o news
- the feature flushing writable branches at remount/unmounting aufs was
removed, since it doesn't seem to be so useful and it may cause a
deadlock in the future.
o bugfix
- aufs: possible bugfix, replace __getname() by au_getname()
- aufs: tiny, replace au_getname() by
Hello Barry,
Barry Kauler:
Now, it may be that these two things can be done in Aufs, but I don't know
how:
First reason
:::
If we have a save partition on top, we also have the SFS files in
that partition. Aufs will refuse to mount those SFS files as they are
already inside a
bl4:
In /var, unmounting had no effect on results. When it failed after
remount -ro, it also failed after umount. Currently fsck after remount
-ro succeeds.
Hmm, in my test environment, fsck after umount never fail. Even if you
execute sync(8) and specify sync,dirsync mount options, fsck
Daniel Vidal:
The rt patch is patch-2.6.33.1-rt10 with date 3-17-2009 from
http://www.kernel.org/pub/linux/kernel/projects/rt/ .
Ok, please test this patch.
It should apply against the aufs2-33 branch in aufs2-standalone.git
tree.
J. R. Okajima
a.patch.bz2
Description: BZip2 compressed
First reason
:::
If we have a save partition on top, we also have the SFS files in
that partition. Aufs will refuse to mount those SFS files as they are
already inside a branch. Consequently, we are forced to copy the main
SFS file into RAM and mount it from there. However,
Dipl.-Ing. Michael Niederle:
Here an excerpt from your original mail (Feb. 13th, 2008):
Here is a patch for the new option 'dirperm1' and 'nodirperm1.'
My local and simple tests are passed, so I want you to test it more.
As you can guess, if this new option has a bug, it can easily become a
o bugfix
- aufs: bugfix, stop setting the inotify watch twice for the same lower
inode.
o news
- aufs: support FSNOTIFY
introduce a new option 'udba=notify' which is equivalen to
udba=inotify before aufs2-31.
in aufs2-31 and later, you can choose either fsnotify or inotify in
aufs
Hello Oleg,
Thanx for the report.
Oleg:
[243074.982168] aufs au_do_pin:423:vsftpd[29191]: err 0
Can this be because of high load of the server? I have a big load average
lately, which sometimes per day grow up to 7 (it's many for 1 processor
machine).
I tried to update kernel with aufs. I
Oleg:
No, every write goes through aufs.
Ok.
Yes, certainly :-)! This would be great.
Please apply this patch. You will need to set the kernel log level to
'debug' to get the debug log. And send me the log.
Usually you can set it by specifying debug to the kernel boot option,
or echo 8
Tommy[D]:
That patch looks fine to me and works as expected.
Thanks, it will be included in next Monday release.
J. R. Okajima
--
Download Intel#174; Parallel Studio Eval
Try the new software tools for yourself.
Tommy[D]:
could you please revert commit dc96db9e700ef1073d519071a6c748467e0f7f79 ?=
I didn't know that LDFLAGS is for CC instead of LD.
This patch will be included in next Monday release.
J. R. Okajima
a.patch.bz2
Description: BZip2 compressed data
o bugfix
- aufs: bugfix, support updating a removed dir
- aufs: bugfix, missing unlock in an error path
- aufs: bugfix, error from kset_create_and_add()
- Install symlinks
- LDFLAGS is for CC instead of LD
J. R. Okajima
If you or your company use and like AUFS project, why not consider
Hello Matthew,
Matthew Harrell:
this all works great. What I would like to do, though, is take an additional
jffs2 partition I have and mount it as something like /etc/preferences where
changes to that directory will bypass the root filesystem and the tmpfs and
only get written to the jffs2
Oleg:
Here a kernel debug file.
There is no difference from the previous one, no necessary info.
Are you sure that you changed the loglevel?
(from my mail)
Please apply this patch. You will need to set the kernel log level to
'debug' to get the debug log. And send me the log.
Usually you
Hello Fabio,
Fabio Erculiani:
out of curiosity, what's the support status of kernel capabilities in AUFS?
If you mean capget(2)/capset(2), then aufs simply follows VFS and branch
fs in it.
If you mean ACL or XATTR, aufs doesn't support it currently. If I can
find a common set of them, I will
Hi,
Tobias Wilken:
we are currently porting our platform to amazon ec2, or we want to. Does
someone has any experiences with ec2 and aufs?
That is interesting.
As underlying hypervisor amazon uses xen, so the newest available kernel is
2.6.16 and I have no possibility to use a newer one.
Hello Phil,
Phil Miller:
I've a problem with aufs2 and my kernel. I'm building for chakra-project.org
the live medias. Since archlinux changed to kernel-33 series I've to update
mine aswell. Now I'm stuck in compiling madness:
:::
o bugfix
- aufs: bugfix, revalidate in setattr
- aufs stdalone: bugfix, include path for an ugly trick in f_op.c
This fix may not be enough. I will try again.
o news
- refine aufs slabs
pros: faster a little
cons: larger a little
I already started my work to get smaller.
- aufs: tiny,
Phil Miller:
I managed to build aufs2 now. I edited my PKGBUILD:
Glad to hear that.
Then I'd ask your opinion.
If I add one line -include priv.mk in config.mk and you
prepare the file priv.mk, then your build script will become much
simpler?
It will include priv.mk if it exists. Otherwise
Hi,
Jan Engelhardt:
13:25 xensrvneu:/ # mount -t aufs none /x -o br:/root=rw
/sbin/mount.aufs:plink.c:223: AUFS_CTL_PLINK_MAINT: Invalid argument
Currently I am guessing your mount(8) passed some incorrect parameters
to mount.aufs(8).
Try strace -f mount -t ... and send me the output please.
Tomas M:
So I decided to do something to help aufs stay in business:
When you donate to AUFS, I will double your donation from my pocket.
Read more here: http://www.slax.org/aufs.php
AMAZING!!
I have no word to represent my appreciation!
I really really appreciate your offer and whoever make
Jan Engelhardt:
Git-over-HTTP is known to be suboptimal; could you perhaps make these
repositories available via git:// transport too? Note that Sourceforge,
where aufs is hosted, does offer support for git, including multiple
repos per project.
Hmm, this not first time users talk about
Hi Phil,
Phil Miller:
I managed to build aufs2 now. I edited my PKGBUILD:
I noticed I was confused your case with concurrent another user's case.
For him, I added CFLAGS_f_op.o in fs/aufs/Makefile in the version
20100412. But I am considering to revert it since I am afraid he just
specified an
Phil Miller:
Also look into my attachments. There are all my files listed I have. There is
no kernel/mutex(-debug).h only in include ...
You are just listing up the contents of package, aren't you?
To build aufs, you need kernel SOURCE files.
Actually kernel/mutex.h exists in linux-2.6.33.2.
Hi Oleg,
Oleg:
This dir is very old :-). And we don't plan to remove or rename it in the
near future.
Why I asked how old is that it contains .wh..wh.opq or something.
If it is so old and you have used it with UnionFS or aufs1 in
unionfs-compat mode, the name may be .wh.diropq or something.
In your case, you (or someone else) tried creating something
video/Films/fileA (fileA was unprintable characters actually). Since
Correction.
- the path was video/Films/dirA/fileB and dirA was unprintable.
- aufs decided the second branch is the most-free-space (mfs).
- aufs tried creating
Jan Engelhardt:
Hans-Peter Jansen:
au_xino_create:659:mount[7044]: xino doesn't support /tmp/.aufs.xino(xfs)
Hmm, /tmp is xfs
Correct.
Aufs2 rejects placing XINO files on XFS while aufs1 allowed it.
Why exactly is that?
Because XFS doesn't maintain some necessary attributes for
Hans-Peter Jansen:
Okay, understood. Unfortunate, but true. Since more and more things are
moved to tmpfses nowadays, it might be a good idea to add another flag to
aufs sysfs interface:
/sys/fs/aufs/si_*/xi_size
You can find it if you mount debugfs.
For aufs1, such entries were located
o bugfix
- aufs bugfix: opaque ancestors
o misc
- aufs stdalone: stop including ../kernel/mutex.h,
and create a new file fs/aufs/mtx.h.
J. R. Okajima
--
- aufs2-2.6.git#aufs2-27 branch
aufs bugfix: opaque ancestors
Oleg:
So, my report:
kerndmesg is empty :-).
Everything works good.
Many thanks!
Thank you for the test and report.
The patch was included in this week's release, so you can use the latest
version in the public GIT tree.
J. R. Okajima
o news
- aufs: dynop, support AIO, Direct I/O and XIP
o misc
- aufs: tiny, revise Documentation/ABI
J. R. Okajima
--
- aufs2-2.6.git#aufs2-27 branch
aufs: tiny, rename fi_h_vm_ops to fi_hvmop
aufs: tiny, rename
Tomas M:
Not exactly, what I meant is the opposite - all donations are just
one time payments, however everybody is welcome to donate again
every month.
Yes, I understand what 'monthly' means. The sentence was not good in my
previous mail.
I know donation from everybody is NOT fixed. It may
Hello Yair,
Yair Yarom:
We have here diskless machines with aufsed /etc (rw tmpfs /etc_rw + ro
root /etc on /etc). When trying to run sungrid's qmaster daemon, we get
an oops. Without the aufsed /etc there's no oops (even when there are
other aufsed directories).
The kernel is 2.6.33.3
Jan Engelhardt:
Git-over-HTTP is known to be suboptimal; could you perhaps make these
repositories available via git:// transport too? Note that Sourceforge,
where aufs is hosted, does offer support for git, including multiple
repos per project.
Hmm, this not first time users talk
Hello Jordi,
Jordi Pujol:
I have found that in the booting process,
the root filesystem, that is mounted usign aufs, can not be modified,
It not works well until the system init procedure has been completed.
:::
+ mount -n -t squashfs -o ro,noatime
Jordi Pujol:
no more libraries are needed, all the libraries related to the commands=20
included in the initramfs file are also included in the initramfs file, and=
the=20
path is not changed until the main init procedure starts,
previous releases of aufs have executed that without
Jordi Pujol:
Sorry, I have not explained correctly, the mount process not stops, the boo=
t=20
script is stopped at a breakpoint (set by an option cheatcode). I should=20
What is breakpoint?
Do you mean it is an expected behaviour, and the problem is that vi
cannot save?
If so, how is your
Jordi Pujol:
a breakpoint is done by some code into the boot script, that tests if one=20
cheatcode is set, and therefore stops the program and brings the user into =
the=20
shell command mode,
I see.
Strictly speaking, the script didn't stop. It just invoked an
interactive shell.
But
Jordi Pujol:
ok, I will do,
Please test this patch. It passed most of my local tests.
J. R. Okajima
a.patch.bz2
Description: BZip2 compressed data
--
Jordi Pujol:
OK, aufs works with this patch.
I have built a linux kernel using the source from 2.6.33-3.slh.9,
adding the patch it works.
Thank you very much.
All of my local tests are passed too.
I refined the patch and it will be included in next Monday release.
J. R. Okajima
o bugfix
- aufs: bugfix, force setting file write operations for future copyup
Also remove dynamic operation for i_fop/f_op.
J. R. Okajima
--
- aufs2-2.6.git
aufs: tiny, revise CONFIG_AUFS_RDU description
aufs:
Hello Wilhelm,
Wilhelm:
is there a problem using make menuconfig in aufs2-33?
If I select some options under aufs I can't deselect them anymore and
the cursor permanently jumps to the notify-config-submenu.
I cannot reproduce the problem.
Will you send me your .config?
J. R. Okajima
Hello Joonwoo,
Joonwoo Park:
I should have mentioned some more.
I'm using 2.6.30.10 + aufs 2-standalone.tree-30-20100308
A r/w ubifs partition and another r/o ubifs partition constitute my
aufs partition.
:::
I've got following errors from r/w UBIFS partition and mounting aufs
is
Wilhelm:
The problem is the empty prompt in Kconfig in the section below. If I
change that to a non empty string as below (the string Bla) all is fine!
