Page MenuHomePhabricator
Feed Advanced Search

Mon, Nov 11

bjope created D70091: WIP: BlockFolding not debug invariant (PR43964).
Mon, Nov 11, 9:36 AM · Restricted Project
bjope added a comment to D66467: [Codegen] skip debug instr to avoid code change.

name: test3
body: |
bb.0:

TEST8rr killed renamable $al, renamable $al, implicit-def $eflags
JCC_1 %bb.1, 5, implicit killed $eflags

from the test case, the JCC should be bb.2, as below, right?

JCC_1 %bb.2, 5, implicit killed $eflags

Mon, Nov 11, 9:17 AM · Restricted Project

Sat, Nov 9

bjope added inline comments to D64630: [DebugInfo] Address performance regression with r364515.
Sat, Nov 9, 6:08 AM · Restricted Project

Thu, Nov 7

bjope added a comment to D66467: [Codegen] skip debug instr to avoid code change.

Here is an test case that currently fail. The test1, test2 and test3 functions only differ by DBG_VALUE and CFI_INSTRUCTION instructions. So I guess we want to see the same real instructions being generated for all three functions.

Thu, Nov 7, 7:32 AM · Restricted Project
bjope added a comment to D66467: [Codegen] skip debug instr to avoid code change.

Unfortunately I see a lot more test cases failing due to codegen not being debug invariant after having merged this patch, compared to the past.

Thu, Nov 7, 7:23 AM · Restricted Project

Sun, Nov 3

bjope committed rGe2549a0a2517: [DebugInfo] Fix for DW_OP_LLVM_fragment in DIExpression::isImplicit() (authored by bjope).
[DebugInfo] Fix for DW_OP_LLVM_fragment in DIExpression::isImplicit()
Sun, Nov 3, 9:04 AM

Fri, Nov 1

bjope committed rG56c22931bdfa: [LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller (authored by bjope).
[LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller
Fri, Nov 1, 8:45 AM
bjope closed D69584: [LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller.
Fri, Nov 1, 8:44 AM · Restricted Project

Thu, Oct 31

bjope added a comment to D69656: [IPCP] Bail on extractvalue's with more than 1 index..

Besides the point: I didn't know anyone was still using this pass. Actually I was going to propose to delete it soon. Is there a desire to keep it or would people switch to IP-SCCP and/or the Attributor (assuming they subsume the functionality)? The one reason to keep this pass is that it is simple and "fast" but it's questionable if that is useful if we run either of the other ones anyway.

The patch LGTM.

Thu, Oct 31, 11:40 AM · Restricted Project

Wed, Oct 30

bjope added a comment to D69584: [LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller.

LGTM; I wonder whether there are other places in the register allocator that need to notify LDV of what's happening, but I don't know enough about allocators.

Wed, Oct 30, 12:10 PM · Restricted Project
bjope added a comment to D69028: [DebugInfo] Correctly place DW_OP_derefs for arguments passed on stack.

Ah blast, I mentally filed this under "things to push upwards when there's spare time" as opposed to "a regression that needs fixing". Sorry for the delay, I'll give it a kick now.

Wed, Oct 30, 11:31 AM · Restricted Project
bjope updated the diff for D69584: [LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller.

Addressed review comments.

Wed, Oct 30, 10:51 AM · Restricted Project
bjope added inline comments to D64630: [DebugInfo] Address performance regression with r364515.
Wed, Oct 30, 10:51 AM · Restricted Project
bjope added a comment to D69028: [DebugInfo] Correctly place DW_OP_derefs for arguments passed on stack.

(Reverse ping!) Any reason why this hasn't landed yet?

