On Mon, 01 Oct 2012 15:06:49 +0200, Mehdi Dogguy wrote: > >I've prepared an NMU for libatomic-ops (versioned as > >7.3~alpha1+git20111031-1.1) and > >uploaded it to DELAYED/2. Please feel free to tell me if I > >should delay it longer. > We may need this patch after all :) It seems that #632604 and sid's > libatomic-ops cannot be unblocked for Wheezy (due to very large > changes in its codebase). Could you please prepare an upload > targetting testing-proposed-updates?
Here we are, the same patch applied against the version in testing, and the build tested in a wheezy i386 cowbuilder chroot. Ian, if your busy I'm happy to upload the fix (if Mehdi is ok with the diff). Cheers, gregor -- .''`. Homepage: http://info.comodo.priv.at/ - OpenPGP key 0xBB3A68018649AA06 : :' : Debian GNU/Linux user, admin, and developer - http://www.debian.org/ `. `' Member of VIBE!AT & SPI, fellow of the Free Software Foundation Europe `- NP: Flying Pickets
diff -u libatomic-ops-7.2~alpha5+cvs20101124/debian/changelog libatomic-ops-7.2~alpha5+cvs20101124/debian/changelog --- libatomic-ops-7.2~alpha5+cvs20101124/debian/changelog +++ libatomic-ops-7.2~alpha5+cvs20101124/debian/changelog @@ -1,3 +1,13 @@ +libatomic-ops (7.2~alpha5+cvs20101124-1.1) testing-proposed-updates; urgency=low + + * Non-maintainer upload. + * Fix "FTBFS on i386": apply patch from upstream git repo: + https://github.com/ivmai/libatomic_ops/commit/30cea1b#diff-0 + Thanks to Emanuele Aina for the pointer. + (Closes: #632604) + + -- gregor herrmann <[email protected]> Tue, 02 Oct 2012 17:31:16 +0200 + libatomic-ops (7.2~alpha5+cvs20101124-1) unstable; urgency=low * Closes: 604753 -- Update from upstream CVS, fixes ARM Thumb-2 mode only in patch2: unchanged: --- libatomic-ops-7.2~alpha5+cvs20101124.orig/src/atomic_ops/sysdeps/gcc/x86.h +++ libatomic-ops-7.2~alpha5+cvs20101124/src/atomic_ops/sysdeps/gcc/x86.h @@ -148,13 +148,16 @@ /* If PIC is turned on, we can't use %ebx as it is reserved for the GOT pointer. We can save and restore %ebx because GCC won't be using it for anything else (such as any of the m operands) */ - __asm__ __volatile__("pushl %%ebx;" /* save ebx used for PIC GOT ptr */ - "movl %6,%%ebx;" /* move new_val2 to %ebx */ + /* We use %edi (for new_val1) instead of a memory operand and swap */ + /* instruction instead of push/pop because some GCC releases have */ + /* a bug in processing memory operands (if address base is %esp) in */ + /* the inline assembly after push. */ + __asm__ __volatile__("xchg %%ebx,%6;" /* swap GOT ptr and new_val1 */ "lock; cmpxchg8b %0; setz %1;" - "pop %%ebx;" /* restore %ebx */ + "xchg %%ebx,%6;" /* restore ebx and edi */ : "=m"(*addr), "=a"(result) : "m"(*addr), "d" (old_val2), "a" (old_val1), - "c" (new_val2), "m" (new_val1) : "memory"); + "c" (new_val2), "D" (new_val1) : "memory"); #else /* We can't just do the same thing in non-PIC mode, because GCC * might be using %ebx as the memory operand. We could have ifdef'd
signature.asc
Description: Digital signature
