Re: [Libunwind-devel] libunwind on x86-64 w/ C++ exceptions

[Mark Rabkin]

Ok, great -- I'll test and let you know.


On 2/27/08 2:05 PM, "Arun Sharma" <address@hidden> wrote:

> Hi Mark,
>
> As I said in the other mail, something along the lines of this completely untested change may be what you and others who're having trouble with _Unwind_*() APIs want:
>
> --- a/src/Makefile.am
> +++ b/src/Makefile.am
> @@ -70,7 +70,10 @@ libunwind_la_SOURCES_local =                                         \
>         mi/Lput_dynamic_unwind_info.c mi/Ldestroy_addr_space.c          \
>         mi/Lget_reg.c   mi/Lset_reg.c                                   \
>         mi/Lget_fpreg.c mi/Lset_fpreg.c                                 \
> -       mi/Lset_caching_policy.c                                        \
> +       mi/Lset_caching_policy.c
> +
> +if SUPPORT_CXX_EXCEPTIONS
> +libunwind_la_SOURCES_local +=                                          \
>         unwind/Backtrace.c unwind/DeleteException.c                     \
>         unwind/FindEnclosingFunction.c unwind/ForcedUnwind.c            \
>         unwind/GetBSP.c unwind/GetCFA.c unwind/GetDataRelBase.c         \
> @@ -78,6 +81,7 @@ libunwind_la_SOURCES_local =                                          \
>         unwind/GetRegionStart.c unwind/GetTextRelBase.c                 \
>         unwind/RaiseException.c unwind/Resume.c                         \
>         unwind/Resume_or_Rethrow.c unwind/SetGR.c unwind/SetIP.c
> +endif
>  --- a/configure.in
> +++ b/configure.in
> @@ -94,6 +94,7 @@ AM_CONDITIONAL(ARCH_PPC32, test x$target_arch = xppc32)
>  AM_CONDITIONAL(ARCH_PPC64, test x$target_arch = xppc64)
>  AM_CONDITIONAL(OS_LINUX, expr x$target_os : xlinux >/dev/null)
>  AM_CONDITIONAL(OS_HPUX, expr x$target_os : xhpux >/dev/null)
> +AM_CONDITIONAL(SUPPORT_CXX_EXCEPTIONS, test x$target_arch = xia64)
>  
>  if test x$target_arch = xppc64; then
>          libdir='${exec_prefix}/lib64'
>
>  -Arun
>
>
> On Tue, Feb 26, 2008 at 10:41 PM, Mark Rabkin <address@hidden> wrote:
> > Hi Arun,
> >
> > My battle to get this working still continues :)  Are you going to have time to take a look into the first crash (the one you can reproduce)?  
> >
> > I made a patch to prevent that 2nd crashing bug below on the fldenv(), as I mentioned before, but now I'm seeing another strange behavior -- in my complicated program, I have exceptions that slip through a try-catch block uncaught (even though they should be), and result in a program termination due to uncaught exception. I'm struggling to get a smaller test case of that working so I can get verification of it.
> >
> > I'll send you my patch in the meantime to see if you think it's acceptable.
> >
> > - Mark
> >
> >
> >
> >
> >
> > On 2/15/08 4:58 PM, "Arun Sharma" <address@hidden> wrote:
> >
> > > On Fri, Feb 15, 2008 at 4:25 PM, Mark Rabkin <address@hidden> wrote:
> > > > $ gdb unwind
> > > > Program received signal SIGABRT, Aborted.
> > > >
> > > > #0  0x0000003bd882f3b0 in raise () from /lib64/libc.so.6
> > > > #1  0x0000003bd8830860 in abort () from /lib64/libc.so.6
> > > > #2  0x0000003bd95bb8c1 in __cxa_get_globals () from /usr/lib64/libstdc++.so.6
> > > > #3  0x0000003bd95bb9b4 in __cxa_get_globals () from /usr/lib64/libstdc++.so.6
> > > > #4  0x0000003bd95bbd07 in __gxx_personality_v0 ()
> > > >    from /usr/lib64/libstdc++.so.6
> > > > #5  0x000000000040100b in _Unwind_Resume (exception_object=0x517060)
> > > >     at unwind/unwind-internal.h:118
> > > > #6  0x0000000000400e84 in throw_int () at unwind.cpp:7
> > > > #7  0x0000000000400e98 in main (argc=1, argv=0x7ffffffcb818) at unwind.cpp:12
> > >
> > > I can reproduce this one.
> > >  
> > > > $ ./unwind-shared
> > > >
> > > > Program received signal SIGSEGV, Segmentation fault.
> > > > _x86_64_setcontext () at x86_64/setcontext.S:34
> > > > 34              fldenv (%r8)
> > >  
> > >
> > > But not this one. Which version of gcc are you using?
> > >
> > >  -Arun