Page MenuHomePhabricator

steven.zhang (Qing Shan Zhang)
User

Projects

User does not belong to any projects.

User Details

User Since
May 15 2018, 2:45 AM (253 w, 5 d)

Compiler Engineer worked for IBM before, and now, for ByteDance

Recent Activity

Thu, Mar 16

steven.zhang added a comment to D144565: dwp check overflow.

We intend this for llvm release/11.x version. Perhaps this can go as a backport?

Thu, Mar 16, 5:08 AM · Restricted Project, Restricted Project

Feb 22 2023

steven.zhang added inline comments to D144565: dwp check overflow.
Feb 22 2023, 6:56 PM · Restricted Project, Restricted Project
steven.zhang added a comment to D144565: dwp check overflow.

I didn't see any tests in the PR. Please add one.

Feb 22 2023, 6:54 PM · Restricted Project, Restricted Project

Oct 3 2022

steven.zhang added inline comments to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Oct 3 2022, 5:15 PM · Restricted Project, Restricted Project

Sep 13 2022

steven.zhang added a comment to D133549: [llvm-dwp] Report the filename if it cannot be found.

I have re-commit as suggested. Thank you for pointing out the testing issue for windows.

Sep 13 2022, 8:34 PM · Restricted Project, Restricted Project
steven.zhang committed rG734843ebc7c3: [llvm-dwp] Report the filename if it cannot be found (authored by steven.zhang).
[llvm-dwp] Report the filename if it cannot be found
Sep 13 2022, 8:34 PM · Restricted Project, Restricted Project
steven.zhang added inline comments to D133549: [llvm-dwp] Report the filename if it cannot be found.
Sep 13 2022, 8:28 PM · Restricted Project, Restricted Project

Sep 12 2022

steven.zhang committed rG4531f5385125: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid (authored by steven.zhang).
[llvm-dwp] Get the DWO file from relative path if the absolute path is not valid
Sep 12 2022, 10:46 PM · Restricted Project, Restricted Project
steven.zhang closed D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 12 2022, 10:46 PM · Restricted Project, Restricted Project
steven.zhang added inline comments to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 12 2022, 10:43 PM · Restricted Project, Restricted Project
steven.zhang committed rG35028d417bb3: [llvm-dwp] Report the filename if it cannot be found (authored by steven.zhang).
[llvm-dwp] Report the filename if it cannot be found
Sep 12 2022, 10:00 PM · Restricted Project, Restricted Project
steven.zhang closed D133549: [llvm-dwp] Report the filename if it cannot be found.
Sep 12 2022, 10:00 PM · Restricted Project, Restricted Project

Sep 9 2022

steven.zhang added a reviewer for D133549: [llvm-dwp] Report the filename if it cannot be found: MTC.
Sep 9 2022, 6:01 PM · Restricted Project, Restricted Project
steven.zhang added a reviewer for D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid: MTC.
Sep 9 2022, 6:01 PM · Restricted Project, Restricted Project
steven.zhang updated the diff for D133549: [llvm-dwp] Report the filename if it cannot be found.

Add tests.

Sep 9 2022, 5:52 PM · Restricted Project, Restricted Project
steven.zhang updated the diff for D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 9 2022, 5:47 PM · Restricted Project, Restricted Project
steven.zhang updated the diff for D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 9 2022, 5:44 PM · Restricted Project, Restricted Project
steven.zhang retitled D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid from [llvm-dwp] Get the DWO file using relative path instead of absolute path to make it work for distribution build to [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 9 2022, 5:07 PM · Restricted Project, Restricted Project
steven.zhang updated the diff for D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

Add tests.

Sep 9 2022, 5:06 PM · Restricted Project, Restricted Project
steven.zhang added a comment to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

fair enough, if this is gold/binutils dwp's behavior (can you provide a pastebin/show this is the behavior there, maybe even point to the code that implements it?) doesn't seem like the worst thing to have in llvm-dwp, though this will need a test case

Sep 9 2022, 4:17 PM · Restricted Project, Restricted Project

Sep 8 2022

steven.zhang requested review of D133549: [llvm-dwp] Report the filename if it cannot be found.
Sep 8 2022, 7:34 PM · Restricted Project, Restricted Project
steven.zhang added a comment to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

I don't think this is correct - the dwp tool might be run from a different directory than the equivalent comp_dir and so treating these as cwd-of-dwp-relative paths seems problematic.

