LLVM and SPIRV-LLVM-Translator pulldown (WW22 2025) #18738

iclsrc · 2025-05-30T02:36:27Z

LLVM: llvm/llvm-project@7a8090c
SPIRV-LLVM-Translator: KhronosGroup/SPIRV-LLVM-Translator@6efaf7767be0aca

Switch from `int64_t` to `int64_t*` to fit with the rest of the implementation. New tentative with some fix. The previous was reverted some time ago. Reviewed in #138010

…nts. (#140162) Recently, I got a report how a user 'hung', and come to find out it's actually because DAP is checking percentage, and on non-deterministic events, we will effectively never send a progress event to DAP. Here we short circuit and don't view percentages for DAP messages and solely depend on the DAP 250ms throttle, which is probably still too aggressive, but better than no updates.

For llvm/llvm-project#140529

if the `cwd` option is empty this means the adapter is created in the home folder. Any relative file saved is in the home folder. The files should be relative to the current workspace folder.

For #140529.

Having the whole clause emission be in a header file ended up being pragmatic, but ended up being a sizable negative for a variety of reasons. This patch moves it to its own .cpp file and makes CIRGenFunction instead call into the visitor via a template instead. This is possible because the valid list of construct kinds is quite finite, and easy to enumerate.

Add support for riscv32 ELF corefiles.

…e (#140620) Adds checks for resource globals that were created for DXIL metadata. The names of the globals and the names of the types will be changing soon. Adding these to the baseline will make it easier to see what is changing.

…#140501) Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

This test seems to be non-determistic recently. Changed it not to rely on offsets.

…duleSummary (Core) to a separate header (#140505) Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

Instead of looking through all the vectorizable tree to find the operand entry, better to store it in a separate map and perform quick lookup, basing on user tree entry and operand index. It allows to remove lots of duplicated code, simplify processing and fix potential future issues with the analysis, affected by the codegen. Also, improves compile time. Reviewers: HanKuanChen, RKSimon, hiraditya Reviewed By: hiraditya Pull Request: llvm/llvm-project#140549

…#140637)

…#140631)

This clarifies the outcome of the discussion in https://discourse.llvm.org/t/clarifiying-the-semantics-of-ptrtoint/83987/54 In the future, we will also introduce a non-capturing pointer -> address conversion using a new `ptrtoaddr` instruction. Reviewed By: krzysz00 Pull Request: llvm/llvm-project#139349

…y (#140650) #140505 dropped the dependency on core but it's still needed as indicated by the failing dynamically linked builds.

Of the 128-bits of buffer descriptor only 48 bits are address bits, so following the discussion on https://discourse.llvm.org/t/clarifiying-the-semantics-of-ptrtoint/83987/54, the logic conclusion is to set the index width to 48 bits instead of the current value of 128. Most of the test changes are mechanical datalayout updates, but there is one actual change: the ptrmask test now uses .i48 instead of .i128 and I had to update SelectionDAGBuilder to correctly extend the mask. Reviewed By: krzysz00 Pull Request: llvm/llvm-project#139419

When linker relaxation is enabled, relaxable relocations are followed by a R_RISCV_RELAX/R_LARCH_RELAX relocation. They are encoded as two fixups by CodeEmitter and expected to have the same `IsResolved` value within MCAssembler::evaluateFixup (they must lead to either 0 or 2 relocations). This scheme wasite space and requires RISCVAsmBackend::shouldForceRelocation to be conservative. This patch introduces MCFixup::NeedsRelax to encode the RELAX relocation implicitly. The fixup will lead to either 0 or 2 relocations. Pull Request: llvm/llvm-project#140494

getArgumentsIfRequest is returning a local variable by value. Using std::move is not needed and may inhibit copy elision. Also fixes the braces around a single-line if.

jsji · 2025-05-31T18:47:49Z

This is ready for review.

libclc cherry-picks and conflict resolution @frasercrmck
[SYCL] Don't set alwaysinline for NativeCPU (#29546) @intel/dpcpp-cfe-reviewers
[NFC] Update intel-fpga-local.cpp after 61b435e @intel/dpcpp-cfe-reviewers
[Clang] Fix conflict resolution in SemaDeclAttr @intel/dpcpp-cfe-reviewers
Revert "Deprecate SPV_INTEL_hw_thread_queries extension from forward … @MrSidims The change in llvm-spirv is breaking ESIMD, so I have to revert it first. If we are sure ESIMD usage can be removed, please work with @intel/dpcpp-esimd-reviewers to remove failing tests and reland.
[SYCL][E2E] Disable NEO Cache for shaderdump check @intel/dpcpp-esimd-reviewers

frasercrmck · 2025-06-03T10:34:41Z

I merged #18626 which caused conflicts in libclc here, so I merged sycl, resolved the conflicts and push the fix. I hope that's alright.

