ctx = ctx
// Algorithm 3
//
// Associated data
// A_1 || ... || A_la || A_∗ <- A where each |A_i| = n and |A_∗| < n
// Auth <- 0^n
// for i = 0 to la − 1 do
// Auth <- Auth ^ EK(0010 || i, A_i+1)
// end
// if A_∗ != nil then
// Auth <- Auth ^ EK(0110 || la, pad10∗(A_∗))
// end
auth: [TAG_SIZE]byte
aad := aad
n := bc_absorb(&st, auth[:], aad, PREFIX_AD_BLOCK, 0)
aad = aad[n*BLOCK_SIZE:]
if l := len(aad); l > 0 {
a_star: [BLOCK_SIZE]byte
copy(a_star[:], aad)
a_star[l] = 0x80
_ = bc_absorb(&st, auth[:], a_star[:], PREFIX_AD_FINAL, n)
}
// Message authentication and tag generation
// M_1 || ... || M_l || M_∗ <- M where each |M_j| = n and |M_∗| < n
// tag <- Auth
// for j = 0 to l − 1 do
// tag <- tag ^ EK(0000 || j, M_j+1)
// end
// if M_∗ != nil then
// tag <- tag ^ EK(0100 || l, pad10∗(M_∗))
// end
// tag <- EK(0001 || 0^4 || N, tag)
m := plaintext
n = bc_absorb(&st, auth[:], m, PREFIX_MSG_BLOCK, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
m_star[l] = 0x80
_ = bc_absorb(&st, auth[:], m_star[:], PREFIX_MSG_FINAL, n)
}
bc_final(&st, auth[:], iv)
// Message encryption
// for j = 0 to l − 1 do
// C_j <- M_j ^ EK(1 || tag ^ j, 0^8 || N)
// end
// if M_∗ != nil then
// C_∗ <- M_* ^ EK(1 || tag ^ l, 0^8 || N)
// end
//
// return (C_1 || ... || C_l || C_∗, tag)
q_iv: [8]u64 = ---
enc_plaintext(&q_iv, iv)
m = plaintext
n = bc_encrypt(&st, dst, m, &q_iv, &auth, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
_ = bc_encrypt(&st, m_star[:], m_star[:], &q_iv, &auth, n)
copy(dst[n*BLOCK_SIZE:], m_star[:])
mem.zero_explicit(&m_star, size_of(m_star))
}
copy(tag, auth[:])
mem.zero_explicit(&st.q_stk, size_of(st.q_stk))
mem.zero_explicit(&st.q_b, size_of(st.q_b))
}
@(private, require_results)
d_ref :: proc "contextless" (ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
st: State_SW = ---
st.ctx = ctx
// Algorithm 4
//
// Message decryption
// C_1 || ... || C_l || C_∗ <- C where each |C_j| = n and |C_∗| < n
// for j = 0 to l − 1 do
// M_j <- C_j ^ EK(1 || tag ^ j, 0^8 || N)
// end
// if C_∗ != nil then
// M_∗ <- C_∗ ^ EK(1 || tag ^ l, 0^8 || N)
// end
q_iv: [8]u64 = ---
enc_plaintext(&q_iv, iv)
auth: [TAG_SIZE]byte
copy(auth[:], tag)
m := ciphertext
n := bc_encrypt(&st, dst, m, &q_iv, &auth, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
_ = bc_encrypt(&st, m_star[:], m_star[:], &q_iv, &auth, n)
copy(dst[n*BLOCK_SIZE:], m_star[:])
mem.zero_explicit(&m_star, size_of(m_star))
}
// Associated data
// A_1 || ... || Al_a || A_∗ <- A where each |Ai_| = n and |A_∗| < n
// Auth <- 0
// for i = 0 to la − 1 do
// Auth <- Auth ^ EK(0010 || i, A_i+1)
// end
// if A∗ != nil then
// Auth <- Auth ^ EK(0110| | l_a, pad10∗(A_∗))
// end
auth = 0
aad := aad
n = bc_absorb(&st, auth[:], aad, PREFIX_AD_BLOCK, 0)
aad = aad[n*BLOCK_SIZE:]
if l := len(aad); l > 0 {
a_star: [BLOCK_SIZE]byte
copy(a_star[:], aad)
a_star[l] = 0x80
_ = bc_absorb(&st, auth[:], a_star[:], PREFIX_AD_FINAL, n)
}
// Message authentication and tag generation
// M_1 || ... || M_l || M_∗ <- M where each |M_j| = n and |M_∗| < n
// tag0 <- Auth
// for j = 0 to l − 1 do
// tag0 <- tag0 ^ EK(0000 || j, M_j+1)
// end
// if M_∗ != nil then
// tag0 <- tag0 ^ EK(0100 || l, pad10∗(M_∗))
// end
// tag0 <- EK(0001 || 0^4 || N, tag0)
m = dst[:len(ciphertext)]
n = bc_absorb(&st, auth[:], m, PREFIX_MSG_BLOCK, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
m_star[l] = 0x80
_ = bc_absorb(&st, auth[:], m_star[:], PREFIX_MSG_FINAL, n)
mem.zero_explicit(&m_star, size_of(m_star))
}
bc_final(&st, auth[:], iv)
// Tag verification
// if tag0 = tag then return (M_1 || ... || M_l || M_∗)
// else return false
ok := crypto.compare_constant_time(auth[:], tag) == 1
mem.zero_explicit(&auth, size_of(auth))
mem.zero_explicit(&st.q_stk, size_of(st.q_stk))
mem.zero_explicit(&st.q_b, size_of(st.q_b))
return ok
}

21
core/crypto/deoxysii/deoxysii_impl_hw_gen.odin Normal file
View File

@@ -0,0 +1,21 @@
#+build !amd64
package deoxysii
@(private = "file")
ERR_HW_NOT_SUPPORTED :: "crypto/deoxysii: hardware implementation unsupported"
// is_hardware_accelerated returns true iff hardware accelerated Deoxys-II
// is supported.
is_hardware_accelerated :: proc "contextless" () -> bool {
return false
}
@(private)
e_hw :: proc "contextless" (ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) #no_bounds_check {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}
@(private, require_results)
d_hw :: proc "contextless" (ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}

434
core/crypto/deoxysii/deoxysii_impl_hw_intel.odin Normal file
View File

@@ -0,0 +1,434 @@
#+build amd64
package deoxysii
import "base:intrinsics"
import "core:crypto"
import "core:crypto/aes"
import "core:mem"
import "core:simd"
import "core:simd/x86"
// This processes a maximum of 4 blocks at a time, as that is suitable
// for most current hardware that doesn't say "Xeon".
@(private = "file")
_BIT_ENC :: x86.__m128i{0x80, 0}
@(private = "file")
_PREFIX_AD_BLOCK :: x86.__m128i{PREFIX_AD_BLOCK << PREFIX_SHIFT, 0}
@(private = "file")
_PREFIX_AD_FINAL :: x86.__m128i{PREFIX_AD_FINAL << PREFIX_SHIFT, 0}
@(private = "file")
_PREFIX_MSG_BLOCK :: x86.__m128i{PREFIX_MSG_BLOCK << PREFIX_SHIFT, 0}
@(private = "file")
_PREFIX_MSG_FINAL :: x86.__m128i{PREFIX_MSG_FINAL << PREFIX_SHIFT, 0}
// is_hardware_accelerated returns true iff hardware accelerated Deoxys-II
// is supported.
is_hardware_accelerated :: proc "contextless" () -> bool {
return aes.is_hardware_accelerated()
}
@(private = "file", enable_target_feature = "sse4.1", require_results)
auth_tweak :: #force_inline proc "contextless" (
prefix: x86.__m128i,
block_nr: int,
) -> x86.__m128i {
return x86._mm_insert_epi64(prefix, i64(intrinsics.byte_swap(u64(block_nr))), 1)
}
@(private = "file", enable_target_feature = "sse2", require_results)
enc_tweak :: #force_inline proc "contextless" (
tag: x86.__m128i,
block_nr: int,
) -> x86.__m128i {
return x86._mm_xor_si128(
x86._mm_or_si128(tag, _BIT_ENC),
x86.__m128i{0, i64(intrinsics.byte_swap(u64(block_nr)))},
)
}
@(private = "file", enable_target_feature = "ssse3", require_results)
h_ :: #force_inline proc "contextless" (tk1: x86.__m128i) -> x86.__m128i {
return transmute(x86.__m128i)h(transmute(simd.u8x16)tk1)
}
@(private = "file", enable_target_feature = "sse2,ssse3,aes", require_results)
bc_x4 :: #force_inline proc "contextless" (
ctx: ^Context,
s_0, s_1, s_2, s_3: x86.__m128i,
tweak_0, tweak_1, tweak_2, tweak_3: x86.__m128i,
) -> (x86.__m128i, x86.__m128i, x86.__m128i, x86.__m128i) #no_bounds_check {
s_0, s_1, s_2, s_3 := s_0, s_1, s_2, s_3
tk1_0, tk1_1, tk1_2, tk1_3 := tweak_0, tweak_1, tweak_2, tweak_3
sk := intrinsics.unaligned_load((^x86.__m128i)(&ctx._subkeys[0]))
stk_0 := x86._mm_xor_si128(tk1_0, sk)
stk_1 := x86._mm_xor_si128(tk1_1, sk)
stk_2 := x86._mm_xor_si128(tk1_2, sk)
stk_3 := x86._mm_xor_si128(tk1_3, sk)
s_0 = x86._mm_xor_si128(s_0, stk_0)
s_1 = x86._mm_xor_si128(s_1, stk_1)
s_2 = x86._mm_xor_si128(s_2, stk_2)
s_3 = x86._mm_xor_si128(s_3, stk_3)
for i in 1 ..= BC_ROUNDS {
sk = intrinsics.unaligned_load((^x86.__m128i)(&ctx._subkeys[i]))
tk1_0 = h_(tk1_0)
tk1_1 = h_(tk1_1)
tk1_2 = h_(tk1_2)
tk1_3 = h_(tk1_3)
stk_0 = x86._mm_xor_si128(tk1_0, sk)
stk_1 = x86._mm_xor_si128(tk1_1, sk)
stk_2 = x86._mm_xor_si128(tk1_2, sk)
stk_3 = x86._mm_xor_si128(tk1_3, sk)
s_0 = x86._mm_aesenc_si128(s_0, stk_0)
s_1 = x86._mm_aesenc_si128(s_1, stk_1)
s_2 = x86._mm_aesenc_si128(s_2, stk_2)
s_3 = x86._mm_aesenc_si128(s_3, stk_3)
}
return s_0, s_1, s_2, s_3
}
@(private = "file", enable_target_feature = "sse2,ssse3,aes", require_results)
bc_x1 :: #force_inline proc "contextless" (
ctx: ^Context,
s: x86.__m128i,
tweak: x86.__m128i,
) -> x86.__m128i #no_bounds_check {
s, tk1 := s, tweak
sk := intrinsics.unaligned_load((^x86.__m128i)(&ctx._subkeys[0]))
stk := x86._mm_xor_si128(tk1, sk)
s = x86._mm_xor_si128(s, stk)
for i in 1 ..= BC_ROUNDS {
sk = intrinsics.unaligned_load((^x86.__m128i)(&ctx._subkeys[i]))
tk1 = h_(tk1)
stk = x86._mm_xor_si128(tk1, sk)
s = x86._mm_aesenc_si128(s, stk)
}
return s
}
@(private = "file", enable_target_feature = "sse2,ssse3,sse4.1,aes", require_results)
bc_absorb :: proc "contextless" (
ctx: ^Context,
tag: x86.__m128i,
src: []byte,
tweak_prefix: x86.__m128i,
stk_block_nr: int,
) -> (x86.__m128i, int) #no_bounds_check {
src, stk_block_nr, tag := src, stk_block_nr, tag
nr_blocks := len(src) / BLOCK_SIZE
for nr_blocks >= 4 {
d_0, d_1, d_2, d_3 := bc_x4(
ctx,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src))),
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src[BLOCK_SIZE:]))),
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src[2*BLOCK_SIZE:]))),
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src[3*BLOCK_SIZE:]))),
auth_tweak(tweak_prefix, stk_block_nr),
auth_tweak(tweak_prefix, stk_block_nr + 1),
auth_tweak(tweak_prefix, stk_block_nr + 2),
auth_tweak(tweak_prefix, stk_block_nr + 3),
)
tag = x86._mm_xor_si128(tag, d_0)
tag = x86._mm_xor_si128(tag, d_1)
tag = x86._mm_xor_si128(tag, d_2)
tag = x86._mm_xor_si128(tag, d_3)
src = src[4*BLOCK_SIZE:]
stk_block_nr += 4
nr_blocks -= 4
}
for nr_blocks > 0 {
d := bc_x1(
ctx,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src))),
auth_tweak(tweak_prefix, stk_block_nr),
)
tag = x86._mm_xor_si128(tag, d)
src = src[BLOCK_SIZE:]
stk_block_nr += 1
nr_blocks -= 1
}
return tag, stk_block_nr
}
@(private = "file", enable_target_feature = "sse2,ssse3,aes", require_results)
bc_final :: proc "contextless" (
ctx: ^Context,
tag: x86.__m128i,
iv: []byte,
) -> x86.__m128i {
tmp: [BLOCK_SIZE]byte
tmp[0] = PREFIX_TAG << PREFIX_SHIFT
copy(tmp[1:], iv)
tweak := intrinsics.unaligned_load((^x86.__m128i)(&tmp))
return bc_x1(ctx, tag, tweak)
}
@(private = "file", enable_target_feature = "sse2,ssse3,aes", require_results)
bc_encrypt :: proc "contextless" (
ctx: ^Context,
dst: []byte,
src: []byte,
iv: x86.__m128i,
tweak_tag: x86.__m128i,
stk_block_nr: int,
) -> int {
dst, src, stk_block_nr := dst, src, stk_block_nr
nr_blocks := len(src) / BLOCK_SIZE
for nr_blocks >= 4 {
d_0, d_1, d_2, d_3 := bc_x4(
ctx,
iv, iv, iv, iv,
enc_tweak(tweak_tag, stk_block_nr),
enc_tweak(tweak_tag, stk_block_nr + 1),
enc_tweak(tweak_tag, stk_block_nr + 2),
enc_tweak(tweak_tag, stk_block_nr + 3),
)
intrinsics.unaligned_store(
(^x86.__m128i)(raw_data(dst)),
x86._mm_xor_si128(
d_0,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src))),
),
)
intrinsics.unaligned_store(
(^x86.__m128i)(raw_data(dst[BLOCK_SIZE:])),
x86._mm_xor_si128(
d_1,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src[BLOCK_SIZE:]))),
),
)
intrinsics.unaligned_store(
(^x86.__m128i)(raw_data(dst[2*BLOCK_SIZE:])),
x86._mm_xor_si128(
d_2,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src[2*BLOCK_SIZE:]))),
),
)
intrinsics.unaligned_store(
(^x86.__m128i)(raw_data(dst[3*BLOCK_SIZE:])),
x86._mm_xor_si128(
d_3,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src[3*BLOCK_SIZE:]))),
),
)
src, dst = src[4*BLOCK_SIZE:], dst[4*BLOCK_SIZE:]
stk_block_nr += 4
nr_blocks -= 4
}
for nr_blocks > 0 {
d := bc_x1(
ctx,
iv,
enc_tweak(tweak_tag, stk_block_nr),
)
intrinsics.unaligned_store(
(^x86.__m128i)(raw_data(dst)),
x86._mm_xor_si128(
d,
intrinsics.unaligned_load((^x86.__m128i)(raw_data(src))),
),
)
src, dst = src[BLOCK_SIZE:], dst[BLOCK_SIZE:]
stk_block_nr += 1
nr_blocks -= 1
}
return stk_block_nr
}
@(private)
e_hw :: proc "contextless" (ctx: ^Context, dst, tag, iv, aad, plaintext: []byte) #no_bounds_check {
tmp: [BLOCK_SIZE]byte
copy(tmp[1:], iv)
iv_ := intrinsics.unaligned_load((^x86.__m128i)(raw_data(&tmp)))
// Algorithm 3
//
// Associated data
// A_1 || ... || A_la || A_∗ <- A where each |A_i| = n and |A_∗| < n
// Auth <- 0^n
// for i = 0 to la − 1 do
// Auth <- Auth ^ EK(0010 || i, A_i+1)
// end
// if A_∗ != nil then
// Auth <- Auth ^ EK(0110 || la, pad10∗(A_∗))
// end
auth: x86.__m128i
n: int
aad := aad
auth, n = bc_absorb(ctx, auth, aad, _PREFIX_AD_BLOCK, 0)
aad = aad[n*BLOCK_SIZE:]
if l := len(aad); l > 0 {
a_star: [BLOCK_SIZE]byte
copy(a_star[:], aad)
a_star[l] = 0x80
auth, _ = bc_absorb(ctx, auth, a_star[:], _PREFIX_AD_FINAL, n)
}
// Message authentication and tag generation
// M_1 || ... || M_l || M_∗ <- M where each |M_j| = n and |M_∗| < n
// tag <- Auth
// for j = 0 to l − 1 do
// tag <- tag ^ EK(0000 || j, M_j+1)
// end
// if M_∗ != nil then
// tag <- tag ^ EK(0100 || l, pad10∗(M_∗))
// end
// tag <- EK(0001 || 0^4 ||N, tag)
m := plaintext
auth, n = bc_absorb(ctx, auth, m, _PREFIX_MSG_BLOCK, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
m_star[l] = 0x80
auth, _ = bc_absorb(ctx, auth, m_star[:], _PREFIX_MSG_FINAL, n)
}
auth = bc_final(ctx, auth, iv)
// Message encryption
// for j = 0 to l − 1 do
// C_j <- M_j ^ EK(1 || tag ^ j, 0^8 || N)
// end
// if M_∗ != nil then
// C_∗ <- M_* ^ EK(1 || tag ^ l, 0^8 || N)
// end
//
// return (C_1 || ... || C_l || C_∗, tag)
m = plaintext
n = bc_encrypt(ctx, dst, m, iv_, auth, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
_ = bc_encrypt(ctx, m_star[:], m_star[:], iv_, auth, n)
copy(dst[n*BLOCK_SIZE:], m_star[:])
}
intrinsics.unaligned_store((^x86.__m128i)(raw_data(tag)), auth)
}
@(private, require_results)
d_hw :: proc "contextless" (ctx: ^Context, dst, iv, aad, ciphertext, tag: []byte) -> bool {
tmp: [BLOCK_SIZE]byte
copy(tmp[1:], iv)
iv_ := intrinsics.unaligned_load((^x86.__m128i)(raw_data(&tmp)))
// Algorithm 4
//
// Message decryption
// C_1 || ... || C_l || C_∗ <- C where each |C_j| = n and |C_∗| < n
// for j = 0 to l − 1 do
// M_j <- C_j ^ EK(1 || tag ^ j, 0^8 || N)
// end
// if C_∗ != nil then
// M_∗ <- C_∗ ^ EK(1 || tag ^ l, 0^8 || N)
// end
auth := intrinsics.unaligned_load((^x86.__m128i)(raw_data(tag)))
m := ciphertext
n := bc_encrypt(ctx, dst, m, iv_, auth, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
_ = bc_encrypt(ctx, m_star[:], m_star[:], iv_, auth, n)
copy(dst[n*BLOCK_SIZE:], m_star[:])
mem.zero_explicit(&m_star, size_of(m_star))
}
// Associated data
// A_1 || ... || Al_a || A_∗ <- A where each |Ai_| = n and |A_∗| < n
// Auth <- 0
// for i = 0 to la − 1 do
// Auth <- Auth ^ EK(0010 || i, A_i+1)
// end
// if A∗ != nil then
// Auth <- Auth ^ EK(0110| | l_a, pad10∗(A_∗))
// end
auth = x86.__m128i{0, 0}
aad := aad
auth, n = bc_absorb(ctx, auth, aad, _PREFIX_AD_BLOCK, 0)
aad = aad[BLOCK_SIZE*n:]
if l := len(aad); l > 0 {
a_star: [BLOCK_SIZE]byte
copy(a_star[:], aad)
a_star[l] = 0x80
auth, _ = bc_absorb(ctx, auth, a_star[:], _PREFIX_AD_FINAL, n)
}
// Message authentication and tag generation
// M_1 || ... || M_l || M_∗ <- M where each |M_j| = n and |M_∗| < n
// tag0 <- Auth
// for j = 0 to l − 1 do
// tag0 <- tag0 ^ EK(0000 || j, M_j+1)
// end
// if M_∗ != nil then
// tag0 <- tag0 ^ EK(0100 || l, pad10∗(M_∗))
// end
// tag0 <- EK(0001 || 0^4 || N, tag0)
m = dst[:len(ciphertext)]
auth, n = bc_absorb(ctx, auth, m, _PREFIX_MSG_BLOCK, 0)
m = m[n*BLOCK_SIZE:]
if l := len(m); l > 0 {
m_star: [BLOCK_SIZE]byte
copy(m_star[:], m)
m_star[l] = 0x80
auth, _ = bc_absorb(ctx, auth, m_star[:], _PREFIX_MSG_FINAL, n)
}
auth = bc_final(ctx, auth, iv)
// Tag verification
// if tag0 = tag then return (M_1 || ... || M_l || M_∗)
// else return false
intrinsics.unaligned_store((^x86.__m128i)(raw_data(&tmp)), auth)
ok := crypto.compare_constant_time(tmp[:], tag) == 1
mem.zero_explicit(&tmp, size_of(tmp))
return ok
}

32
core/crypto/ed25519/ed25519.odin
View File

@@ -81,12 +81,8 @@ private_key_set_bytes :: proc(priv_key: ^Private_Key, b: []byte) -> bool {
// private_key_bytes sets dst to byte-encoding of priv_key.
private_key_bytes :: proc(priv_key: ^Private_Key, dst: []byte) {
if !priv_key._is_initialized {
panic("crypto/ed25519: uninitialized private key")
}
if len(dst) != PRIVATE_KEY_SIZE {
panic("crypto/ed25519: invalid destination size")
}
ensure(priv_key._is_initialized, "crypto/ed25519: uninitialized private key")
ensure(len(dst) == PRIVATE_KEY_SIZE, "crypto/ed25519: invalid destination size")
copy(dst, priv_key._b[:])
}
@@ -98,12 +94,8 @@ private_key_clear :: proc "contextless" (priv_key: ^Private_Key) {
// sign writes the signature by priv_key over msg to sig.
sign :: proc(priv_key: ^Private_Key, msg, sig: []byte) {
if !priv_key._is_initialized {
panic("crypto/ed25519: uninitialized private key")
}
if len(sig) != SIGNATURE_SIZE {
panic("crypto/ed25519: invalid destination size")
}
ensure(priv_key._is_initialized, "crypto/ed25519: uninitialized private key")
ensure(len(sig) == SIGNATURE_SIZE, "crypto/ed25519: invalid destination size")
// 1. Compute the hash of the private key d, H(d) = (h_0, h_1, ..., h_2b-1)
// using SHA-512 for Ed25519. H(d) may be precomputed.
@@ -178,9 +170,7 @@ public_key_set_bytes :: proc "contextless" (pub_key: ^Public_Key, b: []byte) ->
// public_key_set_priv sets pub_key to the public component of priv_key.
public_key_set_priv :: proc(pub_key: ^Public_Key, priv_key: ^Private_Key) {
if !priv_key._is_initialized {
panic("crypto/ed25519: uninitialized public key")
}
ensure(priv_key._is_initialized, "crypto/ed25519: uninitialized public key")
src := &priv_key._pub_key
copy(pub_key._b[:], src._b[:])
@@ -191,21 +181,15 @@ public_key_set_priv :: proc(pub_key: ^Public_Key, priv_key: ^Private_Key) {
// public_key_bytes sets dst to byte-encoding of pub_key.
public_key_bytes :: proc(pub_key: ^Public_Key, dst: []byte) {
if !pub_key._is_initialized {
panic("crypto/ed25519: uninitialized public key")
}
if len(dst) != PUBLIC_KEY_SIZE {
panic("crypto/ed25519: invalid destination size")
}
ensure(pub_key._is_initialized, "crypto/ed25519: uninitialized public key")
ensure(len(dst) == PUBLIC_KEY_SIZE, "crypto/ed25519: invalid destination size")
copy(dst, pub_key._b[:])
}
// public_key_equal returns true iff pub_key is equal to other.
public_key_equal :: proc(pub_key, other: ^Public_Key) -> bool {
if !pub_key._is_initialized || !other._is_initialized {
panic("crypto/ed25519: uninitialized public key")
}
ensure(pub_key._is_initialized && other._is_initialized, "crypto/ed25519: uninitialized public key")
return crypto.compare_constant_time(pub_key._b[:], other._b[:]) == 1
}

13
core/crypto/hmac/hmac.odin
View File

@@ -56,7 +56,7 @@ init :: proc(ctx: ^Context, algorithm: hash.Algorithm, key: []byte) {
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
hash.update(&ctx._i_hash, data)
}
@@ -64,13 +64,10 @@ update :: proc(ctx: ^Context, data: []byte) {
// final finalizes the Context, writes the tag to dst, and calls
// reset on the Context.
final :: proc(ctx: ^Context, dst: []byte) {
assert(ctx._is_initialized)
defer (reset(ctx))
if len(dst) != ctx._tag_sz {
panic("crypto/hmac: invalid destination tag size")
}
ensure(ctx._is_initialized)
ensure(len(dst) == ctx._tag_sz, "crypto/hmac: invalid destination tag size")
hash.final(&ctx._i_hash, dst) // H((k ^ ipad) || text)
@@ -105,14 +102,14 @@ reset :: proc(ctx: ^Context) {
// algorithm returns the Algorithm used by a Context instance.
algorithm :: proc(ctx: ^Context) -> hash.Algorithm {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
return hash.algorithm(&ctx._i_hash)
}
// tag_size returns the tag size of a Context instance in bytes.
tag_size :: proc(ctx: ^Context) -> int {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
return ctx._tag_sz
}

14
core/crypto/kmac/kmac.odin
View File

@@ -36,6 +36,7 @@ sum :: proc(sec_strength: int, dst, msg, key, domain_sep: []byte) {
// tag is valid.
verify :: proc(sec_strength: int, tag, msg, key, domain_sep: []byte, allocator := context.temp_allocator) -> bool {
derived_tag := make([]byte, len(tag), allocator)
defer(delete(derived_tag))
sum(sec_strength, derived_tag, msg, key, domain_sep)
@@ -59,8 +60,6 @@ init_256 :: proc(ctx: ^Context, key, domain_sep: []byte) {
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx.is_initialized)
shake.write((^shake.Context)(ctx), data)
}
@@ -68,12 +67,9 @@ update :: proc(ctx: ^Context, data: []byte) {
// on the Context. This routine will panic if the dst length is less than
// MIN_TAG_SIZE.
final :: proc(ctx: ^Context, dst: []byte) {
assert(ctx.is_initialized)
defer reset(ctx)
if len(dst) < MIN_TAG_SIZE {
panic("crypto/kmac: invalid KMAC tag_size, too short")
}
ensure(len(dst) >= MIN_TAG_SIZE, "crypto/kmac: invalid KMAC tag_size, too short")
_sha3.final_cshake((^_sha3.Context)(ctx), dst)
}
@@ -103,14 +99,12 @@ _init_kmac :: proc(ctx: ^Context, key, s: []byte, sec_strength: int) {
reset(ctx)
}
if len(key) < sec_strength / 8 {
panic("crypto/kmac: invalid KMAC key, too short")
}
ensure(len(key) >= sec_strength / 8, "crypto/kmac: invalid KMAC key, too short")
ctx_ := (^_sha3.Context)(ctx)
_sha3.init_cshake(ctx_, N_KMAC, s, sec_strength)
_sha3.bytepad(ctx_, [][]byte{key}, _sha3.rate_cshake(sec_strength))
}
@(private)
@(private, rodata)
N_KMAC := []byte{'K', 'M', 'A', 'C'}

18
core/crypto/legacy/keccak/keccak.odin
View File

@@ -40,37 +40,37 @@ BLOCK_SIZE_512 :: _sha3.RATE_512
Context :: distinct _sha3.Context
// init_224 initializes a Context for Keccak-224.
init_224 :: proc(ctx: ^Context) {
init_224 :: proc "contextless" (ctx: ^Context) {
ctx.mdlen = DIGEST_SIZE_224
_init(ctx)
}
// init_256 initializes a Context for Keccak-256.
init_256 :: proc(ctx: ^Context) {
init_256 :: proc "contextless" (ctx: ^Context) {
ctx.mdlen = DIGEST_SIZE_256
_init(ctx)
}
// init_384 initializes a Context for Keccak-384.
init_384 :: proc(ctx: ^Context) {
init_384 :: proc "contextless" (ctx: ^Context) {
ctx.mdlen = DIGEST_SIZE_384
_init(ctx)
}
// init_512 initializes a Context for Keccak-512.
init_512 :: proc(ctx: ^Context) {
init_512 :: proc "contextless" (ctx: ^Context) {
ctx.mdlen = DIGEST_SIZE_512
_init(ctx)
}
@(private)
_init :: proc(ctx: ^Context) {
_init :: proc "contextless" (ctx: ^Context) {
ctx.dsbyte = _sha3.DS_KECCAK
_sha3.init((^_sha3.Context)(ctx))
}
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
update :: proc "contextless" (ctx: ^Context, data: []byte) {
_sha3.update((^_sha3.Context)(ctx), data)
}
@@ -79,17 +79,17 @@ update :: proc(ctx: ^Context, data: []byte) {
//
// Iff finalize_clone is set, final will work on a copy of the Context,
// which is useful for for calculating rolling digests.
final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
final :: proc "contextless" (ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
_sha3.final((^_sha3.Context)(ctx), hash, finalize_clone)
}
// clone clones the Context other into ctx.
clone :: proc(ctx, other: ^Context) {
clone :: proc "contextless" (ctx, other: ^Context) {
_sha3.clone((^_sha3.Context)(ctx), (^_sha3.Context)(other))
}
// reset sanitizes the Context. The Context must be re-initialized to
// be used again.
reset :: proc(ctx: ^Context) {
reset :: proc "contextless" (ctx: ^Context) {
_sha3.reset((^_sha3.Context)(ctx))
}

9
core/crypto/legacy/md5/md5.odin
View File

@@ -53,7 +53,7 @@ init :: proc(ctx: ^Context) {
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx.is_initialized)
ensure(ctx.is_initialized)
for i := 0; i < len(data); i += 1 {
ctx.data[ctx.datalen] = data[i]
@@ -72,11 +72,8 @@ update :: proc(ctx: ^Context, data: []byte) {
// Iff finalize_clone is set, final will work on a copy of the Context,
// which is useful for for calculating rolling digests.
final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
assert(ctx.is_initialized)
if len(hash) < DIGEST_SIZE {
panic("crypto/md5: invalid destination digest size")
}
ensure(ctx.is_initialized)
ensure(len(hash) >= DIGEST_SIZE, "crypto/md5: invalid destination digest size")
ctx := ctx
if finalize_clone {

9
core/crypto/legacy/sha1/sha1.odin
View File

@@ -60,7 +60,7 @@ init :: proc(ctx: ^Context) {
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx.is_initialized)
ensure(ctx.is_initialized)
for i := 0; i < len(data); i += 1 {
ctx.data[ctx.datalen] = data[i]
@@ -79,11 +79,8 @@ update :: proc(ctx: ^Context, data: []byte) {
// Iff finalize_clone is set, final will work on a copy of the Context,
// which is useful for for calculating rolling digests.
final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
assert(ctx.is_initialized)
if len(hash) < DIGEST_SIZE {
panic("crypto/sha1: invalid destination digest size")
}
ensure(ctx.is_initialized)
ensure(len(hash) >= DIGEST_SIZE, "crypto/sha1: invalid destination digest size")
ctx := ctx
if finalize_clone {

12
core/crypto/poly1305/poly1305.odin
View File

@@ -60,9 +60,7 @@ Context :: struct {
// init initializes a Context with the specified key. The key SHOULD be
// unique and MUST be unpredictable for each invocation.
init :: proc(ctx: ^Context, key: []byte) {
if len(key) != KEY_SIZE {
panic("crypto/poly1305: invalid key size")
}
ensure(len(key) == KEY_SIZE, "crypto/poly1305: invalid key size")
// r = le_bytes_to_num(key[0..15])
// r = clamp(r) (r &= 0xffffffc0ffffffc0ffffffc0fffffff)
@@ -85,7 +83,7 @@ init :: proc(ctx: ^Context, key: []byte) {
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx._is_initialized)
ensure(ctx._is_initialized)
msg := data
msg_len := len(data)
@@ -124,12 +122,10 @@ update :: proc(ctx: ^Context, data: []byte) {
// final finalizes the Context, writes the tag to dst, and calls
// reset on the Context.
final :: proc(ctx: ^Context, dst: []byte) {
assert(ctx._is_initialized)
defer reset(ctx)
if len(dst) != TAG_SIZE {
panic("poly1305: invalid destination tag size")
}
ensure(ctx._is_initialized)
ensure(len(dst) == TAG_SIZE, "poly1305: invalid destination tag size")
// Process remaining block
if ctx._leftover > 0 {

