Page MenuHomePhabricator

gbedwell (Greg Bedwell)
User

Projects

User does not belong to any projects.

User Details

User Since
Sep 30 2013, 11:56 PM (302 w, 2 d)

Recent Activity

Mon, Jul 15

gbedwell added a comment to D64589: [UpdateTestChecks] Emit warning when invalid value for -check-prefix(es) option.

I've tried the latest update and it seems to work well for me. The only suggestion I'd make at this stage is optionally adding a message like (did you mean --check-prefixes?) if a comma is spotted in the prefix string.

Mon, Jul 15, 4:59 AM · Restricted Project
gbedwell added inline comments to D64589: [UpdateTestChecks] Emit warning when invalid value for -check-prefix(es) option.
Mon, Jul 15, 3:33 AM · Restricted Project

May 8 2019

gbedwell added a comment to D61680: [X86] Avoid SFB - Fix inconsistent codegen with/without debug info .

Hm, can -debugify be run by llc?

May 8 2019, 7:16 AM · Restricted Project

Apr 18 2019

gbedwell added a comment to D60831: [DebugInfo@O2][LoopVectorize] pr39024: Vectorized code linenos step through loop even after completion.

As the lead of a project building profiling tools, I am strongly against having any instructions map to line 0.

This is probably not what you meant, but for completeness I feel like I should point out that there are many legitimate situations where LLVM generates a line 0 location. The most prominent example is instruction merging: Since both LLVM IR and DWARF currently require each PC address to map to exactly one source location, LLVM's will insert a line 0 location when it merges two instructions with distinct source locations. I can't speak for profiling, but at least on the debugger side, the consensus is that potentially misleading information is worse than no information, because if there is no way to distinguish "always correct" from "maybe correct" information, the user can't trust any information.

Apr 18 2019, 2:57 PM · debug-info, Restricted Project

Mar 21 2019

gbedwell added a comment to D59431: [DebugInfo] Terminate more kinds of location-list ranges at the end of basic blocks.

"at the start of compilation" really means in mem2reg, at least for Clang. As you pointed out, there is no way to preserve the dbg.value in the exit block in mem2reg without introducing an phi node, but doing so would change codegen, which is not an option.

Mar 21 2019, 5:34 AM · debug-info, Restricted Project

Mar 5 2019

gbedwell updated subscribers of D58963: [JumpThreading] Retain debug info when replacing branch instructions.
Mar 5 2019, 6:47 AM · Restricted Project

Mar 1 2019

gbedwell added inline comments to D58817: Add wildcard support to all update_*_test_checks.py scripts (PR37500).
Mar 1 2019, 4:37 AM · Restricted Project

Dec 12 2018

gbedwell accepted D55258: Python 2/3 compat - socket server.

LGTM with one minor nit.

Dec 12 2018, 6:27 AM

Dec 5 2018

gbedwell accepted D55310: [utils] Use operator "in" instead of bound function "has_key".

LGTM.

Dec 5 2018, 9:27 AM

Nov 7 2018

gbedwell added a comment to D54187: Add debuginfo-tests that use cdb on Windows.

+gbedwell

Just to throw the idea out there, why not abandon debuginfo-tests entirely and try using Dexter for this. Dexter's design center is debug-info quality, not correctness, but it already knows how to drive several different debuggers on multiple platforms.

Nov 7 2018, 9:38 AM · Restricted Project, debug-info

Oct 29 2018

gbedwell committed rL345499: [llvm-mca][UpdateTestChecks] Don't try to align blocks that have already been….
[llvm-mca][UpdateTestChecks] Don't try to align blocks that have already been…
Oct 29 2018, 6:26 AM
gbedwell updated subscribers of D53390: [DebugInfo][Dexter] Unreachable line stepped onto after SimplifyCFG..
Oct 29 2018, 5:47 AM · debug-info

Oct 23 2018

gbedwell committed rL345019: [lit] Only return a found bash executable on Windows if it can understand….
[lit] Only return a found bash executable on Windows if it can understand…
Oct 23 2018, 4:36 AM
gbedwell closed D52831: [lit] Only return a found bash executable on Windows if it can understand Windows paths.
Oct 23 2018, 4:36 AM

Oct 18 2018

gbedwell committed rL344766: Update the project name in README.md.
Update the project name in README.md
Oct 18 2018, 5:05 PM

Oct 12 2018

gbedwell abandoned D53073: With CMake 3.8+ force Visual Studio generator builds to use the x64 host toolchain if the user is on a 64-bit system.

