Page MenuHomePhabricator
Feed Advanced Search

Yesterday

RKSimon committed rGb743e94cdca9: [TargetLowering] SimplifyDemandedBits - add EXTRACT_SUBVECTOR support. (authored by RKSimon).
[TargetLowering] SimplifyDemandedBits - add EXTRACT_SUBVECTOR support.
Sat, Sep 14, 9:42 AM
RKSimon committed rL371923: [TargetLowering] SimplifyDemandedBits - add EXTRACT_SUBVECTOR support..
[TargetLowering] SimplifyDemandedBits - add EXTRACT_SUBVECTOR support.
Sat, Sep 14, 9:37 AM
RKSimon accepted D67362: [SLP] limit vectorization of Constant subclasses (PR33958).

LGTM - cheers

Sat, Sep 14, 4:24 AM · Restricted Project

Fri, Sep 13

RKSimon updated subscribers of rL371722: AArch64: support arm64_32, an ILP32 slice for watchOS..

ping @t.p.northover

Fri, Sep 13, 10:06 AM
RKSimon added a comment to rL371722: AArch64: support arm64_32, an ILP32 slice for watchOS..

@tnorthover There are a number of EXPENSIVE_CHECKS failures due to this: http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/19653/steps/test-check-all/logs/stdio

Fri, Sep 13, 10:05 AM
RKSimon added a comment to D67362: [SLP] limit vectorization of Constant subclasses (PR33958).

Can we easily get a constexpr test?

Fri, Sep 13, 9:26 AM · Restricted Project
RKSimon accepted D67266: [X86][NFC] Add a `use-aa` feature..

LGTM

Fri, Sep 13, 9:15 AM · Restricted Project
RKSimon created D67557: [DAG][X86] Convert isNegatibleForFree/GetNegatedExpression to a target hook (PR42863).
Fri, Sep 13, 8:39 AM · Restricted Project
RKSimon committed rG930ebc15a639: [X86] negateFMAOpcode - extend to support FMADDSUB/FMSUBADD and output negation. (authored by RKSimon).
[X86] negateFMAOpcode - extend to support FMADDSUB/FMSUBADD and output negation.
Fri, Sep 13, 4:23 AM
RKSimon committed rL371840: [X86] negateFMAOpcode - extend to support FMADDSUB/FMSUBADD and output negation..
[X86] negateFMAOpcode - extend to support FMADDSUB/FMSUBADD and output negation.
Fri, Sep 13, 4:21 AM
RKSimon accepted D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

Accepting this to remove the block

Fri, Sep 13, 1:45 AM · Restricted Project, Restricted Project

Thu, Sep 12

RKSimon committed rGd67661ee2406: [X86] Move negateFMAOpcode helper earlier to help future patch. NFCI. (authored by RKSimon).
[X86] Move negateFMAOpcode helper earlier to help future patch. NFCI.
Thu, Sep 12, 1:39 PM
RKSimon committed rL371770: [X86] Move negateFMAOpcode helper earlier to help future patch. NFCI..
[X86] Move negateFMAOpcode helper earlier to help future patch. NFCI.
Thu, Sep 12, 1:38 PM
RKSimon added a comment to D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

Thanks for looking into the problem - sorry for the delay!

Thu, Sep 12, 10:51 AM · Restricted Project, Restricted Project
RKSimon added a comment to D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

Thanks - this builds cleanly on MSVC now

Thu, Sep 12, 4:05 AM · Restricted Project, Restricted Project
RKSimon committed rGda59a6bf7db1: [DAGCombine] visitFDIV - Use isCheaperToUseNegatedFPOps helper for (fdiv (fneg… (authored by RKSimon).
[DAGCombine] visitFDIV - Use isCheaperToUseNegatedFPOps helper for (fdiv (fneg…
Thu, Sep 12, 4:03 AM
RKSimon committed rL371724: [DAGCombine] visitFDIV - Use isCheaperToUseNegatedFPOps helper for (fdiv (fneg….
[DAGCombine] visitFDIV - Use isCheaperToUseNegatedFPOps helper for (fdiv (fneg…
Thu, Sep 12, 4:02 AM

Wed, Sep 11

RKSimon added a comment to D67259: [X86] Enable -mprefer-vector-width=256 by default for Skylake-avx512 and later Intel CPUs..

