[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

ManuelJBrito added a comment.

It seems the build failure was caused by a known crash  
https://github.com/llvm/llvm-project/issues/55263. I tried to find some 
workaround but unsuccessfully.

So I'm thinking I can drop the end-to-end tests for now and commit them when 
the crash is fixed and for now just rely on the separate frontend and backend 
tests for regressions.

Is this OK with you @RKSimon ?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Simon Pilgrim via Phabricator via cfe-commits]

RKSimon added a comment.

In D144903#4169623 , @ManuelJBrito 
wrote:

> It seems the build failure was caused by a known crash  
> https://github.com/llvm/llvm-project/issues/55263. I tried to find some 
> workaround but unsuccessfully.

@pengfei Do you know what's going on with this bug?

> So I'm thinking I can drop the end-to-end tests for now and commit them when 
> the crash is fixed and for now just rely on the separate frontend and backend 
> tests for regressions.
>
> Is this OK with you @RKSimon ?

Let's see if we can unstick PR55263 first - yak shaving :(


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Juneyoung Lee via Phabricator via cfe-commits]

aqjune added inline comments.



Comment at: clang/test/CodeGen/X86/avx-cast-builtins.c:1
 // RUN: %clang_cc1 %s -O3 -flax-vector-conversions=none -ffreestanding %s 
-triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  
-target-feature +avx512fp16 -S -o - | FileCheck %s
 

This line contains `%s` twice, which seems to cause the crash.

Would removing one of `%s` resolve the crash issue? On my machine removing one 
of those worked well; with two `%s`; it crashed.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Juneyoung Lee via Phabricator via cfe-commits]

aqjune added inline comments.



Comment at: clang/test/CodeGen/X86/avx-cast-builtins.c:1
 // RUN: %clang_cc1 %s -O3 -flax-vector-conversions=none -ffreestanding %s 
-triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  
-target-feature +avx512fp16 -S -o - | FileCheck %s
 

aqjune wrote:
> This line contains `%s` twice, which seems to cause the crash.
> 
> Would removing one of `%s` resolve the crash issue? On my machine removing 
> one of those worked well; with two `%s`; it crashed.
Actually, with FileCheck's argument counted there are three `%s`, but it isn't 
important; I meant two `%s` in clang_cc1's arguments.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

ManuelJBrito added inline comments.



Comment at: clang/test/CodeGen/X86/avx-cast-builtins.c:1
 // RUN: %clang_cc1 %s -O3 -flax-vector-conversions=none -ffreestanding %s 
-triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  
-target-feature +avx512fp16 -S -o - | FileCheck %s
 

aqjune wrote:
> aqjune wrote:
> > This line contains `%s` twice, which seems to cause the crash.
> > 
> > Would removing one of `%s` resolve the crash issue? On my machine removing 
> > one of those worked well; with two `%s`; it crashed.
> Actually, with FileCheck's argument counted there are three `%s`, but it 
> isn't important; I meant two `%s` in clang_cc1's arguments.
Oops .. I should have noticed that...
I will try to commit again with this fix . Thanks for looking at this aqjune!!


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rGe2817933fdd4: [X86] Drop single use check for freeze(undef) 
in LowerAVXCONCAT_VECTORS (authored by ManuelJBrito).

Changed prior to commit:
  https://reviews.llvm.org/D144903?vs=501079&id=503711#toc

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

Files:
  clang/test/CodeGen/X86/avx-cast-builtins.c
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/test/CodeGen/X86/avx512-intrinsics.ll
  llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll

Index: llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
===
--- llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
+++ llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
@@ -1231,10 +1231,7 @@
 define <32 x half> @test_mm512_castph128_ph512_freeze(<8 x half> %a0) nounwind {
 ; CHECK-LABEL: test_mm512_castph128_ph512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:retq
   %a1 = freeze <8 x half> poison
   %res = shufflevector <8 x half> %a0, <8 x half> %a1, <32 x i32> 
Index: llvm/test/CodeGen/X86/avx512-intrinsics.ll
===
--- llvm/test/CodeGen/X86/avx512-intrinsics.ll
+++ llvm/test/CodeGen/X86/avx512-intrinsics.ll
@@ -7495,10 +7495,7 @@
 define <8 x double> @test_mm256_castpd128_pd256_freeze(<2 x double> %a0) nounwind {
 ; CHECK-LABEL: test_mm256_castpd128_pd256_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <2 x double> poison
   %res = shufflevector <2 x double> %a0, <2 x double> %a1, <8 x i32> 
@@ -7520,10 +7517,7 @@
 define <16 x float> @test_mm256_castps128_ps512_freeze(<4 x float> %a0) nounwind {
 ; CHECK-LABEL: test_mm256_castps128_ps512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <4 x float> poison
   %res = shufflevector <4 x float> %a0, <4 x float> %a1, <16x i32> 
@@ -7545,10 +7539,7 @@
 define <8 x i64> @test_mm512_castsi128_si512_freeze(<2 x i64> %a0) nounwind {
 ; CHECK-LABEL: test_mm512_castsi128_si512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <2 x i64> poison
   %res = shufflevector <2 x i64> %a0, <2 x i64> %a1, <8 x i32> 
Index: llvm/lib/Target/X86/X86ISelLowering.cpp
===
--- llvm/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -11656,7 +11656,7 @@
 SDValue SubVec = Op.getOperand(i);
 if (SubVec.isUndef())
   continue;
-if (ISD::isFreezeUndef(SubVec.getNode()) && SubVec.hasOneUse())
+if (ISD::isFreezeUndef(SubVec.getNode()))
   ++NumFreezeUndef;
 else if (ISD::isBuildVectorAllZeros(SubVec.getNode()))
   ++NumZero;
Index: clang/test/CodeGen/X86/avx-cast-builtins.c
===
--- /dev/null
+++ clang/test/CodeGen/X86/avx-cast-builtins.c
@@ -0,0 +1,100 @@
+// RUN: %clang_cc1 -O3 -flax-vector-conversions=none -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  -target-feature +avx512fp16 -S -o - | FileCheck %s
+
+
+#include 
+
+__m256d test_mm256_castpd128_pd256(__m128d A) {
+  // CHECK-LABEL: test_mm256_castpd128_pd256
+  // CHECK: # %bb.0:
+  // CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
+  // CHECK-NEXT:ret{{[l|q]}}
+  return _mm256_castpd128_pd256(A);
+}
+
+__m256 test_mm256_castps128_ps256(__m128 A) {
+  // CHECK-LABEL: test_mm256_castps128_ps256
+  // CHECK: # %bb.0:
+  // CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
+  // CHECK-NEXT:ret{{[l|q]}}
+  return _mm256_castps128_ps256(A);
+}
+
+__m256i test_mm256_castsi128_si256(__

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Eli Friedman via Phabricator via cfe-commits]

efriedma added a comment.

I think the point of the hasOneUse check is to avoid a possible miscompile; if 
a FREEZE has more than one use, all users need to see the same value.  So not 
sure dropping the check is correct in general.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Nuno Lopes via Phabricator via cfe-commits]

nlopes added a comment.

In D144903#4182498 , @efriedma wrote:

> I think the point of the hasOneUse check is to avoid a possible miscompile; 
> if a FREEZE has more than one use, all users need to see the same value.  So 
> not sure dropping the check is correct in general.

Good point.
This patch is not correct, must be reverted.
The shufflevectors create copies of a same 'freeze poison'. These copies must 
see the same value, and that's not the case after this patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

ManuelJBrito added a comment.

Reverted. I'll reassess what can be done here.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

ManuelJBrito created this revision.
ManuelJBrito added a reviewer: RKSimon.
Herald added subscribers: pengfei, hiraditya.
Herald added a project: All.
ManuelJBrito requested review of this revision.
Herald added projects: clang, LLVM.
Herald added subscribers: llvm-commits, cfe-commits.

Ignoring freeze(undef) if it has multiple uses in LowerAVXCONCAT_VECTORS causes 
the custom INSERT_SUBVECTOR for vector widening to be ignored.

For example in https://godbolt.org/z/7hacPe1KM extra vinsertf128 instructions 
are introduced.

This is necessary to lower `mm512_cast*128` intel intrinsics as no-op 
intrinsics.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D144903

Files:
  clang/test/CodeGen/X86/avx-cast-builtins.c
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/test/CodeGen/X86/avx512-intrinsics.ll
  llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll

Index: llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
===
--- llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
+++ llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
@@ -1231,10 +1231,7 @@
 define <32 x half> @test_mm512_castph128_ph512_freeze(<8 x half> %a0) nounwind {
 ; CHECK-LABEL: test_mm512_castph128_ph512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:retq
   %a1 = freeze <8 x half> poison
   %res = shufflevector <8 x half> %a0, <8 x half> %a1, <32 x i32> 
Index: llvm/test/CodeGen/X86/avx512-intrinsics.ll
===
--- llvm/test/CodeGen/X86/avx512-intrinsics.ll
+++ llvm/test/CodeGen/X86/avx512-intrinsics.ll
@@ -7495,10 +7495,7 @@
 define <8 x double> @test_mm256_castpd128_pd256_freeze(<2 x double> %a0) nounwind {
 ; CHECK-LABEL: test_mm256_castpd128_pd256_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <2 x double> poison
   %res = shufflevector <2 x double> %a0, <2 x double> %a1, <8 x i32> 
@@ -7520,10 +7517,7 @@
 define <16 x float> @test_mm256_castps128_ps512_freeze(<4 x float> %a0) nounwind {
 ; CHECK-LABEL: test_mm256_castps128_ps512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <4 x float> poison
   %res = shufflevector <4 x float> %a0, <4 x float> %a1, <16x i32> 
@@ -7545,10 +7539,7 @@
 define <8 x i64> @test_mm512_castsi128_si512_freeze(<2 x i64> %a0) nounwind {
 ; CHECK-LABEL: test_mm512_castsi128_si512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <2 x i64> poison
   %res = shufflevector <2 x i64> %a0, <2 x i64> %a1, <8 x i32> 
Index: llvm/lib/Target/X86/X86ISelLowering.cpp
===
--- llvm/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -11656,7 +11656,7 @@
 SDValue SubVec = Op.getOperand(i);
 if (SubVec.isUndef())
   continue;
-if (ISD::isFreezeUndef(SubVec.getNode()) && SubVec.hasOneUse())
+if (ISD::isFreezeUndef(SubVec.getNode()))
   ++NumFreezeUndef;
 else if (ISD::isBuildVectorAllZeros(SubVec.getNode()))
   ++NumZero;
Index: clang/test/CodeGen/X86/avx-cast-builtins.c
===
--- clang/test/CodeGen/X86/avx-cast-builtins.c
+++ clang/test/CodeGen/X86/avx-cast-builtins.c
@@ -1,4 +1,4 @@
-// RUN: %clang_cc1 %s -O3 -flax-vector-conversions=none -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  -target-feature +avx512fp16 -S -o - | FileCheck %s
+// RUN: %clang_cc1 %s -O3  -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  -target-feature +avx512fp16 -S -o - | FileCheck %s
 
 
 #include 
@@ -38,10 +38,7 @@
 __m512h test_mm512_castph128_ph512(__m128h A) {
   // CHECK-LABEL: test_mm512_castph128_ph512
   // CHECK: # %bb.0:
-  // CHECK-N

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Simon Pilgrim via Phabricator via cfe-commits]

RKSimon added inline comments.



Comment at: clang/test/CodeGen/X86/avx-cast-builtins.c:1
-// RUN: %clang_cc1 %s -O3 -flax-vector-conversions=none -ffreestanding %s 
-triple=x86_64-unknown-unknown -target-feature +avx -target-feature +avx512f  
-target-feature +avx512fp16 -S -o - | FileCheck %s
 

Why did you lose this? 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

ManuelJBrito updated this revision to Diff 501079.
ManuelJBrito added a comment.