I tested it on ubuntu/lucid and debian/lenny.
I tested on my debian lenny system with plain aufs2-33, but I could not
reproduce the problem.
Joonwoo Park:
Sorry for confusion.
# mount -t ubifs -o rw /dev/brabra /mnt
# ls -al /mnt
(error occurs here, .wh.dir doesn't appear)
It seems that is the problem of ubifs.
No, I'm not using aufs2-util at all.
aufs2-util is necessary.
But I don't know that is the root cause of the
o bugfix
- aufs: bugfix, check the file handle strictly
- aufs: fill the prompt string for choice block
o news
- new branch aufs2-34
- begin supporting linux-2.6.35-rcN
J. R. Okajima
--
- aufs2-2.6.git#aufs2-27 branch
Hello Giorgos,
Giorgos Perivolarhs:
Hello, I am trying to compile the Linux kernel 2.6.32.8 with Aufs as a
module in order to create a Linux Live USB of my LFS (Linux From Scratch)
:::
$ git clone http://git.c3sl.ufpr.br/pub/scm/aufs/aufs2-standalone.gitaufs2-=
standalone.git
$ cd
Hello Tommy,
Tommy[D]:
just tried to compile aufs2-standalone against my local kernel sources, b=
ut get the attached failure
for 2.6.33. The kernel is a development kernel from Gentoo with additiona=
l pax and grsec patches.
For a more less patched kernel version 2.6.34 i had no issues
Tommy[D]:
Yes, that is your plain aufs2-standalone from today (this monday), i just=
take the git tree for
aufs2-standalone and the git tree for aufs2-util and put both into a tarb=
all, which then goes onto
mirrors. Depending on the current kernel, the right branch is then used t=
o
Tommy[D]:
just tried to compile aufs2-standalone against my local kernel sources, b=
ut get the attached failure
for 2.6.33. The kernel is a development kernel from Gentoo with additiona=
l pax and grsec patches.
I found the pax and grsec patches declare members as 'const'.
I found the pax and grsec patches declare members as 'const'.
grsecurity-2.1.14-2.6.33.4-201005151340.patch
pax-linux-2.6.33.4-test19.patch
:::
I don't know whether making them 'const' is necessary or not.
If you really need to co-work aufs and grsec/pax, then this patch will
solve
Hello Fariz,
simpleLinux:
I'd like to say thanks to Thomas M. and Junjiro Okajima for both your
Slax Core and AUFS. without you, We'll not have any potential
Malaysian Linux distribution. I would like to donate a big amount of
money once I opened a Paypal...
Glad to hear that.
We are
Hello Oliver,
Oliver Welter:
I am using aufs together with linux-vserver to mount a rootfs with a
data partiton. After having 4 nice years with 2.6.22 kernel I upgraded
to 2.6.33 with current aufs last week. Today, I got this trace on my
mailserver during a period of higher load.
Any ideas
o news
- reducing memory consumption #1, fi_hdir
now the internal pointer array for struct file object is allocated at
opening a dir only.
o misc
- aufs: tiny, update atime in splice_to
It is late at Monday night, Tuesday precisely.
I was late because the change in this week is big and took
Barry Kauler:
I got Aufs2 out of version control yesterday, and downloaded the
2.6.27.47 source. I applied Squashfs4 and Aufs patches, and it
compiled, but right at the end gave this error:
Error: mmap_min_addr [fs/aufs/aufs.ko] undefined!
You need to EXPORT it in security/min_addr.c,
o bugfix
- aufs: possible bugfix, revalidate inode race between readdir and lookup
- aufs: bugfix, the dentry paramter for security funcs
o misc
- aufs: tiny, follow the changes in linux-2.6.35-rcN
- aufs: follow linux-2.6.35, simple_setsize()
I recevied a donation via Tomas M who gathered
Ian Stakenvicius, Aerobiology Research:
This would generally make sense except that the issue is in
/usr/src/linux/include/linux, which as far as I can tell from the
makefiles, is never modified but only read. Indeed,
/usr/src/linux/include also seems that it is not modified except for the
Hello Maxim,
Maxim Litnitskiy:
Just checked out aufs-33.
Pathed he kernel for standalone version.
Got the following error:
:::
In file included from include/linux/gfp.h:4:0,
from include/linux/kmod.h:22,
from include/linux/module.h:13,
o news
- an external patch for pax/grsec users, grsecurity-2.1.14-2.6.33.4.patch
is added to aufs2-standalone.git#aufs2-33 branch.
J. R. Okajima
--
- aufs2-2.6.git
aufs: tiny, debug print [if]_version
-
Hello Robert,
Robert Wotzlaw:
Your script auroot A sample script to build a chroot-ed/jail environment for
Internet service has inspired me to write the script bldchraufs. The script
:::
I will thank You for, if You could take a look at the script.
Good script.
Do you have a plan to
Robert Wotzlaw:
it would be a great pleasure for me, if You could publish the script on You=
r
proposed git repository under the address aufs2-util.git/sample/auroot/. I
Ok, I will do.
It will be next Monday release.
AUFS. I think, the proposed control structures will hide the idea. For
Hello aluno3,
alu...@poczta.onet.pl:
I had problem with aufs 20081208 and 2.6.27.39 kernel. During normal
activity of the operating system I have got call trace (in attachment).
Process networking is my script to set IPs and so on. I know that this
version of aufs is a little old but I had
Hi Tomas,
Tomas M:
Tomas M makes a donation as monthly doubling.
I have tought about what I can do for users, especially for SLAX
users. Actually I have asked but there is no response.
I replied to your question some time ago, I hope you got the reply.
I told you that I really appreciate
Holger Schurig:
I've read documentation for aufs and there is some note that aufs2
utilities are always necessary.
That's not true. I use a basic aufs2 system on my device and
don't use or even have installed aufs utilities. It can all be
done with mount -o
I guess you'll need the
Tomas M:
So I would say the most problems do not lay with squashfs, but rather with
the tmpfs/ondisk_changes. Users would like to see some way of SNAPSHOTing,
Ok, I will keep it in mind.
The next report (after first report) will be about the aubrsync script
which should be useful for such
Hello Merten,
Lipkowski, Merten:
I've downloaded a plain vanilla 2.6.33.5 from www.kernel.org
Afterwards I've applied the real-time patch from Ingo Molnar. (http://www.k=
ernel.org/pub/linux/kernel/projects/rt/patch-2.6.33.5-rt23.bz2)
Although I didn't test by myself, there is a patch for
Hello Nikolaus,
Nikolaus Rath:
I have set up diskless fat clients with kernel 2.6.32. The root filesystem is
aufs. Underlying the aufs is an rw tmpfs and a ro squashfs, the later mounted
from NBD.
The problems is that the fat clients work fine for a little while, but then
reproducibly
Ian Stakenvicius, Aerobiology Research:
If you take a look at your syslog, I bet you will see at least one
additional hung task, that being an aufsd process. I usually see four
or five hung processes (i have a quad-core processor, you may see less
if you have fewer cores?), most of which are
Nikolaus Rath:
There are lots of other tasks here, but aufsd is not among them.
Do you mean all stopped at the similar function, ie. block I/O?
J. R. Okajima
--
This SF.net email is sponsored by Sprint
What will you
Nikolaus Rath:
Do you mean all stopped at the similar function, ie. block I/O?
Since the system becomes totally inaccessible, I have no idea what these
processes are up to. What I mean is that I get kernel messages of the for=
m
Jul 10 14:11:38 beta kernel: [25560.688091] INFO: task
Hello,
C Robinson:
I'm using linux-2.6.34 kernel with xenomai-2.5.3 patch.
:::
cd aufs2-standalone.git
sudo git checkout origin/aufs2
For linux-2.6.34, you need
$ git co origin/aufs2-34
J. R. Okajima
--
Hello Lou,
Thanks for detailed reserach.
Lou Gosselin:
I've been encountering the dir is overlapped error when using aufs.
I'd like to point out there are some scenarios where the overlap is
desirable, and not harmful.
:::
Test Case E # inner mount
mount -t aufs aufs union -o
As I wrote to this ML, I skipped this Monday release.
It was because I accidentally removed some files and lost some
changes. It was totally my fault.
Today I had done it again!
Some files are already restored, but I might loose
something. Unfortunately I don't remember everything and cannot be
Lou Gosselin:
Test Case E # This won't work without the modified aufs...
mkdir -p union a/mnt/original_files
mount -t tmpfs tmpfs a/mnt
mkdir -p a/mnt/new_mount_files
mount -t aufs aufs union -o br:a:a/mnt
find
:::
Therefor, in the resulting union there is no illogical
Hello Roopesh,
Roopesh:
While I was running fstest over an aufs mounted partition, the kernel
paniced at [exception RIP: dy_get+104].
It must be an aufs bug.
Please test this patch. Unfortunately I didn't test it by myself.
J. R. Okajima
a.patch.bz2
Description: BZip2 compressed data
Hello Thayumanavar,
Thayumanavar Sachithanantham:
In au_br_del, we free up br in au_br_do_del and later on access br.
Avoid deference on br to prevent crash.
Exactly.
Thank you very much.
J. R. Okajima
--
This
Thayumanavar Sachithanantham:
Fix a slab memory corruption seen when rmdir a directory of length
greater than 242 characters. This happen because in whout.c when cnt++
get large value and directory len become, we write past end of
allocated memory.
Thanx for patch.
I don't think such long
Hi,
Thayumanavar Sachithanantham:
correct name could not go beyond 242 chars and because of cnt/%d the
buffer could be overrun.
I think having
snprintf(p, WH_TMP_LEN+1, %.*x, AUFS_WH_TMP_LEN, cnt++); would be better.
'cnt' is 'short' and its size is guranteed as
Hi Lou,
Lou Gosselin:
Test Case E # This won't work without the modified aufs...
mkdir -p union a/mnt/original_files
mount -t tmpfs tmpfs a/mnt
mkdir -p a/mnt/new_mount_files
mount -t aufs aufs union -o br:a:a/mnt
find
I am reviewing the behaviour of overlapping, and found case
sf...@users.sourceforge.net:
Lou Gosselin:
Test Case E # This won't work without the modified aufs...
mkdir -p union a/mnt/original_files
mount -t tmpfs tmpfs a/mnt
mkdir -p a/mnt/new_mount_files
mount -t aufs aufs union -o br:a:a/mnt
find
I am reviewing the behaviour of
Lipkowski, Merten:
Not knowing what I'm doing, with a little bit try and error, I figured out,
that the 2.6.33.5-rt25 compiles if I reduce the a.patch to this:
:::
It seems as it works, but the tests the next days will show...
Can you tell me something about the risks and side
Hello Joonwoo,
Joonwoo Park:
While running rpm install command from the underneath partition, I got
kernel oops below.
I'm using aufs 2-standalone.tree-30-20100308 which is a bit old.
Can you see if this is aufs2 issue?
I think it is an aufs problem.
If you upgrade aufs, then the problem
Wolfgang Alper:
If the above seems to be correct, my question would be if there is an option
to cache branch information or something similar to avoid re-reading branches
(readdir), since the entire mountPoint can be considered RO (even the first
branch if needed) and no udba is in effect.
sf...@users.sourceforge.net:
The easy way to find which operation is issued to the ftp server is
watching on the ftp server. But you wrote that you don't know which
operation is issued. So here is my suggestion.
Test aufs readdir caching by using branches other than fuse. For
example,
Lou Gosselin:
How do I know which processes are locking a specific branch?
:::
# Any manipulation on the union results in locked file errors
mount union -o remount,del:branch1
mount: /tmp/union is busy
How about the verbose option?
(from the aufs manual)
Lou Gosselin:
I tried the verbose option, but it doesn't produce enough output.
mkdir union a b
mount -t aufs aufs union -o verbose,br:a=rw:b=rw
cat union/lock # in separate term
mount union -o remount,del:a
[ 114.704254] aufs au_br_del:758:mount[1215]: 1 file(s) opened
mount:
Lou Gosselin:
You mention lsof, but I don't think it can identify which file handles
are responsible for blocking a specific aufs branch (my former post had
an example of this).
