In D65000#1616368, @dnsampaio wrote:

I have tested this in our MacOS and linux environments. @thakis @thegameg @phosek, would it be possible for you to check if this works for you?

Tue, Aug 6, 1:57 PM · Restricted Project, Restricted Project

Mon, Aug 5

thegameg added a comment to rG4361da24acab: [Attributor][Fix] Resolve various liveness issues.

Hi, this seems to fail with asan enabled: http://lab.llvm.org:8080/green/job/clang-stage2-cmake-RgSan/6498/.

Mon, Aug 5, 9:48 AM

Wed, Jul 31

thegameg added a comment to D64188: [NewPM] Port the MIR Printing pass to new pass manager..

The MIRPrinter is used to print MIR when used with -stop-before / -stop-after / -run-pass flags in llc (see https://llvm.org/docs/MIRLangRef.html#mir-testing-guide).

Wed, Jul 31, 11:15 AM · Restricted Project

thegameg accepted D65504: [AArch64] Do not allocate unnecessary emergency slot..

This LGTM, thanks!

Wed, Jul 31, 11:01 AM · Restricted Project

Tue, Jul 30

thegameg committed rG84e80979b530: Reland: [Remarks] Add an LLVM-bitstream-based remark serializer (authored by thegameg).

Reland: [Remarks] Add an LLVM-bitstream-based remark serializer

Tue, Jul 30, 5:18 PM

thegameg committed rL367372: Reland: [Remarks] Add an LLVM-bitstream-based remark serializer.

Reland: [Remarks] Add an LLVM-bitstream-based remark serializer

Tue, Jul 30, 5:13 PM

thegameg committed rGd8e7967a22b6: Revert "[Remarks] Add an LLVM-bitstream-based remark serializer" (authored by thegameg).

Revert "[Remarks] Add an LLVM-bitstream-based remark serializer"

Tue, Jul 30, 5:02 PM

thegameg committed rL367370: Revert "[Remarks] Add an LLVM-bitstream-based remark serializer".

Revert "[Remarks] Add an LLVM-bitstream-based remark serializer"

Tue, Jul 30, 5:01 PM

thegameg committed rG6c3c9483e716: [Remarks] Add an LLVM-bitstream-based remark serializer (authored by thegameg).

[Remarks] Add an LLVM-bitstream-based remark serializer

Tue, Jul 30, 4:13 PM

thegameg committed rL367364: [Remarks] Add an LLVM-bitstream-based remark serializer.

[Remarks] Add an LLVM-bitstream-based remark serializer

Tue, Jul 30, 4:13 PM

thegameg closed D63466: [Remarks] Add an LLVM-bitstream-based remark serializer.

Tue, Jul 30, 4:13 PM · Restricted Project

thegameg committed rGd9402ee03f90: [Docs] Fix sphinx warning in OCamlLangImpl5.rst (authored by thegameg).

[Docs] Fix sphinx warning in OCamlLangImpl5.rst

Tue, Jul 30, 9:57 AM

thegameg committed rL367334: [Docs] Fix sphinx warning in OCamlLangImpl5.rst.

[Docs] Fix sphinx warning in OCamlLangImpl5.rst

Tue, Jul 30, 9:57 AM

thegameg committed rG5ed3d146f87e: [Remarks] Add two serialization modes for remarks: separate and standalone (authored by thegameg).

[Remarks] Add two serialization modes for remarks: separate and standalone

Tue, Jul 30, 9:02 AM

thegameg committed rL367328: [Remarks] Add two serialization modes for remarks: separate and standalone.

[Remarks] Add two serialization modes for remarks: separate and standalone

Tue, Jul 30, 9:01 AM

Mon, Jul 29

thegameg committed rGfb3c53af3407: Revert "[compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for… (authored by thegameg).

Revert "[compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for…

Mon, Jul 29, 3:09 PM

thegameg committed rL367265: Revert "[compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for….

Revert "[compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for…

Mon, Jul 29, 3:08 PM

thegameg added a comment to rL367250: [compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for xray (2).

Sorry, I had to revert this in rL367265 to unblock the macOS bots.

Mon, Jul 29, 3:08 PM

thegameg added a comment to rL367250: [compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for xray (2).

http://lab.llvm.org:8080/green/job/clang-stage1-RA/482/console

Mon, Jul 29, 1:50 PM

thegameg added a comment to rL367250: [compiler-rt] Appending COMPILER_RT_LIBCXX_PATH -isystem include for xray (2).

