LLVM and SPIRV-LLVM-Translator pulldown (WW49 2025) #20851
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
More missed target checks. Signed-off-by: Nick Sarnie <nick.sarnie@intel.com>
This is exactly like the 'copy', except the exit operation is a 'delete' instead of a 'copyout'. Also, creating the 'delete' op has one less argument to it, so we have to do some special handling when creating that.
Replaces the default "Success" std::error_code with a more meaningful one if `Magic != file_magic::pdb`.
```
---------------------------------------------------
Benchmark old new
---------------------------------------------------
BM_num_get<bool> 86.5 ns 32.3 ns
BM_num_get<long> 82.1 ns 30.3 ns
BM_num_get<long long> 85.2 ns 33.4 ns
BM_num_get<unsigned short> 85.3 ns 31.2 ns
BM_num_get<unsigned int> 84.2 ns 31.1 ns
BM_num_get<unsigned long> 83.6 ns 31.9 ns
BM_num_get<unsigned long long> 87.7 ns 31.5 ns
BM_num_get<float> 116 ns 114 ns
BM_num_get<double> 114 ns 114 ns
BM_num_get<long double> 113 ns 114 ns
BM_num_get<void*> 151 ns 144 ns
```
This patch applies multiple optimizations:
- Stages two and three of do_get are merged and a custom integer parser
has been implemented
This avoids allocations, removes the need for strto{,u}ll and avoids
__stage2_int_loop (avoiding extra writes to memory)
- std::find has been replaced with __atoms_offset, which uses vector
instructions to look for a character
Fixes #158100
Fixes #158102
…alescing SUBREG_TO_REG"
A SUBREG_TO_REG instruction expresses that the top bits of the result
register are set to a certain value (e.g. 0).
The example below expresses that the result of %1 will have the top 32
bits zeroed and the lower 32bits being equal to the result of INSTR.
```
%0:gpr32 = INSTR
%1:gpr64 = SUBREG_TO_REG 0, %0, sub32
```
When the RegisterCoalescer tries to remove SUBREG_TO_REG instructions by
coalescing %0 into %1, it must keep the same semantics. Currently
however, the RegisterCoalescer would emit:
```
%1.sub32:gpr64 = INSTR
```
which no longer expresses that the top 32-bits of the register are
defined (zeroed) by INSTR.
This may cause issues with e.g. machine copy propagation where the pass
may think it can remove a COPY-like instruction because the MIR says
only the bottom 32-bits are defined/used, even though other uses of the
register rely on the top 32-bits being zeroed by the COPY-like
instruction.
This PR changes the RegisterCoalescer to instead emit:
```
undef %1.sub32:gpr64 = MOVimm32 42, implicit-def %1
```
to express that the entire contents of %1:gpr64 are defined by the
instruction.
This tries to reland #134408 which had to be reverted due to a few reported
failures.
Reverts llvm/llvm-project#169318 Our builders are back online. I see them picking up existing jobs.
This is identical to 'copy' and 'copyin', except it uses 'create' and 'copyout' as its entry/exit op. This patch adds the same tests, and similar code for all of it.
The `#warning` causes diagnostics if system headers include deprecated headers. #168041 will add a way to deprecated headers properly, which then also interacts nicely with system header suppression.
Update getAllocTokenModeFromString() to recognize "default" as a valid mode string, mapping it to `DefaultAllocTokenMode`.
Signed-off-by: Tikhomirova, Kseniya <kseniya.tikhomirova@intel.com>
`[[nodiscard]]` should be applied to functions where discarding the return value is most likely a correctness issue. - https://libcxx.llvm.org/CodingGuidelines.html#apply-nodiscard-where-relevant
This one is another that is effectively identical to copy, copyin, and copyout, except its entry/exit ops pair is create/delete.
…n (#162458) The inliner uses DirectSP to check if a function has instructions that modify the SP. Exceptions are stack Push and Pop instructions. We can also allow pointer signing and authenticating instructions. The inliner removes the Return instructions from the inlined functions. If it is a fused pointer-authentication-and-return (e.g. RETAA), we have to generate a new authentication instruction.
