diff mbox

[v2] libstdc++: Make certain exceptions transaction_safe.

Message ID 1453125282.26597.422.camel@localhost.localdomain
State New
Headers show

Commit Message

Torvald Riegel Jan. 18, 2016, 1:54 p.m. UTC
On Sun, 2016-01-17 at 18:30 -0500, David Edelsohn wrote:
> On Sun, Jan 17, 2016 at 3:21 PM, Torvald Riegel <triegel@redhat.com> wrote:
> > On Sat, 2016-01-16 at 15:38 -0500, David Edelsohn wrote:
> >> On Sat, Jan 16, 2016 at 8:35 AM, Jakub Jelinek <jakub@redhat.com> wrote:
> >> > On Sat, Jan 16, 2016 at 07:47:33AM -0500, David Edelsohn wrote:
> >> >> stage1 libstdc++ builds just fine.  the problem is stage2 configure
> >> >> fails due to missing ITM_xxx symbols when configure tries to compile
> >> >> and run conftest programs.
> >> >
> >> > On x86_64-linux, the _ITM_xxx symbols are undef weak ones and thus it is
> >> > fine to load libstdc++ without libitm and libstdc++ doesn't depend on
> >> > libitm.
> >> >
> >> > So, is AIX defining __GXX_WEAK__ or not?  Perhaps some other macro or
> >> > configure check needs to be used to determine if undefined weak symbols
> >> > work the way libstdc++ needs them to.
> >>
> >> __GXX_WEAK__ appears to be defined by gcc/c-family/c-cppbuiltin.c
> >> based on  SUPPORTS_ONE_ONLY.  gcc/defaults.h defines SUPPORTS_ONE_ONLY
> >> if the target supports MAKE_DECL_ONE_ONLY and link-once semantics.
> >> AIX weak correctly supports link-once semantics.  AIX also supports
> >> the definition of __GXX_WEAK__ in gcc/doc/cpp.texi, namely collapsing
> >> symbols with vague linkage in multiple translation units.
> >>
> >> libstdc++/src/c++11/cow-stdexcept.cc appears to be using __GXX_WEAK__
> >> and __attribute__ ((weak)) for references to symbols that may not be
> >> defined at link time or run time.  AIX does not allow undefined symbol
> >> errors by default.  And the libstdc++ inference about the semantics of
> >> __GXX_WEAK__ are different than the documentation.
> >>
> >> AIX supports MAKE_DECL_ONE_ONLY and the documented meaning of
> >> __GXX_WEAK__.  AIX does not support extension of the meaning to
> >> additional SVR4 semantics not specified in the documentation.
> >
> > I see, so we might be assuming that __GXX_WEAK__ means more than it
> > actually does (I'm saying "might" because personally, I don't know; your
> > information supports this is the case, but the initial info I got was
> > that __GXX_WEAK__ would mean we could have weak decls without
> > definitions).
> 
> I believe that libstdc++ must continue with the weak undefined
> references to the symbols as designed, but protect them with a
> different macro.  For example, __GXX_WEAK_REF__ or __GXX_WEAK_UNDEF__
> defined in defaults.h based on configure test or simply overridden in
> config/rs6000/aix.h.  Or the macro could be local to libstdc++ and
> overridden in config/os/aix/os_defines.h.

OK.  I'm currently testing the attached patch on x86_64-linux.  David,
if there are no objections from you and those CC'ed, could you give this
one a try on AIX, please?

	* include/bits/c++config (_GLIBCXX_USE_WEAK_REF): New.
	(_GLIBCXX_TXN_SAFE, _GLIBCXX_TXN_SAFE_DYN): Use	_GLIBCXX_USE_WEAK_REF
	and move after its definition.
	* config/os/aix/os_defines.h (_GLIBCXX_USE_WEAK_REF): Override.
	* src/c++11/cow-stdexcept.cc: Use _GLIBCXX_USE_WEAK_REF instead of
	__GXX_WEAK__, and only provide transactional clones if
	_GLIBCXX_USE_WEAK_REF is true.  Don't provide stubs of libitm
	functions.