48
core/crypto/ristretto255/ristretto255.odin
View File

@@ -16,7 +16,7 @@ ELEMENT_SIZE :: 32
// group element.
WIDE_ELEMENT_SIZE :: 64
@(private)
@(private, rodata)
FE_NEG_ONE := field.Tight_Field_Element {
2251799813685228,
2251799813685247,
@@ -24,7 +24,7 @@ FE_NEG_ONE := field.Tight_Field_Element {
2251799813685247,
2251799813685247,
}
@(private)
@(private, rodata)
FE_INVSQRT_A_MINUS_D := field.Tight_Field_Element {
278908739862762,
821645201101625,
@@ -32,7 +32,7 @@ FE_INVSQRT_A_MINUS_D := field.Tight_Field_Element {
1777959178193151,
2118520810568447,
}
@(private)
@(private, rodata)
FE_ONE_MINUS_D_SQ := field.Tight_Field_Element {
1136626929484150,
1998550399581263,
@@ -40,7 +40,7 @@ FE_ONE_MINUS_D_SQ := field.Tight_Field_Element {
118527312129759,
45110755273534,
}
@(private)
@(private, rodata)
FE_D_MINUS_ONE_SQUARED := field.Tight_Field_Element {
1507062230895904,
1572317787530805,
@@ -48,7 +48,7 @@ FE_D_MINUS_ONE_SQUARED := field.Tight_Field_Element {
317374165784489,
1572899562415810,
}
@(private)
@(private, rodata)
FE_SQRT_AD_MINUS_ONE := field.Tight_Field_Element {
2241493124984347,
425987919032274,
@@ -76,7 +76,7 @@ ge_clear :: proc "contextless" (ge: ^Group_Element) {
// ge_set sets `ge = a`.
ge_set :: proc(ge, a: ^Group_Element) {
_ge_assert_initialized([]^Group_Element{a})
_ge_ensure_initialized([]^Group_Element{a})
grp.ge_set(&ge._p, &a._p)
ge._is_initialized = true
@@ -199,9 +199,7 @@ ge_set_bytes :: proc "contextless" (ge: ^Group_Element, b: []byte) -> bool {
// ge_set_wide_bytes sets ge to the result of deriving a ristretto255
// group element, from a wide (512-bit) byte string.
ge_set_wide_bytes :: proc(ge: ^Group_Element, b: []byte) {
if len(b) != WIDE_ELEMENT_SIZE {
panic("crypto/ristretto255: invalid wide input size")
}
ensure(len(b) == WIDE_ELEMENT_SIZE, "crypto/ristretto255: invalid wide input size")
// The element derivation function on an input string b proceeds as
// follows:
@@ -222,10 +220,8 @@ ge_set_wide_bytes :: proc(ge: ^Group_Element, b: []byte) {
// ge_bytes sets dst to the canonical encoding of ge.
ge_bytes :: proc(ge: ^Group_Element, dst: []byte) {
_ge_assert_initialized([]^Group_Element{ge})
if len(dst) != ELEMENT_SIZE {
panic("crypto/ristretto255: invalid destination size")
}
_ge_ensure_initialized([]^Group_Element{ge})
ensure(len(dst) == ELEMENT_SIZE, "crypto/ristretto255: invalid destination size")
x0, y0, z0, t0 := &ge._p.x, &ge._p.y, &ge._p.z, &ge._p.t
@@ -306,7 +302,7 @@ ge_bytes :: proc(ge: ^Group_Element, dst: []byte) {
// ge_add sets `ge = a + b`.
ge_add :: proc(ge, a, b: ^Group_Element) {
_ge_assert_initialized([]^Group_Element{a, b})
_ge_ensure_initialized([]^Group_Element{a, b})
grp.ge_add(&ge._p, &a._p, &b._p)
ge._is_initialized = true
@@ -314,7 +310,7 @@ ge_add :: proc(ge, a, b: ^Group_Element) {
// ge_double sets `ge = a + a`.
ge_double :: proc(ge, a: ^Group_Element) {
_ge_assert_initialized([]^Group_Element{a})
_ge_ensure_initialized([]^Group_Element{a})
grp.ge_double(&ge._p, &a._p)
ge._is_initialized = true
@@ -322,7 +318,7 @@ ge_double :: proc(ge, a: ^Group_Element) {
// ge_negate sets `ge = -a`.
ge_negate :: proc(ge, a: ^Group_Element) {
_ge_assert_initialized([]^Group_Element{a})
_ge_ensure_initialized([]^Group_Element{a})
grp.ge_negate(&ge._p, &a._p)
ge._is_initialized = true
@@ -330,7 +326,7 @@ ge_negate :: proc(ge, a: ^Group_Element) {
// ge_scalarmult sets `ge = A * sc`.
ge_scalarmult :: proc(ge, A: ^Group_Element, sc: ^Scalar) {
_ge_assert_initialized([]^Group_Element{A})
_ge_ensure_initialized([]^Group_Element{A})
grp.ge_scalarmult(&ge._p, &A._p, sc)
ge._is_initialized = true
@@ -344,7 +340,7 @@ ge_scalarmult_generator :: proc "contextless" (ge: ^Group_Element, sc: ^Scalar)
// ge_scalarmult_vartime sets `ge = A * sc` in variable time.
ge_scalarmult_vartime :: proc(ge, A: ^Group_Element, sc: ^Scalar) {
_ge_assert_initialized([]^Group_Element{A})
_ge_ensure_initialized([]^Group_Element{A})
grp.ge_scalarmult_vartime(&ge._p, &A._p, sc)
ge._is_initialized = true
@@ -358,7 +354,7 @@ ge_double_scalarmult_generator_vartime :: proc(
A: ^Group_Element,
b: ^Scalar,
) {
_ge_assert_initialized([]^Group_Element{A})
_ge_ensure_initialized([]^Group_Element{A})
grp.ge_double_scalarmult_basepoint_vartime(&ge._p, a, &A._p, b)
ge._is_initialized = true
@@ -367,7 +363,7 @@ ge_double_scalarmult_generator_vartime :: proc(
// ge_cond_negate sets `ge = a` iff `ctrl == 0` and `ge = -a` iff `ctrl == 1`.
// Behavior for all other values of ctrl are undefined,
ge_cond_negate :: proc(ge, a: ^Group_Element, ctrl: int) {
_ge_assert_initialized([]^Group_Element{a})
_ge_ensure_initialized([]^Group_Element{a})
grp.ge_cond_negate(&ge._p, &a._p, ctrl)
ge._is_initialized = true
@@ -376,7 +372,7 @@ ge_cond_negate :: proc(ge, a: ^Group_Element, ctrl: int) {
// ge_cond_assign sets `ge = ge` iff `ctrl == 0` and `ge = a` iff `ctrl == 1`.
// Behavior for all other values of ctrl are undefined,
ge_cond_assign :: proc(ge, a: ^Group_Element, ctrl: int) {
_ge_assert_initialized([]^Group_Element{ge, a})
_ge_ensure_initialized([]^Group_Element{ge, a})
grp.ge_cond_assign(&ge._p, &a._p, ctrl)
}
@@ -384,7 +380,7 @@ ge_cond_assign :: proc(ge, a: ^Group_Element, ctrl: int) {
// ge_cond_select sets `ge = a` iff `ctrl == 0` and `ge = b` iff `ctrl == 1`.
// Behavior for all other values of ctrl are undefined,
ge_cond_select :: proc(ge, a, b: ^Group_Element, ctrl: int) {
_ge_assert_initialized([]^Group_Element{a, b})
_ge_ensure_initialized([]^Group_Element{a, b})
grp.ge_cond_select(&ge._p, &a._p, &b._p, ctrl)
ge._is_initialized = true
@@ -393,7 +389,7 @@ ge_cond_select :: proc(ge, a, b: ^Group_Element, ctrl: int) {
// ge_equal returns 1 iff `a == b`, and 0 otherwise.
@(require_results)
ge_equal :: proc(a, b: ^Group_Element) -> int {
_ge_assert_initialized([]^Group_Element{a, b})
_ge_ensure_initialized([]^Group_Element{a, b})
// CT_EQ(x1 * y2, y1 * x2) | CT_EQ(y1 * y2, x1 * x2)
ax_by, ay_bx, ay_by, ax_bx: field.Tight_Field_Element = ---, ---, ---, ---
@@ -501,10 +497,8 @@ ge_map :: proc "contextless" (ge: ^Group_Element, b: []byte) {
}
@(private)
_ge_assert_initialized :: proc(ges: []^Group_Element) {
_ge_ensure_initialized :: proc(ges: []^Group_Element) {
for ge in ges {
if !ge._is_initialized {
panic("crypto/ristretto255: uninitialized group element")
}
ensure(ge._is_initialized, "crypto/ristretto255: uninitialized group element")
}
}

8
core/crypto/ristretto255/ristretto255_scalar.odin
View File

@@ -42,9 +42,7 @@ sc_set_bytes :: proc(sc: ^Scalar, b: []byte) -> bool {
// scalar, from a wide (512-bit) byte string by interpreting b as a
// little-endian value, and reducing it mod the group order.
sc_set_bytes_wide :: proc(sc: ^Scalar, b: []byte) {
if len(b) != WIDE_SCALAR_SIZE {
panic("crypto/ristretto255: invalid wide input size")
}
ensure(len(b) == WIDE_SCALAR_SIZE, "crypto/ristretto255: invalid wide input size")
b_ := (^[WIDE_SCALAR_SIZE]byte)(raw_data(b))
grp.sc_set_bytes_wide(sc, b_)
@@ -52,9 +50,7 @@ sc_set_bytes_wide :: proc(sc: ^Scalar, b: []byte) {
// sc_bytes sets dst to the canonical encoding of sc.
sc_bytes :: proc(sc: ^Scalar, dst: []byte) {
if len(dst) != SCALAR_SIZE {
panic("crypto/ristretto255: invalid destination size")
}
ensure(len(dst) == SCALAR_SIZE, "crypto/ristretto255: invalid destination size")
grp.sc_bytes(dst, sc)
}

59
core/crypto/sha2/sha2.odin
View File

@@ -15,9 +15,9 @@ package sha2
zhibog, dotbmp: Initial implementation.
*/
import "core:encoding/endian"
@(require) import "core:encoding/endian"
import "core:math/bits"
import "core:mem"
@(require) import "core:mem"
// DIGEST_SIZE_224 is the SHA-224 digest size in bytes.
DIGEST_SIZE_224 :: 28
@@ -158,7 +158,7 @@ _init :: proc(ctx: ^$T) {
// update adds more data to the Context.
update :: proc(ctx: ^$T, data: []byte) {
assert(ctx.is_initialized)
ensure(ctx.is_initialized)
when T == Context_256 {
CURR_BLOCK_SIZE :: BLOCK_SIZE_256
@@ -194,11 +194,8 @@ update :: proc(ctx: ^$T, data: []byte) {
// Iff finalize_clone is set, final will work on a copy of the Context,
// which is useful for for calculating rolling digests.
final :: proc(ctx: ^$T, hash: []byte, finalize_clone: bool = false) {
assert(ctx.is_initialized)
if len(hash) * 8 < ctx.md_bits {
panic("crypto/sha2: invalid destination digest size")
}
ensure(ctx.is_initialized)
ensure(len(hash) * 8 >= ctx.md_bits, "crypto/sha2: invalid destination digest size")
ctx := ctx
if finalize_clone {
@@ -238,7 +235,7 @@ final :: proc(ctx: ^$T, hash: []byte, finalize_clone: bool = false) {
endian.unchecked_put_u64be(pad[8:], length_lo)
update(ctx, pad[0:16])
}
assert(ctx.bitlength == 0)
assert(ctx.bitlength == 0) // Check for bugs
when T == Context_256 {
for i := 0; i < ctx.md_bits / 32; i += 1 {
@@ -270,8 +267,8 @@ reset :: proc(ctx: ^$T) {
SHA2 implementation
*/
@(private)
sha256_k := [64]u32 {
@(private, rodata)
SHA256_K := [64]u32 {
0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5,
0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3,
@@ -290,8 +287,8 @@ sha256_k := [64]u32 {
0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2,
}
@(private)
sha512_k := [80]u64 {
@(private, rodata)
SHA512_K := [80]u64 {
0x428a2f98d728ae22, 0x7137449123ef65cd,
0xb5c0fbcfec4d3b2f, 0xe9b5dba58189dbbc,
0x3956c25bf348b538, 0x59f111f1b605d019,
@@ -334,6 +331,11 @@ sha512_k := [80]u64 {
0x5fcb6fab3ad6faec, 0x6c44198c4a475817,
}
@(private)
SHA256_ROUNDS :: 64
@(private)
SHA512_ROUNDS :: 80
@(private)
SHA256_CH :: #force_inline proc "contextless" (x, y, z: u32) -> u32 {
return (x & y) ~ (~x & z)
@@ -395,22 +397,29 @@ SHA512_F4 :: #force_inline proc "contextless" (x: u64) -> u64 {
}
@(private)
sha2_transf :: proc "contextless" (ctx: ^$T, data: []byte) {
sha2_transf :: proc "contextless" (ctx: ^$T, data: []byte) #no_bounds_check {
when T == Context_256 {
w: [64]u32
if is_hardware_accelerated_256() {
sha256_transf_hw(ctx, data)
return
}
w: [SHA256_ROUNDS]u32
wv: [8]u32
t1, t2: u32
CURR_BLOCK_SIZE :: BLOCK_SIZE_256
} else when T == Context_512 {
w: [80]u64
w: [SHA512_ROUNDS]u64
wv: [8]u64
t1, t2: u64
CURR_BLOCK_SIZE :: BLOCK_SIZE_512
}
data := data
for len(data) >= CURR_BLOCK_SIZE {
for i := 0; i < 16; i += 1 {
for i in 0 ..< 16 {
when T == Context_256 {
w[i] = endian.unchecked_get_u32be(data[i * 4:])
} else when T == Context_512 {
@@ -419,22 +428,22 @@ sha2_transf :: proc "contextless" (ctx: ^$T, data: []byte) {
}
when T == Context_256 {
for i := 16; i < 64; i += 1 {
for i in 16 ..< SHA256_ROUNDS {
w[i] = SHA256_F4(w[i - 2]) + w[i - 7] + SHA256_F3(w[i - 15]) + w[i - 16]
}
} else when T == Context_512 {
for i := 16; i < 80; i += 1 {
for i in 16 ..< SHA512_ROUNDS {
w[i] = SHA512_F4(w[i - 2]) + w[i - 7] + SHA512_F3(w[i - 15]) + w[i - 16]
}
}
for i := 0; i < 8; i += 1 {
for i in 0 ..< 8 {
wv[i] = ctx.h[i]
}
when T == Context_256 {
for i := 0; i < 64; i += 1 {
t1 = wv[7] + SHA256_F2(wv[4]) + SHA256_CH(wv[4], wv[5], wv[6]) + sha256_k[i] + w[i]
for i in 0 ..< SHA256_ROUNDS {
t1 = wv[7] + SHA256_F2(wv[4]) + SHA256_CH(wv[4], wv[5], wv[6]) + SHA256_K[i] + w[i]
t2 = SHA256_F1(wv[0]) + SHA256_MAJ(wv[0], wv[1], wv[2])
wv[7] = wv[6]
wv[6] = wv[5]
@@ -446,8 +455,8 @@ sha2_transf :: proc "contextless" (ctx: ^$T, data: []byte) {
wv[0] = t1 + t2
}
} else when T == Context_512 {
for i := 0; i < 80; i += 1 {
t1 = wv[7] + SHA512_F2(wv[4]) + SHA512_CH(wv[4], wv[5], wv[6]) + sha512_k[i] + w[i]
for i in 0 ..< SHA512_ROUNDS {
t1 = wv[7] + SHA512_F2(wv[4]) + SHA512_CH(wv[4], wv[5], wv[6]) + SHA512_K[i] + w[i]
t2 = SHA512_F1(wv[0]) + SHA512_MAJ(wv[0], wv[1], wv[2])
wv[7] = wv[6]
wv[6] = wv[5]
@@ -460,7 +469,7 @@ sha2_transf :: proc "contextless" (ctx: ^$T, data: []byte) {
}
}
for i := 0; i < 8; i += 1 {
for i in 0 ..< 8 {
ctx.h[i] += wv[i]
}

15
core/crypto/sha2/sha2_impl_hw_gen.odin Normal file
View File

@@ -0,0 +1,15 @@
#+build !amd64
package sha2
@(private = "file")
ERR_HW_NOT_SUPPORTED :: "crypto/sha2: hardware implementation unsupported"
// is_hardware_accelerated_256 returns true iff hardware accelerated
// SHA-224/SHA-256 is supported.
is_hardware_accelerated_256 :: proc "contextless" () -> bool {
return false
}
sha256_transf_hw :: proc "contextless" (ctx: ^Context_256, data: []byte) {
panic_contextless(ERR_HW_NOT_SUPPORTED)
}

260
core/crypto/sha2/sha2_impl_hw_intel.odin Normal file
View File

@@ -0,0 +1,260 @@
#+build amd64
package sha2
// Based on the public domain code by Jeffrey Walton, though
// realistically, there only is one sensible way to write this
// and Intel's whitepaper covers it.
//
// See: https://github.com/noloader/SHA-Intrinsics
import "base:intrinsics"
import "core:simd"
import "core:simd/x86"
import "core:sys/info"
@(private = "file")
MASK :: x86.__m128i{0x0405060700010203, 0x0c0d0e0f08090a0b}
@(private = "file")
K_0 :: simd.u64x2{0x71374491428a2f98, 0xe9b5dba5b5c0fbcf}
@(private = "file")
K_1 :: simd.u64x2{0x59f111f13956c25b, 0xab1c5ed5923f82a4}
@(private = "file")
K_2 :: simd.u64x2{0x12835b01d807aa98, 0x550c7dc3243185be}
@(private = "file")
K_3 :: simd.u64x2{0x80deb1fe72be5d74, 0xc19bf1749bdc06a7}
@(private = "file")
K_4 :: simd.u64x2{0xefbe4786e49b69c1, 0x240ca1cc0fc19dc6}
@(private = "file")
K_5 :: simd.u64x2{0x4a7484aa2de92c6f, 0x76f988da5cb0a9dc}
@(private = "file")
K_6 :: simd.u64x2{0xa831c66d983e5152, 0xbf597fc7b00327c8}
@(private = "file")
K_7 :: simd.u64x2{0xd5a79147c6e00bf3, 0x1429296706ca6351}
@(private = "file")
K_8 :: simd.u64x2{0x2e1b213827b70a85, 0x53380d134d2c6dfc}
@(private = "file")
K_9 :: simd.u64x2{0x766a0abb650a7354, 0x92722c8581c2c92e}
@(private = "file")
K_10 :: simd.u64x2{0xa81a664ba2bfe8a1, 0xc76c51a3c24b8b70}
@(private = "file")
K_11 :: simd.u64x2{0xd6990624d192e819, 0x106aa070f40e3585}
@(private = "file")
K_12 :: simd.u64x2{0x1e376c0819a4c116, 0x34b0bcb52748774c}
@(private = "file")
K_13 :: simd.u64x2{0x4ed8aa4a391c0cb3, 0x682e6ff35b9cca4f}
@(private = "file")
K_14 :: simd.u64x2{0x78a5636f748f82ee, 0x8cc7020884c87814}
@(private = "file")
K_15 :: simd.u64x2{0xa4506ceb90befffa, 0xc67178f2bef9a3f7}
// is_hardware_accelerated_256 returns true iff hardware accelerated
// SHA-224/SHA-256 is supported.
is_hardware_accelerated_256 :: proc "contextless" () -> bool {
features, ok := info.cpu_features.?
if !ok {
return false
}
req_features :: info.CPU_Features{
.sse2,
.ssse3,
.sse41,
.sha,
}
return features >= req_features
}
@(private, enable_target_feature="sse2,ssse3,sse4.1,sha")
sha256_transf_hw :: proc "contextless" (ctx: ^Context_256, data: []byte) #no_bounds_check {
// Load the state
tmp := intrinsics.unaligned_load((^x86.__m128i)(&ctx.h[0]))
state_1 := intrinsics.unaligned_load((^x86.__m128i)(&ctx.h[4]))
tmp = x86._mm_shuffle_epi32(tmp, 0xb1) // CDAB
state_1 = x86._mm_shuffle_epi32(state_1, 0x1b) // EFGH
state_0 := x86._mm_alignr_epi8(tmp, state_1, 8) // ABEF
// state_1 = x86._mm_blend_epi16(state_1, tmp, 0xf0) // CDGH
state_1 = kludge_mm_blend_epi16_0xf0(state_1, tmp)
data := data
for len(data) >= BLOCK_SIZE_256 {
state_0_save, state_1_save := state_0, state_1
// Rounds 0-3
msg := intrinsics.unaligned_load((^x86.__m128i)(raw_data(data)))
msg_0 := x86._mm_shuffle_epi8(msg, MASK)
msg = x86._mm_add_epi32(msg_0, x86.__m128i(K_0))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
msg = x86._mm_shuffle_epi32(msg, 0xe)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
// Rounds 4-7
msg_1 := intrinsics.unaligned_load((^x86.__m128i)(raw_data(data[16:])))
msg_1 = x86._mm_shuffle_epi8(msg_1, MASK)
msg = x86._mm_add_epi32(msg_1, x86.__m128i(K_1))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
msg = x86._mm_shuffle_epi32(msg, 0xe)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_0 = x86._mm_sha256msg1_epu32(msg_0, msg_1)
// Rounds 8-11
msg_2 := intrinsics.unaligned_load((^x86.__m128i)(raw_data(data[32:])))
msg_2 = x86._mm_shuffle_epi8(msg_2, MASK)
msg = x86._mm_add_epi32(msg_2, x86.__m128i(K_2))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
msg = x86._mm_shuffle_epi32(msg, 0xe)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_1 = x86._mm_sha256msg1_epu32(msg_1, msg_2)
// Rounds 12-15
msg_3 := intrinsics.unaligned_load((^x86.__m128i)(raw_data(data[48:])))
msg_3 = x86._mm_shuffle_epi8(msg_3, MASK)
msg = x86._mm_add_epi32(msg_3, x86.__m128i(K_3))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_3, msg_2, 4)
msg_0 = x86._mm_add_epi32(msg_0, tmp)
msg_0 = x86._mm_sha256msg2_epu32(msg_0, msg_3)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_2 = x86._mm_sha256msg1_epu32(msg_2, msg_3)
// Rounds 16-19
msg = x86._mm_add_epi32(msg_0, x86.__m128i(K_4))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_0, msg_3, 4)
msg_1 = x86._mm_add_epi32(msg_1, tmp)
msg_1 = x86._mm_sha256msg2_epu32(msg_1, msg_0)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_3 = x86._mm_sha256msg1_epu32(msg_3, msg_0)
// Rounds 20-23
msg = x86._mm_add_epi32(msg_1, x86.__m128i(K_5))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_1, msg_0, 4)
msg_2 = x86._mm_add_epi32(msg_2, tmp)
msg_2 = x86._mm_sha256msg2_epu32(msg_2, msg_1)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_0 = x86._mm_sha256msg1_epu32(msg_0, msg_1)
// Rounds 24-27
msg = x86._mm_add_epi32(msg_2, x86.__m128i(K_6))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_2, msg_1, 4)
msg_3 = x86._mm_add_epi32(msg_3, tmp)
msg_3 = x86._mm_sha256msg2_epu32(msg_3, msg_2)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_1 = x86._mm_sha256msg1_epu32(msg_1, msg_2)
// Rounds 28-31
msg = x86._mm_add_epi32(msg_3, x86.__m128i(K_7))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_3, msg_2, 4)
msg_0 = x86._mm_add_epi32(msg_0, tmp)
msg_0 = x86._mm_sha256msg2_epu32(msg_0, msg_3)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_2 = x86._mm_sha256msg1_epu32(msg_2, msg_3)
// Rounds 32-35
msg = x86._mm_add_epi32(msg_0, x86.__m128i(K_8))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_0, msg_3, 4)
msg_1 = x86._mm_add_epi32(msg_1, tmp)
msg_1 = x86._mm_sha256msg2_epu32(msg_1, msg_0)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_3 = x86._mm_sha256msg1_epu32(msg_3, msg_0)
// Rounds 36-39
msg = x86._mm_add_epi32(msg_1, x86.__m128i(K_9))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_1, msg_0, 4)
msg_2 = x86._mm_add_epi32(msg_2, tmp)
msg_2 = x86._mm_sha256msg2_epu32(msg_2, msg_1)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_0 = x86._mm_sha256msg1_epu32(msg_0, msg_1)
// Rounds 40-43
msg = x86._mm_add_epi32(msg_2, x86.__m128i(K_10))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_2, msg_1, 4)
msg_3 = x86._mm_add_epi32(msg_3, tmp)
msg_3 = x86._mm_sha256msg2_epu32(msg_3, msg_2)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_1 = x86._mm_sha256msg1_epu32(msg_1, msg_2)
// Rounds 44-47
msg = x86._mm_add_epi32(msg_3, x86.__m128i(K_11))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_3, msg_2, 4)
msg_0 = x86._mm_add_epi32(msg_0, tmp)
msg_0 = x86._mm_sha256msg2_epu32(msg_0, msg_3)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_2 = x86._mm_sha256msg1_epu32(msg_2, msg_3)
// Rounds 48-51
msg = x86._mm_add_epi32(msg_0, x86.__m128i(K_12))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_0, msg_3, 4)
msg_1 = x86._mm_add_epi32(msg_1, tmp)
msg_1 = x86._mm_sha256msg2_epu32(msg_1, msg_0)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
msg_3 = x86._mm_sha256msg1_epu32(msg_3, msg_0)
// Rounds 52-55
msg = x86._mm_add_epi32(msg_1, x86.__m128i(K_13))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_1, msg_0, 4)
msg_2 = x86._mm_add_epi32(msg_2, tmp)
msg_2 = x86._mm_sha256msg2_epu32(msg_2, msg_1)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
/* Rounds 56-59 */
msg = x86._mm_add_epi32(msg_2, x86.__m128i(K_14))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
tmp = x86._mm_alignr_epi8(msg_2, msg_1, 4)
msg_3 = x86._mm_add_epi32(msg_3, tmp)
msg_3 = x86._mm_sha256msg2_epu32(msg_3, msg_2)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
// Rounds 60-63
msg = x86._mm_add_epi32(msg_3, x86.__m128i(K_15))
state_1 = x86._mm_sha256rnds2_epu32(state_1, state_0, msg)
msg = x86._mm_shuffle_epi32(msg, 0x0e)
state_0 = x86._mm_sha256rnds2_epu32(state_0, state_1, msg)
state_0 = x86._mm_add_epi32(state_0, state_0_save)
state_1 = x86._mm_add_epi32(state_1, state_1_save)
data = data[BLOCK_SIZE_256:]
}
// Write back the updated state
tmp = x86._mm_shuffle_epi32(state_0, 0x1b) // FEBA
state_1 = x86._mm_shuffle_epi32(state_1, 0xb1) // DCHG
// state_0 = x86._mm_blend_epi16(tmp, state_1, 0xf0) // DCBA
state_0 = kludge_mm_blend_epi16_0xf0(tmp, state_1)
state_1 = x86._mm_alignr_epi8(state_1, tmp, 8) // ABEF
intrinsics.unaligned_store((^x86.__m128i)(&ctx.h[0]), state_0)
intrinsics.unaligned_store((^x86.__m128i)(&ctx.h[4]), state_1)
}
@(private = "file")
kludge_mm_blend_epi16_0xf0 :: #force_inline proc "contextless"(a, b: x86.__m128i) -> x86.__m128i {
// HACK HACK HACK: LLVM got rid of `llvm.x86.sse41.pblendw`.
a_ := simd.to_array(a)
b_ := simd.to_array(b)
return x86.__m128i{a_[0], b_[1]}
}

17
core/crypto/siphash/siphash.odin
View File

@@ -219,18 +219,14 @@ verify_4_8 :: proc {
*/
init :: proc(ctx: ^Context, key: []byte, c_rounds, d_rounds: int) {
if len(key) != KEY_SIZE {
panic("crypto/siphash; invalid key size")
}
ensure(len(key) == KEY_SIZE,"crypto/siphash; invalid key size")
ctx.c_rounds = c_rounds
ctx.d_rounds = d_rounds
is_valid_setting :=
(ctx.c_rounds == 1 && ctx.d_rounds == 3) ||
(ctx.c_rounds == 2 && ctx.d_rounds == 4) ||
(ctx.c_rounds == 4 && ctx.d_rounds == 8)
if !is_valid_setting {
panic("crypto/siphash: incorrect rounds set up")
}
ensure(is_valid_setting, "crypto/siphash: incorrect rounds set up")
ctx.k0 = endian.unchecked_get_u64le(key[:8])
ctx.k1 = endian.unchecked_get_u64le(key[8:])
ctx.v0 = 0x736f6d6570736575 ~ ctx.k0
@@ -245,7 +241,7 @@ init :: proc(ctx: ^Context, key: []byte, c_rounds, d_rounds: int) {
}
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx.is_initialized, "crypto/siphash: context is not initialized")
ensure(ctx.is_initialized)
data := data
ctx.total_length += len(data)
@@ -269,7 +265,7 @@ update :: proc(ctx: ^Context, data: []byte) {
}
final :: proc(ctx: ^Context, dst: ^u64) {
assert(ctx.is_initialized, "crypto/siphash: context is not initialized")
ensure(ctx.is_initialized)
tmp: [BLOCK_SIZE]byte
copy(tmp[:], ctx.buf[:ctx.last_block])
@@ -336,9 +332,8 @@ _get_byte :: #force_inline proc "contextless" (byte_num: byte, into: u64) -> byt
@(private)
_collect_output :: #force_inline proc(dst: []byte, hash: u64) {
if len(dst) < DIGEST_SIZE {
panic("crypto/siphash: invalid tag size")
}
ensure(len(dst) >= DIGEST_SIZE, "crypto/siphash: invalid tag size")
dst[0] = _get_byte(7, hash)
dst[1] = _get_byte(6, hash)
dst[2] = _get_byte(5, hash)

13
core/crypto/sm3/sm3.odin
View File

@@ -53,7 +53,7 @@ init :: proc(ctx: ^Context) {
// update adds more data to the Context.
update :: proc(ctx: ^Context, data: []byte) {
assert(ctx.is_initialized)
ensure(ctx.is_initialized)
data := data
ctx.length += u64(len(data))
@@ -83,11 +83,8 @@ update :: proc(ctx: ^Context, data: []byte) {
// Iff finalize_clone is set, final will work on a copy of the Context,
// which is useful for for calculating rolling digests.
final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
assert(ctx.is_initialized)
if len(hash) < DIGEST_SIZE {
panic("crypto/sm3: invalid destination digest size")
}
ensure(ctx.is_initialized)
ensure(len(hash) >= DIGEST_SIZE, "crypto/sm3: invalid destination digest size")
ctx := ctx
if finalize_clone {
@@ -110,7 +107,7 @@ final :: proc(ctx: ^Context, hash: []byte, finalize_clone: bool = false) {
length <<= 3
endian.unchecked_put_u64be(pad[:], length)
update(ctx, pad[0:8])
assert(ctx.bitlength == 0)
assert(ctx.bitlength == 0) // Check for bugs
for i := 0; i < DIGEST_SIZE / 4; i += 1 {
endian.unchecked_put_u32be(hash[i * 4:], ctx.state[i])
@@ -136,7 +133,7 @@ reset :: proc(ctx: ^Context) {
SM3 implementation
*/
@(private)
@(private, rodata)
IV := [8]u32 {
0x7380166f, 0x4914b2b9, 0x172442d7, 0xda8a0600,
0xa96f30bc, 0x163138aa, 0xe38dee4d, 0xb0fb0e4e,

14
core/crypto/x25519/x25519.odin
View File

@@ -15,7 +15,7 @@ SCALAR_SIZE :: 32
// POINT_SIZE is the size of a X25519 point (public key/shared secret) in bytes.
POINT_SIZE :: 32
@(private)
@(private, rodata)
_BASE_POINT: [32]byte = {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
@(private)
@@ -101,15 +101,9 @@ _scalarmult :: proc "contextless" (out, scalar, point: ^[32]byte) {
// scalarmult "multiplies" the provided scalar and point, and writes the
// resulting point to dst.
scalarmult :: proc(dst, scalar, point: []byte) {
if len(scalar) != SCALAR_SIZE {
panic("crypto/x25519: invalid scalar size")
}
if len(point) != POINT_SIZE {
panic("crypto/x25519: invalid point size")
}
if len(dst) != POINT_SIZE {
panic("crypto/x25519: invalid destination point size")
}
ensure(len(scalar) == SCALAR_SIZE, "crypto/x25519: invalid scalar size")
ensure(len(point) == POINT_SIZE, "crypto/x25519: invalid point size")
ensure(len(dst) == POINT_SIZE, "crypto/x25519: invalid destination point size")
// "clamp" the scalar
e: [32]byte = ---

