vsk (Vedant Kumar)
User

Projects

User Details

User Since
Jul 8 2015, 10:26 AM (123 w, 4 d)

Recent Activity

Wed, Nov 15

vsk removed a watcher for debug-info: vsk.
Wed, Nov 15, 3:02 PM
vsk added a member for debug-info: vsk.
Wed, Nov 15, 3:02 PM
vsk created D40104: [SelectionDAG] Consolidate (t|T)ransferDbgValues methods, NFC..
Wed, Nov 15, 2:59 PM · debug-info
vsk added a watcher for debug-info: vsk.
Wed, Nov 15, 9:56 AM

Tue, Nov 14

vsk updated the diff for D39982: [IRBuilder] Set the insert point and debug location together.
  • Add a release note, and rename the C APIs so they use 'WithLoc' instead of '2'.
Tue, Nov 14, 3:21 PM · debug-info
vsk added a comment to D39446: [PGO] Detect more structural changes with the stable hash.

Thanks for the review Alex. There hasn't been any more feedback so I'll commit this soon.

Tue, Nov 14, 3:11 PM
vsk added a comment to D39982: [IRBuilder] Set the insert point and debug location together.

After some consideration, I no longer think that we should provide an update tool or take the one-shot approach.

Tue, Nov 14, 2:46 PM · debug-info
vsk added inline comments to D39982: [IRBuilder] Set the insert point and debug location together.
Tue, Nov 14, 1:54 PM · debug-info
vsk added inline comments to D39982: [IRBuilder] Set the insert point and debug location together.
Tue, Nov 14, 1:44 PM · debug-info
vsk added inline comments to D40042: Make salvageDebugInfo of casts work for dbg.declare and dbg.addr.
Tue, Nov 14, 1:12 PM
vsk added inline comments to D39982: [IRBuilder] Set the insert point and debug location together.
Tue, Nov 14, 11:52 AM · debug-info
vsk added inline comments to D40042: Make salvageDebugInfo of casts work for dbg.declare and dbg.addr.
Tue, Nov 14, 11:43 AM
vsk added inline comments to D39985: [SCEV] Apply a single debug loc when expanding a SCEV.
Tue, Nov 14, 11:22 AM
vsk added inline comments to D39986: [LSR] Expand: Use the replaced value's debug loc (PR25630).
Tue, Nov 14, 11:21 AM

Mon, Nov 13

vsk added inline comments to D39982: [IRBuilder] Set the insert point and debug location together.
Mon, Nov 13, 8:52 PM · debug-info
vsk updated the diff for D39982: [IRBuilder] Set the insert point and debug location together.
  • Address review feedback from Adrian.
Mon, Nov 13, 8:52 PM · debug-info
vsk updated the diff for D39985: [SCEV] Apply a single debug loc when expanding a SCEV.
  • Pass DebugLoc by value, and add a comment explaining what the PrintTo test helper does.
Mon, Nov 13, 8:49 PM
vsk added inline comments to D39985: [SCEV] Apply a single debug loc when expanding a SCEV.
Mon, Nov 13, 8:49 PM
vsk added inline comments to D39986: [LSR] Expand: Use the replaced value's debug loc (PR25630).
Mon, Nov 13, 8:47 PM
vsk updated the diff for D39986: [LSR] Expand: Use the replaced value's debug loc (PR25630).
  • Address review feedback from Quentin and Adrian.
  • Add test coverage for the logic in GenerateIVChain, and for more users of LSRInstance::Expand.
Mon, Nov 13, 8:45 PM
vsk updated the summary of D39985: [SCEV] Apply a single debug loc when expanding a SCEV.
Mon, Nov 13, 3:18 PM
vsk added a dependent revision for D39982: [IRBuilder] Set the insert point and debug location together: D39985: [SCEV] Apply a single debug loc when expanding a SCEV.
Mon, Nov 13, 3:18 PM · debug-info
vsk created D39986: [LSR] Expand: Use the replaced value's debug loc (PR25630).
Mon, Nov 13, 3:18 PM
vsk added a dependent revision for D39985: [SCEV] Apply a single debug loc when expanding a SCEV: D39986: [LSR] Expand: Use the replaced value's debug loc (PR25630).
Mon, Nov 13, 3:18 PM
vsk created D39985: [SCEV] Apply a single debug loc when expanding a SCEV.
Mon, Nov 13, 3:18 PM
vsk updated the diff for D39982: [IRBuilder] Set the insert point and debug location together.
  • Upload a diff with context.
Mon, Nov 13, 3:14 PM · debug-info
vsk created D39982: [IRBuilder] Set the insert point and debug location together.
Mon, Nov 13, 2:47 PM · debug-info

Fri, Nov 10

vsk added inline comments to D39446: [PGO] Detect more structural changes with the stable hash.
Fri, Nov 10, 3:59 PM

Thu, Nov 9

vsk updated the diff for D39446: [PGO] Detect more structural changes with the stable hash.
  • Consider logical nots and some binary comparison operators in the hash, per an offline conversation with @bogner
Thu, Nov 9, 4:42 PM

Tue, Nov 7

vsk added a comment to D39747: [globalisel][tablegen] Generate rule coverage and use it to identify untested rules.