I think the correct thing to do in the situation described is for the compilation to use -fdebug-compilation-dir to make the debug info portable - this will also ensure that a debugger can find the source and find the .dwo files even if they aren't packaged into a .dwp.

Can we do it as what gdb did ? That is, getting the dwo from absolute path first, and if failed, get it from relative path ?

I'm somewhat open to that, though I'd really encourage you to switch your build system/build flags so you don't have to keep adding functionality like that into other debug info consuming tools.

Thank for the suggestion, and I appreciate it. The issue of adding the build flags is that, it will pollute the remote cache. For example:

  • building in workspace A: adding compilation option -fdebug-compilation-dir=path_to_workspace_A, and the objects will be cached in the remote build server.
  • building in workspace B: adding compilation option -fdebug-compilation-dir=path_to_workspace_B, we cannot get the objects from remote cache though all the source code didn't change. We have to rebuild it as the compilation option changed.

If with relative path, we don't need to rebuild the objects as everything is unchanged. Remote cache is important to large project to speed up the building.

I will extend current implementation to support relative path.

Ah, sorry I didn't fully explain - yes (for context: I work at Google/on the compiler we use in the internal version of Bazel - so I'm familiar with the complexities of distributed and reusable build caches). Specifically, if the comp_dir is basically unusable (because if it's accurate/absolute for a local user, it's useless to another user, and it can't be whatever's on the build node either) you can use -fdebug-compilation-dir=. (I think at Google we use /proc/cwd to be precise) and then you'll get the current-working-directory relative behavior without consumers having to implement any special case.

Sep 8 2022, 7:30 PM · Restricted Project, Restricted Project
steven.zhang updated the diff for D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 8 2022, 4:23 PM · Restricted Project, Restricted Project
steven.zhang updated the diff for D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

Address comments.

Sep 8 2022, 4:20 PM · Restricted Project, Restricted Project
steven.zhang added a comment to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

I don't think this is correct - the dwp tool might be run from a different directory than the equivalent comp_dir and so treating these as cwd-of-dwp-relative paths seems problematic.

I think the correct thing to do in the situation described is for the compilation to use -fdebug-compilation-dir to make the debug info portable - this will also ensure that a debugger can find the source and find the .dwo files even if they aren't packaged into a .dwp.

Can we do it as what gdb did ? That is, getting the dwo from absolute path first, and if failed, get it from relative path ?

I'm somewhat open to that, though I'd really encourage you to switch your build system/build flags so you don't have to keep adding functionality like that into other debug info consuming tools.

Thank for the suggestion, and I appreciate it. The issue of adding the build flags is that, it will pollute the remote cache. For example:

Sep 8 2022, 4:16 PM · Restricted Project, Restricted Project
steven.zhang added a comment to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

I don't think this is correct - the dwp tool might be run from a different directory than the equivalent comp_dir and so treating these as cwd-of-dwp-relative paths seems problematic.

I think the correct thing to do in the situation described is for the compilation to use -fdebug-compilation-dir to make the debug info portable - this will also ensure that a debugger can find the source and find the .dwo files even if they aren't packaged into a .dwp.

Sep 8 2022, 8:01 AM · Restricted Project, Restricted Project
steven.zhang added a comment to D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.

I don't think this is correct - the dwp tool might be run from a different directory than the equivalent comp_dir and so treating these as cwd-of-dwp-relative paths seems problematic.

I think the correct thing to do in the situation described is for the compilation to use -fdebug-compilation-dir to make the debug info portable - this will also ensure that a debugger can find the source and find the .dwo files even if they aren't packaged into a .dwp.

Sep 8 2022, 7:54 AM · Restricted Project, Restricted Project
steven.zhang updated the summary of D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 8 2022, 2:55 AM · Restricted Project, Restricted Project
steven.zhang updated the summary of D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 8 2022, 2:54 AM · Restricted Project, Restricted Project
steven.zhang updated the summary of D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 8 2022, 2:53 AM · Restricted Project, Restricted Project
steven.zhang requested review of D133480: [llvm-dwp] Get the DWO file from relative path if the absolute path is not valid.
Sep 8 2022, 2:49 AM · Restricted Project, Restricted Project

Aug 24 2022

steven.zhang added a comment to D131771: [RISCV] : Add support for immediate operands..

I do not have commit access.
Someone who has it, please, push this commit to upstream.

Aug 24 2022, 2:49 AM · Restricted Project, Restricted Project
steven.zhang committed rG8c1f18bd3ee1: [RISCV] : Add support for immediate operands. (authored by MarkGoncharovAl).
[RISCV] : Add support for immediate operands.
Aug 24 2022, 2:49 AM · Restricted Project, Restricted Project
steven.zhang closed D131771: [RISCV] : Add support for immediate operands..
Aug 24 2022, 2:48 AM · Restricted Project, Restricted Project

Aug 2 2021

steven.zhang added a comment to D91050: [NFC] Add the EmitTargetCodeForConstantPool hook for target to customize it with MachineConstantPoolValue.

Passing an alignment in and returning an alignment out at the same time seems unlikely to be useful. DAGCombine can refine the alignment later anyway, if it turns out to be relevant. Maybe rename the version that returns the alignment out?

Passing a constant in,which has the alignment of the constant,return a constantpool,which need to specify the alignment of constantpool. We cannot query the alignment of constantpool now as the result could be cp + offset.

I think you're missing my point. If the caller has a particular idea of what alignment it needs, it will request it. The result will have at least that alignment. If the actual alignment comes out higher, it doesn't really matter.

Aug 2 2021, 1:49 AM · Restricted Project

Jul 29 2021

steven.zhang added a comment to D91053: [PowerPC] Lump the constants to save one addis for each constant access.

A few months ago, steven shared his patch to us. For me, it should be an optimization of TOC layout, it looks odd to me to get it done at isel phase. I think we should figure out a new approach for this. What do you think @nemanjai @qiucf ?

I tend to not pursue this patch and might have a new one to get TOC layout optimization done.

I think it's located at DAG because we lower such constant pools stuff there. If this can be target independent (as Eli commented in previous patch) in DAG, or put it after ISel (I guess you mean that?), it's fine to abandon that.

I don't think after ISEL is a good idea. After ISel, the constant pool and the code pattern to access the constant pool are both settled down. It does not make sense to me to revert or change them later. It is better to generate it like we expect in the first place.

+1

Jul 29 2021, 5:50 PM · Restricted Project, Restricted Project
steven.zhang added a comment to D91050: [NFC] Add the EmitTargetCodeForConstantPool hook for target to customize it with MachineConstantPoolValue.

Passing an alignment in and returning an alignment out at the same time seems unlikely to be useful. DAGCombine can refine the alignment later anyway, if it turns out to be relevant. Maybe rename the version that returns the alignment out?

Also, if the point is literally just "glue a bunch of ConstantPools together", can we teach target-independent code to do that? It doesn't actually require anything target-specific, I think, and it should be useful on any target where computing the address of a constant pool isn't cheap.

Jul 29 2021, 5:49 PM · Restricted Project

Jul 27 2021

steven.zhang added a comment to D106431: [clang-tidy] Fix cppcoreguidelines-init-variables by removing the enum FixIt, and add support for initialization check of scoped enum..

Any thoughts? : )

First, let's first fix that we should still warn for the uninitialised enum case, without a FixIt. That's the issue at hand, right now, Clang-Tidy generates, as you identified, broken output. We can discuss the later steps after this is fixed. Please implement this logic, and update the patch, so we have a snapshot of how that would look like and the thing working.

+1

Jul 27 2021, 7:27 PM · Restricted Project

Jul 21 2021

steven.zhang added a comment to D106431: [clang-tidy] Fix cppcoreguidelines-init-variables by removing the enum FixIt, and add support for initialization check of scoped enum..

Is this the right decision to make, conceptually? It will leave the variable uninitialised still, and reading such an uninit variable is still an issue, even if it is an enum.

