Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
This revision was automatically updated to reflect the committed changes.
Closed by commit rL271267: [libcxxabi] Introduce a -fno-exceptions libc++abi
libary variant (authored by asiri).
Changed prior to commit:
http://reviews.llvm.org/D20677?vs=59009=59044#toc
Repository:
rL LLVM
http://reviews.llvm.org/D20677
Files:
libcxxabi/trunk/CMakeLists.txt
libcxxabi/trunk/src/CMakeLists.txt
libcxxabi/trunk/src/cxa_aux_runtime.cpp
libcxxabi/trunk/src/cxa_handlers.cpp
libcxxabi/trunk/src/cxa_new_delete.cpp
libcxxabi/trunk/src/cxa_noexception.cpp
libcxxabi/trunk/test/CMakeLists.txt
libcxxabi/trunk/test/backtrace_test.pass.cpp
libcxxabi/trunk/test/catch_array_01.pass.cpp
libcxxabi/trunk/test/catch_array_02.pass.cpp
libcxxabi/trunk/test/catch_class_01.pass.cpp
libcxxabi/trunk/test/catch_class_02.pass.cpp
libcxxabi/trunk/test/catch_class_03.pass.cpp
libcxxabi/trunk/test/catch_class_04.pass.cpp
libcxxabi/trunk/test/catch_const_pointer_nullptr.pass.cpp
libcxxabi/trunk/test/catch_function_01.pass.cpp
libcxxabi/trunk/test/catch_function_02.pass.cpp
libcxxabi/trunk/test/catch_in_noexcept.pass.cpp
libcxxabi/trunk/test/catch_member_data_pointer_01.pass.cpp
libcxxabi/trunk/test/catch_member_function_pointer_01.pass.cpp
libcxxabi/trunk/test/catch_member_pointer_nullptr.pass.cpp
libcxxabi/trunk/test/catch_multi_level_pointer.pass.cpp
libcxxabi/trunk/test/catch_pointer_nullptr.pass.cpp
libcxxabi/trunk/test/catch_pointer_reference.pass.cpp
libcxxabi/trunk/test/catch_ptr.pass.cpp
libcxxabi/trunk/test/catch_ptr_02.pass.cpp
libcxxabi/trunk/test/cxa_bad_cast.pass.cpp
libcxxabi/trunk/test/cxa_bad_typeid.pass.cpp
libcxxabi/trunk/test/incomplete_type.sh.cpp
libcxxabi/trunk/test/inherited_exception.pass.cpp
libcxxabi/trunk/test/libcxxabi/test/config.py
libcxxabi/trunk/test/lit.site.cfg.in
libcxxabi/trunk/test/noexception1.pass.cpp
libcxxabi/trunk/test/noexception2.pass.cpp
libcxxabi/trunk/test/noexception3.pass.cpp
libcxxabi/trunk/test/noexception4.pass.cpp
libcxxabi/trunk/test/test_aux_runtime.pass.cpp
libcxxabi/trunk/test/test_aux_runtime_op_array_new.pass.cpp
libcxxabi/trunk/test/test_guard.pass.cpp
libcxxabi/trunk/test/test_vector1.pass.cpp
libcxxabi/trunk/test/test_vector2.pass.cpp
libcxxabi/trunk/test/test_vector3.pass.cpp
libcxxabi/trunk/test/uncaught_exceptions.pass.cpp
libcxxabi/trunk/test/unwind_01.pass.cpp
libcxxabi/trunk/test/unwind_02.pass.cpp
libcxxabi/trunk/test/unwind_03.pass.cpp
libcxxabi/trunk/test/unwind_04.pass.cpp
libcxxabi/trunk/test/unwind_05.pass.cpp
libcxxabi/trunk/test/unwind_06.pass.cpp
Index: libcxxabi/trunk/test/catch_function_01.pass.cpp
===
--- libcxxabi/trunk/test/catch_function_01.pass.cpp
+++ libcxxabi/trunk/test/catch_function_01.pass.cpp
@@ -12,6 +12,7 @@
// GCC incorrectly allows function pointer to be caught by reference.
