Page MenuHomePhabricator

jmorse (Jeremy Morse)
User

Projects

User does not belong to any projects.

User Details

User Since
Mar 19 2018, 2:57 AM (168 w, 5 d)

Recent Activity

Thu, Jun 10

jmorse added a comment to D87302: [IRSim][IROutliner] Adding DebugInfo handling for IR outlined functions..

I think that's certainly possible. If there was a way in the debug info to say that a specific section of IR is from Section A of the program if called from location A, Section B if called from location B etc that would solve the problem. I don't know enough about DebugInfo to really say how difficult that would be though.

Thu, Jun 10, 9:40 AM · debug-info, Restricted Project

Fri, Jun 4

jmorse committed rG4501928eb2fd: Re-land ae4303b42c, "Track PHI values through register coalescing" (authored by jmorse).
Re-land ae4303b42c, "Track PHI values through register coalescing"
Fri, Jun 4, 3:32 AM
jmorse updated subscribers of D86813: [DebugInstrRef][2/3] Track PHI values through register coalescing.

Yes! Bugpoint can already reduce metadata — I'm wondering if this could maybe a mode of bugpoint in the future?

Fri, Jun 4, 2:26 AM · Restricted Project

Thu, Jun 3

jmorse added a reverting change for rGae4303b42cfa: [DebugInstrRef][2/3] Track PHI values through register coalescing: rG0507fc2ffc96: Revert "[DebugInstrRef][2/3] Track PHI values through register coalescing".
Thu, Jun 3, 9:18 AM
jmorse committed rG0507fc2ffc96: Revert "[DebugInstrRef][2/3] Track PHI values through register coalescing" (authored by jmorse).
Revert "[DebugInstrRef][2/3] Track PHI values through register coalescing"
Thu, Jun 3, 9:17 AM
jmorse added a reverting change for D86813: [DebugInstrRef][2/3] Track PHI values through register coalescing: rG0507fc2ffc96: Revert "[DebugInstrRef][2/3] Track PHI values through register coalescing".
Thu, Jun 3, 9:17 AM · Restricted Project
jmorse committed rGae4303b42cfa: [DebugInstrRef][2/3] Track PHI values through register coalescing (authored by jmorse).
[DebugInstrRef][2/3] Track PHI values through register coalescing
Thu, Jun 3, 9:07 AM
jmorse closed D86813: [DebugInstrRef][2/3] Track PHI values through register coalescing.
Thu, Jun 3, 9:07 AM · Restricted Project
jmorse updated the diff for D86814: [DebugInstrRef][3/3] Follow DBG_PHI instructions through LiveDebugValues..

Ping and refresh of this patch; I've replaced a multimap with a sorted vector, used a class to store information about seen DBG_PHIs, and twiddled some documentation.

Thu, Jun 3, 8:42 AM · Restricted Project

Wed, Jun 2

jmorse added a comment to D86813: [DebugInstrRef][2/3] Track PHI values through register coalescing.

@jmorse @aprantl Sorry for the divergence, but regarding this tests reduction in terms of DI Metadata -- do you think that we can automate this process? I've written a small write-up/proposal for a utility for that purpose: https://github.com/djolertrk/llvm-metadataburn -- please let me know what you think.

Wed, Jun 2, 9:32 AM · Restricted Project

Tue, Jun 1

jmorse updated the diff for D86813: [DebugInstrRef][2/3] Track PHI values through register coalescing.

Address feedback, delete various un-needed MIR bits and pieces, un-necessary DILocations.

Tue, Jun 1, 7:38 AM · Restricted Project

Fri, May 28

jmorse accepted D102917: [LiveDebugVariables] Stop trimming locations of non-inlined vars.

LGTM plus nit.

Fri, May 28, 3:37 AM · Restricted Project, debug-info
jmorse accepted D103306: [IR] Make TypeFinder aware of DIArgList values.

LGTM

Fri, May 28, 3:15 AM · Restricted Project, debug-info

Wed, May 26

jmorse committed rG8496fc2ec804: [DebugInstrRef][1/3] Track PHI values through register allocation (authored by jmorse).
[DebugInstrRef][1/3] Track PHI values through register allocation
Wed, May 26, 12:24 PM
jmorse closed D86812: [DebugInstrRef][1/3] Track PHI values through register allocation.
Wed, May 26, 12:24 PM · Restricted Project
jmorse accepted D103162: [DebugInfo] Limit the number of values that may be referenced by a dbg.value.

For the record, this LGTM too. Picking a better cut-off point is probably something we can experiment with in the future.

Wed, May 26, 7:15 AM · Restricted Project, debug-info

Tue, May 25

jmorse added a comment to D86813: [DebugInstrRef][2/3] Track PHI values through register coalescing.

ping here; I can see the tests here can be improved by removing attributes and a large amount of un-necessary MIR headers, otherwise IMO it's ready to go

Tue, May 25, 9:50 AM · Restricted Project
jmorse added a comment to D86812: [DebugInstrRef][1/3] Track PHI values through register allocation.

NB: I'll probably drop this in sometimes soon, on the back of @djtodoro 's LGTM earlier

Tue, May 25, 9:30 AM · Restricted Project

Mon, May 24

jmorse added a comment to D102917: [LiveDebugVariables] Stop trimming locations of non-inlined vars.

This makes sense to me: the range extension mechanism puts in DBG_VALUEs at points where the live value switches from one register to another, something that LiveDebugValues can't know later on, and often gets wrong. Instruction spans with the "wrong" scope in the middle of a block will prematurely terminate ranges -- or in the case of the prologue, it won't cover early instructions. I think there's also potential for an argument that's moved more than once in the prologue to be dropped, not completely certain.

Mon, May 24, 3:31 AM · Restricted Project, debug-info

Fri, May 21

jmorse added a comment to D91722: [DebugInfo] Use variadic debug values to salvage BinOps and GEP instrs with non-const operands.
Fri, May 21, 1:29 PM · debug-info, Restricted Project

Thu, May 20

jmorse updated the diff for D86812: [DebugInstrRef][1/3] Track PHI values through register allocation.

Ping and update for @djtodoro 's comments -- some minor editing and deleting of un-necessary parts of tests.

Thu, May 20, 7:24 AM · Restricted Project
jmorse accepted D102500: [DebugInfo] Produce undef DBG_VALUE instructions for variadic dbg.values when using FastISel or GlobalIsel.

LGTM given that I've seen a reproducer where this causes an assertion failure -- it's clearly best to avoid the assertion failure. We should re-visit this in the future as there are quite legitimate scenarios where this would be important, for example using optimised LLVM-IR with llc -O0.

Thu, May 20, 6:02 AM · Restricted Project, debug-info
jmorse accepted D101148: [dexter] Change line label reference syntax to enable label-relative offsets (2/2).

LGTM

Thu, May 20, 5:29 AM · Restricted Project, debug-info
jmorse accepted D101147: [dexter] Change line label reference syntax to enable label-relative offsets (1/2).

LGTM. While this is a breaking change, AFAIUI the body of Dexter tests "out there" is small. If this is a serious issue for someone, we can revert and come up with a better plan.

Thu, May 20, 5:29 AM · Restricted Project, debug-info
jmorse accepted D102841: [NFC][Debugify][Original DI] Use MapVector insted of DenseMap for DI tracking.

LGTM

Thu, May 20, 5:16 AM · Restricted Project, debug-info
jmorse accepted D100845: [Debugify][Original DI] Test preservation of original debug var intrinsics in optimizations.

LGTM with D102841

Thu, May 20, 5:16 AM · Restricted Project, debug-info
jmorse accepted D99651: [Dexter] Implement DexDeclareFile, a new Dexter command.

LGTM

Thu, May 20, 4:25 AM · Restricted Project, debug-info
jmorse added inline comments to D100845: [Debugify][Original DI] Test preservation of original debug var intrinsics in optimizations.
Thu, May 20, 4:20 AM · Restricted Project, debug-info

Wed, May 19

jmorse added a comment to D100845: [Debugify][Original DI] Test preservation of original debug var intrinsics in optimizations.

Minor inline nits that can be skipped, plus a question about relying on DenseMap ordering.

Wed, May 19, 9:22 AM · Restricted Project, debug-info
jmorse added a comment to D102680: [debuginfo-tests] Stop using installed LLDB and remove redundancy.