155
core/crypto/x448/x448.odin Normal file
View File

@@ -0,0 +1,155 @@
/*
package x448 implements the X448 (aka curve448) Elliptic-Curve
Diffie-Hellman key exchange protocol.
See:
- [[ https://www.rfc-editor.org/rfc/rfc7748 ]]
*/
package x448
import field "core:crypto/_fiat/field_curve448"
import "core:mem"
// SCALAR_SIZE is the size of a X448 scalar (private key) in bytes.
SCALAR_SIZE :: 56
// POINT_SIZE is the size of a X448 point (public key/shared secret) in bytes.
POINT_SIZE :: 56
@(private, rodata)
_BASE_POINT: [56]byte = {
5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0,
}
@(private)
_scalar_bit :: #force_inline proc "contextless" (s: ^[56]byte, i: int) -> u8 {
if i < 0 {
return 0
}
return (s[i >> 3] >> uint(i & 7)) & 1
}
@(private)
_scalarmult :: proc "contextless" (out, scalar, point: ^[56]byte) {
// Montgomery pseudo-multiplication, using the RFC 7748 formula.
t1, t2: field.Loose_Field_Element = ---, ---
// x_1 = u
// x_2 = 1
// z_2 = 0
// x_3 = u
// z_3 = 1
x1: field.Tight_Field_Element = ---
field.fe_from_bytes(&x1, point)
x2, x3, z2, z3: field.Tight_Field_Element = ---, ---, ---, ---
field.fe_one(&x2)
field.fe_zero(&z2)
field.fe_set(&x3, &x1)
field.fe_one(&z3)
// swap = 0
swap: int
// For t = bits-1 down to 0:a
for t := 448 - 1; t >= 0; t -= 1 {
// k_t = (k >> t) & 1
k_t := int(_scalar_bit(scalar, t))
// swap ^= k_t
swap ~= k_t
// Conditional swap; see text below.
// (x_2, x_3) = cswap(swap, x_2, x_3)
field.fe_cond_swap(&x2, &x3, swap)
// (z_2, z_3) = cswap(swap, z_2, z_3)
field.fe_cond_swap(&z2, &z3, swap)
// swap = k_t
swap = k_t
// Note: This deliberately omits reductions after add/sub operations
// if the result is only ever used as the input to a mul/square since
// the implementations of those can deal with non-reduced inputs.
//
// fe_tighten_cast is only used to store a fully reduced
// output in a Loose_Field_Element, or to provide such a
// Loose_Field_Element as a Tight_Field_Element argument.
// A = x_2 + z_2
field.fe_add(&t1, &x2, &z2)
// B = x_2 - z_2
field.fe_sub(&t2, &x2, &z2)
// D = x_3 - z_3
field.fe_sub(field.fe_relax_cast(&z2), &x3, &z3) // (z2 unreduced)
// DA = D * A
field.fe_carry_mul(&x2, field.fe_relax_cast(&z2), &t1)
// C = x_3 + z_3
field.fe_add(field.fe_relax_cast(&z3), &x3, &z3) // (z3 unreduced)
// CB = C * B
field.fe_carry_mul(&x3, &t2, field.fe_relax_cast(&z3))
// z_3 = x_1 * (DA - CB)^2
field.fe_sub(field.fe_relax_cast(&z3), &x2, &x3) // (z3 unreduced)
field.fe_carry_square(&z3, field.fe_relax_cast(&z3))
field.fe_carry_mul(&z3, field.fe_relax_cast(&x1), field.fe_relax_cast(&z3))
// x_3 = (DA + CB)^2
field.fe_add(field.fe_relax_cast(&z2), &x2, &x3) // (z2 unreduced)
field.fe_carry_square(&x3, field.fe_relax_cast(&z2))
// AA = A^2
field.fe_carry_square(&z2, &t1)
// BB = B^2
field.fe_carry_square(field.fe_tighten_cast(&t1), &t2) // (t1 reduced)
// x_2 = AA * BB
field.fe_carry_mul(&x2, field.fe_relax_cast(&z2), &t1)
// E = AA - BB
field.fe_sub(&t2, &z2, field.fe_tighten_cast(&t1)) // (t1 (input) is reduced)
// z_2 = E * (AA + a24 * E)
field.fe_carry_mul_small(field.fe_tighten_cast(&t1), &t2, 39081) // (t1 reduced)
field.fe_add(&t1, &z2, field.fe_tighten_cast(&t1)) // (t1 (input) is reduced)
field.fe_carry_mul(&z2, &t2, &t1)
}
// Conditional swap; see text below.
// (x_2, x_3) = cswap(swap, x_2, x_3)
field.fe_cond_swap(&x2, &x3, swap)
// (z_2, z_3) = cswap(swap, z_2, z_3)
field.fe_cond_swap(&z2, &z3, swap)
// Return x_2 * (z_2^(p - 2))
field.fe_carry_inv(&z2, field.fe_relax_cast(&z2))
field.fe_carry_mul(&x2, field.fe_relax_cast(&x2), field.fe_relax_cast(&z2))
field.fe_to_bytes(out, &x2)
field.fe_clear_vec([]^field.Tight_Field_Element{&x1, &x2, &x3, &z2, &z3})
field.fe_clear_vec([]^field.Loose_Field_Element{&t1, &t2})
}
// scalarmult "multiplies" the provided scalar and point, and writes the
// resulting point to dst.
scalarmult :: proc(dst, scalar, point: []byte) {
ensure(len(scalar) == SCALAR_SIZE, "crypto/x448: invalid scalar size")
ensure(len(point) == POINT_SIZE, "crypto/x448: invalid point size")
ensure(len(dst) == POINT_SIZE, "crypto/x448: invalid destination point size")
// "clamp" the scalar
e: [56]byte = ---
copy_slice(e[:], scalar)
e[0] &= 252
e[55] |= 128
p: [56]byte = ---
copy_slice(p[:], point)
d: [56]byte = ---
_scalarmult(&d, &e, &p)
copy_slice(dst, d[:])
mem.zero_explicit(&e, size_of(e))
mem.zero_explicit(&d, size_of(d))
}
// scalarmult_basepoint "multiplies" the provided scalar with the X448
// base point and writes the resulting point to dst.
scalarmult_basepoint :: proc(dst, scalar: []byte) {
scalarmult(dst, scalar, _BASE_POINT[:])
}

20
core/encoding/json/unmarshal.odin
View File

@@ -117,9 +117,25 @@ assign_int :: proc(val: any, i: $T) -> bool {
case uint: dst = uint (i)
case uintptr: dst = uintptr(i)
case:
is_bit_set_different_endian_to_platform :: proc(ti: ^runtime.Type_Info) -> bool {
if ti == nil {
return false
}
t := runtime.type_info_base(ti)
#partial switch info in t.variant {
case runtime.Type_Info_Integer:
switch info.endianness {
case .Platform: return false
case .Little: return ODIN_ENDIAN != .Little
case .Big: return ODIN_ENDIAN != .Big
}
}
return false
}
ti := type_info_of(v.id)
if _, ok := ti.variant.(^runtime.Type_Info_Bit_Set); ok {
do_byte_swap := !reflect.bit_set_is_big_endian(v)
if info, ok := ti.variant.(^runtime.Type_Info_Bit_Set); ok {
do_byte_swap := is_bit_set_different_endian_to_platform(info.underlying)
switch ti.size * 8 {
case 0: // no-op.
case 8:

23
core/mem/alloc.odin
View File

@@ -953,6 +953,22 @@ make_dynamic_array_len_cap :: proc(
return runtime.make_dynamic_array_len_cap(T, len, cap, allocator, loc)
}
/*
Create a map with no initial allocation.
This procedure creates a map of type `T` with no initial allocation, which will
use the allocator specified by `allocator` as its backing allocator when it
allocates.
*/
@(require_results)
make_map :: proc(
$T: typeid/map[$K]$E,
allocator := context.allocator,
loc := #caller_location,
) -> (m: T) {
return runtime.make_map(T, allocator, loc)
}
/*
Allocate a map.
@@ -961,13 +977,13 @@ This procedure creates a map of type `T` with initial capacity specified by
allocator.
*/
@(require_results)
make_map :: proc(
make_map_cap :: proc(
$T: typeid/map[$K]$E,
#any_int cap: int = 1<<runtime.MAP_MIN_LOG2_CAPACITY,
#any_int cap: int,
allocator := context.allocator,
loc := #caller_location,
) -> (m: T, err: Allocator_Error) {
return runtime.make_map(T, cap, allocator, loc)
return runtime.make_map_cap(T, cap, allocator, loc)
}
/*
@@ -1060,6 +1076,7 @@ make :: proc{
make_dynamic_array_len,
make_dynamic_array_len_cap,
make_map,
make_map_cap,
make_multi_pointer,
make_soa_slice,
make_soa_dynamic_array,

12
core/mem/virtual/arena.odin
View File

@@ -107,8 +107,9 @@ arena_alloc :: proc(arena: ^Arena, size: uint, alignment: uint, loc := #caller_l
switch arena.kind {
case .Growing:
needed := mem.align_forward_uint(size, alignment)
if arena.curr_block == nil || (safe_add(arena.curr_block.used, needed) or_else 0) > arena.curr_block.reserved {
prev_used := 0 if arena.curr_block == nil else arena.curr_block.used
data, err = alloc_from_memory_block(arena.curr_block, size, alignment, default_commit_size=arena.default_commit_size)
if err == .Out_Of_Memory {
if arena.minimum_block_size == 0 {
arena.minimum_block_size = DEFAULT_ARENA_GROWING_MINIMUM_BLOCK_SIZE
arena.minimum_block_size = mem.align_forward_uint(arena.minimum_block_size, DEFAULT_PAGE_SIZE)
@@ -124,6 +125,7 @@ arena_alloc :: proc(arena: ^Arena, size: uint, alignment: uint, loc := #caller_l
max(arena.default_commit_size, arena.minimum_block_size)
}
needed := mem.align_forward_uint(size, alignment)
needed = max(needed, arena.default_commit_size)
block_size := max(needed, arena.minimum_block_size)
@@ -131,10 +133,10 @@ arena_alloc :: proc(arena: ^Arena, size: uint, alignment: uint, loc := #caller_l
new_block.prev = arena.curr_block
arena.curr_block = new_block
arena.total_reserved += new_block.reserved
}
prev_used := arena.curr_block.used
data, err = alloc_from_memory_block(arena.curr_block, size, alignment, default_commit_size=arena.default_commit_size)
prev_used = 0
data, err = alloc_from_memory_block(arena.curr_block, size, alignment, default_commit_size=arena.default_commit_size)
}
arena.total_used += arena.curr_block.used - prev_used
case .Static:
if arena.curr_block == nil {

20
core/mem/virtual/virtual_darwin.odin Normal file
View File

@@ -0,0 +1,20 @@
package mem_virtual
import "core:sys/posix"
_reserve :: proc "contextless" (size: uint) -> (data: []byte, err: Allocator_Error) {
result := posix.mmap(nil, size, {}, {.ANONYMOUS, .PRIVATE})
if result == posix.MAP_FAILED {
assert_contextless(posix.errno() == .ENOMEM)
return nil, .Out_Of_Memory
}
return ([^]byte)(uintptr(result))[:size], nil
}
_decommit :: proc "contextless" (data: rawptr, size: uint) {
MADV_FREE :: 5
posix.mprotect(data, size, {})
posix.posix_madvise(data, size, transmute(posix.MAdvice)i32(MADV_FREE))
}

26
core/mem/virtual/virtual_freebsd.odin Normal file
View File

@@ -0,0 +1,26 @@
package mem_virtual
import "core:sys/posix"
_reserve :: proc "contextless" (size: uint) -> (data: []byte, err: Allocator_Error) {
PROT_MAX :: proc "contextless" (flags: posix.Prot_Flags) -> posix.Prot_Flags {
_PROT_MAX_SHIFT :: 16
return transmute(posix.Prot_Flags)(transmute(i32)flags << _PROT_MAX_SHIFT)
}
result := posix.mmap(nil, size, PROT_MAX({.READ, .WRITE, .EXEC}), {.ANONYMOUS, .PRIVATE})
if result == posix.MAP_FAILED {
assert_contextless(posix.errno() == .ENOMEM)
return nil, .Out_Of_Memory
}
return ([^]byte)(uintptr(result))[:size], nil
}
_decommit :: proc "contextless" (data: rawptr, size: uint) {
MADV_FREE :: 5
posix.mprotect(data, size, {})
posix.posix_madvise(data, size, transmute(posix.MAdvice)i32(MADV_FREE))
}

25
core/mem/virtual/virtual_netbsd.odin Normal file
View File

@@ -0,0 +1,25 @@
package mem_virtual
import "core:sys/posix"
_reserve :: proc "contextless" (size: uint) -> (data: []byte, err: Allocator_Error) {
PROT_MPROTECT :: proc "contextless" (flags: posix.Prot_Flags) -> posix.Prot_Flags {
return transmute(posix.Prot_Flags)(transmute(i32)flags << 3)
}
result := posix.mmap(nil, size, PROT_MPROTECT({.READ, .WRITE, .EXEC}), {.ANONYMOUS, .PRIVATE})
if result == posix.MAP_FAILED {
assert_contextless(posix.errno() == .ENOMEM)
return nil, .Out_Of_Memory
}
return ([^]byte)(uintptr(result))[:size], nil
}
_decommit :: proc "contextless" (data: rawptr, size: uint) {
MADV_FREE :: 6
posix.mprotect(data, size, {})
posix.posix_madvise(data, size, transmute(posix.MAdvice)i32(MADV_FREE))
}

20
core/mem/virtual/virtual_openbsd.odin Normal file
View File

@@ -0,0 +1,20 @@
package mem_virtual
import "core:sys/posix"
_reserve :: proc "contextless" (size: uint) -> (data: []byte, err: Allocator_Error) {
result := posix.mmap(nil, size, {}, {.ANONYMOUS, .PRIVATE})
if result == posix.MAP_FAILED {
assert_contextless(posix.errno() == .ENOMEM)
return nil, .Out_Of_Memory
}
return ([^]byte)(uintptr(result))[:size], nil
}
_decommit :: proc "contextless" (data: rawptr, size: uint) {
MADV_FREE :: 6
posix.mprotect(data, size, {})
posix.posix_madvise(data, size, transmute(posix.MAdvice)i32(MADV_FREE))
}

4
core/mem/virtual/virtual_platform.odin
View File

@@ -15,7 +15,9 @@ platform_memory_alloc :: proc "contextless" (to_commit, to_reserve: uint) -> (bl
to_commit = clamp(to_commit, size_of(Platform_Memory_Block), total_to_reserved)
data := reserve(total_to_reserved) or_return
commit(raw_data(data), to_commit)
commit_err := commit(raw_data(data), to_commit)
assert_contextless(commit_err == nil)
block = (^Platform_Memory_Block)(raw_data(data))
block.committed = to_commit

28
core/mem/virtual/virtual_posix.odin
View File

@@ -4,36 +4,18 @@ package mem_virtual
import "core:sys/posix"
// Define non-posix needed flags:
when ODIN_OS == .Darwin || ODIN_OS == .FreeBSD {
MADV_FREE :: 5 /* pages unneeded, discard contents */
} else when ODIN_OS == .OpenBSD || ODIN_OS == .NetBSD {
MADV_FREE :: 6
}
_reserve :: proc "contextless" (size: uint) -> (data: []byte, err: Allocator_Error) {
flags := posix.Map_Flags{ .ANONYMOUS, .PRIVATE }
result := posix.mmap(nil, size, {}, flags)
if result == posix.MAP_FAILED {
return nil, .Out_Of_Memory
}
return ([^]byte)(uintptr(result))[:size], nil
}
_commit :: proc "contextless" (data: rawptr, size: uint) -> Allocator_Error {
if posix.mprotect(data, size, { .READ, .WRITE }) != .OK {
return .Out_Of_Memory
#partial switch posix.errno() {
case .EACCES, .EPERM: return .Invalid_Pointer
case .ENOTSUP, .EINVAL: return .Invalid_Argument
case: return .Out_Of_Memory
}
}
return nil
}
_decommit :: proc "contextless" (data: rawptr, size: uint) {
posix.mprotect(data, size, {})
posix.posix_madvise(data, size, transmute(posix.MAdvice)i32(MADV_FREE))
}
_release :: proc "contextless" (data: rawptr, size: uint) {
posix.munmap(data, size)
}

5
core/os/dir_unix.odin
View File

@@ -5,9 +5,10 @@ import "core:strings"
@(require_results)
read_dir :: proc(fd: Handle, n: int, allocator := context.allocator) -> (fi: []File_Info, err: Error) {
dupfd := _dup(fd) or_return
context.allocator = allocator
dirp := _fdopendir(dupfd) or_return
dupfd := _dup(fd) or_return
dirp := _fdopendir(dupfd) or_return
defer _closedir(dirp)
dirpath := absolute_path_from_handle(dupfd) or_return

413
core/os/os2/path.odin
View File

@@ -2,12 +2,18 @@ package os2
import "base:runtime"
import "core:path/filepath"
import "core:strings"
Path_Separator :: _Path_Separator // OS-Specific
Path_Separator_String :: _Path_Separator_String // OS-Specific
Path_List_Separator :: _Path_List_Separator // OS-Specific
#assert(_Path_Separator <= rune(0x7F), "The system-specific path separator rune is expected to be within the 7-bit ASCII character set.")
/*
Return true if `c` is a character used to separate paths into directory and
file hierarchies on the current system.
*/
@(require_results)
is_path_separator :: proc(c: byte) -> bool {
return _is_path_separator(c)
@@ -15,22 +21,42 @@ is_path_separator :: proc(c: byte) -> bool {
mkdir :: make_directory
/*
Make a new directory.
If `path` is relative, it will be relative to the process's current working directory.
*/
make_directory :: proc(name: string, perm: int = 0o755) -> Error {
return _mkdir(name, perm)
}
mkdir_all :: make_directory_all
/*
Make a new directory, creating new intervening directories when needed.
If `path` is relative, it will be relative to the process's current working directory.
*/
make_directory_all :: proc(path: string, perm: int = 0o755) -> Error {
return _mkdir_all(path, perm)
}
/*
Delete `path` and all files and directories inside of `path` if it is a directory.
If `path` is relative, it will be relative to the process's current working directory.
*/
remove_all :: proc(path: string) -> Error {
return _remove_all(path)
}
getwd :: get_working_directory
/*
Get the working directory of the current process.
*Allocates Using Provided Allocator*
*/
@(require_results)
get_working_directory :: proc(allocator: runtime.Allocator) -> (dir: string, err: Error) {
return _get_working_directory(allocator)
@@ -38,16 +64,399 @@ get_working_directory :: proc(allocator: runtime.Allocator) -> (dir: string, err
setwd :: set_working_directory
/*
Change the working directory of the current process.
*Allocates Using Provided Allocator*
*/
set_working_directory :: proc(dir: string) -> (err: Error) {
return _set_working_directory(dir)
}
/*
Get the path for the currently running executable.
*Allocates Using Provided Allocator*
*/
@(require_results)
get_executable_path :: proc(allocator: runtime.Allocator) -> (path: string, err: Error) {
return _get_executable_path(allocator)
}
/*
Get the directory for the currently running executable.
*Allocates Using Provided Allocator*
*/
@(require_results)
get_executable_directory :: proc(allocator: runtime.Allocator) -> (path: string, err: Error) {
path = _get_executable_path(allocator) or_return
path, _ = filepath.split(path)
path, _ = split_path(path)
return
}
/*
Compare two paths for exactness without normalization.
This procedure takes into account case-sensitivity on differing systems.
*/
@(require_results)
are_paths_identical :: proc(a, b: string) -> (identical: bool) {
return _are_paths_identical(a, b)
}
/*
Normalize a path.
*Allocates Using Provided Allocator*
This will remove duplicate separators and unneeded references to the current or
parent directory.
*/
@(require_results)
clean_path :: proc(path: string, allocator: runtime.Allocator) -> (cleaned: string, err: Error) {
if path == "" || path == "." {
return strings.clone(".", allocator)
}
TEMP_ALLOCATOR_GUARD()
// The extra byte is to simplify appending path elements by letting the
// loop to end each with a separator. We'll trim the last one when we're done.
buffer := make([]u8, len(path) + 1, temp_allocator()) or_return
// This is the only point where Windows and POSIX differ, as Windows has
// alphabet-based volumes for root paths.
rooted, start := _clean_path_handle_start(path, buffer)
head, buffer_i := start, start
for i, j := start, start; i <= len(path); i += 1 {
if i == len(path) || _is_path_separator(path[i]) {
elem := path[j:i]
j = i + 1
switch elem {
case "", ".":
// Skip duplicate path separators and current directory references.
case "..":
if !rooted && buffer_i == head {
// Only allow accessing further parent directories when the path is relative.
buffer[buffer_i] = '.'
buffer[buffer_i+1] = '.'
buffer[buffer_i+2] = _Path_Separator
buffer_i += 3
head = buffer_i
} else {
// Roll back to the last separator or the head of the buffer.
back_to := head
// `buffer_i` will be equal to 1 + the last set byte, so
// skipping two bytes avoids the final separator we just
// added.
for k := buffer_i-2; k >= head; k -= 1 {
if _is_path_separator(buffer[k]) {
back_to = k + 1
break
}
}
buffer_i = back_to
}
case:
// Copy the path element verbatim and add a separator.
copy(buffer[buffer_i:], elem)
buffer_i += len(elem)
buffer[buffer_i] = _Path_Separator
buffer_i += 1
}
}
}
// Trim the final separator.
// NOTE: No need to check if the last byte is a separator, as we always add it.
if buffer_i > start {
buffer_i -= 1
}
if buffer_i == 0 {
return strings.clone(".", allocator)
}
compact := make([]u8, buffer_i, allocator) or_return
copy(compact, buffer) // NOTE(bill): buffer[:buffer_i] is redundant here
return string(compact), nil
}
/*
Return true if `path` is an absolute path as opposed to a relative one.
*/
@(require_results)
is_absolute_path :: proc(path: string) -> bool {
return _is_absolute_path(path)
}
/*
Get the absolute path to `path` with respect to the process's current directory.
*Allocates Using Provided Allocator*
*/
@(require_results)
get_absolute_path :: proc(path: string, allocator: runtime.Allocator) -> (absolute_path: string, err: Error) {
return _get_absolute_path(path, allocator)
}
/*
Get the relative path needed to change directories from `base` to `target`.
*Allocates Using Provided Allocator*
The result is such that `join_path(base, get_relative_path(base, target))` is equivalent to `target`.
NOTE: This procedure expects both `base` and `target` to be normalized first,
which can be done by calling `clean_path` on them if needed.
This procedure will return an `Invalid_Path` error if `base` begins with a
reference to the parent directory (`".."`). Use `get_working_directory` with
`join_path` to construct absolute paths for both arguments instead.
*/
@(require_results)
get_relative_path :: proc(base, target: string, allocator: runtime.Allocator) -> (path: string, err: Error) {
if _are_paths_identical(base, target) {
return strings.clone(".", allocator)
}
if base == "." {
return strings.clone(target, allocator)
}
// This is the first point where Windows and POSIX differ, as Windows has
// alphabet-based volumes for root paths.
if !_get_relative_path_handle_start(base, target) {
return "", .Invalid_Path
}
if strings.has_prefix(base, "..") && (len(base) == 2 || _is_path_separator(base[2])) {
// We could do the work for the user of getting absolute paths for both
// arguments, but that could make something costly (repeatedly
// normalizing paths) convenient, when it would be better for the user
// to store already-finalized paths and operate on those instead.
return "", .Invalid_Path
}
// This is the other point where Windows and POSIX differ, as Windows is
// case-insensitive.
common := _get_common_path_len(base, target)
// Get the result of splitting `base` and `target` on _Path_Separator,
// comparing them up to their most common elements, then count how many
// unshared parts are in the split `base`.
seps := 0
size := 0
if len(base)-common > 0 {
seps = 1
size = 2
}
// This range skips separators on the ends of the string.
for i in common+1..<len(base)-1 {
if _is_path_separator(base[i]) {
seps += 1
size += 3
}
}
// Handle the rest of the size calculations.
trailing := target[common:]
if len(trailing) > 0 {
// Account for leading separators on the target after cutting the common part.
// (i.e. base == `/home`, target == `/home/a`)
if _is_path_separator(trailing[0]) {
trailing = trailing[1:]
}
size += len(trailing)
if seps > 0 {
size += 1
}
}
if trailing == "." {
trailing = ""
size -= 2
}
// Build the string.
buf := make([]u8, size, allocator) or_return
n := 0
if seps > 0 {
buf[0] = '.'
buf[1] = '.'
n = 2
}
for _ in 1..<seps {
buf[n] = _Path_Separator
buf[n+1] = '.'
buf[n+2] = '.'
n += 3
}
if len(trailing) > 0 {
if seps > 0 {
buf[n] = _Path_Separator
n += 1
}
copy(buf[n:], trailing)
}
path = string(buf)
return
}
/*
Split a path into a directory hierarchy and a filename.
For example, `split_path("/home/foo/bar.tar.gz")` will return `"/home/foo"` and `"bar.tar.gz"`.
*/
@(require_results)
split_path :: proc(path: string) -> (dir, filename: string) {
return _split_path(path)
}
/*
Join all `elems` with the system's path separator and normalize the result.
*Allocates Using Provided Allocator*
For example, `join_path({"/home", "foo", "bar.txt"})` will result in `"/home/foo/bar.txt"`.
*/
@(require_results)
join_path :: proc(elems: []string, allocator: runtime.Allocator) -> (joined: string, err: Error) {
for e, i in elems {
if e != "" {
TEMP_ALLOCATOR_GUARD()
p := strings.join(elems[i:], Path_Separator_String, temp_allocator()) or_return
return clean_path(p, allocator)
}
}
return "", nil
}
/*
Split a filename from its extension.
This procedure splits on the last separator.
If the filename begins with a separator, such as `".readme.txt"`, the separator
will be included in the filename, resulting in `".readme"` and `"txt"`.
For example, `split_filename("foo.tar.gz")` will return `"foo.tar"` and `"gz"`.
*/
@(require_results)
split_filename :: proc(filename: string) -> (base, ext: string) {
i := strings.last_index_byte(filename, '.')
if i <= 0 {
return filename, ""
}
return filename[:i], filename[i+1:]
}
/*
Split a filename from its extension.
This procedure splits on the first separator.
If the filename begins with a separator, such as `".readme.txt.gz"`, the separator
will be included in the filename, resulting in `".readme"` and `"txt.gz"`.
For example, `split_filename_all("foo.tar.gz")` will return `"foo"` and `"tar.gz"`.
*/
@(require_results)
split_filename_all :: proc(filename: string) -> (base, ext: string) {
i := strings.index_byte(filename, '.')
if i == 0 {
j := strings.index_byte(filename[1:], '.')
if j != -1 {
j += 1
}
i = j
}
if i == -1 {
return filename, ""
}
return filename[:i], filename[i+1:]
}
/*
Join `base` and `ext` with the system's filename extension separator.
*Allocates Using Provided Allocator*
For example, `join_filename("foo", "tar.gz")` will result in `"foo.tar.gz"`.
*/
@(require_results)
join_filename :: proc(base: string, ext: string, allocator: runtime.Allocator) -> (joined: string, err: Error) {
if len(base) == 0 {
return strings.clone(ext, allocator)
} else if len(ext) == 0 {
return strings.clone(base, allocator)
}
buf := make([]u8, len(base) + 1 + len(ext), allocator) or_return
copy(buf, base)
buf[len(base)] = '.'
copy(buf[1+len(base):], ext)
return string(buf), nil
}
/*
Split a string that is separated by a system-specific separator, typically used
for environment variables specifying multiple directories.
*Allocates Using Provided Allocator*
For example, there is the "PATH" environment variable on POSIX systems which
this procedure can split into separate entries.
*/
@(require_results)
split_path_list :: proc(path: string, allocator: runtime.Allocator) -> (list: []string, err: Error) {
if path == "" {
return nil, nil
}
start: int
quote: bool
start, quote = 0, false
count := 0
for i := 0; i < len(path); i += 1 {
c := path[i]
switch {
case c == '"':
quote = !quote
case c == Path_List_Separator && !quote:
count += 1
}
}
start, quote = 0, false
list = make([]string, count + 1, allocator) or_return
index := 0
for i := 0; i < len(path); i += 1 {
c := path[i]
switch {
case c == '"':
quote = !quote
case c == Path_List_Separator && !quote:
list[index] = path[start:i]
index += 1
start = i + 1
}
}
assert(index == count)
list[index] = path[start:]
for s0, i in list {
s, new := strings.replace_all(s0, `"`, ``, allocator)
if !new {
s = strings.clone(s, allocator) or_return
}
list[i] = s
}
return list, nil
}

2
core/os/os2/path_linux.odin
View File

@@ -14,7 +14,7 @@ _Path_List_Separator :: ':'
_OPENDIR_FLAGS : linux.Open_Flags : {.NONBLOCK, .DIRECTORY, .LARGEFILE, .CLOEXEC}
_is_path_separator :: proc(c: byte) -> bool {
return c == '/'
return c == _Path_Separator
}
_mkdir :: proc(path: string, perm: int) -> Error {

5
core/os/os2/path_posix.odin
View File

@@ -3,7 +3,6 @@
package os2
import "base:runtime"
import "core:path/filepath"
import "core:sys/posix"
@@ -35,11 +34,11 @@ _mkdir_all :: proc(path: string, perm: int) -> Error {
return .Exist
}
clean_path := filepath.clean(path, temp_allocator())
clean_path := clean_path(path, temp_allocator()) or_return
return internal_mkdir_all(clean_path, perm)
internal_mkdir_all :: proc(path: string, perm: int) -> Error {
dir, file := filepath.split(path)
dir, file := split_path(path)
if file != path && dir != "/" {
if len(dir) > 1 && dir[len(dir) - 1] == '/' {
dir = dir[:len(dir) - 1]

78
core/os/os2/path_posixfs.odin Normal file
View File

@@ -0,0 +1,78 @@
#+private
#+build linux, darwin, netbsd, freebsd, openbsd, wasi
package os2
// This implementation is for all systems that have POSIX-compliant filesystem paths.
import "base:runtime"
import "core:strings"
import "core:sys/posix"
_are_paths_identical :: proc(a, b: string) -> (identical: bool) {
return a == b
}
_clean_path_handle_start :: proc(path: string, buffer: []u8) -> (rooted: bool, start: int) {
// Preserve rooted paths.
if _is_path_separator(path[0]) {
rooted = true
buffer[0] = _Path_Separator
start = 1
}
return
}
_is_absolute_path :: proc(path: string) -> bool {
return len(path) > 0 && _is_path_separator(path[0])
}
_get_absolute_path :: proc(path: string, allocator: runtime.Allocator) -> (absolute_path: string, err: Error) {
rel := path
if rel == "" {
rel = "."
}
TEMP_ALLOCATOR_GUARD()
rel_cstr := strings.clone_to_cstring(rel, temp_allocator())
path_ptr := posix.realpath(rel_cstr, nil)
if path_ptr == nil {
return "", Platform_Error(posix.errno())
}
defer posix.free(path_ptr)
path_str := strings.clone(string(path_ptr), allocator)
return path_str, nil
}
_get_relative_path_handle_start :: proc(base, target: string) -> bool {
base_rooted := len(base) > 0 && _is_path_separator(base[0])
target_rooted := len(target) > 0 && _is_path_separator(target[0])
return base_rooted == target_rooted
}
_get_common_path_len :: proc(base, target: string) -> int {
i := 0
end := min(len(base), len(target))
for j in 0..=end {
if j == end || _is_path_separator(base[j]) {
if base[i:j] == target[i:j] {
i = j
} else {
break
}
}
}
return i
}
_split_path :: proc(path: string) -> (dir, file: string) {
i := len(path) - 1
for i >= 0 && !_is_path_separator(path[i]) {
i -= 1
}
if i == 0 {
return path[:i+1], path[i+1:]
} else if i > 0 {
return path[:i], path[i+1:]
}
return "", path
}

5
core/os/os2/path_wasi.odin
View File

@@ -3,7 +3,6 @@ package os2
import "base:runtime"
import "core:path/filepath"
import "core:sync"
import "core:sys/wasm/wasi"
@@ -35,11 +34,11 @@ _mkdir_all :: proc(path: string, perm: int) -> Error {
return .Exist
}
clean_path := filepath.clean(path, temp_allocator())
clean_path := clean_path(path, temp_allocator())
return internal_mkdir_all(clean_path)
internal_mkdir_all :: proc(path: string) -> Error {
dir, file := filepath.split(path)
dir, file := split_path(path)
if file != path && dir != "/" {
if len(dir) > 1 && dir[len(dir) - 1] == '/' {
dir = dir[:len(dir) - 1]

95
core/os/os2/path_windows.odin
View File

@@ -1,8 +1,9 @@
#+private
package os2
import win32 "core:sys/windows"
import "base:runtime"
import "core:strings"
import win32 "core:sys/windows"
_Path_Separator :: '\\'
_Path_Separator_String :: "\\"
@@ -217,7 +218,7 @@ _fix_long_path_internal :: proc(path: string) -> string {
return path
}
if !_is_abs(path) { // relative path
if !_is_absolute_path(path) { // relative path
return path
}
@@ -257,3 +258,93 @@ _fix_long_path_internal :: proc(path: string) -> string {
return string(path_buf[:w])
}
_are_paths_identical :: strings.equal_fold
_clean_path_handle_start :: proc(path: string, buffer: []u8) -> (rooted: bool, start: int) {
// Preserve rooted paths.
start = _volume_name_len(path)
if start > 0 {
rooted = true
if len(path) > start && _is_path_separator(path[start]) {
// Take `C:` to `C:\`.
start += 1
}
copy(buffer, path[:start])
}
return
}
_is_absolute_path :: proc(path: string) -> bool {
if _is_reserved_name(path) {
return true
}
l := _volume_name_len(path)
if l == 0 {
return false
}
path := path
path = path[l:]
if path == "" {
return false
}
return _is_path_separator(path[0])
}
_get_absolute_path :: proc(path: string, allocator: runtime.Allocator) -> (absolute_path: string, err: Error) {
rel := path
if rel == "" {
rel = "."
}
TEMP_ALLOCATOR_GUARD()
rel_utf16 := win32.utf8_to_utf16(rel, temp_allocator())
n := win32.GetFullPathNameW(raw_data(rel_utf16), 0, nil, nil)
if n == 0 {
return "", Platform_Error(win32.GetLastError())
}
buf := make([]u16, n, temp_allocator()) or_return
n = win32.GetFullPathNameW(raw_data(rel_utf16), u32(n), raw_data(buf), nil)
if n == 0 {
return "", Platform_Error(win32.GetLastError())
}
return win32.utf16_to_utf8(buf, allocator)
}
_get_relative_path_handle_start :: proc(base, target: string) -> bool {
base_root := base[:_volume_name_len(base)]
target_root := target[:_volume_name_len(target)]
return strings.equal_fold(base_root, target_root)
}
_get_common_path_len :: proc(base, target: string) -> int {
i := 0
end := min(len(base), len(target))
for j in 0..=end {
if j == end || _is_path_separator(base[j]) {
if strings.equal_fold(base[i:j], target[i:j]) {
i = j
} else {
break
}
}
}
return i
}
_split_path :: proc(path: string) -> (dir, file: string) {
vol_len := _volume_name_len(path)
i := len(path) - 1
for i >= vol_len && !_is_path_separator(path[i]) {
i -= 1
}
if i == vol_len {
return path[:i+1], path[i+1:]
} else if i > vol_len {
return path[:i], path[i+1:]
}
return "", path
}

5
core/os/os2/process_linux.odin
View File

@@ -10,7 +10,6 @@ import "core:slice"
import "core:strings"
import "core:strconv"
import "core:sys/linux"
import "core:path/filepath"
PIDFD_UNASSIGNED :: ~uintptr(0)
@@ -205,7 +204,7 @@ _process_info_by_pid :: proc(pid: int, selection: Process_Info_Fields, allocator
info.executable_path = strings.clone(cmdline[:terminator], allocator) or_return
info.fields += {.Executable_Path}
} else if cwd_err == nil {
info.executable_path = filepath.join({ cwd, cmdline[:terminator] }, allocator) or_return
info.executable_path = join_path({ cwd, cmdline[:terminator] }, allocator) or_return
info.fields += {.Executable_Path}
} else {
break cmdline_if
@@ -407,7 +406,7 @@ _process_start :: proc(desc: Process_Desc) -> (process: Process, err: Error) {
executable_name := desc.command[0]
if strings.index_byte(executable_name, '/') < 0 {
path_env := get_env("PATH", temp_allocator())
path_dirs := filepath.split_list(path_env, temp_allocator()) or_return
path_dirs := split_path_list(path_env, temp_allocator()) or_return
exe_builder := strings.builder_make(temp_allocator()) or_return

3
core/os/os2/process_posix.odin
View File

@@ -6,7 +6,6 @@ import "base:runtime"
import "core:time"
import "core:strings"
import "core:path/filepath"
import kq "core:sys/kqueue"
import "core:sys/posix"
@@ -62,7 +61,7 @@ _process_start :: proc(desc: Process_Desc) -> (process: Process, err: Error) {
exe_name := desc.command[0]
if strings.index_byte(exe_name, '/') < 0 {
path_env := get_env("PATH", temp_allocator())
path_dirs := filepath.split_list(path_env, temp_allocator())
path_dirs := split_path_list(path_env, temp_allocator()) or_return
found: bool
for dir in path_dirs {

3
core/os/os2/stat.odin
View File

@@ -1,7 +1,6 @@
package os2
import "base:runtime"
import "core:path/filepath"
import "core:strings"
import "core:time"
@@ -25,7 +24,7 @@ File_Info :: struct {
file_info_clone :: proc(fi: File_Info, allocator: runtime.Allocator) -> (cloned: File_Info, err: runtime.Allocator_Error) {
cloned = fi
cloned.fullpath = strings.clone(fi.fullpath, allocator) or_return
cloned.name = filepath.base(cloned.fullpath)
_, cloned.name = split_path(cloned.fullpath)
return
}

3
core/os/os2/stat_linux.odin
View File

@@ -4,7 +4,6 @@ package os2
import "core:time"
import "base:runtime"
import "core:sys/linux"
import "core:path/filepath"
_fstat :: proc(f: ^File, allocator: runtime.Allocator) -> (File_Info, Error) {
impl := (^File_Impl)(f.impl)
@@ -42,7 +41,7 @@ _fstat_internal :: proc(fd: linux.Fd, allocator: runtime.Allocator) -> (fi: File
creation_time = time.Time{i64(s.ctime.time_sec) * i64(time.Second) + i64(s.ctime.time_nsec)}, // regular stat does not provide this
}
fi.creation_time = fi.modification_time
fi.name = filepath.base(fi.fullpath)
_, fi.name = split_path(fi.fullpath)
return
}

5
core/os/os2/stat_posix.odin
View File

@@ -4,13 +4,12 @@ package os2
import "base:runtime"
import "core:path/filepath"
import "core:sys/posix"
import "core:time"
internal_stat :: proc(stat: posix.stat_t, fullpath: string) -> (fi: File_Info) {
fi.fullpath = fullpath
fi.name = filepath.base(fi.fullpath)
_, fi.name = split_path(fi.fullpath)
fi.inode = u128(stat.st_ino)
fi.size = i64(stat.st_size)
@@ -104,7 +103,7 @@ _lstat :: proc(name: string, allocator: runtime.Allocator) -> (fi: File_Info, er
// NOTE: This might not be correct when given "/symlink/foo.txt",
// you would want that to resolve "/symlink", but not resolve "foo.txt".
fullpath := filepath.clean(name, temp_allocator())
fullpath := clean_path(name, temp_allocator()) or_return
assert(len(fullpath) > 0)
switch {
case fullpath[0] == '/':

3
core/os/os2/stat_wasi.odin
View File

@@ -3,13 +3,12 @@ package os2
import "base:runtime"
import "core:path/filepath"
import "core:sys/wasm/wasi"
import "core:time"
internal_stat :: proc(stat: wasi.filestat_t, fullpath: string) -> (fi: File_Info) {
fi.fullpath = fullpath
fi.name = filepath.base(fi.fullpath)
_, fi.name = split_path(fi.fullpath)
fi.inode = u128(stat.ino)
fi.size = i64(stat.size)

70
core/os/os2/stat_windows.odin
View File

@@ -315,57 +315,37 @@ _is_UNC :: proc(path: string) -> bool {
}
_volume_name_len :: proc(path: string) -> int {
if ODIN_OS == .Windows {
if len(path) < 2 {
return 0
}
c := path[0]
if path[1] == ':' {
switch c {
case 'a'..='z', 'A'..='Z':
return 2
}
if len(path) < 2 {
return 0
}
c := path[0]
if path[1] == ':' {
switch c {
case 'a'..='z', 'A'..='Z':
return 2
}
}
// URL: https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx
if l := len(path); l >= 5 && _is_path_separator(path[0]) && _is_path_separator(path[1]) &&
!_is_path_separator(path[2]) && path[2] != '.' {
for n := 3; n < l-1; n += 1 {
if _is_path_separator(path[n]) {
n += 1
if !_is_path_separator(path[n]) {
if path[n] == '.' {
break
}
// URL: https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx
if l := len(path); l >= 5 && _is_path_separator(path[0]) && _is_path_separator(path[1]) &&
!_is_path_separator(path[2]) && path[2] != '.' {
for n := 3; n < l-1; n += 1 {
if _is_path_separator(path[n]) {
n += 1
if !_is_path_separator(path[n]) {
if path[n] == '.' {
break
}
for ; n < l; n += 1 {
if _is_path_separator(path[n]) {
break
}
}
return n
}
break
for ; n < l; n += 1 {
if _is_path_separator(path[n]) {
break
}
}
return n
}
break
}
}
return 0
}
_is_abs :: proc(path: string) -> bool {
if _is_reserved_name(path) {
return true
}
l := _volume_name_len(path)
if l == 0 {
return false
}
path := path
path = path[l:]
if path == "" {
return false
}
return is_path_separator(path[0])
}

2
core/os/os_openbsd.odin
View File

@@ -343,7 +343,7 @@ AT_REMOVEDIR :: 0x08
@(default_calling_convention="c")
foreign libc {
@(link_name="__error") __error :: proc() -> ^c.int ---
@(link_name="__errno") __error :: proc() -> ^c.int ---
@(link_name="fork") _unix_fork :: proc() -> pid_t ---
@(link_name="getthrid") _unix_getthrid :: proc() -> int ---

2
core/simd/x86/ssse3.odin
View File

@@ -21,7 +21,7 @@ _mm_abs_epi32 :: #force_inline proc "c" (a: __m128i) -> __m128i {
_mm_shuffle_epi8 :: #force_inline proc "c" (a, b: __m128i) -> __m128i {
return transmute(__m128i)pshufb128(transmute(u8x16)a, transmute(u8x16)b)
}
@(require_results, enable_target_feature="ssse3")
@(require_results, enable_target_feature="sse2,ssse3")
_mm_alignr_epi8 :: #force_inline proc "c" (a, b: __m128i, $IMM8: u32) -> __m128i {
shift :: IMM8

41
core/sys/freebsd/syscalls.odin
View File

@@ -21,6 +21,7 @@ SYS_close : uintptr : 6
SYS_getpid : uintptr : 20
SYS_recvfrom : uintptr : 29
SYS_accept : uintptr : 30
SYS_getpeername: uintptr : 31
SYS_getsockname: uintptr : 32
SYS_fcntl : uintptr : 92
SYS_fsync : uintptr : 95
@@ -202,24 +203,36 @@ accept_nil :: proc "contextless" (s: Fd) -> (Fd, Errno) {
accept :: proc { accept_T, accept_nil }
getsockname_or_peername :: proc "contextless" (s: Fd, sockaddr: ^$T, is_peer: bool) -> Errno {
// sockaddr must contain a valid pointer, or this will segfault because
// we're telling the syscall that there's memory available to write to.
addrlen: socklen_t = size_of(T)
result, ok := intrinsics.syscall_bsd(
is_peer ? SYS_getpeername : SYS_getsockname,
cast(uintptr)s,
cast(uintptr)sockaddr,
cast(uintptr)&addrlen)
if !ok {
return cast(Errno)result
}
return nil
}
// Get name of connected peer
//
// The getpeername() system call appeared in 4.2BSD.
getpeername :: proc "contextless" (s: Fd, sockaddr: ^$T) -> Errno {
return getsockname_or_peername(s, sockaddr, true)
}
// Get socket name.
//
// The getsockname() system call appeared in 4.2BSD.
getsockname :: proc "contextless" (s: Fd, sockaddr: ^$T) -> Errno {
// sockaddr must contain a valid pointer, or this will segfault because
// we're telling the syscall that there's memory available to write to.
addrlen: socklen_t = size_of(T)
result, ok := intrinsics.syscall_bsd(SYS_getsockname,
cast(uintptr)s,
cast(uintptr)sockaddr,
cast(uintptr)&addrlen)
if !ok {
return cast(Errno)result
}
return nil
return getsockname_or_peername(s, sockaddr, false)
}
// Synchronize changes to a file.

2
core/sys/info/cpu_intel.odin
View File

@@ -23,6 +23,7 @@ CPU_Feature :: enum u64 {
popcnt, // Hamming weight instruction POPCNT.
rdrand, // RDRAND instruction (on-chip random number generator)
rdseed, // RDSEED instruction (on-chip random number generator)
sha, // SHA Extensions (SHA-1, SHA-224, SHA-256)
sse2, // Streaming SIMD extension 2 (always available on amd64)
sse3, // Streaming SIMD extension 3
ssse3, // Supplemental streaming SIMD extension 3
@@ -115,6 +116,7 @@ init_cpu_features :: proc "c" () {
_, ebx7, ecx7, edx7 := cpuid(7, 0)
try_set(&set, .bmi1, 3, ebx7)
try_set(&set, .sha, 29, ebx7)
if os_supports_avx {
try_set(&set, .avx2, 5, ebx7)
}

36
core/sys/linux/bits.odin
View File

@@ -1,5 +1,10 @@
package linux
import "base:intrinsics"
@(private)
log2 :: intrinsics.constant_log2
/*
Represents an error returned by most of syscalls
@@ -1839,22 +1844,23 @@ EPoll_Flags_Bits :: enum {
}
EPoll_Event_Kind :: enum u32 {
IN = 0x001,
PRI = 0x002,
OUT = 0x004,
RDNORM = 0x040,
RDBAND = 0x080,
WRNORM = 0x100,
WRBAND = 0x200,
MSG = 0x400,
ERR = 0x008,
HUP = 0x010,
RDHUP = 0x2000,
EXCLUSIVE = 1<<28,
WAKEUP = 1<<29,
ONESHOT = 1<<30,
ET = 1<<31,
IN = log2(0x001),
PRI = log2(0x002),
OUT = log2(0x004),
RDNORM = log2(0x040),
RDBAND = log2(0x080),
WRNORM = log2(0x100),
WRBAND = log2(0x200),
MSG = log2(0x400),
ERR = log2(0x008),
HUP = log2(0x010),
RDHUP = log2(0x2000),
EXCLUSIVE = log2(1<<28),
WAKEUP = log2(1<<29),
ONESHOT = log2(1<<30),
ET = log2(1<<31),
}
EPoll_Event_Set :: bit_set[EPoll_Event_Kind; u32]
EPoll_Ctl_Opcode :: enum i32 {
ADD = 1,

2
core/sys/linux/types.odin
View File

@@ -1450,7 +1450,7 @@ EPoll_Data :: struct #raw_union {
}
EPoll_Event :: struct #packed {
events: EPoll_Event_Kind,
events: EPoll_Event_Set,
data: EPoll_Data,
}

2
core/sys/posix/signal.odin
View File

@@ -565,7 +565,7 @@ when ODIN_OS == .Darwin {
SS_ONSTACK :: 0x0001
SS_DISABLE :: 0x0004
when ODIN_ARCH == .amd64 || ODIN_ARCH == .arm32 {
when ODIN_ARCH == .arm64 || ODIN_ARCH == .arm32 {
MINSIGSTKSZ :: 1024 * 4
} else when ODIN_ARCH == .amd64 || ODIN_ARCH == .i386 {
MINSIGSTKSZ :: 512 * 4

28
core/sys/wasm/js/events.odin
View File

@@ -346,13 +346,13 @@ add_event_listener :: proc(id: string, kind: Event_Kind, user_data: rawptr, call
return _add_event_listener(id, event_kind_string[kind], kind, user_data, callback, use_capture)
}
remove_event_listener :: proc(id: string, kind: Event_Kind, user_data: rawptr, callback: proc(e: Event)) -> bool {
remove_event_listener :: proc(id: string, kind: Event_Kind, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
@(default_calling_convention="contextless")
foreign dom_lib {
@(link_name="remove_event_listener")
_remove_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc "odin" (Event)) -> bool ---
_remove_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc "odin" (Event), use_capture: bool) -> bool ---
}
return _remove_event_listener(id, event_kind_string[kind], user_data, callback)
return _remove_event_listener(id, event_kind_string[kind], user_data, callback, use_capture)
}
add_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
@@ -364,20 +364,26 @@ add_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback:
return _add_window_event_listener(event_kind_string[kind], kind, user_data, callback, use_capture)
}
remove_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback: proc(e: Event)) -> bool {
remove_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
@(default_calling_convention="contextless")
foreign dom_lib {
@(link_name="remove_window_event_listener")
_remove_window_event_listener :: proc(name: string, user_data: rawptr, callback: proc "odin" (Event)) -> bool ---
_remove_window_event_listener :: proc(name: string, user_data: rawptr, callback: proc "odin" (Event), use_capture: bool) -> bool ---
}
return _remove_window_event_listener(event_kind_string[kind], user_data, callback)
return _remove_window_event_listener(event_kind_string[kind], user_data, callback, use_capture)
}
remove_event_listener_from_event :: proc(e: Event) -> bool {
from_use_capture_false: bool
from_use_capture_true: bool
if e.id == "" {
return remove_window_event_listener(e.kind, e.user_data, e.callback)
from_use_capture_false = remove_window_event_listener(e.kind, e.user_data, e.callback, false)
from_use_capture_true = remove_window_event_listener(e.kind, e.user_data, e.callback, true)
} else {
from_use_capture_false = remove_event_listener(e.id, e.kind, e.user_data, e.callback, false)
from_use_capture_true = remove_event_listener(e.id, e.kind, e.user_data, e.callback, true)
}
return remove_event_listener(e.id, e.kind, e.user_data, e.callback)
return from_use_capture_false || from_use_capture_true
}
add_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
@@ -388,13 +394,13 @@ add_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, c
}
return _add_event_listener(id, name, .Custom, user_data, callback, use_capture)
}
remove_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc(e: Event)) -> bool {
remove_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
@(default_calling_convention="contextless")
foreign dom_lib {
@(link_name="remove_event_listener")
_remove_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc "odin" (Event)) -> bool ---
_remove_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc "odin" (Event), use_capture: bool) -> bool ---
}
return _remove_event_listener(id, name, user_data, callback)
return _remove_event_listener(id, name, user_data, callback, use_capture)
}
get_gamepad_state :: proc "contextless" (index: int, s: ^Gamepad_State) -> bool {

8
core/sys/wasm/js/events_all_targets.odin
View File

@@ -263,7 +263,7 @@ add_event_listener :: proc(id: string, kind: Event_Kind, user_data: rawptr, call
panic("vendor:wasm/js not supported on non JS targets")
}
remove_event_listener :: proc(id: string, kind: Event_Kind, user_data: rawptr, callback: proc(e: Event)) -> bool {
remove_event_listener :: proc(id: string, kind: Event_Kind, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
panic("vendor:wasm/js not supported on non JS targets")
}
@@ -271,7 +271,7 @@ add_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback:
panic("vendor:wasm/js not supported on non JS targets")
}
remove_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback: proc(e: Event)) -> bool {
remove_window_event_listener :: proc(kind: Event_Kind, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
panic("vendor:wasm/js not supported on non JS targets")
}
@@ -282,6 +282,6 @@ remove_event_listener_from_event :: proc(e: Event) -> bool {
add_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
panic("vendor:wasm/js not supported on non JS targets")
}
remove_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc(e: Event)) -> bool {
remove_custom_event_listener :: proc(id: string, name: string, user_data: rawptr, callback: proc(e: Event), use_capture := false) -> bool {
panic("vendor:wasm/js not supported on non JS targets")
}
}

57
core/sys/wasm/js/odin.js
View File

@@ -17,7 +17,7 @@ class WasmMemoryInterface {
constructor() {
this.memory = null;
this.exports = null;
this.listenerMap = {};
this.listenerMap = new Map();
// Size (in bytes) of the integer type, should be 4 on `js_wasm32` and 8 on `js_wasm64p32`
this.intSize = 4;
@@ -1325,18 +1325,20 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
} else if (!line.includes("\n")) {
currentLine[isError] = currentLine[isError].concat(line);
} else {
let lines = line.split("\n");
let lines = line.trimEnd().split("\n");
let printLast = lines.length > 1 && line.endsWith("\n");
println(currentLine[isError].concat(lines[0]));
currentLine[isError] = "";
for (let i = 1; i < lines.length-1; i++) {
println(lines[i]);
}
let last = lines[lines.length-1];
if (printLast) {
println(last);
} else {
currentLine[isError] = last;
if (lines.length > 1) {
let last = lines[lines.length-1];
if (printLast) {
println(last);
} else {
currentLine[isError] = last;
}
}
}
@@ -1403,6 +1405,10 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
info.scrollTop = info.scrollHeight;
};
const listener_key = (id, name, data, callback, useCapture) => {
return `${id}-${name}-data:${data}-callback:${callback}-useCapture:${useCapture}`;
};
let webglContext = new WebGLInterface(wasmMemoryInterface);
const env = {};
@@ -1658,6 +1664,10 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
if (element == undefined) {
return false;
}
let key = listener_key(id, name, data, callback, !!use_capture);
if (wasmMemoryInterface.listenerMap.has(key)) {
return false;
}
let listener = (e) => {
let event_data = {};
@@ -1668,7 +1678,7 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
onEventReceived(event_data, data, callback);
};
wasmMemoryInterface.listenerMap[{data: data, callback: callback}] = listener;
wasmMemoryInterface.listenerMap.set(key, listener);
element.addEventListener(name, listener, !!use_capture);
return true;
},
@@ -1676,6 +1686,11 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
add_window_event_listener: (name_ptr, name_len, name_code, data, callback, use_capture) => {
let name = wasmMemoryInterface.loadString(name_ptr, name_len);
let element = window;
let key = listener_key('window', name, data, callback, !!use_capture);
if (wasmMemoryInterface.listenerMap.has(key)) {
return false;
}
let listener = (e) => {
let event_data = {};
event_data.id_ptr = 0;
@@ -1685,12 +1700,12 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
onEventReceived(event_data, data, callback);
};
wasmMemoryInterface.listenerMap[{data: data, callback: callback}] = listener;
wasmMemoryInterface.listenerMap.set(key, listener);
element.addEventListener(name, listener, !!use_capture);
return true;
},
remove_event_listener: (id_ptr, id_len, name_ptr, name_len, data, callback) => {
remove_event_listener: (id_ptr, id_len, name_ptr, name_len, data, callback, use_capture) => {
let id = wasmMemoryInterface.loadString(id_ptr, id_len);
let name = wasmMemoryInterface.loadString(name_ptr, name_len);
let element = getElement(id);
@@ -1698,24 +1713,28 @@ function odinSetupDefaultImports(wasmMemoryInterface, consoleElement, memory) {
return false;
}
let listener = wasmMemoryInterface.listenerMap[{data: data, callback: callback}];
if (listener == undefined) {
let key = listener_key(id, name, data, callback, !!use_capture);
let listener = wasmMemoryInterface.listenerMap.get(key);
if (listener === undefined) {
return false;
}
element.removeEventListener(name, listener);
wasmMemoryInterface.listenerMap.delete(key);
element.removeEventListener(name, listener, !!use_capture);
return true;
},
remove_window_event_listener: (name_ptr, name_len, data, callback) => {
remove_window_event_listener: (name_ptr, name_len, data, callback, use_capture) => {
let name = wasmMemoryInterface.loadString(name_ptr, name_len);
let element = window;
let key = {data: data, callback: callback};
let listener = wasmMemoryInterface.listenerMap[key];
if (!listener) {
let key = listener_key('window', name, data, callback, !!use_capture);
let listener = wasmMemoryInterface.listenerMap.get(key);
if (listener === undefined) {
return false;
}
wasmMemoryInterface.listenerMap[key] = undefined;
wasmMemoryInterface.listenerMap.delete(key);
element.removeEventListener(name, listener);
element.removeEventListener(name, listener, !!use_capture);
return true;
},

6
examples/all/all_main.odin
View File

@@ -26,12 +26,14 @@ import topological_sort "core:container/topological_sort"
import crypto "core:crypto"
import aead "core:crypto/aead"
import aegis "core:crypto/aegis"
import aes "core:crypto/aes"
import blake2b "core:crypto/blake2b"
import blake2s "core:crypto/blake2s"
import chacha20 "core:crypto/chacha20"
import chacha20poly1305 "core:crypto/chacha20poly1305"
import crypto_hash "core:crypto/hash"
import deoxysii "core:crypto/deoxysii"
import ed25519 "core:crypto/ed25519"
import hkdf "core:crypto/hkdf"
import hmac "core:crypto/hmac"
@@ -48,6 +50,7 @@ import shake "core:crypto/shake"
import sm3 "core:crypto/sm3"
import tuplehash "core:crypto/tuplehash"
import x25519 "core:crypto/x25519"
import x448 "core:crypto/x448"
import pe "core:debug/pe"
import trace "core:debug/trace"
@@ -169,11 +172,13 @@ _ :: topological_sort
_ :: crypto
_ :: crypto_hash
_ :: aead
_ :: aegis
_ :: aes
_ :: blake2b
_ :: blake2s
_ :: chacha20
_ :: chacha20poly1305
_ :: deoxysii
_ :: ed25519
_ :: hmac
_ :: hkdf
@@ -190,6 +195,7 @@ _ :: shake
_ :: sm3
_ :: tuplehash
_ :: x25519
_ :: x448
_ :: pe
_ :: trace
_ :: dynlib

201
src/build_settings.cpp
View File

@@ -109,6 +109,12 @@ gb_global String target_arch_names[TargetArch_COUNT] = {
str_lit("riscv64"),
};
#if defined(GB_SYSTEM_WINDOWS)
#include <llvm-c/Config/llvm-config.h>
#else
#include <llvm/Config/llvm-config.h>
#endif
#include "build_settings_microarch.cpp"
gb_global String target_endian_names[TargetEndian_COUNT] = {
@@ -165,6 +171,7 @@ struct TargetMetrics {
enum Subtarget : u32 {
Subtarget_Default,
Subtarget_iOS,
Subtarget_Android,
Subtarget_COUNT,
};
@@ -172,6 +179,7 @@ enum Subtarget : u32 {
gb_global String subtarget_strings[Subtarget_COUNT] = {
str_lit(""),
str_lit("ios"),
str_lit("android"),
};
@@ -198,20 +206,25 @@ enum BuildModeKind {
BuildMode_COUNT,
};
enum CommandKind : u32 {
enum CommandKind : u64 {
Command_run = 1<<0,
Command_build = 1<<1,
Command_check = 1<<3,
Command_doc = 1<<5,
Command_version = 1<<6,
Command_test = 1<<7,
Command_check = 1<<2,
Command_doc = 1<<3,
Command_version = 1<<4,
Command_test = 1<<5,
Command_strip_semicolon = 1<<8,
Command_bug_report = 1<<9,
Command_strip_semicolon = 1<<6,
Command_bug_report = 1<<7,
Command_bundle_android = 1<<8,
Command_bundle_macos = 1<<9,
Command_bundle_ios = 1<<10,
Command_bundle_orca = 1<<11,
Command__does_check = Command_run|Command_build|Command_check|Command_doc|Command_test|Command_strip_semicolon,
Command__does_build = Command_run|Command_build|Command_test,
Command_all = ~(u32)0,
Command_all = ~(CommandKind)0,
};
gb_global char const *odin_command_strings[32] = {
@@ -222,6 +235,11 @@ gb_global char const *odin_command_strings[32] = {
"version",
"test",
"strip-semicolon",
"",
"bundle android",
"bundle macos",
"bundle ios",
"bundle orca",
};
@@ -521,6 +539,22 @@ struct BuildContext {
String minimum_os_version_string;
bool minimum_os_version_string_given;
int ODIN_ANDROID_API_LEVEL;
String ODIN_ANDROID_SDK;
String ODIN_ANDROID_NDK;
String ODIN_ANDROID_NDK_TOOLCHAIN;
String ODIN_ANDROID_NDK_TOOLCHAIN_LIB;
String ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL;
String ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT;
String ODIN_ANDROID_JAR_SIGNER;
String android_keystore;
String android_keystore_alias;
String android_manifest;
};
gb_global BuildContext build_context = {0};
@@ -544,12 +578,6 @@ gb_internal isize MAX_ERROR_COLLECTOR_COUNT(void) {
return build_context.max_error_count;
}
#if defined(GB_SYSTEM_WINDOWS)
#include <llvm-c/Config/llvm-config.h>
#else
#include <llvm/Config/llvm-config.h>
#endif
// NOTE: AMD64 targets had their alignment on 128 bit ints bumped from 8 to 16 (undocumented of course).
#if LLVM_VERSION_MAJOR >= 18
#define AMD64_MAX_ALIGNMENT 16
@@ -946,6 +974,14 @@ gb_internal bool is_arch_x86(void) {
gb_global String const WIN32_SEPARATOR_STRING = {cast(u8 *)"\\", 1};
gb_global String const NIX_SEPARATOR_STRING = {cast(u8 *)"/", 1};
gb_global String const SEPARATOR_STRING =
#if defined(GB_SYSTEM_WINDOWS)
WIN32_SEPARATOR_STRING;
#else
NIX_SEPARATOR_STRING;
#endif
gb_global String const WASM_MODULE_NAME_SEPARATOR = str_lit("..");
gb_internal String internal_odin_root_dir(void);
@@ -1461,6 +1497,103 @@ gb_internal bool has_ansi_terminal_colours(void) {
return build_context.has_ansi_terminal_colours && !json_errors();
}
gb_internal void init_android_values(bool with_sdk) {
auto *bc = &build_context;
{ // Android SDK/API Level
String default_level = str_lit("34");
if (!bc->minimum_os_version_string_given) {
bc->minimum_os_version_string = default_level;
}
BigInt level = {};
bool success = false;
big_int_from_string(&level, bc->minimum_os_version_string, &success);
if (!success) {
gb_printf_err("Warning: Invalid -minimum-os-version:%.*s for -subtarget:Android, defaulting to %.*s\n", LIT(bc->minimum_os_version_string), LIT(default_level));
bc->minimum_os_version_string = default_level;
big_int_from_string(&level, bc->minimum_os_version_string, &success);
GB_ASSERT(success);
}
i64 new_level = big_int_to_i64(&level);
if (new_level >= 21) {
bc->ODIN_ANDROID_API_LEVEL = cast(int)new_level;
} else {
gb_printf_err("Warning: Invalid -minimum-os-version:%.*s for -subtarget:Android, defaulting to %.*s\n", LIT(bc->minimum_os_version_string), LIT(default_level));
bc->ODIN_ANDROID_API_LEVEL = atoi(cast(char const *)default_level.text);
}
}
bc->ODIN_ANDROID_NDK = normalize_path(permanent_allocator(), make_string_c(gb_get_env("ODIN_ANDROID_NDK", permanent_allocator())), NIX_SEPARATOR_STRING);
bc->ODIN_ANDROID_NDK_TOOLCHAIN = normalize_path(permanent_allocator(), make_string_c(gb_get_env("ODIN_ANDROID_NDK_TOOLCHAIN", permanent_allocator())), NIX_SEPARATOR_STRING);
bc->ODIN_ANDROID_SDK = normalize_path(permanent_allocator(), make_string_c(gb_get_env("ODIN_ANDROID_SDK", permanent_allocator())), NIX_SEPARATOR_STRING);
#if defined(GB_SYSTEM_WINDOWS)
if (bc->ODIN_ANDROID_SDK.len == 0) {
bc->ODIN_ANDROID_SDK = normalize_path(permanent_allocator(),
path_to_fullpath(permanent_allocator(), str_lit("%LocalAppData%/Android/Sdk"), nullptr),
NIX_SEPARATOR_STRING);
}
#endif
if (bc->ODIN_ANDROID_NDK.len != 0 && bc->ODIN_ANDROID_NDK_TOOLCHAIN.len == 0) {
String arch = str_lit("x86_64");
#if defined (GB_CPU_ARM)
// TODO(bill): this is a complete guess
arch = str_lit("aarch64");
#endif
#if defined(GB_SYSTEM_WINDOWS)
bc->ODIN_ANDROID_NDK_TOOLCHAIN = concatenate4_strings(temporary_allocator(), bc->ODIN_ANDROID_NDK, str_lit("toolchains/llvm/prebuilt/"), str_lit("windows-"), arch);
#elif defined(GB_SYSTEM_OSX)
// TODO(bill): is this name even correct?
bc->ODIN_ANDROID_NDK_TOOLCHAIN = concatenate4_strings(temporary_allocator(), bc->ODIN_ANDROID_NDK, str_lit("toolchains/llvm/prebuilt/"), str_lit("darwin-"), arch);
#elif defined(GB_SYSTEM_LINUX)
bc->ODIN_ANDROID_NDK_TOOLCHAIN = concatenate4_strings(temporary_allocator(), bc->ODIN_ANDROID_NDK, str_lit("toolchains/llvm/prebuilt/"), str_lit("linux-"), arch);
#endif
bc->ODIN_ANDROID_NDK_TOOLCHAIN = normalize_path(permanent_allocator(), bc->ODIN_ANDROID_NDK_TOOLCHAIN, NIX_SEPARATOR_STRING);
}
if (bc->ODIN_ANDROID_NDK.len == 0 && !with_sdk) {
gb_printf_err("Error: ODIN_ANDROID_NDK not set");
gb_exit(1);
}
if (bc->ODIN_ANDROID_NDK_TOOLCHAIN.len == 0 && !with_sdk) {
gb_printf_err("Error: ODIN_ANDROID_NDK not set");
gb_exit(1);
}
bc->ODIN_ANDROID_NDK_TOOLCHAIN_LIB = concatenate_strings(permanent_allocator(), bc->ODIN_ANDROID_NDK_TOOLCHAIN, str_lit("sysroot/usr/lib/aarch64-linux-android/"));
char buf[32] = {};
gb_snprintf(buf, gb_size_of(buf), "%d/", bc->ODIN_ANDROID_API_LEVEL);
bc->ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL = concatenate_strings(permanent_allocator(), bc->ODIN_ANDROID_NDK_TOOLCHAIN_LIB, make_string_c(buf));
bc->ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT = concatenate_strings(permanent_allocator(), bc->ODIN_ANDROID_NDK_TOOLCHAIN, str_lit("sysroot/"));
bc->ODIN_ANDROID_JAR_SIGNER = normalize_path(permanent_allocator(), make_string_c(gb_get_env("ODIN_ANDROID_JAR_SIGNER", permanent_allocator())), NIX_SEPARATOR_STRING);
if (with_sdk) {
if (bc->ODIN_ANDROID_SDK.len == 0) {
gb_printf_err("Error: ODIN_ANDROID_SDK not set, which is required for -build-mode:executable for -subtarget:android");
gb_exit(1);
}
if (bc->ODIN_ANDROID_JAR_SIGNER.len == 0) {
gb_printf_err("Error: ODIN_ANDROID_JAR_SIGNER not set, which is required for -build-mode:executable for -subtarget:android");
gb_exit(1);
}
if (bc->android_keystore.len == 0) {
gb_printf_err("Error: -android-keystore:<string> has not been set\n");
gb_exit(1);
}
if (bc->android_keystore_alias.len == 0) {
gb_printf_err("Error: -android-keystore_alias:<string> has not been set\n");
gb_exit(1);
}
}
}
gb_internal bool has_asm_extension(String const &path) {
String ext = path_extension(path);
if (ext == ".asm") {
@@ -1652,6 +1785,15 @@ gb_internal void init_build_context(TargetMetrics *cross_target, Subtarget subta
default:
GB_PANIC("Unknown architecture for darwin");
}
} else if (metrics->os == TargetOs_linux && subtarget == Subtarget_Android) {
switch (metrics->arch) {
case TargetArch_arm64:
bc->metrics.target_triplet = str_lit("aarch64-none-linux-android");
bc->reloc_mode = RelocMode_PIC;
break;
default:
GB_PANIC("Unknown architecture for -subtarget:android");
}
}
if (bc->metrics.os == TargetOs_windows) {
@@ -1706,6 +1848,8 @@ gb_internal void init_build_context(TargetMetrics *cross_target, Subtarget subta
if (subtarget == Subtarget_Default) {
bc->metrics.target_triplet = concatenate_strings(permanent_allocator(), bc->metrics.target_triplet, bc->minimum_os_version_string);
}
} else if (selected_subtarget == Subtarget_Android) {
init_android_values(bc->build_mode == BuildMode_Executable);
}
if (!bc->custom_optimization_level) {
@@ -1749,6 +1893,30 @@ gb_internal void init_build_context(TargetMetrics *cross_target, Subtarget subta
if (bc->metrics.os == TargetOs_freestanding) {
bc->ODIN_DEFAULT_TO_NIL_ALLOCATOR = !bc->ODIN_DEFAULT_TO_PANIC_ALLOCATOR;
}
if (subtarget == Subtarget_Android) {
switch (build_context.build_mode) {
case BuildMode_DynamicLibrary:
case BuildMode_Object:
case BuildMode_Assembly:
case BuildMode_LLVM_IR:
break;
default:
case BuildMode_Executable:
case BuildMode_StaticLibrary:
if ((build_context.command_kind & Command__does_build) != 0) {
gb_printf_err("Unsupported -build-mode for -subtarget:android\n");
gb_printf_err("\tCurrently only supporting: \n");
// gb_printf_err("\t\texe\n");
gb_printf_err("\t\tshared\n");
gb_printf_err("\t\tobject\n");
gb_printf_err("\t\tassembly\n");
gb_printf_err("\t\tllvm-ir\n");
gb_exit(1);
}
break;
}
}
}
#if defined(GB_SYSTEM_WINDOWS)
@@ -1947,7 +2115,10 @@ gb_internal bool init_build_paths(String init_filename) {
output_extension = make_string(nullptr, 0);
String const single_file_extension = str_lit(".odin");
if (build_context.metrics.os == TargetOs_windows) {
if (selected_subtarget == Subtarget_Android) {
// NOTE(bill): It's always shared!
output_extension = STR_LIT("so");
} else if (build_context.metrics.os == TargetOs_windows) {
output_extension = STR_LIT("exe");
} else if (build_context.cross_compiling && selected_target_metrics->metrics == &target_essence_amd64) {
// Do nothing: we don't want the .bin extension

