Bug#284563: libunwind in unstable
On Thu, Dec 09, 2004 at 01:35:07PM +0100, Matthias Klose wrote: David Mosberger writes: On Tue, 7 Dec 2004 09:01:33 +0100, Matthias Klose [EMAIL PROTECTED] said: Matthias glibc now fails to build from source: Matthias undefined reference to `__gcc_personality_v0' Argh, looks like we may have to leave unwind-c.c in LIB2ADDEH. I cc'd HJ Lu since he may remember better what the final conclusion was in regards to the personality routine. IIRC, the resolution was that it needs to be provided by libgcc. works ok until building glibc with the just built compiler. _Unwind_Resume, _Unwind_GetRegionStart are referenced by glibc objects, _Unwind_GetIP, _Unwind_SetGR, _Unwind_SetGR, _Unwind_SetIP are referenced by libgcc_eh.a. Found in the standalone libunwind.a: That is wrong. Glibc doesn't use _Unwind_GetRegionStart. Glibc has its own _Unwind_Resume, _Unwind_ForcedUnwind, _Unwind_GetCFA and _Unwind_GetBSP. H.J. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#284563: libunwind in unstable
On Thu, Dec 09, 2004 at 12:49:38PM +0100, Matthias Klose wrote: H. J. Lu writes: On Tue, Dec 07, 2004 at 12:16:56AM -0800, David Mosberger wrote: On Tue, 7 Dec 2004 09:01:33 +0100, Matthias Klose [EMAIL PROTECTED] said: Matthias glibc now fails to build from source: Matthias undefined reference to `__gcc_personality_v0' Argh, looks like we may have to leave unwind-c.c in LIB2ADDEH. I cc'd HJ Lu since he may remember better what the final conclusion was in regards to the personality routine. IIRC, the resolution was that it needs to be provided by libgcc. We decided that the personality routine shouldn't be the part of libunwind.so. For gcc 3.4/4.0, unwind-sjlj.c and unwind-c.c are in libgcc_eh.a, unwind-sjlj.c, unwind-c.c, unwind-compat.c and unwind-dw2-fde-compat.c are in libgcc_s.so.1. ok, I'm currently bootstrapping gcc-3.3 with the patch attached, a glibc bootstrap using this compiler did succeed. I'll upload the fixed gcc-3.3, when bootstrap and testsuite finishes. Why not just backport the 3.4 libunwind patches to 3.3? H.J. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [PATCH] glibc 2.2.94 - hppa - revised plt relocation patch
On Wed, Sep 25, 2002 at 01:03:06PM -0400, Carlos O'Donell wrote: diff -urN libc/elf/do-rel.h libc/elf/do-rel.h --- libc/elf/do-rel.h 18 Sep 2002 18:28:43 - +++ libc/elf/do-rel.h 23 Sep 2002 11:35:14 - @@ -30,6 +30,12 @@ # define RELCOUNT_IDXVERSYMIDX (DT_RELCOUNT) #endif +#ifndef DO_ELF_MACHINE_REL_RELATIVE +#define DO_ELF_MACHINE_REL_RELATIVE(map, l_addr, reloc) \ + elf_machine_rel_relative (l_addr, reloc, \ + (void *) (l_addr + relative-r_offset)) +#endif + Where does `relative' come from? Why not add it to the arg? H.J. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [PATCH] glibc 2.2.94 - hppa - plt relocations
On Mon, Sep 23, 2002 at 03:13:29PM +0930, Alan Modra wrote: It may be possible to redesign the relocations, or at least the hppa dl-machine.h code, but I don't have time and I think the approach used in the patch is the cleanest available. eg. another approach would be to add the load address to the ltp entry, but RELA targets shouldn't be reading section contents. This would also require a binutils change as ld currently leaves the field zero. I'll also note that the patch as it stands should be rejected for another reason, and that is that it isn't complete. eg. I needed to patch sysdeps/i386/dl-machine.h elf_machine_rela_relative to add the extra param. IA 64 had the similar problem when it was first mergred. Shouldn't HPPA try something similar like DL_RELOCATION_NEEDS_MAP? H.J. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
