Page MenuHomePhabricator

vsk (Vedant Kumar)
User

Projects

User Details

User Since
Jul 8 2015, 10:26 AM (290 w, 15 h)

Recent Activity

Yesterday

vsk added a comment to D94631: [DebugInfo] Add support for multiple value references in debug values, and enable salvaging.

Just chiming in with compile time numbers on the -O3-optimized bitcode file from D74986.

Wed, Jan 27, 4:41 PM · Restricted Project, debug-info
vsk updated subscribers of D95258: [RGT][ProfileData] Correct a test assertion.

The test update looks correct to me: the expansion regions occur in the file "main", not "expanded". @ikudrin, wdyt?

Wed, Jan 27, 8:59 AM · Restricted Project
vsk added a comment to D83890: [DebugInfo] Process DBG_VALUE_LIST in LiveDebugValues.

Paging @vsk -- it was a while ago, but do you still have the bitcode from D7406 handy, and would you have spare cycles to test it against this (combined patch in D94631)?

Taking a high level view: because variadic variable locations can be part of more than one machine location, here @StephenTozer is adding a VarLoc to each machine location the variable location uses, and keeping an index of all VarLocs in LocIndex=0, so that a "cannonical" VarLoc can be looked up.

I'm not concerned about the average performance, as the cost of extra VarLocs will be proportionate to the number of variadic variable locations, which will be small (at least initially). However, keeping everything in a single index risks defeating the CoalescingBitVector improvements by fragmenting the collection of set bits more than it was before, increasing pointer chasing. Hence it'd be good to test against the original motivating example, if possible.

Wed, Jan 27, 8:54 AM · Restricted Project, debug-info

Thu, Jan 21

vsk added inline comments to D94820: Support for instrumenting only selected files or functions.
Thu, Jan 21, 9:23 AM · Restricted Project, Restricted Project

Fri, Jan 8

vsk committed rGe05baf40de8a: [InitLLVM] Ensure SIGPIPE handler installed before sigaction() (authored by vsk).
[InitLLVM] Ensure SIGPIPE handler installed before sigaction()
Fri, Jan 8, 3:13 PM
vsk closed D94324: [InitLLVM] Ensure SIGPIPE handler installed before sigaction().
Fri, Jan 8, 3:13 PM · Restricted Project, Restricted Project
vsk added a comment to D94324: [InitLLVM] Ensure SIGPIPE handler installed before sigaction().

What are the exact conditions to repro Nick's failure? On Fedora 32, I ran py llvm-lit sigpipe-handling.c before this patch and before rGbf401256edd00e921a5d3a0bf4cf6ee66ae51cd6 , the test suceeds, it should not, should it?

Fri, Jan 8, 1:03 PM · Restricted Project, Restricted Project
vsk updated the diff for D94324: [InitLLVM] Ensure SIGPIPE handler installed before sigaction().

Reduce the test a bit further.

Fri, Jan 8, 12:59 PM · Restricted Project, Restricted Project
vsk accepted D93662: [SimplifyCFG] Keep !dgb metadata of moved instruction, if they match..

Looks good to me as-is. Thanks!

Fri, Jan 8, 11:56 AM · Restricted Project
vsk updated the diff for D94324: [InitLLVM] Ensure SIGPIPE handler installed before sigaction().

Fix a typo in the test.

Fri, Jan 8, 11:42 AM · Restricted Project, Restricted Project
vsk updated the diff for D94324: [InitLLVM] Ensure SIGPIPE handler installed before sigaction().

Add a clang driver test.

Fri, Jan 8, 11:41 AM · Restricted Project, Restricted Project
vsk added a comment to D70277: [Signal] Allow llvm clients to opt into one-shot SIGPIPE handling.

The issue is that InitLLVM initializes its private PrettyStackPrinter before installing the one-shot default SIGPIPE handler: the PrettyStackPrinter itself eventually calls RegisterHandlers(), which runs before the SIGPIPE handler is installed and therefore doesn't register a SIGPIPE handler.

Fri, Jan 8, 11:18 AM · Restricted Project, Restricted Project
vsk requested review of D94324: [InitLLVM] Ensure SIGPIPE handler installed before sigaction().
Fri, Jan 8, 11:16 AM · Restricted Project, Restricted Project
vsk committed rGbf401256edd0: [Signal] Re-raise SIGPIPE if the handler is uninstalled (authored by vsk).
[Signal] Re-raise SIGPIPE if the handler is uninstalled
Fri, Jan 8, 11:14 AM
vsk added a comment to rG9a3f892d0182: [Signal] Allow one-shot SIGPIPE handler to be reached.

Thanks, I've landed this change as bf401256edd0.

Fri, Jan 8, 11:14 AM
vsk added a comment to rG9a3f892d0182: [Signal] Allow one-shot SIGPIPE handler to be reached.

Thanks Vedant! Minor: there's still an unexpected side-effect when InitLLVM is called with InstallPipeSignalExitHandler=false or if sys::SetOneShotPipeSignalFunction is used to clear the default function. In that case, the test will fall through, and will go and call llvm::sys::RunSignalHandlers(). Whereas before it used raise the signal again, and return. What do you think?
I'm wondering if the code shouldn't read more like:

if (llvm::is_contained(IntSigs, Sig) || Sig == SIGPIPE) {
  if (auto OldInterruptFunction = InterruptFunction.exchange(nullptr))
    return OldInterruptFunction();

  if (Sig == SIGPIPE)
    if (auto OldOneShotPipeFunction =
            OneShotPipeSignalFunction.exchange(nullptr))
      return OldOneShotPipeFunction();

  raise(Sig); // Execute the default handler.
  return;
}
Fri, Jan 8, 9:49 AM

Thu, Jan 7

vsk added a comment to D70277: [Signal] Allow llvm clients to opt into one-shot SIGPIPE handling.

Unfortunately I can’t get to this right now, but will take a look first thing tomorrow morning.

Thu, Jan 7, 5:53 PM · Restricted Project, Restricted Project

Tue, Jan 5

vsk added a comment to D93662: [SimplifyCFG] Keep !dgb metadata of moved instruction, if they match..

This improvement looks good and preserves more source locations. Awkwardly, as @vsk says it's quite different from the other documented rules, because it relies on the context of where the instructions are going to be placed.

Tue, Jan 5, 9:11 AM · Restricted Project

Sun, Jan 3

vsk added a comment to D87302: [IRSim][IROutliner] Adding DebugInfo handling for IR outlined functions..
In D87302#2277131, @vsk wrote:

CodeExtractor contains a utility to fix up debug info after extraction occurs (I believe it's called fixupDebugInfoPostExtraction). It's solving the same problem - is there any way to generalize/reuse that logic?

I believe it already uses it since the CodeExtractor is used to create the new functions in each case, but this is mainly for the function that moves them all together. The problem is that fixupDebugInfoPostExtraction transfers the information from the original function, which would not be correct in this case since we are merging the information from several different code locations into one. This also addresses @jroelofs disappointment, but since I'm not aware of any mechanism to say that one particular line in IR is from two different source code locations, I think it has to be dropped altogether for correctness.

Sun, Jan 3, 5:08 PM · debug-info, Restricted Project
vsk added a comment to D93662: [SimplifyCFG] Keep !dgb metadata of moved instruction, if they match..

In its current form, the rule for preserving debug locs doesn't seem to apply here (https://llvm.org/docs/HowToUpdateDebugInfo.html#when-to-preserve-an-instruction-location). Imho we ought to revise the guidance there before proceeding with this patch (e.g. by adding some carveout for when the destination debug loc is the same). If you can share, it might also help to have a little context - unless there's some intervening optimization, the dwarf generator's location cascade should kick in in this situation?

Sun, Jan 3, 5:06 PM · Restricted Project
vsk updated subscribers of D57190: [MC] Teach the MachO object writer about N_FUNC_COLD.
Sun, Jan 3, 4:45 PM · Restricted Project

Dec 15 2020

vsk resigned from D91595: [Debugify] Support checking Machine IR debug info.

