date:20170316

[PATCH] D31069: Don't warn about an unreachable fallthrough annotation in a template function

2017-03-16 Thread Richard Smith via Phabricator via cfe-commits

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

LGTM, do you need someone to commit for you?


https://reviews.llvm.org/D31069



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

r298040 - [AVX-512] Add missing typecasts and parentheses to _mm512_mask_i64gather_ps. My macro cleanup script I used on the others last year must have missed it.

2017-03-16 Thread Craig Topper via cfe-commits

Author: ctopper
Date: Fri Mar 17 00:14:37 2017
New Revision: 298040

URL: http://llvm.org/viewvc/llvm-project?rev=298040=rev
Log:
[AVX-512] Add missing typecasts and parentheses to _mm512_mask_i64gather_ps. My 
macro cleanup script I used on the others last year must have missed it.

Modified:
cfe/trunk/lib/Headers/avx512fintrin.h

Modified: cfe/trunk/lib/Headers/avx512fintrin.h
URL: 
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx512fintrin.h?rev=298040=298039=298040=diff
==
--- cfe/trunk/lib/Headers/avx512fintrin.h (original)
+++ cfe/trunk/lib/Headers/avx512fintrin.h Fri Mar 17 00:14:37 2017
@@ -8149,11 +8149,11 @@ _mm512_maskz_getexp_ps (__mmask16 __U, _
(__v8di)(__m512i)(index), (__mmask8)-1, 
\
(int)(scale)); })
 
-#define _mm512_mask_i64gather_ps( __v1_old, __mask, __index,\
-  __addr, __scale) __extension__({\
-__builtin_ia32_gatherdiv16sf ((__v8sf) __v1_old,\
-  __addr,(__v8di) __index, __mask, __scale);\
-})
+#define _mm512_mask_i64gather_ps(v1_old, mask, index, addr, scale) 
__extension__({\
+  (__m256)__builtin_ia32_gatherdiv16sf((__v8sf)(__m256)(v1_old),\
+   (float const *)(addr), \
+   (__v8di)(__m512i)(index), \
+   (__mmask8)(mask), (int)(scale)); })
 
 #define _mm512_i64gather_epi32(index, addr, scale) __extension__ ({\
   (__m256i)__builtin_ia32_gatherdiv16si((__v8si)_mm256_undefined_ps(), \


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

[PATCH] D31069: Don't warn about an unreachable fallthrough annotation in a template function

2017-03-16 Thread Ahmed Asadi via Phabricator via cfe-commits

ahmedasadi updated this revision to Diff 92104.
ahmedasadi added a comment.

Added a test case.


https://reviews.llvm.org/D31069

Files:
  lib/Sema/AnalysisBasedWarnings.cpp
  test/SemaCXX/P30636.cpp


Index: test/SemaCXX/P30636.cpp
===
--- test/SemaCXX/P30636.cpp
+++ test/SemaCXX/P30636.cpp
@@ -0,0 +1,20 @@
+// RUN: %clang_cc1 -fsyntax-only -verify -std=c++11 -Wimplicit-fallthrough %s
+// expected-no-diagnostics
+
+template
+int fallthrough_template(int i)
+{
+  switch (i) {
+case 1:
+  if (param)
+return 3;
+  [[clang::fallthrough]]; // no warning here, for an unreachable 
annotation (in the fallthrough_template case) in a template function
+case 2:
+  return 4;
+default:
+  return 5;
+  }
+}
+  
+template int fallthrough_template(int);
+
Index: lib/Sema/AnalysisBasedWarnings.cpp
===
--- lib/Sema/AnalysisBasedWarnings.cpp
+++ lib/Sema/AnalysisBasedWarnings.cpp
@@ -972,7 +972,7 @@
   }
 }
 
-bool checkFallThroughIntoBlock(const CFGBlock , int ) {
+bool checkFallThroughIntoBlock(const CFGBlock , int , bool 
isTemplateInstantiation) {
   assert(!ReachableBlocks.empty() && "ReachableBlocks empty");
 
   int UnannotatedCnt = 0;
@@ -1002,8 +1002,12 @@
ElemIt != ElemEnd; ++ElemIt) {
 if (Optional CS = ElemIt->getAs()) {
   if (const AttributedStmt *AS = asFallThroughAttr(CS->getStmt())) 
{
-S.Diag(AS->getLocStart(),
-   diag::warn_fallthrough_attr_unreachable);
+// Don't issue a warning for an unreachable fallthrough
+// attribute in template instantiations as it may not be
+// unreachable in all instantiations of the template
+if (!isTemplateInstantiation)
+S.Diag(AS->getLocStart(),
+   diag::warn_fallthrough_attr_unreachable);
 markFallthroughVisited(AS);
 ++AnnotatedCnt;
 break;
@@ -1164,7 +1168,10 @@
 
 int AnnotatedCnt;
 