**Problem** In Rust, checked math functions (like `checked_mul`, `overflowing_mul`, `saturating_mul`) are part of the primitive implementation of integers ([see u64](https://doc.rust-lang.org/std/primitive.u64.html) for instance). The Rust compiler builds the Rust [compiler-builtins](https://github.com/rust-lang/compiler-builtins) crate as a step in the compilation processes, since it contains the math builtins to be lowered in the target. For BPF, however, when using those functions in Rust we hit the following errors: ``` ERROR llvm: <unknown>:0:0: in function func i64 (i64, i64): A call to built-in function '__multi3' is not supported. ERROR llvm: <unknown>:0:0: in function func i64 (i64, i64): only small returns supported ``` Those errors come from the following code: ``` pub fn func(a: u64, b: u64) -> u64 { a.saturating_mul(b) } ``` Those functions invoke underneath the llvm instrinc `{ i64, i1 } @llvm.umul.with.overflow.i64(i64, i64)` or its variants. It is very useful to use safe math operations when writing BPF code in Rust, and I would like to add support for those in the target. **Changes** 1. Create a target feature `allow-builtin-calls` to enable code generation for builtin functions. 2. Implement `CanLowerReturn` to fix the error `only small returns supported`. 3. Add code to correctly invoke lib functions. 4. Add a test case together with the corresponding C code.
This patch fixes a minor typo in the **Kaleidoscope tutorial (Chapter 2)**. The sentence: “checks to see if **if** is too low” has been corrected to: “checks to see if **it** is too low”. This is a documentation-only change and does not affect any semantic behavior or code generation. Thank you for maintaining the tutorial, and please let me know if any further adjustments are needed.
This adds some trivial handling to force emitting of child decls inside C++ records.
…dialect (#169194) Enables `amdgcn.named.barrier` target extension type as a global variable type in MLIR.
…8134) Make sure the CoroSplitter pass correctly handles `#dbg_declare_value` intrinsics. Which means, it should identify them, and convert them to `#dbg_declares` so that any subsequent passes do not need to be amended to support the `#dbg_declare_value` intrinsic. More information here: https://discourse.llvm.org/t/rfc-introduce-new-llvm-dbg-coroframe-entry-intrinsic/88269 This patch is the second and last in a stack of patches, with the one preceding it being: llvm/llvm-project#168132
…#169364) This is to resolve a regression caused by #168534. Now when we have an anonymous object like a struct or union that has a typedef attached, we print the typedef name instead of listing it as anonymous.
This patch adds Mustache HTML tests alongside the legacy HTML backend for namespace output. This way, we can see exactly where the output currently differs before replacing the legacy backend. The same thing will be done for all other tests where the legacy HTML backend is tested.
Changes: Fix a missed update to WidenGEP::usesFirstLaneOnly, and include reduced-case test that was previously hitting the new assert: the underlying reason was that VPWidenGEP::usesScalars was too weak, and the single-scalar WidenGEP was not narrowed by narrowToSingleScalarRecipes. This allows us to strip a special case in VPWidenGEP::execute.
In preparation to extend the work done by dfa665f ([VPlan] Add transformation to narrow interleave groups) to make the narrowing more powerful, pre-commit a test case from #128062.
…n (#169114) The `acc.present` Op as generated by ACCImplicitData does not provide a way to differentiate between `acc.present` ops that are generated implicitly and the ones that are generated as result of an explicit `default(present)` clause in the source code. This differentiation would allow for better communication to the user on the decisions made by the compiler while managing data automatically between the host and the device. This commit adds this information as a discardable attribute on the `acc.present` op.
…#169055) This reverts commit b83e458. Also undo the use of namespace qualifier for `ReducePassList` as that seems to cause build failures.
Just like the last handful of patches that did copy, copyin, copyout,
create, etc, this patch has the exact same behavior, except the
entry op is a present, and the exit is delete.
Adds initial (preview) support for device scope barriers via SPV_INTEL_device_barrier. See: #12092 Original commit: KhronosGroup/SPIRV-LLVM-Translator@36a6c287e3ecbb7
Signed-off-by: Nick Sarnie <nick.sarnie@intel.com>
iclsrc
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
iclsrc
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
iclsrc
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
iclsrc
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
iclsrc
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
…removal (#33284) intel-innersource/applications.compilers.source.llvm-spirv-translator@5458eb2 Removed entry point wrapper, so the offset is smaller.
jsji
force-pushed
the
llvmspirv_pulldown
branch
from
bbe7a9e to
05c2241
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
LLVM: llvm/llvm-project@bd04ef6
SPIRV-LLVM-Translator: KhronosGroup/SPIRV-LLVM-Translator@36a6c287e3ecbb7