In which case I'm happy to go with this.

Wed, Sep 11, 9:55 AM · Restricted Project
RKSimon added a comment to D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

rebase?

Wed, Sep 11, 7:32 AM · Restricted Project, Restricted Project
RKSimon committed rGe0d9a0bd59d8: Fix -Wdocumentation warning - void function doesn't need a @returns. NFCI. (authored by RKSimon).
Fix -Wdocumentation warning - void function doesn't need a @returns. NFCI.
Wed, Sep 11, 4:15 AM
RKSimon added a comment to D67101: Target hooks for custom COPY insertion..

@alex-t This is failing on EXPENSIVE_CHECKS builds: http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/19594/steps/test-check-all/logs/stdio

Wed, Sep 11, 4:15 AM · Restricted Project
RKSimon committed rL371606: Fix -Wdocumentation warning - void function doesn't need a @returns. NFCI..
Fix -Wdocumentation warning - void function doesn't need a @returns. NFCI.
Wed, Sep 11, 4:11 AM

Tue, Sep 10

RKSimon added a comment to D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

Totally missed that - thanks for noticing. I must have forgotten to remove stuff from the header since clang/gcc don't warn about it.
I'll get hold of a Windows machine soon-ish and I'll make sure to fix this problem.
Thanks!

Tue, Sep 10, 3:24 PM · Restricted Project, Restricted Project
RKSimon committed rG937ca6815743: [X86] Add AVX partial dependency tests as noted on D67363 (authored by RKSimon).
[X86] Add AVX partial dependency tests as noted on D67363
Tue, Sep 10, 7:28 AM
RKSimon committed rL371525: [X86] Add AVX partial dependency tests as noted on D67363.
[X86] Add AVX partial dependency tests as noted on D67363
Tue, Sep 10, 7:28 AM

Mon, Sep 9

RKSimon added a reviewer for D67363: [BreakFalseDeps] ignore function with minsize attribute: andreadb.

This would make it policy for -Oz builds to not bother to break dependencies but -Os/-O0+ builds would still do.

Mon, Sep 9, 3:53 PM · Restricted Project
RKSimon added a comment to D66322: [X86] Updated target specific selection dag code to conservatively check for isAtomic in addition to isVolatile.

Thanks @reames - no more comments from me.

Mon, Sep 9, 3:50 PM · Restricted Project
RKSimon added a comment to D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

I am providing definitions in the C++ file - the problem is that they are not available in the header before the extern declaration. The methods are available at the site of the extern definition.
gcc and clang accept this, so does Visual Studio 2019. This feels like an incorrect implementation of extern templates in Visual Studio?

I see two ways to proceed: move everything into a header (would like to avoid this) or silence the warning on VC++ (not great either).
Is there a better way? Which option is less bad from these two?

Mon, Sep 9, 3:01 PM · Restricted Project, Restricted Project
RKSimon committed rG7f37d9a714d3: Fix MSVC "not all control paths return a value" warning. NFCI. (authored by RKSimon).
Fix MSVC "not all control paths return a value" warning. NFCI.
Mon, Sep 9, 2:33 PM
RKSimon committed rL371454: Fix MSVC "not all control paths return a value" warning. NFCI..
Fix MSVC "not all control paths return a value" warning. NFCI.
Mon, Sep 9, 2:33 PM
RKSimon accepted D67312: [SelectionDAG] Remove ISD::FP_ROUND_INREG.

LGTM - cheers

Mon, Sep 9, 10:05 AM · Restricted Project
RKSimon committed rG9ede7c039563: Fix typo in comment noticed in D60295. NFCI. (authored by RKSimon).
Fix typo in comment noticed in D60295. NFCI.
Mon, Sep 9, 9:06 AM
RKSimon committed rL371415: Fix typo in comment noticed in D60295. NFCI..
Fix typo in comment noticed in D60295. NFCI.
Mon, Sep 9, 9:06 AM
RKSimon added a comment to D64146: [Clang Interpreter] Initial patch for the constexpr interpreter.

