Page MenuHomePhabricator
Feed Advanced Search

Today

dsanders added a comment to D83034: [GlobalISel] Don't skip adding predicate matcher.

Could you add a test case or point us at a example of an existing test that goes wrong? The description in the commit message is unlikely to be the full story as we already cover loads with non-null MemoryVT's. My best guess is you are attempting to use builtin predicates and custom predicates together. I don't see a reason why that shouldn't be allowed but it's not something that was intended as the goal was aiming to fully remove the custom C++ from the tablegen headers so that tablegen could do some transformations on sextload/zextload and similar to fix DAGISel vs GlobalISel modelling mismatches.

Mon, Jul 6, 9:42 AM · Restricted Project

Mon, Jun 29

dsanders added a comment to D82765: GlobalISel: Disallow undef generic virtual register uses.

From that it *sounds* like a DBG_VALUE of an undef vreg doesn't make sense?

Mon, Jun 29, 5:13 PM · Restricted Project
dsanders added a comment to D82765: GlobalISel: Disallow undef generic virtual register uses.

My MIR is lacking here: What does it mean for a USE to be UNDEF? DBG_VALUE accepts UNDEF as an operand, but it's not clear to me what an undef vreg means / how it can be created.

You can set the undef flag on the MachineOperand.

ProcessImplicitDefs turns

%0 = IMPLICIT_DEF
FOO %0

into

FOO undef %0

You can also hand write this, but I don't think any pass before ProcessImplicitDefs and co should be using these

Mon, Jun 29, 5:13 PM · Restricted Project
dsanders added reviewers for D82765: GlobalISel: Disallow undef generic virtual register uses: vsk, aprantl.

@vsk + @aprantl: I believe this is going to need an exception for DBG_VALUE. Could you confirm if that's correct?

Mon, Jun 29, 10:48 AM · Restricted Project

Mon, Jun 22

dsanders accepted D81074: [TableGen] Add error messages.

LGTM

Mon, Jun 22, 10:12 AM · Restricted Project

Tue, Jun 16

dsanders committed rG778db88723d7: [gicombiner] Allow disable-rule option to disable all-except-... (authored by dsanders).
[gicombiner] Allow disable-rule option to disable all-except-...
Tue, Jun 16, 5:02 PM
dsanders closed D81889: [gicombiner] Allow disable-rule option to disable all-except-....
Tue, Jun 16, 5:02 PM · Restricted Project
dsanders updated the diff for D81889: [gicombiner] Allow disable-rule option to disable all-except-....

Add only-enable-rule and make it interact with disable-rule correctly.

Tue, Jun 16, 4:29 PM · Restricted Project
dsanders added a comment to D81889: [gicombiner] Allow disable-rule option to disable all-except-....

The point here is to allow people to only enable specific rules, right?

Tue, Jun 16, 3:23 PM · Restricted Project
dsanders committed rGe35ba09961c8: [gicombiner] Allow generated combiners to store additional members (authored by dsanders).
[gicombiner] Allow generated combiners to store additional members
Tue, Jun 16, 2:52 PM
dsanders closed D81863: [gicombiner] Allow generated combiners to store additional members.
Tue, Jun 16, 2:51 PM · Restricted Project
dsanders committed rG28ebdf1a02eb: [gicombiner] Allow generated CombinerHelpers to have additional arguments (authored by dsanders).
[gicombiner] Allow generated CombinerHelpers to have additional arguments
Tue, Jun 16, 2:22 PM
dsanders closed D81862: [gicombiner] Allow generated CombinerHelpers to have additional arguments.
Tue, Jun 16, 2:21 PM · Restricted Project
dsanders added inline comments to D81899: [gicombiner] Unify common state for current targets into CommonTargetCombinerHelperState.
Tue, Jun 16, 11:00 AM · Restricted Project
dsanders added a comment to D81889: [gicombiner] Allow disable-rule option to disable all-except-....

It's not clear to me what "'!foo-bar' - Enable rules from 'foo' to (and including) 'bar'" means. These have some kind of ordering?

Tue, Jun 16, 10:26 AM · Restricted Project

Mon, Jun 15

dsanders updated the summary of D81889: [gicombiner] Allow disable-rule option to disable all-except-....
Mon, Jun 15, 6:14 PM · Restricted Project
dsanders updated the summary of D81889: [gicombiner] Allow disable-rule option to disable all-except-....
Mon, Jun 15, 6:14 PM · Restricted Project
dsanders added a child revision for D81863: [gicombiner] Allow generated combiners to store additional members: D81889: [gicombiner] Allow disable-rule option to disable all-except-....
Mon, Jun 15, 6:14 PM · Restricted Project
dsanders added a parent revision for D81889: [gicombiner] Allow disable-rule option to disable all-except-...: D81863: [gicombiner] Allow generated combiners to store additional members.
Mon, Jun 15, 6:14 PM · Restricted Project
dsanders created D81899: [gicombiner] Unify common state for current targets into CommonTargetCombinerHelperState.
Mon, Jun 15, 5:42 PM · Restricted Project
dsanders added a child revision for D81889: [gicombiner] Allow disable-rule option to disable all-except-...: D81899: [gicombiner] Unify common state for current targets into CommonTargetCombinerHelperState.
Mon, Jun 15, 5:42 PM · Restricted Project
dsanders created D81889: [gicombiner] Allow disable-rule option to disable all-except-....
Mon, Jun 15, 3:29 PM · Restricted Project
dsanders added inline comments to D81863: [gicombiner] Allow generated combiners to store additional members.
Mon, Jun 15, 3:29 PM · Restricted Project
dsanders added a comment to D81862: [gicombiner] Allow generated CombinerHelpers to have additional arguments.

Doesn't https://reviews.llvm.org/D81863 make this unnecessary?

Mon, Jun 15, 3:29 PM · Restricted Project
dsanders added inline comments to D81863: [gicombiner] Allow generated combiners to store additional members.
Mon, Jun 15, 1:14 PM · Restricted Project
dsanders created D81863: [gicombiner] Allow generated combiners to store additional members.
Mon, Jun 15, 12:08 PM · Restricted Project
dsanders added a child revision for D81862: [gicombiner] Allow generated CombinerHelpers to have additional arguments: D81863: [gicombiner] Allow generated combiners to store additional members.
Mon, Jun 15, 12:08 PM · Restricted Project
dsanders created D81862: [gicombiner] Allow generated CombinerHelpers to have additional arguments.
Mon, Jun 15, 12:07 PM · Restricted Project

Fri, Jun 12

dsanders accepted D81587: [MachineVerifier][GlobalISel] Check that branches have a MBB operand or are declared indirect. Add missing properties to G_BRJT, G_BRINDIRECT.
Fri, Jun 12, 9:01 PM · Restricted Project
dsanders edited child revisions for D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled, added: 1; removed: 1.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a parent revision for D81777: [llvmbuild] --enable-optional-components should not be a substring match: D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a parent revision for D81778: [llvmbuild] Add optional_libraries, i.e. a dependency iff the target is enabled: D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders removed a parent revision for D81778: [llvmbuild] Add optional_libraries, i.e. a dependency iff the target is enabled: D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a child revision for D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled: D81778: [llvmbuild] Add optional_libraries, i.e. a dependency iff the target is enabled.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a child revision for D80960: Support other llc-like tools in update_llc_test_checks.py: D81777: [llvmbuild] --enable-optional-components should not be a substring match.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a parent revision for D81777: [llvmbuild] --enable-optional-components should not be a substring match: D80960: Support other llc-like tools in update_llc_test_checks.py.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders removed a parent revision for D81777: [llvmbuild] --enable-optional-components should not be a substring match: D80960: Support other llc-like tools in update_llc_test_checks.py.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders removed a child revision for D80960: Support other llc-like tools in update_llc_test_checks.py: D81777: [llvmbuild] --enable-optional-components should not be a substring match.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a child revision for D81777: [llvmbuild] --enable-optional-components should not be a substring match: D81778: [llvmbuild] Add optional_libraries, i.e. a dependency iff the target is enabled.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders created D81778: [llvmbuild] Add optional_libraries, i.e. a dependency iff the target is enabled.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders added a parent revision for D81778: [llvmbuild] Add optional_libraries, i.e. a dependency iff the target is enabled: D81777: [llvmbuild] --enable-optional-components should not be a substring match.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders created D81777: [llvmbuild] --enable-optional-components should not be a substring match.
Fri, Jun 12, 5:08 PM · Restricted Project
dsanders updated the diff for D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled.

Remove a bit of the next patch that snuck into this one while organizing the patches

Fri, Jun 12, 4:34 PM · Restricted Project
dsanders added a comment to D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled.

Just to give you some context and an overview of the direction, we have some hacks we use to make our LLVM dylib smaller by omitting subcomponents of LLVM that we don't actually use in practice. This is part of a small series of patches to try to eliminate these hacks by rounding out llvmbuild's support for optional libraries. This patch allows libraries to be included by default and optionally disabled, and it will be followed by a small bugfix for --enable-optional-components, and then the addition of an optional_libraries to go with the existing required_libraries.

Fri, Jun 12, 4:34 PM · Restricted Project
dsanders created D81773: [llvmbuild] Allow OptionalLibrary's to be enabled by default and disabled.
Fri, Jun 12, 4:34 PM · Restricted Project

Thu, Jun 11

dsanders accepted D81625: [GISel][NFC]: Add unit test for clarifying CSE behavior.

LGTM

Thu, Jun 11, 10:27 AM · Restricted Project
dsanders added inline comments to D81587: [MachineVerifier][GlobalISel] Check that branches have a MBB operand or are declared indirect. Add missing properties to G_BRJT, G_BRINDIRECT.
Thu, Jun 11, 9:53 AM · Restricted Project

Wed, Jun 10

dsanders committed rGf743975e0ab6: Support other llc-like tools in update_llc_test_checks.py (authored by dsanders).
Support other llc-like tools in update_llc_test_checks.py
Wed, Jun 10, 5:48 PM
dsanders closed D80960: Support other llc-like tools in update_llc_test_checks.py.
Wed, Jun 10, 5:48 PM · Restricted Project

Tue, Jun 9

dsanders accepted D81387: GlobalISel: Set instr/debugloc before any legalizer action.

LGTM

Tue, Jun 9, 9:53 AM · debug-info, Restricted Project
dsanders accepted D81461: GlobalISel: Fix double printing new instructions in legalizer.

I'm not sure this is the correct instance to remove; now the visible
ordering is different. Now you will typically see the one erased
instruction message before all the new instructions in order. I think
this is the more logical view of typical legalization changes,
although it's mechanically backwards from the normal
insert-new-erase-old pattern.

Tue, Jun 9, 9:52 AM · Restricted Project

Jun 3 2020

dsanders accepted D81057: [SVE] Fix ubsan issues in DecodeIITType.

LGTM

Jun 3 2020, 9:53 AM · Restricted Project
dsanders added inline comments to D81074: [TableGen] Add error messages.
Jun 3 2020, 9:53 AM · Restricted Project

Jun 2 2020

dsanders accepted D81021: [TableGen] Avoid generating switch with just default.

LGTM

Jun 2 2020, 12:04 PM · Restricted Project

Jun 1 2020

dsanders added inline comments to rG1c3d9c2f3629: [SVE] Remove IITDescriptor::ScalableVecArgument.
Jun 1 2020, 5:54 PM
dsanders created D80960: Support other llc-like tools in update_llc_test_checks.py.
Jun 1 2020, 3:43 PM · Restricted Project
dsanders added a comment to D80390: [mips] Support 64-bit relative relocations.

LGTM once all other reviewers are happy with this patch.

You don't really need to wait for all of us but you should probably wait for @MaskRay and @atanasyan as they had comments.

Jun 1 2020, 12:59 PM · Restricted Project

May 29 2020

dsanders added inline comments to rG1c3d9c2f3629: [SVE] Remove IITDescriptor::ScalableVecArgument.
May 29 2020, 2:14 PM

May 19 2020

dsanders added inline comments to D80056: GlobalISel: Merge G_PTR_MASK with llvm.ptrmask intrinsic.
May 19 2020, 7:16 PM · Restricted Project
dsanders added a comment to D77446: Add MIR-level debugify with only locations support for now.

Thanks for this work! I've missed this, since it was not marked as 'debug-info' project.

I think we should document this at https://llvm.org/docs/SourceLevelDebugging.html#testing-debug-info-preservation-in-optimizations.

The D80052 will address this comment.

May 19 2020, 7:16 PM · debug-info, Restricted Project
dsanders added a comment to D80052: [docs] Sketch outline for HowToUpdateDebugInfo.rst.

Thanks for working on this. I had a couple inline comments.

May 19 2020, 7:16 PM · debug-info, Restricted Project

May 18 2020

dsanders created D80145: [docs] Enable syntax highlight of MIR (requires pygments 2.6.1).
May 18 2020, 10:48 AM · Restricted Project

May 7 2020

dsanders added a comment to D78796: [Support] Refactor LEB128 encoding into an input iterator.

I'm inclined to agree that the patch series as-is doesn't really warrant the iterators as the interface as no callers have been updated. However, I also don't see much that's iterator specific (ULEBifier would be roughly similar code leaving the iterator portion as trivial wrappers on the ULEBifier) and there are a few places (particularly in tablegen) that are emitting LEB's into containers where the loop to add bytes one by one is just noise and something like std::copy(to_uleb(...), std::back_inserter(Table)); would be somewhat nice. The loop could easily be hidden in something like append_uleb(Table, ...) though I don't think there's a strong argument for (or against) iterators.

What I would suggest is separating out the byte-sequence generation into a ULEBifier as David suggested but still keeping the iterator object as a thin adapter (effectively implementing that while loop) to support inter-operation with STL functions like std::copy.

What is the core issue with the iterator interface that makes it desirable to have something like ULEBifier<T> instead?

May 7 2020, 2:09 PM · Restricted Project
dsanders added a comment to D78796: [Support] Refactor LEB128 encoding into an input iterator.

I'm inclined to agree that the patch series as-is doesn't really warrant the iterators as the interface as no callers have been updated. However, I also don't see much that's iterator specific (ULEBifier would be roughly similar code leaving the iterator portion as trivial wrappers on the ULEBifier) and there are a few places (particularly in tablegen) that are emitting LEB's into containers where the loop to add bytes one by one is just noise and something like std::copy(to_uleb(...), std::back_inserter(Table)); would be somewhat nice. The loop could easily be hidden in something like append_uleb(Table, ...) though I don't think there's a strong argument for (or against) iterators.

May 7 2020, 11:55 AM · Restricted Project

May 5 2020

dsanders accepted D79419: [clang/clang-tools-extra] Fix BZ44437 - add_new_check.py does not work with Python 3.

LGTM

May 5 2020, 1:30 PM · Restricted Project

Apr 29 2020

dsanders added a comment to D78245: [LIT] Make `%T` unique for every test.

@thakis Could you link to previous discussions about %T? I'm sure there are reasons for the way things are today, but I'd like to understand them.

I remember reading the discussion, but I failed to find it just now. @dsanders added the docs that say "deprecated"; maybe he remembers where the thread on moving off %T was?

Apr 29 2020, 2:00 PM · Restricted Project, Restricted Project, Restricted Project

Apr 28 2020

dsanders added inline comments to D78852: [AArch64][GlobalISel] Add a post-legalizer combiner + remove undef stores.
Apr 28 2020, 3:40 PM · Restricted Project
dsanders accepted D79031: [GlobalISel] Assign the correct debug location when combining G_SEXT/G_ANYEXT/G_ZEXT.

LGTM but double check -verify-machineinstrs works with -run-pass

Apr 28 2020, 2:34 PM · Restricted Project

Apr 27 2020

dsanders added inline comments to D78309: [LLVM] Remove wrong DBG_VALUE instruction with one operand in AArch64 test case.
Apr 27 2020, 4:11 PM · Restricted Project
dsanders accepted D78450: CMake: Missing GlobalISel TableGen link.

I don't think it's safe to remove Support since llvm/lib/TableGen/CMakeLists.txt doesn't mention it but adding TableGen to the existing list would LGTM

Apr 27 2020, 3:39 PM · Restricted Project

Apr 21 2020

dsanders accepted D78135: [MachineDebugify] Insert synthetic DBG_VALUE instructions.

LGTM

Apr 21 2020, 2:39 PM · Restricted Project

Apr 17 2020

dsanders committed rG14ad8dc0761a: Don't accidentally create MachineFunctions in mir-debugify/mir-strip-debugify (authored by dsanders).
Don't accidentally create MachineFunctions in mir-debugify/mir-strip-debugify
Apr 17 2020, 2:38 PM
dsanders committed rG701af684f65d: [globalisel][legalizer] Expect to lose DebugLocs in dead code (authored by dsanders).
[globalisel][legalizer] Expect to lose DebugLocs in dead code
Apr 17 2020, 2:06 PM
dsanders committed rG5ef64bbf7ac4: [globalisel][legalizer] Include newly-dead code in artifact combine checks for… (authored by dsanders).
[globalisel][legalizer] Include newly-dead code in artifact combine checks for…
Apr 17 2020, 2:06 PM
dsanders committed rG7f7f98b15448: [globalisel][legalizer] Fix --verify-legalizer-debug-locs values (authored by dsanders).
[globalisel][legalizer] Fix --verify-legalizer-debug-locs values
Apr 17 2020, 2:05 PM
dsanders added a comment to D78388: [globalisel][cse] Merge debug locations when CSE'ing.

I currently don't have an in-tree test case for this but it should be possible to construct one. Essentially it needs CSE enabled, and needs two instructions that need legalization and emit a common instruction during their legalizations with identical constraints (to avoid the copy) and have conflicting debug locations. I'll try to make an AArch64 test case when I get chance.

Apr 17 2020, 11:20 AM · Restricted Project
dsanders created D78388: [globalisel][cse] Merge debug locations when CSE'ing.
Apr 17 2020, 11:20 AM · Restricted Project

Apr 16 2020

dsanders committed rG561cb14e743f: [LLVM] Remove wrong DBG_VALUE instruction with one operand in AArch64 test case (authored by Jac1494).
[LLVM] Remove wrong DBG_VALUE instruction with one operand in AArch64 test case
Apr 16 2020, 12:16 PM
dsanders closed D78309: [LLVM] Remove wrong DBG_VALUE instruction with one operand in AArch64 test case.
Apr 16 2020, 12:15 PM · Restricted Project
dsanders added a comment to D78309: [LLVM] Remove wrong DBG_VALUE instruction with one operand in AArch64 test case.

Hi @dsanders ,

Thank you for accepting this. I don't have commit access please commit this.

Apr 16 2020, 11:41 AM · Restricted Project
dsanders committed rGd9085f65db0b: [globalisel] Add lost debug locations verifier (authored by dsanders).
[globalisel] Add lost debug locations verifier
Apr 16 2020, 11:10 AM
dsanders closed D77576: [globalisel] Add lost debug locations verifier.
Apr 16 2020, 11:09 AM · Restricted Project
dsanders committed rG7c6ca18fffdc: [globalisel] Allow backends to report an issue without triggering fallback. NFC (authored by dsanders).
[globalisel] Allow backends to report an issue without triggering fallback. NFC
Apr 16 2020, 11:09 AM
dsanders closed D78261: [globalisel] Allow backends to report an issue without triggering fallback. NFC.
Apr 16 2020, 11:09 AM · Restricted Project
dsanders accepted D78309: [LLVM] Remove wrong DBG_VALUE instruction with one operand in AArch64 test case.

LGTM. I'll remove the workaround in mir-strip-debug once this lands

Apr 16 2020, 10:35 AM · Restricted Project

Apr 15 2020

dsanders added a comment to D78261: [globalisel] Allow backends to report an issue without triggering fallback. NFC.

Thanks

Apr 15 2020, 7:19 PM · Restricted Project
dsanders accepted D78253: [GIsel][CombinerHelper] Don't consider debug insts in dominance queries [3/10].

LGTM

Apr 15 2020, 7:19 PM · Restricted Project
dsanders created D78261: [globalisel] Allow backends to report an issue without triggering fallback. NFC.
Apr 15 2020, 6:46 PM · Restricted Project
dsanders added a parent revision for D77576: [globalisel] Add lost debug locations verifier: D78261: [globalisel] Allow backends to report an issue without triggering fallback. NFC.
Apr 15 2020, 6:46 PM · Restricted Project
dsanders added a child revision for D78261: [globalisel] Allow backends to report an issue without triggering fallback. NFC: D77576: [globalisel] Add lost debug locations verifier.
Apr 15 2020, 6:46 PM · Restricted Project
dsanders added a comment to D77576: [globalisel] Add lost debug locations verifier.

Hi Daniel,

What is the compile time impact of this change?

That depends on VerifyDebugLocs. When it's None (default for release), there shouldn't be a measurable impact as the observer doesn't get installed. I haven't measured it for Locations (current default for asserts) yet as I've been more focused on fixing the bugs it finds in our out-of-tree target so far. I don't expect it to be noticable as it's a fairly simple matcher on small search spaces but if it is we can change the default to None at the cost of needing to use an extra option in out llvm-lit -Dllc='llc ...' command.

Apr 15 2020, 4:35 PM · Restricted Project
dsanders added a comment to D78137: [GISel][RegBankSelect] Hide assertion failure from LLT::getScalarSizeInBits [2/10].
In D78137#1982421, @vsk wrote:

(lldb) p MF.dump()

  1. Machine code for function phiPropagation: IsSSA, TracksLiveness, Legalized

    bb.0.entry: successors: %bb.2(0x40000000), %bb.1(0x40000000); %bb.2(50.00%), %bb.1(50.00%) liveins: $x0, $x1, $w2 %0:gpr32 = LDRWui killed $x0, 0, debug-location !23 :: (load 4 from %ir.src); /Users/vsk/src/llvm-project-master/llvm/test/CodeGen/AArch64/GlobalISel/arm64-regbankselect.mir:6:1 DBG_VALUE %0:gpr32, $noreg, !"1", !DIExpression(), debug-location !23; /Users/vsk/src/llvm-project-master/llvm/test/CodeGen/AArch64/GlobalISel/arm64-regbankselect.mir:6:1 line no:6 `

If you moved the DBG_VALUE below the %5(s32) = COPY %0 does it still assert? It's interesting that the target instruction means the DBG_VALUE is the first to be processed in that def/use chain but I suspect the real problem is that it's not deriving the register bank from the register class and the special case for COPY deals with that.

No, after moving the DBG_VALUE, I don't hit the assertion:

%0:gpr32 = LDRWui
%5:_(s32) = COPY %0
DBG_VALUE %5

Sorry, I'm not entirely sure what to make of that. Do you suspect -mir-debugify is adding DBG_VALUE insts in the wrong place? In the generic regbankselect path, is it required for operands to go through a COPY before they can be used?

Apr 15 2020, 10:24 AM · Restricted Project

Apr 14 2020

dsanders added a comment to D78137: [GISel][RegBankSelect] Hide assertion failure from LLT::getScalarSizeInBits [2/10].

(lldb) p MF.dump()

  1. Machine code for function phiPropagation: IsSSA, TracksLiveness, Legalized

    bb.0.entry: successors: %bb.2(0x40000000), %bb.1(0x40000000); %bb.2(50.00%), %bb.1(50.00%) liveins: $x0, $x1, $w2 %0:gpr32 = LDRWui killed $x0, 0, debug-location !23 :: (load 4 from %ir.src); /Users/vsk/src/llvm-project-master/llvm/test/CodeGen/AArch64/GlobalISel/arm64-regbankselect.mir:6:1 DBG_VALUE %0:gpr32, $noreg, !"1", !DIExpression(), debug-location !23; /Users/vsk/src/llvm-project-master/llvm/test/CodeGen/AArch64/GlobalISel/arm64-regbankselect.mir:6:1 line no:6 `
Apr 14 2020, 4:52 PM · Restricted Project
dsanders added a comment to D78135: [MachineDebugify] Insert synthetic DBG_VALUE instructions.

There is essentially no attempt made to match up DBG_VALUE register
operands with the local variables they ought to correspond to. I'm not
sure how to improve the situation.

Apr 14 2020, 4:52 PM · Restricted Project
dsanders updated the diff for D77576: [globalisel] Add lost debug locations verifier.

One more last-minute fix: If there was a line-0 location in the input
and the output it was matching them up. If another location folded
into that line-0 location it would falsely be considered lost despite
being handled correctly because the code to treat line-0 locations
as covering everything was not reached.

Apr 14 2020, 4:20 PM · Restricted Project
dsanders accepted D78137: [GISel][RegBankSelect] Hide assertion failure from LLT::getScalarSizeInBits [2/10].

Thanks for fixing this. LGTM

Apr 14 2020, 3:45 PM · Restricted Project

Apr 13 2020

dsanders added a comment to D77576: [globalisel] Add lost debug locations verifier.

Hi Daniel,

What is the compile time impact of this change?

Apr 13 2020, 2:41 PM · Restricted Project
dsanders updated the diff for D77576: [globalisel] Add lost debug locations verifier.

static const
Doxygen comment
Make a function private

Apr 13 2020, 11:17 AM · Restricted Project

Apr 10 2020

dsanders added inline comments to D77576: [globalisel] Add lost debug locations verifier.
Apr 10 2020, 5:47 PM · Restricted Project
dsanders updated the diff for D77576: [globalisel] Add lost debug locations verifier.

Add missing opcodes that never have locations
Predicate
Early returns

Apr 10 2020, 5:47 PM · Restricted Project