Today

efriedma added a comment to D48332: [AArch64] Add custom lowering for v4i8 trunc store.

I wonder if we should prefer to widen <2 x i8> and <4 x i8> to <8 x i8> instead of promoting to <4 x i16>. It would make stores like this a bit cheaper. Maybe an interesting experiment at some point (mostly just modifying AArch64TargetLowering::getPreferredVectorAction, I think, and seeing what happens to the generated code).

Tue, Jun 19, 1:10 PM
spatel added a comment to D48085: [DAGCombiner] restrict (float)((int) f) --> ftrunc with no-signed-zeros.

Should this adjust the ReleaseNotes?

The docs provide extra warning/workaround, and this change doesn't affect that language IMO (there's less chance we're going to break code, but we don't have to disclose that).

For reference, here are links to the docs that we added with the previous patches (last updated with D46236 I think):
https://clang.llvm.org/docs/ReleaseNotes.html#new-compiler-flags
https://clang.llvm.org/docs/UsersManual.html#controlling-code-generation
http://llvm.org/docs/ReleaseNotes.html#non-comprehensive-list-of-changes-in-this-release

Let me know if you see anything that can be improved.

The text in first and last links suggests that such an optimization is always being done.
But now it will only be done in presence of no-signed-zeros-fp-math attribute.
Which is controlled by -ffast-math (or maybe some more fine-grained option, too?)

Tue, Jun 19, 1:05 PM
stefanp updated the diff for D48294: [Power9] Add __float128 support for compare operations.

Fixed test case command.

Tue, Jun 19, 1:02 PM
vsk added inline comments to D48305: [IR] Introduce helpers to skip debug instructions (NFC).
Tue, Jun 19, 1:00 PM
teemperor updated the diff for D48303: Don't take the address of an xvalue when printing an expr result.
  • Generalized test case a bit more and no longer checking only for 0.
Tue, Jun 19, 12:48 PM
arphaman committed rC335073: Revert r335063 as it causes bot failures.
Revert r335063 as it causes bot failures
Tue, Jun 19, 12:48 PM
arphaman committed rL335073: Revert r335063 as it causes bot failures.
Revert r335063 as it causes bot failures
Tue, Jun 19, 12:47 PM
ddcc added a comment to D48324: [Analyzer] [WIP] Fix wrong comparison generation of the ranges generated by the refutation manager.

I think the correct type should always be ptrdiff_t, regardless of the type of the pointers involved. ASTContext has a method has a getPointerDiffType() so maybe we should you that instead of Ctx.getIntTypeForBitwidth(Ctx.getTypeSize(LTy), true)?

Tue, Jun 19, 12:47 PM
vsk updated subscribers of D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.
Tue, Jun 19, 12:43 PM
mikhail.ramalho added a comment to D48324: [Analyzer] [WIP] Fix wrong comparison generation of the ranges generated by the refutation manager.

Changing it to:

Shouldn't we then use maximum of two type sizes? [I'm not sure what are the exact conversion rules there]

Tue, Jun 19, 12:42 PM
ctopper committed rC335072: Revert r335070 "[X86] Rewrite the max and min reduction intrinsics to make….
Revert r335070 "[X86] Rewrite the max and min reduction intrinsics to make…
Tue, Jun 19, 12:41 PM
ctopper committed rL335072: Revert r335070 "[X86] Rewrite the max and min reduction intrinsics to make….
Revert r335070 "[X86] Rewrite the max and min reduction intrinsics to make…
Tue, Jun 19, 12:41 PM
zatrazz created D48332: [AArch64] Add custom lowering for v4i8 trunc store.
Tue, Jun 19, 12:38 PM
lebedev.ri added a comment to D48085: [DAGCombiner] restrict (float)((int) f) --> ftrunc with no-signed-zeros.

Should this adjust the ReleaseNotes?

The docs provide extra warning/workaround, and this change doesn't affect that language IMO (there's less chance we're going to break code, but we don't have to disclose that).

For reference, here are links to the docs that we added with the previous patches (last updated with D46236 I think):
https://clang.llvm.org/docs/ReleaseNotes.html#new-compiler-flags
https://clang.llvm.org/docs/UsersManual.html#controlling-code-generation
http://llvm.org/docs/ReleaseNotes.html#non-comprehensive-list-of-changes-in-this-release

Let me know if you see anything that can be improved.

Tue, Jun 19, 12:35 PM
efriedma accepted D48029: [DAGCombine] Fix alignment for offset loads/stores.

LGTM

Tue, Jun 19, 12:31 PM
spatel updated the diff for D48085: [DAGCombiner] restrict (float)((int) f) --> ftrunc with no-signed-zeros.