@nand The MSVC warnings are self explanatory - you've declared a number of methods (visitIndirectMember, emitConv and getPtrConstFn) but not provided definitions, as they're on template classes MSVC complains, a lot.

Mon, Sep 9, 8:59 AM · Restricted Project, Restricted Project
RKSimon added a comment to D66936: [DFAPacketizer] Track resources for packetized instructions.

Thanks @jmolloy rL371399 looks OK

Mon, Sep 9, 7:52 AM · Restricted Project
RKSimon added a comment to D67312: [SelectionDAG] Remove ISD::FP_ROUND_INREG.

We should probably mention this in the release notes

Mon, Sep 9, 6:07 AM · Restricted Project
RKSimon accepted D67349: [Inliner][NFC] Make test less brittle..

LGTM

Mon, Sep 9, 6:04 AM · Restricted Project
RKSimon added a comment to D66936: [DFAPacketizer] Track resources for packetized instructions.

@jmolloy This is causing "compiler is out of heap space" errors on my VS2017 and VS2019 all targets builds:

e:\llvm\ninja17\lib\target\hexagon\hexagongendfapacketizer.inc(4892) : fatal error C1002: compiler is out of heap space in pass 2
Mon, Sep 9, 5:40 AM · Restricted Project
RKSimon committed rG462e3d8050fa: Revert rL371198 from llvm/trunk: [DFAPacketizer] Track resources for packetized… (authored by RKSimon).
Revert rL371198 from llvm/trunk: [DFAPacketizer] Track resources for packetized…
Mon, Sep 9, 5:36 AM
RKSimon committed rL371393: Revert rL371198 from llvm/trunk: [DFAPacketizer] Track resources for packetized….
Revert rL371198 from llvm/trunk: [DFAPacketizer] Track resources for packetized…
Mon, Sep 9, 5:32 AM
RKSimon added a reverting change for rL371198: [DFAPacketizer] Track resources for packetized instructions: rL371393: Revert rL371198 from llvm/trunk: [DFAPacketizer] Track resources for packetized….
Mon, Sep 9, 5:32 AM
RKSimon added a comment to D67334: [InstCombine] fold sign-bit compares of srem.

Slightly beyond the scope of this patch, but would it be realistic to support non-constant cases? I know at present m_Power2 can't handle this

Mon, Sep 9, 4:48 AM · Restricted Project
RKSimon accepted D67289: [X86] Allow _MM_FROUND_CUR_DIRECTION and _MM_FROUND_NO_EXC to be used together on instructions that only support SAE and not embedded rounding..

LGTM

Mon, Sep 9, 4:46 AM · Restricted Project
RKSimon added a comment to D66936: [DFAPacketizer] Track resources for packetized instructions.

@jmolloy This is causing "compiler is out of heap space" errors on my VS2017 and VS2019 all targets builds:

e:\llvm\ninja17\lib\target\hexagon\hexagongendfapacketizer.inc(4892) : fatal error C1002: compiler is out of heap space in pass 2
Mon, Sep 9, 3:53 AM · Restricted Project

Sun, Sep 8

RKSimon committed rGe0ea746215c8: [X86][SSE] SimplifyDemandedVectorEltsForTargetNode - add faux shuffle support. (authored by RKSimon).
[X86][SSE] SimplifyDemandedVectorEltsForTargetNode - add faux shuffle support.
Sun, Sep 8, 2:40 PM
RKSimon committed rL371353: [X86][SSE] SimplifyDemandedVectorEltsForTargetNode - add faux shuffle support..
[X86][SSE] SimplifyDemandedVectorEltsForTargetNode - add faux shuffle support.
Sun, Sep 8, 2:37 PM
RKSimon committed rG178cd2cd3a69: [X86][SSE] Fix out of range shift introduced in D67070/rL371328 (authored by RKSimon).
[X86][SSE] Fix out of range shift introduced in D67070/rL371328
Sun, Sep 8, 5:49 AM
RKSimon committed rL371330: [X86][SSE] Fix out of range shift introduced in D67070/rL371328.
[X86][SSE] Fix out of range shift introduced in D67070/rL371328
Sun, Sep 8, 5:43 AM
RKSimon committed rG9d570020708d: [X86] Add test case for PR32546 (authored by RKSimon).
[X86] Add test case for PR32546
Sun, Sep 8, 4:58 AM
RKSimon committed rL371329: [X86] Add test case for PR32546.
[X86] Add test case for PR32546
Sun, Sep 8, 4:57 AM
RKSimon committed rG3262084384c7: [X86][SSE] Add support for <64 x i1> bool reduction (authored by RKSimon).
[X86][SSE] Add support for <64 x i1> bool reduction
Sun, Sep 8, 4:49 AM
RKSimon committed rL371328: [X86][SSE] Add support for <64 x i1> bool reduction.
[X86][SSE] Add support for <64 x i1> bool reduction
Sun, Sep 8, 4:48 AM
RKSimon closed D67070: [X86][SSE] Add support for <64 x i1> bool reduction.
Sun, Sep 8, 4:48 AM · Restricted Project

Sat, Sep 7

RKSimon added inline comments to D43339: [SLP] Vectorize reversed stores..
Sat, Sep 7, 11:11 AM · Restricted Project
RKSimon committed rG879ed20bde22: Fix typo. NFCI (authored by RKSimon).
Fix typo. NFCI
Sat, Sep 7, 11:09 AM
RKSimon committed rL371317: Fix typo. NFCI.
Fix typo. NFCI
Sat, Sep 7, 11:08 AM
RKSimon committed rG08692e5dd1f0: [X86] Avoid uses of getZextValue(). NFCI. (authored by RKSimon).
[X86] Avoid uses of getZextValue(). NFCI.
Sat, Sep 7, 9:14 AM
RKSimon committed rG31c98abda3f9: [X86][AVX] Add 'f5' v4f64 shuffle test mentioned in D66004 (authored by RKSimon).
[X86][AVX] Add 'f5' v4f64 shuffle test mentioned in D66004
Sat, Sep 7, 9:13 AM
RKSimon committed rL371315: [X86] Avoid uses of getZextValue(). NFCI..
[X86] Avoid uses of getZextValue(). NFCI.
Sat, Sep 7, 9:12 AM
RKSimon committed rL371314: [X86][AVX] Add 'f5' v4f64 shuffle test mentioned in D66004.
[X86][AVX] Add 'f5' v4f64 shuffle test mentioned in D66004
Sat, Sep 7, 9:12 AM
RKSimon accepted D66309: Introduce infrastructure for an incremental port of SelectionDAG atomic load/store handling.

LGTM, sorry for sidetracking the patch

Sat, Sep 7, 7:44 AM · Restricted Project
RKSimon added a comment to D67266: [X86][NFC] Add a `use-aa` feature..

Makes sense to me - any one else have any comments?

Sat, Sep 7, 5:36 AM · Restricted Project
RKSimon added a comment to D62341: [DAGCombine][X86][AArch64][AMDGPU][MIPS][PPC] (sub x, c) -> (add x, -c) vector edition..

Yes the vXi8 shifts for instance

Sat, Sep 7, 5:31 AM · Restricted Project
RKSimon added a comment to D67259: [X86] Enable -mprefer-vector-width=256 by default for Skylake-avx512 and later Intel CPUs..

Do we need to include some benchmark numbers?

Sat, Sep 7, 5:31 AM · Restricted Project
RKSimon accepted D67314: [X86] Fix pshuflw formation from repeated shuffle mask (PR43230).

Nice catch! LGTM, as @lebedev.ri suggested please pre-commit the test so when the fix is committed it shows the codegen diff.

Sat, Sep 7, 5:01 AM · Restricted Project
RKSimon accepted D67311: [X86] Add prefer-128-bit subtarget feature..

LGTM - this could be useful!

Sat, Sep 7, 4:46 AM · Restricted Project
RKSimon committed rGd7d8bb937ad0: Fix MSVC "32-bit shift implicitly converted to 64 bits" warnings. NFCI. (authored by RKSimon).
Fix MSVC "32-bit shift implicitly converted to 64 bits" warnings. NFCI.
Sat, Sep 7, 4:04 AM
RKSimon committed rL371302: Fix MSVC "32-bit shift implicitly converted to 64 bits" warnings. NFCI..
Fix MSVC "32-bit shift implicitly converted to 64 bits" warnings. NFCI.
Sat, Sep 7, 4:03 AM
RKSimon added a reviewer for D67312: [SelectionDAG] Remove ISD::FP_ROUND_INREG: efriedma.

Code coverage agrees that this is dead:
http://lab.llvm.org:8080/coverage/coverage-reports/llvm/coverage/Users/buildslave/jenkins/workspace/clang-stage2-coverage-R/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp.html

Sat, Sep 7, 2:06 AM · Restricted Project

Fri, Sep 6

RKSimon requested changes to D60748: Fix i386 struct and union parameter alignment.

Thanks for the information!
We have reverted the patch and will resubmit it when we have a complete fix.

Fri, Sep 6, 8:25 AM · Restricted Project
RKSimon added a comment to D37365: [x86] Enable f128 as a legal type in 64-bit mode if SSE is enabled rather than if MMX is enabled..

@craig.topper I noticed in D67128 that it looks like this was dealt with at rL370682?

Fri, Sep 6, 8:21 AM
RKSimon requested changes to D60318: [ExpandMemCmp][MergeICmps] Move passes out of CodeGen into opt pipeline..

reopening until the regressions have been investigated

Fri, Sep 6, 7:45 AM · Restricted Project
RKSimon added a comment to D60318: [ExpandMemCmp][MergeICmps] Move passes out of CodeGen into opt pipeline..

@courbet What's happening with this patch?

Fri, Sep 6, 7:08 AM · Restricted Project
RKSimon added a comment to D31724: [SelectionDAG] Remove special call to LHS computeKnownBits for ANDs with constant RHS..

@craig.topper Is this still relevant? At least some of these changes have been fixed by improvements to SimplifyDemandedBits

Fri, Sep 6, 7:08 AM
RKSimon added a comment to D64672: [X86] Prevent passing vectors of __int128 as <X x i128> in llvm IR.

@craig.topper Wasn't this committed at rL371169 ?

Fri, Sep 6, 7:08 AM · Restricted Project
RKSimon added a comment to D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic.

rebase after D67071 has landed? (and add equivalent scale == 0 tests)

Fri, Sep 6, 6:52 AM · Restricted Project
RKSimon added a comment to D62341: [DAGCombine][X86][AArch64][AMDGPU][MIPS][PPC] (sub x, c) -> (add x, -c) vector edition..

For vectors this is a big increase in constant pool usage - are we sure we want to do this?

Fri, Sep 6, 6:43 AM · Restricted Project
RKSimon added inline comments to D67259: [X86] Enable -mprefer-vector-width=256 by default for Skylake-avx512 and later Intel CPUs..
Fri, Sep 6, 6:17 AM · Restricted Project
RKSimon added inline comments to D66309: Introduce infrastructure for an incremental port of SelectionDAG atomic load/store handling.
Fri, Sep 6, 3:54 AM · Restricted Project
RKSimon accepted D67071: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering::expandFixedPointMul.

Makes sense - LGTM

Fri, Sep 6, 3:51 AM · Restricted Project
RKSimon added inline comments to D67070: [X86][SSE] Add support for <64 x i1> bool reduction.
Fri, Sep 6, 1:49 AM · Restricted Project
RKSimon added inline comments to D67212: [x86] Adding support for some missing intrinsics: _castf32_u32, _castf64_u64, _castu32_f32, _castu64_f64.
Fri, Sep 6, 1:46 AM · Restricted Project

Thu, Sep 5

RKSimon added a comment to D67071: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering::expandFixedPointMul.

Worth adding a vector test ?

Thu, Sep 5, 2:36 PM · Restricted Project
RKSimon added a comment to D67070: [X86][SSE] Add support for <64 x i1> bool reduction.

ping?

