Package: cvs
Version: 1.12.9-15
Severity: important

My CVS repository on an AMD64 can be accessed when cvs 1.12.9-15 is
running in an x86 chroot but not when the same version is used in
native AMD64 on the same machine. I suspect a void*/int problem or
suchlike. It seems to depend on the state of the repository, and the
way it is now, it's 100% reproducible, though it has never done this
in the past.

Output from the client:

# cvs commit -m ''
cvs commit: Examining .
cvs commit: Examining test
Terminated with fatal signal 11

At this point, the kernel log on the AMD64 server prints 

cvs[21411]: segfault at 0000000000000000 rip 000000000042ff5b rsp 
0000007fbffff500 error 4

If there is an easy way of getting a backtrace of that cvs process,
please let me know and I'll make one.

I don't think it's a hardware fault or file-system corruption because
(1) cvs in the x86 chroot works fine and (2) I copied the repository
to a different location and that does exactly the same.

Both client and server machines (AMD64 and x86 chroot) run Debian
sid. pserver is disabled, all communication is via ssh, which then
invokes cvs on the server.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to