Debian Bug report logs - #317479
mount /boot ro just before suspend, if possible

Toggle useless messages

Message #3 received at submit@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: hibernate messes up menu.lst file

Date: Sat, 9 Jul 2005 01:16:08 +0200

[Message part 1 (text/plain, inline)]

Package: hibernate
Version: 1.09-1
Severity: important

I can reproduce this at will: if I hibernate and reboot, grub will
only show the prompt but no menu. At this time, the attached grub
menu.lst file is in place. If I run `hibernate --restore-grub`, the
boot works again as expected with the menu.

I can't figure out what could cause grub to b0rk to the menu when
the attached file is in use.

-- System Information:
Debian Release: 3.1
Architecture: i386 (i686)
Kernel: Linux 2.6.12-wing
Locale: LANG=en_GB, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8)

-- no debconf information

-- 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer, admin, user, and author
`. `'`
  `-  Debian - when you have better things to do than fixing a system
 
Invalid/expired PGP subkeys? Use subkeys.pgp.net as keyserver!
 
"i might disagree with what you have to say,
 but I'll defend to the death your right to say it."
                                                           -- voltaire

[menu.lst (text/plain, attachment)]

[signature.asc (application/pgp-signature, inline)]

Message #6 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: 317479@bugs.debian.org

Subject: double grub menu change

Date: Sat, 9 Jul 2005 01:56:08 +0200

[Message part 1 (text/plain, inline)]

Curiously, if I set ChangeGrubMenu to 'no' in hibernate.conf, the
menu is still being changed, but now the problem does not occur.
Thus, it seems as if somehow the change is done twice, once
according to the config file, and once in all cases.

-- 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system
 
Invalid/expired PGP subkeys? Use subkeys.pgp.net as keyserver!
 
"time flies like an arrow. fruit flies like a banana."
                                                       -- groucho marx

[signature.asc (application/pgp-signature, inline)]

Message #9 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: 317479@bugs.debian.org

Subject: Re: double grub menu change

Date: Sat, 9 Jul 2005 02:10:31 +0200

[Message part 1 (text/plain, inline)]

also sprach martin f krafft <madduck@debian.org> [2005.07.09.0156 +0200]:
> Curiously, if I set ChangeGrubMenu to 'no' in hibernate.conf, the
> menu is still being changed, but now the problem does not occur.
> Thus, it seems as if somehow the change is done twice, once
> according to the config file, and once in all cases.

Sorry, the file does *not* get rewritten. I still had an old version
in place. If I set ChangeGrubMenu to yes, it is changed in a way
that makes grub display only a prompt. If I set it to no, everything
works as it should, though no grub menu indication exists.

-- 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system
 
Invalid/expired PGP subkeys? Use subkeys.pgp.net as keyserver!
 
"of course the music is a great difficulty.
 you see, if one plays good music, people don't listen,
 and if one plays bad music people don't talk."
                                                        -- oscar wilde

[signature.asc (application/pgp-signature, inline)]

Message #14 received at submit@bugs.debian.org (full text, mbox, reply):

From: Cameron Patrick <cameron@patrick.wattle.id.au>

To: martin f krafft <madduck@debian.org>, 317479@bugs.debian.org

Cc: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: Bug#317479: hibernate messes up menu.lst file

Date: Sat, 9 Jul 2005 11:50:25 +0800

[Message part 1 (text/plain, inline)]

On Sat, Jul 09, 2005 at 01:16:08AM +0200, martin f krafft wrote:

> I can't figure out what could cause grub to b0rk to the menu when
> the attached file is in use.

> ### BEGIN HIBERNATE SENTINEL
> 
> title _____________________________________________________________________
> configfile dummy
> title WARNING: Linux 2.6.12-wing is suspended via Software Suspend!"
> configfile dummy
> ### END HIBERNATE SENTINEL

Does getting rid of the trailing quote on the second title line help?

Cameron

[signature.asc (application/pgp-signature, inline)]

Message #22 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Cameron Patrick <cameron@patrick.wattle.id.au>

Cc: 317479@bugs.debian.org

Subject: XFS problem

Date: Sat, 9 Jul 2005 11:12:15 +0200

[Message part 1 (text/plain, inline)]

retitle 317479 hibernate does not play well with XFS /boot
thanks

also sprach Cameron Patrick <cameron@patrick.wattle.id.au> [2005.07.09.0550 +0200]:
> > title WARNING: Linux 2.6.12-wing is suspended via Software Suspend!"
> 
> Does getting rid of the trailing quote on the second title line help?

