Files
Odin/core/os/os.odin
Tetralux 99121d6ff2 Implement core:thread and core:sync on Unix using pthreads
Also do some cleanup and refactoring of the thread, sync and time APIs.

- remove 'semaphore_release' because 'post' and 'wait' is easier to understand

- change 'semaphore_wait' to '*_wait_for' to match Condition

- pthreads can be given a stack, but doing so requires the user to set up the guard
  pages manually. BE WARNED. The alignment requirements of the stack are also
  platform-dependant; it may need to be page size aligned on some systems.
  Unclear which systems, however. See 'os.get_page_size', and 'mem.make_aligned'.
  HOWEVER: I was unable to get custom stacks with guard pages working reliably,
  so while you can do it, the API does not support it.

- add 'os.get_page_size', 'mem.make_aligned', and 'mem.new_aligned'.

- removed thread return values because windows and linux are not consistent; windows returns 'i32'
  and pthreads return 'void*'; besides which, if you really wanted to communicate how the
  thread exited, you probably wouldn't do it with the thread's exit code.

- fixed 'thread.is_done' on Windows; it didn't report true immediately after calling 'thread.join'.

- moved time related stuff out of 'core:os' to 'core:time'.

- add 'mem.align_backward'

- fixed default allocator alignment
  The heap on Windows, and calloc on Linux, both have no facility to request alignment.
  It's a bit of hack, but the heap_allocator now overallocates; `size + alignment` bytes,
  and aligns things to at least 2.
  It does both of these things to ensure that there is at least two bytes before the payload,
  which it uses to store how much padding it needed to insert in order to fulfil the alignment
  requested.

- make conditions more sane by matching the Windows behaviour.
  The fact that they were signalled now lingers until a thread tries to wait,
  causing them to just pass by uninterrupted, without sleeping or locking the
  underlying mutex, as it would otherwise need to do.
  This means that a thread no longer has to be waiting in order to be signalled, which
  avoids timing bugs that causes deadlocks that are hard to debug and fix.
  See the comment on the `sync.Condition.flag` field.

- add thread priority: `thread.create(worker_proc, .High)`
2019-12-01 00:46:23 +00:00

194 lines
4.7 KiB
Odin

package os
import "core:mem"
import "core:strconv"
import "core:unicode/utf8"
write_string :: proc(fd: Handle, str: string) -> (int, Errno) {
return write(fd, cast([]byte)str);
}
write_byte :: proc(fd: Handle, b: byte) -> (int, Errno) {
return write(fd, []byte{b});
}
write_rune :: proc(fd: Handle, r: rune) -> (int, Errno) {
if r < utf8.RUNE_SELF {
return write_byte(fd, byte(r));
}
b, n := utf8.encode_rune(r);
return write(fd, b[:n]);
}
write_encoded_rune :: proc(fd: Handle, r: rune) {
write_byte(fd, '\'');
switch r {
case '\a': write_string(fd, "\\a");
case '\b': write_string(fd, "\\b");
case '\e': write_string(fd, "\\e");
case '\f': write_string(fd, "\\f");
case '\n': write_string(fd, "\\n");
case '\r': write_string(fd, "\\r");
case '\t': write_string(fd, "\\t");
case '\v': write_string(fd, "\\v");
case:
if r < 32 {
write_string(fd, "\\x");
b: [2]byte;
s := strconv.append_bits(b[:], u64(r), 16, true, 64, strconv.digits, nil);
switch len(s) {
case 0: write_string(fd, "00");
case 1: write_rune(fd, '0');
case 2: write_string(fd, s);
}
} else {
write_rune(fd, r);
}
}
write_byte(fd, '\'');
}
file_size_from_path :: proc(path: string) -> i64 {
fd, err := open(path, O_RDONLY, 0);
if err != 0 {
return -1;
}
defer close(fd);
length: i64;
if length, err = file_size(fd); err != 0 {
return -1;
}
return length;
}
read_entire_file :: proc(name: string) -> (data: []byte, success: bool) {
fd, err := open(name, O_RDONLY, 0);
if err != 0 {
return nil, false;
}
defer close(fd);
length: i64;
if length, err = file_size(fd); err != 0 {
return nil, false;
}
if length <= 0 {
return nil, true;
}
data = make([]byte, int(length));
if data == nil {
return nil, false;
}
bytes_read, read_err := read(fd, data);
if read_err != 0 {
delete(data);
return nil, false;
}
return data[0:bytes_read], true;
}
write_entire_file :: proc(name: string, data: []byte, truncate := true) -> (success: bool) {
flags: int = O_WRONLY|O_CREATE;
if truncate {
flags |= O_TRUNC;
}
fd, err := open(name, flags, 0);
if err != 0 {
return false;
}
defer close(fd);
_, write_err := write(fd, data);
return write_err == 0;
}
write_ptr :: proc(fd: Handle, data: rawptr, len: int) -> (int, Errno) {
s := transmute([]byte)mem.Raw_Slice{data, len};
return write(fd, s);
}
read_ptr :: proc(fd: Handle, data: rawptr, len: int) -> (int, Errno) {
s := transmute([]byte)mem.Raw_Slice{data, len};
return read(fd, s);
}
heap_allocator_proc :: proc(allocator_data: rawptr, mode: mem.Allocator_Mode,
size, alignment: int,
old_memory: rawptr, old_size: int, flags: u64 = 0, loc := #caller_location) -> rawptr {
//
// NOTE(tetra, 2019-11-10): The heap doesn't respect alignment.
// HACK: Overallocate, align forwards, and then use the two bytes immediately before
// the address we return, to store the padding we inserted.
// This allows us to pass the original pointer we got back from the heap to `free` later.
//
align_and_store_padding :: proc(ptr: rawptr, alignment: int) -> rawptr {
ptr := mem.ptr_offset(cast(^u8) ptr, 2);
new_ptr := cast(^u8) mem.align_forward(ptr, uintptr(alignment));
offset := mem.ptr_sub(new_ptr, cast(^u8) ptr) + 2;
assert(offset < int(max(u16)));
(^[2]u8)(mem.ptr_offset(new_ptr, -2))^ = transmute([2]u8) u16(offset);
return new_ptr;
}
recover_original_pointer :: proc(ptr: rawptr) -> rawptr {
ptr := cast(^u8) ptr;
offset := transmute(u16) (^[2]u8)(mem.ptr_offset(ptr, -2))^;
ptr = mem.ptr_offset(ptr, -int(offset));
return ptr;
}
aligned_heap_alloc :: proc(size: int, alignment: int) -> rawptr {
// NOTE(tetra): Alignment 1 will mean we only have one extra byte.
// This is not enough for a u16 - so we ensure there is at least two bytes extra.
// This also means that the pointer is always aligned to at least 2.
extra := alignment;
if extra <= 1 do extra = 2;
orig := cast(^u8) heap_alloc(size + extra);
if orig == nil do return nil;
ptr := align_and_store_padding(orig, alignment);
assert(recover_original_pointer(ptr) == orig);
return ptr;
}
switch mode {
case .Alloc:
return aligned_heap_alloc(size, alignment);
case .Free:
assert(old_memory != nil);
ptr := recover_original_pointer(old_memory);
heap_free(ptr);
return nil;
case .Free_All:
// NOTE(bill): Does nothing
case .Resize:
if old_memory == nil {
return aligned_heap_alloc(size, alignment);
}
ptr := recover_original_pointer(old_memory);
ptr = heap_resize(ptr, size);
assert(ptr != nil);
return align_and_store_padding(ptr, alignment);
}
return nil;
}
heap_allocator :: proc() -> mem.Allocator {
return mem.Allocator{
procedure = heap_allocator_proc,
data = nil,
};
}