Hi @xiangzhangllvm, apologies for the delay.

Dec 15 2020, 11:49 AM · Restricted Project

Dec 11 2020

vsk added a comment to D84467: Add support for Branch Coverage in LLVM Source-Based Code Coverage.
In D84467#2423636, @vsk wrote:

Thank you, lgtm!

Thank you! I also would like to ask if you could commit it for me as I don't have access for that. Note that a handful of tests have binary files which were not uploaded with the diff. What do you advise I do to get those up here so that I can commit it (unfortunately my system doesn't have arcanist setup and that would take some time). Thanks again for all of your help!

Dec 11 2020, 9:01 AM · Restricted Project, Restricted Project, Restricted Project

Dec 3 2020

vsk added inline comments to D91595: [Debugify] Support checking Machine IR debug info.
Dec 3 2020, 10:51 AM · Restricted Project

Dec 2 2020

vsk added inline comments to D91595: [Debugify] Support checking Machine IR debug info.
Dec 2 2020, 10:04 AM · Restricted Project
vsk added inline comments to D92001: [ubsan] Fix crash on __builtin_assume_aligned.
Dec 2 2020, 9:49 AM · Restricted Project, Restricted Project

Dec 1 2020

vsk added inline comments to D91595: [Debugify] Support checking Machine IR debug info.
Dec 1 2020, 1:56 PM · Restricted Project

Nov 30 2020

vsk added inline comments to D91595: [Debugify] Support checking Machine IR debug info.
Nov 30 2020, 12:05 PM · Restricted Project
vsk resigned from D92074: [llvm-profdata] Emit Error when Invalid MemOpSize Section is Created by llvm-profdata.
Nov 30 2020, 11:51 AM · Restricted Project
vsk updated subscribers of D92074: [llvm-profdata] Emit Error when Invalid MemOpSize Section is Created by llvm-profdata.

The error handling additions look good to me, but I don't have the background to lgtm InstrProfWriter::validateRecord. cc @xur.

Nov 30 2020, 11:51 AM · Restricted Project
vsk accepted D84467: Add support for Branch Coverage in LLVM Source-Based Code Coverage.

Thank you, lgtm!

Nov 30 2020, 11:47 AM · Restricted Project, Restricted Project, Restricted Project
vsk added a comment to D91960: [llvm][unittests] Fix protential nullptr dereferences due to unchecked return value or EXPECT_* macro.

Hi @OikawaKirie, thanks for the patch! Would you mind splitting up this patch? This should make each individual patch easier to review, and make it possible for changes to land incrementally (& faster). Generally we ask for patches to be split up into self-contained units: it's not an exact science, but splitting things up by directory might be a reasonable heuristic to start with.

Nov 30 2020, 11:44 AM · Restricted Project

Nov 11 2020

vsk committed rGfc8c1ea9aff3: [lldb/test] Add missing decorators import (authored by vsk).
[lldb/test] Add missing decorators import
Nov 11 2020, 10:49 AM
vsk committed rGb7c06dcb739f: [ThreadPlan] Add a test for `thread step-in -r`, NFC (reapply) (authored by vsk).
[ThreadPlan] Add a test for `thread step-in -r`, NFC (reapply)
Nov 11 2020, 10:44 AM
vsk committed rGd76e01a6a7c2: [MachO] Allow the LC_IDENT load command (authored by vsk).
[MachO] Allow the LC_IDENT load command
Nov 11 2020, 10:16 AM
vsk closed D91221: [MachO] Allow the LC_IDENT load command.
Nov 11 2020, 10:16 AM · Restricted Project

Nov 10 2020

vsk requested review of D91221: [MachO] Allow the LC_IDENT load command.
Nov 10 2020, 5:03 PM · Restricted Project
vsk committed rGae3640e386cc: [ThreadPlan] Add a test for `thread step-in -r`, NFC (authored by vsk).
[ThreadPlan] Add a test for `thread step-in -r`, NFC
Nov 10 2020, 4:38 PM
vsk committed rG9922dded4737: [test] Delete redundant lldbutil import, NFC (authored by vsk).
[test] Delete redundant lldbutil import, NFC
Nov 10 2020, 4:38 PM
vsk closed D91220: [ThreadPlan] Add a test for `thread step-in -r`, NFC.
Nov 10 2020, 4:38 PM · Restricted Project
vsk updated the diff for D91220: [ThreadPlan] Add a test for `thread step-in -r`, NFC.

Simplify, using run_to_source_breakpoint.

Nov 10 2020, 4:27 PM · Restricted Project
vsk committed rG04cd6c62176c: [ThreadPlan] Delete unused ThreadPlanStepInRange code, NFC (authored by vsk).
[ThreadPlan] Delete unused ThreadPlanStepInRange code, NFC
Nov 10 2020, 4:16 PM
vsk committed rG34d56b05fd78: [ThreadPlan] Reflow docs to fit the 80 column limit, NFC (authored by vsk).
[ThreadPlan] Reflow docs to fit the 80 column limit, NFC
Nov 10 2020, 4:16 PM
vsk committed rGba21376883d4: [Command] Fix accidental word concatenation in Options.td (authored by vsk).
[Command] Fix accidental word concatenation in Options.td
Nov 10 2020, 4:16 PM
vsk requested review of D91220: [ThreadPlan] Add a test for `thread step-in -r`, NFC.
Nov 10 2020, 4:10 PM · Restricted Project

Nov 5 2020

vsk committed rG65d15fefe339: [TargetList] Delete the destructor (authored by vsk).
[TargetList] Delete the destructor
Nov 5 2020, 4:57 PM
vsk closed D90895: [TargetList] Delete the destructor.
Nov 5 2020, 4:56 PM · Restricted Project
vsk requested review of D90895: [TargetList] Delete the destructor.
Nov 5 2020, 4:49 PM · Restricted Project
vsk committed rG16e5a347e70b: [TargetList] Simplify dummy target creation (authored by vsk).
[TargetList] Simplify dummy target creation
Nov 5 2020, 4:04 PM
vsk closed D90872: [TargetList] Simplify dummy target creation.
Nov 5 2020, 4:04 PM · Restricted Project
vsk added inline comments to D90872: [TargetList] Simplify dummy target creation.
Nov 5 2020, 2:51 PM · Restricted Project
vsk accepted D90689: [CodeExtractor] Replace uses of extracted bitcasts in out-of-region lifetime markers.

Thanks, lgtm.

Nov 5 2020, 2:01 PM · Restricted Project
vsk added inline comments to D90872: [TargetList] Simplify dummy target creation.
Nov 5 2020, 1:43 PM · Restricted Project
vsk updated the diff for D90872: [TargetList] Simplify dummy target creation.

Make the dummy target per-debugger.

Nov 5 2020, 1:40 PM · Restricted Project
vsk added inline comments to D90872: [TargetList] Simplify dummy target creation.
Nov 5 2020, 1:40 PM · Restricted Project
vsk updated the diff for D90872: [TargetList] Simplify dummy target creation.

Delete some dead code.

Nov 5 2020, 12:02 PM · Restricted Project
vsk requested review of D90872: [TargetList] Simplify dummy target creation.
Nov 5 2020, 11:54 AM · Restricted Project

Nov 4 2020

vsk accepted D90706: [crashlog] Pass the debugger around instead of relying on lldb.debugger.

Looks good to me.

Nov 4 2020, 12:50 PM · Restricted Project
vsk updated subscribers of D90689: [CodeExtractor] Replace uses of extracted bitcasts in out-of-region lifetime markers.
In D90689#2371522, @vsk wrote:

In the example, what pass/component is responsible for creating %1 = bitcast i32* %0 to i8*?

If it's CodeExtractor itself, it would be nice to solve the problem earlier by getting rid of bitcasts which only exist to wire up lifetime markers. I tried doing that in 26ee8aff2, but had to revert (099bffe7f) because llvm doesn't support specializing its lifetime intrinsics on opaque pointer types. That's a bug, I think: if we fix that, perhaps this whole problem goes away.

OTOH, if this %1 = bitcast i32* %0 to i8* value comes from a frontend or another pass, I think the approach taken in this patch is reasonable. Please add a test (perhaps in the CodeExtractor unittest).

Many thanks for the quick reply @vsk! The bitcast is generated by the frontend, not by CodeExtractor. Our use case is outlining a merged parallel region that includes sequential code which include bitcasts like that. I am going to add a unit test.

I have two questions about the CodeExtractor.

  1. I have second thoughts on changing the IR within the findAllocas. Should I create a helper data structure, e.g., ReplaceBitcastUses as in for SinkCands, and do the transformation inside the extractCodeRegion caller?
  2. Is the assumption that findAllocas is always called before findInputsOutputs? If findAllocas is not called, my concern is that findInputsOutputs will report an output on the use of the bitcast although this output would be removed if findAllocas has been called. This problem does not appear for extractCodeRegion because it always calls findAllocas but there may be other modules using the CodeExtractor interface. I could change findInputsOutputs not to include bitcast uses to lifetime markers as outputs regardless of whether findAllocas has been called. Is that reasonable or does that create other problems?
Nov 4 2020, 9:41 AM · Restricted Project
vsk added inline comments to D90706: [crashlog] Pass the debugger around instead of relying on lldb.debugger.
Nov 4 2020, 9:19 AM · Restricted Project

Nov 3 2020

vsk added a comment to D90689: [CodeExtractor] Replace uses of extracted bitcasts in out-of-region lifetime markers.

In the example, what pass/component is responsible for creating %1 = bitcast i32* %0 to i8*?

Nov 3 2020, 9:46 AM · Restricted Project
vsk accepted D89959: UBSAN: emit distinctive traps in trapping mode.

Thanks, lgtm with two comments --

Nov 3 2020, 9:11 AM · Restricted Project

Nov 2 2020

vsk added a comment to D89959: UBSAN: emit distinctive traps in trapping mode.

I might've missed it, but the debug location merging behavior could use a test. Here's one way to write it: https://godbolt.org/z/Yb9PY9.

Nov 2 2020, 8:09 AM · Restricted Project

Oct 30 2020

vsk added inline comments to D89283: [trace][intel-pt] Implement the basic decoding functionality.
Oct 30 2020, 11:16 AM · Restricted Project, Restricted Project

Oct 29 2020

vsk accepted D90423: [llvm-cov][NFC] Remove unused prefixes from FileCheck usage.

Thanks, lgtm!

Oct 29 2020, 3:07 PM · Restricted Project
vsk added inline comments to D89959: UBSAN: emit distinctive traps in trapping mode.
Oct 29 2020, 9:42 AM · Restricted Project
vsk added inline comments to D89959: UBSAN: emit distinctive traps in trapping mode.
Oct 29 2020, 9:30 AM · Restricted Project

Oct 28 2020

vsk added a reverting change for rG99053462216c: [DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927): rGffba94a9acff: Revert "[DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927)".
Oct 28 2020, 6:58 PM
vsk committed rGffba94a9acff: Revert "[DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927)" (authored by vsk).
Revert "[DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927)"
Oct 28 2020, 6:58 PM
vsk added a reverting change for rG2ce36ebca544: [DebugInfo] Shorten legacy [s|z]ext dwarf expressions: rG4fe81b6b6a73: Revert "[DebugInfo] Shorten legacy [s|z]ext dwarf expressions".
Oct 28 2020, 6:58 PM
vsk committed rG4fe81b6b6a73: Revert "[DebugInfo] Shorten legacy [s|z]ext dwarf expressions" (authored by vsk).
Revert "[DebugInfo] Shorten legacy [s|z]ext dwarf expressions"
Oct 28 2020, 6:57 PM
vsk added a reverting change for D89838: [DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927): rGffba94a9acff: Revert "[DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927)".
Oct 28 2020, 6:57 PM · Restricted Project
vsk added a comment to D89838: [DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927).

I'll revert.

Oct 28 2020, 6:55 PM · Restricted Project
vsk accepted D90340: [NFC] Add more tests for DISubprogram verifier.

It looks like verifyFunction asserts that the DISubprogram attached to any function, declaration or definition, is distinct:

Oct 28 2020, 3:29 PM · Restricted Project
vsk committed rG99053462216c: [DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927) (authored by vsk).
[DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927)
Oct 28 2020, 12:06 PM
vsk committed rG158f3360438f: [lldb] Delete lldb/utils/test (authored by vsk).
[lldb] Delete lldb/utils/test
Oct 28 2020, 12:06 PM
vsk committed rG2ce36ebca544: [DebugInfo] Shorten legacy [s|z]ext dwarf expressions (authored by vsk).
[DebugInfo] Shorten legacy [s|z]ext dwarf expressions
Oct 28 2020, 12:06 PM
vsk closed D89838: [DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927).
Oct 28 2020, 12:06 PM · Restricted Project
vsk abandoned D90276: [lldb/utils] Add the lldb-env tool.

Jonas walked me through some of the issues here offline:

  • Looks like this makes testing depend on having a configured lldb-env tool, this will make it hard to run testing against an lldb root.
  • If we need to launch ./bin/lldb in some other workflow, we can consider having lldb-env derive all the info it needs from a root lit.site.cfg.py.
Oct 28 2020, 11:17 AM · Restricted Project
vsk added a comment to D90276: [lldb/utils] Add the lldb-env tool.

Re: making the lit.site.cfg self-contained, IIUC the situation is that there are a couple places where we shell out via subprocess, but that you'd like to get rid of these? I wonder whether we could bundle up the necessary scripts along with lit.site.cfg instead.

My remark was not so much about the need for another tool but rather the need to configure it. Right now I can hand-write a lit.site.cfg.py, omit a bunch of values, and point lit to it. With this tool I'd need to also configure it (i.e. replacing @LLVM_USE_SANITIZER@ and @CMAKE_CXX_COMPILER@.

Oct 28 2020, 9:45 AM · Restricted Project

Oct 27 2020

vsk added a comment to D90276: [lldb/utils] Add the lldb-env tool.

Can you explain the added value of the tool a bit more. I can see how it's nice to be able to run export $(lldb-env); path/to/lldb but I wonder how often you'd end up doing that. When I invoke lldb directly, it's mostly to debug a test case, in which case I'd need to run lldb under lldb and set the environment variables myself. It also means we have to configure an additional file which goes against my effort of making the lit.site.cfg file self contained (to be able to run the test suite against a different toolchain).

Oct 27 2020, 6:26 PM · Restricted Project
vsk requested review of D90276: [lldb/utils] Add the lldb-env tool.
Oct 27 2020, 5:42 PM · Restricted Project
vsk added a comment to D90258: [InstCombine] Do not introduce bitcasts for swifterror arguments..

Test changes look reasonable to me!

Oct 27 2020, 3:02 PM · Restricted Project
vsk added inline comments to D90258: [InstCombine] Do not introduce bitcasts for swifterror arguments..
Oct 27 2020, 12:07 PM · Restricted Project
vsk added a comment to D89838: [DebugInfo] Fix legacy ZExt emission when FromBits >= 64 (PR47927).

Friendly ping.

Oct 27 2020, 10:41 AM · Restricted Project
vsk added a comment to D89283: [trace][intel-pt] Implement the basic decoding functionality.

@wallace thanks for winnowing the test objects. I left an inline suggestion about simplifying the error-handling in IntelPTInstruction. Other than that, mechanically this is looking good. Thanks!

Oct 27 2020, 10:25 AM · Restricted Project, Restricted Project
vsk committed rG5a3ef55a524b: [Utils] Skip RemoveRedundantDbgInstrs in MergeBlockIntoPredecessor (PR47746) (authored by vsk).
[Utils] Skip RemoveRedundantDbgInstrs in MergeBlockIntoPredecessor (PR47746)
Oct 27 2020, 10:13 AM
vsk closed D88928: [Utils] Skip RemoveRedundantDbgInstrs in MergeBlockIntoPredecessor (PR47746).
Oct 27 2020, 10:13 AM · Restricted Project

Oct 26 2020

vsk committed rGa77a739abcfa: [profile] Suppress spurious 'expected profile to require unlock' warning (authored by vsk).
[profile] Suppress spurious 'expected profile to require unlock' warning
Oct 26 2020, 4:25 PM
vsk committed rG905f874c449c: [cmake] Add LLVM_UBSAN_FLAGS, to allow overriding UBSan flags (authored by vsk).
[cmake] Add LLVM_UBSAN_FLAGS, to allow overriding UBSan flags
Oct 26 2020, 3:48 PM
vsk closed D89439: [cmake] Add LLVM_UBSAN_FLAGS, to allow overriding UBSan flags.
Oct 26 2020, 3:48 PM · Restricted Project
vsk added a comment to D89818: [Compile Time] Make it possible to skip redundant debuginst removal.

Friendly ping. On the assumption that the debug inst cleanup in MergeBlockIntoPredecessor was intended to be a compile-time optimization, I propose making it opt-in as the performance impact can be unexpected. I.e., I suggest that we land D88928 soon-ish (tomorrow barring any concerns/objections).

Oct 26 2020, 3:46 PM · Restricted Project
vsk added inline comments to D89439: [cmake] Add LLVM_UBSAN_FLAGS, to allow overriding UBSan flags.
Oct 26 2020, 3:18 PM · Restricted Project

Oct 22 2020

vsk committed rGda27bebb2346: [test] HotColdSplit: cover use of opaque pointer type (authored by vsk).
[test] HotColdSplit: cover use of opaque pointer type
Oct 22 2020, 12:26 PM
vsk added a reverting change for rG26ee8aff2b85: [CodeExtractor] Don't create bitcasts when inserting lifetime markers (NFCI): rG099bffe7f7df: Revert "[CodeExtractor] Don't create bitcasts when inserting lifetime markers….
Oct 22 2020, 12:26 PM
vsk committed rG099bffe7f7df: Revert "[CodeExtractor] Don't create bitcasts when inserting lifetime markers… (authored by vsk).
Revert "[CodeExtractor] Don't create bitcasts when inserting lifetime markers…
Oct 22 2020, 12:26 PM
vsk committed rG3419252a792a: [InstCombine] Remove dbg.values describing contents of dead allocas (authored by vsk).
[InstCombine] Remove dbg.values describing contents of dead allocas
Oct 22 2020, 10:00 AM
vsk closed D85555: [InstCombine] Remove dbg.values describing contents of dead allocas.
Oct 22 2020, 10:00 AM · Restricted Project

Oct 21 2020

vsk added a comment to D89818: [Compile Time] Make it possible to skip redundant debuginst removal.

Comparing with the D88928 patch (in its current state), that patch removes the RemoveRedundantDbgInstrs call from MergeBlockIntoPredecessor, but only replaces that with extra calls to RemoveRedundantDbgInstrs in a few places where MergeBlockIntoPredecessor is used (so afaict that patch currently removes a lot of DbgInstr cleanups?). IMO this patch makes it more of an active choice to decide if the removal can be deferred or not. With D88928 it is easier to forget about such cleanups.

Oct 21 2020, 12:51 PM · Restricted Project
vsk added a comment to D88609: Use uint64_t for branch weights instead of uint32_t.

To my knowledge, we're not relying on the fact that branch_weights are 32-bit for bitcode compatibility purposes, or for PGO. We do support using older .profdata files with newer builds, but (afaict) this change to branch_weights does not affect that.

Oct 21 2020, 11:59 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project