[Bug gas/16765] Assertion failure in create_unwind_entry

[nickc at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16765

Nick Clifton <nickc at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #4 from Nick Clifton <nickc at redhat dot com> ---
Hi Fabian,

  This is a gcc bug, as you guessed.  So you need to report it on the gcc-bugs
mailing list.  When you do, you might like to include the arm.c patch uploaded
here which will fix the problem(*).

  Gas should not be generating an internal error however, even when given bad
assembler input, so I have applied a small patch to replace the assertion with
an error message.

Cheers
  Nick

PS. FYI I was not able to reproduce the bug using the assembler file that you
uploaded but I was able to reproduce it when building a toolchain with the
configuration options you specified.  I was also able to capture the gas
command line and assembler input by adding "--save-temps -v" to the gcc command
line.

(*) After applying the patches in this PR I tried continuing the build.  It
failed with:

libstdc++-v3/libsupc++/eh_throw.cc: In function 'void
__cxxabiv1::__cxa_throw(void*, std::type_info*, void (*)(void*))':
/work/sources/gcc/current/libstdc++-v3/libsupc++/eh_throw.cc:80:57: error:
'_Unwind_SjLj_RaiseException' was not declared in this scope
   _Unwind_SjLj_RaiseException (&header->exc.unwindHeader);
                                                         ^
libstdc++-v3/libsupc++/eh_throw.cc: In function 'void
__cxxabiv1::__cxa_rethrow(...)':
/work/sources/gcc/current/libstdc++-v3/libsupc++/eh_throw.cc:113:60: error:
'_Unwind_SjLj_Resume_or_Rethrow' was not declared in this scope
       _Unwind_SjLj_Resume_or_Rethrow (&header->unwindHeader);
                                                            ^
make[3]: *** [eh_throw.lo] Error 1

I suspect that setjmp/longjmp exception handling support may have bit-rotted. 
At least for the ARM anyway.

-- 
You are receiving this mail because:
You are on the CC list for the bug.