Page MenuHomePhabricator

mattd (Matt Davis)
User

Projects

User does not belong to any projects.

User Details

User Since
Nov 16 2017, 6:22 PM (65 w, 5 d)

Recent Activity

Today

mattd accepted D58456: [llvm-readelf]Test a couple of corner-cases for --section-mapping.

LGTM. Thanks for adding these.

Wed, Feb 20, 9:12 AM · Restricted Project
mattd added inline comments to D58416: [llvm-cxxfilt] Split and demangle stdin input on certain non-alphanumerics..
Wed, Feb 20, 8:55 AM · Restricted Project

Yesterday

mattd added inline comments to D58416: [llvm-cxxfilt] Split and demangle stdin input on certain non-alphanumerics..
Tue, Feb 19, 5:17 PM · Restricted Project
mattd added a comment to D58416: [llvm-cxxfilt] Split and demangle stdin input on certain non-alphanumerics..

Previously we would have emitted Foo [[:space:]] . right?

Tue, Feb 19, 4:49 PM · Restricted Project
mattd created D58416: [llvm-cxxfilt] Split and demangle stdin input on certain non-alphanumerics..
Tue, Feb 19, 3:53 PM · Restricted Project

Mon, Feb 18

mattd added a comment to D57149: [StackProtector] Skip analysing dead users in HasAddressTaken, PR40436.

Hi @uabelho

The test case in your patch definitely causes a problem, but that's because the -start-before/-stop-after is used. If I run that same test through llc, the UnreachableBlockElim pass kicks-in and drops the dead code.
I took a look at the code in your Janurary 24th comment. I was unable to get the same dead block to surface, but most likely I had the wrong flag set. How did you build the sample in that comment?

It can be reproduced with the full llc -O1 pipe on current trunk using:

llc -O1 -o - pr40436.ll

Mon, Feb 18, 4:52 PM

Thu, Feb 14

mattd committed rG582274329e3d: [llvm-cxxfilt] Fix a comment typo. NFC. (authored by mattd).
[llvm-cxxfilt] Fix a comment typo. NFC.
Thu, Feb 14, 6:44 PM
mattd committed rL354094: [llvm-cxxfilt] Fix a comment typo. NFC..
[llvm-cxxfilt] Fix a comment typo. NFC.
Thu, Feb 14, 6:43 PM
mattd committed rG123be5d4c0dd: [symbolizer] Avoid collecting symbols belonging to invalid sections. (authored by mattd).
[symbolizer] Avoid collecting symbols belonging to invalid sections.
Thu, Feb 14, 3:54 PM
mattd committed rL354083: [symbolizer] Avoid collecting symbols belonging to invalid sections..
[symbolizer] Avoid collecting symbols belonging to invalid sections.
Thu, Feb 14, 3:54 PM
mattd closed D58146: [symbolizer] Avoid collecting symbols belonging to invalid sections..
Thu, Feb 14, 3:54 PM · Restricted Project

Tue, Feb 12

mattd created D58146: [symbolizer] Avoid collecting symbols belonging to invalid sections..
Tue, Feb 12, 1:45 PM · Restricted Project
mattd accepted D58066: [MCA][Scheduler] Use latency information to further classify busy instructions..
Tue, Feb 12, 9:26 AM · Restricted Project

Mon, Feb 11

mattd committed rG22c21934cefc: [llvm-cxxfilt] Split and demangle stdin input (authored by mattd).
[llvm-cxxfilt] Split and demangle stdin input
Mon, Feb 11, 12:31 PM
mattd committed rL353743: [llvm-cxxfilt] Split and demangle stdin input.
[llvm-cxxfilt] Split and demangle stdin input
Mon, Feb 11, 12:30 PM
mattd closed D57350: [llvm-cxxfilt] Split and demangle stdin input.
Mon, Feb 11, 12:30 PM · Restricted Project
mattd added a comment to D58066: [MCA][Scheduler] Use latency information to further classify busy instructions..

LGTM. I had a few suggestions but nothing looks wrong to me.

Mon, Feb 11, 11:16 AM · Restricted Project

Thu, Feb 7

mattd updated the diff for D57350: [llvm-cxxfilt] Split and demangle stdin input.

I've updated the patch to replicate GNU's c++filt behavior. This patch now only splits strings that are passed via stdin. It does not split strings that are passed as arguments to llvm-cxxfilt via command line.

Thu, Feb 7, 2:02 PM · Restricted Project

Wed, Feb 6

mattd added a comment to D57149: [StackProtector] Skip analysing dead users in HasAddressTaken, PR40436.

The test case in your patch definitely causes a problem, but that's because the -start-before/-stop-after is used. If I run that same test through llc, the UnreachableBlockElim pass kicks-in and drops the dead code.
I took a look at the code in your Janurary 24th comment. I was unable to get the same dead block to surface, but most likely I had the wrong flag set. How did you build the sample in that comment?

Wed, Feb 6, 10:28 AM

Tue, Feb 5

mattd committed rG0d0e9c08a49d: [llvm-readobj] Display sections that do not belong to a segment in the section… (authored by mattd).
[llvm-readobj] Display sections that do not belong to a segment in the section…
Tue, Feb 5, 1:02 PM
mattd committed rLLD353217: [llvm-readobj] Display sections that do not belong to a segment in the section….
[llvm-readobj] Display sections that do not belong to a segment in the section…
Tue, Feb 5, 1:02 PM
mattd committed rL353217: [llvm-readobj] Display sections that do not belong to a segment in the section….
[llvm-readobj] Display sections that do not belong to a segment in the section…
Tue, Feb 5, 1:02 PM
mattd closed D57700: [llvm-readobj] Display sections that do not belong to a segment in the section-mapping.
Tue, Feb 5, 1:02 PM · Restricted Project
mattd added a comment to D57700: [llvm-readobj] Display sections that do not belong to a segment in the section-mapping.

Hello Davide,
The present format is good enough to convey the information. It is however not sufficient or sometimes difficult for present users of GNU tools (which are quite ubiquitous) to make a
switch as it is complete redesign of the post processing tools they may have based on.

-Hemant

It's not immediately clear from the original introduction of GNU style whether complete output identity was the original goal, or just similarity. Similarly, it's not clear from my previously attempt to gather information as to users expectations. I've subscribed @khemant in case he's got any further comments, since he made the original change. My reading of rL260430 which commits it is that it is intended to be like GNU readelf, but doesn't necessarily have to be identical. I'm open to further discussion on this (and I actually have a BoF on this subject at the next Euro LLVM, but that's still a couple of months away).

Tue, Feb 5, 9:40 AM · Restricted Project

Mon, Feb 4

mattd retitled D57700: [llvm-readobj] Display sections that do not belong to a segment in the section-mapping from [llvm-readobj] Display sections that do not belong to a segment in the segment-mapping to [llvm-readobj] Display sections that do not belong to a segment in the section-mapping.
Mon, Feb 4, 10:49 AM · Restricted Project
mattd retitled D57700: [llvm-readobj] Display sections that do not belong to a segment in the section-mapping from [llvm-readobj] Display sections that do not belong to a segment in the section-to-segment dump. to [llvm-readobj] Display sections that do not belong to a segment in the segment-mapping.
Mon, Feb 4, 10:34 AM · Restricted Project
mattd created D57700: [llvm-readobj] Display sections that do not belong to a segment in the section-mapping.
Mon, Feb 4, 10:34 AM · Restricted Project

Fri, Feb 1

mattd committed rL352896: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
[llvm-readobj] Add a flag to dump just the section-to-segment mapping.
Fri, Feb 1, 10:51 AM
mattd closed D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
Fri, Feb 1, 10:51 AM · Restricted Project
mattd added inline comments to D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
Fri, Feb 1, 10:47 AM · Restricted Project
mattd committed rL352891: [llvm-nm] Report '.comment' ELF sections as 'n' instead of '?'.
[llvm-nm] Report '.comment' ELF sections as 'n' instead of '?'
Fri, Feb 1, 9:38 AM
mattd closed D57544: [llvm-nm] Report '.comment' ELF sections as 'n' instead of '?'.
Fri, Feb 1, 9:37 AM · Restricted Project

