Page MenuHomePhabricator

bogner (Justin Bogner)
User

Projects

User does not belong to any projects.

User Details

User Since
Mar 12 2014, 3:00 PM (267 w, 4 h)

Recent Activity

Mon, Apr 22

bogner committed rGe90d5c8db069: [IPSCCP] Add missing `AssumptionCacheTracker` dependency (authored by bogner).
[IPSCCP] Add missing `AssumptionCacheTracker` dependency
Mon, Apr 22, 10:39 AM
bogner committed rL358903: [IPSCCP] Add missing `AssumptionCacheTracker` dependency.
[IPSCCP] Add missing `AssumptionCacheTracker` dependency
Mon, Apr 22, 10:39 AM

Sun, Apr 14

bogner added a comment to D60664: [git-svn] Add a username option for push command.

Should we (either instead or as well) make the default username try git config llvm.username before falling back to the local user? That way people can set it with git config llvm.username xyz or, if they work with a lot of llvm checkouts, git config --global llvm.username xyz instead of having to specify it every time.

Sun, Apr 14, 12:03 PM · Restricted Project

Tue, Apr 9

bogner committed rGc60d09597c6b: [CMake] Move configuration of LLVM_CXX_STD to HandleLLVMOptions.cmake (authored by bogner).
[CMake] Move configuration of LLVM_CXX_STD to HandleLLVMOptions.cmake
Tue, Apr 9, 1:15 AM
bogner added a comment to D60399: [CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD.

I moved where we set the default into HandleLLVMOptions in r357976, this should fix both issues. Sorry for the breakage!

Tue, Apr 9, 1:15 AM · Restricted Project
bogner committed rL357976: [CMake] Move configuration of LLVM_CXX_STD to HandleLLVMOptions.cmake.
[CMake] Move configuration of LLVM_CXX_STD to HandleLLVMOptions.cmake
Tue, Apr 9, 1:15 AM
bogner added a comment to D60434: Add a default for LLVM_CXX_STD in HandleLLVMOptions.

This shouldn't be necessary any more after r357976. Sorry for the breakage!

Tue, Apr 9, 1:15 AM · Restricted Project

Mon, Apr 8

bogner added a comment to D60399: [CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD.
Mon, Apr 8, 6:27 AM · Restricted Project
bogner committed rG25de7691a0e2: [CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD (authored by bogner).
[CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD
Mon, Apr 8, 3:18 AM
bogner committed rL357899: [CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD.
[CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD
Mon, Apr 8, 3:17 AM
bogner closed D60399: [CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD.
Mon, Apr 8, 3:17 AM · Restricted Project
bogner created D60399: [CMake] Replace LLVM_ENABLE_CXX1Y and friends with LLVM_CXX_STD.
Mon, Apr 8, 1:38 AM · Restricted Project

Thu, Apr 4

bogner accepted D60006: [SelectionDAG] Add fcmp UNDEF handling to SelectionDAG::FoldSetCC.

lgtm

Thu, Apr 4, 1:47 PM · Restricted Project

Wed, Mar 27

bogner committed rGb1650f0da92b: [LegalizeVectorTypes] Allow single loads and stores for more short vectors (authored by bogner).
[LegalizeVectorTypes] Allow single loads and stores for more short vectors
Wed, Mar 27, 1:36 PM
bogner committed rL357120: [LegalizeVectorTypes] Allow single loads and stores for more short vectors.
[LegalizeVectorTypes] Allow single loads and stores for more short vectors
Wed, Mar 27, 1:36 PM
bogner closed D56201: [LegalizeVectorTypes] Allow single loads and stores for more short vectors.
Wed, Mar 27, 1:36 PM · Restricted Project
bogner accepted D56201: [LegalizeVectorTypes] Allow single loads and stores for more short vectors.
Wed, Mar 27, 1:36 PM · Restricted Project

Tue, Mar 26

bogner requested changes to D56201: [LegalizeVectorTypes] Allow single loads and stores for more short vectors.

I hit a couple of minor issues trying to apply the patch:

Tue, Mar 26, 1:18 PM · Restricted Project

Mar 25 2019

bogner accepted D56201: [LegalizeVectorTypes] Allow single loads and stores for more short vectors.