Oh, that issue. No, it does not fix it, but it's a separate bug.

No, this time round it's really weird.

If I suspend, the grub prompt is shown.
If I reboot normally, the menu is shown.

I once rebooted shortly after the kernel started loading after
seeing the prompt and then got the menu.

Thus I think something must be happening at the early stages of
initialisation which makes grub work on subsequent reboots.

Here's another peculiarity:

In the grub prompt, I can

  cat /grub/menu.lst.working

(that's a backup copy), but when I do

  cat /grub/menu.lst

I get: Error 2: Bad file or directory type
"This error is returned if a file requested is not a regular file,
but something like a symbolic link, directory, or FIFO."

But I cannot determine what the problem is. Booting off a CD and
mounting the /boot filesystem shows a normal file.

And now I found something even more interesting. This is probably
the same as #246111 and all the merged blocks. My /boot is an XFS
filesystem, and menu.lst does not get committed in time. If
I repair a filesystem dump:

wing:/tmp# xfs_repair -L part1
[...]
imap claims in-use inode 136 is free, correcting imap
[...]
entry "menu.lst" at block 0 offset 1608 in directory inode 132 references free inode 156
  clearing inode number in entry at offset 1608...
[...]
disconnected inode 136, moving to lost+found
[...]

you can guess what inode 136 holds... menu.lst.

I don't know of a fix to this problem, but I have contacted the XFS
mailing list about it.

-- 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system
 
Invalid/expired PGP subkeys? Use subkeys.pgp.net as keyserver!
 
whatever you do will be insignificant,
but it is very important that you do it.
                                                     -- mahatma gandhi

[signature.asc (application/pgp-signature, inline)]

Message #29 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Cameron Patrick <cameron@patrick.wattle.id.au>

To: martin f krafft <madduck@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: XFS problem

Date: Sat, 9 Jul 2005 18:26:01 +0800

[Message part 1 (text/plain, inline)]

On Sat, Jul 09, 2005 at 11:12:15AM +0200, martin f krafft wrote:

> I don't know of a fix to this problem, but I have contacted the XFS
> mailing list about it.

Could you try running xfs_freeze and syncing after modifying the
menu.lst file, then xfs_freeze -u after resuming before changing it
back?  XFS does freaky caching of its own, so the "sync" that suspend2
does is not good enough to make sure that the filesystem state is
updated on disc.

Cameron.

[signature.asc (application/pgp-signature, inline)]

Message #34 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Henrique de Moraes Holschuh <hmh@debian.org>

To: 317479@bugs.debian.org, 317479-submitter@bugs.debian.org

Date: Tue, 30 May 2006 20:44:30 -0300

tag 317479 moreinfo
thanks

Martin, unless we have a way to call fsync() from the shell, a read-only
mount (known good solution) or a xfs_freeze (must be tested) is probably the
only way to work around it.

Does current XFS still exibit the issue you described?  Do they still cache
inodes in memory through sync()?

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Message #42 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Henrique de Moraes Holschuh <hmh@debian.org>, 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 02:43:46 +0200

[Message part 1 (text/plain, inline)]

also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.05.31.0144 +0200]:
> Martin, unless we have a way to call fsync() from the shell, a read-only
> mount (known good solution) or a xfs_freeze (must be tested) is probably the
> only way to work around it.

Is there anything to be said against mounting /boot r/o on
hibernate? Can we do it as a final last step, after most everything
has been suspended? Or is it not possible? I bet there are problems
if /boot is on the / partition.

> Does current XFS still exibit the issue you described?  Do they
> still cache inodes in memory through sync()?

I will check this when I get back to kernel-patch-suspend2 in a bit
of time, but I did not notice any changes, so I assume yes.

Btw: are you now working on hibernate as well?