559
src/build_settings_microarch.cpp
View File

@@ -1,3 +1,561 @@
#ifndef LLVM_VERSION_MAJOR
#error "LLVM_VERSION_MAJOR is not defined!"
#endif
#if LLVM_VERSION_MAJOR >= 20
// Generated with the featuregen script in `misc/featuregen`
gb_global String target_microarch_list[TargetArch_COUNT] = {
// TargetArch_Invalid:
str_lit(""),
// TargetArch_amd64:
str_lit("alderlake,amdfam10,arrowlake,arrowlake-s,arrowlake_s,athlon,athlon-4,athlon-fx,athlon-mp,athlon-tbird,athlon-xp,athlon64,athlon64-sse3,atom,atom_sse4_2,atom_sse4_2_movbe,barcelona,bdver1,bdver2,bdver3,bdver4,bonnell,broadwell,btver1,btver2,c3,c3-2,cannonlake,cascadelake,clearwaterforest,cooperlake,core-avx-i,core-avx2,core2,core_2_duo_sse4_1,core_2_duo_ssse3,core_2nd_gen_avx,core_3rd_gen_avx,core_4th_gen_avx,core_4th_gen_avx_tsx,core_5th_gen_avx,core_5th_gen_avx_tsx,core_aes_pclmulqdq,core_i7_sse4_2,corei7,corei7-avx,diamondrapids,emeraldrapids,generic,geode,goldmont,goldmont-plus,goldmont_plus,gracemont,grandridge,graniterapids,graniterapids-d,graniterapids_d,haswell,i386,i486,i586,i686,icelake-client,icelake-server,icelake_client,icelake_server,ivybridge,k6,k6-2,k6-3,k8,k8-sse3,knl,knm,lakemont,lunarlake,meteorlake,mic_avx512,nehalem,nocona,opteron,opteron-sse3,pantherlake,penryn,pentium,pentium-m,pentium-mmx,pentium2,pentium3,pentium3m,pentium4,pentium4m,pentium_4,pentium_4_sse3,pentium_ii,pentium_iii,pentium_iii_no_xmm_regs,pentium_m,pentium_mmx,pentium_pro,pentiumpro,prescott,raptorlake,rocketlake,sandybridge,sapphirerapids,sierraforest,silvermont,skx,skylake,skylake-avx512,skylake_avx512,slm,tigerlake,tremont,westmere,winchip-c6,winchip2,x86-64,x86-64-v2,x86-64-v3,x86-64-v4,yonah,znver1,znver2,znver3,znver4,znver5"),
// TargetArch_i386:
str_lit("alderlake,amdfam10,arrowlake,arrowlake-s,arrowlake_s,athlon,athlon-4,athlon-fx,athlon-mp,athlon-tbird,athlon-xp,athlon64,athlon64-sse3,atom,atom_sse4_2,atom_sse4_2_movbe,barcelona,bdver1,bdver2,bdver3,bdver4,bonnell,broadwell,btver1,btver2,c3,c3-2,cannonlake,cascadelake,clearwaterforest,cooperlake,core-avx-i,core-avx2,core2,core_2_duo_sse4_1,core_2_duo_ssse3,core_2nd_gen_avx,core_3rd_gen_avx,core_4th_gen_avx,core_4th_gen_avx_tsx,core_5th_gen_avx,core_5th_gen_avx_tsx,core_aes_pclmulqdq,core_i7_sse4_2,corei7,corei7-avx,diamondrapids,emeraldrapids,generic,geode,goldmont,goldmont-plus,goldmont_plus,gracemont,grandridge,graniterapids,graniterapids-d,graniterapids_d,haswell,i386,i486,i586,i686,icelake-client,icelake-server,icelake_client,icelake_server,ivybridge,k6,k6-2,k6-3,k8,k8-sse3,knl,knm,lakemont,lunarlake,meteorlake,mic_avx512,nehalem,nocona,opteron,opteron-sse3,pantherlake,penryn,pentium,pentium-m,pentium-mmx,pentium2,pentium3,pentium3m,pentium4,pentium4m,pentium_4,pentium_4_sse3,pentium_ii,pentium_iii,pentium_iii_no_xmm_regs,pentium_m,pentium_mmx,pentium_pro,pentiumpro,prescott,raptorlake,rocketlake,sandybridge,sapphirerapids,sierraforest,silvermont,skx,skylake,skylake-avx512,skylake_avx512,slm,tigerlake,tremont,westmere,winchip-c6,winchip2,x86-64,x86-64-v2,x86-64-v3,x86-64-v4,yonah,znver1,znver2,znver3,znver4,znver5"),
// TargetArch_arm32:
str_lit("arm1020e,arm1020t,arm1022e,arm10e,arm10tdmi,arm1136j-s,arm1136jf-s,arm1156t2-s,arm1156t2f-s,arm1176jz-s,arm1176jzf-s,arm710t,arm720t,arm7tdmi,arm7tdmi-s,arm8,arm810,arm9,arm920,arm920t,arm922t,arm926ej-s,arm940t,arm946e-s,arm966e-s,arm968e-s,arm9e,arm9tdmi,cortex-a12,cortex-a15,cortex-a17,cortex-a32,cortex-a35,cortex-a5,cortex-a510,cortex-a53,cortex-a55,cortex-a57,cortex-a7,cortex-a710,cortex-a72,cortex-a73,cortex-a75,cortex-a76,cortex-a76ae,cortex-a77,cortex-a78,cortex-a78ae,cortex-a78c,cortex-a8,cortex-a9,cortex-m0,cortex-m0plus,cortex-m1,cortex-m23,cortex-m3,cortex-m33,cortex-m35p,cortex-m4,cortex-m52,cortex-m55,cortex-m7,cortex-m85,cortex-r4,cortex-r4f,cortex-r5,cortex-r52,cortex-r52plus,cortex-r7,cortex-r8,cortex-x1,cortex-x1c,cyclone,ep9312,exynos-m3,exynos-m4,exynos-m5,generic,iwmmxt,krait,kryo,mpcore,mpcorenovfp,neoverse-n1,neoverse-n2,neoverse-v1,sc000,sc300,star-mc1,strongarm,strongarm110,strongarm1100,strongarm1110,swift,xscale"),
// TargetArch_arm64:
str_lit("a64fx,ampere1,ampere1a,ampere1b,apple-a10,apple-a11,apple-a12,apple-a13,apple-a14,apple-a15,apple-a16,apple-a17,apple-a7,apple-a8,apple-a9,apple-m1,apple-m2,apple-m3,apple-m4,apple-s4,apple-s5,carmel,cobalt-100,cortex-a34,cortex-a35,cortex-a510,cortex-a520,cortex-a520ae,cortex-a53,cortex-a55,cortex-a57,cortex-a65,cortex-a65ae,cortex-a710,cortex-a715,cortex-a72,cortex-a720,cortex-a720ae,cortex-a725,cortex-a73,cortex-a75,cortex-a76,cortex-a76ae,cortex-a77,cortex-a78,cortex-a78ae,cortex-a78c,cortex-r82,cortex-r82ae,cortex-x1,cortex-x1c,cortex-x2,cortex-x3,cortex-x4,cortex-x925,cyclone,exynos-m3,exynos-m4,exynos-m5,falkor,fujitsu-monaka,generic,grace,kryo,neoverse-512tvb,neoverse-e1,neoverse-n1,neoverse-n2,neoverse-n3,neoverse-v1,neoverse-v2,neoverse-v3,neoverse-v3ae,oryon-1,saphira,thunderx,thunderx2t99,thunderx3t110,thunderxt81,thunderxt83,thunderxt88,tsv110"),
// TargetArch_wasm32:
str_lit("bleeding-edge,generic,lime1,mvp"),
// TargetArch_wasm64p32:
str_lit("bleeding-edge,generic,lime1,mvp"),
// TargetArch_riscv64:
str_lit("generic,generic-rv32,generic-rv64,mips-p8700,rocket,rocket-rv32,rocket-rv64,rp2350-hazard3,sifive-7-series,sifive-e20,sifive-e21,sifive-e24,sifive-e31,sifive-e34,sifive-e76,sifive-p450,sifive-p470,sifive-p550,sifive-p670,sifive-s21,sifive-s51,sifive-s54,sifive-s76,sifive-u54,sifive-u74,sifive-x280,spacemit-x60,syntacore-scr1-base,syntacore-scr1-max,syntacore-scr3-rv32,syntacore-scr3-rv64,syntacore-scr4-rv32,syntacore-scr4-rv64,syntacore-scr5-rv32,syntacore-scr5-rv64,syntacore-scr7,tt-ascalon-d8,veyron-v1,xiangshan-nanhu"),
};
// Generated with the featuregen script in `misc/featuregen`
gb_global String target_features_list[TargetArch_COUNT] = {
// TargetArch_Invalid:
str_lit(""),
// TargetArch_amd64:
str_lit("16bit-mode,32bit-mode,64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-avx512,amx-bf16,amx-complex,amx-fp16,amx-fp8,amx-int8,amx-movrs,amx-tf32,amx-tile,amx-transpose,avx,avx10.1-256,avx10.1-512,avx10.2-256,avx10.2-512,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vp2intersect,avx512vpopcntdq,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,branch-hint,branchfusion,ccmp,cf,cldemote,clflushopt,clwb,clzero,cmov,cmpccxadd,crc32,cx16,cx8,egpr,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-lzcnt-tzcnt,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-popcnt,false-deps-range,fast-11bytenop,fast-15bytenop,fast-7bytenop,fast-bextr,fast-dpwssd,fast-gather,fast-hops,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fast-vector-shift-masks,faster-shift-than-shuffle,fma,fma4,fsgsbase,fsrm,fxsr,gfni,harden-sls-ijmp,harden-sls-ret,hreset,idivl-to-divb,idivq-to-divl,inline-asm-use-gpr32,invpcid,kl,lea-sp,lea-uses-ag,lvi-cfi,lvi-load-hardening,lwp,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,movrs,mwaitx,ndd,nf,no-bypass-delay,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pad-short-functions,pclmul,pconfig,pku,popcnt,ppx,prefer-128-bit,prefer-256-bit,prefer-mask-registers,prefer-movmsk-over-vtest,prefer-no-gather,prefer-no-scatter,prefetchi,prfchw,ptwrite,push2pop2,raoint,rdpid,rdpru,rdrnd,rdseed,retpoline,retpoline-external-thunk,retpoline-indirect-branches,retpoline-indirect-calls,rtm,sahf,sbb-dep-breaking,serialize,seses,sgx,sha,sha512,shstk,slow-3ops-lea,slow-incdec,slow-lea,slow-pmaddwd,slow-pmulld,slow-shld,slow-two-mem-ops,slow-unaligned-mem-16,slow-unaligned-mem-32,sm3,sm4,soft-float,sse,sse-unaligned-mem,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tagged-globals,tbm,tsxldtrk,tuning-fast-imm-vector-shift,uintr,use-glm-div-sqrt-costs,use-slm-arith-costs,usermsr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,widekl,x87,xop,xsave,xsavec,xsaveopt,xsaves,zu"),
// TargetArch_i386:
str_lit("16bit-mode,32bit-mode,64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-avx512,amx-bf16,amx-complex,amx-fp16,amx-fp8,amx-int8,amx-movrs,amx-tf32,amx-tile,amx-transpose,avx,avx10.1-256,avx10.1-512,avx10.2-256,avx10.2-512,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vp2intersect,avx512vpopcntdq,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,branch-hint,branchfusion,ccmp,cf,cldemote,clflushopt,clwb,clzero,cmov,cmpccxadd,crc32,cx16,cx8,egpr,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-lzcnt-tzcnt,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-popcnt,false-deps-range,fast-11bytenop,fast-15bytenop,fast-7bytenop,fast-bextr,fast-dpwssd,fast-gather,fast-hops,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fast-vector-shift-masks,faster-shift-than-shuffle,fma,fma4,fsgsbase,fsrm,fxsr,gfni,harden-sls-ijmp,harden-sls-ret,hreset,idivl-to-divb,idivq-to-divl,inline-asm-use-gpr32,invpcid,kl,lea-sp,lea-uses-ag,lvi-cfi,lvi-load-hardening,lwp,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,movrs,mwaitx,ndd,nf,no-bypass-delay,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pad-short-functions,pclmul,pconfig,pku,popcnt,ppx,prefer-128-bit,prefer-256-bit,prefer-mask-registers,prefer-movmsk-over-vtest,prefer-no-gather,prefer-no-scatter,prefetchi,prfchw,ptwrite,push2pop2,raoint,rdpid,rdpru,rdrnd,rdseed,retpoline,retpoline-external-thunk,retpoline-indirect-branches,retpoline-indirect-calls,rtm,sahf,sbb-dep-breaking,serialize,seses,sgx,sha,sha512,shstk,slow-3ops-lea,slow-incdec,slow-lea,slow-pmaddwd,slow-pmulld,slow-shld,slow-two-mem-ops,slow-unaligned-mem-16,slow-unaligned-mem-32,sm3,sm4,soft-float,sse,sse-unaligned-mem,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tagged-globals,tbm,tsxldtrk,tuning-fast-imm-vector-shift,uintr,use-glm-div-sqrt-costs,use-slm-arith-costs,usermsr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,widekl,x87,xop,xsave,xsavec,xsaveopt,xsaves,zu"),
// TargetArch_arm32:
str_lit("32bit,8msecext,a12,a15,a17,a32,a35,a5,a53,a55,a57,a7,a72,a73,a75,a76,a77,a78c,a8,a9,aapcs-frame-chain,aclass,acquire-release,aes,armv4,armv4t,armv5t,armv5te,armv5tej,armv6,armv6-m,armv6j,armv6k,armv6kz,armv6s-m,armv6t2,armv7-a,armv7-m,armv7-r,armv7e-m,armv7k,armv7s,armv7ve,armv8-a,armv8-m.base,armv8-m.main,armv8-r,armv8.1-a,armv8.1-m.main,armv8.2-a,armv8.3-a,armv8.4-a,armv8.5-a,armv8.6-a,armv8.7-a,armv8.8-a,armv8.9-a,armv9-a,armv9.1-a,armv9.2-a,armv9.3-a,armv9.4-a,armv9.5-a,armv9.6-a,atomics-32,avoid-movs-shop,avoid-muls,avoid-partial-cpsr,bf16,big-endian-instructions,branch-align-64,cde,cdecp0,cdecp1,cdecp2,cdecp3,cdecp4,cdecp5,cdecp6,cdecp7,cheap-predicable-cpsr,clrbhb,cortex-a510,cortex-a710,cortex-a78,cortex-a78ae,cortex-x1,cortex-x1c,crc,crypto,d32,db,dfb,disable-postra-scheduler,dont-widen-vmovs,dotprod,dsp,execute-only,expand-fp-mlx,exynos,fix-cmse-cve-2021-35465,fix-cortex-a57-aes-1742098,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp16fml,fp64,fpao,fpregs,fpregs16,fpregs64,fullfp16,fuse-aes,fuse-literals,harden-sls-blr,harden-sls-nocomdat,harden-sls-retbr,hwdiv,hwdiv-arm,i8mm,iwmmxt,iwmmxt2,krait,kryo,lob,long-calls,loop-align,m3,m55,m7,m85,mclass,mp,muxed-units,mve,mve.fp,mve1beat,mve2beat,mve4beat,nacl-trap,neon,neon-fpmovs,neonfp,neoverse-v1,no-branch-predictor,no-bti-at-return-twice,no-movt,no-neg-immediates,noarm,nonpipelined-vfp,pacbti,perfmon,prefer-ishst,prefer-vmovsr,prof-unpr,r4,r5,r52,r52plus,r7,ras,rclass,read-tp-tpidrprw,read-tp-tpidruro,read-tp-tpidrurw,reserve-r9,ret-addr-stack,sb,sha2,slow-fp-brcc,slow-load-D-subreg,slow-odd-reg,slow-vdup32,slow-vgetlni32,slowfpvfmx,slowfpvmlx,soft-float,splat-vfp-neon,strict-align,swift,thumb-mode,thumb2,trustzone,use-mipipeliner,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.1m.main,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8.8a,v8.9a,v8m,v8m.main,v9.1a,v9.2a,v9.3a,v9.4a,v9.5a,v9.6a,v9a,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,vldn-align,vmlx-forwarding,vmlx-hazards,wide-stride-vfp,xscale,zcz"),
// TargetArch_arm64:
str_lit("CONTEXTIDREL2,a35,a510,a520,a520ae,a53,a55,a57,a64fx,a65,a710,a715,a72,a720,a720ae,a73,a75,a76,a77,a78,a78ae,a78c,addr-lsl-slow-14,aes,aggressive-fma,all,alternate-sextload-cvt-f32-pattern,altnzcv,alu-lsl-fast,am,ampere1,ampere1a,ampere1b,amvs,apple-a10,apple-a11,apple-a12,apple-a13,apple-a14,apple-a15,apple-a16,apple-a17,apple-a7,apple-m4,arith-bcc-fusion,arith-cbz-fusion,ascend-store-address,avoid-ldapur,balance-fp-ops,bf16,brbe,bti,call-saved-x10,call-saved-x11,call-saved-x12,call-saved-x13,call-saved-x14,call-saved-x15,call-saved-x18,call-saved-x8,call-saved-x9,carmel,ccdp,ccidx,ccpp,chk,clrbhb,cmp-bcc-fusion,cmpbr,complxnum,cortex-a725,cortex-r82,cortex-r82ae,cortex-x1,cortex-x2,cortex-x3,cortex-x4,cortex-x925,cpa,crc,crypto,cssc,d128,disable-latency-sched-heuristic,disable-ldp,disable-stp,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,exynos-cheap-as-move,exynosm3,exynosm4,f32mm,f64mm,f8f16mm,f8f32mm,falkor,faminmax,fgt,fix-cortex-a53-835769,flagm,fmv,force-32bit-jump-tables,fp-armv8,fp16fml,fp8,fp8dot2,fp8dot4,fp8fma,fpac,fprcvt,fptoint,fujitsu-monaka,fullfp16,fuse-address,fuse-addsub-2reg-const1,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,gcs,harden-sls-blr,harden-sls-nocomdat,harden-sls-retbr,hbc,hcx,i8mm,ite,jsconv,kryo,ldp-aligned-only,lor,ls64,lse,lse128,lse2,lsfe,lsui,lut,mec,mops,mpam,mte,neon,neoverse512tvb,neoversee1,neoversen1,neoversen2,neoversen3,neoversev1,neoversev2,neoversev3,neoversev3AE,nmi,no-bti-at-return-twice,no-neg-immediates,no-sve-fp-ld1r,no-zcz-fp,nv,occmo,oryon-1,outline-atomics,pan,pan-rwv,pauth,pauth-lr,pcdphint,perfmon,pops,predictable-select-expensive,predres,prfm-slc-target,rand,ras,rasv2,rcpc,rcpc-immo,rcpc3,rdm,reserve-lr-for-ra,reserve-x1,reserve-x10,reserve-x11,reserve-x12,reserve-x13,reserve-x14,reserve-x15,reserve-x18,reserve-x2,reserve-x20,reserve-x21,reserve-x22,reserve-x23,reserve-x24,reserve-x25,reserve-x26,reserve-x27,reserve-x28,reserve-x3,reserve-x4,reserve-x5,reserve-x6,reserve-x7,reserve-x9,rme,saphira,sb,sel2,sha2,sha3,slow-misaligned-128store,slow-paired-128,slow-strqro-store,sm4,sme,sme-b16b16,sme-f16f16,sme-f64f64,sme-f8f16,sme-f8f32,sme-fa64,sme-i16i64,sme-lutv2,sme-mop4,sme-tmop,sme2,sme2p1,sme2p2,spe,spe-eef,specres2,specrestrict,ssbs,ssve-aes,ssve-bitperm,ssve-fp8dot2,ssve-fp8dot4,ssve-fp8fma,store-pair-suppress,stp-aligned-only,strict-align,sve,sve-aes,sve-aes2,sve-b16b16,sve-bfscale,sve-bitperm,sve-f16f32mm,sve2,sve2-aes,sve2-bitperm,sve2-sha3,sve2-sm4,sve2p1,sve2p2,tagged-globals,the,thunderx,thunderx2t99,thunderx3t110,thunderxt81,thunderxt83,thunderxt88,tlb-rmi,tlbiw,tme,tpidr-el1,tpidr-el2,tpidr-el3,tpidrro-el0,tracev8.4,trbe,tsv110,uaops,use-experimental-zeroing-pseudos,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,use-reciprocal-square-root,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8.8a,v8.9a,v8a,v8r,v9.1a,v9.2a,v9.3a,v9.4a,v9.5a,v9.6a,v9a,vh,wfxt,xs,zcm,zcz,zcz-fp-workaround,zcz-gp"),
// TargetArch_wasm32:
str_lit("atomics,bulk-memory,bulk-memory-opt,call-indirect-overlong,exception-handling,extended-const,fp16,multimemory,multivalue,mutable-globals,nontrapping-fptoint,reference-types,relaxed-simd,sign-ext,simd128,tail-call,wide-arithmetic"),
// TargetArch_wasm64p32:
str_lit("atomics,bulk-memory,bulk-memory-opt,call-indirect-overlong,exception-handling,extended-const,fp16,multimemory,multivalue,mutable-globals,nontrapping-fptoint,reference-types,relaxed-simd,sign-ext,simd128,tail-call,wide-arithmetic"),
// TargetArch_riscv64:
str_lit("32bit,64bit,a,auipc-addi-fusion,b,c,conditional-cmv-fusion,d,disable-latency-sched-heuristic,dlen-factor-2,e,experimental,experimental-rvm23u32,experimental-sdext,experimental-sdtrig,experimental-smctr,experimental-ssctr,experimental-svukte,experimental-xqcia,experimental-xqciac,experimental-xqcicli,experimental-xqcicm,experimental-xqcics,experimental-xqcicsr,experimental-xqciint,experimental-xqcilo,experimental-xqcilsm,experimental-xqcisls,experimental-zalasr,experimental-zicfilp,experimental-zicfiss,experimental-zvbc32e,experimental-zvkgs,f,forced-atomics,h,i,ld-add-fusion,lui-addi-fusion,m,mips-p8700,no-default-unroll,no-rvc-hints,no-sink-splat-operands,no-trailing-seq-cst-fence,optimized-nf2-segment-load-store,optimized-nf3-segment-load-store,optimized-nf4-segment-load-store,optimized-nf5-segment-load-store,optimized-nf6-segment-load-store,optimized-nf7-segment-load-store,optimized-nf8-segment-load-store,optimized-zero-stride-load,predictable-select-expensive,prefer-w-inst,relax,reserve-x1,reserve-x10,reserve-x11,reserve-x12,reserve-x13,reserve-x14,reserve-x15,reserve-x16,reserve-x17,reserve-x18,reserve-x19,reserve-x2,reserve-x20,reserve-x21,reserve-x22,reserve-x23,reserve-x24,reserve-x25,reserve-x26,reserve-x27,reserve-x28,reserve-x29,reserve-x3,reserve-x30,reserve-x31,reserve-x4,reserve-x5,reserve-x6,reserve-x7,reserve-x8,reserve-x9,rva20s64,rva20u64,rva22s64,rva22u64,rva23s64,rva23u64,rvb23s64,rvb23u64,rvi20u32,rvi20u64,save-restore,sha,shcounterenw,shgatpa,shifted-zextw-fusion,short-forward-branch-opt,shtvala,shvsatpa,shvstvala,shvstvecd,sifive7,smaia,smcdeleg,smcsrind,smdbltrp,smepmp,smmpm,smnpm,smrnmi,smstateen,ssaia,ssccfg,ssccptr,sscofpmf,sscounterenw,sscsrind,ssdbltrp,ssnpm,sspm,ssqosid,ssstateen,ssstrict,sstc,sstvala,sstvecd,ssu64xl,supm,svade,svadu,svbare,svinval,svnapot,svpbmt,svvptc,tagged-globals,unaligned-scalar-mem,unaligned-vector-mem,use-postra-scheduler,v,ventana-veyron,vxrm-pipeline-flush,xcvalu,xcvbi,xcvbitmanip,xcvelw,xcvmac,xcvmem,xcvsimd,xmipscmove,xmipslsp,xsfcease,xsfvcp,xsfvfnrclipxfqf,xsfvfwmaccqqq,xsfvqmaccdod,xsfvqmaccqoq,xsifivecdiscarddlone,xsifivecflushdlone,xtheadba,xtheadbb,xtheadbs,xtheadcmo,xtheadcondmov,xtheadfmemidx,xtheadmac,xtheadmemidx,xtheadmempair,xtheadsync,xtheadvdot,xventanacondops,xwchc,za128rs,za64rs,zaamo,zabha,zacas,zalrsc,zama16b,zawrs,zba,zbb,zbc,zbkb,zbkc,zbkx,zbs,zca,zcb,zcd,zce,zcf,zcmop,zcmp,zcmt,zdinx,zexth-fusion,zextw-fusion,zfa,zfbfmin,zfh,zfhmin,zfinx,zhinx,zhinxmin,zic64b,zicbom,zicbop,zicboz,ziccamoa,ziccif,zicclsm,ziccrse,zicntr,zicond,zicsr,zifencei,zihintntl,zihintpause,zihpm,zimop,zk,zkn,zknd,zkne,zknh,zkr,zks,zksed,zksh,zkt,zmmul,ztso,zvbb,zvbc,zve32f,zve32x,zve64d,zve64f,zve64x,zvfbfmin,zvfbfwma,zvfh,zvfhmin,zvkb,zvkg,zvkn,zvknc,zvkned,zvkng,zvknha,zvknhb,zvks,zvksc,zvksed,zvksg,zvksh,zvkt,zvl1024b,zvl128b,zvl16384b,zvl2048b,zvl256b,zvl32768b,zvl32b,zvl4096b,zvl512b,zvl64b,zvl65536b,zvl8192b"),
};
// Generated with the featuregen script in `misc/featuregen`
gb_global int target_microarch_counts[TargetArch_COUNT] = {
// TargetArch_Invalid:
0,
// TargetArch_amd64:
129,
// TargetArch_i386:
129,
// TargetArch_arm32:
95,
// TargetArch_arm64:
82,
// TargetArch_wasm32:
4,
// TargetArch_wasm64p32:
4,
// TargetArch_riscv64:
39,
};
// Generated with the featuregen script in `misc/featuregen`
gb_global MicroarchFeatureList microarch_features_list[] = {
// TargetArch_amd64:
{ str_lit("alderlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("amdfam10"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,idivq-to-divl,lzcnt,nopl,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4a,vzeroupper,x87") },
{ str_lit("arrowlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("arrowlake-s"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("arrowlake_s"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("athlon"), str_lit("64bit-mode,cmov,cx8,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon-4"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon-fx"), str_lit("64bit,64bit-mode,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon-mp"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon-tbird"), str_lit("64bit-mode,cmov,cx8,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon-xp"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon64"), str_lit("64bit,64bit-mode,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon64-sse3"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("atom"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-imm16,fxsr,idivl-to-divb,idivq-to-divl,lea-sp,lea-uses-ag,mmx,movbe,no-bypass-delay,nopl,pad-short-functions,sahf,slow-two-mem-ops,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("atom_sse4_2"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87") },
{ str_lit("atom_sse4_2_movbe"), str_lit("64bit,64bit-mode,aes,clflushopt,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fsgsbase,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("barcelona"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,idivq-to-divl,lzcnt,nopl,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4a,vzeroupper,x87") },
{ str_lit("bdver1"), str_lit("64bit,64bit-mode,aes,avx,branchfusion,cmov,crc32,cx16,cx8,fast-11bytenop,fast-scalar-shift-masks,fma4,fxsr,idivq-to-divl,lwp,lzcnt,mmx,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vzeroupper,x87,xop,xsave") },
{ str_lit("bdver2"), str_lit("64bit,64bit-mode,aes,avx,bmi,branchfusion,cmov,crc32,cx16,cx8,f16c,fast-11bytenop,fast-bextr,fast-movbe,fast-scalar-shift-masks,fma,fma4,fxsr,idivq-to-divl,lwp,lzcnt,mmx,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tbm,vzeroupper,x87,xop,xsave") },
{ str_lit("bdver3"), str_lit("64bit,64bit-mode,aes,avx,bmi,branchfusion,cmov,crc32,cx16,cx8,f16c,fast-11bytenop,fast-bextr,fast-movbe,fast-scalar-shift-masks,fma,fma4,fsgsbase,fxsr,idivq-to-divl,lwp,lzcnt,mmx,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tbm,vzeroupper,x87,xop,xsave,xsaveopt") },
{ str_lit("bdver4"), str_lit("64bit,64bit-mode,aes,avx,avx2,bmi,bmi2,branchfusion,cmov,crc32,cx16,cx8,f16c,fast-11bytenop,fast-bextr,fast-movbe,fast-scalar-shift-masks,fma,fma4,fsgsbase,fxsr,idivq-to-divl,lwp,lzcnt,mmx,movbe,mwaitx,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tbm,vzeroupper,x87,xop,xsave,xsaveopt") },
{ str_lit("bonnell"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-imm16,fxsr,idivl-to-divb,idivq-to-divl,lea-sp,lea-uses-ag,mmx,movbe,no-bypass-delay,nopl,pad-short-functions,sahf,slow-two-mem-ops,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("broadwell"), str_lit("64bit,64bit-mode,adx,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("btver1"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-15bytenop,fast-imm16,fast-scalar-shift-masks,fast-vector-shift-masks,fxsr,idivq-to-divl,lzcnt,mmx,nopl,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4a,ssse3,vzeroupper,x87") },
{ str_lit("btver2"), str_lit("64bit,64bit-mode,aes,avx,bmi,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-hops,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-shift-masks,fast-vector-shift-masks,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,x87,xsave,xsaveopt") },
{ str_lit("c3"), str_lit("64bit-mode,mmx,prfchw,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("c3-2"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("cannonlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vl,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,sha,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("cascadelake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,avx512vnni,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("clearwaterforest"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,usermsr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("cooperlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bf16,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,avx512vnni,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("core-avx-i"), str_lit("64bit,64bit-mode,avx,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fsgsbase,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core-avx2"), str_lit("64bit,64bit-mode,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core2"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("core_2_duo_sse4_1"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,sse4.1,ssse3,vzeroupper,x87") },
{ str_lit("core_2_duo_ssse3"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("core_2nd_gen_avx"), str_lit("64bit,64bit-mode,avx,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_3rd_gen_avx"), str_lit("64bit,64bit-mode,avx,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fsgsbase,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_4th_gen_avx"), str_lit("64bit,64bit-mode,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_4th_gen_avx_tsx"), str_lit("64bit,64bit-mode,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_5th_gen_avx"), str_lit("64bit,64bit-mode,adx,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_5th_gen_avx_tsx"), str_lit("64bit,64bit-mode,adx,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_aes_pclmulqdq"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("core_i7_sse4_2"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("corei7"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("corei7-avx"), str_lit("64bit,64bit-mode,avx,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("diamondrapids"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-avx512,amx-bf16,amx-complex,amx-fp16,amx-fp8,amx-int8,amx-movrs,amx-tf32,amx-tile,amx-transpose,avx,avx10.1-256,avx10.1-512,avx10.2-256,avx10.2-512,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,branch-hint,ccmp,cf,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,egpr,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,movrs,ndd,nf,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,ppx,prefer-256-bit,prefetchi,prfchw,ptwrite,push2pop2,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,usermsr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves,zu") },
{ str_lit("emeraldrapids"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-bf16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("generic"), str_lit("64bit,64bit-mode,cx8,fast-15bytenop,fast-scalar-fsqrt,idivq-to-divl,macrofusion,slow-3ops-lea,sse,sse2,vzeroupper,x87") },
{ str_lit("geode"), str_lit("64bit-mode,cx8,mmx,prfchw,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("goldmont"), str_lit("64bit,64bit-mode,aes,clflushopt,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-imm16,fast-movbe,fsgsbase,fxsr,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("goldmont-plus"), str_lit("64bit,64bit-mode,aes,clflushopt,cmov,crc32,cx16,cx8,fast-imm16,fast-movbe,fsgsbase,fxsr,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("goldmont_plus"), str_lit("64bit,64bit-mode,aes,clflushopt,cmov,crc32,cx16,cx8,fast-imm16,fast-movbe,fsgsbase,fxsr,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("gracemont"), str_lit("64bit,64bit-mode,adx,aes,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,nopl,pclmul,pconfig,pku,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("grandridge"), str_lit("64bit,64bit-mode,adx,aes,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,nopl,pclmul,pconfig,pku,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("graniterapids"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-bf16,amx-fp16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,branch-hint,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("graniterapids-d"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-bf16,amx-complex,amx-fp16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,branch-hint,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("graniterapids_d"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-bf16,amx-complex,amx-fp16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,branch-hint,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("haswell"), str_lit("64bit,64bit-mode,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("i386"), str_lit("64bit-mode,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("i486"), str_lit("64bit-mode,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("i586"), str_lit("64bit-mode,cx8,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("i686"), str_lit("64bit-mode,cmov,cx8,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("icelake-client"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("icelake-server"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("icelake_client"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("icelake_server"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("ivybridge"), str_lit("64bit,64bit-mode,avx,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fsgsbase,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("k6"), str_lit("64bit-mode,cx8,mmx,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("k6-2"), str_lit("64bit-mode,cx8,mmx,prfchw,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("k6-3"), str_lit("64bit-mode,cx8,mmx,prfchw,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("k8"), str_lit("64bit,64bit-mode,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("k8-sse3"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("knl"), str_lit("64bit,64bit-mode,adx,aes,avx,avx2,avx512cd,avx512f,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,fast-gather,fast-imm16,fast-movbe,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prefer-mask-registers,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,slow-incdec,slow-pmaddwd,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,x87,xsave,xsaveopt") },
{ str_lit("knm"), str_lit("64bit,64bit-mode,adx,aes,avx,avx2,avx512cd,avx512f,avx512vpopcntdq,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,fast-gather,fast-imm16,fast-movbe,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prefer-mask-registers,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,slow-incdec,slow-pmaddwd,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,x87,xsave,xsaveopt") },
{ str_lit("lakemont"), str_lit("64bit-mode,cx8,slow-unaligned-mem-16,sse,sse2,vzeroupper") },
{ str_lit("lunarlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("meteorlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("mic_avx512"), str_lit("64bit,64bit-mode,adx,aes,avx,avx2,avx512cd,avx512f,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,fast-gather,fast-imm16,fast-movbe,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prefer-mask-registers,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,slow-incdec,slow-pmaddwd,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,x87,xsave,xsaveopt") },
{ str_lit("nehalem"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("nocona"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("opteron"), str_lit("64bit,64bit-mode,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("opteron-sse3"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("pantherlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("penryn"), str_lit("64bit,64bit-mode,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,sse4.1,ssse3,vzeroupper,x87") },
{ str_lit("pentium"), str_lit("64bit-mode,cx8,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium-m"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium-mmx"), str_lit("64bit-mode,cx8,mmx,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium2"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium3"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium3m"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium4"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium4m"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_4"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_4_sse3"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("pentium_ii"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_iii"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_iii_no_xmm_regs"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_m"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_mmx"), str_lit("64bit-mode,cx8,mmx,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_pro"), str_lit("64bit-mode,cmov,cx8,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentiumpro"), str_lit("64bit-mode,cmov,cx8,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("prescott"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("raptorlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("rocketlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("sandybridge"), str_lit("64bit,64bit-mode,avx,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("sapphirerapids"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,amx-bf16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("sierraforest"), str_lit("64bit,64bit-mode,adx,aes,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,nopl,pclmul,pconfig,pku,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("silvermont"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87") },
{ str_lit("skx"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("skylake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("skylake-avx512"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("skylake_avx512"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("slm"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87") },
{ str_lit("tigerlake"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vp2intersect,avx512vpopcntdq,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("tremont"), str_lit("64bit,64bit-mode,aes,clflushopt,clwb,cmov,crc32,cx16,cx8,fast-imm16,fast-movbe,fsgsbase,fxsr,gfni,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("westmere"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("winchip-c6"), str_lit("64bit-mode,mmx,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("winchip2"), str_lit("64bit-mode,mmx,prfchw,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("x86-64"), str_lit("64bit,64bit-mode,cmov,cx8,fxsr,idivq-to-divl,macrofusion,mmx,nopl,slow-3ops-lea,slow-incdec,sse,sse2,vzeroupper,x87") },
{ str_lit("x86-64-v2"), str_lit("64bit,64bit-mode,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,nopl,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("x86-64-v3"), str_lit("64bit,64bit-mode,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fxsr,idivq-to-divl,lzcnt,macrofusion,mmx,movbe,nopl,popcnt,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave") },
