Yesterday

ctopper committed rL314025: [InstCombine] Add constant splat handling to one of the ICMP_SLT/SGT cases in….
[InstCombine] Add constant splat handling to one of the ICMP_SLT/SGT cases in…
Fri, Sep 22, 12:55 PM
spatel committed rL314024: [x86] shiftRightAlgebraic -> shiftRightArithmetic; NFC.
[x86] shiftRightAlgebraic -> shiftRightArithmetic; NFC
Fri, Sep 22, 12:51 PM
kcc added a comment to D38056: Partial fix for bug 34607: ASan misses global underflow in first symbol of data section.

Please address the comment around (uptr) and update the commit message.

Fri, Sep 22, 12:49 PM
alexshap closed D38151: [clang] Fix isExternC matcher docs by committing rL314022: [clang] Fix isExternC matcher docs.
Fri, Sep 22, 12:46 PM
sylvestre.ledru created D38186: Add the new -Wnull-pointer-arithmetic warnings to the release notes.
Fri, Sep 22, 12:41 PM · Restricted Project
protze.joachim updated subscribers of D38185: Implementation of OMPT as specified in OpenMP 5.0 Preview 1.
Fri, Sep 22, 12:37 PM
protze.joachim created D38185: Implementation of OMPT as specified in OpenMP 5.0 Preview 1.
Fri, Sep 22, 12:36 PM
Nebiroth updated the diff for D36150: [clangd] LSP extension to switch between source/header file.

Rebased on latest version.
Corrected code style issues in test file.

Fri, Sep 22, 12:35 PM
reames added a comment to D36258: Disable loop peeling during full unrolling pass..

That's clearly interesting. This change, however, seems like it should be more-or-less a nop. If you're going to fully unroll, why would peeling help? Do you end up with a shorter critical path through the branches if we "peel" first?

In our pass pipeline, we'd end up with loop-unrolling being quite a bit earlier than upstream. This wasn't so much an intentional choice as an accident of history that we really hadn't evaluated up until now. My best guess is that the peeling was accidentally exposing information to other transformations and that the actual unrolling was incidental to the performance swing. Someone on my team is investigating that a bit further to see what actually happened; what I just said was just an educated guess.

Fri, Sep 22, 12:32 PM
alexshap committed rL314022: [clang] Fix isExternC matcher docs.
[clang] Fix isExternC matcher docs
Fri, Sep 22, 12:31 PM
wdng edited reviewers for D37348: Implement custom lowering for ISD::CTTZ_ZERO_UNDEF and ISD::CTTZ., added: kzhuravl; removed: craig.topper.
Fri, Sep 22, 12:29 PM
fjricci committed rL314021: Fix fuchsia builds broken by r313999.
Fix fuchsia builds broken by r313999
Fri, Sep 22, 12:24 PM
ibiryukov committed rL314020: [clangd] Updated gold for completion tests after cfe changes..
[clangd] Updated gold for completion tests after cfe changes.
Fri, Sep 22, 12:09 PM
ibiryukov committed rL314019: Set completion priority of destructors and operators to CCP_Unlikely..
Set completion priority of destructors and operators to CCP_Unlikely.
Fri, Sep 22, 12:09 PM
Diffusion closed D38081: Set completion priority of destructors and operators to CCP_Unlikely. by committing rL314019: Set completion priority of destructors and operators to CCP_Unlikely..
Fri, Sep 22, 12:09 PM
ruiu added a comment to D38167: [ELF] Ensure that .init/.fini sections are padded with nops instead of traps.

Well, this is not a bug, so this patch is not a fix but a workaround for existing FreeBSD/MIPS kernel's issue.

Fri, Sep 22, 12:06 PM · lld
yaxunl added inline comments to D37568: [AMDGPU] Allow flexible register names in inline asm constraints.
Fri, Sep 22, 12:06 PM
haicheng updated the diff for D33946: [InlineCost] Find identical loads in the callee.

Fix a typo in the test case. Please take a look, thank you.

Fri, Sep 22, 12:05 PM
yaxunl updated the diff for D37568: [AMDGPU] Allow flexible register names in inline asm constraints.

Fix typo.