Wed, Oct 30, 10:26 AM · Restricted Project
bjope added inline comments to D69584: [LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller.
Wed, Oct 30, 8:52 AM · Restricted Project

Tue, Oct 29

bjope created D69584: [LDV][RAGreedy] Inform LiveDebugVariables about new VRegs added by InlineSpiller.
Tue, Oct 29, 1:32 PM · Restricted Project
bjope updated subscribers of D68816: [NFC] Replace a linked list in LiveDebugVariables pass with a DenseMap.
Tue, Oct 29, 3:19 AM · Restricted Project, debug-info

Mon, Oct 28

bjope committed rG80cb2cecc657: [utils] InlineFunction: fix for debug info affecting optimizations (authored by bjope).
[utils] InlineFunction: fix for debug info affecting optimizations
Mon, Oct 28, 10:19 AM
bjope closed D68633: [utils] InlineFunction: fix for debug info affecting optimizations.
Mon, Oct 28, 10:19 AM · Restricted Project
bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

@bjope thanks for help, to be honest, I don't know the process or execution of commit, I thought I have no commit permission right now.

Mon, Oct 28, 8:33 AM · Restricted Project
bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

Not sure why this isn't in "Accepted" state. I know @aprantl and @vsk asked for a new revision on an earlier diff. So is it perhaps because not all comments have been marked as "Done"?

Mon, Oct 28, 5:59 AM · Restricted Project
bjope added a comment to D66467: [Codegen] skip debug instr to avoid code change.

Looks like this one was accepted a long time ago. But I can't see that it has been committed yet. So what is the status?

Mon, Oct 28, 5:54 AM · Restricted Project

Wed, Oct 23

bjope added inline comments to D68267: [MBB LiveIn lists, MachineVerifier, SystemZ] New method isLiveOut() and mverifier improvement..
Wed, Oct 23, 4:10 PM · Restricted Project

Tue, Oct 22

bjope added inline comments to D68267: [MBB LiveIn lists, MachineVerifier, SystemZ] New method isLiveOut() and mverifier improvement..
Tue, Oct 22, 12:58 PM · Restricted Project
bjope added inline comments to D68267: [MBB LiveIn lists, MachineVerifier, SystemZ] New method isLiveOut() and mverifier improvement..
Tue, Oct 22, 12:49 PM · Restricted Project

Mon, Oct 21

bjope committed rG78424e5f8417: Prune include of DataLayout.h from include/clang/Basic/TargetInfo.h. NFC (authored by bjope).
Prune include of DataLayout.h from include/clang/Basic/TargetInfo.h. NFC
Mon, Oct 21, 11:05 AM
bjope closed D69262: Prune include of DataLayout.h from include/clang/Basic/TargetInfo.h. NFC.
Mon, Oct 21, 11:05 AM · Restricted Project
bjope committed rL375438: Prune include of DataLayout.h from include/clang/Basic/TargetInfo.h. NFC.
Prune include of DataLayout.h from include/clang/Basic/TargetInfo.h. NFC
Mon, Oct 21, 10:56 AM
bjope committed rG1f43ea41c330: Prune Pass.h include from DataLayout.h. NFCI (authored by bjope).
Prune Pass.h include from DataLayout.h. NFCI
Mon, Oct 21, 10:56 AM
bjope committed rL375436: Prune Pass.h include from DataLayout.h. NFCI.
Prune Pass.h include from DataLayout.h. NFCI
Mon, Oct 21, 10:56 AM
bjope closed D69261: Prune Pass.h include from DataLayout.h. NFCI.
Mon, Oct 21, 10:55 AM · Restricted Project, Restricted Project
bjope created D69262: Prune include of DataLayout.h from include/clang/Basic/TargetInfo.h. NFC.
Mon, Oct 21, 7:58 AM · Restricted Project
bjope created D69261: Prune Pass.h include from DataLayout.h. NFCI.
Mon, Oct 21, 7:53 AM · Restricted Project, Restricted Project
bjope accepted D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

Thanks for the patience with all my remarks, turning this around completely compared to the original patch.

Mon, Oct 21, 5:57 AM · Restricted Project

Fri, Oct 18

bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

Thanks! I like this solution.