-if (!FM.checkFallThroughIntoBlock(*B, AnnotatedCnt))
+bool isTemplateInstantiation = false;
+if (const FunctionDecl *Function = dyn_cast(AC.getDecl()))
+isTemplateInstantiation = Function->isTemplateInstantiation();
+if (!FM.checkFallThroughIntoBlock(*B, AnnotatedCnt, 
isTemplateInstantiation))
   continue;
 
 S.Diag(Label->getLocStart(),


Index: test/SemaCXX/P30636.cpp
===
--- test/SemaCXX/P30636.cpp
+++ test/SemaCXX/P30636.cpp
@@ -0,0 +1,20 @@
+// RUN: %clang_cc1 -fsyntax-only -verify -std=c++11 -Wimplicit-fallthrough %s
+// expected-no-diagnostics
+
+template
+int fallthrough_template(int i)
+{
+  switch (i) {
+case 1:
+  if (param)
+return 3;
+  [[clang::fallthrough]]; // no warning here, for an unreachable annotation (in the fallthrough_template case) in a template function
+case 2:
+  return 4;
+default:
+  return 5;
+  }
+}
+  
+template int fallthrough_template(int);
+
Index: lib/Sema/AnalysisBasedWarnings.cpp
===
--- lib/Sema/AnalysisBasedWarnings.cpp
+++ lib/Sema/AnalysisBasedWarnings.cpp
@@ -972,7 +972,7 @@
   }
 }
 
-bool checkFallThroughIntoBlock(const CFGBlock , int ) {
+bool checkFallThroughIntoBlock(const CFGBlock , int , bool isTemplateInstantiation) {
   assert(!ReachableBlocks.empty() && "ReachableBlocks empty");
 
   int UnannotatedCnt = 0;
@@ -1002,8 +1002,12 @@
ElemIt != ElemEnd; ++ElemIt) {
 if (Optional CS = ElemIt->getAs()) {
   if (const AttributedStmt *AS = asFallThroughAttr(CS->getStmt())) {
-S.Diag(AS->getLocStart(),
-   diag::warn_fallthrough_attr_unreachable);
+// Don't issue a warning for an unreachable fallthrough
+// attribute in template instantiations as it may not be
+// unreachable in all instantiations of the template
+if (!isTemplateInstantiation)
+S.Diag(AS->getLocStart(),
+   diag::warn_fallthrough_attr_unreachable);
 markFallthroughVisited(AS);
 ++AnnotatedCnt;
 break;
@@ -1164,7 +1168,10 @@
 
 int AnnotatedCnt;
 
-if (!FM.checkFallThroughIntoBlock(*B, AnnotatedCnt))
+bool isTemplateInstantiation = false;
+if (const FunctionDecl *Function = dyn_cast(AC.getDecl()))
+isTemplateInstantiation = Function->isTemplateInstantiation();
+if (!FM.checkFallThroughIntoBlock(*B, AnnotatedCnt, isTemplateInstantiation))
   continue;
 
 S.Diag(Label->getLocStart(),

[PATCH] D31069: Don't warn about an unreachable fallthrough annotation in a template function

2017-03-16 Thread Richard Smith via Phabricator via cfe-commits

rsmith added a comment.

This needs a test case, but the change itself looks fine to me.


Repository:
  rL LLVM

https://reviews.llvm.org/D31069



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

[PATCH] D30806: [nonnull] Teach Clang to attach the nonnull LLVM attribute to declarations and calls instead of just definitions, and then teach it to not attach such attributes even if the source c

2017-03-16 Thread Hal Finkel via Phabricator via cfe-commits

hfinkel added inline comments.



Comment at: include/clang/AST/ASTContext.h:1868
+  bool *OverrideNonnullReturn = nullptr,
+  unsigned *OverrideNonnullArgs = nullptr,
   unsigned *IntegerConstantArgs = nullptr) const;

chandlerc wrote:
> hfinkel wrote:
> > It seems like you had to touch more code than necessary because you decided 
> > to add these parameters before IntegerConstantArgs. Why don't you add them 
> > afterward instead?
> It seemed more natural But I'm happy to change the order. I'm not worreid 
> about how much code I have to touch, I'd rather just get the options in the 
> right place. What order would you suggest?
I suggest the other order because it is more common to want the 
IntegerConstantArgs than the nonnull args (which essentially only one caller 
wants).


https://reviews.llvm.org/D30806



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

Re: [PATCH] D30415: Fix -mno-altivec cannot overwrite -maltivec option

2017-03-16 Thread Eric Christopher via cfe-commits

Great, thanks! Can you make this change instead Sean? :)

-eric

On Thu, Mar 16, 2017 at 7:20 PM Hal Finkel  wrote:

>
> On 03/16/2017 08:11 PM, Eric Christopher wrote:
>
>
>
> On Thu, Mar 16, 2017 at 5:45 PM Hal Finkel  wrote:
>
>
> On 03/16/2017 07:40 PM, Eric Christopher wrote:
>
>
>
> On Thu, Mar 16, 2017 at 5:37 PM Hal Finkel via Phabricator <
> revi...@reviews.llvm.org> wrote:
>
> hfinkel added a comment.
>
> In https://reviews.llvm.org/D30415#703398, @echristo wrote:
>
> > Different suggestion:
> >
> > Remove the faltivec option. Even gcc doesn't support it anymore afaict.
>
>
> What are you suggesting? Always having the language extensions on? Or
> explicitly tying the language extensions to the underlying target feature?
>
>
> I was thinking the latter.
>
>
> Is that what GCC now does?
>
>
> That would be my guess given the option isn't listed anymore, but what it
> does is this:
>
> echristo@dzur ~/s/gcc-git> grep -r faltivec *
> gcc/testsuite/gcc.target/powerpc/stabs-attrib-vect-darwin.c:/* {
> dg-options "-gstabs+ -fno-eliminate-unused-debug-types -faltivec" } */
> gcc/testsuite/ChangeLog-1993-2007: * g++.dg/ext/altivec-8.C: Use
> '-maltivec' instead of '-faltivec';
> gcc/config/rs6000/darwin.h:   the kernel or some such.  The "-faltivec"
> option should have been
> gcc/config/rs6000/darwin.h:  %{faltivec:-maltivec -include altivec.h}
> %{fno-altivec:-mno-altivec} \
> gcc/config/rs6000/darwin.h:  % gcc/config/rs6000/darwin.opt:faltivec
> gcc/ChangeLog-2003: -maltivec is specified, not the non-existent
> -faltivec.
> gcc/ChangeLog-2010: * config/rs6000/darwin.h (CC1_SPEC): Handle -faltivec
> and -fno-altivec.
> gcc/ChangeLog-2010: * config/rs6000/darwin.opt (Waltivec-long-deprecated,
> faltivec,
>
> and only on darwin. I don't see anything that treats the faltivec alias as
> anything language specific anywhere. It basically just says "pass the
> include and turn on maltivec".
>
> At this point I'm pretty sure that -faltivec can just be ignored.
>
>
> It is certainly fair to say that having the ability to use -fno-altivec
> was much more important when -faltivec included altivec.h (which injected
> names like vec_add into the global namespace). I'm fine with enabling the
> vector syntax extensions when targeting altivec is enabled (they're
> extensions to extensions anyway).
>
>
>  -Hal
>
>
>
> -eric
>
>
>
>  -Hal
>
>
>
> -eric
>
>
> > (Go ahead and commit the zvector part if you'd like).
> >
> > -eric
>
>
>
>
> https://reviews.llvm.org/D30415
>
>
>
>
> --
> Hal Finkel
> Lead, Compiler Technology and Programming Languages
> Leadership Computing Facility
> Argonne National Laboratory
>
>
> --
> Hal Finkel
> Lead, Compiler Technology and Programming Languages
> Leadership Computing Facility
> Argonne National Laboratory
>
>
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D30415: Fix -mno-altivec cannot overwrite -maltivec option

2017-03-16 Thread Hal Finkel via cfe-commits

On 03/16/2017 08:11 PM, Eric Christopher wrote:

On Thu, Mar 16, 2017 at 5:45 PM Hal Finkel > wrote:

On 03/16/2017 07:40 PM, Eric Christopher wrote:

On Thu, Mar 16, 2017 at 5:37 PM Hal Finkel via Phabricator
> wrote:

hfinkel added a comment.

In https://reviews.llvm.org/D30415#703398, @echristo wrote:

> Different suggestion:
>
> Remove the faltivec option. Even gcc doesn't support it
anymore afaict.

What are you suggesting? Always having the language
extensions on? Or explicitly tying the language extensions to
the underlying target feature?

I was thinking the latter.

Is that what GCC now does?

That would be my guess given the option isn't listed anymore, but what
it does is this:

echristo@dzur ~/s/gcc-git> grep -r faltivec *
gcc/testsuite/gcc.target/powerpc/stabs-attrib-vect-darwin.c:/* {
dg-options "-gstabs+ -fno-eliminate-unused-debug-types -faltivec" } */
gcc/testsuite/ChangeLog-1993-2007:* g++.dg/ext/altivec-8.C: Use
'-maltivec' instead of '-faltivec';
gcc/config/rs6000/darwin.h: the kernel or some such. The "-faltivec"
option should have been
gcc/config/rs6000/darwin.h: %{faltivec:-maltivec -include altivec.h}
%{fno-altivec:-mno-altivec} \

gcc/config/rs6000/darwin.h: %gcc/ChangeLog-2010:* config/rs6000/darwin.h (CC1_SPEC): Handle
-faltivec and -fno-altivec.
gcc/ChangeLog-2010:* config/rs6000/darwin.opt
(Waltivec-long-deprecated, faltivec,

and only on darwin. I don't see anything that treats the faltivec
alias as anything language specific anywhere. It basically just says
"pass the include and turn on maltivec".

At this point I'm pretty sure that -faltivec can just be ignored.

It is certainly fair to say that having the ability to use -fno-altivec
was much more important when -faltivec included altivec.h (which
injected names like vec_add into the global namespace). I'm fine with
enabling the vector syntax extensions when targeting altivec is enabled
(they're extensions to extensions anyway).

1 2 >

1 - 100 of 109 matches

Mail list logo