Page MenuHomePhabricator

pftbest (Vadzim Dambrouski)
User

Projects

User does not belong to any projects.

User Details

User Since
Apr 24 2016, 5:50 PM (160 w, 3 d)

Recent Activity

Apr 2 2019

pftbest added a comment to D60033: [MSP430] Expand Atomic nodes.

But MSP430 has interrupts and they are similar to threads in some ways. For example if I need to increment some shared counter, I would like it to be done in a single instruction. That way if some interrupt triggers at the same time it would not corrupt the value, because a single instruction can not be split. I had to use inline assembly to make such kind of counter, because volatile doesn't optimize to a single instruction in debug builds, only in release.

Apr 2 2019, 1:10 PM · Restricted Project

Oct 25 2018

pftbest added a comment to D53661: [MSP430] Add MC layer.

Now looks good, thank you.

Oct 25 2018, 3:01 PM

Oct 24 2018

pftbest added a comment to D53661: [MSP430] Add MC layer.

I've tested this patch with asan and it found some stack-use-after-scope. Can you please take a look?
Log: https://gist.github.com/pftbest/e7ca0967ab96e2318a86289a31fe8fe4

Oct 24 2018, 11:43 PM

Jul 2 2018

pftbest updated the diff for D48846: [ARM] Fix PR37382: Don't optimize mul.with.overflow on thumbv6m..

Explicit check for __aeabi_lmul

Jul 2 2018, 1:51 PM
pftbest added a comment to D48846: [ARM] Fix PR37382: Don't optimize mul.with.overflow on thumbv6m..

I've added a RUN line in overflow-intrinsic-optimizations.ll test, and it will cause the same exact assertion if tested without the fix. So I think adding another test would be redundant.

Jul 2 2018, 12:49 PM
pftbest created D48846: [ARM] Fix PR37382: Don't optimize mul.with.overflow on thumbv6m..
Jul 2 2018, 12:03 PM

Sep 19 2017

pftbest added a comment to D36061: [MSP430] Implement multiplication by a constant.

ping, please review

Sep 19 2017, 2:19 PM

Aug 30 2017

pftbest added a comment to D36061: [MSP430] Implement multiplication by a constant.

ping

Aug 30 2017, 5:46 AM

Aug 23 2017

pftbest added a comment to D36061: [MSP430] Implement multiplication by a constant.

ping

Aug 23 2017, 4:22 AM

Aug 16 2017

pftbest updated the diff for D36061: [MSP430] Implement multiplication by a constant.

Fixed an issue with i8 not being promoted to i16 in case we fall back
to libcall.

Aug 16 2017, 5:52 PM
pftbest added a comment to D36165: [MSP430] Align functions on 2-byte boundary instead of 4..

Should I add the tests for this change?

Aug 16 2017, 5:07 PM
pftbest updated the diff for D36061: [MSP430] Implement multiplication by a constant.

Moved the algorithm into SelectionDAG
Added an option to limit the number of operations

Aug 16 2017, 5:07 PM

Aug 1 2017

pftbest created D36165: [MSP430] Align functions on 2-byte boundary instead of 4..
Aug 1 2017, 11:13 AM

Jul 31 2017

pftbest added a comment to D36061: [MSP430] Implement multiplication by a constant.

@efriedma
Good catch, I accidentally used i8 type instead of i16 and didn't notice, sorry.
The correct code is rather large indeed (27 ops), so I'll add the option, and set the default limit to 12.

Jul 31 2017, 5:12 PM
pftbest added a comment to D36061: [MSP430] Implement multiplication by a constant.

Do you need to worry about codesize here? Lowering something like "a * 0x3333" to an inline sequence like this is going to generate a lot of code.

Jul 31 2017, 4:09 PM
pftbest added a comment to D36061: [MSP430] Implement multiplication by a constant.

This looks like target-independent code

Jul 31 2017, 7:53 AM

Jul 30 2017

pftbest created D36061: [MSP430] Implement multiplication by a constant.
Jul 30 2017, 9:02 AM

Jun 28 2017

pftbest added inline comments to D34472: [DAG] Rewrite areNonVolatileConsecutiveLoads to use BaseIndexOffset.
Jun 28 2017, 1:52 AM

Jun 23 2017

pftbest added a comment to D34562: [MSP430] Fix data layout string..

shall we support only EABI or... 2 different ABIs?