Comments

Torvald Riegel Jan. 18, 2016, 4:30 p.m. UTC | #1
On Mon, 2016-01-18 at 14:54 +0100, Torvald Riegel wrote:
> On Sun, 2016-01-17 at 18:30 -0500, David Edelsohn wrote:
> > On Sun, Jan 17, 2016 at 3:21 PM, Torvald Riegel <triegel@redhat.com> wrote:
> > > On Sat, 2016-01-16 at 15:38 -0500, David Edelsohn wrote:
> > >> On Sat, Jan 16, 2016 at 8:35 AM, Jakub Jelinek <jakub@redhat.com> wrote:
> > >> > On Sat, Jan 16, 2016 at 07:47:33AM -0500, David Edelsohn wrote:
> > >> >> stage1 libstdc++ builds just fine.  the problem is stage2 configure
> > >> >> fails due to missing ITM_xxx symbols when configure tries to compile
> > >> >> and run conftest programs.
> > >> >
> > >> > On x86_64-linux, the _ITM_xxx symbols are undef weak ones and thus it is
> > >> > fine to load libstdc++ without libitm and libstdc++ doesn't depend on
> > >> > libitm.
> > >> >
> > >> > So, is AIX defining __GXX_WEAK__ or not?  Perhaps some other macro or
> > >> > configure check needs to be used to determine if undefined weak symbols
> > >> > work the way libstdc++ needs them to.
> > >>
> > >> __GXX_WEAK__ appears to be defined by gcc/c-family/c-cppbuiltin.c
> > >> based on  SUPPORTS_ONE_ONLY.  gcc/defaults.h defines SUPPORTS_ONE_ONLY
> > >> if the target supports MAKE_DECL_ONE_ONLY and link-once semantics.
> > >> AIX weak correctly supports link-once semantics.  AIX also supports
> > >> the definition of __GXX_WEAK__ in gcc/doc/cpp.texi, namely collapsing
> > >> symbols with vague linkage in multiple translation units.
> > >>
> > >> libstdc++/src/c++11/cow-stdexcept.cc appears to be using __GXX_WEAK__
> > >> and __attribute__ ((weak)) for references to symbols that may not be
> > >> defined at link time or run time.  AIX does not allow undefined symbol
> > >> errors by default.  And the libstdc++ inference about the semantics of
> > >> __GXX_WEAK__ are different than the documentation.
> > >>
> > >> AIX supports MAKE_DECL_ONE_ONLY and the documented meaning of
> > >> __GXX_WEAK__.  AIX does not support extension of the meaning to
> > >> additional SVR4 semantics not specified in the documentation.
> > >
> > > I see, so we might be assuming that __GXX_WEAK__ means more than it
> > > actually does (I'm saying "might" because personally, I don't know; your
> > > information supports this is the case, but the initial info I got was
> > > that __GXX_WEAK__ would mean we could have weak decls without
> > > definitions).
> > 
> > I believe that libstdc++ must continue with the weak undefined
> > references to the symbols as designed, but protect them with a
> > different macro.  For example, __GXX_WEAK_REF__ or __GXX_WEAK_UNDEF__
> > defined in defaults.h based on configure test or simply overridden in
> > config/rs6000/aix.h.  Or the macro could be local to libstdc++ and
> > overridden in config/os/aix/os_defines.h.
> 
> OK.  I'm currently testing the attached patch on x86_64-linux.

