rfs4, 2.5.53: jnode_flush failed with err = -12

rfs4, 2.5.53: jnode_flush failed with err = -12

[Ookhoi]

Dear Developers and list readers,

First of all, a happy new year to all of you.

With kernel 2.5.53 and reiserfs4 snapshot of 24 Dec 2002 I get the
following error when I try to untar a file on it.

jnode_flush failed with err = -12

16 times.

then:

 reiser4[pdflush(5)]: jnode_flush (fs/reiser4/flush.c:1050)[jmacd-16739]:
WARNING: flush failed: -12

and 
jnode_flush failed with err = -12

two more times.


I make a bootfloppy with the 2.5.53 kernel and rfs4 compiled in (get an
insmod "Unknow symbol" error if compiled as module (balance_dirty_pages
IIRC).

Boot the system, up the network, wget scsi modules, libs and binaries,
insmod the scsi modules, fdisk the disk (one partition), mkfs it without
options, mount it, wget base2_2.tar from debian potato onto it (44M),
untar it and get the above messages.

The untar seems stuck and I can't get a prompt back (sash). sysrq works,
so I could get some info out of it if it is useful.

Please let me know if I should provide more info.

Re: rfs4, 2.5.53: jnode_flush failed with err = -12

[Oleg Drokin]

Hello!

On Mon, Jan 06, 2003 at 12:22:17PM +0100, Ookhoi wrote:

> With kernel 2.5.53 and reiserfs4 snapshot of 24 Dec 2002 I get the
> following error when I try to untar a file on it.
> jnode_flush failed with err = -12

Errno 12 means "Out of Memory".
How much RAM do you have on your test box?
Currently reiser4 is somewhat memory-hungry and the logic that deals
with sort-memory situations is not implemented nicely yet.

Thank you for trying out reiser4 (and for reporting problems of course).

Bye,
    Oleg

Re: rfs4, 2.5.53: jnode_flush failed with err = -12

[Ookhoi]

Oleg Drokin wrote (ao):
> On Mon, Jan 06, 2003 at 12:22:17PM +0100, Ookhoi wrote:
> 
> > With kernel 2.5.53 and reiserfs4 snapshot of 24 Dec 2002 I get the
> > following error when I try to untar a file on it.
> > jnode_flush failed with err = -12
> 
> Errno 12 means "Out of Memory".
> How much RAM do you have on your test box?

512MB.
I don't have swap space. Could that be a problem?
The ramdisk of the bootfloppy is only 10MB unzipped.

sysrq-m says:
Mem-info:
DMA per-cpu:
cpu 0 hot: low 2, high 6, batch 1
cpu 0 cold: low 0, high 2, batch 1
Normal per-cpu:
cpu 0 hot: low 32, high 96, batch 16
cpu 0 cold: low 0, high 32, batch 16
HighMem per-cpu: empty

Free pages:      442720kB (0kB HighMem)
Active:3155 inactive:13374 dirty:12963 writeback:0 free:110680
DMA free:14352kB min:128kB low:256kB high:384kB active:0kB inactive:0kB
Normal free:428368kB min:1020kB low:2040kB high:3060kB active:12620kB inactive:53496kB
HighMem free:0kB min:0kB low:0kB high:0kB active:0kB inactive:0kB
DMA: 2*4kB 1*8kB 2*16kB 3*32kB 2*64kB 4*128kB 3*256kB 1*512kB 2*1024kB 1*2048kB 2*4096kB = 14352kB
HighMem: empty
Swap cache: add 0, delete 0, find 0/0, race 0+0
Free swap:            0kB
131072 pages of RAM
0 pages of HIGHMEM
1794 reserved pages
14081 pages shared
0 pages swap cached


> Currently reiser4 is somewhat memory-hungry and the logic that deals
> with sort-memory situations is not implemented nicely yet.
> 
> Thank you for trying out reiser4 (and for reporting problems of course).

Very welcome.

Re: rfs4, 2.5.53: jnode_flush failed with err = -12

[Oleg Drokin]

Hello!

On Mon, Jan 06, 2003 at 12:55:01PM +0100, Ookhoi wrote:
> > > With kernel 2.5.53 and reiserfs4 snapshot of 24 Dec 2002 I get the
> > > following error when I try to untar a file on it.
> > > jnode_flush failed with err = -12
> > Errno 12 means "Out of Memory".
> > How much RAM do you have on your test box?
> 512MB.

Hm. Very strange.

> I don't have swap space. Could that be a problem?

Not likely. We were able (in fact we do that quite often) to untar compiled
kernel source that is bigger than your tar file both compressed and uncompressed
on 240M system.
I guess we need to release another snapshot with current code and see if it will
break for you again.
I am now merging our kernel bits to see how 2.5.54 will live with reiser4.

Bye,
    Oleg

Re: rfs4, 2.5.53: jnode_flush failed with err = -12

[Ookhoi]

Oleg Drokin wrote (ao):
> On Mon, Jan 06, 2003 at 12:55:01PM +0100, Ookhoi wrote:
> > I don't have swap space. Could that be a problem?
> 
> Not likely. We were able (in fact we do that quite often) to untar
> compiled kernel source that is bigger than your tar file both
> compressed and uncompressed on 240M system. I guess we need to release
> another snapshot with current code and see if it will break for you
> again.
> I am now merging our kernel bits to see how 2.5.54 will live with
> reiser4.

Would love to try. I pulled reiser4 and 2.5 from bk (and plain 2.5.54).
It gives me this error:

ookhoi $ make > /dev/null
fs/reiser4/flush.c:772: macro `write_syscall_trace' used with just one arg
fs/reiser4/flush.c:1087: macro `write_syscall_trace' used with just one arg
make[2]: *** [fs/reiser4/flush.o] Error 1
make[1]: *** [fs/reiser4] Error 2
make: *** [fs] Error 2

Unfortunately source code is almost like Russian (or Chinese) to me, so
I guess I'll just wait for a new snapshot :-)