// See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69372
// XFAIL: gcc
+// UNSUPPORTED: libcxxabi-no-exceptions
#include
Index: libcxxabi/trunk/test/test_vector3.pass.cpp
===
--- libcxxabi/trunk/test/test_vector3.pass.cpp
+++ libcxxabi/trunk/test/test_vector3.pass.cpp
@@ -7,6 +7,8 @@
//
//===--===//
+// UNSUPPORTED: libcxxabi-no-exceptions
+
#include "cxxabi.h"
#include
Index: libcxxabi/trunk/test/test_vector1.pass.cpp
===
--- libcxxabi/trunk/test/test_vector1.pass.cpp
+++ libcxxabi/trunk/test/test_vector1.pass.cpp
@@ -47,8 +47,19 @@
int gConstructorThrowTarget;
int gDestructorCounter;
int gDestructorThrowTarget;
-void throw_construct ( void *p ) { if ( gConstructorCounter == gConstructorThrowTarget ) throw 1; ++gConstructorCounter; }
-void throw_destruct ( void *p ) { if ( ++gDestructorCounter == gDestructorThrowTarget ) throw 2; }
+void throw_construct ( void *p ) {
+#ifndef LIBCXXABI_HAS_NO_EXCEPTIONS
+if ( gConstructorCounter == gConstructorThrowTarget )
+throw 1;
+++gConstructorCounter;
+#endif
+}
+void throw_destruct ( void *p ) {
+#ifndef LIBCXXABI_HAS_NO_EXCEPTIONS
+if ( ++gDestructorCounter == gDestructorThrowTarget )
+throw 2;
+#endif
+}
#if __cplusplus >= 201103L
# define CAN_THROW noexcept(false)
@@ -146,6 +157,7 @@
return retVal;
}
+#ifndef LIBCXXABI_HAS_NO_EXCEPTIONS
// Make sure the constructors and destructors are matched
int test_exception_in_constructor ( ) {
int retVal = 0;
@@ -202,7 +214,9 @@
return retVal;
}
+#endif
+#ifndef LIBCXXABI_HAS_NO_EXCEPTIONS
// Make sure the constructors and destructors are matched
int test_exception_in_destructor ( ) {
int retVal = 0;
@@ -253,12 +267,15 @@
return retVal;
}
+#endif
int main ( int argc, char *argv [] ) {
int retVal = 0;
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment. In http://reviews.llvm.org/D20677#443997, @rmaprath wrote: > Thanks Eric. > > I think I noted a few other `assert(true)` statements elsewhere (with similar > application), will update those separately. > > Btw, do you use some systematic way to detect ABI breakages? I've only bumped > into http://lvc.github.io/abi-compliance-checker/ but didn't go as far as > setting it up. Not really, the two primary ways I try to detect ABI breakage are using: - the scripts under `libcxx/utils/sym_check/`. They provide a way to diff the symbols between two dylibs `sym_diff.py old-lib.so new-lib.so`. - Compiling against an old library and swapping in a new library at runtime. I also used libabigail for a while but it was a bit too noisy with false positives. It may be better now though. > / Asiri http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. Thanks Eric. I think I noted a few other `assert(true)` statements elsewhere (with similar application), will update those separately. Btw, do you use some systematic way to detect ABI breakages? I've only bumped into http://lvc.github.io/abi-compliance-checker/ but didn't go as far as setting it up. / Asiri http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF accepted this revision.
EricWF added a comment.
This revision is now accepted and ready to land.
Just for reference here is a list of symbols which have been removed:
https://gist.github.com/EricWF/bf00dbc5fccc78b779f8d260727a0710
LGTM. Your free to submit it after addressing the inline comments.
Comment at: test/cxa_bad_cast.pass.cpp:40
@@ +39,3 @@
+} catch (std::bad_cast) {
+assert(true);
+} catch (...) {
return 0;
Comment at: test/cxa_bad_cast.pass.cpp:45
@@ +44,3 @@
+#endif
+return 0;
+}
return 1;
Comment at: test/cxa_bad_typeid.pass.cpp:39
@@ +38,3 @@
+} catch (std::bad_typeid) {
+assert(true);
+} catch (...) {
return 0 here instead. "assert(true)" is pretty much un-observable.
Comment at: test/cxa_bad_typeid.pass.cpp:44
@@ +43,3 @@
+#endif
+return 0;
+}
This should return non-zero.
Comment at: test/noexception1.pass.cpp:26
@@ +25,3 @@
+int main ()
+{
+std::set_terminate(my_terminate);
Before setting the terminate handling call the function once with null to make
sure it returns.
Comment at: test/noexception2.pass.cpp:26
@@ +25,3 @@
+int main ()
+{
+std::set_terminate(my_terminate);
Before setting the terminate handling call the function once with null to make
sure it returns.
Comment at: test/noexception3.pass.cpp:27
@@ +26,3 @@
+{
+std::set_terminate(my_terminate);
+__cxxabiv1::__cxa_rethrow_primary_exception((void*) );
Before setting the terminate handling call the function once with null to make
sure it returns.
