Bug in nfsd from kernel 2.6.16.21-0.13
Hello, I discovered a bug in the nfs daemon delivered with the kernel update to kernel 2.6.16.21-0.13. When accessing a large amount of files over an nfs export from a machine with SuSE 10.1 and kernel 2.6.16.21-0.13, stale NFS handles occur. After a short period (3-5s) the files are accessible again. "grep -R hello" * triggers the bug for example. My configuration: SuSE 10.1 with kernel 2.6.16.21-0.13, NFS export of an XFS filesystem mounted on different clients with autofs (UDP and TCP). Degrading back the kernel to 2.6.16.13-4 and all works fine again. Regards, Marco
I thought this list was for SECURITY issues, which is why I subscribed to it. Aren't there any other lists for discussing normal system bugs that this sort of traffic should go to? I find a lot of unnecessary chatter on non-security issues. I really hate to censor anything, but on these lists that aren't moderated (presuming that it's not monitored) sometimes discussions get carried away. Just thought I would raise the point. Mike Siedelberg Jackson National Life Insurance-IT Security Desk Phone 517-367-3546 Cell Phone 517-230-0922 Marco Munderloh <munderl@tnt.uni-hannover.de> 07/25/2006 12:31 PM To suse-security@suse.com cc Subject [suse-security] Bug in nfsd from kernel 2.6.16.21-0.13 Hello, I discovered a bug in the nfs daemon delivered with the kernel update to kernel 2.6.16.21-0.13. When accessing a large amount of files over an nfs export from a machine with SuSE 10.1 and kernel 2.6.16.21-0.13, stale NFS handles occur. After a short period (3-5s) the files are accessible again. "grep -R hello" * triggers the bug for example. My configuration: SuSE 10.1 with kernel 2.6.16.21-0.13, NFS export of an XFS filesystem mounted on different clients with autofs (UDP and TCP). Degrading back the kernel to 2.6.16.13-4 and all works fine again. Regards, Marco -- Check the headers for your unsubscription address For additional commands, e-mail: suse-security-help@suse.com Security-related bug reports go to security@suse.de, not here
I would call his bug, a bug and a denial of service that any user could trigger. Thanks, LDB Mike Siedelberg wrote:
I thought this list was for SECURITY issues, which is why I subscribed to it. Aren't there any other lists for discussing normal system bugs that this sort of traffic should go to?
I find a lot of unnecessary chatter on non-security issues. I really hate to censor anything, but on these lists that aren't moderated (presuming that it's not monitored) sometimes discussions get carried away.
Just thought I would raise the point.
Mike Siedelberg Jackson National Life Insurance-IT Security Desk Phone 517-367-3546 Cell Phone 517-230-0922
Marco Munderloh <munderl@tnt.uni-hannover.de> 07/25/2006 12:31 PM
To suse-security@suse.com cc
Subject [suse-security] Bug in nfsd from kernel 2.6.16.21-0.13
Hello,
I discovered a bug in the nfs daemon delivered with the kernel update to kernel 2.6.16.21-0.13.
When accessing a large amount of files over an nfs export from a machine with SuSE 10.1 and kernel 2.6.16.21-0.13, stale NFS handles occur. After a short period (3-5s) the files are accessible again. "grep -R hello" * triggers the bug for example.
My configuration: SuSE 10.1 with kernel 2.6.16.21-0.13, NFS export of an XFS filesystem mounted on different clients with autofs (UDP and TCP).
Degrading back the kernel to 2.6.16.13-4 and all works fine again.
Regards, Marco
LDB wrote:
I would call his bug, a bug and a denial of service that any user could trigger.
The provided kernel update fixes _a lot_ of security issues but with this denial-of-service bug there is no way to actually use this kernel. So, for security concerns, this bug should be fixed soon so everybody can install the security update and get a save system without troubling with an unstable system and unsatisfied users. regards, Marco
Mike Siedelberg schrieb: Hi Mike,
I thought this list was for SECURITY issues
Marco did indeed describe a security relevant regression. Actuallly it is a netowrk exploitable(*) denial of service attack issue.
When accessing a large amount of files over an nfs export from a machine with SuSE 10.1 and kernel 2.6.16.21-0.13, stale NFS handles occur.
Regards, --martin
Hello,
I discovered a bug in the nfs daemon delivered with the kernel update to kernel 2.6.16.21-0.13.
When accessing a large amount of files over an nfs export from a machine with SuSE 10.1 and kernel 2.6.16.21-0.13, stale NFS handles occur. After a short period (3-5s) the files are accessible again. "grep -R hello" * triggers the bug for example.
My configuration: SuSE 10.1 with kernel 2.6.16.21-0.13, NFS export of an XFS filesystem mounted on different clients with autofs (UDP and TCP).
Degrading back the kernel to 2.6.16.13-4 and all works fine again.
I opened a bug for this now, but you could have done yourself. I have cced Marco in the report. Ciao, Marcus
I opened a bug for this now, but you could have done yourself. I have cced Marco in the report. I already did that myself: https://bugzilla.novell.com/show_bug.cgi?id=195040
The bug seems to occur in SLES9 before. Adding "no_subtree_check" to the export options in /etc/exports works as a workaround. Nevertheless I don't think it's a good idea that a novell account is needed to post a bug in Bugzilla. Giving there full address surely scares some good people off. Regards, Marco
participants (5)
-
LDB -
Marco Munderloh -
Marcus Meissner -
Martin Konold -
Mike Siedelberg