On Thu, Dec 14, 2006 at 21:36:51 +0100, Julien Cristau wrote: > I reproduced this bug (I had seen it a while ago, but never took the > time to debug it), and got a core dump (without debug symbols). > Then I installed your packages and reran cvs up to get a useful core > file, but this time it didn't fail :( > This is weird. cvs built with "-O0 -g" seems to work fine cvs built with "-O2 -g" gets a sigbus, but when running gdb on the core file I get the following: /build/buildd/gdb-6.4.90.dfsg/gdb/regcache.c:926: internal-error: register_offset_hack: Assertion `regnum >= 0 && regnum < descr->nr_cooked_registers' failed. A problem internal to GDB has been detected, further debugging may prove unreliable.
In any case, gdb gives the following backtrace: #0 0x00081e00 in md5_read_ctx (ctx= ) at md5.c:91 #1 0x0006c7c4 in update_fileproc (callerdat=<value optimized out>, finfo=0xff12532d) at update.c:1644 #2 0x00055e90 in do_file_proc (p=0x0, closure=0xff125494) at recurse.c:959 #3 0x0002fbb0 in walklist (list=<value optimized out>, proc=0xff1254b4, closure=0xce918) at hash.c:419 #4 0x00054a14 in do_recursion (frame=0xce750) at recurse.c:847 #5 0x00055904 in do_dir_proc (p=<value optimized out>, closure=0xcd3c8) at recurse.c:1232 #6 0x0002fbb0 in walklist (list=<value optimized out>, proc=0xff12564c, closure=0x994d0) at hash.c:419 #7 0x000548fc in do_recursion (frame=0xcdf78) at recurse.c:877 #8 0x00055904 in do_dir_proc (p=<value optimized out>, closure=0xcca92) at recurse.c:1232 #9 0x0002fbb0 in walklist (list=<value optimized out>, proc=0xff1257ec, closure=0x994d0) at hash.c:419 #10 0x000548fc in do_recursion (frame=0xcaed8) at recurse.c:877 #11 0x000550fc in start_recursion (fileproc=<value optimized out>, filesdoneproc=0x994d0 <__PRETTY_FUNCTION__.8742+984>, direntproc=0x1, dirleaveproc=0xff12592c, callerdat=0x1bd38, argc=0, argv= /build/buildd/gdb-6.4.90.dfsg/gdb/regcache.c:175: internal-error: register_size: Assertion `regnum >= 0 && regnum < (NUM_REGS + NUM_PSEUDO_REGS)' failed. A problem internal to GDB has been detected, Hope this helps, and let me know if I can do any more debugging. Cheers, Julien -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]