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment. Nevermind. I figured out why it wasn't working. Phab didn't know the libc++ patch was in a different repo and so it wouldn't apply the patch. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment. For some reason I'm having trouble applying your patch. Could you please upload it with more context? (ie git diff -U999 or similar) http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath marked 6 inline comments as done. rmaprath added a comment. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath updated this revision to Diff 59009. rmaprath added a comment. Addressed review comments from @EricWF: - Addressed the nits - Disabled `backtrace_test.pass.cpp` for the moment, need to check if using `_Unwind_backtrace` with no unwinding tables (-fno-exceptions) makes sense - Enabled a small part of `test_vector1.pass.cpp`, most of this test and other vector tests are to do with how exceptions are dealt with - Added tests for the new functions introduced to `cxa_noexceptions.cpp`. These are mostly trivial. - Added tests for the `__cxa_bad_cast` and `__cxa_bad_typeid` functions. Could not manage to invoke `__cxa_throw_bad_array_new_length`, which does not seem to work like documented in [1]. Need to take this offline. [1] http://en.cppreference.com/w/cpp/memory/new/bad_array_new_length http://reviews.llvm.org/D20677 Files: CMakeLists.txt src/CMakeLists.txt src/cxa_aux_runtime.cpp src/cxa_handlers.cpp src/cxa_new_delete.cpp src/cxa_noexception.cpp test/CMakeLists.txt test/backtrace_test.pass.cpp test/catch_array_01.pass.cpp test/catch_array_02.pass.cpp test/catch_class_01.pass.cpp test/catch_class_02.pass.cpp test/catch_class_03.pass.cpp test/catch_class_04.pass.cpp test/catch_const_pointer_nullptr.pass.cpp test/catch_function_01.pass.cpp test/catch_function_02.pass.cpp test/catch_in_noexcept.pass.cpp test/catch_member_data_pointer_01.pass.cpp test/catch_member_function_pointer_01.pass.cpp test/catch_member_pointer_nullptr.pass.cpp test/catch_multi_level_pointer.pass.cpp test/catch_pointer_nullptr.pass.cpp test/catch_pointer_reference.pass.cpp test/catch_ptr.pass.cpp test/catch_ptr_02.pass.cpp test/cxa_bad_cast.pass.cpp test/cxa_bad_typeid.pass.cpp test/incomplete_type.sh.cpp test/inherited_exception.pass.cpp test/libcxxabi/test/config.py test/lit.site.cfg.in test/noexception1.pass.cpp test/noexception2.pass.cpp test/noexception3.pass.cpp test/noexception4.pass.cpp test/test_aux_runtime.pass.cpp test/test_aux_runtime_op_array_new.pass.cpp test/test_guard.pass.cpp test/test_vector1.pass.cpp test/test_vector2.pass.cpp test/test_vector3.pass.cpp test/uncaught_exceptions.pass.cpp test/unwind_01.pass.cpp test/unwind_02.pass.cpp test/unwind_03.pass.cpp test/unwind_04.pass.cpp test/unwind_05.pass.cpp test/unwind_06.pass.cpp Index: test/unwind_06.pass.cpp === --- test/unwind_06.pass.cpp +++ test/unwind_06.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_05.pass.cpp === --- test/unwind_05.pass.cpp +++ test/unwind_05.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_04.pass.cpp === --- test/unwind_04.pass.cpp +++ test/unwind_04.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_03.pass.cpp === --- test/unwind_03.pass.cpp +++ test/unwind_03.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_02.pass.cpp === --- test/unwind_02.pass.cpp +++ test/unwind_02.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/unwind_01.pass.cpp === --- test/unwind_01.pass.cpp +++ test/unwind_01.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/uncaught_exceptions.pass.cpp === --- test/uncaught_exceptions.pass.cpp +++ test/uncaught_exceptions.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/test_vector3.pass.cpp === --- test/test_vector3.pass.cpp +++ test/test_vector3.pass.cpp @@ -7,6 +7,8 @@ // //===--===//
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment.
Looking good. I would like to see some tests for the methods we have specificly
defined to have behavior in -fno-exceptions mode.
Comment at: src/cxa_noexception.cpp:27
@@ +26,3 @@
+
+_LIBCXXABI_FUNC_VIS void
+__cxa_increment_exception_refcount(void *thrown_object) throw() {
I think either all of the functions in this file should have
`_LIBCXXABI_FUNC_VIS` or none of them should. My preference is none since the
declarations should have them.
Comment at: src/cxa_noexception.cpp:29
@@ +28,3 @@
+__cxa_increment_exception_refcount(void *thrown_object) throw() {
+if (thrown_object != NULL)
+std::terminate();
nit: nullptr
Comment at: src/cxa_noexception.cpp:45
@@ +44,3 @@
+void
+__cxa_rethrow_primary_exception(void*) { return; }
+
I think we should terminate if `void*` is non-null.
Comment at: test/backtrace_test.pass.cpp:19
@@ -19,2 +19,2 @@
}
Does this test have any value in -fno-exceptions mode?
Comment at: test/libcxxabi/test/config.py:46
@@ +45,3 @@
+if self.get_lit_bool('enable_exceptions', True):
+self.cxx.compile_flags += ['-funwind-tables',
'-DLIBCXXABI_ENABLE_EXCEPTIONS']
+else:
Nit: I would prefer `LIBCXXABI_HAS_NO_EXCEPTIONS` since the default should be
on. The fact that a macro is missing shouldn't disable half the tests.
Comment at: test/test_vector1.pass.cpp:8
@@ -7,3 +7,3 @@
//
//===--===//
Don't you want these tests to run? They aren't part of the exception API AFAIK.
Comment at: test/uncaught_exceptions.pass.cpp:10
@@ -9,1 +9,3 @@
+// UNSUPPORTED: libcxxabi-no-exceptions
+
We provided a different implementation so we should test it, not disable the
test.