Thu, Jan 31

mattd created D57544: [llvm-nm] Report '.comment' ELF sections as 'n' instead of '?'.
Thu, Jan 31, 2:34 PM · Restricted Project
mattd added inline comments to D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
Thu, Jan 31, 1:25 PM · Restricted Project
mattd updated the diff for D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
  • Update two tests to use FileChecks --implicit-check-not
  • Remove some cruft that, I think, was in an earlier version of this patch.
  • Update the CHECK-COUNT test to simply perform a CHECK followed by a CHECK-NOT.
Thu, Jan 31, 1:24 PM · Restricted Project
mattd committed rL352785: [ELF] Return the section name when calling getSymbolName on a section symbol..
[ELF] Return the section name when calling getSymbolName on a section symbol.
Thu, Jan 31, 11:42 AM
mattd closed D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..
Thu, Jan 31, 11:42 AM
mattd added a comment to D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..

Okay, the change as-is LGTM, assuming that the following has been done:

I think before putting this in as is, it's probably worth reviewing all the call sites and seeing if there is anywhere where returning the section name would be harmful. I suspect it's probably mostly fine, given the lack of test failures.

A few things to look out for:

  1. Some code paths may already do the extra work needed to look up the section name, if the symbol doesn't have a name. These should be tidied up to not do this any more (it can be a later change).
  2. Some code paths may always print the name returned, but this may be undesirable for certain situations (I can't think of any off the top of my head however). This wouldn't be picked up by tests necessarily, due to the way FileCheck tests tend to be written.
  3. Some code may be using the name in some kind of map, if non-empty. This could cause issues e.g. with relocations in LLD, when attempting to map references to symbols via names. In practice, it shouldn't be an issue, but we should double check.
Thu, Jan 31, 9:41 AM

Wed, Jan 30

mattd updated the diff for D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
  • I took a hint and decided to more the decision to print the section mapping with the program headers down into the dumper classes. This is similar to how the printSymbols and printDynamicSymbols are handled.
Wed, Jan 30, 8:12 PM · Restricted Project
mattd added a comment to D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..

It occurred to me that if we were to implement section mapping for other output styles, e.g. LLVM, or formats (I don't know if that concept applies to COFF/MachO/Wasm etc mind you!), then we probably don't want to print the section mapping when program-headers is specified, except in GNU output style. This implies that the decision on whether to print the section mapping should be down to the dumper layer, not the command-line handling layer. This is very similar to the situation with --symbols/--dyn-symbols for GNU output style, and I suspect the solution should be essentially the same as rL351960.

Wed, Jan 30, 2:54 PM · Restricted Project
mattd updated the diff for D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
  • Removed a test
  • Replaced the uses of grep in the tests with CHECK-NOT and CHECK-COUNT
Wed, Jan 30, 2:50 PM · Restricted Project
mattd updated the diff for D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..
  • Update a comment, it was bothering me.
Wed, Jan 30, 10:18 AM
mattd updated the diff for D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..
  • Updated the test to also look at the presentation of other sections/symbols, not just the debug sections.
  • Replaced the use of an auto with the actual type.
  • Rebased against master.
Wed, Jan 30, 10:09 AM
mattd added inline comments to D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
Wed, Jan 30, 9:28 AM · Restricted Project

Tue, Jan 29

mattd updated the diff for D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..

Thanks for the feedback, and my apologies for not originally providing as thorough of
testing as I should have.

Tue, Jan 29, 6:18 PM · Restricted Project
mattd updated the diff for D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
  • Moved the call to printSectionMapping after printSectionHeaders`
  • Specifying --program-headers will still print the section mapping; however, the calls to printSectionMapping has been removed from the body of printProgramHeaders
Tue, Jan 29, 12:52 PM · Restricted Project
mattd added inline comments to D57350: [llvm-cxxfilt] Split and demangle stdin input.
Tue, Jan 29, 9:34 AM · Restricted Project
mattd added inline comments to D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
Tue, Jan 29, 9:32 AM · Restricted Project
mattd added a comment to D57350: [llvm-cxxfilt] Split and demangle stdin input.