Fznamznon · 2025-06-03T13:52:25Z

clang/lib/CodeGen/CodeGenModule.cpp

@@ -6899,7 +6899,7 @@ void CodeGenModule::EmitGlobalFunctionDefinition(GlobalDecl GD,
    if (GD.getKernelReferenceKind() == KernelReferenceKind::Stub &&
        !D->hasAttr<NoInlineAttr>() &&
        !Fn->hasFnAttribute(llvm::Attribute::NoInline) &&
-        !D->hasAttr<OptimizeNoneAttr>() &&
+        !D->hasAttr<OptimizeNoneAttr>() && !LangOpts.SYCLIsNativeCPU &&


Shouldn't we also add !LangOpts.SYCLIsDevice?

Sure, we can.

sarnex

NEO_CACHE_PERSISTENT lgtm!

sarnex · 2025-06-03T18:38:43Z

/merge

bb-sycl · 2025-06-03T18:39:19Z

Tue 03 Jun 2025 06:39:18 PM UTC --- Start to merge the commit into sycl branch. It will take several minutes.

jsji · 2025-06-03T18:39:19Z

@intel/llvm-gatekeepers This is ready for merge -- UR CI failures are common to sycl branch.

bb-sycl · 2025-06-03T19:00:40Z

Tue 03 Jun 2025 07:00:40 PM UTC --- Merge the branch in this PR to base automatically. Will close the PR later.

clementval and others added 30 commits May 19, 2025 15:02

[flang][cuda] Use a reference for asyncObject (#140614)

f5609aa

Switch from `int64_t` to `int64_t*` to fit with the rest of the implementation. New tentative with some fix. The previous was reverted some time ago. Reviewed in #138010

[lldb-dap] Remove unused enum (NFC)

2dd6a8c

[NFC] Pre-commit UBSAN src:*=sanitize test (#140602)

91a45a3

For llvm/llvm-project#140529

[lldb-dap] launch the adapter from the workspace folder. (#140604)

2523a83

if the `cwd` option is empty this means the adapter is created in the home folder. Any relative file saved is in the home folder. The files should be relative to the current workspace folder.

Merge from 'main' to 'sycl-web' (221 commits)

f1076da

[NFC] Extend ubsan-src-ignorelist-category.test

b93bc77

For #140529.

[NFC] Extend ubsan-src-ignorelist-category.test

5db4aea

For #140529.

[RISCV] Fix copy/paste mistake in test comments. NFC

050e49a

[lldb] Support riscv32 ELF corefiles (#115408)

dfabd61

Add support for riscv32 ELF corefiles.

[gn build] Port dfabd61

c519190

Merge from 'sycl' to 'sycl-web' (16 commits)

b64487d

[NFC][MemProf] Fix typo in type name (#140500)

7268c4e

[NFC][MemProf] Move Radix tree methods to their own header and cpp. (…

a53b306

…#140501) Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

[NFC][MemProf] Move getGUID out of IndexedMemProfRecord (#140502)

ad3c1d2

Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

[NFC][MemProf] Move IndexedMemProfData to its own header. (#140503)

0528848

Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

[NFC][MemProf] Add the LLVM license text and minor clean up. (#140504)

c2fd63c

Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

[dsym][test][nfc] Remove DIE offsets. (#140628)

ad45031

This test seems to be non-determistic recently. Changed it not to rely on offsets.

[NFC][MemProf] Move types shared between Analysis, ProfileData and Mo…

4cfbe55

…duleSummary (Core) to a separate header (#140505) Part of a larger refactoring with the following goals 1. Reduce the size of MemProf.h 2. Avoid including ModuleSummaryIndex just for a couple of types

[gn build] Port a53b306

0a71879

[flang][cuda] Set implicit CUDA device attribute in block construct (…

73c638f

…#140637)

[RISCV] Add LD_RV32/SD_RV32 to RISCVInstrInfo::canFoldIntoAddrMode. (…

e9c0840

…#140631)

[ProfileData] Restore Core as a dependency for the ProfileData librar…

90daed3

…y (#140650) #140505 dropped the dependency on core but it's still needed as indicated by the failing dynamically linked builds.