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath updated this revision to Diff 59000. rmaprath added a comment. Addressing review comments from @EricWF: - Rather than explicitly decoupling the no-exceptions `libc++` library from the `__cxa_*` routines (http://reviews.llvm.org/D20784), provide a placeholder implementation of those functions within the no-exceptions `libc++abi` library. We still have some symbols like `__cxa_throw` removed from the no-exceptions `libc++abi` library (those symbols are generated from the compiler, rather than found in the library), so the original requirement of not allowing linking with-exceptions code with no-exceptions `libc++abi` library is satisfied. http://reviews.llvm.org/D20677 Files: CMakeLists.txt src/CMakeLists.txt src/cxa_aux_runtime.cpp src/cxa_handlers.cpp src/cxa_new_delete.cpp src/cxa_noexception.cpp test/CMakeLists.txt test/backtrace_test.pass.cpp test/catch_array_01.pass.cpp test/catch_array_02.pass.cpp test/catch_class_01.pass.cpp test/catch_class_02.pass.cpp test/catch_class_03.pass.cpp test/catch_class_04.pass.cpp test/catch_const_pointer_nullptr.pass.cpp test/catch_function_01.pass.cpp test/catch_function_02.pass.cpp test/catch_in_noexcept.pass.cpp test/catch_member_data_pointer_01.pass.cpp test/catch_member_function_pointer_01.pass.cpp test/catch_member_pointer_nullptr.pass.cpp test/catch_multi_level_pointer.pass.cpp test/catch_pointer_nullptr.pass.cpp test/catch_pointer_reference.pass.cpp test/catch_ptr.pass.cpp test/catch_ptr_02.pass.cpp test/incomplete_type.sh.cpp test/inherited_exception.pass.cpp test/libcxxabi/test/config.py test/lit.site.cfg.in test/test_aux_runtime.pass.cpp test/test_aux_runtime_op_array_new.pass.cpp test/test_guard.pass.cpp test/test_vector1.pass.cpp test/test_vector2.pass.cpp test/test_vector3.pass.cpp test/uncaught_exceptions.pass.cpp test/unwind_01.pass.cpp test/unwind_02.pass.cpp test/unwind_03.pass.cpp test/unwind_04.pass.cpp test/unwind_05.pass.cpp test/unwind_06.pass.cpp Index: test/unwind_06.pass.cpp === --- test/unwind_06.pass.cpp +++ test/unwind_06.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_05.pass.cpp === --- test/unwind_05.pass.cpp +++ test/unwind_05.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_04.pass.cpp === --- test/unwind_04.pass.cpp +++ test/unwind_04.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_03.pass.cpp === --- test/unwind_03.pass.cpp +++ test/unwind_03.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_02.pass.cpp === --- test/unwind_02.pass.cpp +++ test/unwind_02.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/unwind_01.pass.cpp === --- test/unwind_01.pass.cpp +++ test/unwind_01.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/uncaught_exceptions.pass.cpp === --- test/uncaught_exceptions.pass.cpp +++ test/uncaught_exceptions.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/test_vector3.pass.cpp === --- test/test_vector3.pass.cpp +++ test/test_vector3.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector2.pass.cpp === --- test/test_vector2.pass.cpp +++ test/test_vector2.pass.cpp @@ -7,6 +7,8 @@ //
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath marked an inline comment as done. rmaprath added a comment. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath updated this revision to Diff 58929. rmaprath added a comment. Fixed a few typos. http://reviews.llvm.org/D20677 Files: CMakeLists.txt src/CMakeLists.txt src/cxa_aux_runtime.cpp src/cxa_handlers.cpp src/cxa_new_delete.cpp test/CMakeLists.txt test/backtrace_test.pass.cpp test/catch_array_01.pass.cpp test/catch_array_02.pass.cpp test/catch_class_01.pass.cpp test/catch_class_02.pass.cpp test/catch_class_03.pass.cpp test/catch_class_04.pass.cpp test/catch_const_pointer_nullptr.pass.cpp test/catch_function_01.pass.cpp test/catch_function_02.pass.cpp test/catch_in_noexcept.pass.cpp test/catch_member_data_pointer_01.pass.cpp test/catch_member_function_pointer_01.pass.cpp test/catch_member_pointer_nullptr.pass.cpp test/catch_multi_level_pointer.pass.cpp test/catch_pointer_nullptr.pass.cpp test/catch_pointer_reference.pass.cpp test/catch_ptr.pass.cpp test/catch_ptr_02.pass.cpp test/incomplete_type.sh.cpp test/inherited_exception.pass.cpp test/libcxxabi/test/config.py test/lit.site.cfg.in test/test_aux_runtime.pass.cpp test/test_aux_runtime_op_array_new.pass.cpp test/test_guard.pass.cpp test/test_vector1.pass.cpp test/test_vector2.pass.cpp test/test_vector3.pass.cpp test/uncaught_exceptions.pass.cpp test/unwind_01.pass.cpp test/unwind_02.pass.cpp test/unwind_03.pass.cpp test/unwind_04.pass.cpp test/unwind_05.pass.cpp test/unwind_06.pass.cpp Index: test/unwind_06.pass.cpp === --- test/unwind_06.pass.cpp +++ test/unwind_06.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_05.pass.cpp === --- test/unwind_05.pass.cpp +++ test/unwind_05.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_04.pass.cpp === --- test/unwind_04.pass.cpp +++ test/unwind_04.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_03.pass.cpp === --- test/unwind_03.pass.cpp +++ test/unwind_03.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_02.pass.cpp === --- test/unwind_02.pass.cpp +++ test/unwind_02.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/unwind_01.pass.cpp === --- test/unwind_01.pass.cpp +++ test/unwind_01.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/uncaught_exceptions.pass.cpp === --- test/uncaught_exceptions.pass.cpp +++ test/uncaught_exceptions.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/test_vector3.pass.cpp === --- test/test_vector3.pass.cpp +++ test/test_vector3.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector2.pass.cpp === --- test/test_vector2.pass.cpp +++ test/test_vector2.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector1.pass.cpp === --- test/test_vector1.pass.cpp +++ test/test_vector1.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_guard.pass.cpp === --- test/test_guard.pass.cpp +++ test/test_guard.pass.cpp