Yeah, that's right. However, it's much more difficult to give enum an initial value than an integer.

Could we consider the alternative of warning the user about the uninitialized variable, just not offering an automatic (and potentially bad / incorrect) fix?

Make sense, we (ByteDance) are also hesitating whether we should provide automatic repair for uninitialized variables, because automatic fix may change the program semantics.

Jul 21 2021, 5:03 AM · Restricted Project
steven.zhang added a reviewer for D106431: [clang-tidy] Fix cppcoreguidelines-init-variables by removing the enum FixIt, and add support for initialization check of scoped enum.: whisperity.
Jul 21 2021, 5:00 AM · Restricted Project
steven.zhang added inline comments to D106431: [clang-tidy] Fix cppcoreguidelines-init-variables by removing the enum FixIt, and add support for initialization check of scoped enum..
Jul 21 2021, 4:58 AM · Restricted Project

May 17 2021

steven.zhang added a reviewer for D101759: [PowerPC] Scalar IBM MASS library conversion pass: Restricted Project.
May 17 2021, 10:24 PM · Restricted Project, Restricted Project, Restricted Project

Apr 11 2021

steven.zhang committed rGd69c236e1d6b: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time (authored by steven.zhang).
[NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time
Apr 11 2021, 11:58 PM
steven.zhang closed D100162: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time .
Apr 11 2021, 11:58 PM · Restricted Project
steven.zhang added a comment to D100162: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time .

Does this solve the problem?

Apr 11 2021, 11:29 PM · Restricted Project
steven.zhang accepted D92374: [PowerPC] Support f128 under VSX.

LGTM. But please do a complete test before enabling it by default.

Apr 11 2021, 7:41 PM · Restricted Project
steven.zhang committed rG37388b8ad13f: Update personal info in CREDITS.TXT (authored by steven.zhang).
Update personal info in CREDITS.TXT
Apr 11 2021, 4:26 AM
steven.zhang updated steven.zhang.
Apr 11 2021, 4:19 AM
steven.zhang added a comment to D100162: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time .

Hi, @djtodoro, do you have any concern on this ? If not, I will land this patch days later.

Apr 11 2021, 3:05 AM · Restricted Project

Apr 8 2021

steven.zhang retitled D100162: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time from [Debug] Fix unnecessary deep-copy for vecto to save compiling time to [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time .
Apr 8 2021, 10:00 PM · Restricted Project
steven.zhang updated the diff for D100162: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time .
Apr 8 2021, 9:04 PM · Restricted Project
steven.zhang requested review of D100162: [NFC][Debug] Fix unnecessary deep-copy for vector to save compiling time .
Apr 8 2021, 9:02 PM · Restricted Project
steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

Sorry, I don't mean to accept as PowerPC now ...

Apr 8 2021, 3:48 AM · Restricted Project
steven.zhang accepted D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

LGTM. But please hold on for some days to see if @nemanjai has more comments.

Apr 8 2021, 3:48 AM · Restricted Project

Mar 30 2021

steven.zhang added inline comments to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.
Mar 30 2021, 7:06 PM · Restricted Project
steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.
Mar 30 2021, 7:05 PM · Restricted Project

Mar 3 2021

steven.zhang accepted D92815: [PowerPC] [Clang] Enable float128 feature on VSX targets.

LGTM now. But please hold on for at least one week to see if @nemanjai has concern.

Mar 3 2021, 11:32 PM · Restricted Project

Mar 2 2021

steven.zhang added a comment to D86183: [DAGCombine] Remove dead node when it is created by getNegatedExpression.

It seems that, craig is working on this bugzallia issue. Thank you for reporting this.

Mar 2 2021, 8:42 PM · Restricted Project

Jan 24 2021

steven.zhang committed rGffc3e800c65e: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero (authored by steven.zhang).
[NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero
Jan 24 2021, 8:03 PM
steven.zhang closed D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.
Jan 24 2021, 8:03 PM · Restricted Project

Jan 21 2021

steven.zhang added a comment to D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.

The bif/bic difference seems logically fine and probably better overall since we're using less registers, but someone who knows AArch should comment.
Also, it would be interesting to know why the code changed because we are seemingly producing the same set of SDNode ops? There might be another combine/lowering opportunity.

Jan 21 2021, 7:12 PM · Restricted Project
steven.zhang updated the diff for D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.
Jan 21 2021, 7:08 PM · Restricted Project

Jan 20 2021

steven.zhang updated the diff for D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

Split the BR_CC into another patch which will enhance the legalizer to support it. For SETCC/SELECT_CC, we will do it inside PowerPC as what other target did.

Jan 20 2021, 9:41 PM · Restricted Project
steven.zhang retitled D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled from [PowerPC] Lower the SETCC/SELECT_CC/BR_CC as libcall for fp128 with Power9 vector disabled to [PowerPC] Lower the SETCC/SELECT_CC as libcall for fp128 with Power9 vector disabled.
Jan 20 2021, 9:36 PM · Restricted Project
steven.zhang updated the diff for D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.

Address comments.

Jan 20 2021, 9:16 PM · Restricted Project
steven.zhang accepted D94464: [PowerPC] Duplicate inherited heuristic from base scheduler.

LGTM.

Jan 20 2021, 12:21 AM · Restricted Project

Jan 19 2021

steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

The description isn't really adequate - it does not answer my question at all really. The question is why not set the legalization action for BR_CC as Expand? Namely, what does the legalizer do that we don't want it to do if we let it expand the node? You just answered the question with the equivalent of "The legalizer does its job and that's not what we want." So please explain why that is not what we want.

Jan 19 2021, 1:52 AM · Restricted Project

Jan 18 2021

steven.zhang added a comment to D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.

Gentle ping...

Jan 18 2021, 4:34 PM · Restricted Project
steven.zhang added reviewers for D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero: RKSimon, dmgreen.
Jan 18 2021, 4:34 PM · Restricted Project

Jan 17 2021

steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

@nemanjai Do you have any more comments on this ? Thank you!

Jan 17 2021, 7:55 PM · Restricted Project
steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

Ping...

Jan 17 2021, 4:22 PM · Restricted Project

Jan 12 2021

steven.zhang abandoned D94462: [PowerPC][AIX] Support calling convention for IEEE Long Double.
Jan 12 2021, 5:47 PM · Restricted Project
steven.zhang updated the summary of D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.
Jan 12 2021, 2:27 AM · Restricted Project
steven.zhang requested review of D94480: [NFC] [DAGCombine] Correct the result for sqrt even the iteration is zero.
Jan 12 2021, 2:12 AM · Restricted Project
steven.zhang accepted D94159: [PowerPC] [NFC] Add AIX triple to some regression tests.

Thank you for doing this. LGTM now.

Jan 12 2021, 1:52 AM · Restricted Project

Jan 11 2021

steven.zhang added a comment to D94464: [PowerPC] Duplicate inherited heuristic from base scheduler.

This is the same as what I want to do in D86681. As D86684 is still pending, it makes sense to split from it now.

Jan 11 2021, 11:14 PM · Restricted Project
steven.zhang updated the summary of D94462: [PowerPC][AIX] Support calling convention for IEEE Long Double.
Jan 11 2021, 6:14 PM · Restricted Project
steven.zhang requested review of D94462: [PowerPC][AIX] Support calling convention for IEEE Long Double.
Jan 11 2021, 6:12 PM · Restricted Project
steven.zhang accepted D94054: [PowerPC] Try to fold sqrt/sdiv test results with the branch..
Jan 11 2021, 6:03 PM · Restricted Project
steven.zhang added a comment to D94054: [PowerPC] Try to fold sqrt/sdiv test results with the branch..

LGTM. Please hold on for days to see if @nemanjai has more comments.

Jan 11 2021, 6:03 PM · Restricted Project

Jan 10 2021

steven.zhang updated the diff for D86684: [Refactor] Add the SchedHeuristic for Scheduler to allow platform customizing the heuristics.

Rebase the patch.

Jan 10 2021, 10:44 PM · Restricted Project
steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

Ping ...

Jan 10 2021, 8:10 PM · Restricted Project
steven.zhang accepted D92473: [Legalizer] Promote result type in expanding FP_TO_XINT.