Do you mean that you want to find the tail process accessing
branch1_file, ie. want to make sure that
Waoh, most of the patch becomes unnecessary so soon.
I'll take a look at 2.6.33.5-rt25 next week.
For 2.6.33.5-rt25, you need the patch for fs/aufs/mtx.h which you
mentioned and the patch I've told its URL in my first reply
(fs/aufs/rwsem.h).
J. R. Okajima
Lou Gosselin:
I tried to detect locked branches based on the paths as mentioned below.
However I'm finding that the below algorithm does not reliably determine
the branch that a file lock is on.
As I wrote, there are some complicated cases in real world.
mkdir union a b
cp /bin/sh a
Alexander Rau:
On the other hand, in the env with installed util mount.aufs is found,
but the branch is not created, failing with unexpected reloc type in
static...
I am afraid you failed building /sbin/mount.aufs and it is broken.
For instance, 32bit binary on 64 bit kernel without
Lou Gosselin:
However, as shown in last email, knowing which file is causing the lock
is insufficient because mmaped files could be sourced from any branch.
There are of course many scenarios which could cause this: copy-up, file
unlink, remount prepend, etc.
So, in the case of mmaped
I am afraid you won't be happy even if I develop a new feature such like
- ioctl interface to get the opened file path
- /sys/fs/aufs/*/opened_files
- or something.
Because it is essentially equivalent to Magic SysRq.
I've got a tiny idea.
Try commenting-out a line in
- aufs2-2.6.git
aufs: more verbose at deleting a branch
aufs: tiny, note about the vdir cache
- aufs2-standalone.git
ditto
- aufs2-util.git
aufs: tiny, note about the vdir cache
tiny, fix typo
J. R. Okajima
T o n g:
- aufs: allow a loopback-mounted branch
:::
Would you elaborate more on this please?
If you have the git tree, try git show 94a8852 and you will see
aufs: allow a loopback-mounted branch
It is rejected to add a branch which is a loopback-mount in another
T o n g:
If my RW branch only overwrite old file or write to new files, can this
aubrsync handle multiple RO branches? If not, what are my other choices
are?
The simple interface of the script doesn't support more than two
branches. With the complex interface (parameters), you may be able
T o n g:
hmm, am I doing something wrong, or hit another bug?
% mount -vt aufs -o br:./rw:./ro2:./ro1 none ./u
none on /dev/shm/aumt/u type aufs (rw,relatime,si=a5a291785125542e)
% mount -o remount,ro,shwh,mod:ro2=ro+wh u
mount: /dev/shm/aumt/u not mounted already, or bad option
You can
T o n g:
So to conclude, if *all* my middle layers are mounted with +wh, eg:
mount -t aufs -o br:./rw:./ro2=ro+wh:./ro1 none ./u
and I move RW branch down layer by layer, the above script will work
regardless of number of middle layers (subject to aufs max), Is that so?
I guess so, but
Thayumanavar Sachithanantham:
The kernel is 2.6.27.45 SLES kernel with aufs ( recent ver as taken
git for 2.6.27) pulled into it.
How recent is it?
For instance, I made some bugfixes recently. Do you think they are
related to your problem?
4946ca0 2010-07-21 aufs: bugfix, buffer overrun in
Thayumanavar Sachithanantham:
The kernel is 2.6.27.45 SLES kernel with aufs ( recent ver as taken
git for 2.6.27) pulled into it.
What is the difference between vanilla linux-2.6.27.45 and SUSE Linux
Enterprise Server? Can I get the diff or full source files?
If SUSE backports kmemleak, then
Thayumanavar Sachithanantham:
i tried the patch that you seem to have provided but seem to have been
hit the issue. Once hit the issue i provide the details of the log.
I don't understand you well. If you apply the debug patch successfully,
reproduce the problem, and get the debug log, then
701 - 800 of 1884 matches
Mail list logo