Oh drat. Well, I guess maybe this does manage to fall into the area of undefined behaviour then. Weirdly, that didn't happen on my local system, although I did see something similar when setting the PreferredToolArchitecture environment variable for the environment that CMake spawns in, but not in the global environment when running the VS IDE (CMAKE_C[XX]_COMPILER would not match the actual compiler being invoked). Thanks for taking a look anyway!

Oct 12 2018, 8:27 AM

Oct 10 2018

gbedwell updated the diff for D52831: [lit] Only return a found bash executable on Windows if it can understand Windows paths.

Another iteration :)

Oct 10 2018, 7:47 AM
gbedwell added a comment to D53073: With CMake 3.8+ force Visual Studio generator builds to use the x64 host toolchain if the user is on a 64-bit system.

I find the CMake documentation very unclear here. On the related page for https://cmake.org/cmake/help/latest/variable/CMAKE_GENERATOR_TOOLSET.html it says:

Oct 10 2018, 5:19 AM
gbedwell updated the diff for D53073: With CMake 3.8+ force Visual Studio generator builds to use the x64 host toolchain if the user is on a 64-bit system.
Oct 10 2018, 4:55 AM
gbedwell created D53073: With CMake 3.8+ force Visual Studio generator builds to use the x64 host toolchain if the user is on a 64-bit system.
Oct 10 2018, 4:50 AM

Oct 4 2018

gbedwell added a comment to D52831: [lit] Only return a found bash executable on Windows if it can understand Windows paths.
In D52831#1255765, @rnk wrote:

Running cmd.exe is probably even less supported than running bash. If someone is accidentally using execute_external=True, I'd prefer it if they kept using bash and not cmd. So, I'd stick with the first patch.

Oct 4 2018, 3:01 PM
gbedwell committed rL343783: [utils] Ensure that update_mca_test_checks.py writes prefixes in alphabetical….
[utils] Ensure that update_mca_test_checks.py writes prefixes in alphabetical…
Oct 4 2018, 7:46 AM
gbedwell committed rL343782: [utils] simple refactor in update_mca_test_checks.py to make intent more….
[utils] simple refactor in update_mca_test_checks.py to make intent more…
Oct 4 2018, 7:46 AM
gbedwell requested review of D52831: [lit] Only return a found bash executable on Windows if it can understand Windows paths.
Oct 4 2018, 4:03 AM

Oct 3 2018

gbedwell updated the diff for D52831: [lit] Only return a found bash executable on Windows if it can understand Windows paths.

Thanks both for reviewing. Given that it appears that there doesn't, as I'd worried, appear to be some special hidden use-case for bash as an external shell on Windows, but is more of a historical artifact, I'm going to propose an alternative approach. To keep the code all in the same review, and as it's in a completely separate file, I've just added it to this review in the same patch instead. To be clear, I'm talking about committing only one of these files, not both.

Oct 3 2018, 3:40 PM
gbedwell created D52831: [lit] Only return a found bash executable on Windows if it can understand Windows paths.
Oct 3 2018, 8:17 AM

Sep 28 2018

gbedwell committed rL343332: [utils] Stricter checking from update_mca_test_checks.py.
[utils] Stricter checking from update_mca_test_checks.py
Sep 28 2018, 8:41 AM
gbedwell committed rL343333: [utils] Cope with the binary having a .exe extension in update_mca_test_checks..
[utils] Cope with the binary having a .exe extension in update_mca_test_checks.
Sep 28 2018, 8:40 AM
gbedwell closed D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.
Sep 28 2018, 8:40 AM
gbedwell committed rL343331: [utils] Allow better identification of matching blocks in….
[utils] Allow better identification of matching blocks in…
Sep 28 2018, 8:40 AM
gbedwell closed D52560: [utils] Allow better identification of matching blocks in update_mca_test_checks.py.
Sep 28 2018, 8:40 AM

Sep 26 2018

gbedwell updated the diff for D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.

Rebase on top of D52560 which improves the block analysis significantly.

Sep 26 2018, 9:39 AM
gbedwell added a parent revision for D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py: D52560: [utils] Allow better identification of matching blocks in update_mca_test_checks.py.
Sep 26 2018, 9:37 AM
gbedwell added a child revision for D52560: [utils] Allow better identification of matching blocks in update_mca_test_checks.py: D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.
Sep 26 2018, 9:37 AM
gbedwell created D52560: [utils] Allow better identification of matching blocks in update_mca_test_checks.py.
Sep 26 2018, 9:37 AM
gbedwell added a comment to D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.

I've chatted with @andreadb and @RKSimon offline a bit. It would be nice to try and avoid some of the obviously duplicated blocks in the tests. For example, the added blocks in option-all-stats-1.s above. That's not actually an issue related to this specific patch, and the generated tests are still valid, but sometimes look a bit messy.

Sep 26 2018, 4:23 AM

Sep 25 2018

gbedwell added inline comments to D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.
Sep 25 2018, 9:52 AM
gbedwell updated the diff for D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.

Rebased.

Sep 25 2018, 9:47 AM

Aug 17 2018

gbedwell added a reviewer for D50887: [DWARF] Missing location debug information with -O2.: vsk.
Aug 17 2018, 2:36 AM

Aug 2 2018

gbedwell committed rL338761: Add poster and link to slides/poster for "Measuring the User Debugging….
Add poster and link to slides/poster for "Measuring the User Debugging…
Aug 2 2018, 12:40 PM

Jul 18 2018

gbedwell accepted D49003: [Debugify] Export per-pass debug info loss statistics.

Modulo my nit above about the error message text, this LGTM.

Jul 18 2018, 3:44 AM

Jul 17 2018

gbedwell added a comment to D49297: expose debugify as an LLVM option in clang.
In D49297#1165863, @vsk wrote:

First off, @tyb0807, I really appreciate the work you've been doing on LLVM.

With that said, I think this is trying to solve the wrong problem, and I don't think it makes sense to proceed.

Jul 17 2018, 4:18 PM

Jul 13 2018

gbedwell added a comment to D49297: expose debugify as an LLVM option in clang.

Thanks for working on this! I've not reviewed the code, but I did try applying the patch and making use of it, and it seems to work perfectly so far for my use-case of being able to automatically enable automatic debugify checking over large codebases in build systems that I don't necessarily have control of :)

Jul 13 2018, 9:53 AM

Jul 9 2018

gbedwell added a comment to D49003: [Debugify] Export per-pass debug info loss statistics.
In D49003#1154631, @vsk wrote:

At a high-level is it fine to use CSV here? I've found it really convenient because Google Sheets, Numbers, python scripts etc. can ingest the data easily.

Jul 9 2018, 9:05 AM

Jul 6 2018

gbedwell added a comment to D49003: [Debugify] Export per-pass debug info loss statistics.

This is really nice. I agree it will be really useful, albeit with a minor concern about false positives in the stats themselves due to DCE transforms like what we ran into in https://llvm.org/PR37942 .

Jul 6 2018, 8:29 AM

Jun 22 2018

gbedwell added a comment to D28896: [FileCheck] Add directive for checking for blank lines.

This change would be really really useful for addressing http://llvm.org/PR37871 so I'd like to have this.

Jun 22 2018, 10:10 AM

Jun 19 2018

gbedwell added inline comments to D48079: [X86] Add sched class WriteLAHFSAHF and fix values..
Jun 19 2018, 9:32 AM
gbedwell updated subscribers of D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.

Forgot to add llvm-commits

Jun 19 2018, 9:28 AM
gbedwell added inline comments to D48079: [X86] Add sched class WriteLAHFSAHF and fix values..
Jun 19 2018, 9:03 AM

Jun 18 2018

gbedwell updated the diff for D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.
Jun 18 2018, 6:53 AM
gbedwell added a comment to D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.

It's a bit hard to review, the patch misses context (-U99999).

Jun 18 2018, 6:37 AM
gbedwell added inline comments to D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.
Jun 18 2018, 5:35 AM
gbedwell created D48276: [llvm-mca] Stricter checking from update_mca_test_checks.py.
Jun 18 2018, 5:11 AM

Jun 15 2018

gbedwell added inline comments to D48209: [MCA] Add -summary-view option.
Jun 15 2018, 6:41 AM
gbedwell added a comment to D48209: [MCA] Add -summary-view option.

Given that Clement (and presumably Greg) are okay with this change, then I won't oppose it. Sorry for being pedantic.

Jun 15 2018, 6:14 AM
gbedwell added a comment to D48209: [MCA] Add -summary-view option.

Actually make ninja check-llvm-tools-llvm-mca pass, as suggested/noted by @gbedwell

Jun 15 2018, 5:40 AM
gbedwell added inline comments to D48209: [MCA] Add -summary-view option.
Jun 15 2018, 4:44 AM

Jun 6 2018

gbedwell added a comment to D47796: Use the correct project name in README.txt.

How about changing it to "LLVM Compiler Infrastructure". Accepted with that tweak, thanks!

Jun 6 2018, 4:21 AM
gbedwell committed rL334087: Update the project name in README.txt.
Update the project name in README.txt
Jun 6 2018, 4:21 AM
gbedwell closed D47796: Use the correct project name in README.txt.
Jun 6 2018, 4:21 AM

Jun 5 2018

gbedwell created D47796: Use the correct project name in README.txt.
Jun 5 2018, 1:37 PM
gbedwell committed rL334029: [UpdateTestChecks] Error if --llvm-mca-binary gets an empty string.
[UpdateTestChecks] Error if --llvm-mca-binary gets an empty string
Jun 5 2018, 10:20 AM

Jun 4 2018

gbedwell committed rL333894: [llvm-mca][UpdateTestChecks] Prevent an IndexError being raised when given….
[llvm-mca][UpdateTestChecks] Prevent an IndexError being raised when given…
Jun 4 2018, 5:34 AM
gbedwell committed rL333893: [llvm-mca] Regenerate a test to remove a double newline.
[llvm-mca] Regenerate a test to remove a double newline
Jun 4 2018, 5:34 AM
gbedwell accepted D47697: [llvm-mca] Make sure not to end the test files with an empty line..

LGTM. Thanks!

Jun 4 2018, 4:39 AM
gbedwell added inline comments to D47697: [llvm-mca] Make sure not to end the test files with an empty line..
Jun 4 2018, 4:14 AM
gbedwell added a reviewer for D47697: [llvm-mca] Make sure not to end the test files with an empty line.: gbedwell.
Jun 4 2018, 1:22 AM
gbedwell added a comment to D47697: [llvm-mca] Make sure not to end the test files with an empty line..

Thanks for the patch! Annoying that my git client never complained about this one to me.

Jun 4 2018, 1:18 AM

Jun 1 2018

gbedwell accepted D47485: [X86] Help update_llc_test_checks.py to recognise retl/retq to reduce CHECK duplication (PR35003).

LGTM too :)

Jun 1 2018, 6:13 AM

May 29 2018

gbedwell added inline comments to D47485: [X86] Help update_llc_test_checks.py to recognise retl/retq to reduce CHECK duplication (PR35003).
May 29 2018, 10:04 AM

May 24 2018

gbedwell committed rL333204: [UpdateTestChecks] Improved update_mca_test_checks block analysis.
[UpdateTestChecks] Improved update_mca_test_checks block analysis
May 24 2018, 9:40 AM
gbedwell closed D47321: [UpdateTestChecks] Improved update_mca_test_checks block analysis.
May 24 2018, 9:40 AM
gbedwell updated the diff for D47321: [UpdateTestChecks] Improved update_mca_test_checks block analysis.

minor update to formatting to get rid of blank lines around single check lines.

May 24 2018, 7:47 AM
gbedwell created D47321: [UpdateTestChecks] Improved update_mca_test_checks block analysis.
May 24 2018, 6:30 AM

May 17 2018

gbedwell added inline comments to D47015: [llvm-exegesis] Improve documentation..
May 17 2018, 8:03 AM

May 10 2018

gbedwell added inline comments to D46701: [RFC][AArch64] Use the new MCSchedPredicate to rewrite a couple of predicates..
May 10 2018, 3:19 PM

Apr 25 2018

gbedwell committed rL330815: [UpdateTestChecks] Change update_mca_test_checks.py file mode to match the….
[UpdateTestChecks] Change update_mca_test_checks.py file mode to match the…
Apr 25 2018, 4:24 AM

Apr 23 2018

gbedwell added a comment to D40522: Make check-compiler-rt subtarget respect LLVM_LIT_TOOLS_DIR.

Delayed ping. Apologies for the delay. I started paternity leave shortly after starting this review, and never got a chance to come back to it until now.

Apr 23 2018, 4:41 AM

Apr 20 2018

gbedwell committed rL330419: [UpdateTestChecks] Fix update_mca_test_checks.py slowness issue.
[UpdateTestChecks] Fix update_mca_test_checks.py slowness issue
Apr 20 2018, 4:43 AM

Apr 19 2018

gbedwell added a comment to D45787: [llvm-exegesis] Fix PfmIssueCountersTable creation.

I've raised PR37176 to track this

Apr 19 2018, 4:35 AM
gbedwell added a comment to D45787: [llvm-exegesis] Fix PfmIssueCountersTable creation.