Thu, Sep 5, 2:34 PM · Restricted Project
RKSimon added inline comments to D57059: [SLP] Initial support for the vectorization of the non-power-of-2 vectors..
Thu, Sep 5, 2:34 PM · Restricted Project
RKSimon added a comment to D67212: [x86] Adding support for some missing intrinsics: _castf32_u32, _castf64_u64, _castu32_f32, _castu64_f64.

Also, as its x86 specific miscellaneous-builtins.c should be called x86-builtins.c (or similar).

Thu, Sep 5, 2:31 PM · Restricted Project
RKSimon added a reviewer for D67175: [llvm] [cmake] Add possibility to use ChooseMSVCCRT.cmake when include LLVM library: rnk.
Thu, Sep 5, 2:23 PM · Restricted Project
RKSimon added a comment to D67192: [X86] Use MOVSX instead of CBW to extend i8 to AX for i8 sdiv..

I'm not clear on why we can't peephole replace MOVSX with CBW at a later stage.

Thu, Sep 5, 10:28 AM · Restricted Project
RKSimon added a comment to D67210: [x86] bug fix for https://reviews.llvm.org/D64551.

rL371078 should fix this

Thu, Sep 5, 8:36 AM · Restricted Project
RKSimon added inline comments to D29641: [SLP] Fix for PR31847: Assertion failed: (isLoopInvariant(Operands[i], L) && "SCEVAddRecExpr operand is not loop-invariant!").
Thu, Sep 5, 8:29 AM · Restricted Project
RKSimon committed rG29361c704dfa: [X86][SSE] EltsFromConsecutiveLoads - ignore non-zero offset base loads… (authored by RKSimon).
[X86][SSE] EltsFromConsecutiveLoads - ignore non-zero offset base loads…
Thu, Sep 5, 8:08 AM
RKSimon committed rL371078: [X86][SSE] EltsFromConsecutiveLoads - ignore non-zero offset base loads….
[X86][SSE] EltsFromConsecutiveLoads - ignore non-zero offset base loads…
Thu, Sep 5, 8:08 AM
RKSimon added a comment to D67210: [x86] bug fix for https://reviews.llvm.org/D64551.

Please abandon this, it isn't a valid solution to the issue (raised at PR43227). I have a WIP fix that will address this correctly.

Thu, Sep 5, 5:43 AM · Restricted Project
RKSimon committed rG215910eeb2e1: [X86][SSE] Add (failing) test case for PR43227 (authored by RKSimon).
[X86][SSE] Add (failing) test case for PR43227
Thu, Sep 5, 5:37 AM
RKSimon committed rL371061: [X86][SSE] Add (failing) test case for PR43227.
[X86][SSE] Add (failing) test case for PR43227
Thu, Sep 5, 5:37 AM
RKSimon added a comment to D64551: [X86] EltsFromConsecutiveLoads - support common source loads.

I've raised https://bugs.llvm.org/show_bug.cgi?id=43227 to handle this

Thu, Sep 5, 5:29 AM · Restricted Project
RKSimon added inline comments to D64551: [X86] EltsFromConsecutiveLoads - support common source loads.
Thu, Sep 5, 4:29 AM · Restricted Project
RKSimon added a comment to D67189: [AArch64][GlobalISel] Teach AArch64CallLowering to handle basic sibling calls.

@paquette I'm sorry but I had to revert rL370996 at rL371051 to fix the EXPENSIVE_CHECKS builds: http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/19511/

Thu, Sep 5, 3:42 AM · Restricted Project
RKSimon committed rG071287c5a932: Revert rL370996 from llvm/trunk: [AArch64][GlobalISel] Teach… (authored by RKSimon).
Revert rL370996 from llvm/trunk: [AArch64][GlobalISel] Teach…
Thu, Sep 5, 3:42 AM
RKSimon committed rL371051: Revert rL370996 from llvm/trunk: [AArch64][GlobalISel] Teach….
Revert rL370996 from llvm/trunk: [AArch64][GlobalISel] Teach…
Thu, Sep 5, 3:41 AM