This looks good to me, but it'd be worth getting another +1.

Tue, Nov 7, 11:40 AM
vsk added inline comments to D39747: [globalisel][tablegen] Generate rule coverage and use it to identify untested rules.
Tue, Nov 7, 10:39 AM

Mon, Nov 6

vsk added inline comments to D39267: [GISel]: Change Legalization from top down to bottom up + DCE.
Mon, Nov 6, 5:54 PM
vsk added a comment to D39628: [DebugInfo] Unify logic to merge DILocations. NFC..

What about the common-parent-scope-finding logic?

Mon, Nov 6, 11:10 AM
vsk added a comment to D39628: [DebugInfo] Unify logic to merge DILocations. NFC..

It looks like this will cause DILocation::getMergedLocation() to create new DILocations? When this happens at the MIR level this will cause problems with the textual MIR output because the new locations aren't printed. Should we introduce a flag to disable creation of new metadata?

Mon, Nov 6, 10:52 AM

Fri, Nov 3

vsk created D39628: [DebugInfo] Unify logic to merge DILocations. NFC..
Fri, Nov 3, 7:24 PM

Wed, Nov 1

vsk updated the diff for D39447: [PGO] Bump the indexed profile format version.
  • Clarify comment.
Wed, Nov 1, 6:22 PM
vsk updated the diff for D39446: [PGO] Detect more structural changes with the stable hash.
  • Handle loop nesting, conditions, and out-of-order control flow.
  • Improve test coverage. Add a format compatibility test, and check that functions which were previously hashed the same way get different hashes now.
Wed, Nov 1, 6:16 PM

Tue, Oct 31

vsk added inline comments to D39355: [dsymutil] Implement the --threads option.
Tue, Oct 31, 4:09 PM
vsk planned changes to D39446: [PGO] Detect more structural changes with the stable hash.

Apart from Alex's point, this patch is missing a compatibility test for v5 of the profile format, and is missing coverage for the new entries in HashType.

Tue, Oct 31, 9:51 AM

Mon, Oct 30

vsk created D39447: [PGO] Bump the indexed profile format version.
Mon, Oct 30, 7:51 PM
vsk created D39446: [PGO] Detect more structural changes with the stable hash.
Mon, Oct 30, 7:48 PM

Fri, Oct 27

vsk accepted D39355: [dsymutil] Implement the --threads option.
Fri, Oct 27, 4:59 PM

Thu, Oct 26

vsk added inline comments to D39355: [dsymutil] Implement the --threads option.
Thu, Oct 26, 6:38 PM
vsk added a comment to D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.

Thanks!

Thu, Oct 26, 2:09 PM
vsk added inline comments to D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.
Thu, Oct 26, 11:42 AM
vsk updated the diff for D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.
  • Tighten test case to show the dbg loc on return instructions.
Thu, Oct 26, 11:42 AM
vsk added inline comments to D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.
Thu, Oct 26, 10:55 AM

Wed, Oct 25

vsk added inline comments to D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.
Wed, Oct 25, 6:21 PM
vsk updated the diff for D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.
  • Simplify the checks by not hardcoding a line number.
Wed, Oct 25, 5:23 PM
vsk created D39310: [CGBlocks] Improve line info in backtraces containing *_helper_block.
Wed, Oct 25, 5:11 PM

Tue, Oct 24

vsk accepted D39212: Check special-case-list regex before insertion..

Yep, lgtm.

Tue, Oct 24, 4:38 PM
vsk created D39257: [Coverage] Provide a stable order for getInstantiationGroups.
Tue, Oct 24, 1:46 PM
vsk added a comment to D39245: [ADT] Shuffle containers before sorting to uncover non-deterministic behavior.

@mgrang thanks, should be fixed in r316490.

Tue, Oct 24, 1:04 PM
vsk added inline comments to D39212: Check special-case-list regex before insertion..
Tue, Oct 24, 11:30 AM

Mon, Oct 23

vsk added inline comments to D39212: Check special-case-list regex before insertion..
Mon, Oct 23, 4:25 PM
vsk added a comment to D27607: [ubsan] Treat ObjC's BOOL as if its range is always {0, 1}.

On platforms where BOOL == signed char, is it actually undefined behavior (or is it just bad programming practice) to store a value other than 0 or 1 in your BOOL? I can't find any language specs suggesting that it is, and given that it's just a typedef for a signed char, I don't see why it would be.

Mon, Oct 23, 11:43 AM

Oct 18 2017

vsk added inline comments to D38777: [wasm] readSection: Avoid reading past eof (fixes oss-fuzz #3219).
Oct 18 2017, 10:35 PM
vsk updated the diff for D38777: [wasm] readSection: Avoid reading past eof (fixes oss-fuzz #3219).
  • Address review feedback, and set aside the dwarfdump changes for later.
Oct 18 2017, 10:34 PM

Oct 13 2017

vsk abandoned D38903: [ubsan] Only use indirect RTTI in prologues on Darwin.

D38913 should make this unnecessary.

Oct 13 2017, 6:38 PM
vsk added a comment to D38903: [ubsan] Only use indirect RTTI in prologues on Darwin.

@pcc made an alternate suggestion which led to D38913. We're still discussing whether the new patch is a sufficient fix.

Oct 13 2017, 6:02 PM
vsk created D38913: [ubsan] Don't emit function signatures for virtual methods.
Oct 13 2017, 5:41 PM
vsk added inline comments to D38777: [wasm] readSection: Avoid reading past eof (fixes oss-fuzz #3219).
Oct 13 2017, 3:06 PM
vsk added inline comments to D38777: [wasm] readSection: Avoid reading past eof (fixes oss-fuzz #3219).
Oct 13 2017, 3:02 PM
vsk added a comment to D38903: [ubsan] Only use indirect RTTI in prologues on Darwin.

Sounds good. This doesn't seem too controversial, since it just takes us back to the old behavior on all platforms except Darwin. I'll wait an hour or so before committing in case there are any more comments.

Oct 13 2017, 2:59 PM
vsk created D38903: [ubsan] Only use indirect RTTI in prologues on Darwin.
Oct 13 2017, 2:50 PM
vsk accepted D38889: [llvm-cov] Reland sources-specified.test with addition of "-path-equivalence"..

Ah, sorry I missed the path issue. This looks good.

Oct 13 2017, 10:25 AM

Oct 12 2017

vsk added inline comments to D38861: [CodeGen] Error on unsupported checked multiplies early.
Oct 12 2017, 6:35 PM
vsk accepted D38859: [clang] Enable clang build with LLVM_BUILD_INSTRUMENTED without setting LLVM_PROFTDATA.

Thanks!

Oct 12 2017, 5:57 PM
vsk added a comment to D38859: [clang] Enable clang build with LLVM_BUILD_INSTRUMENTED without setting LLVM_PROFTDATA.

yeah, i agree, this is not a good idea. My thoughts were different - right now it's not particularly convenient that one has to specify LLVM_PROFDATA when it's not actually used by the build.
Maybe we can create the target "generate-profdata" only if LLVM_PROFDATA is set (but don't fail otherwise) ?

Oct 12 2017, 5:29 PM
vsk accepted D38860: [llvm-cov] Generate "report" for given source paths if sources are specified..

Looks great. Thank you very much!

Oct 12 2017, 5:25 PM
vsk added a comment to D38859: [clang] Enable clang build with LLVM_BUILD_INSTRUMENTED without setting LLVM_PROFTDATA.

llvm-profdata is tightly coupled with the host compiler: while this setup may work if you get lucky, I don't think it's resilient to changes in libProfData. Also, using the instrumented llvm-profdata will be slow and create extra profiles.

Oct 12 2017, 5:20 PM
vsk added a comment to D38861: [CodeGen] Error on unsupported checked multiplies early.

Okay. Sounds good to me.

We intend to actually implement the generic lowering, I hope?

Oct 12 2017, 4:03 PM
vsk added inline comments to D38861: [CodeGen] Error on unsupported checked multiplies early.
Oct 12 2017, 4:00 PM
vsk updated the diff for D38861: [CodeGen] Error on unsupported checked multiplies early.
  • Update to check against a whitelist of supported targets.
Oct 12 2017, 3:59 PM
vsk created D38861: [CodeGen] Error on unsupported checked multiplies early.
Oct 12 2017, 3:26 PM

Oct 10 2017

vsk created D38777: [wasm] readSection: Avoid reading past eof (fixes oss-fuzz #3219).
Oct 10 2017, 6:58 PM
vsk abandoned D38567: [config] Warn when POSIX_C_SOURCE breaks threading support on Darwin.

For those following along, Alex worked out that this doesn't affect apple-clang 802. We took a closer look and found that the build break just affects clang-900, and was introduced in this r290889. The fix (r293167) didn't make it into clang-900. Adding a warning here wouldn't be the right solution, since it would be better to just cherry pick r293167.

Oct 10 2017, 10:38 AM

Oct 4 2017

vsk added a comment to D38567: [config] Warn when POSIX_C_SOURCE breaks threading support on Darwin.

I'm not sure how to test the warning against anything but the macOS SDK. When I tried, I hit a -Wincompatible-sysroot issue. I can leave those changes out of this patch if we want to be more conservative.

Oct 4 2017, 3:45 PM
vsk created D38567: [config] Warn when POSIX_C_SOURCE breaks threading support on Darwin.
Oct 4 2017, 3:43 PM

Oct 3 2017

vsk planned changes to D38210: [ubsan] Port the function sanitizer to C.
In D38210#887635, @pcc wrote:

Wouldn't we get false positives if there is an indirect call in C++ code that calls into C code (or vice versa)?

Oct 3 2017, 3:53 PM
vsk added a reviewer for D38210: [ubsan] Port the function sanitizer to C: arphaman.

Ping.

Oct 3 2017, 3:18 PM
vsk closed D37542: [ubsan] Save a ptrtoint when emitting alignment checks.

Landed as r314749

Oct 3 2017, 11:06 AM
vsk accepted D38507: [llvm-cov] Fix showing title when filtering and not outputting to a directory.

Great, thank you! Lgtm.

Oct 3 2017, 10:59 AM
vsk added a comment to D38507: [llvm-cov] Fix showing title when filtering and not outputting to a directory.

It seems easy enough to add a regression test for this, and I think we should.

Oct 3 2017, 10:28 AM

Oct 2 2017

vsk accepted D38431: [ProfileData] Fix data racing in merging indexed profiles.

Thanks! Lgtm.

Oct 2 2017, 2:12 PM
vsk accepted D38457: [llvm-cov] Hide files with no coverage from the index when filtering by name.

Thanks!

Oct 2 2017, 10:43 AM
vsk added a comment to D38457: [llvm-cov] Hide files with no coverage from the index when filtering by name.

Could you upload a diff with some source context (e.g 'git diff -U1000')?

Oct 2 2017, 8:48 AM

Oct 1 2017

vsk accepted D38441: [compiler-rt] [cmake] Add a separate CMake var to control profile runtime.

Thanks, lgtm.

Oct 1 2017, 7:39 PM

Sep 29 2017

vsk added inline comments to D38431: [ProfileData] Fix data racing in merging indexed profiles.
Sep 29 2017, 4:32 PM

Sep 28 2017

vsk added inline comments to D38382: Fix for the issue 12176 llvm.org/PR12176.
Sep 28 2017, 4:37 PM · Restricted Project
vsk accepted D37810: [test] Enable LeakSanitizer on 64-bit Darwin ASan clang builds.

LGTM, thanks for the pointer.

Sep 28 2017, 1:37 PM

Sep 27 2017

vsk updated subscribers of D37810: [test] Enable LeakSanitizer on 64-bit Darwin ASan clang builds.

I'm inclined to just accept this because it looks simple and harmless. Just to be sure, could you point me to any explanation of why LSan is disabled on Darwin in the first place? CC'ing @gottesmm, since he's looked at using LSan on Darwin before.

Sep 27 2017, 4:04 PM
vsk accepted D38280: [llvm-cov] Create HTML directory structure when filtering using -name*= options.

LGTM with two changes. Thanks!

Sep 27 2017, 8:03 AM

Sep 26 2017

vsk added a comment to D38280: [llvm-cov] Create HTML directory structure when filtering using -name*= options.

Thanks. Could you update the textual coverage logic as well and re-upload the diff with context lines (e.g git diff -U10000)?

Sep 26 2017, 4:15 PM

Sep 25 2017

vsk updated the summary of D38210: [ubsan] Port the function sanitizer to C.
Sep 25 2017, 5:03 AM
vsk updated the diff for D38210: [ubsan] Port the function sanitizer to C.
  • Remove some noisy changes.
Sep 25 2017, 5:02 AM
vsk created D38211: [ubsan] Test -fsanitize=function with a C program.
Sep 25 2017, 5:02 AM
vsk created D38210: [ubsan] Port the function sanitizer to C.
Sep 25 2017, 5:02 AM
vsk added a dependent revision for D38210: [ubsan] Port the function sanitizer to C: D38211: [ubsan] Test -fsanitize=function with a C program.
Sep 25 2017, 5:02 AM

Sep 22 2017

vsk updated the diff for D37544: [ubsan] Skip alignment checks which are folded away.
  • Tighten up lit test.
Sep 22 2017, 10:54 AM