What's the more general plan for regression testing exegesis? It looks like the issue with pfm failing to match counters was a regression introduced by rL329675 due to the trailing commas. It would be nice to be able to have regression tests for this sort of thing, especially as it's under active development so the churn is likely to be quite high along with the risk of introducing regressions. I know it gets complicated due to the libpfm dependency and the fact it's CPU specific, but it should still be a solvable problem.

Apr 19 2018, 3:30 AM

Apr 18 2018

gbedwell committed rL330246: [UpdateTestChecks] Add update_mca_test_checks.py script.
[UpdateTestChecks] Add update_mca_test_checks.py script
Apr 18 2018, 3:31 AM
gbedwell closed D45369: [UpdateTestChecks] Add update_mca_test_checks.py script.
Apr 18 2018, 3:31 AM

Apr 13 2018

gbedwell updated the diff for D45369: [UpdateTestChecks] Add update_mca_test_checks.py script.

Rebased and made some minor edits. The stderr output is now a bit more verbose about what's actually been changed and it will now only prefix the advert if it has actually successfully written some check lines to that file.

Apr 13 2018, 9:25 AM

Apr 6 2018

gbedwell created D45369: [UpdateTestChecks] Add update_mca_test_checks.py script.
Apr 6 2018, 7:10 AM
gbedwell added a comment to D45272: [UpdateTestChecks] Add update_analyze_test_checks.py for cost model analysis generation.

Yeah. LGTM too. In general we should probably put some effort into refactoring the scripts so that they work with both Python versions but it seems when the default was discussed a couple of months back ( http://lists.llvm.org/pipermail/llvm-dev/2018-February/thread.html#120907 ) there were still too many OSs in use shipping with only 2.7 by default, so it makes sense to leave this in line with the existing scripts until there can be a coordinated effort to switch.

Apr 6 2018, 4:57 AM

Apr 5 2018

gbedwell added inline comments to D45272: [UpdateTestChecks] Add update_analyze_test_checks.py for cost model analysis generation.
Apr 5 2018, 5:28 AM

Mar 16 2018

gbedwell added inline comments to D44519: Add llvm-exegesis tool..
Mar 16 2018, 8:47 AM

Nov 29 2017

gbedwell committed rL319329: Make check-lit tests respect LLVM_LIT_TOOLS_DIR.
Make check-lit tests respect LLVM_LIT_TOOLS_DIR
Nov 29 2017, 10:06 AM
gbedwell committed rL319330: [cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases….
[cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases…
Nov 29 2017, 10:06 AM
gbedwell closed D40520: Make check-lit tests respect LLVM_LIT_TOOLS_DIR by committing rL319329: Make check-lit tests respect LLVM_LIT_TOOLS_DIR.
Nov 29 2017, 10:06 AM
gbedwell closed D40603: [cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases -Weverything by committing rL319330: [cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases….
Nov 29 2017, 10:05 AM
gbedwell added a comment to D40603: [cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases -Weverything.
In D40603#939278, @rnk wrote:

lgtm

This has the side effect of enabling -Wextra for Windows self-host builds. Are we sure we want to do that?

Nov 29 2017, 9:54 AM
gbedwell updated the diff for D40603: [cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases -Weverything.
Nov 29 2017, 7:33 AM
gbedwell created D40603: [cmake] Replace -Wall with /W4 in clang-cl options now that -Wall aliases -Weverything.
Nov 29 2017, 7:14 AM

Nov 27 2017

gbedwell added a comment to D40520: Make check-lit tests respect LLVM_LIT_TOOLS_DIR.

See D40522 for the compiler-rt part.

Nov 27 2017, 2:16 PM
gbedwell created D40522: Make check-compiler-rt subtarget respect LLVM_LIT_TOOLS_DIR.
Nov 27 2017, 2:15 PM
gbedwell created D40520: Make check-lit tests respect LLVM_LIT_TOOLS_DIR.
Nov 27 2017, 2:09 PM

Nov 13 2017

gbedwell committed rL318037: Allow compiler-rt test targets to work with multi-config CMake generators.
Allow compiler-rt test targets to work with multi-config CMake generators
Nov 13 2017, 4:59 AM
gbedwell closed D38471: Allow compiler-rt test targets to work with multi-config CMake generators by committing rL318037: Allow compiler-rt test targets to work with multi-config CMake generators.
Nov 13 2017, 4:59 AM