Re: cvsup broken on amd64?

From: Kostik Belousov <kostikbel_at_gmail.com>
Date: Fri, 9 Sep 2011 17:55:13 +0300
On Fri, Sep 09, 2011 at 04:34:54PM +0200, Oliver Lehmann wrote:
> 
> Kostik Belousov <kostikbel_at_gmail.com> wrote:
> 
> >On Fri, Sep 09, 2011 at 04:19:42PM +0200, Oliver Lehmann wrote:
> 
> >>(gdb) bt
> >>#0  0x00000000004d24c6 in tzload ()
> >
> >Try to do "disas 0x4d24c6 0x4d24c6+30" from gdb prompt with the loaded 
> >core.
> 
> (gdb) disas 0x4d24c6 0x4d24c6+30
> Dump of assembler code from 0x4d24c6 to 0x4d24e4:
> 0x00000000004d24c6 <tzload+86>: callq  0x4db370 <issetugid>
> 0x00000000004d24cb <tzload+91>: test   %eax,%eax
> 0x00000000004d24cd <tzload+93>: jne    0x4d25e0 <tzload+368>
> 0x00000000004d24d3 <tzload+99>: movzbl (%rbx),%ebp
> 0x00000000004d24d6 <tzload+102>:        cmp    $0x3a,%bpl
> 0x00000000004d24da <tzload+106>:        jne    0x4d24e3 <tzload+115>
> 0x00000000004d24dc <tzload+108>:        add    $0x1,%rbx
> 0x00000000004d24e0 <tzload+112>:        movzbl (%rbx),%ebp
> 0x00000000004d24e3 <tzload+115>:        cmp    $0x2f,%bpl
> End of assembler dump.

Ok, please do the following:
run cvsup under the gdb. When SIGSEGV is raised, from the gdb prompt, do:
1. info registers $rsp
2. info program
	This should print you the pid of the process, then do
3. shell procstat -v <pid>

I suspect that modula 3 system uses the kind of green threads, and
the default thread stack size is simply too small for amd64. This is
consistent with SIGILL when running standalone, but SIGSEGV under
debugger.

Received on Fri Sep 09 2011 - 12:55:19 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:17 UTC