Fri, Sep 22, 12:04 PM
mstojanovic retitled D35946: Fix printing of alias instructions by removing redundant spacing from Fix redundant spacing of mips alias instructions in llvm-objdump disassemble output to Fix printing of alias instructions by removing redundant spacing.
Fri, Sep 22, 12:00 PM
ctopper committed rL314018: [InstCombine] Move the call to isSignBitCheck into getDemandedBitsLHSMask….
[InstCombine] Move the call to isSignBitCheck into getDemandedBitsLHSMask…
Fri, Sep 22, 11:59 AM
ctopper committed rL314017: [InstCombine] Simplify check for RHS being a splat constant in….
[InstCombine] Simplify check for RHS being a splat constant in…
Fri, Sep 22, 11:59 AM
ctopper committed rL314016: [InstCombine] Make cases for ICMP_UGT/ICMP_ULT use similar formatting since….
[InstCombine] Make cases for ICMP_UGT/ICMP_ULT use similar formatting since…
Fri, Sep 22, 11:59 AM
MatzeB added a comment to D38164: [MachineScheduler] Favor instructions that do not increase pressure..
  • I vaguely remember trying something like this and having some crypto benchmarks produce bad schedules; I'll see if I can remember/find it
  • tryPressure seems like the wrong place to me, as it is used in 3 different contexts: (compared with target limits, compared with increase region limits, and the current max). From your description it sounds like we only want this behavior once.
  • Do you have a specific example where this helps?
Fri, Sep 22, 11:57 AM
vitalybuka committed rL314015: [sanitizer] Replace AddressSanitizer with correct tool name.
[sanitizer] Replace AddressSanitizer with correct tool name
Fri, Sep 22, 11:51 AM
kuba.brecka committed rL314014: Deflake the "xpc-race.mm" test..
Deflake the "xpc-race.mm" test.
Fri, Sep 22, 11:51 AM
bruno added a comment to D37299: [Modules] Add ability to specify module name to module file mapping in a file.

Hi Boris,

Fri, Sep 22, 11:50 AM
arichardson added a comment to D38167: [ELF] Ensure that .init/.fini sections are padded with nops instead of traps.
In D38167#879065, @ruiu wrote:

At least for x86, null bytes are not nop instructions, so it feels like it just happened to work for MIPS but it wasn't really guaranteed. How did you get a gap in .fini section?

Fri, Sep 22, 11:49 AM · lld
ilya-biryukov added a comment to D38081: Set completion priority of destructors and operators to CCP_Unlikely..

Thanks for the review!
Sorry for inappropriate ping, I'll make sure to stick to the policy.

Fri, Sep 22, 11:46 AM
ilya-biryukov updated the diff for D38081: Set completion priority of destructors and operators to CCP_Unlikely..
  • Removed redundant NameKind:: qualifiers.
Fri, Sep 22, 11:46 AM
vedantk committed rL314013: [lit.cfg] Avoid concatenating which(clang-func-mapping) if it's missing.
[lit.cfg] Avoid concatenating which(clang-func-mapping) if it's missing
Fri, Sep 22, 11:44 AM
alexshap committed rL314011: [clang] Fix printf fixit for objc specific types.
[clang] Fix printf fixit for objc specific types
Fri, Sep 22, 11:43 AM
alexshap closed D38159: [clang] Fix printf fixit for objc specific types by committing rL314011: [clang] Fix printf fixit for objc specific types.
Fri, Sep 22, 11:43 AM
ruiu added a comment to D38167: [ELF] Ensure that .init/.fini sections are padded with nops instead of traps.

At least for x86, null bytes are not nop instructions, so it feels like it just happened to work for MIPS but it wasn't really guaranteed. How did you get a gap in .fini section?

Fri, Sep 22, 11:42 AM · lld
rafael committed rL314012: Move code to a helper function. NFC..
Move code to a helper function. NFC.
Fri, Sep 22, 11:41 AM
nemanjai accepted D38165: [CodeGenPrepare][NFC] Rename TargetTransformInfo::expandMemCmp -> TargetTransformInfo::enableMemCmpExpansion..

Yeah, the new name makes a lot more sense. LGTM.

Fri, Sep 22, 11:41 AM
rafael committed rL314010: llvm-ar: align the first archive member consistently..
llvm-ar: align the first archive member consistently.
Fri, Sep 22, 11:37 AM
compnerd closed D37573: [bindings] add Cursor.linkage.

SVN r314009

Fri, Sep 22, 11:36 AM
compnerd committed rL314009: bindings: expose Linkage to the python bindings.
bindings: expose Linkage to the python bindings
Fri, Sep 22, 11:36 AM
ruiu added inline comments to D37520: [ELF] - Fix removing of unused synthetic sections..
Fri, Sep 22, 11:35 AM