Sorry for the delay, I somehow completely missed this. The change looks good - I'll commit it for you today or tomorrow.

Mar 25 2019, 4:47 PM · Restricted Project

Mar 19 2019

bogner committed rGb353d6887ea5: [DAGCombine] Fix a miscompile when reducing BUILD_VECTORs to a shuffle (authored by bogner).
[DAGCombine] Fix a miscompile when reducing BUILD_VECTORs to a shuffle
Mar 19 2019, 9:52 AM
bogner committed rL356476: [DAGCombine] Fix a miscompile when reducing BUILD_VECTORs to a shuffle.
[DAGCombine] Fix a miscompile when reducing BUILD_VECTORs to a shuffle
Mar 19 2019, 9:52 AM
bogner closed D59507: [DAGCombine] Fix a miscompile when reducing BUILD_VECTORs to a shuffle.
Mar 19 2019, 9:52 AM · Restricted Project

Mar 18 2019

bogner accepted D49837: [SelectionDAG] Handle unary SelectPatternFlavor for ABS case in SelectionDAGBuilder::visitSelect..
Mar 18 2019, 4:58 PM · Restricted Project
bogner created D59507: [DAGCombine] Fix a miscompile when reducing BUILD_VECTORs to a shuffle.
Mar 18 2019, 1:14 PM · Restricted Project

Feb 18 2019

bogner added inline comments to D52954: Annotate timeline in Instruments with passes and other timed regions..
Feb 18 2019, 3:18 PM · Restricted Project

Feb 13 2019

bogner accepted D52954: Annotate timeline in Instruments with passes and other timed regions..

Is there a more generic term for this than "signpost"? If so, it might be nice to use that to avoid confusion when/if other implementations show up. If not, that's fine.

Feb 13 2019, 4:35 PM · Restricted Project

Jan 29 2019

bogner accepted D57334: [CMake] Accept ENTITLEMENTS in llvm_add_library().

LGTM

Jan 29 2019, 11:59 AM

Dec 17 2018

bogner accepted D55668: Add FMF management to common fp intrinsics in GlobalIsel.

One minor style nit, but this LGTM.

Dec 17 2018, 7:34 PM
bogner accepted D55694: GlobalISel: Improve crash on invalid mapping.

I don’t think I understand what this is trying to accomplish. ISTM that this will now hit a null dereference on line 123 when we read BreadDown in the case that NumBreakDowns is zero, but this is to avoid hitting an assert somewhere else? If this is to make it clearer as to what’s failing, shouldn’t we assert NumBreakDowns != 0 here instead of just hoping to crash on a null dereference?

No, that is what this avoids. Right now there is a null dereference on line 123. Now this function exits, so this continues until it hits
Assertion failed: (getOperandMapping(Idx).isValid() && "We must have a mapping for reg operands"), function verify, file ../lib/CodeGen/GlobalISel/RegisterBankInfo.cpp, line 572.

Dec 17 2018, 3:19 PM

Dec 15 2018

bogner added a comment to D55694: GlobalISel: Improve crash on invalid mapping.

I don’t think I understand what this is trying to accomplish. ISTM that this will now hit a null dereference on line 123 when we read BreadDown in the case that NumBreakDowns is zero, but this is to avoid hitting an assert somewhere else? If this is to make it clearer as to what’s failing, shouldn’t we assert NumBreakDowns != 0 here instead of just hoping to crash on a null dereference?

Dec 15 2018, 11:34 AM

Dec 10 2018

bogner accepted D55294: [GISel]: Refactor MachineIRBuilder to allow passing additional parameters to build Instrs.
Dec 10 2018, 2:17 PM
bogner added a comment to D55294: [GISel]: Refactor MachineIRBuilder to allow passing additional parameters to build Instrs.

This looks like a big improvement to the API overall, thanks for working on it.

Dec 10 2018, 11:46 AM
bogner added a comment to D55294: [GISel]: Refactor MachineIRBuilder to allow passing additional parameters to build Instrs.

I also added{F7649492}


which is a clang-tidy implementation to automatically add {} to most buildInstr calls.

Dec 10 2018, 11:20 AM

Nov 6 2018

bogner committed rL346252: [utils] Update SmallVector lldb formatter for r337514.
[utils] Update SmallVector lldb formatter for r337514
Nov 6 2018, 10:54 AM

Nov 5 2018

bogner committed rL346185: Specify REQUIRES: default_triple in two debuginfo tests.
Specify REQUIRES: default_triple in two debuginfo tests
Nov 5 2018, 4:19 PM

Oct 31 2018

bogner accepted D53950: Fix clang -Wimplicit-fallthrough warnings across llvm, NFC.

lgtm as long as there's general consent to the idea in the llvm-dev thread.

Oct 31 2018, 3:49 PM
bogner accepted D52366: [tblgen][disasm] Separate encodings from instructions.
Oct 31 2018, 3:45 PM
bogner added inline comments to D53950: Fix clang -Wimplicit-fallthrough warnings across llvm, NFC.
Oct 31 2018, 3:24 PM
bogner added inline comments to D52366: [tblgen][disasm] Separate encodings from instructions.
Oct 31 2018, 10:24 AM
bogner added inline comments to D53414: Add instructions for migrating branches from one git repository to another..
Oct 31 2018, 9:25 AM

Oct 24 2018

bogner added inline comments to D53414: Add instructions for migrating branches from one git repository to another..
Oct 24 2018, 12:03 PM

Oct 22 2018

bogner committed rL344942: Reapply "[MachineCopyPropagation] Reimplement CopyTracker in terms of register….
Reapply "[MachineCopyPropagation] Reimplement CopyTracker in terms of register…
Oct 22 2018, 12:53 PM

Oct 9 2018

bogner committed rL344087: [LV] Move test for r343954 into x86 subdirectory.
[LV] Move test for r343954 into x86 subdirectory
Oct 9 2018, 3:41 PM

Sep 25 2018

bogner accepted D49881: docs: Emphasize ArrayRef over SmallVectorImpl.

LGTM once dberris's suggestions are taken into account

Sep 25 2018, 11:12 AM
bogner committed rL342993: Revert "[DebugInfo] Do not generate address info for removed debug labels.".
Revert "[DebugInfo] Do not generate address info for removed debug labels."
Sep 25 2018, 10:31 AM

Sep 24 2018

bogner committed rL342942: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units.
[MachineCopyPropagation] Reimplement CopyTracker in terms of register units
Sep 24 2018, 10:18 PM
bogner closed D52374: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units.
Sep 24 2018, 10:18 PM
bogner committed rL342940: [MachineCopyPropagation] Rework how we manage RegMask clobbers.
[MachineCopyPropagation] Rework how we manage RegMask clobbers
Sep 24 2018, 9:46 PM
bogner closed D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers.
Sep 24 2018, 9:46 PM
bogner added a comment to D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers.

I'm convinced now the clobbering works correctly because as clobberRegister does not just clobber the register at hand but the whole destination register of the copy touched by the COPY at hand.
LGTM

Sep 24 2018, 5:06 PM
bogner added inline comments to D52374: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units.
Sep 24 2018, 5:00 PM
bogner updated the diff for D52374: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units.

Clarify a bit based on Matthias's comments.

Sep 24 2018, 4:59 PM
bogner added inline comments to D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers.
Sep 24 2018, 3:40 PM
bogner updated the diff for D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers.

I'd accidentally sent an old version of the patch for the review originally, which didn't quite work. This one is correct and also incorporates Matthias's feedback.

Sep 24 2018, 3:40 PM

Sep 21 2018

bogner added a parent revision for D52374: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units: D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers.
Sep 21 2018, 12:08 PM
bogner added a child revision for D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers: D52374: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units.
Sep 21 2018, 12:08 PM
bogner created D52374: [MachineCopyPropagation] Reimplement CopyTracker in terms of register units.
Sep 21 2018, 12:08 PM
bogner created D52370: [MachineCopyPropagation] Rework how we manage RegMask clobbers.
Sep 21 2018, 11:03 AM

Sep 20 2018

bogner committed rL342703: [MachineCopyPropagation] Refactor copy tracking into a class. NFC.
[MachineCopyPropagation] Refactor copy tracking into a class. NFC
Sep 20 2018, 5:52 PM
bogner accepted D49064: DAG: Add helper for creating shifts with correct type.

Patch LGTM.

Sep 20 2018, 5:31 PM
bogner committed rL342700: [MachineCopyPropagation] Minor clang-formatting. NFC.
[MachineCopyPropagation] Minor clang-formatting. NFC
Sep 20 2018, 5:12 PM

Aug 23 2018

bogner committed rL340594: [SDAG] Add versions of computeKnownBits that return a value.
[SDAG] Add versions of computeKnownBits that return a value
Aug 23 2018, 7:43 PM

Aug 19 2018

bogner committed rL340153: [SimplifyCFG] Replace some uses of bitwise or with logical or.
[SimplifyCFG] Replace some uses of bitwise or with logical or
Aug 19 2018, 11:38 PM

Aug 16 2018

bogner committed rL339949: [docs] Try to clarify the FuzzingLLVM docs.
[docs] Try to clarify the FuzzingLLVM docs
Aug 16 2018, 2:55 PM
bogner accepted D50701: [MI] Change the array of `MachineMemOperand` pointers to be a generically extensible collection of extra info attached to a `MachineInstr`..

lg

Aug 16 2018, 1:52 PM

Aug 15 2018

bogner added inline comments to D50701: [MI] Change the array of `MachineMemOperand` pointers to be a generically extensible collection of extra info attached to a `MachineInstr`..
Aug 15 2018, 10:32 AM

Aug 13 2018

bogner accepted D50680: [SDAG] Remove the reliance on MI's allocation strategy for `MachineMemOperand` pointers attached to `MachineSDNodes` and instead have the `SelectionDAG` fully manage the memory for this array..

LGTM as long as Hal's good with your explanation of why we can't use TinyPtrVector.

Aug 13 2018, 6:18 PM
bogner added a comment to D50594: [analyzer] [NFC] Introduce separate targets for testing the analyzer: check-clang-analyzer and check-clang-analyzer-z3.

The cmake and lit bits all look correct

Aug 13 2018, 1:37 PM

Aug 10 2018

bogner added a comment to D50285: [MC] Remove MCRegisterClass::getSize.

What exactly are out of tree targets supposed to migrate to?

Aug 10 2018, 12:51 AM

Aug 7 2018

bogner accepted D50283: Refactor FileCheck to be used as API.

LGTM.

Aug 7 2018, 2:16 PM

Aug 3 2018

bogner added a comment to D49973: [LegalizeDAG] Fix FCOPYSIGN expansion.

This landed shortly after the 7.0 release branch was created. Should we merge it?

Aug 3 2018, 2:36 PM

Aug 1 2018

bogner added inline comments to D48850: [GISel]: Testing LegalizerHelper using Unit tests.
Aug 1 2018, 1:17 PM

Jul 26 2018

bogner accepted D49843: [CMake] Followup for r337366: Only export LLVM_LINK_LLVM_DYLIB it it's set to ON.

I believe this fixes the issues I was hitting.

Jul 26 2018, 9:44 PM

Jul 24 2018

bogner accepted D49771: CodeGen: use non-zero memset when possible for automatic variables.

Seems straightforward and correct to me.

Jul 24 2018, 5:52 PM

Jul 23 2018

bogner committed rL337718: [lit] Move the shtest-xunit-output check lines into shtest-format.
[lit] Move the shtest-xunit-output check lines into shtest-format
Jul 23 2018, 11:08 AM

Jul 19 2018

bogner added a comment to D48850: [GISel]: Testing LegalizerHelper using Unit tests.

Conceptually this looks pretty good, but we should definitely sink a lot of this implementation into lib/Support/FileCheck.cpp and anything that stays in the header should be moved into the llvm:: namespace, since this is library code now. There are a lot of overly generic names in the implementation as is (like Pattern), which is fine as long as they end up in anonymous namespaces in FileCheck.cpp but will have to be changed if they're really part of the API.

