[Top][All Lists]

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

looking for snapshot hurd build (was: Re: system crash when much disk io

From: Joost van Baal
Subject: looking for snapshot hurd build (was: Re: system crash when much disk io)
Date: Mon, 8 Apr 2002 18:33:37 +0200
User-agent: Mutt/1.3.27i

Hi Neal,

[Sorry for crossposting, Mail-Followup-To: bug-hurd@gnu.org .]

Sorry for this ridiculously late reply; I've been quite busy (see below
on doing what... ;-)

On Sun, Mar 31, 2002 at 10:12:38PM -0500, Neal H Walfield wrote:
> > > Running dselect or apt-get very often gets my hurd box on it's knees.
> > > When it's 'Reading Package Lists...' , the system hangs, after giving:
> > >
> > > ext2fs.static: ../../ext2fs/pager.c:396: file_pager_write_page:
> > > Assertion `block' failed.
> Can you send us the output of `tune2fs -l /dev/PART' where part is the
> partition with the package files?

here it is:

root@stegun-hurd ~# tune2fs -l /dev/hd0s6
tune2fs 1.25 (20-Sep-2001)
Filesystem volume name:   <none>
Last mounted on:          <not available>
Filesystem UUID:          423a23a2-b3d7-4279-b545-ac0fa1753573
Filesystem magic number:  0xEF53
Filesystem revision #:    1 (dynamic)
Filesystem features:      sparse_super
Filesystem state:         not clean
Errors behavior:          Continue
Filesystem OS type:       GNU/Hurd
Inode count:              130560
Block count:              522049
Reserved block count:     26102
Free blocks:              221293
Free inodes:              102876
First block:              1
Block size:               1024
Fragment size:            1024
Blocks per group:         8192
Fragments per group:      8192
Inodes per group:         2040
Inode blocks per group:   255
Last mount time:          Thu Jan  2 23:24:26 1997
Last write time:          Tue Apr  2 00:34:11 2002
Mount count:              0
Maximum mount count:      25
Last checked:             Tue Apr  2 00:34:11 2002
Check interval:           15552000 (6 months)
Next check after:         Sun Sep 29 00:34:11 2002
Reserved blocks uid:      0 (user root)
Reserved blocks gid:      0 (group root)
First inode:              11
Inode size:               128

Hmm.. I suffer from clock skewness...

> I think that this patch will solve your problem:
> Index: pager.c
> ===================================================================
> RCS file: /cvsroot/hurd/hurd/ext2fs/pager.c,v

> This code is only run in the case of partial pages, i.e. where disk
> blocks are less than the size of virtual memory pages.  This is not
> the normal case; most users will have partitions >512MB and disk
> blocks of size 4k (ext2fs selects the default disk block size based on
> the size of the partition: if it is >512MB, the disk block size is set
> to 4k).  Thus, this code is rarely exercised.

Indeed, the partition on my system has 1k blocks.

> Can you try this out and tell me if you get the error any more.

I applied the patch to the hurd_20020103-2 Debian source package.  (I
prefer to install from Debian packages.)  I ran

 joostvb@stegun-hurd /usr/local/src/hurd/hurd-20020103$ ./debian/rules binary

After entering hurd-20020103/build/libports, the build failed:

 mig -cc cat - /dev/null -subrprefix __ -prefix ports_  \
    -sheader interrupt_S.h -server interruptServer.c \
    -user /dev/null -header /dev/null < interrupt.sdefsi
 cat: invalid option -- d
 Try `cat --help' for more information.
 mig: fatal: no SubSystem declaration
 make[2]: *** [interruptServer.c] Error 1





in /bin/mig, as I did before (see
works around this (what is the "right" way to do this?  Am I missing
something here?), the build succeeded.  (Well, it kept me busy for about
5 days: the machine is low on cpu, low on memory and low on diskspace.
And it occasionally crashes.  O well.)

I did

 root@stegun-hurd /usr/local/src/hurd/hurd-20020103# dpkg -i 
 root@stegun-hurd /usr/local/src/hurd/hurd-20020103# reboot

The system booted, sshd started, console login prompt appeared.  However,

 login> login root

just hangs.  One _can_ run ps as the not-logged-in user, which gives pretty
convincing output.  catting files works fine too, as does writing files
in /tmp/ .

On Sat, Apr 06, 2002 at 04:47:38PM -0500, Neal H Walfield wrote:
> You need the second patch that I posted on bug-hurd.  The error that I
> fixed exposed another.

OK, I guess that's
http://mail.gnu.org/pipermail/bug-hurd/2002-April/007388.html .  I saw
it in CVS too.

Problem is: I no longer have a hurd build system now :(  (And I've heard
people stating crosscompiling the hurd is a major pita.)  It would be
cool if anybody could build a snapshot hurd Debian binary package for
me.  I guess I could install it from within the GNU/Linux system running
on the box, using dpkg's --root=/gnu flag, or, even better, running

 dpkg-deb --extract hurd_foo.deb /gnu/

to prevent the postinst script to fail.  I only want to reinstall the
hurd if _really_ _everything_ else fails.

> Thanks.

Thank *you* !



                               . .                  http://mdcc.cx/
Joost van Baal                .   .
                              .   .
                               . .            http://logreport.org/

Attachment: pgp2dAzJgInup.pgp
Description: PGP signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]