Jun 23 2017, 11:37 AM
pftbest created D34562: [MSP430] Fix data layout string..
Jun 23 2017, 11:20 AM
pftbest added a child revision for D34561: [MSP430] Fix data layout string.: D34562: [MSP430] Fix data layout string..
Jun 23 2017, 11:20 AM
pftbest created D34561: [MSP430] Fix data layout string..
Jun 23 2017, 10:37 AM

May 31 2017

pftbest abandoned D33469: Add .DS_Store file in .gitignore for case sensitive macOS systems..
May 31 2017, 2:42 AM

May 30 2017

pftbest added a comment to D33696: TargetMachine: Indicate whether machine verifier passes..

MSP430 was fixed, it should pass the tests now.

May 30 2017, 2:59 PM

May 23 2017

pftbest created D33469: Add .DS_Store file in .gitignore for case sensitive macOS systems..
May 23 2017, 3:06 PM
pftbest created D33452: [MSP430] Fix PR33050: Don't use ADD16ri to lower FrameIndex..
May 23 2017, 11:24 AM
pftbest added a comment to D33335: [MSP430] Add subtarget features for hardware multiplier..

I'd just like a test for the -mhwmult command line option as well as the -mattr version.

May 23 2017, 6:15 AM

May 18 2017

pftbest created D33335: [MSP430] Add subtarget features for hardware multiplier..
May 18 2017, 1:16 PM

May 11 2017

pftbest accepted D32676: [MSP430] Generate EABI-compliant libcalls.

LGTM, thank you @awygle

May 11 2017, 12:40 PM

May 8 2017

pftbest added a comment to D32676: [MSP430] Generate EABI-compliant libcalls.

Also MSP430_BUILTIN has the same id (93) as AMDGPU_HS

May 8 2017, 7:50 PM
pftbest added a comment to D32676: [MSP430] Generate EABI-compliant libcalls.

Sorry, I didn't notice this earlier, your previous revision (97399) had hwmult tests, but latest revision (98033) doesn't have them anymore.

May 8 2017, 6:58 PM

May 6 2017

pftbest added a comment to D32676: [MSP430] Generate EABI-compliant libcalls.

Looks good to me. Will do more tests on a real hardware

May 6 2017, 5:52 AM

May 5 2017

pftbest added a comment to D32676: [MSP430] Generate EABI-compliant libcalls.

this code produces a lot of warnings:

May 5 2017, 2:33 PM

Apr 25 2017

pftbest updated the diff for D32473: [MSP430] Fix PR32769: Select8 and Select16 need to have SR in Uses..

Removed the comment.

Apr 25 2017, 5:09 PM
pftbest edited reviewers for D32473: [MSP430] Fix PR32769: Select8 and Select16 need to have SR in Uses., added: efriedma; removed: eli.friedman.
Apr 25 2017, 1:52 AM

Apr 24 2017

pftbest created D32473: [MSP430] Fix PR32769: Select8 and Select16 need to have SR in Uses..
Apr 24 2017, 11:19 PM

Mar 2 2017

pftbest accepted D29069: [MSP430] Add SRet support to MSP430 target.

Looks good, Thank you.

Mar 2 2017, 12:28 PM

Mar 1 2017

pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

Looks like D29916 broke this patch. Also CanLowerReturn should be marked with override to prevent compile warnings.

Mar 1 2017, 2:13 AM

Feb 17 2017

pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

@asl can you please commit this change?

Feb 17 2017, 2:50 AM

Feb 4 2017

pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

I can confirm that now it's passing the tests

Feb 4 2017, 5:12 PM
pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

Sorry, not the vararg test, but this 3 tests:
CodeGen/MSP430/2009-08-25-DynamicStackAlloc.ll
CodeGen/MSP430/2009-12-22-InlineAsm.ll
CodeGen/MSP430/cc_ret.ll

Feb 4 2017, 3:53 PM
pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

This new revision is not good. It crashes on CodeGen/MSP430/vararg.ll test.

Feb 4 2017, 3:45 PM
pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

As I see it, CurrentArgIndex is overloaded with two (or more) tasks, so the code is confusing. Maybe it would be better if CurrentArgIndex was just a counter, and the other conditions can be handled by introducing boolean flags, for example.

Feb 4 2017, 1:57 PM
pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

Whitespace is gone, thank you.
Also I did some testing: compiled llvm with expensive checks on and address sanitizer, found no issues.
Compiled a non-trivial program and ran it on CC430 MCU, it worked fine.
So this patch LGTM

Feb 4 2017, 11:29 AM
pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

Sorry for being pedantic, but your new diff adds a lot of trailing spaces to the code.

Feb 4 2017, 10:02 AM

Feb 2 2017

pftbest added a comment to D29069: [MSP430] Add SRet support to MSP430 target.

This change looks good, but I don't have any rights to accept it, sorry. We'll have to wait for aKor to review it, you can try to ask him on irc if you see him.