This actually breaks the compatibility with the existing tools. The quoted string is supposed to be interpreted as a *single* entry. I think that if you want to support something like this, it should be a separate flag for this behaviour.

Tue, Jan 29, 9:07 AM · Restricted Project

Mon, Jan 28

mattd created D57365: [llvm-readobj] Add a flag to dump just the section-to-segment mapping..
Mon, Jan 28, 5:19 PM · Restricted Project
mattd created D57350: [llvm-cxxfilt] Split and demangle stdin input.
Mon, Jan 28, 11:51 AM · Restricted Project

Fri, Jan 25

mattd added a comment to D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..

My instinct is that the place you've made the change is too deep in the hierarchy, and that's why you have the problem in llvm-objdump. My feeling is that Symbol::getName() should probably return the name of the symbol, and leave it up to the caller to fall back and get the section name for section symbols.

Fri, Jan 25, 8:32 AM

Thu, Jan 24

mattd requested review of D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..
Thu, Jan 24, 11:54 AM
mattd updated the diff for D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..
  • Added a comment to the change in ELFObjectFile.h.
  • Updated the condition to retrieve the section name if the symbol is a section and has no name.
  • I discovered that my original patch broke an aarch64 test. In that case, llvm-objdump was using the section name (produced from my patch) instead of ignoring the section in general. That had the effect of the section name being displayed in the disassembly instead of the aarch64/arm mapping name "$x." This patch now ignores section names and preserves the mapping name in that case. I'm not sure that the change to llvm-objdump is the best choice, but it doesn't seem to break any llvm tests or any test I generated while fiddling around with this patch.
Thu, Jan 24, 11:54 AM
mattd added a comment to D57149: [StackProtector] Skip analysing dead users in HasAddressTaken, PR40436.

I like this patch, and I like the DominatorTree clean up/refactor. As you pointed out, your patch does make the DT required, which makes sense because your solution must make use of that analysis. I don't see that as an issue. Overall, this patch makes sense to me, but I'm curious as to what others think.

Thu, Jan 24, 11:06 AM

Wed, Jan 23

mattd planned changes to D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..

I have a few changes to this patch. Namely, I want to avoid breaking the llvm-objdump/AArch64/elf-aarch64-mapping-symbols.test. I didn't realize I had broken that test earlier, but I caught it in my second round of testing. It turns out that llvm-objdump presents a data mapping symbol ($x) when the test works. However, with the patch above, that mapping symbol is gone and the section name is displayed instead. I'm working on fixing my patch to avoid that situation.

Wed, Jan 23, 4:13 PM
mattd created D57105: [ELF] Return the section name when calling getSymbolName on a section symbol..
Wed, Jan 23, 9:57 AM

Tue, Jan 22

mattd added a comment to D57056: [MC][X86] Correctly model additional operand latency caused by transfer delays from the integer to the floating point unit..

LGTM. I'd wait a bit to see if others have anything else to add.

Tue, Jan 22, 10:11 AM

Jan 10 2019

mattd added a comment to D55039: [sema] Warn of mangling change if function parameters are noexcept..

*Ping* to see if anyone else has any feedback towards this patch.

Jan 10 2019, 3:25 PM
mattd committed rL350880: [GVN] Update BlockRPONumber prior to use..
[GVN] Update BlockRPONumber prior to use.
Jan 10 2019, 11:59 AM
mattd closed D55974: [GVN] Update BlockRPONumber prior to use..
Jan 10 2019, 11:59 AM

Jan 9 2019

mattd updated the diff for D55974: [GVN] Update BlockRPONumber prior to use..
  • Moved the logic for controlling access to performPRE/performScalarPRE functions into a separate patch.
  • Added an assertion that will trigger on 9 tests if the BlockRPONumber map is invalid. In particular, test/Transforms/GVN/nonescaping-malloc.ll, which also requires an assert build when testing.
Jan 9 2019, 4:35 PM
mattd created D56515: [GVN] Check the EnablePRE flag prior to calling performScalarPRE.
Jan 9 2019, 3:04 PM

Jan 8 2019

mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..

Update the SelectionDAG handling of the mca_code_region_start and mca_code_region_end intrinsics.

