Feoramund
920487be04
Zero only needed part of memory when changing size categories
2025-05-08 11:06:03 -04:00
Feoramund
92fdb4d79a
Cache contiguous free slabs per heap superpage
2025-05-08 11:06:03 -04:00
Feoramund
e038c3cf4e
Make superpage caching linear time instead of quadratic
2025-05-08 11:06:03 -04:00
Feoramund
dcd9bd7888
Make slab map caching linear time instead of quadratic
2025-05-08 11:06:03 -04:00
Feoramund
fa1f6abe78
Don't atomically subtract zero from remote_free_count
2025-05-08 11:06:03 -04:00
Feoramund
a0425ca220
Do not return superpages that had remote frees to the OS
2025-05-08 11:06:03 -04:00
Feoramund
5e773aaf56
Adjust heap allocator benchmarks
2025-05-08 11:06:03 -04:00
Feoramund
acd5d0d40b
Fix slab iteration bug in the remote free collection phase
...
Previously, a wide slab could be freed during this phase which could
trigger the superpage itself to be freed, resulting in iteration across
potentially invalid memory if the superpage was returned to the host and
not placed in the orphanage.
`heap_free_wide_slab` has had its responsibilities reduced to just
freeing the wide slab, and `heap_free_superpage_if_empty_and_unused` has
been added to better clarify what is happening where.
2025-05-08 11:06:02 -04:00
Feoramund
2d0547dcaf
Remove unneeded call to min
...
At this point, we know we're expanding the wide slab, so `old_size` will
always be the right size.
2025-05-08 11:06:02 -04:00
Feoramund
44a7015609
Fix in-place expansion of wide slabs
2025-05-08 11:06:02 -04:00
Feoramund
8638bd3b89
Fix slab iteration bug in heap_cache_register_superpage
...
Increment of `i` is handled near the start of this loop.
2025-05-08 11:06:02 -04:00
Feoramund
ee3d859c6b
Remove HEAP_PANIC_ON_FREE_NIL
2025-05-08 11:06:02 -04:00
Feoramund
f5226d01f7
Revert to unoptimized behavior on Darwin
2025-05-08 11:06:02 -04:00
Feoramund
4a31e1f635
Ensure remapped memory has desired protection on Darwin
2025-05-08 11:06:02 -04:00
Feoramund
0947914e0d
Use mach_task_self_ global instead
2025-05-08 11:06:02 -04:00
Feoramund
247ce3cc97
Favor rescheduling superpage cache entry instead
...
This should be more robust during merges that happen simultaneously with
remote frees.
Previously, we would preserve the cache entry, but there is an edge case
where a slab of a lower index could be remotely freed and missed during
a merge.
2025-05-08 11:06:02 -04:00
Feoramund
e86bd35387
Optimize virtual memory resizing on Darwin
2025-05-08 11:06:02 -04:00
Feoramund
c52c29856c
Unify heap allocators
...
- Simplify the malloc-based allocator
- Correct the type signatures on malloc/calloc/realloc
- Make `heap_alloc/heap_free/heap_resize` API consistent across platforms
2025-05-08 11:06:02 -04:00
Feoramund
30e600699d
Strengthen order to prevent reordering
2025-05-08 11:06:01 -04:00
Feoramund
0532b73563
Simplify loop
2025-05-08 11:06:01 -04:00
Feoramund
541443fe33
Don't build feoramalloc on web platforms
...
Their allocation pathways are already tuned for their respective
platforms.
2025-05-08 11:06:01 -04:00
Feoramund
e9bf350cca
Add heap allocator exception for Orca
...
This is a copy of the libc malloc-based allocator that we previously
had. It's preferable to use the Orca runtime allocator over our own.
2025-05-08 11:06:01 -04:00
Feoramund
ef72a373b2
Use VM_INHERIT_COPY instead
...
This aligns with the use of `MAP_PRIVATE` in `mmap`-based
implementations.