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added inline comments. Comment at: CMakeLists.txt:111 @@ -110,2 +110,3 @@ # Define options. +option(LIBCXXABI_ENABLE_EXCEPTIONS "Use exceptions." OFF) option(LIBCXXABI_ENABLE_ASSERTIONS "Enable assertions independent of build mode." ON) My bad, this should be ON by default (leftover from testing). http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath updated this revision to Diff 58927. rmaprath added a comment. OK, managed to solve this problem by removing `cxa_exception.cpp` and `cxa_personality.cpp` from the no-exceptions libcxxabi build. In order to do this, a small patch was needed for `exception.cpp` of libcxx: http://reviews.llvm.org/D20784 @EricWF, @mclow.lists: WDYT? Thanks. / Asiri http://reviews.llvm.org/D20677 Files: CMakeLists.txt src/CMakeLists.txt src/cxa_aux_runtime.cpp src/cxa_handlers.cpp src/cxa_new_delete.cpp test/CMakeLists.txt test/backtrace_test.pass.cpp test/catch_array_01.pass.cpp test/catch_array_02.pass.cpp test/catch_class_01.pass.cpp test/catch_class_02.pass.cpp test/catch_class_03.pass.cpp test/catch_class_04.pass.cpp test/catch_const_pointer_nullptr.pass.cpp test/catch_function_01.pass.cpp test/catch_function_02.pass.cpp test/catch_in_noexcept.pass.cpp test/catch_member_data_pointer_01.pass.cpp test/catch_member_function_pointer_01.pass.cpp test/catch_member_pointer_nullptr.pass.cpp test/catch_multi_level_pointer.pass.cpp test/catch_pointer_nullptr.pass.cpp test/catch_pointer_reference.pass.cpp test/catch_ptr.pass.cpp test/catch_ptr_02.pass.cpp test/incomplete_type.sh.cpp test/inherited_exception.pass.cpp test/libcxxabi/test/config.py test/lit.site.cfg.in test/test_aux_runtime.pass.cpp test/test_aux_runtime_op_array_new.pass.cpp test/test_guard.pass.cpp test/test_vector1.pass.cpp test/test_vector2.pass.cpp test/test_vector3.pass.cpp test/uncaught_exceptions.pass.cpp test/unwind_01.pass.cpp test/unwind_02.pass.cpp test/unwind_03.pass.cpp test/unwind_04.pass.cpp test/unwind_05.pass.cpp test/unwind_06.pass.cpp Index: test/unwind_06.pass.cpp === --- test/unwind_06.pass.cpp +++ test/unwind_06.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_05.pass.cpp === --- test/unwind_05.pass.cpp +++ test/unwind_05.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_04.pass.cpp === --- test/unwind_04.pass.cpp +++ test/unwind_04.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_03.pass.cpp === --- test/unwind_03.pass.cpp +++ test/unwind_03.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_02.pass.cpp === --- test/unwind_02.pass.cpp +++ test/unwind_02.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/unwind_01.pass.cpp === --- test/unwind_01.pass.cpp +++ test/unwind_01.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/uncaught_exceptions.pass.cpp === --- test/uncaught_exceptions.pass.cpp +++ test/uncaught_exceptions.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/test_vector3.pass.cpp === --- test/test_vector3.pass.cpp +++ test/test_vector3.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector2.pass.cpp === --- test/test_vector2.pass.cpp +++ test/test_vector2.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector1.pass.cpp === --- test/test_vector1.pass.cpp +++ test/test_vector1.pass.cpp @@ -7,6 +7,8 @@ //
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. > I have a small concern about how difficult this would make things for us > though - our toolchain setup uses `-ffunction-sections` and `-fdata-sections` > along with linker's unused-section elimination to get rid of most of the > stuff not necessary for the final image, but I can't remember if > unused-section elimination happens before or after checking for missing > symbols. If it is the latter, then we might be in trouble with this approach. > I will have a look into this when I get back to work. Come to think of it, I'm 90% sure it's the latter (I don't see how the former can work - the linker has to load all the objects while resolving all the dependencies, then only it can get rid of the stuff the final image doesn't really need). The linker will give up as soon as it hits a missing symbol referred from a section which is actually unused in the final image (but is within some object which contains sections that are actually needed for the image). Of course I need to confirm that 10% when I get my hands on the linker, but I have a feeling this is going to be a blocker for us. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. In http://reviews.llvm.org/D20677#442713, @EricWF wrote: > In http://reviews.llvm.org/D20677#442687, @rmaprath wrote: > > > In http://reviews.llvm.org/D20677#442559, @EricWF wrote: > > > > > I have an issue with this change since it allows a libc++abi built > > > without exceptions to be used in a program compiled with them. I assert > > > that this should not be supported in any way. > > > > > > My personal preference would be to remove as much of the exception API > > > from the library when it's built without exceptions. This should cause > > > link errors if a program contains any code that throws. If this is not > > > done it creates bugs within code like `__terminate` as mentioned in an > > > inline comment. > > > > > > I would like to see another revision of this patch that removes as much > > > of the exception API as possible. Looking at the symbols defined in > > > `cxxabi.h` I would say ever symbol in sections 2.4 and 2.5 should not > > > longer be defined in the library. > > > > > > That sounds generally OK to me. The toolchain I'm working on (ARM Compiler > > 6) has an elaborate mechanism (based on build attributes) to select the > > correct variant of libraries. In other words, if any of your objects are > > compiled with exceptions, you are guaranteed to no select any of the > > no-exceptions library variants. This avoids the problem you mentioned > > altogether. > > > Note that that behavior will cause different issues in some cases. If you > link an executable to two libraries, `libfoo.so` and `libbar.so`, where one > was compiled with exceptions and the other without you'll get two copies of > libc++abi in