No regressions in the libstdc++ and libitm tests on x86_64-linux.
Jonathan Wakely Jan. 18, 2016, 8:07 p.m. UTC | #2
On 18/01/16 17:30 +0100, Torvald Riegel wrote:
>On Mon, 2016-01-18 at 14:54 +0100, Torvald Riegel wrote:
>> On Sun, 2016-01-17 at 18:30 -0500, David Edelsohn wrote:
>> > On Sun, Jan 17, 2016 at 3:21 PM, Torvald Riegel <triegel@redhat.com> wrote:
>> > > On Sat, 2016-01-16 at 15:38 -0500, David Edelsohn wrote:
>> > >> On Sat, Jan 16, 2016 at 8:35 AM, Jakub Jelinek <jakub@redhat.com> wrote:
>> > >> > On Sat, Jan 16, 2016 at 07:47:33AM -0500, David Edelsohn wrote:
>> > >> >> stage1 libstdc++ builds just fine.  the problem is stage2 configure
>> > >> >> fails due to missing ITM_xxx symbols when configure tries to compile
>> > >> >> and run conftest programs.
>> > >> >
>> > >> > On x86_64-linux, the _ITM_xxx symbols are undef weak ones and thus it is
>> > >> > fine to load libstdc++ without libitm and libstdc++ doesn't depend on
>> > >> > libitm.
>> > >> >
>> > >> > So, is AIX defining __GXX_WEAK__ or not?  Perhaps some other macro or
>> > >> > configure check needs to be used to determine if undefined weak symbols
>> > >> > work the way libstdc++ needs them to.
>> > >>
>> > >> __GXX_WEAK__ appears to be defined by gcc/c-family/c-cppbuiltin.c
>> > >> based on  SUPPORTS_ONE_ONLY.  gcc/defaults.h defines SUPPORTS_ONE_ONLY
>> > >> if the target supports MAKE_DECL_ONE_ONLY and link-once semantics.
>> > >> AIX weak correctly supports link-once semantics.  AIX also supports
>> > >> the definition of __GXX_WEAK__ in gcc/doc/cpp.texi, namely collapsing
>> > >> symbols with vague linkage in multiple translation units.
>> > >>
>> > >> libstdc++/src/c++11/cow-stdexcept.cc appears to be using __GXX_WEAK__
>> > >> and __attribute__ ((weak)) for references to symbols that may not be
>> > >> defined at link time or run time.  AIX does not allow undefined symbol
>> > >> errors by default.  And the libstdc++ inference about the semantics of
>> > >> __GXX_WEAK__ are different than the documentation.
>> > >>
>> > >> AIX supports MAKE_DECL_ONE_ONLY and the documented meaning of
>> > >> __GXX_WEAK__.  AIX does not support extension of the meaning to
>> > >> additional SVR4 semantics not specified in the documentation.
>> > >
>> > > I see, so we might be assuming that __GXX_WEAK__ means more than it
>> > > actually does (I'm saying "might" because personally, I don't know; your
>> > > information supports this is the case, but the initial info I got was
>> > > that __GXX_WEAK__ would mean we could have weak decls without
>> > > definitions).
>> >
>> > I believe that libstdc++ must continue with the weak undefined
>> > references to the symbols as designed, but protect them with a
>> > different macro.  For example, __GXX_WEAK_REF__ or __GXX_WEAK_UNDEF__
>> > defined in defaults.h based on configure test or simply overridden in
>> > config/rs6000/aix.h.  Or the macro could be local to libstdc++ and
>> > overridden in config/os/aix/os_defines.h.
>>
>> OK.  I'm currently testing the attached patch on x86_64-linux.
>
>No regressions in the libstdc++ and libitm tests on x86_64-linux.

Builds OK on AIX, OK for trunk, thanks.
diff mbox

Patch

commit a5a8819bce824815a94ef8d58f6d4123db92f1d4
Author: Torvald Riegel <triegel@redhat.com>
Date:   Mon Jan 18 14:42:21 2016 +0100

    libstdc++: Fix usage of __GXX_WEAK__ in TM TS support.
    
    	* include/bits/c++config (_GLIBCXX_USE_WEAK_REF): New.
    	(_GLIBCXX_TXN_SAFE, _GLIBCXX_TXN_SAFE_DYN): Use	_GLIBCXX_USE_WEAK_REF
    	and move after its definition.
    	* config/os/aix/os_defines.h (_GLIBCXX_USE_WEAK_REF): Override.
    	* src/c++11/cow-stdexcept.cc: Use _GLIBCXX_USE_WEAK_REF instead of
    	__GXX_WEAK__, and only provide transactional clones if
    	_GLIBCXX_USE_WEAK_REF is true.  Don't provide stubs of libitm
    	functions.

diff --git a/libstdc++-v3/config/os/aix/os_defines.h b/libstdc++-v3/config/os/aix/os_defines.h
index d895471..0949446 100644
--- a/libstdc++-v3/config/os/aix/os_defines.h
+++ b/libstdc++-v3/config/os/aix/os_defines.h
@@ -48,4 +48,7 @@ 
 #define __COMPATMATH__
 #endif
 
+// No support for referencing weak symbols without a definition.
+#define _GLIBCXX_USE_WEAK_REF 0
+
 #endif
diff --git a/libstdc++-v3/include/bits/c++config b/libstdc++-v3/include/bits/c++config
index 387a7bb..57024e4 100644
--- a/libstdc++-v3/include/bits/c++config
+++ b/libstdc++-v3/include/bits/c++config
@@ -483,20 +483,6 @@  namespace std
 
 #define _GLIBCXX_USE_ALLOCATOR_NEW
 
-// Conditionally enable annotations for the Transactional Memory TS on C++11.
-// Most of the following conditions are due to limitations in the current
-// implementation.
-#if __cplusplus >= 201103L && _GLIBCXX_USE_CXX11_ABI			\
-  && _GLIBCXX_USE_DUAL_ABI && __cpp_transactional_memory >= 201505L	\
-  &&  !_GLIBCXX_FULLY_DYNAMIC_STRING && __GXX_WEAK__ 			\
-  && _GLIBCXX_USE_ALLOCATOR_NEW
-#define _GLIBCXX_TXN_SAFE transaction_safe
-#define _GLIBCXX_TXN_SAFE_DYN transaction_safe_dynamic
-#else
-#define _GLIBCXX_TXN_SAFE
-#define _GLIBCXX_TXN_SAFE_DYN
-#endif
-
 #else // !__cplusplus
 # define _GLIBCXX_BEGIN_EXTERN_C
 # define _GLIBCXX_END_EXTERN_C
@@ -524,6 +510,28 @@  namespace std
 # define _GLIBCXX_WEAK_DEFINITION
 #endif
 