Cheers,

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #47 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Henrique de Moraes Holschuh <hmh@debian.org>

To: martin f krafft <madduck@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Thu, 1 Jun 2006 23:48:26 -0300

On Fri, 02 Jun 2006, martin f krafft wrote:
> Is there anything to be said against mounting /boot r/o on
> hibernate? Can we do it as a final last step, after most everything
> has been suspended? Or is it not possible? I bet there are problems
> if /boot is on the / partition.

IMO /boot should be read-only almost all the time, the same goes for /usr.

I don't see why we couldn't add a scriptlet to mount partitions read-only
before suspend, though.  Sort of like the automated module unload/load.
Patches are welcome, I am sure :-)

As for /boot on /, I am afraid mount -o ro,remount is NOT an option.  If
there *is* a way to tell XFS to *really* flush everything, it would be nice
to do that, otherwise it is time to fix it in the kernel.

> > Does current XFS still exibit the issue you described?  Do they
> > still cache inodes in memory through sync()?
> 
> I will check this when I get back to kernel-patch-suspend2 in a bit
> of time, but I did not notice any changes, so I assume yes.
> 
> Btw: are you now working on hibernate as well?

I want my T43 to work right, you see :-)  That means I am doing an once-over
everything, so far I fixed the nsc-ircc kernel driver, triaged ACPI bugs and
got at least one on the way to a fix, triaged and fixed crappage in
hibernate, wrote a lot of acpi scripts that I will find a way to package
somehow, found out why ThinkPad HDs die early (it is amazing people NEVER
noticed IBM sets the hd APM mode and that causes them to park heads all the
time...)

It's been a lot of fun so far ;-)

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Message #50 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Henrique de Moraes Holschuh <hmh@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 08:33:44 +0200

[Message part 1 (text/plain, inline)]

also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.0448 +0200]:
> IMO /boot should be read-only almost all the time, the same goes
> for /usr.

That's not the issue. :)

> I don't see why we couldn't add a scriptlet to mount partitions
> read-only before suspend, though.  Sort of like the automated
> module unload/load. Patches are welcome, I am sure :-)

Would it work? Normally on a running system you can't just remount
partitions r/o.

> As for /boot on /, I am afraid mount -o ro,remount is NOT an
> option.  If there *is* a way to tell XFS to *really* flush
> everything, it would be nice to do that, otherwise it is time to
> fix it in the kernel.

I think it sounds like the latter. Argh.

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #55 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Henrique de Moraes Holschuh <hmh@debian.org>

To: martin f krafft <madduck@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 08:17:05 -0300

On Fri, 02 Jun 2006, martin f krafft wrote:
> Would it work? Normally on a running system you can't just remount
> partitions r/o.

Yes, you can. As long as you don't have open files there.  If /boot is
indeed a partition, there is very little chance of open files there...

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Message #58 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Henrique de Moraes Holschuh <hmh@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 13:35:35 +0200

[Message part 1 (text/plain, inline)]

also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1317 +0200]:
> Yes, you can. As long as you don't have open files there.  If /boot is
> indeed a partition, there is very little chance of open files there...

Sure, but if the system's a single / filesystem, you lose.

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #63 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Henrique de Moraes Holschuh <hmh@debian.org>

To: martin f krafft <madduck@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 10:34:59 -0300

Hi martin!

On Fri, 02 Jun 2006, martin f krafft wrote:

> also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1317 +0200]:
> > Yes, you can. As long as you don't have open files there.  If /boot is
> > indeed a partition, there is very little chance of open files there...
> 
> Sure, but if the system's a single / filesystem, you lose.

Yes, you lose.  Don't use XFS for /, or /boot for that matter.  It is not
safe.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Message #66 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Henrique de Moraes Holschuh <hmh@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 15:38:44 +0200

[Message part 1 (text/plain, inline)]

also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1534 +0200]:
> Yes, you lose.  Don't use XFS for /, or /boot for that matter.  It
> is not safe.

But yes it is. In fact, I consider it the safest around. But
generally only if you have a battery-backed power-supply, which
curiously doesn't apply to laptops the way I mean it.