This works fine with our workflow at least, so a soft LGTM from me. I'd suggest waiting for a couple of days for more feedback though.

Wed, May 19, 9:04 AM · Restricted Project
jmorse added a comment to D101064: [WIP] improve debug-info in stack-slot-coloring.

[Digging back through my inbox,]

Wed, May 19, 8:08 AM · Restricted Project
jmorse added a comment to D101247: [Debug-Info] strict dwarf for DW_OP_stack_value.

We need a refactor here to change all places that call addBlock(, location_attr, ) to use DIEDwarfExpression

Wed, May 19, 7:43 AM · Restricted Project, debug-info
jmorse updated the diff for D102158: [DebugInfo][Docs] Add corrected docs for how InstrRefBasedLDV works.

Adjust wording from review; I've added a 'tikzrender.tex' latex header and instructions on how to produce a png from a .tikz diagram. Alas, I'm more familiar with tikz than graphviz.

Wed, May 19, 7:15 AM · Restricted Project

Mon, May 17

jmorse updated subscribers of D101247: [Debug-Info] strict dwarf for DW_OP_stack_value.

Hi @shchenz, is it feasible to use DIExpression::isImplicit at a higher level to filter out these kinds of expressions? I believe it's possible to parse a DIExpression before it's lowered to a DWARF expression, and there are places where we rely on the isImplicit method for correctness already. Also paging @StephenTozer who's touched this area recently.

Mon, May 17, 10:12 AM · Restricted Project, debug-info
jmorse updated subscribers of D102589: [WebAssembly] Nullify DBG_VALUE_LISTs in DebugValueManager.

In the code you've seen, does the debug info all or mostly consist of DBG_VALUE_LISTs now, or is it still mostly DBG_VALUE, or a mix?
I'm just trying to get a sense for how much of a degradation in debug info quality we have now.

Mon, May 17, 9:48 AM · Restricted Project

Fri, May 14

jmorse added a comment to D100165: [GVNHoist] fix a bug where GVNHoist preserves the debug location when it should be dropped.

NB: do you need this and D100270 to be pushed for you?

Fri, May 14, 6:17 AM · Restricted Project

May 11 2021

jmorse added a comment to D101523: [DebugInfo] Fix updateDbgUsersToReg to support DBG_VALUE_LIST.

I'll open a bugzilla ticket about the issue, but since it exists before this patch I think the fix can come later.

May 11 2021, 5:52 AM · Restricted Project, debug-info
jmorse added a comment to D101523: [DebugInfo] Fix updateDbgUsersToReg to support DBG_VALUE_LIST.

Hmmmmm. What if we have:

May 11 2021, 2:26 AM · Restricted Project, debug-info

May 10 2021

jmorse added a comment to D102158: [DebugInfo][Docs] Add corrected docs for how InstrRefBasedLDV works.

NB: a pre-render of the text and diagrams can be found here:

May 10 2021, 3:06 AM · Restricted Project
jmorse requested review of D102158: [DebugInfo][Docs] Add corrected docs for how InstrRefBasedLDV works.
May 10 2021, 3:05 AM · Restricted Project

May 5 2021

jmorse added inline comments to D101523: [DebugInfo] Fix updateDbgUsersToReg to support DBG_VALUE_LIST.
May 5 2021, 7:11 AM · Restricted Project, debug-info
jmorse accepted D101523: [DebugInfo] Fix updateDbgUsersToReg to support DBG_VALUE_LIST.

This LGTM modulo one inline comment, plus: the test is in the "Generic" directory, but it is not fully generic. It depends on the armv6 triple being supported, so things like the powerpc build bots will likely trip on it. You should add an appropriate REQUIRES line.

May 5 2021, 6:07 AM · Restricted Project, debug-info

May 4 2021

jmorse added a comment to D101736: [WebAssembly] Allow DBG_VALUE after terminator in MachineVerifier.

I'm not familiar with WebAssembly, but thought I'd describe how this would works in other backends -- it might inspire you to take a different route. Sorry if it doesn't apply to WebAssembly.

May 4 2021, 11:39 AM · Restricted Project

Apr 29 2021

jmorse added inline comments to D101523: [DebugInfo] Fix updateDbgUsersToReg to support DBG_VALUE_LIST.
Apr 29 2021, 9:18 AM · Restricted Project, debug-info
jmorse added a comment to D101523: [DebugInfo] Fix updateDbgUsersToReg to support DBG_VALUE_LIST.