+// By default, we assume that __GXX_WEAK__ also means that there is support
+// for declaring functions as weak while not defining such functions.  This
+// allows for referring to functions provided by other libraries (e.g.,
+// libitm) without depending on them if the respective features are not used.
+#ifndef _GLIBCXX_USE_WEAK_REF
+# define _GLIBCXX_USE_WEAK_REF __GXX_WEAK__
+#endif
+
+// Conditionally enable annotations for the Transactional Memory TS on C++11.
+// Most of the following conditions are due to limitations in the current
+// implementation.
+#if __cplusplus >= 201103L && _GLIBCXX_USE_CXX11_ABI			\
+  && _GLIBCXX_USE_DUAL_ABI && __cpp_transactional_memory >= 201505L	\
+  &&  !_GLIBCXX_FULLY_DYNAMIC_STRING && _GLIBCXX_USE_WEAK_REF		\
+  && _GLIBCXX_USE_ALLOCATOR_NEW
+#define _GLIBCXX_TXN_SAFE transaction_safe
+#define _GLIBCXX_TXN_SAFE_DYN transaction_safe_dynamic
+#else
+#define _GLIBCXX_TXN_SAFE
+#define _GLIBCXX_TXN_SAFE_DYN
+#endif
+
 
 // The remainder of the prewritten config is automatic; all the
 // user hooks are listed above.
diff --git a/libstdc++-v3/src/c++11/cow-stdexcept.cc b/libstdc++-v3/src/c++11/cow-stdexcept.cc
index eddb36b..a0f505c 100644
--- a/libstdc++-v3/src/c++11/cow-stdexcept.cc
+++ b/libstdc++-v3/src/c++11/cow-stdexcept.cc
@@ -179,6 +179,13 @@  _GLIBCXX_END_NAMESPACE_VERSION
 // Furthermore, _Rep will always have been allocated or deallocated via
 // global new or delete, so nontransactional writes we do to _Rep cannot
 // interfere with transactional accesses.
+
+// We depend on having support for referencing functions declared weak that
+// are not defined by us.  Without such support, the exceptions will not be
+// declared transaction-safe, so we just don't provide transactional clones
+// in this case.
+#if _GLIBCXX_USE_WEAK_REF
+
 extern "C" {
 
 #ifndef _GLIBCXX_MANGLE_SIZE_T
@@ -195,7 +202,6 @@  extern "C" {
 # define ITM_REGPARM
 #endif
 
-#if __GXX_WEAK__
 // Declare all libitm symbols we rely on, but make them weak so that we do
 // not depend on libitm.
 extern void* _ZGTtnaX (size_t sz) __attribute__((weak));
@@ -213,19 +219,6 @@  extern void _ITM_memcpyRnWt(void *, const void *, size_t)
 extern void _ITM_addUserCommitAction(void (*)(void *), uint64_t, void *)
   ITM_REGPARM __attribute__((weak));
 
-#else
-// If there is no support for weak symbols, create dummies.  The exceptions
-// will not be declared transaction_safe in this case.
-void* _ZGTtnaX (size_t) { return NULL; }
-void _ZGTtdlPv (void*) { }
-uint8_t _ITM_RU1(const uint8_t *) { return 0; }
-uint32_t _ITM_RU4(const uint32_t *) { return 0; }
-uint64_t _ITM_RU8(const uint64_t *) { return 0; }
-void _ITM_memcpyRtWn(void *, const void *, size_t) { }
-void _ITM_memcpyRnWt(void *, const void *, size_t) { }
-void _ITM_addUserCommitAction(void (*)(void *), uint64_t, void *) { };
-#endif
-
 }
 
 // A transactional version of basic_string::basic_string(const char *s)
@@ -441,3 +434,5 @@  CTORDTOR(14overflow_error, std::overflow_error, runtime_error)
 CTORDTOR(15underflow_error, std::underflow_error, runtime_error)
 
 }
+
+#endif  // _GLIBCXX_USE_WEAK_REF