mikros/rust
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
3,850 Commits 2 Branches 0 Tags
Commit Graph

1595 Commits

Author SHA1 Message Date
bjorn3
e6e2f00d21 Fix incorrect implementation of several vendor intrinsics 2023-11-25 09:14:37 +00:00
bjorn3
3b49b9efd5 Implement the int part of the gather family vendor intrinsics 2023-11-24 20:45:22 +01:00
bjorn3
d5a7ae7976 Implement the float part of the gather family vendor intrinsics 2023-11-24 20:45:22 +01:00
bjorn3
c8729e9216 Implement _mm256_zeroupper vendor intrinsic 2023-11-24 20:45:22 +01:00
bjorn3
65da671694 Implement *fmaddsub_p*, *fmsubadd_p* and *fnmadd_p* vendor intrinsics 2023-11-24 20:45:22 +01:00
bjorn3
705031d017 Implement _mm_cvttps_epi32 2023-11-24 20:45:22 +01:00
bjorn3
3ec8d7aa4a Implement _mm_cmpestri and _mm_cmpestrm using inline asm 2023-11-24 13:39:51 +00:00
bjorn3
9c958196e7 Fix polymorphization for coroutines 
Fixes rust-lang/rustc_codegen_cranelift#1429
 2023-11-24 10:46:22 +01:00
bjorn3
430ab4e923 Allow function pointers in CValue::const_val 
Fixes rust-lang/rustc_codegen_cranelift#1430
 2023-11-24 10:22:59 +01:00
bjorn3
a5fcfad890 Fix _mm_sad_epu8 implementation 
Fixes rust-lang/rustc_codegen_cranelift#1426
 2023-11-21 17:00:43 +01:00
bjorn3
cd7218aa9f Update Cranelift to 0.102 2023-11-21 12:10:37 +01:00
bjorn3
ca85cc3c7b Implement SHA256 intrinsics using inline asm 2023-11-12 17:17:57 +00:00
bjorn3
813f8b4fd0 Use inline asm for _mm_clmulepi64_si128 
This is a lot more compact and significantly faster
 2023-11-12 17:10:46 +00:00
bjorn3
dc60334777 Implement AES-NI intrinsics using inline asm 2023-11-12 17:10:46 +00:00
bjorn3
6ef877c8c2 Support narrowing casts in mir_operand_get_const_val 2023-11-12 17:10:46 +00:00
bjorn3
2769ac03ba Inline codegen_xgetbv into llvm_x86.rs 2023-11-12 17:10:46 +00:00
bjorn3
9205667c0d Extract codegen_inline_asm_inner function and use in codegen_xgetbv 2023-11-12 17:10:46 +00:00
bjorn3
7947070789 Minor changes to codegen_inline_asm_terminator 2023-11-12 17:10:46 +00:00
bjorn3
82487a9447
Merge pull request #1417 from rust-lang/implement_xgetbv 
Implement a lot of SIMD intrinsics
 2023-11-07 19:49:50 +01:00
bjorn3
864973135a Implement all vendor intrinsics used by the simd-json crate 2023-11-07 18:10:12 +00:00
bjorn3
ecf79a304a Implement all vendor intrinsics used by the fimg crate 2023-11-07 13:32:11 +00:00
bjorn3
0a35232c85 Implement all vendor intrinsics used by the httparse crate 2023-11-07 09:42:01 +00:00
bjorn3
61e38ceea7 Implement all SSE intrinsics used by the jpeg-decoder crate 2023-11-07 08:59:16 +00:00
bjorn3
438194980b Implement all avx2 intrinsics used by the image crate 2023-11-05 19:49:30 +00:00
bjorn3
6a53acefd8 Implement _mm256_permute2f128_ps and _mm256_permute2f128_pd intrinsics 2023-11-05 18:24:39 +00:00
Afonso Bordado
70a6abfd29 Add unsigned saturating add/sub intrinsics for aarch64 2023-11-04 19:37:36 +00:00
Afonso Bordado
88c2e7896b Implement aarch64 addp intrinsics 2023-11-04 19:11:15 +00:00
Afonso Bordado
1f09bae6a8 Implement min/max neon intrisics 2023-11-04 19:07:13 +00:00
bjorn3
f6a8c3afb5 Add real implementation of _xgetbv() 2023-11-02 16:10:15 +00:00
bjorn3
909513ef74 Use Value instead of CValue in CInlineAsmOperand 2023-11-02 15:32:44 +00:00
bjorn3
c04ceb4342 Fix workaround for the int $$0x29 issue to not crash on empty inline asm 2023-11-02 09:52:21 +00:00
bjorn3
361585e06d Sync from rust 75b064d26970ca8e7a487072f51835ebb057d575 2023-11-02 09:38:34 +00:00
George Bateman
03c9acdd8f Support enum variants in offset_of! 2023-10-31 23:25:54 +00:00
bjorn3
48ca2d9703 Implement llvm.fma.v* intrinsics 
cc #1405
 2023-10-31 18:38:31 +00:00
Guillaume Gomez
aed0ed2875 Rollup merge of #117317 - RalfJung:track-caller, r=oli-obk 
share some track_caller logic between interpret and codegen

Also move the code that implements the track_caller intrinsics out of the core interpreter engine -- it's just a helper creating a const-allocation, doesn't need to be part of the interpreter core.
 2023-10-30 17:33:16 +01:00
bjorn3
41dcb52153 Merge commit 'dde58803fd6cbb270c7a437f36a8a3a29fbef679' into sync_cg_clif-2023-10-29 2023-10-29 20:30:50 +00:00
bjorn3
e281e6fdb2 Sync from rust e5cfc55477eceed1317a02189fdf77a4a98f2124 2023-10-29 20:18:11 +00:00
bjorn3
69c6aa560c Avoid infinite recursion when cranelift is the default codegen backend 2023-10-29 19:45:30 +00:00
bjorn3
35453ac0eb Stabilize inline asm usage on all platforms 
But exclude sym operands for now as they are somewhat broken.
 2023-10-29 16:40:52 +00:00
bjorn3
b1e705146a Unconditionally handle int $$0x29 using a shim 2023-10-29 16:08:16 +00:00
bjorn3
827a6d8d9f Use the LLVM rustc backend as external assembler 
The LLVM backend is generally available, while the gnu assembler is not
on Windows and many other platforms by default.
 2023-10-29 16:06:32 +00:00
Ralf Jung
c6f5090294 share the track_caller handling within a mir::Body 2023-10-28 16:16:15 +02:00
Ralf Jung
bad4be6e29 interpret: call caller_location logic the same way codegen does, and share some code 2023-10-28 15:40:03 +02:00
bors
9e20870c2e Auto merge of #116609 - eduardosm:bump-stdarch, r=workingjubilee 
Bump stdarch submodule and remove special handling for LLVM intrinsics that are no longer needed

Bumps stdarch to pull https://github.com/rust-lang/stdarch/pull/1477, which reimplemented some functions with portable SIMD intrinsics instead of arch specific LLVM intrinsics.

Handling of those LLVM intrinsics is removed from cranelift codegen and miri.

cc `@RalfJung` `@bjorn3`
 2023-10-28 11:26:34 +00:00
bjorn3
a302610016 Merge commit '93a5433f17ab5ed48cc88f1e69b0713b16183373' into sync_cg_clif-2023-10-24 2023-10-24 12:22:23 +00:00
bjorn3
a558968dbe Implement all llvm intrinsics necessary for the image crate 
Fixes #1379
 2023-10-23 14:10:22 +00:00
bjorn3
e5ba1e8417 Implement llvm intrinsics necessary for rav1e 
Fixes #1399
 2023-10-23 14:06:47 +02:00
bjorn3
ed8c515cc4 Avoid collect_and_partition_mono_items call when not doing codegen 
Fixes #1401
 2023-10-23 09:35:15 +00:00
bjorn3
56c6c86661 Simplify FunctionCx::create_stack_slot a bit 2023-10-22 18:10:59 +00:00
bjorn3
55380a5ffd Dynamically realign local variables with large alignment 
This works around the lack of a way to specify the alignment of a stack
slot in Cranelift.

Fixes #1230
Fixes #1381
 2023-10-22 17:11:11 +00:00