the program. Now you have two copies of the global terminate > handler and two copies of `__terminate` for the linker to select from. > If it selects the nothrow `__terminate` and the terminate handler throws an > exception your in trouble. In this case, I expect the linker to only select one variant of libc++abi (the one with exceptions). The idea behind build-attributes is to capture user intentions, if the user links his program with `libbar.a` and `libfoo.a` user libraries (with and without exceptions), the linker would assume that the user meant to use exceptions rather than not, and hence select the with-exceptions libc++abi variant. The theory behind build-attributes is bit more complicated that that though, I'm not the best person to go on about this. In any case, I agree that getting the linker to produce an error (when linking with-exceptions objects with a no-exceptions libc++abi) is much more clearer. I have a small concern about how difficult this would make things for us though - our toolchain setup uses `-ffunction-sections` and `-fdata-sections` along with linker's unused-section elimination to get rid of most of the stuff not necessary for the final image, but I can't remember if unused-section elimination happens before or after checking for missing symbols. If it is the latter, then we might be in trouble with this approach. I will have a look into this when I get back to work. Cheers, / Asiri http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment. In http://reviews.llvm.org/D20677#442687, @rmaprath wrote: > In http://reviews.llvm.org/D20677#442559, @EricWF wrote: > > > I have an issue with this change since it allows a libc++abi built without > > exceptions to be used in a program compiled with them. I assert that this > > should not be supported in any way. > > > > My personal preference would be to remove as much of the exception API from > > the library when it's built without exceptions. This should cause link > > errors if a program contains any code that throws. If this is not done it > > creates bugs within code like `__terminate` as mentioned in an inline > > comment. > > > > I would like to see another revision of this patch that removes as much of > > the exception API as possible. Looking at the symbols defined in `cxxabi.h` > > I would say ever symbol in sections 2.4 and 2.5 should not longer be > > defined in the library. > > > That sounds generally OK to me. The toolchain I'm working on (ARM Compiler 6) > has an elaborate mechanism (based on build attributes) to select the correct > variant of libraries. In other words, if any of your objects are compiled > with exceptions, you are guaranteed to no select any of the no-exceptions > library variants. This avoids the problem you mentioned altogether. Note that that behavior will cause different issues in some cases. If you link an executable to two libraries, `libfoo.so` and `libbar.so`, where one was compiled with exceptions and the other without you'll get two copies of libc++abi in the program. Now you have two copies of the global terminate handler and two copies of `__terminate` for the linker to select from. If it selects the nothrow `__terminate` and the terminate handler throws an exception your in trouble. That's not a reason to hold up this patch, but it is a potential problem you should be aware of. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. In http://reviews.llvm.org/D20677#442559, @EricWF wrote: > I have an issue with this change since it allows a libc++abi built without > exceptions to be used in a program compiled with them. I assert that this > should not be supported in any way. > > My personal preference would be to remove as much of the exception API from > the library when it's built without exceptions. This should cause link errors > if a program contains any code that throws. If this is not done it creates > bugs within code like `__terminate` as mentioned in an inline comment. > > I would like to see another revision of this patch that removes as much of > the exception API as possible. Looking at the symbols defined in `cxxabi.h` I > would say ever symbol in sections 2.4 and 2.5 should not longer be defined in > the library. That sounds generally OK to me. The toolchain I'm working on (ARM Compiler 6) has an elaborate mechanism (based on build attributes) to select the correct variant of libraries. In other words, if any of your objects are compiled with exceptions, you are guaranteed to no select any of the no-exceptions library variants. This avoids the problem you mentioned altogether. But I can see that this can be a problem for any other toolchain that does not have such a functionality, and a link-time error would be more suitable. I'll re-spin a patch soon. Thanks. / Asiri http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment.
I have an issue with this change since it allows a libc++abi built without
exceptions to be used in a program compiled with them. I assert that this
should not be supported in any way.
My personal preference would be to remove as much of the exception API from the
library when it's built without exceptions. This should cause link errors if a
program contains any code that throws. If this is not done it creates bugs
within code like `__terminate` as mentioned in an inline comment.
I would like to see another revision of this patch that removes as much of the
exception API as possible. Looking at the symbols defined in `cxxabi.h` I would
say ever symbol in sections 2.4 and 2.5 should not longer be defined in the
library.
Comment at: src/cxa_handlers.cpp:62
@@ -61,3 +61,3 @@
void
__terminate(terminate_handler func) _NOEXCEPT
{
If any part of your executable was compiled with exceptions on, and it sets a
terminate handler that throws an exception this does the wrong thing. I don't
really like that although I admit that would certainly be an exceptional
situation.
At minimum this requirement needs to be documented explicitly somewhere. Within
the CMake option description seems fine for now.