Feb 2 2017, 12:33 PM

Jan 24 2017

pftbest added inline comments to D29069: [MSP430] Add SRet support to MSP430 target.
Jan 24 2017, 8:24 AM

Dec 5 2016

pftbest added a comment to D27154: [CodeGen] Fix result type for SMULO/UMULO legalization.

ping

Dec 5 2016, 2:38 AM

Nov 30 2016

pftbest added a comment to D27154: [CodeGen] Fix result type for SMULO/UMULO legalization.

I don't have access to svn, so please commit it for me.

Nov 30 2016, 11:04 AM

Nov 28 2016

pftbest updated the diff for D27154: [CodeGen] Fix result type for SMULO/UMULO legalization.

Added an assertion, and also getting the size in bits from TopHalf directly

Nov 28 2016, 1:37 PM
pftbest updated the diff for D27154: [CodeGen] Fix result type for SMULO/UMULO legalization.

Check if types are different before truncate

Nov 28 2016, 9:10 AM

Nov 27 2016

pftbest retitled D27154: [CodeGen] Fix result type for SMULO/UMULO legalization from to [CodeGen] Fix result type for SMULO/UMULO legalization.
Nov 27 2016, 4:24 PM

Nov 7 2016

pftbest added a comment to D23718: Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes..

ping

Nov 7 2016, 5:34 AM
pftbest updated subscribers of D23718: Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes..
Nov 7 2016, 5:34 AM

Oct 27 2016

pftbest accepted D26027: [ARM] Add newline char to test..

LGTM, thanks.

Oct 27 2016, 3:50 AM
pftbest added a comment to D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target..

I didn't make this change in my diff.

Oct 27 2016, 3:30 AM
pftbest added a comment to D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target..

I think there is something wrong with the last line in the commit, looks like line ending is missing

Oct 27 2016, 3:26 AM
pftbest added a comment to D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target..

ping

Oct 27 2016, 1:52 AM

Oct 22 2016

pftbest added a comment to D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target..

If anyone has svn access, please commit.

Oct 22 2016, 7:18 AM
pftbest updated subscribers of D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target..
Oct 22 2016, 6:28 AM
pftbest updated the diff for D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target..

Added test for thumbv7em (Cortex-M4) target.

Oct 22 2016, 4:13 AM

Oct 21 2016

pftbest retitled D25890: [ARM] Do not generate UMAAL instruction on thumbv7m target. from to [ARM] Do not generate UMAAL instruction on thumbv7m target..
Oct 21 2016, 6:05 PM

Oct 5 2016

pftbest updated the diff for D23718: Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes..

Update for API changes.

Oct 5 2016, 12:57 PM

Aug 27 2016

pftbest updated the diff for D23718: Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes..

Rename AllVRegsAllocated to NoVRegs.

Aug 27 2016, 5:15 AM

Aug 19 2016

pftbest retitled D23718: Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes. from to Fix PR27500: on MSP430 the branch destination offset is measured in words, not bytes..
Aug 19 2016, 9:16 AM

Aug 17 2016

pftbest added a reviewer for D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass: dexonsmith.

Please review this diff.

Aug 17 2016, 3:04 AM

Aug 13 2016

pftbest updated the diff for D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

Updated diff due to API changes.

Aug 13 2016, 2:04 AM

Jun 19 2016

pftbest added a comment to D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

ping

Jun 19 2016, 11:32 PM

Jun 11 2016

pftbest added a comment to D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

My thoughts on block renumbering:

  1. It happens only once per function in common case
  2. In a rare case when we need some branches to be expanded, block offsets are updated manually in the loop, so blocks are not renumbered
  3. In extremely rare case when both successors are out of range, only then blocks are renumbered and remeasured.
Jun 11 2016, 8:08 AM

Jun 6 2016

pftbest added a comment to D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

ping

Jun 6 2016, 1:00 PM

May 30 2016

pftbest updated the diff for D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

Constants are moved inside isInRange function, BlockOffsets is now local and is passed as a reference to other functions, correctBlockOffsets function is merged with measureFunction.

May 30 2016, 5:11 PM
pftbest added inline comments to D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.
May 30 2016, 1:46 PM
pftbest added a comment to D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

Hi! Any luck getting this diff reviewed? Is there some way I can help?

May 30 2016, 5:14 AM

May 19 2016

pftbest added a comment to D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass.

ping.
please review this change.

May 19 2016, 6:18 AM

May 11 2016

pftbest retitled D20162: [MSP430] PR27500 CodeGen: Rework branch-select pass from to [MSP430] PR27500 CodeGen: Rework branch-select pass.
May 11 2016, 6:10 AM