Since hibernate only really applies to laptops and desktops, and
they aren't really XFS candidates, I guess the bug is a wontfix for
now. :(

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #71 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Henrique de Moraes Holschuh <hmh@debian.org>

To: martin f krafft <madduck@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 11:05:06 -0300

On Fri, 02 Jun 2006, martin f krafft wrote:
> also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1545 +0200]:
> > No, it is not.  XFS has bad data-ordering-flush semanthics for
> > resilience, they are superb for speed and performance, NOT for
> > data safety.  It won't guarantee data safety unless you fsync(),
> > umount or mount -o ro,remount... and it does *not* guarantee data
> > safety on global sync(), which is what this bug is about.
> 
> I suppose you have thought about this before. :)

I have been on the bad side of XFS / before, I take my time to warn people
off that every time I see someone too happy about an all-XFS system :-)

> > I still think an hibernate scriptlet to mount partitions readonly
> > (instead of just umounting them) is a valid, useful wish.  And it
> > would fix the issue with /boot as XFS (but not of / as XFS).
> 
> Okay. Retitle the bug then?

Please do, I am only spending a little time on hibernate, and I am not even
a co-maintainer, so I'd rather the bug owner did this :-)

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Message #76 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Henrique de Moraes Holschuh <hmh@debian.org>

To: martin f krafft <madduck@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 10:45:54 -0300

On Fri, 02 Jun 2006, martin f krafft wrote:
> also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1534 +0200]:
> > Yes, you lose.  Don't use XFS for /, or /boot for that matter.  It
> > is not safe.
> 
> But yes it is. In fact, I consider it the safest around. But
> generally only if you have a battery-backed power-supply, which
> curiously doesn't apply to laptops the way I mean it.

No, it is not.  XFS has bad data-ordering-flush semanthics for resilience,
they are superb for speed and performance, NOT for data safety.  It won't
guarantee data safety unless you fsync(), umount or mount -o ro,remount...
and it does *not* guarantee data safety on global sync(), which is what this
bug is about.

Also, XFS does not tolerate use of its three first sectors for MBR /
bootblocks, which is not a defect or anything, but it *is* something you
must remember or you will corrupt its first superblock.

> Since hibernate only really applies to laptops and desktops, and
> they aren't really XFS candidates, I guess the bug is a wontfix for
> now. :(

If you *do* take the time to make sure sync() is NOT flushing everything XFS
to disk, *please* file a bug on kernel.org, otherwise it will never be fixed
where it really matters.

I still think an hibernate scriptlet to mount partitions readonly (instead
of just umounting them) is a valid, useful wish.  And it would fix the issue
with /boot as XFS (but not of / as XFS).

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Message #79 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Henrique de Moraes Holschuh <hmh@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 16:11:22 +0200

[Message part 1 (text/plain, inline)]

severity 317479 wishlist
retitle 317479 mount /boot ro just before suspend, if possible
thanks

also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1605 +0200]:
> > Okay. Retitle the bug then?
> 
> Please do, I am only spending a little time on hibernate, and I am not even
> a co-maintainer, so I'd rather the bug owner did this :-)

Done. Thanks. Please read the rest of the thread (not you, hmh) for
the reasoning.

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #82 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Henrique de Moraes Holschuh <hmh@debian.org>

Cc: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Fri, 2 Jun 2006 15:52:45 +0200

[Message part 1 (text/plain, inline)]

also sprach Henrique de Moraes Holschuh <hmh@debian.org> [2006.06.02.1545 +0200]:
> No, it is not.  XFS has bad data-ordering-flush semanthics for
> resilience, they are superb for speed and performance, NOT for
> data safety.  It won't guarantee data safety unless you fsync(),
> umount or mount -o ro,remount... and it does *not* guarantee data
> safety on global sync(), which is what this bug is about.

I suppose you have thought about this before. :)

> If you *do* take the time to make sure sync() is NOT flushing
> everything XFS to disk, *please* file a bug on kernel.org,
> otherwise it will never be fixed where it really matters.

Yes.

> I still think an hibernate scriptlet to mount partitions readonly
> (instead of just umounting them) is a valid, useful wish.  And it
> would fix the issue with /boot as XFS (but not of / as XFS).

Okay. Retitle the bug then?

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #97 received at 317479-close@bugs.debian.org (full text, mbox, reply):

From: madduck@debian.org (martin f. krafft)

To: 317479-close@bugs.debian.org

Subject: Bug#317479: fixed in hibernate 1.91-2

Date: Mon, 05 Jun 2006 08:17:10 -0700

Source: hibernate
Source-Version: 1.91-2

We believe that the bug you reported is fixed in the latest version of
hibernate, which is due to be installed in the Debian FTP archive:

hibernate_1.91-2.diff.gz
  to pool/main/h/hibernate/hibernate_1.91-2.diff.gz
hibernate_1.91-2.dsc
  to pool/main/h/hibernate/hibernate_1.91-2.dsc
hibernate_1.91-2_all.deb
  to pool/main/h/hibernate/hibernate_1.91-2_all.deb



A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to 317479@bugs.debian.org,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
martin f. krafft <madduck@debian.org> (supplier of updated hibernate package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing ftpmaster@debian.org)


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Format: 1.7
Date: Mon,  5 Jun 2006 16:49:39 +0200
Source: hibernate
Binary: hibernate
Architecture: source all
Version: 1.91-2
Distribution: unstable
Urgency: low
Maintainer: martin f. krafft <madduck@debian.org>
Changed-By: martin f. krafft <madduck@debian.org>
Description: 
 hibernate  - smartly puts your computer to sleep (suspend to RAM or disk)
Closes: 317479 354341 354350
Changes: 
 hibernate (1.91-2) unstable; urgency=low
 .
   * Taking over principal maintainership.
   * Let's not reverse the patch for #369632. Sorry.
   * Added scriptlet to remount /boot read-only if /boot is a separate
     partition, contains an XFS filesystem, hibernate was asked to update the
     Grub menu file, and this feature has not been disabled in the
     configuration file (closes: #317479).
   * Changed BackupGrubMenuFile in example configuration to Debian location.
   * Reworded package description and copyright file a bit.
   * Added debian/watch file.
   * Just noticed that hibernate.conf(5) documents the order with which options
     are processed if specified multiple times (closes: #354341, #354350).
Files: 
 24de62c77ab766a7ce8dea73f22ddf49 631 utils extra hibernate_1.91-2.dsc
 869a46a659dd5740e46285bd80df2dcc 9258 utils extra hibernate_1.91-2.diff.gz
 b19ec61616b9f0a1a7047b0f43f7bc78 74596 utils extra hibernate_1.91-2_all.deb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)

iD8DBQFEhEW9IgvIgzMMSnURAkMlAJ9vkV/8jla8FIg5oG2536lYQr5LRwCfYFaN
1gr8wdU2Y71Q4Q3fIx8hKTU=
=uUc7
-----END PGP SIGNATURE-----

Message #102 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Bernard Blackham <bernard@blackham.com.au>

To: 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Mon, 5 Jun 2006 23:51:43 +0800

On Fri, Jun 02, 2006 at 10:45:54AM -0300, Henrique de Moraes Holschuh wrote:
> No, it is not.  XFS has bad data-ordering-flush semanthics for resilience,
> they are superb for speed and performance, NOT for data safety.  It won't
> guarantee data safety unless you fsync(), umount or mount -o ro,remount...
> and it does *not* guarantee data safety on global sync(), which is what this
> bug is about.

Does XFS support remounting with the wsync option? Perhaps something
like:
    mount -o remount,sync,wsync /boot

If that does the right thing (gets data sync'd to disk ASAP), then
it might be a cleaner solution to include that into the grub
scriptlet, and extendable to systems without a /boot.

Bernard.

-- 
 Bernard Blackham <bernard at blackham dot com dot au>

Message #105 received at 317479@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: Bernard Blackham <bernard@blackham.com.au>, 317479@bugs.debian.org

Subject: Re: Bug#317479: (no subject)

Date: Mon, 5 Jun 2006 19:16:58 +0200

[Message part 1 (text/plain, inline)]

reopen 317479
retitle mount partition containing /boot sync,wsync just before suspend
thanks

also sprach Bernard Blackham <bernard@blackham.com.au> [2006.06.05.1751 +0200]:
> Does XFS support remounting with the wsync option? Perhaps something
> like:
>     mount -o remount,sync,wsync /boot

It does. Lemme verify that it works. Might take a while.

> If that does the right thing (gets data sync'd to disk ASAP), then
> it might be a cleaner solution to include that into the grub
> scriptlet, and extendable to systems without a /boot.

Yeah true, I can do that to the / partition while it's mounted.

Great. Thanks!

-- 
Please do not send copies of list mail to me; I read the list!
 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system

[signature.asc (application/pgp-signature, inline)]

Message #114 received at 317479-submitter@bugs.debian.org (full text, mbox, reply):

From: martin f krafft <madduck@debian.org>

To: 317479-submitter@bugs.debian.org, 370713-submittter@bugs.debian.org, 377389-submitter@bugs.debian.org, 378282-submitter@bugs.debian.org, 378347-submitter@bugs.debian.org, 381033-submitter@bugs.debian.org, 381903-submitter@bugs.debian.org, 381927-submitter@bugs.debian.org, 384140-submitter@bugs.debian.org, 384471-submitter@bugs.debian.org, 386426-submitter@bugs.debian.org, 386623-submitter@bugs.debian.org, 386799-submitter@bugs.debian.org

Subject: please test latest hibernate package: 1.93.svn1064-1~unreleased.2

Date: Wed, 13 Sep 2006 17:39:07 +0200

[Message part 1 (text/plain, inline)]

Please test the latest hibernate package, which fixes a bug you've
reported. Do note that the package may still change in certains ways
while I am trying to synchronise with upstream. I cannot guarantee
an upgrade path from this version to the next one. Thanks for your
understanding.

http://debian.madduck.net/repo/dists/UNRELEASED/main/source/utils/hibernate_1.93.svn1064-1~unreleased.2.dsc
http://debian.madduck.net/repo/dists/UNRELEASED/main/binary-all/utils/hibernate_1.93.svn1064-1~unreleased.2_all.deb
deb http://debian.madduck.net/repo UNRELEASED main
deb-src http://debian.madduck.net/repo UNRELEASED main

Comments welcome. I'd appreciate if you sent them to the relevant
bug report, or at least CC'ed it.

Thanks,

-- 
 .''`.     martin f. krafft <madduck@debian.org>
: :'  :    proud Debian developer, author, administrator, and user
`. `'`     http://people.debian.org/~madduck http://debiansystem.info
  `-  Debian - when you have better things to do than fixing systems

[signature.asc (application/pgp-signature, inline)]

Message #119 received at 317479-close@bugs.debian.org (full text, mbox, reply):

From: madduck@debian.org (martin f. krafft)

To: 317479-close@bugs.debian.org

Subject: Bug#317479: fixed in hibernate 1.94-1

Date: Tue, 10 Oct 2006 03:05:50 -0700

Source: hibernate
Source-Version: 1.94-1

We believe that the bug you reported is fixed in the latest version of
hibernate, which is due to be installed in the Debian FTP archive:

hibernate_1.94-1.diff.gz
  to pool/main/h/hibernate/hibernate_1.94-1.diff.gz
hibernate_1.94-1.dsc
  to pool/main/h/hibernate/hibernate_1.94-1.dsc
hibernate_1.94-1_all.deb
  to pool/main/h/hibernate/hibernate_1.94-1_all.deb
hibernate_1.94.orig.tar.gz
  to pool/main/h/hibernate/hibernate_1.94.orig.tar.gz



A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to 317479@bugs.debian.org,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
martin f. krafft <madduck@debian.org> (supplier of updated hibernate package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing ftpmaster@debian.org)