Comment at: src/cxa_personality.cpp:1204
@@ -1203,2 +1203,3 @@
}
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
try
This is yucky. We are in exception handling code and we are saying it has been
built without exceptions. I understand this "gets libc++abi to compile" without
exceptions but it doesn't really make sense.
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
mclow.lists added a comment. If you're really going to do this, you should probably empty out most of the routines in src/cxa_exception.cpp as well. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. In http://reviews.llvm.org/D20677#441598, @EricWF wrote: > I want to take a look at this as well. I'll review it tomorrow. Thanks! will wait until we have ironed out all the details, no hurry. / Asiri http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. In http://reviews.llvm.org/D20677#441407, @bcraig wrote: > https://mentorembedded.github.io/cxx-abi/abi-eh.html mentions the functions, > but it doesn't really say what they are for. The C++ standard mentions when > an exception of std::bad_cast and std::bad_typeid are thrown. Put those two > documents together, and you get my earlier statements. Granted, it's just a > well informed guess, but a guess I'm pretty confident in. > > Leaving those functions in should be harmless, other than the size of the > dead code. There are also all the private type_info classes that are dead > weight if RTTI is turned off. > > While it may make sense to remove those functions, that's probably best left > for a different code review. Thanks, added that to my arsenal of references :) http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a reviewer: mclow.lists. EricWF added a subscriber: mclow.lists. EricWF added a comment. Adding @mclow.lists to this as well. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
EricWF added a comment. I want to take a look at this as well. I'll review it tomorrow. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
bcraig added a comment. LGTM. Probably want a "LGTM" from at least one other person though. In http://reviews.llvm.org/D20677#441085, @rmaprath wrote: > In http://reviews.llvm.org/D20677#441061, @jroelofs wrote: > > > This is the canonical reference for the Itanium ABI: > > https://mentorembedded.github.io/cxx-abi/abi.html > > > I was indeed looking at this, but couldn't find those exact function > signatures (or something that looks like it). Perhaps it's covered there as a > more general concept. I'll have a read. https://mentorembedded.github.io/cxx-abi/abi-eh.html mentions the functions, but it doesn't really say what they are for. The C++ standard mentions when an exception of std::bad_cast and std::bad_typeid are thrown. Put those two documents together, and you get my earlier statements. Granted, it's just a well informed guess, but a guess I'm pretty confident in. Leaving those functions in should be harmless, other than the size of the dead code. There are also all the private type_info classes that are dead weight if RTTI is turned off. While it may make sense to remove those functions, that's probably best left for a different code review. http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath added a comment. In http://reviews.llvm.org/D20677#441061, @jroelofs wrote: > This is the canonical reference for the Itanium ABI: > https://mentorembedded.github.io/cxx-abi/abi.html I was indeed looking at this, but couldn't find those exact function signatures (or something that looks like it). Perhaps it's covered there as a more general concept. I'll have a read. Thanks! / Asiri http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
jroelofs added a comment. This is the canonical reference for the Itanium ABI: https://mentorembedded.github.io/cxx-abi/abi.html http://reviews.llvm.org/D20677 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath marked an inline comment as done.
Comment at: src/cxa_aux_runtime.cpp:19-25
@@ -18,5 +18,9 @@
extern "C" {
_LIBCXXABI_FUNC_VIS LIBCXXABI_NORETURN void __cxa_bad_cast(void) {
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
throw std::bad_cast();
+#else
+ std::terminate();
+#endif
}
rmaprath wrote:
> I'm not very familiar with how RTTI works, I was simply getting rid of the
> `throw` statements which get in the way of compiling with `-fno-exceptions`.
>
> So, I presume these `__cxa_bad_xxx` calls are inserted by the compiler?
>
> The tests `dynamic_cast_xxx` all seem to pass, which I found a bit weird
> because I would've expected some of those to fail because of the
> `std::terminate()`. Or are they not related to RTTI?
Thanks. I will try to get clang to generate these (btw, do you have a reference
for those RTTI functions? I couldn't find much on the interwebs.
I suppose leaving these functions (which call `std::terminate()`) in the
-fno-exceptions variant is not a problem?
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
bcraig added inline comments.
Comment at: src/cxa_aux_runtime.cpp:19-24
@@ -18,4 +18,8 @@
extern "C" {
_LIBCXXABI_FUNC_VIS LIBCXXABI_NORETURN void __cxa_bad_cast(void) {
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
throw std::bad_cast();
+#else
+ std::terminate();
+#endif
}
bcraig wrote:
> So you're turning off exceptions, but leaving RTTI? You might want to keep
> some of the RTTI tests around too if that's the case.
_cxa_bad_cast should be inserted by the compiler. It is called when a
dynamic_cast to a reference type fails.
_cxa_bad_typeid should be inserted by the compiler. It is called when typeid
is used on a null pointer.
_cxa_throw_bad_array_new_length should probably stay put. The most common
reason for that to get hit is if sizeof(obj) * number_of_elts > SIZE_T_MAX.
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath marked 2 inline comments as done.
Comment at: src/cxa_aux_runtime.cpp:19-20
@@ -18,3 +18,4 @@
extern "C" {
_LIBCXXABI_FUNC_VIS LIBCXXABI_NORETURN void __cxa_bad_cast(void) {
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
throw std::bad_cast();
I'm not very familiar with how RTTI works, I was simply getting rid of the
`throw` statements which get in the way of compiling with `-fno-exceptions`.