Hey Puyan,

Mon, Jul 29, 1:25 PM

thegameg committed rG72d00802d848: [Remarks] Update error message format string (authored by thegameg).

[Remarks] Update error message format string

Mon, Jul 29, 10:44 AM

thegameg committed rL367243: [Remarks] Update error message format string.

[Remarks] Update error message format string

Mon, Jul 29, 10:44 AM

thegameg committed rGd42289e29166: [RISCV] Fix uninitialized variable after call to evaluateConstantImm (authored by thegameg).

[RISCV] Fix uninitialized variable after call to evaluateConstantImm

Mon, Jul 29, 8:53 AM

thegameg added a comment to D65347: [RISCV] Fix uninitialized variable after call to evaluateConstantImm.

Thanks @asb!

Mon, Jul 29, 8:53 AM · Restricted Project

thegameg committed rL367230: [RISCV] Fix uninitialized variable after call to evaluateConstantImm.

[RISCV] Fix uninitialized variable after call to evaluateConstantImm

Mon, Jul 29, 8:51 AM

thegameg closed D65347: [RISCV] Fix uninitialized variable after call to evaluateConstantImm.

Mon, Jul 29, 8:51 AM · Restricted Project

Fri, Jul 26

thegameg committed rGf5a338369be4: [Remarks] Silence Wreturn-type warning (authored by thegameg).

[Remarks] Silence Wreturn-type warning

Fri, Jul 26, 3:45 PM

thegameg committed rL367162: [Remarks] Silence Wreturn-type warning.

[Remarks] Silence Wreturn-type warning

Fri, Jul 26, 3:45 PM

thegameg committed rGec63dc2c35fb: [Remarks] Update unit test to use StringRef::lower (authored by thegameg).

[Remarks] Update unit test to use StringRef::lower

Fri, Jul 26, 3:37 PM

thegameg committed rL367161: [Remarks] Update unit test to use StringRef::lower.

[Remarks] Update unit test to use StringRef::lower

Fri, Jul 26, 3:37 PM

thegameg added inline comments to rL367155: Reland: [Remarks] Support parsing remark metadata in the YAML remark parser.

Fri, Jul 26, 3:13 PM

thegameg committed rG64a5f9e11257: Reland: [Remarks] Support parsing remark metadata in the YAML remark parser (authored by thegameg).

Reland: [Remarks] Support parsing remark metadata in the YAML remark parser

Fri, Jul 26, 2:04 PM

thegameg committed rL367155: Reland: [Remarks] Support parsing remark metadata in the YAML remark parser.

Reland: [Remarks] Support parsing remark metadata in the YAML remark parser

Fri, Jul 26, 2:03 PM

thegameg committed rGcdc74e219782: Revert "[Remarks] Support parsing remark metadata in the YAML remark parser" (authored by thegameg).

Revert "[Remarks] Support parsing remark metadata in the YAML remark parser"

Fri, Jul 26, 1:55 PM

thegameg committed rL367151: Revert "[Remarks] Support parsing remark metadata in the YAML remark parser".

Revert "[Remarks] Support parsing remark metadata in the YAML remark parser"

Fri, Jul 26, 1:54 PM

thegameg added inline comments to D65000: [ARM] Set default alignment to 64bits.

Fri, Jul 26, 1:50 PM · Restricted Project, Restricted Project

thegameg created D65347: [RISCV] Fix uninitialized variable after call to evaluateConstantImm.

Fri, Jul 26, 1:35 PM · Restricted Project

thegameg committed rGa41f61625a96: [Remarks] Support parsing remark metadata in the YAML remark parser (authored by thegameg).

[Remarks] Support parsing remark metadata in the YAML remark parser

Fri, Jul 26, 1:14 PM

thegameg committed rL367148: [Remarks] Support parsing remark metadata in the YAML remark parser.

[Remarks] Support parsing remark metadata in the YAML remark parser

Fri, Jul 26, 1:11 PM

Thu, Jul 25

thegameg committed rG2d8fdcae9617: Reland: [Remarks] Add support for serializing metadata for every remark streamer (authored by thegameg).

Reland: [Remarks] Add support for serializing metadata for every remark streamer

Thu, Jul 25, 6:35 PM

thegameg committed rL367078: Reland: [Remarks] Add support for serializing metadata for every remark streamer.