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Format: 1.7
Date: Tue, 10 Oct 2006 09:41:11 +0200
Source: hibernate
Binary: hibernate
Architecture: source all
Version: 1.94-1
Distribution: unstable
Urgency: low
Maintainer: martin f. krafft <madduck@debian.org>
Changed-By: martin f. krafft <madduck@debian.org>
Description: 
 hibernate  - smartly puts your computer to sleep (suspend to RAM or disk)
Closes: 317479 317479 370713 377389 378282 378347 381033 381903 381927 384140 384471 386426 386623 386799
Changes: 
 hibernate (1.94-1) unstable; urgency=low
 .
   * New upstream release:
     - properly escape amarok process listing (closes: #378282).
     - does not fail if KDE sessions are already locked (closes: #384471).
     - enables verbose output earlier (closes: #370713).
   * Merged several patches into upstream code:
     - Patch by Joey Hess to save/restore brightness levels with GFX cards
       (closes: #384140).
     - Patch against vbetool scriptlet to log stderr output to log file, and
       only let output through to real stderr if verbosity >=2 is selected
       (closes: #381927).
     - Corrected typo in hibernate.conf manpage (closes: #386799).
     - Reworded the init.d script message about invalidating stale signatures
       (closes: #381903).
     - Added ath_pci to blacklisted modules (see #383169).
     - Added scriptlet to remount /boot read-only if /boot is a separate
       partition, contains an XFS filesystem, hibernate was asked to update the
       Grub menu file, and this feature has not been disabled in the
       configuration file (closes: #317479, #317479, #386623).
       Also added README.xfs to document caveats for users of the XFS
       filesystem.
     - Improved ususpend method. Now supports USuspendRamForce (boolean) to
       pass the -f option to s2ram. Also tests with s2ram's -n option if s2ram
       is used and the option is not set. I also made disk.conf and ram.conf
       prefer the uswsusp methods, if they are available. If you do not want to
       use them, use the sysfs-{ram,disk}.conf files, which are the former
       ram.conf and disk.conf files.
     - Use /proc/net/dev instead of ifconfig to enumerate network devices,
       because the latter would ignore madwifi interfaces.
   * Now actually installs the ususpend configuration (though not
     ususpend.conf), which I had previously forgotten. Thanks to René van
     Bevern and Christian Schuerer for pointing it out (closes: #378347).
   * Added recommendation on vbetool, thanks to Minh Ha Duong.
   * Added recommendation on uswsusp, so that the new uswsusp code gets used
     if supported by the kernel (>= 2.6.17rc1), and suspend2 is not available.
   * Fixed up recommendations (other packages that most people will want to
     use) and suggestions (other packages used but not needed by many).
   * Added a hibernation HOWTO by Barak A. Pearlmutter to
     /usr/share/doc/hibernate (closes: #377389).
   * Moved vbetool options from sysfs-ram.conf to ram.conf; they're also
     applicable to ususpend.
   * Create log file with 0644:root:adm permissions (closes: #381033).
   * Added hibernate-ram and hibernate-disk symlinks which automatically use
     ram.conf or disk.conf as configuration file, rather than hibernate.conf
     (closes: #386426).
Files: 
 e825e6c5be851155747a3f223978de23 701 utils extra hibernate_1.94-1.dsc
 f8826d4c96e320902382f29c91694c04 72928 utils extra hibernate_1.94.orig.tar.gz
 7a1bf5abd28f5f29fc1debb54d2cec13 12416 utils extra hibernate_1.94-1.diff.gz
 3b41e5e2609a2bb2177a449fdd180a7c 85936 utils extra hibernate_1.94-1_all.deb

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)

iD8DBQFFK1eCIgvIgzMMSnURAsxKAJ9DAlMDbmMTJl3I9Sf5B4ULXiauSwCfRN2f
Jl+FH+8aSyhqUWnIP4iaIaI=
=jK/A
-----END PGP SIGNATURE-----

Message #129 received at 317479@bugs.debian.org (full text, mbox, reply):

From: Oleg Verych <olecom@flower.upol.cz>

To: safe data on filesystem <317479@bugs.debian.org>

Subject: LKML opinion on safe data on filesystem [Re: [PATCH] Freeze bdevs when freezing processes.]

Date: Tue, 24 Oct 2006 22:03:17 +0200

Message-ID: <20061024170857.GB17956@infradead.org>

----- Forwarded message from Christoph Hellwig -----

On Tue, Oct 24, 2006 at 04:20:31PM +0000, Oleg Verych wrote:
> > Do you mean calling sys_sync() after the userspace has been frozen
> > may not be sufficient?
> 
> Please see
> <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=317479;msg=105;att=0>
> 
> it's bottom of
> <http://bugs.debian.org/317479>
> 
> IMHO it's may be helpful.

It's not.  It'sa step in the wrong direction.  The only way to guarantee
a filesystem (not just xfs, _any_ filesystem - xfs is just most sensitive)
is to call the write_super_lockfs method.

----- End forwarded message -----

Send a report that this bug log contains spam.