2025-05-08 11:06:01 -04:00
Feoramund
4c59019a57
Restrict superpage allocation on Darwin to AMD64 and 2MiB
...
There's a possibility that Darwin does not support anything else, in
terms of size or architecture.
2025-05-08 11:06:01 -04:00
Feoramund
1e62f2c8c4
Rearrange slab iteration conditionals
2025-05-08 11:06:01 -04:00
Feoramund
76a2877d7d
Hoist reused calculations out into variables
2025-05-08 11:06:01 -04:00
Feoramund
17b38c1bca
Strengthen Consume to Acquire in heap allocator
2025-05-08 11:06:01 -04:00
Feoramund
6d99d9571c
Fix indentation
2025-05-08 11:06:01 -04:00
Feoramund
102fc3f058
Add heap allocator tests to CI
2025-05-08 11:06:00 -04:00
Feoramund
a4a776634c
Add native heap allocator
...
- Add the test bench for the allocator
- Move old allocator code to the test bench
- Fix `heap_resize` usage in `os2/env_linux.odin` to fit new API
requiring `old_size`
2025-05-08 11:06:00 -04:00
Feoramund
f142043db9
Add virtual memory procedures to base
2025-05-08 11:06:00 -04:00
Feoramund
13dca77d3d
Add get_current_thread_id to base
2025-05-08 11:06:00 -04:00
gingerBill
4a4294c2af
Merge pull request #5095 from Lperlind/temp-collisions
...
os2: Resolve temp allocator collisions
2025-05-08 15:11:51 +01:00
gingerBill
204c6f514a
Remove use of do
2025-05-08 14:03:17 +01:00
gingerBill
0ce8373528
Formatting Fix
2025-05-08 13:23:22 +01:00
gingerBill
d7fc3d1770
Merge pull request #5121 from Badaxis/badaxis/xaudio2
...
XAudio2 Implementation
2025-05-08 13:11:51 +01:00
gingerBill
92df892f25
Merge pull request #5064 from harold-b/hb/objc-classes
...
Add support for Objective-C class implementation
2025-05-08 12:58:33 +01:00
Lucas Perlind
f5b947ea73
Fix linux
2025-05-08 17:45:29 +10:00
Lucas Perlind
14c5096b81
Improve assert information
2025-05-08 17:41:03 +10:00
Lucas Perlind
9f2d008a8a
Remove TEMP_ALLOCATOR_GUARD wrapper
2025-05-08 17:41:03 +10:00
Lucas Perlind
5292a7f4f3
Fix posix
2025-05-08 17:41:03 +10:00
Lucas Perlind
1bea59ee68
Fix linux
2025-05-08 17:41:03 +10:00
Lucas Perlind
190960e103
fix -vet
2025-05-08 17:41:03 +10:00
Lucas Perlind
b9db1dd3e0
os2: Resolve temp allocator collisions
2025-05-08 17:41:01 +10:00
gingerBill
4a709086a4
Merge pull request #5131 from laytan/fix-builder-dispose
...
fix disposing builder when not created
2025-05-07 14:03:42 +01:00
Laytan Laats
ad4866653a
fix disposing builder when not created
...
Fixes #5128
p->builder is created in lb_begin_procedure_body, but that isn't called
if there is no body, and we were still calling dispose at that point.
Moved it into lb_end_procedure_body to match.
2025-05-07 14:39:51 +02:00
gingerBill
90a30a145a
Merge pull request #5122 from Lperlind/asan-allocators
...
Add asan support for various allocators and stack unpoisoning
2025-05-07 10:21:16 +01:00
Lucas Perlind
46e0c7ad74
Cleanup
2025-05-07 11:30:58 +10:00
Laytan
7c1a9f1e7a
Merge pull request #5127 from laytan/fix-process-windows-handles-handling
...
Fix process windows handles handling
2025-05-06 22:31:54 +02:00
laytan
9b218a2922
don't need to hang on to the null handle
2025-05-06 19:42:52 +02:00