Jan 8 2019, 4:09 PM
mattd updated the diff for D55974: [GVN] Update BlockRPONumber prior to use..
  • Add a bool to GVN that tracks the status (valid or invalid) of the BlockRPONumber map.
  • Use that flag to indicate that the BlockRPONumber is invalid when blocks are added to the function being analyzed.
  • Make the BlockRPONumber map: DenseMap<AssertingVH<BasicBlock>, uint32_t>
Jan 8 2019, 1:07 PM

Jan 7 2019

mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Rebased against master
  • Moved most of the added llvm-mca CodeGen/X86 tests to CodeGen/Generic.
Jan 7 2019, 9:08 AM

Jan 4 2019

mattd added inline comments to D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
Jan 4 2019, 3:58 PM
mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Update the symbol name parsing for the llvm-mca region markers.
Jan 4 2019, 3:54 PM

Jan 3 2019

mattd retitled D55974: [GVN] Update BlockRPONumber prior to use. from [GVN] Initialize BlockRPONumber when prior to use. to [GVN] Update BlockRPONumber prior to use..
Jan 3 2019, 4:18 PM
mattd updated the diff for D55974: [GVN] Update BlockRPONumber prior to use..

This patch aims to clean-up two items in GVN.

Jan 3 2019, 4:18 PM

Jan 2 2019

mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Renamed the assembly comment that gets added to assembly files when using the llvm.mca.code.region.start intrinsic.
Jan 2 2019, 11:59 AM
mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Removed the check in IR/Verifier that rejected programs if llvm-mca code regions span multiple blocks.

In short, we let llvm-mca handle multiple blocks as it always has. To be fair here, llvm-mca doesn't handle branch instructions, but a user can currently place LLVM-MCA assembly comments such that the instructions cross multiple blocks.

Jan 2 2019, 11:39 AM

Dec 21 2018

mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Corrected some comments and removed unnecessary code, as a result of the last two patches.
  • Rebased
Dec 21 2018, 4:10 PM
mattd added a comment to D55974: [GVN] Update BlockRPONumber prior to use..

Thanks for the feedback @efriedma.

Dec 21 2018, 9:18 AM

Dec 20 2018

mattd created D55974: [GVN] Update BlockRPONumber prior to use..
Dec 20 2018, 5:43 PM

Dec 19 2018

mattd committed rL349703: [DwarfExpression] Fix a typo in a doxygen comment. NFC..
[DwarfExpression] Fix a typo in a doxygen comment. NFC.
Dec 19 2018, 4:05 PM
mattd committed rL349662: [llvm-mca] Rename an error variable..
[llvm-mca] Rename an error variable.
Dec 19 2018, 11:00 AM
mattd committed rL349659: [llvm-mca] Add an error handler for error from parseCodeRegions.
[llvm-mca] Add an error handler for error from parseCodeRegions
Dec 19 2018, 10:30 AM
mattd closed D55803: [llvm-mca] Add an error handler for error from parseCodeRegions.
Dec 19 2018, 10:30 AM

Dec 18 2018

mattd added a comment to D55039: [sema] Warn of mangling change if function parameters are noexcept..

Apart from the comments I think your patch LGTM, but I'll let someone else hav the final say.

Dec 18 2018, 8:05 AM

Dec 17 2018

mattd added a comment to D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Following some discussion of the RFC on the mailing list, this patch makes a few improvements:
  • Object files and executables are supported. This is accomplished by scanning the symbol table for mca_code_region_start and mca_code_region_end symbols.
  • This solution does not rely on target specific relocations.
  • Regions cannot be nested, so a start of region label/symbol must be followed by an end (this has always been the case).
  • Symbol names are encoded with the user's defined region number." That number is just for cosmetic purposes and is only helpful for the user, llvm-mca can make use of that number to annotate its analysis reports.
Dec 17 2018, 8:24 PM
mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Following some discussion of the RFC on the mailing list, this patch makes a few improvements:
  • Object files and executables are supported. This is accomplished by scanning the symbol table for mca_code_region_start and mca_code_region_end symbols.
  • This solution does not rely on target specific relocations.
  • Regions cannot be nested, so a start of region label/symbol must be followed by an end (this has always been the case).
  • Symbol names are encoded with the user's defined region number." That number is just for cosmetic purposes and is only helpful for the user, llvm-mca can make use of that number to annotate its analysis reports.
Dec 17 2018, 8:23 PM
mattd updated the diff for D55803: [llvm-mca] Add an error handler for error from parseCodeRegions.
  • Added context
Dec 17 2018, 8:15 PM
mattd created D55803: [llvm-mca] Add an error handler for error from parseCodeRegions.
Dec 17 2018, 6:22 PM

Dec 13 2018

mattd accepted D55557: [llvm-mca] Move llvm-mca library to llvm/lib/MCA..
Dec 13 2018, 8:18 AM

Dec 12 2018

mattd added inline comments to D55039: [sema] Warn of mangling change if function parameters are noexcept..
Dec 12 2018, 11:24 AM
mattd added a comment to D55557: [llvm-mca] Move llvm-mca library to llvm/lib/MCA..

Thanks @courbet , these changes LGTM. I'd wait a day or two to see if anyone else has any feedback from the RFC on the mailinglist:
https://lists.llvm.org/pipermail/llvm-dev/2018-December/128435.html

Dec 12 2018, 8:33 AM

Dec 5 2018

mattd added a comment to D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..

Just pinging this patch and RFC to see if anyone else, aside from Andrea, has any feedback. The discussion on the RFC is here:
https://lists.llvm.org/pipermail/llvm-dev/2018-December/128218.html

Dec 5 2018, 5:08 PM

Dec 4 2018

mattd added inline comments to D55039: [sema] Warn of mangling change if function parameters are noexcept..
Dec 4 2018, 9:48 AM

Dec 3 2018

mattd updated the diff for D55039: [sema] Warn of mangling change if function parameters are noexcept..
  • Move the HasNoexcept lambda to its own static function.
  • Added an additional check in HasNoexcept to recursively check return values.
  • Modified the parameter check in Sema::CheckFunctionDeclaration to use llvm::any_of, all we need to know is if any of the parameters might have a mangling change.
  • Updated the test to include the example provided by @Rakete1111.
Dec 3 2018, 4:51 PM
mattd accepted D53832: Make instrprof-set-dir-mode test tolerant of group ID.

@troyj, good catch! Thanks, and LGTM.

Dec 3 2018, 10:55 AM

Nov 29 2018

mattd added a comment to D55039: [sema] Warn of mangling change if function parameters are noexcept..

Thanks for the reviews and added example. I'll update the logic to account for the return type, and add @Rakete1111 's example to the existing test.

Nov 29 2018, 3:07 PM

Nov 28 2018

mattd created D55039: [sema] Warn of mangling change if function parameters are noexcept..
Nov 28 2018, 5:08 PM
mattd accepted D54957: [llvm-mca][MC] Add the ability to declare which processor resources model load/store queues (PR36666)..

Thanks for answering my questions.

Nov 28 2018, 11:17 AM
mattd accepted D55000: [llvm-mca] Return the total number of cycles from method Pipeline::run()..

LGTM!

Nov 28 2018, 8:14 AM

Nov 27 2018

mattd added inline comments to D54957: [llvm-mca][MC] Add the ability to declare which processor resources model load/store queues (PR36666)..
Nov 27 2018, 10:50 AM
mattd added inline comments to D54957: [llvm-mca][MC] Add the ability to declare which processor resources model load/store queues (PR36666)..
Nov 27 2018, 10:48 AM
mattd added a comment to D54957: [llvm-mca][MC] Add the ability to declare which processor resources model load/store queues (PR36666)..

LGTM, I'll let others weigh in. For the most part I think this patch is fine, as long as my questions do not point out any bugs.

Nov 27 2018, 10:34 AM
mattd added inline comments to D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
Nov 27 2018, 8:38 AM

Nov 26 2018

mattd updated the diff for D54603: [llvm-mca][RFC] Adding binary support to llvm-mca..
  • Rebased this patch.
  • Added an additional IRVerifier check to generate a compiler error if a llvm-mca code region belongs to more than one basic-block (this was discussed in the llvm-dev mailing list thread).
Nov 26 2018, 3:47 PM