Looks mechanically correct, test wanted seeing how this was missed before.

Apr 29 2021, 5:27 AM · Restricted Project, debug-info

Apr 28 2021

jmorse added inline comments to rGed29dbaafa49: [DebugInfo] Remove some users of DBG_VALUEs IsIndirect field.
Apr 28 2021, 8:53 AM
jmorse added a comment to D101064: [WIP] improve debug-info in stack-slot-coloring.

Looks good as a starting place -- I took this for test drive, see inline comments. On a clang-3.4 build, the "PC ranges covered" statistic ticks down from 66% to 65%, however I believe that's due to rounding. Using llvm-locstats, roughly six thousand variables move out of the 100% covered bucket, or about 0.2% of all variables in clang-3.4. 0% covered goes up by a negligible amount. In my opinion: this is a reasonable sacrifice to make in the name of accuracy, and some of those dropped locations will be incorrect. Plus, there's a longer term plan for not needing this.

Apr 28 2021, 6:40 AM · Restricted Project

Apr 27 2021

jmorse added inline comments to rGed29dbaafa49: [DebugInfo] Remove some users of DBG_VALUEs IsIndirect field.
Apr 27 2021, 5:18 AM

Apr 26 2021

jmorse added a comment to D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.

Pushed up; you might get buildbot emails about tests failing, as commits are tested in batches. I'll keep an eye on those.

Apr 26 2021, 5:15 AM · Restricted Project
jmorse committed rGcc7803ee3f76: [LSR][DebugInfo] Don't unnecessarily drop DebugLocs (authored by yuanboli233).
[LSR][DebugInfo] Don't unnecessarily drop DebugLocs
Apr 26 2021, 5:15 AM
jmorse closed D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.
Apr 26 2021, 5:15 AM · Restricted Project
jmorse committed rG3c9bcf0e3549: [Clang][Coroutine][DebugInfo] Relax test ordering requirement (authored by jmorse).
[Clang][Coroutine][DebugInfo] Relax test ordering requirement
Apr 26 2021, 2:07 AM
jmorse closed D100298: [Clang][Coroutine][DebugInfo] Follow-up: reduce a tests ordering requirements.
Apr 26 2021, 2:07 AM · Restricted Project

Apr 23 2021

jmorse committed rG7deb970efb0f: Drop a REQUIRES: lldb on a dexter regression test (authored by jmorse).
Drop a REQUIRES: lldb on a dexter regression test
Apr 23 2021, 9:42 AM
jmorse added a comment to D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.

Can you help me to commit the changes as well. I don't have commit access. Thanks, @jmorse

Apr 23 2021, 8:18 AM · Restricted Project
jmorse accepted D100270: [debug-info] MemCpyOpt should merge the debug location when replace multiple block-local instruction with a new memset.

LGTM

Apr 23 2021, 8:16 AM · debug-info, Restricted Project
jmorse accepted D100165: [GVNHoist] fix a bug where GVNHoist preserves the debug location when it should be dropped.

LGTM

Apr 23 2021, 8:14 AM · Restricted Project

Apr 22 2021

jmorse added a comment to D90418: [debuginfo] Vectorizing a loop doesn't terminate all vectorized variable locations.

Looking at the output of the test, it looks like the dbg.value(undef) is in the scalar tail of the loop, rather than vector.body, right? This is slightly cross purposes to what we want, I think: the induction variable value is unavailable during the vectorized portion of the loop and should be undef, but the scalar part isn't optimised, and so the value is still available.

Apr 22 2021, 5:43 AM · Restricted Project, debug-info

Apr 21 2021

jmorse accepted D100572: [DebugInfo] Ensure DIArgList in bitcode has no null or forward metadata refs.

LGTM

Apr 21 2021, 8:54 AM · Restricted Project, debug-info
jmorse accepted D100307: [dexter] Change --source-root-dir and add --debugger-use-relative-paths.

LGTM with a nit

Apr 21 2021, 6:53 AM · Restricted Project, debug-info
jmorse accepted D100572: [DebugInfo] Ensure DIArgList in bitcode has no null or forward metadata refs.

LGTM, I enjoy the minimal extension of a test to improve coverage too.

Apr 21 2021, 6:31 AM · Restricted Project, debug-info

Apr 15 2021

jmorse updated the diff for D100298: [Clang][Coroutine][DebugInfo] Follow-up: reduce a tests ordering requirements.

Right you are -- another thing I didn't clock was that this test was running all of the LLVM passes (generating many instances of the coroutine function). Latest revision disables those passes so that there are only DILocalVariables for the un-coro-split function. I've confirmed this fails before the fix landed and passes afterwards.

Apr 15 2021, 9:33 AM · Restricted Project
jmorse added a comment to D100165: [GVNHoist] fix a bug where GVNHoist preserves the debug location when it should be dropped.

Orlando wrote:

the linked document also says "a transformation should also preserve the debug location of an instruction that is moved between basic blocks, if the destination block already contains an instruction with an identical debug location."

Digging deeper, it looks like this was discussed in the patch that introduces the wording to the document (D93662). cc @fhahn @vsk @jmorse. wdyt should happen here?

Apr 15 2021, 7:35 AM · Restricted Project
jmorse accepted D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.

LGTM

Apr 15 2021, 6:23 AM · Restricted Project
jmorse added a comment to D100270: [debug-info] MemCpyOpt should merge the debug location when replace multiple block-local instruction with a new memset.

This is a good find; style nit and test comments inline.

Apr 15 2021, 6:21 AM · debug-info, Restricted Project

Apr 14 2021

jmorse added a comment to D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.

I think this is good to go, with some revisions to the test.

Apr 14 2021, 9:43 AM · Restricted Project
jmorse added inline comments to D98659: [MachineCopyPropagation] Do more backward copy propagations.
Apr 14 2021, 8:56 AM · Restricted Project

Apr 12 2021

jmorse added a comment to D97533: [Clang][Coroutine][DebugInfo] remove useless debug info for coroutine parameters.

NB, while I flagged that CHECK-NEXT added too much ordering, I forgot that CHECK still requires some ordering. It causes us some difficulties downstream; there's a follow-up patch here D100298 for consideration.

Apr 12 2021, 4:12 AM · Restricted Project, debug-info
jmorse requested review of D100298: [Clang][Coroutine][DebugInfo] Follow-up: reduce a tests ordering requirements.
Apr 12 2021, 4:11 AM · Restricted Project

Apr 9 2021

jmorse accepted D99423: [DebugInfo] Fix incorrect updating of SDNode dependencies for variadic debug values.

LGTM with latest memory jiggery pokery.

Apr 9 2021, 9:29 AM · Restricted Project, debug-info

Apr 7 2021

jmorse accepted D98644: [DebugInfo] Fix incorrect handling of variadic salvaging in Loop Strength Reduce.

LGTM with a nit, and modulo @markus 's comment.

Apr 7 2021, 7:02 AM · Restricted Project, debug-info

Apr 6 2021

jmorse requested changes to D99651: [Dexter] Implement DexDeclareFile, a new Dexter command.

I ended up taking this for a test drive, and found some difficulty when using the LLDB bindings -- _add_conditional_breakpoint ends up being passed a PurePath object for the file_, which is then passed to the LLDB bindings, which croak. This is easily fixed by calling str(file_) to convert the PurePath object to a string, but this might affect other debugger drivers & need some assertions adding.

Apr 6 2021, 2:54 PM · Restricted Project, debug-info
jmorse accepted D99651: [Dexter] Implement DexDeclareFile, a new Dexter command.

LGTM, with a couple of nits, plus Orlandos comments should be addressed.

Apr 6 2021, 9:45 AM · Restricted Project, debug-info

Mar 31 2021

jmorse added inline comments to D99048: [RFC][DebugInfo] Do not use the DBG_VALUE to calculate debug info of spill location.
Mar 31 2021, 7:40 AM · Restricted Project
jmorse added a comment to D98659: [MachineCopyPropagation] Do more backward copy propagations.

Uh, I just figured out how mir tests work.

Mar 31 2021, 7:27 AM · Restricted Project

Mar 26 2021

jmorse added a comment to D99048: [RFC][DebugInfo] Do not use the DBG_VALUE to calculate debug info of spill location.

Thanks for the reproducer; I think I see the same as you, when LiveDebugValues runs:

Mar 26 2021, 10:24 AM · Restricted Project

Mar 25 2021

jmorse added a comment to D97533: [Clang][Coroutine][DebugInfo] remove useless debug info for coroutine parameters.

The test now passes without the rest of the patch applied; as mentioned above, it needs to affirmatively identify the contents of retainedNodes to guarantee detection of extra entries. Alternately, use CHECK-NOT or --implicit-check-not to confirm the absence of the extra DILocalVariables.

Mar 25 2021, 10:21 AM · Restricted Project, debug-info
jmorse added a comment to D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.

Happily we can consult https://llvm.org/docs/HowToUpdateDebugInfo.html -- the example that you've given looks like's a good test for the "positive" case, where we should keep the source location attached to the compare instruction. The comparison doesn't get moved between blocks, so we match "When to preserve an instruction location".

Mar 25 2021, 9:23 AM · Restricted Project
jmorse added a comment to D98644: [DebugInfo] Fix incorrect handling of variadic salvaging in Loop Strength Reduce.

I like the direction of the SelectionDAG change; reducing duplication of information is a worthy goal. My knee-jerk reaction is that we need a test for it, but I imagine it's hard to hit something that deep into DAG Combining. If you've got a reproducer to hand, it'd be good to add that as a test, if not maybe we can skip it. IMO it's also worth breaking this off from the LSR modification.

Mar 25 2021, 7:43 AM · Restricted Project, debug-info
jmorse added a comment to D99273: [DebugInfo] Support for signed constants inside DIExpression.

Sounds good overall, with a comment inline about the test for isSignedConstant. IMO, there should be some test coverage of expressions that shouldn't be recognised as a signed constant too.

Mar 25 2021, 4:51 AM · Restricted Project, debug-info
jmorse added a comment to D82547: [Debugify] Expose original debug info preservation check as CC1 option.

FTR, the patch LGTM. I haven't really been following the development of debugify in detail so I have no feeling for whether this is good from a high level; if you're confident the mailing list discussion finished with this being the right direction, I'd say this is ready to land.

Mar 25 2021, 3:42 AM · Restricted Project, Restricted Project, debug-info

Mar 24 2021

jmorse added a comment to D98659: [MachineCopyPropagation] Do more backward copy propagations.

There isn't a lot of difference to how debug-info is updated, as far as I can see, but the summary says a fault is resolved. Is this patch just masking the debug-info issue by doing more propagation? (I might have missed something in the implementation).

Mar 24 2021, 10:14 AM · Restricted Project
jmorse added a comment to D99048: [RFC][DebugInfo] Do not use the DBG_VALUE to calculate debug info of spill location.

Hmmmm, do you have a reduced reproducer in llvm-ir that could go in a bug report? There are a number of things that could be going on here, and we can't be sure which without an example.

Mar 24 2021, 9:18 AM · Restricted Project
jmorse added inline comments to D97673: [RFC] [[Coroutine] [Debug] Salvage dbg.values.
Mar 24 2021, 8:39 AM · debug-info, Restricted Project
jmorse added a comment to D97673: [RFC] [[Coroutine] [Debug] Salvage dbg.values.

When looping over the debug users of instructions and replacing the dbg.values operands, you should now use the "replaceVariableLocationOp" helper instead -- this avoids re-creating a dbg.value intrinsic, and should handle variadic variable locations (the patches for which are 95% landed) seamlessly.

Mar 24 2021, 8:34 AM · debug-info, Restricted Project
jmorse added a comment to D97533: [Clang][Coroutine][DebugInfo] remove useless debug info for coroutine parameters.

Drive-by review: some minor nits commented on, looks very reasonable overall.

Mar 24 2021, 7:01 AM · Restricted Project, debug-info

Mar 18 2021

jmorse added a comment to D98699: [dexter] Fix DexLimitSteps when breakpoint can't be set at requested location.

To my mind, the two attribute names, _conditional_bp_ranges and _conditional_bps are quite similar -- but AFAICT the former is a on-initialization record of where conditional bps should be placed, while the latter is a "live" record of conditional bp handles, is that right? IMO, better to have an indication of this distinction in the name, may I suggest _conditional_bp_handles?