Jul 19 2018, 11:03 AM

Jul 10 2018

bogner closed D48797: [CMake] Teach the build system to codesign built products.

Committed in r336708.

Jul 10 2018, 10:38 AM
bogner committed rL336708: [CMake] Teach the build system to codesign built products.
[CMake] Teach the build system to codesign built products
Jul 10 2018, 10:37 AM

Jul 9 2018

bogner updated the diff for D48797: [CMake] Teach the build system to codesign built products.

Update to set path to codesign_allocate correctly and set the extra tool paths in platforms/iOS.cmake.

Jul 9 2018, 2:35 PM

Jul 3 2018

bogner accepted D48905: [ADT] Guard our use of function pointers in a `PointerUnion` by a dynamic check that the pointer is actually sufficiently aligned..

This fixes the segfault / asan error I was seeing in ADTTests with -Os on macOS, and the trade off with the runtime check seems pretty reasonable to me.

Jul 3 2018, 5:07 PM

Jun 29 2018

bogner created D48797: [CMake] Teach the build system to codesign built products.
Jun 29 2018, 3:14 PM

Jun 28 2018

bogner committed rL335900: [CMake] Respect CMAKE_STRIP and CMAKE_DSYMUTIL on apple platforms.
[CMake] Respect CMAKE_STRIP and CMAKE_DSYMUTIL on apple platforms
Jun 28 2018, 11:41 AM

Jun 27 2018

bogner committed rL335759: [Object] Allow iterating over an IRObjectFile's modules.
[Object] Allow iterating over an IRObjectFile's modules
Jun 27 2018, 11:03 AM

Jun 26 2018

bogner committed rL335698: [ADT] Pass DerivedT from pointe{e,r}_iterator to iterator_adaptor_base.
[ADT] Pass DerivedT from pointe{e,r}_iterator to iterator_adaptor_base
Jun 26 2018, 5:59 PM

Jun 14 2018

bogner added a comment to D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.

Already saw the bot and reverted, thanks. Fix on the way.

Jun 14 2018, 12:30 PM
bogner committed rL334756: Re-apply "[VirtRegRewriter] Avoid clobbering registers when expanding copy….
Re-apply "[VirtRegRewriter] Avoid clobbering registers when expanding copy…
Jun 14 2018, 12:28 PM
bogner added a comment to D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 14 2018, 12:21 PM
bogner committed rL334754: Revert "[VirtRegRewriter] Avoid clobbering registers when expanding copy….
Revert "[VirtRegRewriter] Avoid clobbering registers when expanding copy…
Jun 14 2018, 12:16 PM
bogner committed rL334750: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
[VirtRegRewriter] Avoid clobbering registers when expanding copy bundles
Jun 14 2018, 11:39 AM
bogner closed D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 14 2018, 11:39 AM
bogner added inline comments to D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 14 2018, 11:33 AM
bogner updated the diff for D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 14 2018, 11:33 AM

Jun 13 2018

bogner added a comment to D47999: MIR YAML TracksRegLiveness default to true in yaml parser.

Why? (As in, please update the commit message to explain why)

Jun 13 2018, 11:15 PM
bogner added inline comments to D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 13 2018, 11:09 PM
bogner updated the diff for D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 13 2018, 11:09 PM
bogner added a comment to D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.

Take a look at RAGreedy::selectOrSplit for an example on how to report fatal errors in the backend.

Jun 13 2018, 5:20 PM
bogner updated the diff for D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 13 2018, 5:20 PM
bogner created D48154: [VirtRegRewriter] Avoid clobbering registers when expanding copy bundles.
Jun 13 2018, 3:02 PM
bogner accepted D47959: [DAG] Avoid needing to walk out legalization tables. NFCI..

A couple of very minor nitpicks about use of auto, below, but overall this looks very nice. LGTM.

Jun 13 2018, 2:37 PM

Jun 11 2018

bogner committed rL334452: [CMake] Fix dropped dependency in install-llvm-headers.
[CMake] Fix dropped dependency in install-llvm-headers
Jun 11 2018, 4:09 PM