On Mon, 2014-09-15 at 10:17 +0300, Arto Jantunen wrote: > Hi, > > There is a fairly major regression in the NFSv4 server in version > 3.2.60. This doesn't appear to be fixed in 3.2.63. Ubuntu has fixed this > in their kernels with this patch: > https://lists.ubuntu.com/archives/kernel-team/2014-September/048156.html > > The bug has been filed in launchpad as #1348670 and the Debian BTS as > #754420. > > This could be considered to have security implications, the server needs > a reboot to recover after all nfs kernel threads have been killed by a > client doing suitable operations (cp -a from one directory to another on > the same NFS mount seems to trigger it at least fairly often), making > this a potent denial of service attack (but only for clients that can > access a writable NFSv4 share). > > Based on the Ubuntu report the regression is caused by commit fa1850b, > which only exists in the 3.2 stable series. But the same change was backported into all stable branches for 3.2 up to 3.13 inclusive. I believe I understand the bug, and the patch makes sense to me. Sergio, please can you submit the patch to the stable list with the Developer's Certificate of Origin (Signed-off-by)? Also the commit message should include a brief explanation of why it isn't needed upstream. Ben. -- Ben Hutchings Everything should be made as simple as possible, but not simpler. - Albert Einstein
Attachment:
signature.asc
Description: This is a digitally signed message part