Mar 18 2021, 6:30 AM · Restricted Project

Mar 17 2021

jmorse accepted D98767: [lit] Harmonize test timing data between Unix and Windows.

Passes on Windows just fine, thanks for the follow up!

Mar 17 2021, 4:40 AM · Restricted Project

Mar 16 2021

jmorse committed rG264f101ae6b4: Tweak spelling of system-windows UNSUPPORTED line (authored by jmorse).
Tweak spelling of system-windows UNSUPPORTED line
Mar 16 2021, 9:52 AM
jmorse added a comment to D98179: [lit] Sort test start times based on prior test timing data.

It turns out there's a plain old windows-has-different-slashes problem going on. In the Test object __init__ method, the call to os.sep.join in this patch produces the string "subdir\ccc.txt", which won't match "subdir/ccc.txt" that you load from the the .lit_test_times.txt file. That then leads to no time being loaded for ccc.txt, and it gets mis-ordered.

Mar 16 2021, 7:11 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project
jmorse added a comment to D98179: [lit] Sort test start times based on prior test timing data.

I'm experiencing reorder.py failing intermittently on a Windows system -- it seems to pick up the .lit_test_times.txt produced in the build directory for the Inputs tests after the first run, and use that for ordering. And seeing how that just contains noise, the Inputs tests are then run in the wrong order. Is this supposed to be accounted for already?

Mar 16 2021, 5:55 AM · Restricted Project, Restricted Project, Restricted Project, Restricted Project

Mar 15 2021

jmorse added inline comments to D41226: [LiveDebugValues] recognize spilled register that is killed in instruction after the spill.
Mar 15 2021, 3:27 AM · Restricted Project

Mar 11 2021

jmorse added a comment to D91722: [DebugInfo] Use variadic debug values to salvage BinOps and GEP instrs with non-const operands.

Modified salvageDebugInfoForDbgValues to only produce DIArgLists for dbg.values, and not dbg.addr or dbg.declare. This is necessary for the time being, as the current implementation of variadic debug values only works with DW_OP_stack_value for reasons previously discussed on the mailing list and in the root patch of this stack. In the next patch to follow, which will make the current DBG_VALUE_LIST into the default debug value instruction, the DIExpression representation will be changed to support non-stack_value variadic debug values, and this change can be reverted.

Mar 11 2021, 4:46 AM · debug-info, Restricted Project

Mar 9 2021

jmorse added a comment to D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.

In principle this looks good, but as @djtodoro points out it needs a test.

Mar 9 2021, 9:12 AM · Restricted Project

Mar 8 2021

jmorse updated subscribers of D98218: [LSR] fix a issue that LoopStrengthReduction drop debug location unnecessarily.
Mar 8 2021, 2:05 PM · Restricted Project
jmorse added a comment to D86377: [tsan] Use large address space mapping on Apple Silicon Macs.

I've uploaded a new version of the patch (just moving the definition of HAS_48_BIT_ADDRESS_SPACE outside of the SANITIZER_GO branch). @dyung @jmorse could I ask you to check this patch whether it builds correctly on x86_64/linux? Thanks!

Mar 8 2021, 8:33 AM · Restricted Project
jmorse added a reverting change for rGbde2e5607167: [tsan] Use large address space mapping on Apple Silicon Macs: rG66e3a4abe99c: Revert "[tsan] Use large address space mapping on Apple Silicon Macs".
Mar 8 2021, 1:54 AM
jmorse committed rG66e3a4abe99c: Revert "[tsan] Use large address space mapping on Apple Silicon Macs" (authored by jmorse).
Revert "[tsan] Use large address space mapping on Apple Silicon Macs"
Mar 8 2021, 1:54 AM
jmorse added a reverting change for D86377: [tsan] Use large address space mapping on Apple Silicon Macs: rG66e3a4abe99c: Revert "[tsan] Use large address space mapping on Apple Silicon Macs".
Mar 8 2021, 1:54 AM · Restricted Project

Feb 24 2021

jmorse accepted D95463: Avoid redundant or out-of-order debug value sinking in InstCombine.

LGTM, but pls2leave it 24 hours or so in case @rnk chimes in.

Feb 24 2021, 9:52 AM · Restricted Project, debug-info