Re: [Cooker] Reproducible Crash

1999-11-10 Thread Axalon Bloodstone



On Wed, 10 Nov 1999, Eugenio Diaz wrote:

> I may have encountered a serious bug.
> 
> My box has crashed two times now (in a period of less than an hour).
> 
> In both occations, I was copying files from the floppy while heavily
> using X/Enlightenment by changing desktops or windows quickly using the
> keyboard.

E used to have a pretty severe bug when changeing desktops, but i haven't
been able to reproduce it for awhile now. What Xserver, does the kernel
freeze (sysrq keys work?). 

Try to reproduce it without the floppy IO enable edge-flipping and set
resistance to 0 now put the mouse in a corner (bottom right with default
4 workspaces) and make circles with the mouse so it flips between all the
desktops.
 
> I don't know if this is related, but normally while I use "grip" (which
> uses cdparanoia) to rip CD's, the ppp network response becomes extremely
> slow. I was not using "grip" while the crashes happened, but mentioned
> it anyway just in case it is related. I think this one is related to
> some bad interrupt handling.

Turn the error correction down,
 
> I have no idea how to trace this back, plus I dont't want to reproduce
> it since I have software striping on two 10G HDs and every time it crash
> it takes like half hour to fsck, plus i don't want to loose all those
> mp3's and jps's (Damn! I need to put that CDR to work) ;-). I have a
> Gateway G6-200 (PPro 200) running 2.2.13-28mdk.

 because this isn't all that fast now adays, and cdparanoia with full
error correction will eat lots.. 

> The only rare thing I could find in the various logs is this entry from
> the kernel ring buffer (dmesg) which appears repeated a lot (but don't
> know how to associate it with a time to see if it happened just before
> the crash):

The first field is the month, second date, and third time
like so,
Nov 10 12:29:14 

> Directory sread (sector 0x15) failed
> attempt to access beyond end of device
> 02:00: rw=0, want=11, limit=4
> dev 02:00 blksize=512 blocknr=21 sector=21 size=512 count=1
> 
> I almost sure "dev 02:00" is the floppy, since /proc/devices says:
> 
> Block devices:
>   1 ramdisk
>   2 fd
>   3 ide0
>   9 md
>  22 ide1
> 
> Anyone can help me track this down? Any ideas on were to look for more
> clues?

Yep, it's the floppy. Try dd'ing the floppy to harddisc and mounting it as
a loop device. dd will error if there are funny sectors

> --
> Eugenio Diaz, BSEE/BSCE
> Linux Engineer
> [EMAIL PROTECTED]
> 
> 
> 



[Cooker] Reproducible Crash

1999-11-09 Thread Eugenio Diaz

I may have encountered a serious bug.

My box has crashed two times now (in a period of less than an hour).

In both occations, I was copying files from the floppy while heavily
using X/Enlightenment by changing desktops or windows quickly using the
keyboard.

I don't know if this is related, but normally while I use "grip" (which
uses cdparanoia) to rip CD's, the ppp network response becomes extremely
slow. I was not using "grip" while the crashes happened, but mentioned
it anyway just in case it is related. I think this one is related to
some bad interrupt handling.

I have no idea how to trace this back, plus I dont't want to reproduce
it since I have software striping on two 10G HDs and every time it crash
it takes like half hour to fsck, plus i don't want to loose all those
mp3's and jps's (Damn! I need to put that CDR to work) ;-). I have a
Gateway G6-200 (PPro 200) running 2.2.13-28mdk.

The only rare thing I could find in the various logs is this entry from
the kernel ring buffer (dmesg) which appears repeated a lot (but don't
know how to associate it with a time to see if it happened just before
the crash):

Directory sread (sector 0x15) failed
attempt to access beyond end of device
02:00: rw=0, want=11, limit=4
dev 02:00 blksize=512 blocknr=21 sector=21 size=512 count=1

I almost sure "dev 02:00" is the floppy, since /proc/devices says:

Block devices:
  1 ramdisk
  2 fd
  3 ide0
  9 md
 22 ide1

Anyone can help me track this down? Any ideas on were to look for more
clues?

--
Eugenio Diaz, BSEE/BSCE
Linux Engineer
[EMAIL PROTECTED]