Reland: [Remarks] Add support for serializing metadata for every remark streamer

Thu, Jul 25, 6:35 PM

thegameg committed rG0503add6dab5: [CodeGen] Don't resolve the stack protector frame accesses until PEI (authored by thegameg).

[CodeGen] Don't resolve the stack protector frame accesses until PEI

Thu, Jul 25, 3:28 PM

thegameg committed rL367068: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

[CodeGen] Don't resolve the stack protector frame accesses until PEI

Thu, Jul 25, 3:27 PM

thegameg closed D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

Thu, Jul 25, 3:27 PM · Restricted Project

thegameg added a comment to D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

I filed PR42764 for the missing documentation.

Thu, Jul 25, 3:21 PM · Restricted Project

thegameg added a comment to rL366946: [Remarks] Add support for serializing metadata for every remark streamer.

Seems that the issue is that /externalfile becomes C:/externalfile.

Thu, Jul 25, 9:22 AM

Wed, Jul 24

thegameg committed rGab56cf89142d: [Remarks][NFC] Rename remarks::Parser to remarks::RemarkParser (authored by thegameg).

[Remarks][NFC] Rename remarks::Parser to remarks::RemarkParser

Wed, Jul 24, 5:21 PM

thegameg committed rL366965: [Remarks][NFC] Rename remarks::Parser to remarks::RemarkParser.

[Remarks][NFC] Rename remarks::Parser to remarks::RemarkParser

Wed, Jul 24, 5:20 PM

thegameg added a comment to D15774: [PEI] Calculate stack protector frame offset before local stack allocation..

A fix for this got in: rL366371. Can we close this revision?

Wed, Jul 24, 4:44 PM

thegameg committed rG62388e384651: [Remarks] Add support for serializing metadata for every remark streamer (authored by thegameg).

[Remarks] Add support for serializing metadata for every remark streamer

Wed, Jul 24, 2:30 PM

thegameg committed rL366946: [Remarks] Add support for serializing metadata for every remark streamer.

[Remarks] Add support for serializing metadata for every remark streamer

Wed, Jul 24, 2:30 PM

thegameg added a comment to D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

In D64759#1599892, @efriedma wrote:

If there isn't any free register, the scavenger will spill a register to the emergency spill slot. That doesn't have to be any particular register; it can be basically any allocatable register. So if one or two registers are particularly sensitive, we could specifically forbid the scavenger from spilling them.

Wed, Jul 24, 2:30 PM · Restricted Project

thegameg committed rGff4b515a7788: [Remarks][NFC] Rename remarks::Serializer to remarks::RemarkSerializer (authored by thegameg).

[Remarks][NFC] Rename remarks::Serializer to remarks::RemarkSerializer

Wed, Jul 24, 12:50 PM

thegameg committed rL366939: [Remarks][NFC] Rename remarks::Serializer to remarks::RemarkSerializer.

[Remarks][NFC] Rename remarks::Serializer to remarks::RemarkSerializer

Wed, Jul 24, 12:47 PM

thegameg committed rGc5cc9efa075b: [Remarks] Simplify the creation of remark serializers (authored by thegameg).

[Remarks] Simplify the creation of remark serializers

Wed, Jul 24, 9:38 AM

thegameg committed rL366919: [Remarks] Simplify the creation of remark serializers.

[Remarks] Simplify the creation of remark serializers

Wed, Jul 24, 9:36 AM

thegameg updated the diff for D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

Make stack_guard_remat.ll more explicit.

Wed, Jul 24, 9:35 AM · Restricted Project

thegameg added a comment to D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

In D64759#1588730, @thegameg wrote:

In D64759#1588217, @efriedma wrote:

It's hard for me to imagine a scenario where this actually makes stack protection substantially more effective; if you have a wild write, presumably you can use it to do something more useful than just corrupt the address of the guard. But I guess the performance cost is small.

The only important thing I see is that it can be used to bypass the stack protector.

Is it possible for register scavenging to scavenge the register containing the address of the guard after the frame index is resolved?

I am looking into this.

Wed, Jul 24, 9:04 AM · Restricted Project

Tue, Jul 23

thegameg committed rG4287c95b08ab: [Remarks] String tables should be move-only (authored by thegameg).

[Remarks] String tables should be move-only

Tue, Jul 23, 3:53 PM

thegameg committed rL366864: [Remarks] String tables should be move-only.

[Remarks] String tables should be move-only

Tue, Jul 23, 3:53 PM

thegameg committed rGc5b5cc457545: [Remarks] Introduce a new format: yaml-strtab (authored by thegameg).

[Remarks] Introduce a new format: yaml-strtab

Tue, Jul 23, 1:44 PM

thegameg committed rL366849: [Remarks] Introduce a new format: yaml-strtab.

[Remarks] Introduce a new format: yaml-strtab

Tue, Jul 23, 1:42 PM

thegameg committed rGcbbdc4183837: [Remarks][NFC] Move the YAML serializer to its own header (authored by thegameg).

[Remarks][NFC] Move the YAML serializer to its own header

Tue, Jul 23, 12:29 PM

thegameg committed rL366842: [Remarks][NFC] Move the YAML serializer to its own header.

[Remarks][NFC] Move the YAML serializer to its own header

Tue, Jul 23, 12:27 PM

thegameg committed rG78c92d2ec3a5: [Remarks] Add unit tests for YAML serialization (authored by thegameg).

[Remarks] Add unit tests for YAML serialization

Tue, Jul 23, 11:10 AM

thegameg committed rL366832: [Remarks] Add unit tests for YAML serialization.

[Remarks] Add unit tests for YAML serialization

Tue, Jul 23, 11:09 AM

thegameg updated the diff for D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

Rebase
Update tests added in r366371
Update stack_guard_remat.ll

Tue, Jul 23, 10:48 AM · Restricted Project

Jul 17 2019

thegameg added a comment to D64887: [llvm-bcanalyzer] Fixed error 'Expected<T> must be checked before access or destruction'.

In D64887#1590340, @dendibakh wrote:

In D64887#1590334, @thegameg wrote:

LGTM, thanks! Not sure how I missed this test.

Can you please double check on your side that it fails without this change?

Jul 17 2019, 3:04 PM · Restricted Project

thegameg accepted D64887: [llvm-bcanalyzer] Fixed error 'Expected<T> must be checked before access or destruction'.

LGTM, thanks! Not sure how I missed this test.

Jul 17 2019, 2:51 PM · Restricted Project

thegameg committed rG9f2b290addfc: [PEI] Don't re-allocate a pre-allocated stack protector slot (authored by thegameg).

[PEI] Don't re-allocate a pre-allocated stack protector slot

Jul 17 2019, 1:48 PM

thegameg committed rG39fc2843e4eb: [CodeGen] Add stack protector tests where the guard gets re-assigned (authored by thegameg).

[CodeGen] Add stack protector tests where the guard gets re-assigned

Jul 17 2019, 1:48 PM

thegameg committed rG90ba54bf67c4: [CodeGen][NFC] Simplify checks for stack protector index checking (authored by thegameg).

[CodeGen][NFC] Simplify checks for stack protector index checking

Jul 17 2019, 1:47 PM

thegameg committed rL366371: [PEI] Don't re-allocate a pre-allocated stack protector slot.

[PEI] Don't re-allocate a pre-allocated stack protector slot

Jul 17 2019, 1:47 PM

thegameg closed D64757: [PEI] Don't re-allocate a pre-allocated stack protector slot.

Jul 17 2019, 1:47 PM · Restricted Project

thegameg committed rL366370: [CodeGen] Add stack protector tests where the guard gets re-assigned.

[CodeGen] Add stack protector tests where the guard gets re-assigned

Jul 17 2019, 1:47 PM

thegameg committed rL366369: [CodeGen][NFC] Simplify checks for stack protector index checking.

[CodeGen][NFC] Simplify checks for stack protector index checking

Jul 17 2019, 1:47 PM

Jul 16 2019

thegameg added a comment to D64759: [CodeGen] Don't resolve the stack protector frame accesses until PEI.

In D64759#1588217, @efriedma wrote:

It's hard for me to imagine a scenario where this actually makes stack protection substantially more effective; if you have a wild write, presumably you can use it to do something more useful than just corrupt the address of the guard. But I guess the performance cost is small.

Jul 16 2019, 5:40 PM · Restricted Project

thegameg updated the diff for D64757: [PEI] Don't re-allocate a pre-allocated stack protector slot.

Add test from PR25610.

Jul 16 2019, 1:40 PM · Restricted Project

thegameg updated the diff for D64757: [PEI] Don't re-allocate a pre-allocated stack protector slot.