Fri, Oct 18, 5:08 AM · Restricted Project
bjope committed rG6456252dbf67: [InstCombine] Fix miscompile bug in canEvaluateShuffled (authored by bjope).
[InstCombine] Fix miscompile bug in canEvaluateShuffled
Fri, Oct 18, 12:46 AM
bjope committed rG459134064dae: [InstCombine] Pre-commit of test case showing miscompile bug in… (authored by bjope).
[InstCombine] Pre-commit of test case showing miscompile bug in…
Fri, Oct 18, 12:46 AM
bjope committed rL375208: [InstCombine] Fix miscompile bug in canEvaluateShuffled.
[InstCombine] Fix miscompile bug in canEvaluateShuffled
Fri, Oct 18, 12:46 AM
bjope closed D69038: [InstCombine] Fix miscompile bug in canEvaluateShuffled.
Fri, Oct 18, 12:45 AM · Restricted Project
bjope committed rL375207: [InstCombine] Pre-commit of test case showing miscompile bug in….
[InstCombine] Pre-commit of test case showing miscompile bug in…
Fri, Oct 18, 12:45 AM

Wed, Oct 16

bjope updated the diff for D69038: [InstCombine] Fix miscompile bug in canEvaluateShuffled.
  • Fixup mistake regarding FDiv (it was not supposed to be included in the new restriction).
  • Added test cases for sdiv/urem/udiv (showing that the new restriction works).
  • Added test cases for fdiv/frem (showing that we can eliminate the shufflevector).
Wed, Oct 16, 10:37 AM · Restricted Project
bjope added inline comments to D69038: [InstCombine] Fix miscompile bug in canEvaluateShuffled.
Wed, Oct 16, 8:55 AM · Restricted Project
bjope added inline comments to D69038: [InstCombine] Fix miscompile bug in canEvaluateShuffled.
Wed, Oct 16, 7:13 AM · Restricted Project
bjope created D69038: [InstCombine] Fix miscompile bug in canEvaluateShuffled.
Wed, Oct 16, 7:11 AM · Restricted Project

Tue, Oct 15

bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.
In D68633#1709729, @vsk wrote:

Hey @yechunliang, thanks for working on this.

This part of the inliner looks like it has two issues: 1) it's not fully aware of debug instructions, and 2) it erroneously extends the lifetimes of inlined variables. The current iteration of your patch might address the first issue, but it's not DRY, and it doesn't solve the second.

Tue, Oct 15, 1:00 PM · Restricted Project

Oct 11 2019

bjope added a reviewer for D68633: [utils] InlineFunction: fix for debug info affecting optimizations: fhahn.

So the root cause is rather that we treat an alloca being immediately preceeded by another alloca differrently from the case when it is preceeded by another kind of instruction. This happens also when having other instructions in between, and is not specific to dbg intrinsics (could be interesting to add a test case where you replace the dbg intrinsics by something else).

Yes I think so, if the other instruction is not dbg instr which exist between two allocas, the InlineFunction with and without "-strip-debug" will make the same behavior, that should both erase second use_empty alloca. This patch is to fix the issue that debug instr impact InlineFunction generate different output.

So I think that the solution might be based on one of these ideas:

  1. Remove the check for use_empty in the outer loop.
  2. Add a check for !use_empty in the inner loop.
  3. Remove the inner loop (i.e only splice one alloca at a time).

These good ideas should be talking about the design change of alloca inline or improvement of splice.
Read from the code, I think about the alloca inline behavior like this: First detect one !use_empty alloca, if next immediate instructions are allocas, even they are use_empty, they will all added one after one and move to caller together with first alloca. if other instruction (whatever dbg or others instrs) exist between allocas, the next alloca will check if is use_empty or not, if is use_empty then erase. Does this behavior correct, or could be improve? I don't know much about alloca inline. but seems the code run many years with this design.

Oct 11 2019, 2:41 PM · Restricted Project
bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

The code is written in a way that it skips any instruction, but moves contigous blocks of allocas in one splice (not sure exactly why, is that really faster?).

I also could not understand why continue to scan allocas block after first none use_empty alloca instruction, here is the first commit has some reason: https://github.com/llvm/llvm-project/commit/6f8865bf9

Maybe the difference is that the check for AI->useEmpty() only is done for the first alloca in a sequence of alloca instructions? Or can't we just remove the loop at line 1847 (only moving one alloca at a time).

with this example test case, second alloca is use_empty, and will insert to caller together with first alloca (!use_empty). But if there is dbg instruction between first alloca and second alloca instruction. the continue scan will break,
then with the debug instruction, the program will goto the front for() loop, and handle the second alloca as use_empty (because it has no use list like "xxx.sroa_cast = bitcast %rec1198* %volatileloadslot to i8*") and eraseFromParent.
this is difference as no-dbg inline will not erase second alloca instruction.

Oct 11 2019, 12:59 AM · Restricted Project

Oct 10 2019

bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

I'm still confused. IMO moving or not moving dbg instructions is a different question (not being discussed here). Still no explanation why result differs depending on if we move dbg instructuons or not. And this patch now blindly moves some dbg instructions, regardless if they are related to the allocas or not. Consider an instruction sequence like this:

alloca
bitcast
dbg.declare
bitcast
alloca
bitcast
dbg.value

The scanning for allocas will find the two alloca instructions and call splice twice, only moving the alloca instructions, right?

Oct 10 2019, 9:27 AM · Restricted Project

Oct 9 2019

bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

I do not understand how this helps. The code is written in a way that it skips any instruction, but moves contigous blocks of allocas in one splice (not sure exactly why, is that really faster?). Maybe the difference is that the check for AI->useEmpty() only is done for the first alloca in a sequence of alloca instructions? Or can't we just remove the loop at line 1847 (only moving one alloca at a time).

Oct 9 2019, 12:20 AM · Restricted Project

Oct 8 2019

bjope added a comment to D68633: [utils] InlineFunction: fix for debug info affecting optimizations.

I do not understand how this helps. The code is written in a way that it skips any instruction, but moves contigous blocks of allocas in one splice (not sure exactly why, is that really faster?). Maybe the difference is that the check for AI->useEmpty() only is done for the first alloca in a sequence of alloca instructions? Or can't we just remove the loop at line 1847 (only moving one alloca at a time).

Oct 8 2019, 5:46 AM · Restricted Project
bjope added inline comments to D68006: DSE miscompile when store is clobbered across loop iterations.
Oct 8 2019, 3:55 AM

Sep 26 2019

bjope committed rG163c54d288b2: [InstCombine] Don't assume CmpInst has been visited in… (authored by bjope).
[InstCombine] Don't assume CmpInst has been visited in…
Sep 26 2019, 5:16 AM
bjope committed rL372972: [InstCombine] Don't assume CmpInst has been visited in….
[InstCombine] Don't assume CmpInst has been visited in…
Sep 26 2019, 5:16 AM
bjope closed D68022: [InstCombine] Don't assume CmpInst has been visited in getFlippedStrictnessPredicateAndConstant.
Sep 26 2019, 5:16 AM · Restricted Project
bjope updated the diff for D68022: [InstCombine] Don't assume CmpInst has been visited in getFlippedStrictnessPredicateAndConstant.

I don't think we need to talk about the "edge cases" at all any longer (after reading the comment from Roman once more, and analysing the code a little bit).

Sep 26 2019, 2:05 AM · Restricted Project

Sep 25 2019

bjope added inline comments to D68022: [InstCombine] Don't assume CmpInst has been visited in getFlippedStrictnessPredicateAndConstant.
Sep 25 2019, 8:40 AM · Restricted Project
bjope created D68022: [InstCombine] Don't assume CmpInst has been visited in getFlippedStrictnessPredicateAndConstant.
Sep 25 2019, 7:15 AM · Restricted Project

Sep 20 2019

bjope committed rG169cb63478aa: [AMDGPU] Use std::make_tuple to make some toolchains happy again (authored by bjope).
[AMDGPU] Use std::make_tuple to make some toolchains happy again
Sep 20 2019, 5:14 AM
bjope committed rL372384: [AMDGPU] Use std::make_tuple to make some toolchains happy again.
[AMDGPU] Use std::make_tuple to make some toolchains happy again
Sep 20 2019, 5:14 AM

Sep 18 2019

bjope added a comment to D67234: [MergedLoadStoreMotion] Sink stores if they have common GEP.

Sorry, I haven't had time to look at this.

Sep 18 2019, 11:16 AM · Restricted Project

Sep 16 2019

bjope abandoned D52092: [ValueTracking] Generalize isBytewiseValue into isSplatValue.

Not pursuing this. Should have been abandoned a long time ago.

Sep 16 2019, 5:21 AM · Restricted Project

Sep 13 2019

bjope committed rL371829: Request commit access for bjope.
Request commit access for bjope
Sep 13 2019, 2:18 AM

Sep 11 2019

bjope added inline comments to D67417: [LICM/AST] Check if the AliasAny set is removed from the tracker..
Sep 11 2019, 5:24 AM · Restricted Project

Sep 7 2019

bjope committed rGd065c811649f: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering… (authored by bjope).
[CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering…
Sep 7 2019, 5:16 AM
bjope committed rG5e331e4ce85a: [Intrinsic] Add the llvm.umul.fix.sat intrinsic (authored by bjope).
[Intrinsic] Add the llvm.umul.fix.sat intrinsic
Sep 7 2019, 5:16 AM
bjope committed rL371308: [Intrinsic] Add the llvm.umul.fix.sat intrinsic.
[Intrinsic] Add the llvm.umul.fix.sat intrinsic
Sep 7 2019, 5:15 AM
bjope committed rL371309: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering….
[CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering…
Sep 7 2019, 5:15 AM
bjope closed D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic.
Sep 7 2019, 5:15 AM · Restricted Project
bjope closed D67071: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering::expandFixedPointMul.
Sep 7 2019, 5:15 AM · Restricted Project
bjope committed rG2b698a13a11e: [DwarfExpression] Disallow some rewrites to avoid undefined behavior (authored by bjope).
[DwarfExpression] Disallow some rewrites to avoid undefined behavior
Sep 7 2019, 4:40 AM
bjope committed rGe85acf946d3b: [DebugInfo] Pre-commit of test case for DW_OP_breg/DW_OP_fbreg folds (authored by bjope).
[DebugInfo] Pre-commit of test case for DW_OP_breg/DW_OP_fbreg folds
Sep 7 2019, 4:40 AM
bjope committed rL371304: [DwarfExpression] Disallow some rewrites to avoid undefined behavior.
[DwarfExpression] Disallow some rewrites to avoid undefined behavior
Sep 7 2019, 4:39 AM
bjope closed D67263: [DwarfExpression] Disallow some rewrites to avoid undefined behavior.
Sep 7 2019, 4:39 AM · Restricted Project, debug-info
bjope committed rL371303: [DebugInfo] Pre-commit of test case for DW_OP_breg/DW_OP_fbreg folds.
[DebugInfo] Pre-commit of test case for DW_OP_breg/DW_OP_fbreg folds
Sep 7 2019, 4:39 AM

Sep 6 2019

bjope updated the diff for D67263: [DwarfExpression] Disallow some rewrites to avoid undefined behavior.

Now using std::numeric_limits and C++ style casts (thanks JDevlieghere!).

Sep 6 2019, 9:58 AM · Restricted Project, debug-info
bjope added a comment to D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic.

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

Sep 6 2019, 7:01 AM · Restricted Project
bjope added inline comments to D67263: [DwarfExpression] Disallow some rewrites to avoid undefined behavior.
Sep 6 2019, 5:14 AM · Restricted Project, debug-info
bjope created D67263: [DwarfExpression] Disallow some rewrites to avoid undefined behavior.
Sep 6 2019, 4:47 AM · Restricted Project, debug-info

Sep 5 2019

bjope added a comment to D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic.

Friendly ping.
(totally understand if tihs hasn't been on top of prio list for reviewers, and that it might be a lot to digest)

Sep 5 2019, 11:51 PM · Restricted Project
bjope added a comment to D67071: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering::expandFixedPointMul.

Worth adding a vector test ?

Sep 5 2019, 3:20 PM · Restricted Project

Sep 4 2019

bjope added inline comments to D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic.
Sep 4 2019, 10:51 PM · Restricted Project
bjope accepted D66234: [MergedLoadStoreMotion] Sink stores to BB with more than 2 predecessors.

LGTM (apart from the comment about maybe saying something about the iteration order).

Sep 4 2019, 10:34 AM · Restricted Project
bjope added inline comments to D66234: [MergedLoadStoreMotion] Sink stores to BB with more than 2 predecessors.
Sep 4 2019, 10:23 AM · Restricted Project

Sep 3 2019

bjope committed rGb0eb394417a7: [CodeGen] Use FSHR in DAGTypeLegalizer::ExpandIntRes_MULFIX (authored by bjope).
[CodeGen] Use FSHR in DAGTypeLegalizer::ExpandIntRes_MULFIX
Sep 3 2019, 1:19 PM
bjope committed rL370813: [CodeGen] Use FSHR in DAGTypeLegalizer::ExpandIntRes_MULFIX.
[CodeGen] Use FSHR in DAGTypeLegalizer::ExpandIntRes_MULFIX
Sep 3 2019, 12:36 PM
bjope closed D67036: [CodeGen] Use FSHR in DAGTypeLegalizer::ExpandIntRes_MULFIX.
Sep 3 2019, 12:36 PM · Restricted Project
bjope added a comment to D67074: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit.

Thanks for all the feedback. Landed as rL370721.

Sep 3 2019, 2:46 AM · Restricted Project
bjope committed rGdd18ce4501e1: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit (authored by bjope).
[LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit
Sep 3 2019, 2:34 AM
bjope committed rG0760d348eb71: [LV] Precommit test case showing miscompile from PR43166. NFC (authored by bjope).
[LV] Precommit test case showing miscompile from PR43166. NFC
Sep 3 2019, 2:34 AM
bjope committed rL370721: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit.
[LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit
Sep 3 2019, 2:33 AM
bjope closed D67074: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit.
Sep 3 2019, 2:33 AM · Restricted Project
bjope committed rL370720: [LV] Precommit test case showing miscompile from PR43166. NFC.
[LV] Precommit test case showing miscompile from PR43166. NFC
Sep 3 2019, 2:33 AM
bjope closed D67072: [LV] Precommit test case showing miscompile from PR43166. NFC.
Sep 3 2019, 2:33 AM · Restricted Project

Sep 2 2019

bjope updated the diff for D67074: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit.

Updated description for AllowedExit.

Sep 2 2019, 6:49 AM · Restricted Project
bjope added a child revision for D67072: [LV] Precommit test case showing miscompile from PR43166. NFC: D67074: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit.
Sep 2 2019, 5:27 AM · Restricted Project
bjope added a parent revision for D67074: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit: D67072: [LV] Precommit test case showing miscompile from PR43166. NFC.
Sep 2 2019, 5:27 AM · Restricted Project
bjope created D67074: [LV] Fix miscompiles by adding non-header PHI nodes to AllowedExit.
Sep 2 2019, 5:27 AM · Restricted Project
bjope created D67072: [LV] Precommit test case showing miscompile from PR43166. NFC.
Sep 2 2019, 5:27 AM · Restricted Project
bjope added a child revision for D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic: D67071: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering::expandFixedPointMul.
Sep 2 2019, 4:46 AM · Restricted Project
bjope added a parent revision for D67071: [CodeGen] Handle SMULFIXSAT with scale zero in TargetLowering::expandFixedPointMul: D57836: [Intrinsic] Unsigned Fixed Point Saturation Multiplication Intrinsic.
Sep 2 2019, 4:46 AM · Restricted Project