[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#659111: [2.6.31 -> 2.6.32 regression] Files on NFS4 become unwritable,, but OK after explicit stat



Hi,

On 03/21/2012 06:39 PM, Jonathan Nieder wrote:
Rik Theys wrote:

Can this commit be backported to the 3.2 kernel in sid?
It applies and I have tested it in the past against the 3.2.4 kernel.

Just for the record: we are talking about

  v3.3-rc1~116^2~1 "NFSv4: Save the owner/group name string when doing
                   open", 2012-01-07

It looks safe to me, too.  It is too large to follow the rules of
Greg's 3.2.y-stable kernel to the letter, but I think we should apply
it.

It seems the patch still has not made it into the sid kernel. Any reason this patch hasn't been cherry picked yet?

For 2.6.32.y-longterm, I would be happier with a more minimal patch.
Where can one find more information about the rpciod hang described
in the commit message to 80e52aced138b ("NFSv4: Don't do idmapper
upcalls for asynchronous RPC calls", 2009-08-09)?  Maybe there is
some alternative fix to that.

I would also be very much interested to hear how well the following
series against 2.6.32.y does:

  v2.6.33-rc1~54^2~7 "rpc: add a new priority in RPC task", 2009-12-14
  v2.6.33-rc1~54^2~5 "nfs: make recovery state manager operations
                     privileged", 2009-12-14
  v3.3-rc1~116^2~1 "NFSv4: Save the owner/group name string when doing
                   open", 2012-01-07

I've been running a patched squeeze kernel with those patches applied for about two weeks now and it fixes the bug. I have not seen any regressions in other nfs functionality using these patches.

It would be great to have this fixed in squeeze if not considered too intrusive.

But I would definitely like to see these patches land in the sid/wheezy kernel, so Wheezy will not have this bug.

Regards,

Rik



Reply to: