r180909 - Only evaluate __has_feature(c_thread_local) and __has_feature(cxx_thread_local) true when the target supports thread-local storage.

[Douglas Gregor]

r180925, thanks!

	- Doug

On May 2, 2013, at 9:14 AM, jahanian <fjahanian at apple.com> wrote:

> has_feature_c1x.c test has started to fail on Windows for us.
> Because of:
> 
> #if __has_feature(c_thread_local)
> int has_thread_local();
> #else
> int no_thread_local();
> #endif
> 
> // CHECK-1X: has_thread_local
> 
> When -std=c1x option is passed to cc1. Is this expected to pass everywhere? 
> 
> - Fariborz
> 
> 
> On May 2, 2013, at 6:17 AM, Douglas Gregor <dgregor at apple.com> wrote:
> 
>> 
>> On May 2, 2013, at 1:00 AM, Jean-Daniel Dupas <devlists at shadowlab.org> wrote:
>> 
>>> Is it OK to use the same condition for both while they don't require the same Target support ? 
>>> 
>>> AFAK, C++ TLS require __cxa_thread_atexit support, and so the target should support thread_atexit, while C TLS doesn't need it.
>> 
>> As Richard notes, whether __cxa_thread_atexit is available or not is a library issue, and we don't have visibility into that within the frontend. However, we do know whether the target supports thread-local storage in general (it's part of target information), and my change makes __has_feature(c(xx)?_thread_local) match what Sema actually allows.
>> 
>>> Actually, I think darwin has support for C thread local, but no C++ thread local.
>> 
>> x86 Darwin fully supports thread-local, although the __cxa_thread_atexit equivalent is called __tlv_atexit. ARM Darwin does not support thread-local (at all).
>> 
>> 	- Doug
>> 
>> 
>>> Le 2 mai 2013 à 07:28, Douglas Gregor <dgregor at apple.com> a écrit :
>>> 
>>>> Author: dgregor
>>>> Date: Thu May  2 00:28:32 2013
>>>> New Revision: 180909
>>>> 
>>>> URL: http://llvm.org/viewvc/llvm-project?rev=180909&view=rev
>>>> Log:
>>>> Only evaluate __has_feature(c_thread_local) and __has_feature(cxx_thread_local) true when the target supports thread-local storage.
>>>> 
>>>> Modified:
>>>>  cfe/trunk/lib/Lex/PPMacroExpansion.cpp
>>>>  cfe/trunk/test/Lexer/has_feature_cxx0x.cpp
>>>> 
>>>> Modified: cfe/trunk/lib/Lex/PPMacroExpansion.cpp
>>>> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Lex/PPMacroExpansion.cpp?rev=180909&r1=180908&r2=180909&view=diff
>>>> ==============================================================================
>>>> --- cfe/trunk/lib/Lex/PPMacroExpansion.cpp (original)
>>>> +++ cfe/trunk/lib/Lex/PPMacroExpansion.cpp Thu May  2 00:28:32 2013
>>>> @@ -751,7 +751,8 @@ static bool HasFeature(const Preprocesso
>>>>          .Case("c_atomic", LangOpts.C11)
>>>>          .Case("c_generic_selections", LangOpts.C11)
>>>>          .Case("c_static_assert", LangOpts.C11)
>>>> -           .Case("c_thread_local", LangOpts.C11)
>>>> +           .Case("c_thread_local", 
>>>> +                 LangOpts.C11 && PP.getTargetInfo().isTLSSupported())
>>>>          // C++11 features
>>>>          .Case("cxx_access_control_sfinae", LangOpts.CPlusPlus11)
>>>>          .Case("cxx_alias_templates", LangOpts.CPlusPlus11)
>>>> @@ -783,7 +784,8 @@ static bool HasFeature(const Preprocesso
>>>>          .Case("cxx_rvalue_references", LangOpts.CPlusPlus11)
>>>>          .Case("cxx_strong_enums", LangOpts.CPlusPlus11)
>>>>          .Case("cxx_static_assert", LangOpts.CPlusPlus11)
>>>> -           .Case("cxx_thread_local", LangOpts.CPlusPlus11)
>>>> +           .Case("cxx_thread_local", 
>>>> +                 LangOpts.CPlusPlus11 && PP.getTargetInfo().isTLSSupported())
>>>>          .Case("cxx_trailing_return", LangOpts.CPlusPlus11)
>>>>          .Case("cxx_unicode_literals", LangOpts.CPlusPlus11)
>>>>          .Case("cxx_unrestricted_unions", LangOpts.CPlusPlus11)
>>>> 
>>>> Modified: cfe/trunk/test/Lexer/has_feature_cxx0x.cpp
>>>> URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Lexer/has_feature_cxx0x.cpp?rev=180909&r1=180908&r2=180909&view=diff
>>>> ==============================================================================
>>>> --- cfe/trunk/test/Lexer/has_feature_cxx0x.cpp (original)
>>>> +++ cfe/trunk/test/Lexer/has_feature_cxx0x.cpp Thu May  2 00:28:32 2013
>>>> @@ -1,5 +1,6 @@
>>>> -// RUN: %clang_cc1 -E -std=c++11 %s -o - | FileCheck --check-prefix=CHECK-0X %s
>>>> -// RUN: %clang_cc1 -E %s -o - | FileCheck --check-prefix=CHECK-NO-0X %s
>>>> +// RUN: %clang_cc1 -E -triple x86_64-linux-gnu -std=c++11 %s -o - | FileCheck --check-prefix=CHECK-0X %s
>>>> +// RUN: %clang_cc1 -E -triple armv7-apple-darwin -std=c++11 %s -o - | FileCheck --check-prefix=CHECK-NO-TLS %s
>>>> +// RUN: %clang_cc1 -E -triple x86_64-linux-gnu %s -o - | FileCheck --check-prefix=CHECK-NO-0X %s
>>>> 
>>>> #if __has_feature(cxx_atomic)
>>>> int has_atomic();
>>>> @@ -290,3 +291,4 @@ int no_thread_local();
>>>> 
>>>> // CHECK-0X: has_thread_local
>>>> // CHECK-NO-0X: no_thread_local
>>>> +// CHECK-NO-TLS: no_thread_local
>>>> 
>>>> 
>>>> _______________________________________________
>>>> cfe-commits mailing list
>>>> cfe-commits at cs.uiuc.edu
>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
>>> 
>>> -- Jean-Daniel
>>> 
>>> 
>>> 
>>> 
>> 
>> 
>> _______________________________________________
>> cfe-commits mailing list
>> cfe-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20130502/84fb2748/attachment.html>