{ str_lit("x86-64-v4"), str_lit("64bit,64bit-mode,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fxsr,idivq-to-divl,lzcnt,macrofusion,mmx,movbe,nopl,popcnt,prefer-256-bit,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave") },
{ str_lit("yonah"), str_lit("64bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("znver1"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,branchfusion,clflushopt,clzero,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,mwaitx,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver2"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,mwaitx,nopl,pclmul,popcnt,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver3"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,mwaitx,nopl,pclmul,pku,popcnt,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver4"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,evex512,f16c,fast-15bytenop,fast-bextr,fast-dpwssd,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,mwaitx,nopl,pclmul,pku,popcnt,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,shstk,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver5"), str_lit("64bit,64bit-mode,adx,aes,allow-light-256-bit,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vp2intersect,avx512vpopcntdq,avxvnni,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,evex512,f16c,fast-15bytenop,fast-bextr,fast-dpwssd,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,mwaitx,nopl,pclmul,pku,popcnt,prefetchi,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,shstk,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
// TargetArch_i386:
{ str_lit("alderlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("amdfam10"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,idivq-to-divl,lzcnt,nopl,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4a,vzeroupper,x87") },
{ str_lit("arrowlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("arrowlake-s"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("arrowlake_s"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("athlon"), str_lit("32bit-mode,cmov,cx8,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("athlon-4"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("athlon-fx"), str_lit("32bit-mode,64bit,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon-mp"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("athlon-tbird"), str_lit("32bit-mode,cmov,cx8,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("athlon-xp"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,prfchw,slow-shld,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("athlon64"), str_lit("32bit-mode,64bit,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("athlon64-sse3"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("atom"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-imm16,fxsr,idivl-to-divb,idivq-to-divl,lea-sp,lea-uses-ag,mmx,movbe,no-bypass-delay,nopl,pad-short-functions,sahf,slow-two-mem-ops,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("atom_sse4_2"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87") },
{ str_lit("atom_sse4_2_movbe"), str_lit("32bit-mode,64bit,aes,clflushopt,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fsgsbase,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("barcelona"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,idivq-to-divl,lzcnt,nopl,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4a,vzeroupper,x87") },
{ str_lit("bdver1"), str_lit("32bit-mode,64bit,aes,avx,branchfusion,cmov,crc32,cx16,cx8,fast-11bytenop,fast-scalar-shift-masks,fma4,fxsr,idivq-to-divl,lwp,lzcnt,mmx,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vzeroupper,x87,xop,xsave") },
{ str_lit("bdver2"), str_lit("32bit-mode,64bit,aes,avx,bmi,branchfusion,cmov,crc32,cx16,cx8,f16c,fast-11bytenop,fast-bextr,fast-movbe,fast-scalar-shift-masks,fma,fma4,fxsr,idivq-to-divl,lwp,lzcnt,mmx,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tbm,vzeroupper,x87,xop,xsave") },
{ str_lit("bdver3"), str_lit("32bit-mode,64bit,aes,avx,bmi,branchfusion,cmov,crc32,cx16,cx8,f16c,fast-11bytenop,fast-bextr,fast-movbe,fast-scalar-shift-masks,fma,fma4,fsgsbase,fxsr,idivq-to-divl,lwp,lzcnt,mmx,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tbm,vzeroupper,x87,xop,xsave,xsaveopt") },
{ str_lit("bdver4"), str_lit("32bit-mode,64bit,aes,avx,avx2,bmi,bmi2,branchfusion,cmov,crc32,cx16,cx8,f16c,fast-11bytenop,fast-bextr,fast-movbe,fast-scalar-shift-masks,fma,fma4,fsgsbase,fxsr,idivq-to-divl,lwp,lzcnt,mmx,movbe,mwaitx,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,tbm,vzeroupper,x87,xop,xsave,xsaveopt") },
{ str_lit("bonnell"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-imm16,fxsr,idivl-to-divb,idivq-to-divl,lea-sp,lea-uses-ag,mmx,movbe,no-bypass-delay,nopl,pad-short-functions,sahf,slow-two-mem-ops,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("broadwell"), str_lit("32bit-mode,64bit,adx,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("btver1"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-15bytenop,fast-imm16,fast-scalar-shift-masks,fast-vector-shift-masks,fxsr,idivq-to-divl,lzcnt,mmx,nopl,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4a,ssse3,vzeroupper,x87") },
{ str_lit("btver2"), str_lit("32bit-mode,64bit,aes,avx,bmi,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-hops,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-shift-masks,fast-vector-shift-masks,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prfchw,sahf,sbb-dep-breaking,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,x87,xsave,xsaveopt") },
{ str_lit("c3"), str_lit("32bit-mode,mmx,prfchw,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("c3-2"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("cannonlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vl,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,sha,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("cascadelake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,avx512vnni,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("clearwaterforest"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,usermsr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("cooperlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bf16,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,avx512vnni,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("core-avx-i"), str_lit("32bit-mode,64bit,avx,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fsgsbase,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core-avx2"), str_lit("32bit-mode,64bit,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core2"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("core_2_duo_sse4_1"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,sse4.1,ssse3,vzeroupper,x87") },
{ str_lit("core_2_duo_ssse3"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,ssse3,vzeroupper,x87") },
{ str_lit("core_2nd_gen_avx"), str_lit("32bit-mode,64bit,avx,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_3rd_gen_avx"), str_lit("32bit-mode,64bit,avx,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fsgsbase,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_4th_gen_avx"), str_lit("32bit-mode,64bit,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_4th_gen_avx_tsx"), str_lit("32bit-mode,64bit,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_5th_gen_avx"), str_lit("32bit-mode,64bit,adx,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_5th_gen_avx_tsx"), str_lit("32bit-mode,64bit,adx,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("core_aes_pclmulqdq"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("core_i7_sse4_2"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("corei7"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("corei7-avx"), str_lit("32bit-mode,64bit,avx,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("diamondrapids"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,amx-avx512,amx-bf16,amx-complex,amx-fp16,amx-fp8,amx-int8,amx-movrs,amx-tf32,amx-tile,amx-transpose,avx,avx10.1-256,avx10.1-512,avx10.2-256,avx10.2-512,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,branch-hint,ccmp,cf,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,egpr,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,movrs,ndd,nf,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,ppx,prefer-256-bit,prefetchi,prfchw,ptwrite,push2pop2,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,usermsr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves,zu") },
{ str_lit("emeraldrapids"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,amx-bf16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("generic"), str_lit("32bit-mode,64bit,cx8,fast-15bytenop,fast-scalar-fsqrt,idivq-to-divl,macrofusion,slow-3ops-lea,vzeroupper,x87") },
{ str_lit("geode"), str_lit("32bit-mode,cx8,mmx,prfchw,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("goldmont"), str_lit("32bit-mode,64bit,aes,clflushopt,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-imm16,fast-movbe,fsgsbase,fxsr,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("goldmont-plus"), str_lit("32bit-mode,64bit,aes,clflushopt,cmov,crc32,cx16,cx8,fast-imm16,fast-movbe,fsgsbase,fxsr,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("goldmont_plus"), str_lit("32bit-mode,64bit,aes,clflushopt,cmov,crc32,cx16,cx8,fast-imm16,fast-movbe,fsgsbase,fxsr,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("gracemont"), str_lit("32bit-mode,64bit,adx,aes,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,nopl,pclmul,pconfig,pku,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("grandridge"), str_lit("32bit-mode,64bit,adx,aes,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,nopl,pclmul,pconfig,pku,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("graniterapids"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,amx-bf16,amx-fp16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,branch-hint,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("graniterapids-d"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,amx-bf16,amx-complex,amx-fp16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,branch-hint,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("graniterapids_d"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,amx-bf16,amx-complex,amx-fp16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,branch-hint,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("haswell"), str_lit("32bit-mode,64bit,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("i386"), str_lit("32bit-mode,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("i486"), str_lit("32bit-mode,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("i586"), str_lit("32bit-mode,cx8,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("i686"), str_lit("32bit-mode,cmov,cx8,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("icelake-client"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("icelake-server"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("icelake_client"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("icelake_server"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("ivybridge"), str_lit("32bit-mode,64bit,avx,cmov,crc32,cx16,cx8,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fsgsbase,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,rdrnd,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("k6"), str_lit("32bit-mode,cx8,mmx,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("k6-2"), str_lit("32bit-mode,cx8,mmx,prfchw,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("k6-3"), str_lit("32bit-mode,cx8,mmx,prfchw,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("k8"), str_lit("32bit-mode,64bit,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("k8-sse3"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("knl"), str_lit("32bit-mode,64bit,adx,aes,avx,avx2,avx512cd,avx512f,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,fast-gather,fast-imm16,fast-movbe,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prefer-mask-registers,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,slow-incdec,slow-pmaddwd,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,x87,xsave,xsaveopt") },
{ str_lit("knm"), str_lit("32bit-mode,64bit,adx,aes,avx,avx2,avx512cd,avx512f,avx512vpopcntdq,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,fast-gather,fast-imm16,fast-movbe,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prefer-mask-registers,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,slow-incdec,slow-pmaddwd,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,x87,xsave,xsaveopt") },
{ str_lit("lakemont"), str_lit("32bit-mode,cx8,slow-unaligned-mem-16,vzeroupper") },
{ str_lit("lunarlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("meteorlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("mic_avx512"), str_lit("32bit-mode,64bit,adx,aes,avx,avx2,avx512cd,avx512f,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,fast-gather,fast-imm16,fast-movbe,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,nopl,pclmul,popcnt,prefer-mask-registers,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,slow-incdec,slow-pmaddwd,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,x87,xsave,xsaveopt") },
{ str_lit("nehalem"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("nocona"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("opteron"), str_lit("32bit-mode,64bit,cmov,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("opteron-sse3"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fast-scalar-shift-masks,fxsr,mmx,nopl,prfchw,sbb-dep-breaking,slow-shld,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("pantherlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint16,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prefetchi,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,sha512,shstk,slow-3ops-lea,sm3,sm4,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("penryn"), str_lit("32bit-mode,64bit,cmov,cx16,cx8,fxsr,macrofusion,mmx,nopl,sahf,slow-unaligned-mem-16,sse,sse2,sse3,sse4.1,ssse3,vzeroupper,x87") },
{ str_lit("pentium"), str_lit("32bit-mode,cx8,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("pentium-m"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium-mmx"), str_lit("32bit-mode,cx8,mmx,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("pentium2"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("pentium3"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("pentium3m"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("pentium4"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium4m"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_4"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_4_sse3"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("pentium_ii"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("pentium_iii"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("pentium_iii_no_xmm_regs"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,vzeroupper,x87") },
{ str_lit("pentium_m"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,vzeroupper,x87") },
{ str_lit("pentium_mmx"), str_lit("32bit-mode,cx8,mmx,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("pentium_pro"), str_lit("32bit-mode,cmov,cx8,nopl,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("pentiumpro"), str_lit("32bit-mode,cmov,cx8,nopl,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("prescott"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("raptorlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,f16c,false-deps-perm,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,idivq-to-divl,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-movmsk-over-vtest,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("rocketlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("sandybridge"), str_lit("32bit-mode,64bit,avx,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsaveopt") },
{ str_lit("sapphirerapids"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,amx-bf16,amx-int8,amx-tile,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512fp16,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,avxvnni,bmi,bmi2,cldemote,clflushopt,clwb,cmov,crc32,cx16,cx8,enqcmd,ermsb,evex512,f16c,false-deps-getmant,false-deps-mulc,false-deps-mullq,false-deps-perm,false-deps-range,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pconfig,pku,popcnt,prefer-256-bit,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tsxldtrk,tuning-fast-imm-vector-shift,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("sierraforest"), str_lit("32bit-mode,64bit,adx,aes,avx,avx2,avxifma,avxneconvert,avxvnni,avxvnniint8,bmi,bmi2,cldemote,clflushopt,clwb,cmov,cmpccxadd,crc32,cx16,cx8,enqcmd,f16c,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,gfni,hreset,invpcid,kl,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,nopl,pclmul,pconfig,pku,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,serialize,sha,shstk,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,uintr,vaes,vpclmulqdq,vzeroupper,waitpkg,widekl,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("silvermont"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87") },
{ str_lit("skx"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("skylake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,clflushopt,cmov,crc32,cx16,cx8,ermsb,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("skylake-avx512"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("skylake_avx512"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,faster-shift-than-shuffle,fma,fsgsbase,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdrnd,rdseed,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("slm"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-7bytenop,fast-imm16,fast-movbe,fxsr,idivq-to-divl,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,rdrnd,sahf,slow-incdec,slow-lea,slow-pmulld,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-slm-arith-costs,vzeroupper,x87") },
{ str_lit("tigerlake"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vp2intersect,avx512vpopcntdq,bmi,bmi2,clflushopt,clwb,cmov,crc32,cx16,cx8,ermsb,evex512,f16c,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,no-bypass-delay-blend,no-bypass-delay-mov,no-bypass-delay-shuffle,nopl,pclmul,pku,popcnt,prefer-256-bit,prfchw,rdpid,rdrnd,rdseed,sahf,sha,shstk,sse,sse2,sse3,sse4.1,sse4.2,ssse3,tuning-fast-imm-vector-shift,vaes,vpclmulqdq,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("tremont"), str_lit("32bit-mode,64bit,aes,clflushopt,clwb,cmov,crc32,cx16,cx8,fast-imm16,fast-movbe,fsgsbase,fxsr,gfni,mmx,movbe,no-bypass-delay,nopl,pclmul,popcnt,prfchw,ptwrite,rdpid,rdrnd,rdseed,sahf,sha,slow-incdec,slow-lea,slow-two-mem-ops,sse,sse2,sse3,sse4.1,sse4.2,ssse3,use-glm-div-sqrt-costs,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("westmere"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,fxsr,idivq-to-divl,macrofusion,mmx,no-bypass-delay-mov,nopl,pclmul,popcnt,sahf,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("winchip-c6"), str_lit("32bit-mode,mmx,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("winchip2"), str_lit("32bit-mode,mmx,prfchw,slow-unaligned-mem-16,vzeroupper,x87") },
{ str_lit("x86-64"), str_lit("32bit-mode,64bit,cmov,cx8,fxsr,idivq-to-divl,macrofusion,mmx,nopl,slow-3ops-lea,slow-incdec,sse,sse2,vzeroupper,x87") },
{ str_lit("x86-64-v2"), str_lit("32bit-mode,64bit,cmov,crc32,cx16,cx8,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fxsr,idivq-to-divl,macrofusion,mmx,nopl,popcnt,sahf,slow-3ops-lea,slow-unaligned-mem-32,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87") },
{ str_lit("x86-64-v3"), str_lit("32bit-mode,64bit,allow-light-256-bit,avx,avx2,bmi,bmi2,cmov,crc32,cx16,cx8,f16c,false-deps-lzcnt-tzcnt,false-deps-popcnt,fast-15bytenop,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fma,fxsr,idivq-to-divl,lzcnt,macrofusion,mmx,movbe,nopl,popcnt,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave") },
{ str_lit("x86-64-v4"), str_lit("32bit-mode,64bit,allow-light-256-bit,avx,avx2,avx512bw,avx512cd,avx512dq,avx512f,avx512vl,bmi,bmi2,cmov,crc32,cx16,cx8,evex512,f16c,false-deps-popcnt,fast-15bytenop,fast-gather,fast-scalar-fsqrt,fast-shld-rotate,fast-variable-crosslane-shuffle,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fxsr,idivq-to-divl,lzcnt,macrofusion,mmx,movbe,nopl,popcnt,prefer-256-bit,sahf,slow-3ops-lea,sse,sse2,sse3,sse4.1,sse4.2,ssse3,vzeroupper,x87,xsave") },
{ str_lit("yonah"), str_lit("32bit-mode,cmov,cx8,fxsr,mmx,nopl,slow-unaligned-mem-16,sse,sse2,sse3,vzeroupper,x87") },
{ str_lit("znver1"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,branchfusion,clflushopt,clzero,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,mwaitx,nopl,pclmul,popcnt,prfchw,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vzeroupper,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver2"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fxsr,idivq-to-divl,lzcnt,mmx,movbe,mwaitx,nopl,pclmul,popcnt,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver3"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,f16c,fast-15bytenop,fast-bextr,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,mwaitx,nopl,pclmul,pku,popcnt,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver4"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vpopcntdq,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,evex512,f16c,fast-15bytenop,fast-bextr,fast-dpwssd,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,mwaitx,nopl,pclmul,pku,popcnt,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,shstk,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
{ str_lit("znver5"), str_lit("32bit-mode,64bit,adx,aes,allow-light-256-bit,avx,avx2,avx512bf16,avx512bitalg,avx512bw,avx512cd,avx512dq,avx512f,avx512ifma,avx512vbmi,avx512vbmi2,avx512vl,avx512vnni,avx512vp2intersect,avx512vpopcntdq,avxvnni,bmi,bmi2,branchfusion,clflushopt,clwb,clzero,cmov,crc32,cx16,cx8,evex512,f16c,fast-15bytenop,fast-bextr,fast-dpwssd,fast-imm16,fast-lzcnt,fast-movbe,fast-scalar-fsqrt,fast-scalar-shift-masks,fast-variable-perlane-shuffle,fast-vector-fsqrt,fma,fsgsbase,fsrm,fxsr,gfni,idivq-to-divl,invpcid,lzcnt,macrofusion,mmx,movbe,movdir64b,movdiri,mwaitx,nopl,pclmul,pku,popcnt,prefetchi,prfchw,rdpid,rdpru,rdrnd,rdseed,sahf,sbb-dep-breaking,sha,shstk,slow-shld,sse,sse2,sse3,sse4.1,sse4.2,sse4a,ssse3,vaes,vpclmulqdq,vzeroupper,wbnoinvd,x87,xsave,xsavec,xsaveopt,xsaves") },
// TargetArch_arm32:
{ str_lit("arm1020e"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm1020t"), str_lit("armv5t,v4t,v5t") },
{ str_lit("arm1022e"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm10e"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm10tdmi"), str_lit("armv5t,v4t,v5t") },
{ str_lit("arm1136j-s"), str_lit("armv6,dsp,v4t,v5t,v5te,v6") },
{ str_lit("arm1136jf-s"), str_lit("armv6,dsp,fp64,fpregs,fpregs64,slowfpvmlx,v4t,v5t,v5te,v6,vfp2,vfp2sp") },
{ str_lit("arm1156t2-s"), str_lit("armv6t2,dsp,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v8m") },
{ str_lit("arm1156t2f-s"), str_lit("armv6t2,dsp,fp64,fpregs,fpregs64,slowfpvmlx,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v8m,vfp2,vfp2sp") },
{ str_lit("arm1176jz-s"), str_lit("armv6kz,trustzone,v4t,v5t,v5te,v6,v6k") },
{ str_lit("arm1176jzf-s"), str_lit("armv6kz,fp64,fpregs,fpregs64,slowfpvmlx,trustzone,v4t,v5t,v5te,v6,v6k,vfp2,vfp2sp") },
{ str_lit("arm710t"), str_lit("armv4t,v4t") },
{ str_lit("arm720t"), str_lit("armv4t,v4t") },
{ str_lit("arm7tdmi"), str_lit("armv4t,v4t") },
{ str_lit("arm7tdmi-s"), str_lit("armv4t,v4t") },
{ str_lit("arm8"), str_lit("armv4") },
{ str_lit("arm810"), str_lit("armv4") },
{ str_lit("arm9"), str_lit("armv4t,v4t") },
{ str_lit("arm920"), str_lit("armv4t,v4t") },
{ str_lit("arm920t"), str_lit("armv4t,v4t") },
{ str_lit("arm922t"), str_lit("armv4t,v4t") },
{ str_lit("arm926ej-s"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm940t"), str_lit("armv4t,v4t") },
{ str_lit("arm946e-s"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm966e-s"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm968e-s"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm9e"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("arm9tdmi"), str_lit("armv4t,v4t") },
{ str_lit("cortex-a12"), str_lit("a12,aclass,armv7-a,avoid-partial-cpsr,d32,db,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,ret-addr-stack,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,vmlx-forwarding") },
{ str_lit("cortex-a15"), str_lit("a15,aclass,armv7-a,avoid-partial-cpsr,d32,db,dont-widen-vmovs,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,muxed-units,neon,perfmon,ret-addr-stack,splat-vfp-neon,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,vldn-align") },
{ str_lit("cortex-a17"), str_lit("a17,aclass,armv7-a,avoid-partial-cpsr,d32,db,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,ret-addr-stack,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,vmlx-forwarding") },
{ str_lit("cortex-a32"), str_lit("aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a35"), str_lit("a35,aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a5"), str_lit("a5,aclass,armv7-a,d32,db,dsp,fp16,fp64,fpregs,fpregs64,mp,neon,perfmon,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,vmlx-forwarding") },
{ str_lit("cortex-a510"), str_lit("aclass,acquire-release,armv9-a,bf16,cortex-a710,crc,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp16fml,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,i8mm,mp,neon,perfmon,ras,sb,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8m,v9a,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a53"), str_lit("a53,aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpao,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a55"), str_lit("a55,aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a57"), str_lit("a57,aclass,acquire-release,aes,armv8-a,avoid-partial-cpsr,cheap-predicable-cpsr,crc,crypto,d32,db,dsp,fix-cortex-a57-aes-1742098,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpao,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a7"), str_lit("a7,aclass,armv7-a,d32,db,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,vmlx-forwarding,vmlx-hazards") },
{ str_lit("cortex-a710"), str_lit("aclass,acquire-release,armv9-a,bf16,cortex-a710,crc,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp16fml,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,i8mm,mp,neon,perfmon,ras,sb,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8m,v9a,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a72"), str_lit("a72,aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dsp,fix-cortex-a57-aes-1742098,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a73"), str_lit("a73,aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a75"), str_lit("a75,aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a76"), str_lit("a76,aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a76ae"), str_lit("a76,aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a77"), str_lit("a77,aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a78"), str_lit("aclass,acquire-release,aes,armv8.2-a,cortex-a78,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a78ae"), str_lit("aclass,acquire-release,aes,armv8.2-a,cortex-a78ae,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a78c"), str_lit("a78c,aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-a8"), str_lit("a8,aclass,armv7-a,d32,db,dsp,fp64,fpregs,fpregs64,neon,nonpipelined-vfp,perfmon,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vmlx-forwarding,vmlx-hazards") },
{ str_lit("cortex-a9"), str_lit("a9,aclass,armv7-a,avoid-partial-cpsr,d32,db,dsp,expand-fp-mlx,fp16,fp64,fpregs,fpregs64,mp,muxed-units,neon,neon-fpmovs,perfmon,prefer-vmovsr,ret-addr-stack,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vldn-align,vmlx-forwarding,vmlx-hazards") },
{ str_lit("cortex-m0"), str_lit("armv6-m,db,mclass,no-branch-predictor,noarm,strict-align,thumb-mode,v4t,v5t,v5te,v6,v6m") },
{ str_lit("cortex-m0plus"), str_lit("armv6-m,db,mclass,no-branch-predictor,noarm,strict-align,thumb-mode,v4t,v5t,v5te,v6,v6m") },
{ str_lit("cortex-m1"), str_lit("armv6-m,db,mclass,no-branch-predictor,noarm,strict-align,thumb-mode,v4t,v5t,v5te,v6,v6m") },
{ str_lit("cortex-m23"), str_lit("8msecext,acquire-release,armv8-m.base,db,hwdiv,mclass,no-branch-predictor,no-movt,noarm,strict-align,thumb-mode,v4t,v5t,v5te,v6,v6m,v7clrex,v8m") },
{ str_lit("cortex-m3"), str_lit("armv7-m,db,hwdiv,loop-align,m3,mclass,no-branch-predictor,noarm,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m") },
{ str_lit("cortex-m33"), str_lit("8msecext,acquire-release,armv8-m.main,avoid-muls,db,dsp,fix-cmse-cve-2021-35465,fp-armv8d16sp,fp16,fpregs,hwdiv,loop-align,mclass,no-branch-predictor,noarm,slowfpvfmx,slowfpvmlx,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,v8m.main,vfp2sp,vfp3d16sp,vfp4d16sp") },
{ str_lit("cortex-m35p"), str_lit("8msecext,acquire-release,armv8-m.main,db,dsp,fix-cmse-cve-2021-35465,fp-armv8d16sp,fp16,fpregs,hwdiv,loop-align,mclass,no-branch-predictor,noarm,slowfpvfmx,slowfpvmlx,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,v8m.main,vfp2sp,vfp3d16sp,vfp4d16sp") },
{ str_lit("cortex-m4"), str_lit("armv7e-m,db,dsp,fp16,fpregs,hwdiv,loop-align,mclass,no-branch-predictor,noarm,slowfpvfmx,slowfpvmlx,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2sp,vfp3d16sp,vfp4d16sp") },
{ str_lit("cortex-m52"), str_lit("8msecext,acquire-release,armv8.1-m.main,db,dsp,fp-armv8d16,fp-armv8d16sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,lob,loop-align,mclass,mve,mve.fp,mve1beat,no-branch-predictor,noarm,pacbti,ras,slowfpvmlx,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8.1m.main,v8m,v8m.main,vfp2,vfp2sp,vfp3d16,vfp3d16sp,vfp4d16,vfp4d16sp") },
{ str_lit("cortex-m55"), str_lit("8msecext,acquire-release,armv8.1-m.main,db,dsp,fix-cmse-cve-2021-35465,fp-armv8d16,fp-armv8d16sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,lob,loop-align,m55,mclass,mve,mve.fp,no-branch-predictor,noarm,ras,slowfpvmlx,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8.1m.main,v8m,v8m.main,vfp2,vfp2sp,vfp3d16,vfp3d16sp,vfp4d16,vfp4d16sp") },
{ str_lit("cortex-m7"), str_lit("armv7e-m,branch-align-64,db,dsp,fp-armv8d16,fp-armv8d16sp,fp16,fp64,fpregs,fpregs64,hwdiv,m7,mclass,noarm,thumb-mode,thumb2,use-mipipeliner,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3d16,vfp3d16sp,vfp4d16,vfp4d16sp") },
{ str_lit("cortex-m85"), str_lit("8msecext,acquire-release,armv8.1-m.main,branch-align-64,db,dsp,fp-armv8d16,fp-armv8d16sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,lob,m85,mclass,mve,mve.fp,noarm,pacbti,ras,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8.1m.main,v8m,v8m.main,vfp2,vfp2sp,vfp3d16,vfp3d16sp,vfp4d16,vfp4d16sp") },
{ str_lit("cortex-r4"), str_lit("armv7-r,avoid-partial-cpsr,db,dsp,hwdiv,perfmon,r4,rclass,ret-addr-stack,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m") },
{ str_lit("cortex-r4f"), str_lit("armv7-r,avoid-partial-cpsr,db,dsp,fp64,fpregs,fpregs64,hwdiv,perfmon,r4,rclass,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3d16,vfp3d16sp") },
{ str_lit("cortex-r5"), str_lit("armv7-r,avoid-partial-cpsr,db,dsp,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,perfmon,r5,rclass,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3d16,vfp3d16sp") },
{ str_lit("cortex-r52"), str_lit("acquire-release,armv8-r,crc,d32,db,dfb,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpao,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,r52,rclass,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-r52plus"), str_lit("acquire-release,armv8-r,crc,d32,db,dfb,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpao,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,r52plus,rclass,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-r7"), str_lit("armv7-r,avoid-partial-cpsr,db,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,perfmon,r7,rclass,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3d16,vfp3d16sp") },
{ str_lit("cortex-r8"), str_lit("armv7-r,avoid-partial-cpsr,db,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,perfmon,rclass,ret-addr-stack,slow-fp-brcc,slowfpvfmx,slowfpvmlx,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3d16,vfp3d16sp") },
{ str_lit("cortex-x1"), str_lit("aclass,acquire-release,aes,armv8.2-a,cortex-x1,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cortex-x1c"), str_lit("aclass,acquire-release,aes,armv8.2-a,cortex-x1c,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("cyclone"), str_lit("aclass,acquire-release,aes,armv8-a,avoid-movs-shop,avoid-partial-cpsr,crc,crypto,d32,db,disable-postra-scheduler,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,neonfp,perfmon,ret-addr-stack,sha2,slowfpvfmx,slowfpvmlx,swift,thumb2,trustzone,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,zcz") },
{ str_lit("ep9312"), str_lit("armv4t,v4t") },
{ str_lit("exynos-m3"), str_lit("aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dont-widen-vmovs,dsp,expand-fp-mlx,exynos,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,fuse-aes,fuse-literals,hwdiv,hwdiv-arm,mp,neon,perfmon,prof-unpr,ret-addr-stack,sha2,slow-fp-brcc,slow-vdup32,slow-vgetlni32,slowfpvfmx,slowfpvmlx,splat-vfp-neon,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,wide-stride-vfp,zcz") },
{ str_lit("exynos-m4"), str_lit("aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dont-widen-vmovs,dotprod,dsp,expand-fp-mlx,exynos,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,fuse-aes,fuse-literals,hwdiv,hwdiv-arm,mp,neon,perfmon,prof-unpr,ras,ret-addr-stack,sha2,slow-fp-brcc,slow-vdup32,slow-vgetlni32,slowfpvfmx,slowfpvmlx,splat-vfp-neon,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,wide-stride-vfp,zcz") },
{ str_lit("exynos-m5"), str_lit("aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dont-widen-vmovs,dotprod,dsp,expand-fp-mlx,exynos,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,fuse-aes,fuse-literals,hwdiv,hwdiv-arm,mp,neon,perfmon,prof-unpr,ras,ret-addr-stack,sha2,slow-fp-brcc,slow-vdup32,slow-vgetlni32,slowfpvfmx,slowfpvmlx,splat-vfp-neon,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization,wide-stride-vfp,zcz") },
{ str_lit("generic"), str_lit("") },
{ str_lit("iwmmxt"), str_lit("armv5te,v4t,v5t,v5te") },
{ str_lit("krait"), str_lit("aclass,armv7-a,avoid-partial-cpsr,d32,db,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,krait,muxed-units,neon,perfmon,ret-addr-stack,thumb2,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,vldn-align,vmlx-forwarding") },
{ str_lit("kryo"), str_lit("aclass,acquire-release,aes,armv8-a,crc,crypto,d32,db,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,kryo,mp,neon,perfmon,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("mpcore"), str_lit("armv6k,fp64,fpregs,fpregs64,slowfpvmlx,v4t,v5t,v5te,v6,v6k,vfp2,vfp2sp") },
{ str_lit("mpcorenovfp"), str_lit("armv6k,v4t,v5t,v5te,v6,v6k") },
{ str_lit("neoverse-n1"), str_lit("aclass,acquire-release,aes,armv8.2-a,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("neoverse-n2"), str_lit("aclass,acquire-release,armv9-a,bf16,crc,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp16fml,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,i8mm,mp,neon,perfmon,ras,sb,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8m,v9a,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("neoverse-v1"), str_lit("aclass,acquire-release,aes,armv8.4-a,bf16,crc,crypto,d32,db,dotprod,dsp,fp-armv8,fp-armv8d16,fp-armv8d16sp,fp-armv8sp,fp16,fp64,fpregs,fpregs16,fpregs64,fullfp16,hwdiv,hwdiv-arm,i8mm,mp,neon,perfmon,ras,sha2,thumb2,trustzone,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8,v8.1a,v8.2a,v8.3a,v8.4a,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,virtualization") },
{ str_lit("sc000"), str_lit("armv6-m,db,mclass,no-branch-predictor,noarm,strict-align,thumb-mode,v4t,v5t,v5te,v6,v6m") },
{ str_lit("sc300"), str_lit("armv7-m,db,hwdiv,m3,mclass,no-branch-predictor,noarm,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m") },
{ str_lit("star-mc1"), str_lit("8msecext,acquire-release,armv8-m.main,db,dsp,fix-cmse-cve-2021-35465,fp-armv8d16sp,fp16,fpregs,hwdiv,loop-align,mclass,no-branch-predictor,noarm,slowfpvfmx,slowfpvmlx,thumb-mode,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,v8m.main,vfp2sp,vfp3d16sp,vfp4d16sp") },
{ str_lit("strongarm"), str_lit("armv4") },
{ str_lit("strongarm110"), str_lit("armv4") },
{ str_lit("strongarm1100"), str_lit("armv4") },
{ str_lit("strongarm1110"), str_lit("armv4") },
{ str_lit("swift"), str_lit("aclass,armv7-a,avoid-movs-shop,avoid-partial-cpsr,d32,db,disable-postra-scheduler,dsp,fp16,fp64,fpregs,fpregs64,hwdiv,hwdiv-arm,mp,neon,neonfp,perfmon,prefer-ishst,prof-unpr,ret-addr-stack,slow-load-D-subreg,slow-odd-reg,slow-vdup32,slow-vgetlni32,slowfpvfmx,slowfpvmlx,swift,thumb2,use-misched,v4t,v5t,v5te,v6,v6k,v6m,v6t2,v7,v7clrex,v8m,vfp2,vfp2sp,vfp3,vfp3d16,vfp3d16sp,vfp3sp,vfp4,vfp4d16,vfp4d16sp,vfp4sp,vmlx-hazards,wide-stride-vfp") },
{ str_lit("xscale"), str_lit("armv5te,v4t,v5t,v5te") },
// TargetArch_arm64:
{ str_lit("a64fx"), str_lit("CONTEXTIDREL2,a64fx,aes,aggressive-fma,arith-bcc-fusion,ccpp,complxnum,crc,el2vmsa,el3,fp-armv8,fullfp16,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rdm,sha2,store-pair-suppress,sve,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("ampere1"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,altnzcv,alu-lsl-fast,am,ampere1,amvs,arith-bcc-fusion,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-literals,i8mm,jsconv,ldp-aligned-only,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,stp-aligned-only,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh") },
{ str_lit("ampere1a"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,altnzcv,alu-lsl-fast,am,ampere1a,amvs,arith-bcc-fusion,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fptoint,fullfp16,fuse-address,fuse-addsub-2reg-const1,fuse-adrp-add,fuse-aes,fuse-literals,i8mm,jsconv,ldp-aligned-only,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,sm4,specrestrict,ssbs,store-pair-suppress,stp-aligned-only,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh") },
{ str_lit("ampere1b"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,altnzcv,alu-lsl-fast,am,ampere1b,amvs,arith-bcc-fusion,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,cssc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,fgt,flagm,fp-armv8,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-literals,hcx,i8mm,jsconv,ldp-aligned-only,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,sm4,specrestrict,ssbs,store-pair-suppress,stp-aligned-only,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,vh,wfxt,xs") },
{ str_lit("apple-a10"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,apple-a10,arith-bcc-fusion,arith-cbz-fusion,crc,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fuse-aes,fuse-crypto-eor,lor,neon,pan,perfmon,rdm,sha2,store-pair-suppress,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a11"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,apple-a11,arith-bcc-fusion,arith-cbz-fusion,ccpp,crc,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fullfp16,fuse-aes,fuse-crypto-eor,lor,lse,neon,pan,pan-rwv,perfmon,ras,rdm,sha2,store-pair-suppress,uaops,v8.1a,v8.2a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a12"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,apple-a12,arith-bcc-fusion,arith-cbz-fusion,ccpp,complxnum,crc,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fullfp16,fuse-aes,fuse-crypto-eor,jsconv,lor,lse,neon,pan,pan-rwv,pauth,perfmon,ras,rcpc,rdm,sha2,store-pair-suppress,uaops,v8.1a,v8.2a,v8.3a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a13"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,am,apple-a13,arith-bcc-fusion,arith-cbz-fusion,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,el2vmsa,el3,flagm,fp-armv8,fp16fml,fullfp16,fuse-aes,fuse-crypto-eor,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,ras,rcpc,rcpc-immo,rdm,sel2,sha2,sha3,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a14"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,alternate-sextload-cvt-f32-pattern,altnzcv,am,apple-a14,arith-bcc-fusion,arith-cbz-fusion,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,el2vmsa,el3,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-address,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a15"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,altnzcv,am,amvs,apple-a15,arith-bcc-fusion,arith-cbz-fusion,bf16,bti,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fp16fml,fpac,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a16"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,altnzcv,am,amvs,apple-a16,arith-bcc-fusion,arith-cbz-fusion,bf16,bti,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fp16fml,fpac,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,hcx,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a17"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,altnzcv,am,amvs,apple-a17,arith-bcc-fusion,arith-cbz-fusion,bf16,bti,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fp16fml,fpac,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,hcx,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-a7"), str_lit("aes,alternate-sextload-cvt-f32-pattern,apple-a7,arith-bcc-fusion,arith-cbz-fusion,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fuse-aes,fuse-crypto-eor,neon,perfmon,sha2,store-pair-suppress,v8a,zcm,zcz,zcz-fp-workaround,zcz-gp") },
{ str_lit("apple-a8"), str_lit("aes,alternate-sextload-cvt-f32-pattern,apple-a7,arith-bcc-fusion,arith-cbz-fusion,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fuse-aes,fuse-crypto-eor,neon,perfmon,sha2,store-pair-suppress,v8a,zcm,zcz,zcz-fp-workaround,zcz-gp") },
{ str_lit("apple-a9"), str_lit("aes,alternate-sextload-cvt-f32-pattern,apple-a7,arith-bcc-fusion,arith-cbz-fusion,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fuse-aes,fuse-crypto-eor,neon,perfmon,sha2,store-pair-suppress,v8a,zcm,zcz,zcz-fp-workaround,zcz-gp") },
{ str_lit("apple-m1"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,alternate-sextload-cvt-f32-pattern,altnzcv,am,apple-a14,arith-bcc-fusion,arith-cbz-fusion,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,el2vmsa,el3,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-address,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-m2"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,altnzcv,am,amvs,apple-a15,arith-bcc-fusion,arith-cbz-fusion,bf16,bti,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fp16fml,fpac,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-m3"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,altnzcv,am,amvs,apple-a16,arith-bcc-fusion,arith-cbz-fusion,bf16,bti,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fp16fml,fpac,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,hcx,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,specrestrict,ssbs,store-pair-suppress,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-m4"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,altnzcv,am,amvs,apple-m4,arith-bcc-fusion,arith-cbz-fusion,bf16,bti,ccdp,ccpp,complxnum,crc,disable-latency-sched-heuristic,dit,dotprod,ecv,el2vmsa,el3,fgt,flagm,fp-armv8,fp16fml,fpac,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-crypto-eor,fuse-csel,fuse-literals,hcx,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,sme,sme-f64f64,sme-i16i64,sme2,specrestrict,tlb-rmi,tracev8.4,uaops,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,vh,wfxt,xs,zcm,zcz,zcz-gp") },
{ str_lit("apple-s4"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,apple-a12,arith-bcc-fusion,arith-cbz-fusion,ccpp,complxnum,crc,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fullfp16,fuse-aes,fuse-crypto-eor,jsconv,lor,lse,neon,pan,pan-rwv,pauth,perfmon,ras,rcpc,rdm,sha2,store-pair-suppress,uaops,v8.1a,v8.2a,v8.3a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("apple-s5"), str_lit("CONTEXTIDREL2,aes,alternate-sextload-cvt-f32-pattern,apple-a12,arith-bcc-fusion,arith-cbz-fusion,ccpp,complxnum,crc,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fullfp16,fuse-aes,fuse-crypto-eor,jsconv,lor,lse,neon,pan,pan-rwv,pauth,perfmon,ras,rcpc,rdm,sha2,store-pair-suppress,uaops,v8.1a,v8.2a,v8.3a,v8a,vh,zcm,zcz,zcz-gp") },
{ str_lit("carmel"), str_lit("CONTEXTIDREL2,aes,carmel,ccpp,crc,el2vmsa,el3,fp-armv8,fullfp16,lor,lse,neon,pan,pan-rwv,ras,rdm,sha2,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cobalt-100"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,neoversen2,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("cortex-a34"), str_lit("a35,aes,crc,el2vmsa,el3,fp-armv8,neon,perfmon,sha2,v8a") },
{ str_lit("cortex-a35"), str_lit("a35,aes,crc,el2vmsa,el3,fp-armv8,neon,perfmon,sha2,v8a") },
{ str_lit("cortex-a510"), str_lit("CONTEXTIDREL2,a510,altnzcv,am,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,el2vmsa,el3,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("cortex-a520"), str_lit("CONTEXTIDREL2,a520,altnzcv,am,amvs,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cortex-a520ae"), str_lit("CONTEXTIDREL2,a520ae,altnzcv,am,amvs,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cortex-a53"), str_lit("a53,aes,balance-fp-ops,crc,el2vmsa,el3,fp-armv8,fuse-adrp-add,fuse-aes,neon,perfmon,sha2,use-postra-scheduler,v8a") },
{ str_lit("cortex-a55"), str_lit("CONTEXTIDREL2,a55,aes,ccpp,crc,dotprod,el2vmsa,el3,fp-armv8,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,ras,rcpc,rdm,sha2,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a57"), str_lit("a57,addr-lsl-slow-14,aes,balance-fp-ops,crc,el2vmsa,el3,enable-select-opt,fp-armv8,fuse-adrp-add,fuse-aes,fuse-literals,neon,perfmon,predictable-select-expensive,sha2,use-postra-scheduler,v8a") },
{ str_lit("cortex-a65"), str_lit("CONTEXTIDREL2,a65,aes,ccpp,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-literals,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,ssbs,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a65ae"), str_lit("CONTEXTIDREL2,a65,aes,ccpp,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-literals,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,ssbs,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a710"), str_lit("CONTEXTIDREL2,a710,altnzcv,alu-lsl-fast,am,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("cortex-a715"), str_lit("CONTEXTIDREL2,a715,altnzcv,alu-lsl-fast,am,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("cortex-a72"), str_lit("a72,addr-lsl-slow-14,aes,crc,el2vmsa,el3,enable-select-opt,fp-armv8,fuse-adrp-add,fuse-aes,fuse-literals,neon,perfmon,predictable-select-expensive,sha2,v8a") },
{ str_lit("cortex-a720"), str_lit("CONTEXTIDREL2,a720,altnzcv,alu-lsl-fast,am,amvs,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cortex-a720ae"), str_lit("CONTEXTIDREL2,a720ae,altnzcv,alu-lsl-fast,am,amvs,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cortex-a725"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,amvs,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,cortex-a725,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cortex-a73"), str_lit("a73,addr-lsl-slow-14,aes,crc,el2vmsa,el3,enable-select-opt,fp-armv8,fuse-adrp-add,fuse-aes,neon,perfmon,predictable-select-expensive,sha2,v8a") },
{ str_lit("cortex-a75"), str_lit("CONTEXTIDREL2,a75,addr-lsl-slow-14,aes,ccpp,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a76"), str_lit("CONTEXTIDREL2,a76,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,ssbs,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a76ae"), str_lit("CONTEXTIDREL2,a76,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,ssbs,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a77"), str_lit("CONTEXTIDREL2,a77,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,cmp-bcc-fusion,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,ssbs,uaops,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a78"), str_lit("CONTEXTIDREL2,a78,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,cmp-bcc-fusion,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,spe,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a78ae"), str_lit("CONTEXTIDREL2,a78ae,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,cmp-bcc-fusion,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,spe,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-a78c"), str_lit("CONTEXTIDREL2,a78c,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,cmp-bcc-fusion,crc,dotprod,el2vmsa,el3,enable-select-opt,flagm,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,spe,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-r82"), str_lit("CONTEXTIDREL2,ccdp,ccpp,complxnum,cortex-r82,crc,dit,dotprod,flagm,fp-armv8,fp16fml,fullfp16,jsconv,lse,neon,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8r") },
{ str_lit("cortex-r82ae"), str_lit("CONTEXTIDREL2,ccdp,ccpp,complxnum,cortex-r82ae,crc,dit,dotprod,flagm,fp-armv8,fp16fml,fullfp16,jsconv,lse,neon,pan,pan-rwv,pauth,perfmon,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8r") },
{ str_lit("cortex-x1"), str_lit("CONTEXTIDREL2,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,cmp-bcc-fusion,cortex-x1,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,spe,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-x1c"), str_lit("CONTEXTIDREL2,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,cmp-bcc-fusion,cortex-x1,crc,dotprod,el2vmsa,el3,enable-select-opt,flagm,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,lse2,neon,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,ras,rcpc,rcpc-immo,rdm,sha2,spe,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("cortex-x2"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,cortex-x2,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("cortex-x3"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,avoid-ldapur,bf16,bti,ccdp,ccidx,ccpp,complxnum,cortex-x3,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("cortex-x4"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,amvs,avoid-ldapur,bf16,bti,ccdp,ccidx,ccpp,complxnum,cortex-x4,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cortex-x925"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,amvs,avoid-ldapur,bf16,bti,ccdp,ccidx,ccpp,complxnum,cortex-x925,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("cyclone"), str_lit("aes,alternate-sextload-cvt-f32-pattern,apple-a7,arith-bcc-fusion,arith-cbz-fusion,disable-latency-sched-heuristic,el2vmsa,el3,fp-armv8,fuse-aes,fuse-crypto-eor,neon,perfmon,sha2,store-pair-suppress,v8a,zcm,zcz,zcz-fp-workaround,zcz-gp") },
{ str_lit("exynos-m3"), str_lit("aes,alu-lsl-fast,crc,el2vmsa,el3,exynos-cheap-as-move,exynosm3,force-32bit-jump-tables,fp-armv8,fuse-address,fuse-adrp-add,fuse-aes,fuse-csel,fuse-literals,neon,perfmon,predictable-select-expensive,sha2,store-pair-suppress,use-postra-scheduler,v8a") },
{ str_lit("exynos-m4"), str_lit("CONTEXTIDREL2,aes,alu-lsl-fast,arith-bcc-fusion,arith-cbz-fusion,ccpp,crc,dotprod,el2vmsa,el3,exynos-cheap-as-move,exynosm4,force-32bit-jump-tables,fp-armv8,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-csel,fuse-literals,lor,lse,neon,pan,pan-rwv,perfmon,ras,rdm,sha2,store-pair-suppress,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh,zcz,zcz-gp") },
{ str_lit("exynos-m5"), str_lit("CONTEXTIDREL2,aes,alu-lsl-fast,arith-bcc-fusion,arith-cbz-fusion,ccpp,crc,dotprod,el2vmsa,el3,exynos-cheap-as-move,exynosm4,force-32bit-jump-tables,fp-armv8,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-arith-logic,fuse-csel,fuse-literals,lor,lse,neon,pan,pan-rwv,perfmon,ras,rdm,sha2,store-pair-suppress,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh,zcz,zcz-gp") },
{ str_lit("falkor"), str_lit("aes,alu-lsl-fast,crc,el2vmsa,el3,falkor,fp-armv8,neon,perfmon,predictable-select-expensive,rdm,sha2,slow-strqro-store,store-pair-suppress,use-postra-scheduler,v8a,zcz,zcz-gp") },
{ str_lit("fujitsu-monaka"), str_lit("CONTEXTIDREL2,aes,altnzcv,am,amvs,arith-bcc-fusion,bf16,bti,ccdp,ccidx,ccpp,clrbhb,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,faminmax,fgt,flagm,fp-armv8,fp16fml,fp8,fp8dot2,fpac,fptoint,fujitsu-monaka,fullfp16,hbc,hcx,i8mm,jsconv,lor,ls64,lse,lse2,lut,mec,mops,mpam,neon,nmi,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,rme,sb,sel2,sha2,sha3,sm4,specres2,specrestrict,ssbs,sve,sve-aes,sve-bitperm,sve2,sve2-sha3,sve2-sm4,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8.8a,v8a,v9.1a,v9.2a,v9.3a,v9a,vh,wfxt,xs") },
{ str_lit("generic"), str_lit("enable-select-opt,ete,fp-armv8,fuse-adrp-add,fuse-aes,neon,trbe,use-postra-scheduler") },
{ str_lit("grace"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,avoid-ldapur,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,neoversev2,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("kryo"), str_lit("aes,alu-lsl-fast,crc,el2vmsa,el3,fp-armv8,kryo,neon,perfmon,predictable-select-expensive,sha2,store-pair-suppress,use-postra-scheduler,v8a,zcz,zcz-gp") },
{ str_lit("neoverse-512tvb"), str_lit("CONTEXTIDREL2,aes,alu-lsl-fast,am,bf16,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,flagm,fp-armv8,fp16fml,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,neon,neoverse512tvb,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,rand,ras,rcpc,rcpc-immo,rdm,sel2,sha2,sha3,sm4,spe,ssbs,sve,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8a,vh") },
{ str_lit("neoverse-e1"), str_lit("CONTEXTIDREL2,aes,ccpp,crc,dotprod,el2vmsa,el3,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,neoversee1,pan,pan-rwv,perfmon,ras,rcpc,rdm,sha2,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("neoverse-n1"), str_lit("CONTEXTIDREL2,addr-lsl-slow-14,aes,alu-lsl-fast,ccpp,crc,dotprod,el2vmsa,el3,enable-select-opt,fp-armv8,fullfp16,fuse-adrp-add,fuse-aes,lor,lse,neon,neoversen1,pan,pan-rwv,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,spe,ssbs,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
{ str_lit("neoverse-n2"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,neoversen2,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,ras,rcpc,rcpc-immo,rdm,sb,sel2,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("neoverse-n3"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,amvs,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,neoversen3,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("neoverse-v1"), str_lit("CONTEXTIDREL2,addr-lsl-slow-14,aes,alu-lsl-fast,am,bf16,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,flagm,fp-armv8,fp16fml,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,neon,neoversev1,no-sve-fp-ld1r,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,rand,ras,rcpc,rcpc-immo,rdm,sel2,sha2,sha3,sm4,spe,ssbs,sve,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8a,vh") },
{ str_lit("neoverse-v2"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,avoid-ldapur,bf16,bti,ccdp,ccidx,ccpp,cmp-bcc-fusion,complxnum,crc,dit,dotprod,el2vmsa,el3,enable-select-opt,ete,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,i8mm,jsconv,lor,lse,lse2,mpam,mte,neon,neoversev2,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,spe,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-fixed-over-scalable-if-equal-cost,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8a,v9a,vh") },
{ str_lit("neoverse-v3"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,amvs,avoid-ldapur,bf16,brbe,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,ls64,lse,lse2,mpam,mte,neon,neoversev3,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,rme,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("neoverse-v3ae"), str_lit("CONTEXTIDREL2,altnzcv,alu-lsl-fast,am,amvs,avoid-ldapur,bf16,brbe,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,ete,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-adrp-add,fuse-aes,hcx,i8mm,jsconv,lor,ls64,lse,lse2,mpam,mte,neon,neoversev3AE,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,predres,rand,ras,rcpc,rcpc-immo,rdm,rme,sb,sel2,spe,spe-eef,specrestrict,ssbs,sve,sve-bitperm,sve2,tlb-rmi,tracev8.4,trbe,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8.7a,v8a,v9.1a,v9.2a,v9a,vh,wfxt,xs") },
{ str_lit("oryon-1"), str_lit("CONTEXTIDREL2,aes,altnzcv,am,amvs,bf16,bti,ccdp,ccidx,ccpp,complxnum,crc,dit,dotprod,ecv,el2vmsa,el3,enable-select-opt,fgt,flagm,fp-armv8,fp16fml,fptoint,fullfp16,fuse-address,fuse-adrp-add,fuse-aes,fuse-crypto-eor,i8mm,jsconv,lor,lse,lse2,mpam,neon,nv,oryon-1,pan,pan-rwv,pauth,perfmon,predres,rand,ras,rcpc,rcpc-immo,rdm,sb,sel2,sha2,sha3,sm4,spe,specrestrict,ssbs,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8.5a,v8.6a,v8a,vh") },
{ str_lit("saphira"), str_lit("CONTEXTIDREL2,aes,alu-lsl-fast,am,ccidx,ccpp,complxnum,crc,dit,dotprod,el2vmsa,el3,flagm,fp-armv8,jsconv,lor,lse,lse2,mpam,neon,nv,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,ras,rcpc,rcpc-immo,rdm,saphira,sel2,sha2,spe,store-pair-suppress,tlb-rmi,tracev8.4,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8.4a,v8a,vh,zcz,zcz-gp") },
{ str_lit("thunderx"), str_lit("aes,crc,el2vmsa,el3,fp-armv8,neon,perfmon,predictable-select-expensive,sha2,store-pair-suppress,thunderx,use-postra-scheduler,v8a") },
{ str_lit("thunderx2t99"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,arith-bcc-fusion,crc,el2vmsa,el3,fp-armv8,lor,lse,neon,pan,predictable-select-expensive,rdm,sha2,store-pair-suppress,thunderx2t99,use-postra-scheduler,v8.1a,v8a,vh") },
{ str_lit("thunderx3t110"), str_lit("CONTEXTIDREL2,aes,aggressive-fma,arith-bcc-fusion,balance-fp-ops,ccidx,ccpp,complxnum,crc,el2vmsa,el3,fp-armv8,jsconv,lor,lse,neon,pan,pan-rwv,pauth,perfmon,predictable-select-expensive,ras,rcpc,rdm,sha2,store-pair-suppress,strict-align,thunderx3t110,uaops,use-postra-scheduler,v8.1a,v8.2a,v8.3a,v8a,vh") },
{ str_lit("thunderxt81"), str_lit("aes,crc,el2vmsa,el3,fp-armv8,neon,perfmon,predictable-select-expensive,sha2,store-pair-suppress,thunderxt81,use-postra-scheduler,v8a") },
{ str_lit("thunderxt83"), str_lit("aes,crc,el2vmsa,el3,fp-armv8,neon,perfmon,predictable-select-expensive,sha2,store-pair-suppress,thunderxt83,use-postra-scheduler,v8a") },
{ str_lit("thunderxt88"), str_lit("aes,crc,el2vmsa,el3,fp-armv8,neon,perfmon,predictable-select-expensive,sha2,store-pair-suppress,thunderxt88,use-postra-scheduler,v8a") },
{ str_lit("tsv110"), str_lit("CONTEXTIDREL2,aes,ccpp,complxnum,crc,dotprod,el2vmsa,el3,fp-armv8,fp16fml,fullfp16,fuse-aes,jsconv,lor,lse,neon,pan,pan-rwv,perfmon,ras,rdm,sha2,spe,store-pair-suppress,tsv110,uaops,use-postra-scheduler,v8.1a,v8.2a,v8a,vh") },
// TargetArch_wasm32:
{ str_lit("bleeding-edge"), str_lit("atomics,bulk-memory,bulk-memory-opt,call-indirect-overlong,exception-handling,extended-const,fp16,multimemory,multivalue,mutable-globals,nontrapping-fptoint,reference-types,relaxed-simd,sign-ext,simd128,tail-call") },
{ str_lit("generic"), str_lit("bulk-memory,bulk-memory-opt,call-indirect-overlong,multivalue,mutable-globals,nontrapping-fptoint,reference-types,sign-ext") },
{ str_lit("lime1"), str_lit("bulk-memory-opt,call-indirect-overlong,extended-const,multivalue,mutable-globals,nontrapping-fptoint,sign-ext") },
{ str_lit("mvp"), str_lit("") },
// TargetArch_wasm64p32:
{ str_lit("bleeding-edge"), str_lit("atomics,bulk-memory,bulk-memory-opt,call-indirect-overlong,exception-handling,extended-const,fp16,multimemory,multivalue,mutable-globals,nontrapping-fptoint,reference-types,relaxed-simd,sign-ext,simd128,tail-call") },
{ str_lit("generic"), str_lit("bulk-memory,bulk-memory-opt,call-indirect-overlong,multivalue,mutable-globals,nontrapping-fptoint,reference-types,sign-ext") },
{ str_lit("lime1"), str_lit("bulk-memory-opt,call-indirect-overlong,extended-const,multivalue,mutable-globals,nontrapping-fptoint,sign-ext") },
{ str_lit("mvp"), str_lit("") },
// TargetArch_riscv64:
{ str_lit("generic"), str_lit("64bit,i,optimized-nf2-segment-load-store") },
{ str_lit("generic-rv32"), str_lit("32bit,i,optimized-nf2-segment-load-store") },
{ str_lit("generic-rv64"), str_lit("64bit,i,optimized-nf2-segment-load-store") },
{ str_lit("mips-p8700"), str_lit("64bit,a,c,d,f,i,m,mips-p8700,xmipscmove,xmipslsp,zaamo,zalrsc,zba,zbb,zicsr,zifencei,zmmul") },
{ str_lit("rocket"), str_lit("") },
{ str_lit("rocket-rv32"), str_lit("32bit,i,zicsr,zifencei") },
{ str_lit("rocket-rv64"), str_lit("64bit,i,zicsr,zifencei") },
{ str_lit("rp2350-hazard3"), str_lit("32bit,a,c,i,m,zaamo,zalrsc,zba,zbb,zbkb,zbs,zca,zcb,zcmp,zicsr,zifencei,zmmul") },
{ str_lit("sifive-7-series"), str_lit("no-default-unroll,short-forward-branch-opt,sifive7,use-postra-scheduler") },
{ str_lit("sifive-e20"), str_lit("32bit,c,i,m,zicsr,zifencei,zmmul") },
{ str_lit("sifive-e21"), str_lit("32bit,a,c,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-e24"), str_lit("32bit,a,c,f,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-e31"), str_lit("32bit,a,c,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-e34"), str_lit("32bit,a,c,f,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-e76"), str_lit("32bit,a,c,f,i,m,no-default-unroll,short-forward-branch-opt,sifive7,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-p450"), str_lit("64bit,a,auipc-addi-fusion,b,c,conditional-cmv-fusion,d,f,i,lui-addi-fusion,m,no-default-unroll,unaligned-scalar-mem,unaligned-vector-mem,use-postra-scheduler,za64rs,zaamo,zalrsc,zba,zbb,zbs,zfhmin,zic64b,zicbom,zicbop,zicboz,ziccamoa,ziccif,zicclsm,ziccrse,zicntr,zicsr,zifencei,zihintntl,zihintpause,zihpm,zkt,zmmul") },
{ str_lit("sifive-p470"), str_lit("64bit,a,auipc-addi-fusion,b,c,conditional-cmv-fusion,d,f,i,lui-addi-fusion,m,no-default-unroll,no-sink-splat-operands,unaligned-scalar-mem,unaligned-vector-mem,use-postra-scheduler,v,vxrm-pipeline-flush,xsifivecdiscarddlone,xsifivecflushdlone,za64rs,zaamo,zalrsc,zba,zbb,zbs,zfhmin,zic64b,zicbom,zicbop,zicboz,ziccamoa,ziccif,zicclsm,ziccrse,zicntr,zicsr,zifencei,zihintntl,zihintpause,zihpm,zkt,zmmul,zvbb,zvbc,zve32f,zve32x,zve64d,zve64f,zve64x,zvkb,zvkg,zvkn,zvknc,zvkned,zvkng,zvknhb,zvks,zvksc,zvksed,zvksg,zvksh,zvkt,zvl128b,zvl32b,zvl64b") },
{ str_lit("sifive-p550"), str_lit("64bit,a,auipc-addi-fusion,c,conditional-cmv-fusion,d,f,i,lui-addi-fusion,m,no-default-unroll,use-postra-scheduler,zaamo,zalrsc,zba,zbb,zicsr,zifencei,zmmul") },
{ str_lit("sifive-p670"), str_lit("64bit,a,auipc-addi-fusion,b,c,conditional-cmv-fusion,d,f,i,lui-addi-fusion,m,no-default-unroll,no-sink-splat-operands,unaligned-scalar-mem,unaligned-vector-mem,use-postra-scheduler,v,vxrm-pipeline-flush,za64rs,zaamo,zalrsc,zba,zbb,zbs,zfhmin,zic64b,zicbom,zicbop,zicboz,ziccamoa,ziccif,zicclsm,ziccrse,zicntr,zicsr,zifencei,zihintntl,zihintpause,zihpm,zkt,zmmul,zvbb,zvbc,zve32f,zve32x,zve64d,zve64f,zve64x,zvkb,zvkg,zvkn,zvknc,zvkned,zvkng,zvknhb,zvks,zvksc,zvksed,zvksg,zvksh,zvkt,zvl128b,zvl32b,zvl64b") },
{ str_lit("sifive-s21"), str_lit("64bit,a,c,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-s51"), str_lit("64bit,a,c,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-s54"), str_lit("64bit,a,c,d,f,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-s76"), str_lit("64bit,a,c,d,f,i,m,no-default-unroll,short-forward-branch-opt,sifive7,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zihintpause,zmmul") },
{ str_lit("sifive-u54"), str_lit("64bit,a,c,d,f,i,m,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-u74"), str_lit("64bit,a,c,d,f,i,m,no-default-unroll,short-forward-branch-opt,sifive7,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("sifive-x280"), str_lit("64bit,a,c,d,dlen-factor-2,f,i,m,no-default-unroll,optimized-nf2-segment-load-store,optimized-zero-stride-load,short-forward-branch-opt,sifive7,use-postra-scheduler,v,zaamo,zalrsc,zba,zbb,zfh,zfhmin,zicsr,zifencei,zmmul,zve32f,zve32x,zve64d,zve64f,zve64x,zvfh,zvfhmin,zvl128b,zvl256b,zvl32b,zvl512b,zvl64b") },
{ str_lit("spacemit-x60"), str_lit("64bit,a,b,c,d,dlen-factor-2,f,i,m,optimized-nf2-segment-load-store,optimized-nf3-segment-load-store,optimized-nf4-segment-load-store,ssccptr,sscofpmf,sscounterenw,sstc,sstvala,sstvecd,svade,svbare,svinval,svnapot,svpbmt,unaligned-scalar-mem,v,vxrm-pipeline-flush,za64rs,zaamo,zalrsc,zba,zbb,zbc,zbkc,zbs,zfh,zfhmin,zic64b,zicbom,zicbop,zicboz,ziccamoa,ziccif,zicclsm,ziccrse,zicntr,zicond,zicsr,zifencei,zihintpause,zihpm,zkt,zmmul,zve32f,zve32x,zve64d,zve64f,zve64x,zvfh,zvfhmin,zvkt,zvl128b,zvl256b,zvl32b,zvl64b") },
{ str_lit("syntacore-scr1-base"), str_lit("32bit,c,i,no-default-unroll,zicsr,zifencei") },
{ str_lit("syntacore-scr1-max"), str_lit("32bit,c,i,m,no-default-unroll,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr3-rv32"), str_lit("32bit,c,i,m,no-default-unroll,use-postra-scheduler,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr3-rv64"), str_lit("64bit,a,c,i,m,no-default-unroll,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr4-rv32"), str_lit("32bit,c,d,f,i,m,no-default-unroll,use-postra-scheduler,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr4-rv64"), str_lit("64bit,a,c,d,f,i,m,no-default-unroll,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr5-rv32"), str_lit("32bit,a,c,d,f,i,m,no-default-unroll,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr5-rv64"), str_lit("64bit,a,c,d,f,i,m,no-default-unroll,use-postra-scheduler,zaamo,zalrsc,zicsr,zifencei,zmmul") },
{ str_lit("syntacore-scr7"), str_lit("64bit,a,c,d,f,i,m,no-default-unroll,use-postra-scheduler,v,zaamo,zalrsc,zba,zbb,zbc,zbkb,zbkc,zbkx,zbs,zicsr,zifencei,zkn,zknd,zkne,zknh,zmmul,zve32f,zve32x,zve64d,zve64f,zve64x,zvl128b,zvl32b,zvl64b") },
{ str_lit("tt-ascalon-d8"), str_lit("64bit,a,b,c,d,f,h,i,m,no-default-unroll,optimized-zero-stride-load,sha,shcounterenw,shgatpa,shtvala,shvsatpa,shvstvala,shvstvecd,smaia,ssaia,ssccptr,sscofpmf,sscounterenw,ssnpm,ssstateen,ssstrict,sstc,sstvala,sstvecd,ssu64xl,supm,svade,svbare,svinval,svnapot,svpbmt,unaligned-scalar-mem,unaligned-vector-mem,use-postra-scheduler,v,za64rs,zaamo,zalrsc,zawrs,zba,zbb,zbs,zca,zcb,zcmop,zfa,zfbfmin,zfh,zfhmin,zic64b,zicbom,zicbop,zicboz,ziccamoa,ziccif,zicclsm,ziccrse,zicntr,zicond,zicsr,zifencei,zihintntl,zihintpause,zihpm,zimop,zkt,zmmul,zvbb,zvbc,zve32f,zve32x,zve64d,zve64f,zve64x,zvfbfmin,zvfbfwma,zvfh,zvfhmin,zvkb,zvkg,zvkn,zvkned,zvkng,zvknhb,zvkt,zvl128b,zvl256b,zvl32b,zvl64b") },
{ str_lit("veyron-v1"), str_lit("64bit,a,auipc-addi-fusion,c,d,f,i,ld-add-fusion,lui-addi-fusion,m,shifted-zextw-fusion,ventana-veyron,xventanacondops,zaamo,zalrsc,zba,zbb,zbc,zbs,zexth-fusion,zextw-fusion,zicbom,zicbop,zicboz,zicntr,zicsr,zifencei,zihintpause,zihpm,zmmul") },
{ str_lit("xiangshan-nanhu"), str_lit("64bit,a,c,d,f,i,m,no-default-unroll,shifted-zextw-fusion,svinval,zaamo,zalrsc,zba,zbb,zbc,zbkb,zbkc,zbkx,zbs,zexth-fusion,zextw-fusion,zicbom,zicboz,zicsr,zifencei,zkn,zknd,zkne,zknh,zksed,zksh,zmmul") },
};
#else
// Generated with the featuregen script in `misc/featuregen`
gb_global String target_microarch_list[TargetArch_COUNT] = {
// TargetArch_Invalid:
@@ -514,3 +1072,4 @@ gb_global MicroarchFeatureList microarch_features_list[] = {
{ str_lit("veyron-v1"), str_lit("64bit,a,auipc-addi-fusion,c,d,f,ld-add-fusion,lui-addi-fusion,m,shifted-zextw-fusion,ventana-veyron,xventanacondops,zba,zbb,zbc,zbs,zexth-fusion,zextw-fusion,zicbom,zicbop,zicboz,zicntr,zicsr,zifencei,zihintpause,zihpm") },
{ str_lit("xiangshan-nanhu"), str_lit("64bit,a,c,d,f,m,svinval,zba,zbb,zbc,zbkb,zbkc,zbkx,zbs,zicbom,zicboz,zicsr,zifencei,zkn,zknd,zkne,zknh,zksed,zksh") },
};
#endif

209
src/bundle_command.cpp Normal file
View File

@@ -0,0 +1,209 @@
i32 bundle_android(String init_directory);
i32 bundle(String init_directory) {
switch (build_context.command_kind) {
case Command_bundle_android:
return bundle_android(init_directory);
}
gb_printf_err("Unknown odin package <platform>\n");
return 1;
}
i32 bundle_android(String original_init_directory) {
TEMPORARY_ALLOCATOR_GUARD();
i32 result = 0;
init_android_values(/*with_sdk*/true);
bool init_directory_ok = false;
String init_directory = path_to_fullpath(temporary_allocator(), original_init_directory, &init_directory_ok);
if (!init_directory_ok) {
gb_printf_err("Error: '%.*s' is not a valid directory", LIT(original_init_directory));
return 1;
}
init_directory = normalize_path(temporary_allocator(), init_directory, NIX_SEPARATOR_STRING);
int const ODIN_ANDROID_API_LEVEL = build_context.ODIN_ANDROID_API_LEVEL;
String android_sdk_build_tools = concatenate3_strings(temporary_allocator(),
build_context.ODIN_ANDROID_SDK, str_lit("build-tools"), NIX_SEPARATOR_STRING);
Array<FileInfo> list = {};
ReadDirectoryError rd_err = read_directory(android_sdk_build_tools, &list);
defer (array_free(&list));
switch (rd_err) {
case ReadDirectory_InvalidPath:
gb_printf_err("Invalid path: %.*s\n", LIT(android_sdk_build_tools));
return 1;
case ReadDirectory_NotExists:
gb_printf_err("Path does not exist: %.*s\n", LIT(android_sdk_build_tools));
return 1;
case ReadDirectory_Permission:
gb_printf_err("Unknown error whilst reading path %.*s\n", LIT(android_sdk_build_tools));
return 1;
case ReadDirectory_NotDir:
gb_printf_err("Expected a directory for a package, got a file: %.*s\n", LIT(android_sdk_build_tools));
return 1;
case ReadDirectory_Empty:
gb_printf_err("Empty directory: %.*s\n", LIT(android_sdk_build_tools));
return 1;
case ReadDirectory_Unknown:
gb_printf_err("Unknown error whilst reading path %.*s\n", LIT(android_sdk_build_tools));
return 1;
}
auto possible_valid_dirs = array_make<FileInfo>(heap_allocator(), 0, list.count);
defer (array_free(&possible_valid_dirs));
for (FileInfo fi : list) if (fi.is_dir) {
bool all_numbers = true;
for (isize i = 0; i < fi.name.len; i++) {
u8 c = fi.name[i];
if ('0' <= c && c <= '9') {
// true
} else if (i == 0) {
all_numbers = false;
} else if (c == '.') {
break;
} else {
all_numbers = false;
}
}
if (all_numbers) {
array_add(&possible_valid_dirs, fi);
}
}
if (possible_valid_dirs.count == 0) {
gb_printf_err("Unable to find any Android SDK/API Level in %.*s\n", LIT(android_sdk_build_tools));
return 1;
}
int *dir_numbers = gb_alloc_array(temporary_allocator(), int, possible_valid_dirs.count);
char buf[1024] = {};
for_array(i, possible_valid_dirs) {
FileInfo fi = possible_valid_dirs[i];
isize n = gb_min(gb_size_of(buf)-1, fi.name.len);
memcpy(buf, fi.name.text, n);
buf[n] = 0;
dir_numbers[i] = atoi(buf);
}
isize closest_number_idx = -1;
for (isize i = 0; i < possible_valid_dirs.count; i++) {
if (dir_numbers[i] >= ODIN_ANDROID_API_LEVEL) {
if (closest_number_idx < 0) {
closest_number_idx = i;
} else if (dir_numbers[i] < dir_numbers[closest_number_idx]) {
closest_number_idx = i;
}
}
}
if (closest_number_idx < 0) {
gb_printf_err("Unable to find any Android SDK/API Level in %.*s meeting the minimum API level of %d\n", LIT(android_sdk_build_tools), ODIN_ANDROID_API_LEVEL);
return 1;
}
String api_number = possible_valid_dirs[closest_number_idx].name;
android_sdk_build_tools = concatenate_strings(temporary_allocator(), android_sdk_build_tools, api_number);
String android_sdk_platforms = concatenate_strings(temporary_allocator(),
build_context.ODIN_ANDROID_SDK,
make_string_c(gb_bprintf("platforms/android-%d/", dir_numbers[closest_number_idx]))
);
android_sdk_build_tools = normalize_path(temporary_allocator(), android_sdk_build_tools, NIX_SEPARATOR_STRING);
android_sdk_platforms = normalize_path(temporary_allocator(), android_sdk_platforms, NIX_SEPARATOR_STRING);
gbString cmd = gb_string_make(heap_allocator(), "");
defer (gb_string_free(cmd));
String current_directory = normalize_path(temporary_allocator(), get_working_directory(temporary_allocator()), NIX_SEPARATOR_STRING);
defer (set_working_directory(current_directory));
if (current_directory.len != 0) {
bool ok = set_working_directory(init_directory);
if (!ok) {
gb_printf_err("Error: Unable to currectly set the current working directory to '%.*s'\n", LIT(init_directory));
}
}
String output_filename = str_lit("test");
String output_apk = path_remove_extension(output_filename);
TIME_SECTION("Android aapt");
{
TEMPORARY_ALLOCATOR_GUARD();
gb_string_clear(cmd);
String manifest = {};
if (build_context.android_manifest.len != 0) {
manifest = concatenate_strings(temporary_allocator(), current_directory, build_context.android_manifest);
} else {
manifest = concatenate_strings(temporary_allocator(), init_directory, str_lit("AndroidManifest.xml"));
}
cmd = gb_string_append_length(cmd, android_sdk_build_tools.text, android_sdk_build_tools.len);
cmd = gb_string_appendc(cmd, "aapt");
cmd = gb_string_appendc(cmd, " package -f");
if (manifest.len != 0) {
cmd = gb_string_append_fmt(cmd, " -M \"%.*s\"", LIT(manifest));
}
cmd = gb_string_append_fmt(cmd, " -I \"%.*sandroid.jar\"", LIT(android_sdk_platforms));
cmd = gb_string_append_fmt(cmd, " -F \"%.*s.apk-build\"", LIT(output_apk));
result = system_exec_command_line_app("android-aapt", cmd);
if (result) {
return result;
}
}
TIME_SECTION("Android jarsigner");
{
TEMPORARY_ALLOCATOR_GUARD();
gb_string_clear(cmd);
cmd = gb_string_append_length(cmd, build_context.ODIN_ANDROID_JAR_SIGNER.text, build_context.ODIN_ANDROID_JAR_SIGNER.len);
cmd = gb_string_append_fmt(cmd, " -storepass android");
if (build_context.android_keystore.len != 0) {
String keystore = concatenate_strings(temporary_allocator(), current_directory, build_context.android_keystore);
cmd = gb_string_append_fmt(cmd, " -keystore \"%.*s\"", LIT(keystore));
}
cmd = gb_string_append_fmt(cmd, " \"%.*s.apk-build\"", LIT(output_apk));
if (build_context.android_keystore_alias.len != 0) {
String keystore_alias = build_context.android_keystore_alias;
cmd = gb_string_append_fmt(cmd, " \"%.*s\"", LIT(keystore_alias));
}
result = system_exec_command_line_app("android-jarsigner", cmd);
if (result) {
return result;
}
}
TIME_SECTION("Android zipalign");
{
TEMPORARY_ALLOCATOR_GUARD();
gb_string_clear(cmd);
cmd = gb_string_append_length(cmd, android_sdk_build_tools.text, android_sdk_build_tools.len);
cmd = gb_string_appendc(cmd, "zipalign");
cmd = gb_string_appendc(cmd, " -f 4");
cmd = gb_string_append_fmt(cmd, " \"%.*s.apk-build\" \"%.*s.apk\"", LIT(output_apk), LIT(output_apk));
result = system_exec_command_line_app("android-zipalign", cmd);
if (result) {
return result;
}
}
return 0;
}

21
src/check_builtin.cpp
View File

@@ -645,6 +645,13 @@ gb_internal bool check_builtin_simd_operation(CheckerContext *c, Operand *operan
break;
}
if (!are_types_identical(x.type, y.type)) {
gbString tx = type_to_string(x.type);
gbString ty = type_to_string(y.type);
error(call, "Mismatched types to '%.*s', '%s' vs '%s'", LIT(builtin_name), tx, ty);
gb_string_free(ty);
gb_string_free(tx);
}
Type *vt = base_type(x.type);
GB_ASSERT(vt->kind == Type_SimdVector);
@@ -1675,12 +1682,16 @@ gb_internal bool check_builtin_procedure_directive(CheckerContext *c, Operand *o
}
if (ce->args.count > 0) {
Ast *arg = ce->args[0];
Operand o = {};
Entity *e = check_ident(c, &o, arg, nullptr, nullptr, true);
if (e == nullptr || (e->flags & EntityFlag_Param) == 0) {
error(ce->args[0], "'#caller_expression' expected a valid earlier parameter name");
if (arg->kind != Ast_Ident) {
error(arg, "'#caller_expression' expected an identifier");
} else {
Operand o = {};
Entity *e = check_ident(c, &o, arg, nullptr, nullptr, true);
if (e == nullptr || (e->flags & EntityFlag_Param) == 0) {
error(arg, "'#caller_expression' expected a valid earlier parameter name");
}
arg->Ident.entity = e;
}
arg->Ident.entity = e;
}
operand->type = t_string;

13
src/check_expr.cpp
View File

@@ -8977,8 +8977,14 @@ gb_internal ExprKind check_or_else_expr(CheckerContext *c, Operand *o, Ast *node
o->expr = node;
return Expr_Expr;
}
Type *left_type = nullptr;
Type *right_type = nullptr;
check_or_else_split_types(c, &x, name, &left_type, &right_type);
add_type_and_value(c, arg, x.mode, x.type, x.value);
bool y_is_diverging = false;
check_expr_base(c, &y, default_value, x.type);
check_expr_base(c, &y, default_value, left_type);
switch (y.mode) {
case Addressing_NoValue:
if (is_diverging_expr(y.expr)) {
@@ -9003,11 +9009,6 @@ gb_internal ExprKind check_or_else_expr(CheckerContext *c, Operand *o, Ast *node
return Expr_Expr;
}
Type *left_type = nullptr;
Type *right_type = nullptr;
check_or_else_split_types(c, &x, name, &left_type, &right_type);
add_type_and_value(c, arg, x.mode, x.type, x.value);
if (left_type != nullptr) {
if (!y_is_diverging) {
check_assignment(c, &y, left_type, name);

1
src/checker.cpp
View File

@@ -1149,6 +1149,7 @@ gb_internal void init_universal(void) {
GlobalEnumValue values[Subtarget_COUNT] = {
{"Default", Subtarget_Default},
{"iOS", Subtarget_iOS},
{"Android", Subtarget_Android},
};
auto fields = add_global_enum_type(str_lit("Odin_Platform_Subtarget_Type"), values, gb_count_of(values));

207
src/linker.cpp
View File

@@ -7,19 +7,15 @@ struct LinkerData {
Array<String> output_temp_paths;
String output_base;
String output_name;
#if defined(GB_SYSTEM_OSX)
b8 needs_system_library_linked;
#endif
bool needs_system_library_linked;
};
gb_internal i32 system_exec_command_line_app(char const *name, char const *fmt, ...);
gb_internal bool system_exec_command_line_app_output(char const *command, gbString *output);
#if defined(GB_SYSTEM_OSX)
gb_internal void linker_enable_system_library_linking(LinkerData *ld) {
ld->needs_system_library_linked = 1;
ld->needs_system_library_linked = true;
}
#endif
gb_internal void linker_data_init(LinkerData *ld, CheckerInfo *info, String const &init_fullpath) {
gbAllocator ha = heap_allocator();
@@ -28,9 +24,7 @@ gb_internal void linker_data_init(LinkerData *ld, CheckerInfo *info, String cons
array_init(&ld->foreign_libraries, ha, 0, 1024);
ptr_set_init(&ld->foreign_libraries_set, 1024);
#if defined(GB_SYSTEM_OSX)
ld->needs_system_library_linked = 0;
#endif
ld->needs_system_library_linked = false;
if (build_context.out_filepath.len == 0) {
ld->output_name = remove_directory_from_path(init_fullpath);
@@ -136,6 +130,9 @@ gb_internal i32 linker_stage(LinkerData *gen) {
return result;
}
bool is_cross_linking = false;
bool is_android = false;
if (build_context.cross_compiling && selected_target_metrics->metrics == &target_essence_amd64) {
#if defined(GB_SYSTEM_UNIX)
result = system_exec_command_line_app("linker", "x86_64-essence-gcc \"%.*s.o\" -o \"%.*s\" %.*s %.*s",
@@ -147,22 +144,29 @@ gb_internal i32 linker_stage(LinkerData *gen) {
);
#endif
} else if (build_context.cross_compiling && build_context.different_os) {
gb_printf_err("Linking for cross compilation for this platform is not yet supported (%.*s %.*s)\n",
LIT(target_os_names[build_context.metrics.os]),
LIT(target_arch_names[build_context.metrics.arch])
);
build_context.keep_object_files = true;
switch (selected_subtarget) {
case Subtarget_Android:
is_cross_linking = true;
is_android = true;
goto try_cross_linking;
default:
gb_printf_err("Linking for cross compilation for this platform is not yet supported (%.*s %.*s)\n",
LIT(target_os_names[build_context.metrics.os]),
LIT(target_arch_names[build_context.metrics.arch])
);
build_context.keep_object_files = true;
break;
}
} else {
try_cross_linking:;
#if defined(GB_SYSTEM_WINDOWS)
bool is_windows = true;
bool is_windows = build_context.metrics.os == TargetOs_windows;
#else
bool is_windows = false;
#endif
#if defined(GB_SYSTEM_OSX)
bool is_osx = true;
#else
bool is_osx = false;
#endif
bool is_osx = build_context.metrics.os == TargetOs_darwin;
if (is_windows) {
@@ -414,23 +418,27 @@ gb_internal i32 linker_stage(LinkerData *gen) {
} else {
timings_start_section(timings, str_lit("ld-link"));
int const ODIN_ANDROID_API_LEVEL = build_context.ODIN_ANDROID_API_LEVEL;
String ODIN_ANDROID_NDK = build_context.ODIN_ANDROID_NDK;
String ODIN_ANDROID_NDK_TOOLCHAIN = build_context.ODIN_ANDROID_NDK_TOOLCHAIN;
String ODIN_ANDROID_NDK_TOOLCHAIN_LIB = build_context.ODIN_ANDROID_NDK_TOOLCHAIN_LIB;
String ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL = build_context.ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL;
String ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT = build_context.ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT;
// Link using `clang`, unless overridden by `ODIN_CLANG_PATH` environment variable.
const char* clang_path = gb_get_env("ODIN_CLANG_PATH", permanent_allocator());
if (clang_path == NULL) {
clang_path = "clang";
}
// NOTE(vassvik): get cwd, for used for local shared libs linking, since those have to be relative to the exe
char cwd[256];
#if !defined(GB_SYSTEM_WINDOWS)
getcwd(&cwd[0], 256);
#endif
//printf("%s\n", cwd);
// NOTE(vassvik): needs to add the root to the library search paths, so that the full filenames of the library
// files can be passed with -l:
gbString lib_str = gb_string_make(heap_allocator(), "-L/");
gbString lib_str = gb_string_make(heap_allocator(), "");
defer (gb_string_free(lib_str));
#if !defined(GB_SYSTEM_WINDOWS)
lib_str = gb_string_appendc(lib_str, "-L/ ");
#endif
StringSet asm_files = {};
string_set_init(&asm_files, 64);
@@ -496,19 +504,20 @@ gb_internal i32 linker_stage(LinkerData *gen) {
}
String obj_format;
#if defined(GB_ARCH_64_BIT)
if (is_osx) {
obj_format = str_lit("macho64");
if (build_context.metrics.ptr_size == 8) {
if (is_osx) {
obj_format = str_lit("macho64");
} else {
obj_format = str_lit("elf64");
}
} else {
obj_format = str_lit("elf64");
GB_ASSERT(build_context.metrics.ptr_size == 4);
if (is_osx) {
obj_format = str_lit("macho32");
} else {
obj_format = str_lit("elf32");
}
}
#elif defined(GB_ARCH_32_BIT)
if (is_osx) {
obj_format = str_lit("macho32");
} else {
obj_format = str_lit("elf32");
}
#endif // GB_ARCH_*_BIT
if (build_context.metrics.arch == TargetArch_riscv64) {
result = system_exec_command_line_app("clang",
@@ -618,6 +627,78 @@ gb_internal i32 linker_stage(LinkerData *gen) {
gbString object_files = gb_string_make(heap_allocator(), "");
defer (gb_string_free(object_files));
if (is_android) { // NOTE(bill): glue code needed for Android
TIME_SECTION("Android Native App Glue Compile");
String android_glue_object = {};
String android_glue_static_lib = {};
char hash_buf[64] = {};
gb_snprintf(hash_buf, gb_size_of(hash_buf), "%p", &hash_buf);
String hash = make_string_c(hash_buf);
String temp_dir = normalize_path(temporary_allocator(), temporary_directory(temporary_allocator()), NIX_SEPARATOR_STRING);
android_glue_object = concatenate4_strings(temporary_allocator(), temp_dir, str_lit("android_native_app_glue-"), hash, str_lit(".o"));
android_glue_static_lib = concatenate4_strings(permanent_allocator(), temp_dir, str_lit("libandroid_native_app_glue-"), hash, str_lit(".a"));
gbString glue = gb_string_make(heap_allocator(), clang_path);
defer (gb_string_free(glue));
glue = gb_string_append_fmt(glue, " --target=aarch64-linux-android%d ", ODIN_ANDROID_API_LEVEL);
glue = gb_string_appendc(glue, "-c \"");
glue = gb_string_append_length(glue, ODIN_ANDROID_NDK.text, ODIN_ANDROID_NDK.len);
glue = gb_string_appendc(glue, "sources/android/native_app_glue/android_native_app_glue.c");
glue = gb_string_appendc(glue, "\" ");
glue = gb_string_appendc(glue, "-o \"");
glue = gb_string_append_length(glue, android_glue_object.text, android_glue_object.len);
glue = gb_string_appendc(glue, "\" ");
glue = gb_string_appendc(glue, "\"-I");
glue = gb_string_append_length(glue, ODIN_ANDROID_NDK_TOOLCHAIN.text, ODIN_ANDROID_NDK_TOOLCHAIN.len);
glue = gb_string_appendc(glue, "sysroot/usr/include/");
glue = gb_string_appendc(glue, "\" ");
glue = gb_string_appendc(glue, "\"-I");
glue = gb_string_append_length(glue, ODIN_ANDROID_NDK_TOOLCHAIN.text, ODIN_ANDROID_NDK_TOOLCHAIN.len);
glue = gb_string_appendc(glue, "sysroot/usr/include/aarch64-linux-android/");
glue = gb_string_appendc(glue, "\" ");
glue = gb_string_appendc(glue, "-Wno-macro-redefined ");
result = system_exec_command_line_app("android-native-app-glue-compile", glue);
if (result) {
return result;
}
TIME_SECTION("Android Native App Glue ar");
gbString ar = gb_string_make_length(heap_allocator(), ODIN_ANDROID_NDK_TOOLCHAIN.text, ODIN_ANDROID_NDK_TOOLCHAIN.len);
defer (gb_string_free(ar));
ar = gb_string_appendc(ar, "bin/llvm-ar");
ar = gb_string_appendc(ar, " rcs ");
ar = gb_string_appendc(ar, "\"");
ar = gb_string_append_length(ar, android_glue_static_lib.text, android_glue_static_lib.len);
ar = gb_string_appendc(ar, "\" ");
ar = gb_string_appendc(ar, "\"");
ar = gb_string_append_length(ar, android_glue_object.text, android_glue_object.len);
ar = gb_string_appendc(ar, "\" ");
result = system_exec_command_line_app("android-native-app-glue-ar", ar);
if (result) {
return result;
}
object_files = gb_string_append_fmt(object_files, "\"%.*s\" ", LIT(android_glue_static_lib));
}
for (String object_path : gen->output_object_paths) {
object_files = gb_string_append_fmt(object_files, "\"%.*s\" ", LIT(object_path));
}
@@ -661,7 +742,9 @@ gb_internal i32 linker_stage(LinkerData *gen) {
link_settings = gb_string_appendc(link_settings, "-Wl,-init,'_odin_entry_point' ");
link_settings = gb_string_appendc(link_settings, "-Wl,-fini,'_odin_exit_point' ");
}
} else if (is_android) {
// Always shared even in android!
link_settings = gb_string_appendc(link_settings, "-shared ");
}
if (build_context.build_mode == BuildMode_Executable && build_context.reloc_mode == RelocMode_PIC) {
@@ -670,6 +753,7 @@ gb_internal i32 linker_stage(LinkerData *gen) {
if (build_context.metrics.os != TargetOs_openbsd
&& build_context.metrics.os != TargetOs_haiku
&& build_context.metrics.arch != TargetArch_riscv64
&& !is_android
) {
// OpenBSD and Haiku default to PIE executable. do not pass -no-pie for it.
link_settings = gb_string_appendc(link_settings, "-no-pie ");
@@ -701,6 +785,29 @@ gb_internal i32 linker_stage(LinkerData *gen) {
// This points the linker to where the entry point is
link_settings = gb_string_appendc(link_settings, "-e _main ");
}
} else if (build_context.metrics.os == TargetOs_openbsd) {
// OpenBSD ports install shared libraries in /usr/local/lib. Also, we must explicitly link libpthread.
platform_lib_str = gb_string_appendc(platform_lib_str, "-lpthread -Wl,-L/usr/local/lib ");
// Until the LLVM back-end can be adapted to emit endbr64 instructions on amd64, we
// need to pass -z nobtcfi in order to allow the resulting program to run under
// OpenBSD 7.4 and newer. Once support is added at compile time, this can be dropped.
platform_lib_str = gb_string_appendc(platform_lib_str, "-Wl,-z,nobtcfi ");
}
if (is_android) {
GB_ASSERT(ODIN_ANDROID_NDK_TOOLCHAIN_LIB.len != 0);
GB_ASSERT(ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL.len != 0);
GB_ASSERT(ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT.len != 0);
platform_lib_str = gb_string_appendc(platform_lib_str, "\"-L");
platform_lib_str = gb_string_append_length(platform_lib_str, ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL.text, ODIN_ANDROID_NDK_TOOLCHAIN_LIB_LEVEL.len);
platform_lib_str = gb_string_appendc(platform_lib_str, "\" ");
platform_lib_str = gb_string_appendc(platform_lib_str, "\"--sysroot=");
platform_lib_str = gb_string_append_length(platform_lib_str, ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT.text, ODIN_ANDROID_NDK_TOOLCHAIN_SYSROOT.len);
platform_lib_str = gb_string_appendc(platform_lib_str, "\" ");
link_settings = gb_string_appendc(link_settings, "-u ANativeActivity_onCreate ");
}
if (!build_context.no_rpath) {
@@ -709,24 +816,31 @@ gb_internal i32 linker_stage(LinkerData *gen) {
if (build_context.metrics.os == TargetOs_darwin) {
link_settings = gb_string_appendc(link_settings, "-Wl,-rpath,@loader_path ");
} else {
link_settings = gb_string_appendc(link_settings, "-Wl,-rpath,\\$ORIGIN ");
if (is_android) {
// ignore
} else {
link_settings = gb_string_appendc(link_settings, "-Wl,-rpath,\\$ORIGIN ");
}
}
}
if (!build_context.no_crt) {
platform_lib_str = gb_string_appendc(platform_lib_str, "-lm ");
lib_str = gb_string_appendc(lib_str, "-lm ");
if (build_context.metrics.os == TargetOs_darwin) {
// NOTE: adding this causes a warning about duplicate libraries, I think it is
// automatically assumed/added by clang when you don't do `-nostdlib`.
// platform_lib_str = gb_string_appendc(platform_lib_str, "-lSystem ");
// lib_str = gb_string_appendc(lib_str, "-lSystem ");
} else {
platform_lib_str = gb_string_appendc(platform_lib_str, "-lc ");
lib_str = gb_string_appendc(lib_str, "-lc ");
}
}
gbString link_command_line = gb_string_make(heap_allocator(), clang_path);
defer (gb_string_free(link_command_line));
if (is_android) {
link_command_line = gb_string_append_fmt(link_command_line, " --target=aarch64-linux-android%d ", ODIN_ANDROID_API_LEVEL);
}
link_command_line = gb_string_appendc(link_command_line, " -Wno-unused-command-line-argument ");
link_command_line = gb_string_appendc(link_command_line, object_files);
link_command_line = gb_string_append_fmt(link_command_line, " -o \"%.*s\" ", LIT(output_filename));
@@ -736,6 +850,11 @@ gb_internal i32 linker_stage(LinkerData *gen) {
link_command_line = gb_string_append_fmt(link_command_line, " %.*s ", LIT(build_context.extra_linker_flags));
link_command_line = gb_string_append_fmt(link_command_line, " %s ", link_settings);
if (is_android) {
TIME_SECTION("Linking");
}
if (build_context.linker_choice == Linker_lld) {
link_command_line = gb_string_append_fmt(link_command_line, " -fuse-ld=lld");
result = system_exec_command_line_app("lld-link", link_command_line);

Some files were not shown because too many files have changed in this diff Show More