Recover mistakenly drop flag in avx-cast-builtins.
I was experimenting and forgot i had removed it.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

Files:
  clang/test/CodeGen/X86/avx-cast-builtins.c
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/test/CodeGen/X86/avx512-intrinsics.ll
  llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll

Index: llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
===
--- llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
+++ llvm/test/CodeGen/X86/avx512fp16-intrinsics.ll
@@ -1231,10 +1231,7 @@
 define <32 x half> @test_mm512_castph128_ph512_freeze(<8 x half> %a0) nounwind {
 ; CHECK-LABEL: test_mm512_castph128_ph512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:retq
   %a1 = freeze <8 x half> poison
   %res = shufflevector <8 x half> %a0, <8 x half> %a1, <32 x i32> 
Index: llvm/test/CodeGen/X86/avx512-intrinsics.ll
===
--- llvm/test/CodeGen/X86/avx512-intrinsics.ll
+++ llvm/test/CodeGen/X86/avx512-intrinsics.ll
@@ -7495,10 +7495,7 @@
 define <8 x double> @test_mm256_castpd128_pd256_freeze(<2 x double> %a0) nounwind {
 ; CHECK-LABEL: test_mm256_castpd128_pd256_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <2 x double> poison
   %res = shufflevector <2 x double> %a0, <2 x double> %a1, <8 x i32> 
@@ -7520,10 +7517,7 @@
 define <16 x float> @test_mm256_castps128_ps512_freeze(<4 x float> %a0) nounwind {
 ; CHECK-LABEL: test_mm256_castps128_ps512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <4 x float> poison
   %res = shufflevector <4 x float> %a0, <4 x float> %a1, <16x i32> 
@@ -7545,10 +7539,7 @@
 define <8 x i64> @test_mm512_castsi128_si512_freeze(<2 x i64> %a0) nounwind {
 ; CHECK-LABEL: test_mm512_castsi128_si512_freeze:
 ; CHECK:   # %bb.0:
-; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-; CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+; CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
 ; CHECK-NEXT:ret{{[l|q]}}
   %a1 = freeze <2 x i64> poison
   %res = shufflevector <2 x i64> %a0, <2 x i64> %a1, <8 x i32> 
Index: llvm/lib/Target/X86/X86ISelLowering.cpp
===
--- llvm/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -11656,7 +11656,7 @@
 SDValue SubVec = Op.getOperand(i);
 if (SubVec.isUndef())
   continue;
-if (ISD::isFreezeUndef(SubVec.getNode()) && SubVec.hasOneUse())
+if (ISD::isFreezeUndef(SubVec.getNode()))
   ++NumFreezeUndef;
 else if (ISD::isBuildVectorAllZeros(SubVec.getNode()))
   ++NumZero;
Index: clang/test/CodeGen/X86/avx-cast-builtins.c
===
--- clang/test/CodeGen/X86/avx-cast-builtins.c
+++ clang/test/CodeGen/X86/avx-cast-builtins.c
@@ -38,10 +38,7 @@
 __m512h test_mm512_castph128_ph512(__m128h A) {
   // CHECK-LABEL: test_mm512_castph128_ph512
   // CHECK: # %bb.0:
-  // CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-  // CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-  // CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-  // CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+  // CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $zmm0
   // CHECK-NEXT:ret{{[l|q]}}
   return _mm512_castph128_ph512(A);
 }
@@ -73,10 +70,7 @@
 __m512d test_mm512_castpd128_pd512(__m128d A){
   // CHECK-LABEL: test_mm512_castpd128_pd512
   // CHECK: # %bb.0:
-  // CHECK-NEXT:# kill: def $xmm0 killed $xmm0 def $ymm0
-  // CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm1
-  // CHECK-NEXT:vinsertf128 $1, %xmm0, %ymm0, %ymm0
-  // CHECK-NEXT:vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
+  // CHECK-NEXT:# kill: def $xmm0 k

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Simon Pilgrim via Phabricator via cfe-commits]

RKSimon accepted this revision.
RKSimon added a comment.
This revision is now accepted and ready to land.

LGTM


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Simon Pilgrim via Phabricator via cfe-commits]

RKSimon added a comment.

@ManuelJBrito Any luck with getting this committed? Your first attempt was 
reverted but was it just because of the bad Differential Revision tag?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D144903: [X86] Drop single use check for freeze(undef) in LowerAVXCONCAT_VECTORS

[Manuel Brito via Phabricator via cfe-commits]

ManuelJBrito added a comment.

I originally reverted it because of the wrong tag, but there were also some 
buildbot failures (see 
https://lab.llvm.org/buildbot/#/builders/139/builds/36736).
It appears to be failing an assert in DiagnosticsEngine::DiagStateMap::append, 
but i'm not very familiar with this part so i will have to investigate further.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144903/new/

https://reviews.llvm.org/D144903

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits