gingerBill
80f640c18d
Fix cond_wait synchronizing on the condition variable; match objects within each other; mark more Win32 waits
2026-10-08 23:14:14 +01:00
gingerBill
8bf769b039
Add @(synchronizes=...) to Win32 SRW locks and critical sections
2026-10-08 23:09:05 +01:00
gingerBill
3c7034c158
Add @(synchronizes) to channels; Model guards deferred unlocks when inferring
2026-10-08 23:05:45 +01:00
gingerBill
4508144727
Add @(synchronizes_shared=...) for shared locks; Infer @(synchronizes) for wrappers
2026-10-08 22:56:54 +01:00
gingerBill
e0c1f9e325
Add @(synchronizes=...) to more primitives
2026-10-08 17:17:50 +01:00
gingerBill
5de6a99129
Add @(synchronizes=<Enum>); Vet plain writes; Warn on mixed-sized atomics; Count fences in callers and other branches
2026-10-08 17:16:09 +01:00
bbodi
1e8419d1d4
Fix core:simd/x86 AVX wrappers that did not build
...
Calling these failed to compile: `_mm256_round_pd` passed a u8 where
the intrinsic takes a u32, `_mm256_blend_pd` misspelt its parameter,
`_mm256_dp_ps` took its mask as an i8, `_mm_permute_ps/pd` used
SSE/SSE2 helpers without enabling them, and `_mm256_permute2f128_pd`
did not return its result.
2026-10-08 15:51:24 +02:00
bbodi
1e1ea828a5
Fix intrinsic signatures in core:simd/x86
...
`_mm_cvtsd_ss` passed two `__m128d` to `llvm.x86.sse2.cvtsd2ss`,
which takes `(<4 x float>, <2 x double>)`, so LLVM failed with
"Intrinsic has incorrect argument type". `_mm_cvtss_sd` used
`llvm.x86.sse2.cvtss2sd`, which LLVM no longer has. Both wrappers
also took the wrong operand types; use Intel's (a: __m128, b: __m128d)
and (a: __m128d, b: __m128), and express `_mm_cvtss_sd` as a lane
conversion as clang does.
`_mm_prefetch` called `llvm.prefetch` without its pointer mangling and
`__rdtscp` declared `llvm.x86.rdtscp` as taking a pointer when it
returns `{i64, i32}`; both were rejected by LLVM too.
2026-10-08 15:12:18 +02:00
bbodi
26912ceb95
Replace removed LLVM intrinsics in core:simd/x86
...
LLVM has dropped many x86 intrinsics in favour of generic IR, and
auto-upgrades them only when reading bitcode. Odin declares them
directly, so the calls became undefined external symbols and failed
to link, e.g. `undefined reference to llvm.x86.sse41.pmaxsb`.
Express the affected wrappers the way clang's headers now do: scalar
lane ops, simd.min/max/abs/sqrt, shuffles for the immediate blends,
masked or narrowed multiplies, conversions and an unaligned store.
`_addcarryx_*` now use `llvm.x86.addcarry.*`, as clang does.
`_mm_sqrt_ss` took an unused second argument; it now takes one, as
in Intel's definition. It could not have been used before.
2026-10-08 15:12:18 +02:00
Jeroen van Rijn
a961bb5ddb
Merge pull request #7818 from bbodi/fix/sha3-init-round-start
...
Reset keccak_round_start in _sha3.init
2026-10-08 02:51:19 -07:00
bbodi
f7f3e63191
Reset keccak_round_start in _sha3.init
...
init left the round start untouched. A context that skipped zeroing
with `= ---`, or one reused after TurboSHAKE, kept a stale value, so
SHA-3 and SHAKE ran too few rounds and returned wrong hashes silently.
2026-10-08 11:02:47 +02:00
Jeroen van Rijn
4d09219ff4
Merge pull request #7616 from wardjm/fix/parse_int_overflow
...
Fixed strconv overflow noted in issue#7608.
2026-10-08 01:57:40 -07:00
Jeffrey Ward
c2684b43cb
Added tests from issue 7608 comments.
...
Fixed "_" only and no digits. Now returns an error since this is effectively passing nothing to the parser.
2026-10-07 21:28:43 -04:00
gingerBill
b0c36ab7ca
Merge pull request #7816 from odin-lang/bill/atomic-memory-ordering-pass
...
Atomic Memory Ordering and Futex Operation Passes
2026-10-07 23:36:22 +01:00
gingerBill
d2d1fe8372
Remove the redundant t.flags += {.Joined} in _join on Unix, as it was already set atomically
2026-10-07 23:27:48 +01:00
gingerBill
71e6288049
Use relaxed ordering for the early load in futex_wait and futex_wait_with_timeout, as every caller acquires itself after waking
2026-10-07 23:19:25 +01:00
gingerBill
ecedcbc0bf
Add @(futex=.Wait) and @(futex=.Wake) with intrinsics.Futex_Operation, so the atomic analysis checks futex words, and use relaxed ordering for Atomic_Cond's counter
2026-10-07 23:18:02 +01:00
gingerBill
7571a2676f
Use relaxed ordering for nbio's mpsc head counter and pool's destroy exchange, as their acquire paired with nothing
2026-10-07 22:44:36 +01:00
gingerBill
7743370a2d
Merge pull request #7815 from SOBEX/llvm-fake-use
...
Adds binding for `llvm.fake.use`
2026-10-07 22:13:03 +01:00
gingerBill
711f08a641
Make is_type_lock_free follow what each target lowers without library calls, and reject atomics on types which are not lock-free
2026-10-07 21:58:32 +01:00
gingerBill
403738e5cd
Merge pull request #7802 from odin-lang/bill/compiler-improvements-2026-10
...
Flow-based Escape Analysis of Stack Memory + Show where parapoly procedures and records were instantiated in errors within them
2026-10-07 21:19:06 +01:00
SOBEX
88c4c7a0f0
Adds binding for llvm.fake.use
2026-10-07 20:18:18 +02:00
gingerBill
e6d0112104
Add f16 variants to the core:sys/llvm standard C library bindings
2026-10-07 17:04:22 +01:00
gingerBill
6961b36425
Fix the core:sys/llvm bindings for current LLVM, add the missing ones; allow #const parameters to take their defaults, and reuse an intrinsic already declared by the compiler
2026-10-07 16:56:25 +01:00
gingerBill
9863858570
Fix the deterministic CBOR encoding of maps with non-string keys
2026-10-07 15:52:48 +01:00
gingerBill
c713bede2f
Stop core and tests leaving references to stack memory behind
2026-10-07 15:49:33 +01:00
Jeroen van Rijn
45a6f83679
Merge pull request #7801 from bplu4t2f/fix-7779
...
Fix #7779 - Make encoding/hex behave well on errors.
2026-10-07 02:29:31 -07:00
Jeroen van Rijn
c1ac93f9a6
`delete(dst, allocator)
2026-10-07 11:21:07 +02:00
bplu4t2f
758cdd19ae
Fix #7779 - Make encoding/hex behave well on errors.
2026-10-07 10:47:57 +02:00
Alexander Zhura
620a7ab1c6
Impl simd x86 avx2 intrinsics set
2026-10-05 22:20:52 +03:00
Jeroen van Rijn
8ca9cc4486
Merge pull request #7755 from pJotoro/patch-1
...
Add GetCursor and GetDpiForSystem procedures
2026-10-05 04:15:50 -07:00
Jeroen van Rijn
4781f71276
Merge pull request #7751 from alexthed1rk/fix-simd-x86-sse2-insert
...
Impl simd x86 sse2 _mm_insert_epi16 fix
2026-10-05 02:41:21 -07:00
pJotoro
e6d352495c
Add GetCursor and GetDpiForSystem procedures
2026-10-05 11:28:11 +02:00
bplu4t2f
a496d00f69
Fix incorrect number of bytes written when formatting floats with h/H
2026-10-05 10:41:27 +02:00
Alexander Zhura
6fd8888986
Impl simd x86 sse2 _mm_insert_epi16 fix
2026-10-05 10:03:06 +03:00
gingerBill
19cbd86501
Merge pull request #7738 from staledonuts/master
...
Fix glsl mix_vec4 and mix_dvec4 implementations
2026-10-04 18:24:41 +01:00
Jeroen van Rijn
dc0b041d59
#force_inline math.lerp, unlerp, saturate
2026-10-04 19:20:29 +02:00
Jeroen van Rijn
667efeb840
Merge pull request #7731 from Arthur031221/fix/timezone-utc-nanoseconds-20261004-s29
...
Preserve fractional timestamps in timezone conversions
2026-10-04 05:24:55 -07:00
DeadDonuts Corp.
0ade92df56
Fix glsl mix_vec4 and mix_dvec4 implementations
...
Fix mix implementation typo bug.
Line 763 and 766 never used the target value to progress y channel.
2026-10-04 11:01:30 +02:00
Jeroen van Rijn
930cb5dab3
Merge pull request #7733 from BradLewis/fix/odin-parser-for-stmt-comma
...
Fix odin parser assert with invalid for-in stmt
2026-10-03 15:26:13 -07:00
Brad Lewis
66ccc1011c
Fix odin parser assert with invalid for-in stmt
2026-10-04 09:11:04 +11:00
gingerBill
b079c1a23b
Merge pull request #7672 from MatzeOGH/fast_float_swar_parse
...
Second Part of the Fast Float Port (SWAR based scanning)
2026-10-03 22:31:02 +01:00
Arthur031221 and dozn
d046345202
Preserve fractional timestamps in timezone conversions
...
Apply timezone offsets to the normalized nanosecond timestamp.
Test fractional roundtrips, dates before 1970 and nanosecond carry.
Co-authored-by: dozn <16659513+dozn@users.noreply.github.com >
Signed-off-by: Arthur031221 <levi74108520963@gmail.com >
2026-10-04 03:39:11 +08:00
Jeroen van Rijn
d57f6d8232
Add a few more #no_bounds_check to XXH3.
2026-10-03 16:15:37 +02:00
corley
b785a3fc77
adds a config for opting out of Eisel-Lemire
2026-10-03 15:39:22 +03:00
Artur Leite
d8a598b1a3
Accept .Recv channels in chan.recv and chan.try_recv
2026-10-03 13:01:17 +02:00
gingerBill
1c7acfb1cf
Merge pull request #7707 from odin-lang/bill/doc-comment-markdown
...
Adjust doc comments that misrender as CommonMark
2026-10-02 14:36:30 +01:00
gingerBill
56e7efb1d4
Adjust doc comments that misrender as CommonMark
2026-10-02 14:36:03 +01:00
Arthur031221
eaa1cb34eb
Fix CBOR bit set byte order conversion
2026-10-02 18:57:49 +08:00
Arthur031221
f01284d372
[core:encoding/csv] Read the final multiline record before EOF
...
The multiline line reader returned EOF with buffered data, so record
parsing discarded the last row when the file had no trailing newline.
Clear EOF while data remains, as the normal line reader does.
Add regressions for final records, empty input and unterminated quotes.
The normal core suite passes 627 tests with the new CSV tests included.
2026-10-02 14:10:58 +08:00
gingerBill