[lldb-dap] Remove spurious move (NFC) (#140641)

df0358f

getArgumentsIfRequest is returning a local variable by value. Using std::move is not needed and may inhibit copy elision. Also fixes the braces around a single-line if.

jsji temporarily deployed to WindowsCILock May 31, 2025 17:43 — with GitHub Actions Inactive

Merge remote-tracking branch 'origin/sycl' into llvmspirv_pulldown

2736632

frasercrmck temporarily deployed to WindowsCILock June 3, 2025 10:33 — with GitHub Actions Inactive

frasercrmck temporarily deployed to WindowsCILock June 3, 2025 10:34 — with GitHub Actions Inactive

frasercrmck temporarily deployed to WindowsCILock June 3, 2025 11:03 — with GitHub Actions Inactive

frasercrmck temporarily deployed to WindowsCILock June 3, 2025 11:09 — with GitHub Actions Inactive

Fznamznon reviewed Jun 3, 2025

View reviewed changes

jsji had a problem deploying to WindowsCILock June 3, 2025 14:24 — with GitHub Actions Error

jsji had a problem deploying to WindowsCILock June 3, 2025 14:25 — with GitHub Actions Error

jsji had a problem deploying to WindowsCILock June 3, 2025 14:27 — with GitHub Actions Error

jsji requested a review from Fznamznon June 3, 2025 14:29

Fznamznon approved these changes Jun 3, 2025

View reviewed changes

jsji added 2 commits June 3, 2025 09:36

[SYCL] Also disable alwasyinline for SYCL device

1631cf7

Merge remote-tracking branch 'origin/sycl' into llvmspirv_pulldown

cdf4060

jsji force-pushed the llvmspirv_pulldown branch from 91dbbfc to cdf4060 Compare June 3, 2025 16:37

jsji had a problem deploying to WindowsCILock June 3, 2025 16:37 — with GitHub Actions Failure

jsji temporarily deployed to WindowsCILock June 3, 2025 16:37 — with GitHub Actions Inactive

jsji temporarily deployed to WindowsCILock June 3, 2025 18:04 — with GitHub Actions Inactive

sarnex approved these changes Jun 3, 2025

View reviewed changes

bb-sycl approved these changes Jun 3, 2025

View reviewed changes

bb-sycl merged commit 19964c3 into sycl Jun 3, 2025
42 of 54 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

LLVM and SPIRV-LLVM-Translator pulldown (WW22 2025) #18738

LLVM and SPIRV-LLVM-Translator pulldown (WW22 2025) #18738

iclsrc commented May 30, 2025

Uh oh!

jsji commented May 31, 2025 •

edited

Loading

Uh oh!

frasercrmck commented Jun 3, 2025

Uh oh!

Fznamznon Jun 3, 2025

Uh oh!

jsji Jun 3, 2025

Uh oh!

sarnex left a comment

Uh oh!

sarnex commented Jun 3, 2025

Uh oh!

bb-sycl commented Jun 3, 2025

Uh oh!

jsji commented Jun 3, 2025

Uh oh!

bb-sycl commented Jun 3, 2025

Uh oh!

Uh oh!

Uh oh!

LLVM and SPIRV-LLVM-Translator pulldown (WW22 2025) #18738

LLVM and SPIRV-LLVM-Translator pulldown (WW22 2025) #18738

Conversation

iclsrc commented May 30, 2025

Uh oh!

jsji commented May 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

frasercrmck commented Jun 3, 2025

Uh oh!

Fznamznon Jun 3, 2025

Choose a reason for hiding this comment

Uh oh!

jsji Jun 3, 2025

Choose a reason for hiding this comment

Uh oh!

sarnex left a comment

Choose a reason for hiding this comment

Uh oh!

sarnex commented Jun 3, 2025

Uh oh!

bb-sycl commented Jun 3, 2025

Uh oh!

jsji commented Jun 3, 2025

Uh oh!

bb-sycl commented Jun 3, 2025

Uh oh!

Uh oh!

Uh oh!

jsji commented May 31, 2025 •

edited

Loading