Patch updated:
Added a 'TODO' comment about using a FABS-based sequence if we can't ignore -0.0.

Tue, Jun 19, 12:30 PM
lebedev.ri added inline comments to D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.
Tue, Jun 19, 12:29 PM
spatel added a comment to D48085: [DAGCombiner] restrict (float)((int) f) --> ftrunc with no-signed-zeros.

Should this adjust the ReleaseNotes?

Tue, Jun 19, 12:28 PM
craig.topper added a comment to D47401: [X86] Rewrite the max and min reduction intrinsics to make better use of other functions and to reduce width to 256 and 128 bits were possible..

Fast-isel tests were added for previous codegen in r335068 and updated for new codegen in r335071.

Tue, Jun 19, 12:24 PM
gramanas updated the diff for D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.

Add RUN: clause to the test

Tue, Jun 19, 12:20 PM
ctopper committed rL335071: [X86] Update fast-isel tests for clang's avx512f reduction intrinsics to match….
[X86] Update fast-isel tests for clang's avx512f reduction intrinsics to match…
Tue, Jun 19, 12:19 PM
ctopper committed rC335070: [X86] Rewrite the max and min reduction intrinsics to make better use of other….
[X86] Rewrite the max and min reduction intrinsics to make better use of other…
Tue, Jun 19, 12:18 PM
ctopper committed rL335070: [X86] Rewrite the max and min reduction intrinsics to make better use of other….
[X86] Rewrite the max and min reduction intrinsics to make better use of other…
Tue, Jun 19, 12:18 PM
Diffusion closed D47401: [X86] Rewrite the max and min reduction intrinsics to make better use of other functions and to reduce width to 256 and 128 bits were possible..
Tue, Jun 19, 12:18 PM
efriedma accepted D48170: ARM: use "add" instead of "orr" for code size.

LGTM

Tue, Jun 19, 12:16 PM
guiguiiiiiiii added inline comments to D48259: [clang-format] Fix bug with UT_Always when there is less than one full tab.
Tue, Jun 19, 12:14 PM · Restricted Project
Hahnfeld committed rOMP335069: Remove liboffload from repository.
Remove liboffload from repository
Tue, Jun 19, 12:13 PM
george.karpenkov added a comment to D48324: [Analyzer] [WIP] Fix wrong comparison generation of the ranges generated by the refutation manager.

Changing it to:

Tue, Jun 19, 12:13 PM
lei added inline comments to D48184: [Power9] [CLANG] Add __float128 exponent GET and SET builtins.
Tue, Jun 19, 12:13 PM
Hahnfeld committed rL335069: Remove liboffload from repository.
Remove liboffload from repository
Tue, Jun 19, 12:13 PM
lebedev.ri added inline comments to D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.
Tue, Jun 19, 12:08 PM
efriedma added inline comments to D48230: [PredicateInfo] Order instructions in different BBs by DFSNumIn..
Tue, Jun 19, 12:08 PM
gramanas added inline comments to D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.
Tue, Jun 19, 12:08 PM
mikhail.ramalho added a comment to D48324: [Analyzer] [WIP] Fix wrong comparison generation of the ranges generated by the refutation manager.

The code already tries to handle ptrdiff_t specially, see the comment right above, but it is only executed if RetTy is non-null. Instead, I think it should be refactored so that e.g. a local isBool is always computed, and this is passed to the final call to Z3Expr::fromBinOp.

Tue, Jun 19, 12:08 PM
gramanas updated the summary of D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.
Tue, Jun 19, 12:07 PM
gramanas created D48331: [DebugInfo][InstCombine] Preserve DI after merging instructions.
Tue, Jun 19, 12:06 PM
baloghadamsoftware added a comment to D35110: [Analyzer] Constraint Manager Negates Difference.

I tested all parts of the Iterator Checkers, all tests passed.

Tue, Jun 19, 12:03 PM
efriedma added a comment to D48330: [GVN] Avoid casting a vector of size less than 8 bits to i8.

It should be possible to implement this sort of coercion with an appropriate cast, rather than just bailing out, I think. But I won't block the patch on that.

Tue, Jun 19, 12:01 PM
craig.topper added a comment to D48225: [llvm-mca][X86] Teach how to identify register writes that implicitly clear the upper portion of a super-register..

Ok the register class/feature stuff looks good to me now.

Tue, Jun 19, 12:00 PM
ctopper committed rL335068: [X86] Add fast-isel tests for clang's AVX512F vector reduction intrinsics..
[X86] Add fast-isel tests for clang's AVX512F vector reduction intrinsics.
Tue, Jun 19, 11:56 AM