So, I presume these `__cxa_bad_xxx` calls are inserted by the compiler?
The tests `dynamic_cast_xxx` all seem to pass, which I found a bit weird
because I would've expected some of those to fail because of the
`std::terminate()`. Or are they not related to RTTI?
Comment at: src/cxa_new_delete.cpp:85-86
@@ -79,2 +84,4 @@
+#endif
p = ::operator new(size);
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
}
I agree. I've update it to be inline with `new.cpp` from libcxx.
At least the users will have some method of recovery if we return null, calling
`std::terminate()` would be a bit rude.
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
rmaprath updated this revision to Diff 58620. http://reviews.llvm.org/D20677 Files: CMakeLists.txt src/cxa_aux_runtime.cpp src/cxa_handlers.cpp src/cxa_new_delete.cpp src/cxa_personality.cpp test/CMakeLists.txt test/backtrace_test.pass.cpp test/catch_array_01.pass.cpp test/catch_array_02.pass.cpp test/catch_class_01.pass.cpp test/catch_class_02.pass.cpp test/catch_class_03.pass.cpp test/catch_class_04.pass.cpp test/catch_const_pointer_nullptr.pass.cpp test/catch_function_01.pass.cpp test/catch_function_02.pass.cpp test/catch_in_noexcept.pass.cpp test/catch_member_data_pointer_01.pass.cpp test/catch_member_function_pointer_01.pass.cpp test/catch_member_pointer_nullptr.pass.cpp test/catch_multi_level_pointer.pass.cpp test/catch_pointer_nullptr.pass.cpp test/catch_pointer_reference.pass.cpp test/catch_ptr.pass.cpp test/catch_ptr_02.pass.cpp test/incomplete_type.sh.cpp test/inherited_exception.pass.cpp test/libcxxabi/test/config.py test/lit.site.cfg.in test/test_aux_runtime.pass.cpp test/test_aux_runtime_op_array_new.pass.cpp test/test_guard.pass.cpp test/test_vector1.pass.cpp test/test_vector2.pass.cpp test/test_vector3.pass.cpp test/uncaught_exceptions.pass.cpp test/unwind_01.pass.cpp test/unwind_02.pass.cpp test/unwind_03.pass.cpp test/unwind_04.pass.cpp test/unwind_05.pass.cpp test/unwind_06.pass.cpp Index: test/unwind_06.pass.cpp === --- test/unwind_06.pass.cpp +++ test/unwind_06.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_05.pass.cpp === --- test/unwind_05.pass.cpp +++ test/unwind_05.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_04.pass.cpp === --- test/unwind_04.pass.cpp +++ test/unwind_04.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_03.pass.cpp === --- test/unwind_03.pass.cpp +++ test/unwind_03.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/unwind_02.pass.cpp === --- test/unwind_02.pass.cpp +++ test/unwind_02.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/unwind_01.pass.cpp === --- test/unwind_01.pass.cpp +++ test/unwind_01.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include struct A Index: test/uncaught_exceptions.pass.cpp === --- test/uncaught_exceptions.pass.cpp +++ test/uncaught_exceptions.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include #include #include Index: test/test_vector3.pass.cpp === --- test/test_vector3.pass.cpp +++ test/test_vector3.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector2.pass.cpp === --- test/test_vector2.pass.cpp +++ test/test_vector2.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_vector1.pass.cpp === --- test/test_vector1.pass.cpp +++ test/test_vector1.pass.cpp @@ -7,6 +7,8 @@ // //===--===// +// UNSUPPORTED: libcxxabi-no-exceptions + #include "cxxabi.h" #include Index: test/test_guard.pass.cpp === --- test/test_guard.pass.cpp +++ test/test_guard.pass.cpp @@ -41,6 +41,7 @@ // When initialization
Re: [PATCH] D20677: Make it possible to build a -fno-exceptions libc++abi variant.
bcraig added inline comments.
Comment at: CMakeLists.txt:359
@@ -350,2 +358,3 @@
else()
- add_subdirectory(test)
+ # libc++abi tests are mostly exceptions related. The only reason we want to
+ # build libc++abi without exceptions is to support the -fno-exceptions libc++
There's one area / test that I would still like to see when exceptions are off.
"magic" / thread safe static initialization has a test (test_guard.pass.cpp).
I think it would be pretty easy to #if out the one sub-test that uses
exceptions there.
There's also some demangling and backtracing tests, but I'm much less concerned
with those. If you go through the trouble to get any tests working with
exceptions turned off, you might as well get those too.
Comment at: src/cxa_aux_runtime.cpp:19
@@ -18,2 +18,3 @@
extern "C" {
_LIBCXXABI_FUNC_VIS LIBCXXABI_NORETURN void __cxa_bad_cast(void) {
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
So you're turning off exceptions, but leaving RTTI? You might want to keep
some of the RTTI tests around too if that's the case.
Comment at: src/cxa_new_delete.cpp:85
@@ -79,1 +84,3 @@
+#endif
p = ::operator new(size);
+#ifndef _LIBCXXABI_NO_EXCEPTIONS
Question: Should "throwing" new call terminate on allocation failure, or is
returning null from "throwing" new fine?
I'm leaning towards returning null, as I would rapidly grow tired of
std::nothrow everywhere, but it does seem a bit dangerous.
http://reviews.llvm.org/D20677
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