LGTM, but please hold on for days to see if @craig.topper or @nemanjai have more comments.

Jan 10 2021, 7:50 PM · Restricted Project
steven.zhang committed rG7539c75bb438: [DAGCombine] Remove the check for unsafe-fp-math when we are checking the AFN (authored by steven.zhang).
[DAGCombine] Remove the check for unsafe-fp-math when we are checking the AFN
Jan 10 2021, 6:28 PM
steven.zhang closed D93891: [DAGCombine] Remove the check for unsafe-fp-math when we are checking the AFN.
Jan 10 2021, 6:28 PM · Restricted Project

Jan 7 2021

steven.zhang added inline comments to D94282: [PowerPC] Support ppc-asm-full-reg-names for AIX.
Jan 7 2021, 11:32 PM · Restricted Project
steven.zhang updated the summary of D94162: [PowerPC] Add variants of 64-bit vector types for vec_sel..
Jan 7 2021, 6:18 PM · Restricted Project
steven.zhang added inline comments to D94211: [NFC][PowerPC] Format and clean the logic for setOperationActions to make it easier to maintain.
Jan 7 2021, 2:04 AM · Restricted Project, Restricted Project
steven.zhang added a comment to D94159: [PowerPC] [NFC] Add AIX triple to some regression tests.

It surprises to me that you didn't see the difference between LE and AIX, which doesn't make sense to me. Please double confirm to see if there is any conflict when update the test with the script.

Jan 7 2021, 2:02 AM · Restricted Project
steven.zhang added inline comments to D94211: [NFC][PowerPC] Format and clean the logic for setOperationActions to make it easier to maintain.
Jan 7 2021, 1:33 AM · Restricted Project, Restricted Project

Jan 6 2021

steven.zhang updated the summary of D94211: [NFC][PowerPC] Format and clean the logic for setOperationActions to make it easier to maintain.
Jan 6 2021, 7:18 PM · Restricted Project, Restricted Project
steven.zhang requested review of D94211: [NFC][PowerPC] Format and clean the logic for setOperationActions to make it easier to maintain.
Jan 6 2021, 7:17 PM · Restricted Project, Restricted Project

Jan 5 2021

steven.zhang added inline comments to D92473: [Legalizer] Promote result type in expanding FP_TO_XINT.
Jan 5 2021, 10:06 PM · Restricted Project
steven.zhang added a comment to D92083: [PowerPC] Lower f128 SETCC/SELECT_CC as libcall if p9vector disabled.

Gentle ping ...

Jan 5 2021, 10:01 PM · Restricted Project
steven.zhang added a comment to D93891: [DAGCombine] Remove the check for unsafe-fp-math when we are checking the AFN.

Hmm, so we can try to remove those parts that not relative with the parameter arsenm mentioned. I guess most of the check in the DAGCombiner could be removed if I understand correctly. To remove it completely, some work is needed for parameters or others that current IR cannot represent the semantics of the global setting. Is it right ?

Jan 5 2021, 10:00 PM · Restricted Project
steven.zhang added a comment to D91050: [NFC] Add the EmitTargetCodeForConstantPool hook for target to customize it with MachineConstantPoolValue.
  • Reason why we need to change the interface
Jan 5 2021, 9:49 PM · Restricted Project
steven.zhang added a comment to D91050: [NFC] Add the EmitTargetCodeForConstantPool hook for target to customize it with MachineConstantPoolValue.

Ping...

Jan 5 2021, 4:10 PM · Restricted Project

Jan 4 2021

steven.zhang committed rG2962f1149c8f: [NFC] Add the getSizeInBytes() interface for MachineConstantPoolValue (authored by steven.zhang).
[NFC] Add the getSizeInBytes() interface for MachineConstantPoolValue
Jan 4 2021, 7:26 PM
steven.zhang closed D89108: [NFC] Add the getSizeInBytes() interface for MachineConstantPoolValue.
Jan 4 2021, 7:26 PM · Restricted Project
steven.zhang added inline comments to D94054: [PowerPC] Try to fold sqrt/sdiv test results with the branch..
Jan 4 2021, 6:44 PM · Restricted Project