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

Bug#603301: linux-image-2.6.32-5-amd64: FS-Cache doesn't working with my NFS mount



tags 603301 + moreinfo
quit

In September, 2011, Jonathan Nieder wrote:
> Piotr Meyer wrote:

>> Adter enabling FS-Cache on my NFS mount (busy WWW node) I got
>> following error
>> (more in dmesg below):
>>
>> CacheFiles: I/O Error: Unlink failed
>> FS-Cache: Cache cachefiles stopped due to I/O error
>
> On 2010-11-15, Ben Hutchings wrote:

[...]
>>     CacheFiles: Fix occasional EIO on call to vfs_unlink()
[...]
> Looks like that patch requires the following in order to apply:
[...]
>    CacheFiles: Fix a race in cachefiles_delete_object() vs rename
[...]
> They both seem sane at a glance and apply cleanly to 2.6.32.y.

Piotr, do you still have access to a setup reproducing this?  If so,
here are instructions for testing the patches.

 0. prerequisites:
	apt-get install git build-essential

 1. get the kernel history, if you don't already have it:
	git clone \
	  git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git

 2. fetch 2.6.32.y+drm33.z tree:
	cd linux
	git remote add squeeze-drm \
	  git://kernel.ubuntu.com/smb/linux-2.6.32.y-drm33.z.git
	git fetch squeeze-drm

 3. configure, build, test:
	git checkout squeeze-drm/master
	cp /boot/config-$(uname -r) .config; # current configuration
	scripts/config --disable DEBUG_INFO
	make localmodconfig; # optional: minimize configuration
	make; # optionally with -j<num> for parallel build
	fakeroot -u make deb-pkg
	dpkg -i ../<name of package>; # as root
	reboot
	... test test test ...

    Hopefully it reproduces the bug.  So

 4. try the patches:
	cd linux
	git cherry-pick -x -s 8f9941aeccc3 c61ea31dac03
	make; # maybe with -j4
	fakeroot -u make deb-pkg
	dpkg -i ../<name of package>; # as root
	reboot
	... test test test ...

Once the patches have been tested to work, we can try applying them to
squeeze and pass them upstream for the benefit of others using
kernel.org's 2.6.32.y kernels.

Hope that helps,
Jonathan


Reply to: