[RHEL6.1] rpc.mountd segfault on NFSv3 server after "umount" followed by "showmount -a" on client system
Issue
- showmount sometimes fails with the following message:
rpc mount dump: RPC: Unable to receive; errno = Connection reset by peer
- when showmount fails, rpc.mountd segfaults with a message similar to the following, with 'segfault' showing an address near 'sp', which means the crash is because stack has been exceeded:
Aug 7 19:27:22 hostname1 kernel: rpc.mountd[27257]: segfault at 7fff1ab57ff8 ip 00007f7776b596d1 sp 00007fff1ab58000 error 6 in libc-2.12.so[7f7776ad7000+187000]
- messages file may indicate an 'unmount' request just prior to the rpc.mountd segfault, similar to the following:
Oct 31 21:01:07 hostname1 rpc.mountd[9193]: authenticated unmount request from 1.1.1.2:661 for /export/somedir (/foo/bar) Oct 31 21:01:10 hostname1 kernel: rpc.mountd[9193] general protection ip:7ff86d60154f sp:7fff89755f78 error:0 in libc-2.12.so[7ff86d4dc000+186000]
- rpc.mountd segfault with NFSv3 server when "umount" followed by "showmount -a" is done on client system
- The following steps produce the rpc.mountd segfault:
1. Create an NFS export on test system.
2. On some remote system mount the NFS export with
mount -t nfs <ip of test system>:/srv/redhat-dvd -o nfsvers=3,nolock /mnt/nfs.
3. On remote system run showmount -a <ip of test system>
4. On remote system run umount /mnt/nfs
5. On remote system run showmount -a <ip of test system>
6. Segfault occurs
Environment
- Red Hat Enterprise Linux 6.1
- nfs-utils-1.2.3-7.el6.x86_64.rpm
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.