Page MenuHomePhabricator

joechrisellis (Joe Ellis)
User

Projects

User does not belong to any projects.

User Details

User Since
Oct 15 2020, 5:36 AM (6 w, 2 d)

Recent Activity

Thu, Nov 26

joechrisellis updated the diff for D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.
  • Implement canonicalisation for the fixed case.
  • Move intrinsic into llvm.experimental namespace.
  • Update documentation and tests accordingly.
Thu, Nov 26, 7:42 AM · Restricted Project
joechrisellis added a comment to D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.

@lebedev.ri, hi! We submitted an RFC for named shuffle intrinsics to the llvm-dev mailing list (here). Do you think this is sufficient? Note there are still a few changes left to do on this patch (mainly moving the intrinsics to the experimental namespace and adding canonicalisation to shufflevector for the fixed case). 😄

Thu, Nov 26, 2:10 AM · Restricted Project

Wed, Nov 25

joechrisellis committed rG06654a5348bf: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert (authored by joechrisellis).
[SVE] Fix TypeSize warning in RuntimePointerChecking::insert
Wed, Nov 25, 8:59 AM
joechrisellis closed D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.
Wed, Nov 25, 8:59 AM · Restricted Project
joechrisellis updated the diff for D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.

Move and re-flow comment lines.

Wed, Nov 25, 7:40 AM · Restricted Project

Tue, Nov 24

joechrisellis planned changes to D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.

Planning to add canonicalization to shufflevector for fixed-width vectors.

Tue, Nov 24, 4:03 AM · Restricted Project
joechrisellis added a comment to D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.

Thank you @fhahn, that comment was super helpful. I've opted to create a new test here because I didn't want to change the relationship between the IR and the C++ at the top of memcheck-off-by-on-error.ll. If you think this is the wrong thing to do, just let me know. 😄

Tue, Nov 24, 3:49 AM · Restricted Project
joechrisellis updated the diff for D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.

Add test to defend against accidentally using alloc size instead of store size.

Tue, Nov 24, 3:46 AM · Restricted Project

Mon, Nov 23

joechrisellis updated the diff for D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.

Fix failing tests (forgot to write stderr to %t).

Mon, Nov 23, 9:17 AM · Restricted Project
joechrisellis updated the diff for D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.

Auto-generate tests.

Mon, Nov 23, 7:56 AM · Restricted Project
joechrisellis updated the diff for D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.

Address @paulwalker-arm's comments.

Mon, Nov 23, 7:36 AM · Restricted Project
joechrisellis committed rG3c696a212ba4: [AArch64][SVE] Allow lax conversion between VLATs and GNU vectors (authored by joechrisellis).
[AArch64][SVE] Allow lax conversion between VLATs and GNU vectors
Mon, Nov 23, 2:48 AM
joechrisellis closed D91696: [AArch64][SVE] Allow lax conversion between VLATs and GNU vectors.
Mon, Nov 23, 2:47 AM · Restricted Project
joechrisellis added a comment to D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.

@fhahn Hi! Do you have any objections to landing this as-is without the additional test? Alternatively, any recommendations on how to test this? 😄

Mon, Nov 23, 2:06 AM · Restricted Project

Thu, Nov 19

joechrisellis committed rG1e2da3839cc3: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors (authored by joechrisellis).
[AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors
Thu, Nov 19, 3:19 AM
joechrisellis closed D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.
Thu, Nov 19, 3:18 AM · Restricted Project
joechrisellis updated the diff for D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.

Avoid doing FirstType->getAs<BuiltinType>().

Thu, Nov 19, 3:17 AM · Restricted Project
joechrisellis added inline comments to D91696: [AArch64][SVE] Allow lax conversion between VLATs and GNU vectors.
Thu, Nov 19, 2:20 AM · Restricted Project
joechrisellis updated the diff for D91696: [AArch64][SVE] Allow lax conversion between VLATs and GNU vectors.

Address @c-rhodes's comments regarding lax conversion when __ARM_FEATURE_SVE_BITS != N for GNU vectors.

Thu, Nov 19, 2:20 AM · Restricted Project

Wed, Nov 18

joechrisellis updated the diff for D91696: [AArch64][SVE] Allow lax conversion between VLATs and GNU vectors.

Remove redundant tests from clang/test/Sema/attr-arm-sve-vector-bits.c.

Wed, Nov 18, 4:08 AM · Restricted Project
joechrisellis requested review of D91696: [AArch64][SVE] Allow lax conversion between VLATs and GNU vectors.
Wed, Nov 18, 3:51 AM · Restricted Project

Tue, Nov 17

joechrisellis committed rG1d7abcf99e3d: [AArch64][SVE] Add tests for VLST -> VLAT lax conversions (authored by joechrisellis).
[AArch64][SVE] Add tests for VLST -> VLAT lax conversions
Tue, Nov 17, 10:04 AM
joechrisellis closed D91642: [AArch64][SVE] Add tests for VLST -> VLAT lax conversions.
Tue, Nov 17, 10:04 AM · Restricted Project
joechrisellis requested review of D91642: [AArch64][SVE] Add tests for VLST -> VLAT lax conversions.
Tue, Nov 17, 9:34 AM · Restricted Project
joechrisellis committed rG23a96b84a8d9: [AArch64][SVE] Support implicit lax vector conversions for SVE types (authored by joechrisellis).
[AArch64][SVE] Support implicit lax vector conversions for SVE types
Tue, Nov 17, 6:50 AM
joechrisellis closed D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.
Tue, Nov 17, 6:50 AM · Restricted Project
joechrisellis updated the diff for D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.

Address @c-rhodes's comment regarding comment change.

Tue, Nov 17, 6:29 AM · Restricted Project
joechrisellis added inline comments to D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.
Tue, Nov 17, 6:25 AM · Restricted Project
joechrisellis added inline comments to D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.
Tue, Nov 17, 5:56 AM · Restricted Project
joechrisellis updated the diff for D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.

Address @c-rhodes's and @peterwaller-arm's comments.

Tue, Nov 17, 5:55 AM · Restricted Project
joechrisellis updated the diff for D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.
Tue, Nov 17, 2:02 AM · Restricted Project

Mon, Nov 16

joechrisellis updated the diff for D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.

Remove failing test; it was checking that a conversion _failed_, although the conversion should now _pass_ given the changes in this patch.

Mon, Nov 16, 6:49 AM · Restricted Project

Fri, Nov 13

joechrisellis updated the diff for D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.

Address @sdesmalen and @paulwalker-arm review comments.

Fri, Nov 13, 9:28 AM · Restricted Project
joechrisellis added inline comments to D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.
Fri, Nov 13, 1:16 AM · Restricted Project

Thu, Nov 12

joechrisellis requested review of D91362: [SelectionDAG] Add llvm.vector.{extract,insert} intrinsics.
Thu, Nov 12, 8:49 AM · Restricted Project
joechrisellis accepted D91333: [clang][SveEmitter] Fix enum declarations. [NFCI].
Thu, Nov 12, 7:33 AM · Restricted Project
joechrisellis added a comment to D91333: [clang][SveEmitter] Fix enum declarations. [NFCI].

Thanks for the patch @fpetrogalli! One small comment from me, but not a blocker for me, so approved.

Thu, Nov 12, 7:32 AM · Restricted Project
joechrisellis updated the diff for D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.
  • Support C lax vector conversions.
  • Test C lax vector conversions.
Thu, Nov 12, 6:00 AM · Restricted Project
joechrisellis added inline comments to D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.
Thu, Nov 12, 5:26 AM · Restricted Project
joechrisellis updated the diff for D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.

Address @c-rhodes's comments.

Thu, Nov 12, 5:26 AM · Restricted Project

Wed, Nov 11

joechrisellis added inline comments to D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.
Wed, Nov 11, 9:34 AM · Restricted Project
joechrisellis updated the diff for D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.

Address @fpetrogalli's comments.

Wed, Nov 11, 9:33 AM · Restricted Project
joechrisellis updated the diff for D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.

Add tests for C-style casts to/from scalable/fixed float vector types.

Wed, Nov 11, 7:20 AM · Restricted Project
joechrisellis requested review of D91262: [AArch64][SVE] Allow C-style casts between fixed-size and scalable vectors.
Wed, Nov 11, 7:12 AM · Restricted Project
joechrisellis added inline comments to D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.
Wed, Nov 11, 4:06 AM · Restricted Project
joechrisellis updated the diff for D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.

Address @fpetrogalli's comments.

Wed, Nov 11, 4:06 AM · Restricted Project

Tue, Nov 10

joechrisellis committed rG462dd4f80358: [SVE][AArch64] Improve specificity of vectorization legality TypeSize test (authored by joechrisellis).
[SVE][AArch64] Improve specificity of vectorization legality TypeSize test
Tue, Nov 10, 2:56 AM
joechrisellis closed D90685: [SVE][AArch64] Improve specificity of vectorization legality TypeSize test.
Tue, Nov 10, 2:55 AM · Restricted Project

Mon, Nov 9

joechrisellis added a comment to D90685: [SVE][AArch64] Improve specificity of vectorization legality TypeSize test.

@fpetrogalli sure!

Mon, Nov 9, 5:41 AM · Restricted Project
joechrisellis updated the summary of D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.
Mon, Nov 9, 5:36 AM · Restricted Project
joechrisellis added inline comments to D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.
Mon, Nov 9, 5:25 AM · Restricted Project
joechrisellis updated the diff for D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.

Address @peterwaller-arm's comment regarding unused types.

Mon, Nov 9, 5:25 AM · Restricted Project
joechrisellis requested review of D91067: [AArch64][SVE] Support implicit lax vector conversions for SVE types.
Mon, Nov 9, 4:00 AM · Restricted Project

Tue, Nov 3

joechrisellis committed rGcf637a69e742: [SVE][InstCombine] Improve specificity of InstCombine TypeSize test (authored by joechrisellis).
[SVE][InstCombine] Improve specificity of InstCombine TypeSize test
Tue, Nov 3, 7:27 AM
joechrisellis closed D90684: [SVE][InstCombine] Improve specificity of InstCombine TypeSize test.
Tue, Nov 3, 7:27 AM · Restricted Project
joechrisellis updated the diff for D90684: [SVE][InstCombine] Improve specificity of InstCombine TypeSize test.

Get rid of -mtriple and -mattr flags, since these are also superfluous.

Tue, Nov 3, 6:58 AM · Restricted Project
joechrisellis requested review of D90685: [SVE][AArch64] Improve specificity of vectorization legality TypeSize test.
Tue, Nov 3, 6:44 AM · Restricted Project
joechrisellis added inline comments to D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.
Tue, Nov 3, 6:44 AM · Restricted Project
joechrisellis requested review of D90684: [SVE][InstCombine] Improve specificity of InstCombine TypeSize test.
Tue, Nov 3, 6:42 AM · Restricted Project
joechrisellis updated the diff for D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.

Address @fhahn's comments.

Tue, Nov 3, 3:11 AM · Restricted Project

Oct 27 2020

joechrisellis added inline comments to D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.
Oct 27 2020, 3:11 AM · Restricted Project
joechrisellis updated the diff for D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.

Address @fhahn's suggestions.

Oct 27 2020, 3:11 AM · Restricted Project

Oct 26 2020

joechrisellis committed rGbf60bb26ecbf: [SVE] Fix TypeSize warning in llvm::getGEPInductionOperand (authored by joechrisellis).
[SVE] Fix TypeSize warning in llvm::getGEPInductionOperand
Oct 26 2020, 10:41 AM
joechrisellis committed rG0f8350559325: [SVE][InstCombine] Fix TypeSize warning in canReplaceGEPIdxWithZero (authored by joechrisellis).
[SVE][InstCombine] Fix TypeSize warning in canReplaceGEPIdxWithZero
Oct 26 2020, 10:41 AM
joechrisellis committed rG0383a1a8c230: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis (authored by joechrisellis).
[SVE][AArch64] Fix TypeSize warning in GEP cost analysis
Oct 26 2020, 10:41 AM
joechrisellis committed rG467e5cf40f5d: [SVE][AArch64] Fix TypeSize warning in loop vectorization legality (authored by joechrisellis).
[SVE][AArch64] Fix TypeSize warning in loop vectorization legality
Oct 26 2020, 10:41 AM
joechrisellis closed D90146: [SVE] Fix TypeSize warning in llvm::getGEPInductionOperand.
Oct 26 2020, 10:41 AM · Restricted Project
joechrisellis closed D89968: [SVE][InstCombine] Fix TypeSize warning in canReplaceGEPIdxWithZero.
Oct 26 2020, 10:41 AM · Restricted Project
joechrisellis closed D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.
Oct 26 2020, 10:41 AM · Restricted Project
joechrisellis closed D89798: [SVE][AArch64] Fix TypeSize warning in loop vectorization legality.
Oct 26 2020, 10:41 AM · Restricted Project
joechrisellis requested review of D90171: [SVE] Fix TypeSize warning in RuntimePointerChecking::insert.
Oct 26 2020, 10:24 AM · Restricted Project
joechrisellis committed rG4604441386dc: [SVE][CodeGen][DAGCombiner] Fix TypeSize warning in redundant store elimination (authored by peterwaller-arm).
[SVE][CodeGen][DAGCombiner] Fix TypeSize warning in redundant store elimination
Oct 26 2020, 9:25 AM
joechrisellis accepted D90158: [Aarch64] Fix assumption that Windows implies x86.

LGTM. Also see @arm-chrjan01's old patch here: https://reviews.llvm.org/D44822

Oct 26 2020, 8:16 AM · Restricted Project
joechrisellis requested review of D90146: [SVE] Fix TypeSize warning in llvm::getGEPInductionOperand.
Oct 26 2020, 4:36 AM · Restricted Project
joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Address sdesmalen's review comments.

Oct 26 2020, 3:19 AM · Restricted Project

Oct 22 2020

joechrisellis updated the diff for D89968: [SVE][InstCombine] Fix TypeSize warning in canReplaceGEPIdxWithZero.

Oops -- be sure to use getFixedSize.

Oct 22 2020, 9:23 AM · Restricted Project
joechrisellis requested review of D89968: [SVE][InstCombine] Fix TypeSize warning in canReplaceGEPIdxWithZero.
Oct 22 2020, 8:52 AM · Restricted Project
joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Use pointer element type to check if this is a scalar GEP.

Oct 22 2020, 3:51 AM · Restricted Project
joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Move scalable vector check to the top of the function as-per sdesmalen's suggestion.

Oct 22 2020, 3:10 AM · Restricted Project

Oct 21 2020

joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Use isa<...> instead of dyn_cast<...>.

Oct 21 2020, 9:13 AM · Restricted Project
joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Address fpetrogalli's suggestion.

Oct 21 2020, 8:17 AM · Restricted Project
joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Bail out early if we have scalable vectors.

Oct 21 2020, 7:48 AM · Restricted Project
joechrisellis updated the diff for D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.

Accidentally changed int64_t to uint64_t in the last diff. Reverted this.

Oct 21 2020, 6:03 AM · Restricted Project
joechrisellis requested review of D89872: [SVE][AArch64] Fix TypeSize warning in GEP cost analysis.
Oct 21 2020, 4:34 AM · Restricted Project
joechrisellis updated the diff for D89798: [SVE][AArch64] Fix TypeSize warning in loop vectorization legality.

Revert to single-line if statement, add description to the test, and add an
additional check line.

Oct 21 2020, 2:45 AM · Restricted Project

Oct 20 2020

joechrisellis updated the diff for D89798: [SVE][AArch64] Fix TypeSize warning in loop vectorization legality.

Address peterwaller-arm's comments regarding control flow readability

Oct 20 2020, 9:13 AM · Restricted Project
joechrisellis requested review of D89798: [SVE][AArch64] Fix TypeSize warning in loop vectorization legality.
Oct 20 2020, 8:41 AM · Restricted Project