File corruption
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 8 lines · Ralf Baechle
Quoted 5 lines · Ulf Carlsson
This is *really* annoying, I can't do anything without getting interrupted by some bug on the Indy. I can't stand this file corruption. I just lost an entire directory. This isn't because of crashes, stuff just vanish, or gets corrupted without any reason. I for example had a gzipped patch, I decompressed it and I got a lot of junk in the middle of the file.
Quoted 5 lines
Question, is this effect repeatable, that is if you decompress the file a second time, will you still see the filesystem corruption? The latter would mean that the data decompressed data get corrupted on the fly and the SCSI driver isn't involved. Or does the gzip'ed file itself already contain the corrupted data?
Since the file is decompressed without warnings I assume that the scp went fine. Anyhow, when I had decompressed the file I got corruption in the middle of the patch. The corruption affected only a couple of lines, 10 or so.
The problems I experience are not repeatable. Everything went smooth after a reboot. I usually reboot the machine when it's messing with me, that helps. At least for a while.
Quoted 1 line
And under which kernel version did this start to happen?
2.2.1 I think.
Quoted 1 line
Could you resend me your hinv output?
Unfortunately I don't have IRIX. However, I have a 133 MHz R4600 CPU with 512 k board cache. I have two 1 Gb SCSI driver connected.
Quoted 6 lines
Quoted 3 lines
Earlier today the cached version of
/lib/libpam_misc.so.0.64was corrupted or something, and corrupted ELF headers were reported. A reboot fixed that one.It seems that you observe similar problems like me. For me however the problem has just vanished and was rare anyway.
They're rare for me as well, but I've had a couple in just a few days now. Maybe the file corruption became heavier since 2.3.6?
Quoted 1 line
I understand that very well. Btw, I started to store important stuff on NFS.
I can't stand NFS. Files, or entire directories vanish from ls, but I may still open them if I specify their names directly. I think I've been talking to you about this before.
- Ulf
Quoted 3 lines · Ulf Carlsson
Quoted 1 line
And under which kernel version did this start to happen?
2.2.1 I think.
Are you shure? The problem Alan is tracking started to hit from 2.2.7 on. If 2.2.1 already starts making these kind of troubles then we probably track two different problems.
Quoted 2 lines
Unfortunately I don't have IRIX. However, I have a 133 MHz R4600 CPU with 512 k board cache. I have two 1 Gb SCSI driver connected.
Is this a low-mem configuration? The problem Alan is tracking apparently seems to hit low mem systems more often.
Ralf
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
And under which kernel version did this start to happen?
2.2.1 I think.
Are you shure? The problem Alan is tracking started to hit from 2.2.7 on. If 2.2.1 already starts making these kind of troubles then we probably track two different problems.
Probably. I have a new problem with the 2.3.8 kernel now as well: Page->owner becomes -1 at line 1218 in buffer.c, thus forcing an oops. This happens when I try to cat a couple of big files into a new file. This is even reproduceable...
I get some error messages before the oops as well:
attempt to access beyond end of device
08:01: rw=0, want=122156967, limit=929559
attempt to access beyond end of device
08:01: rw=0, want=886680107, limit=929559
Quoted 5 lines
Quoted 2 lines
Unfortunately I don't have IRIX. However, I have a 133 MHz R4600 CPU with 512 k board cache. I have two 1 Gb SCSI driver connected.
Is this a low-mem configuration? The problem Alan is tracking apparently seems to hit low mem systems more often.
Not really, I have 128 Mb RAM, but no swap.
Regards,
Ulf
Thomas Bogendoerfer
maintainerQuoted 1 line · Ulf Carlsson
Probably. I have a new problem with the 2.3.8 kernel now as well:
Page->owner
[...]
Quoted 6 lines
I get some error messages before the oops as well:
attempt to access beyond end of device 08:01: rw=0, want=122156967, limit=929559 attempt to access beyond end of device 08:01: rw=0, want=886680107, limit=929559
this seems to be a normal 2.3 error. At least I get this impression, when reading the linux-kernel mailinglist.
Thomas.
Signature · 3 lines
This device has completely bogus header. Compaq scores again :-|
It's a host bridge, but it should be called ghost bridge instead ;^)
[Martin `MJ' Mares on linux-kernel]I have been running 2.2.1 on my indy for about 3 weeks now and have encountered problems like this,
in fact i am surprised at how stable the box is, i have been doing some quite large builds of qt, kde and the like too. If this is reproducible then i would be prepared to experiment on one of my boxes.
The only gripe i have is that the scsi driver breaks when i attach external devices, any ideas?
Andy
----- Original Message -----
From: Ralf Baechle <[email protected]>
To: <[email protected]>
Sent: Thursday, June 24, 1999 11:28 PM
Subject: Re: File corruption
Quoted 11 lines · Ulf Carlsson
Quoted 4 lines
Quoted 2 lines
And under which kernel version did this start to happen?
2.2.1 I think.
Are you shure? The problem Alan is tracking started to hit from 2.2.7 on. If 2.2.1 already starts making these kind of troubles then we probably track two different problems.
Quoted 1 line
Unfortunately I don't have IRIX. However, I have a 133 MHz R4600 CPU
with 512 k
Quoted 7 lines
Quoted 1 line
board cache. I have two 1 Gb SCSI driver connected.
Is this a low-mem configuration? The problem Alan is tracking apparently seems to hit low mem systems more often.
Ralf
Quoted 5 lines · Andrew Linfoot
I have been running 2.2.1 on my indy for about 3 weeks now and have encountered problems like this, in fact i am surprised at how stable the box is, i have been doing some quite large builds of qt, kde and the like too. If this is reproducible then i would be prepared to experiment on one of my boxes.
It doesn't crash, only file corruption..
Quoted 2 lines
The only gripe i have is that the scsi driver breaks when i attach external devices, any ideas?
I have file corruption without any external devices. It looks like the problems appear when more than one SCSI devices are present. You only have one internal SCSI drive, right?
Do you mind telling us more exactly what type of file corruption you suffer from? It would also help if you could attach an output from hinv, or just tell us what hardware you have.
Regards,
Ulf
On a couple of occasions i have lost entire directories but this was some time ago
----- Original Message -----
From: Ulf Carlsson <[email protected]>
To: Andrew Linfoot <[email protected]>
Cc: <[email protected]>
Sent: Friday, June 25, 1999 5:59 PM
Subject: Re: File corruption
Quoted 3 lines · Andrew Linfoot
Quoted 2 lines
I have been running 2.2.1 on my indy for about 3 weeks now and have encountered problems like this, in fact i am surprised at how stable the
box
Quoted 1 line
Quoted 1 line
is, i have been doing some quite large builds of qt, kde and the like
too. If
Quoted 2 lines
Quoted 2 lines
this is reproducible then i would be prepared to experiment on one of my boxes.
I have had a couple of instances where i lost complete directories, but this was during the first few days of running Linux on my box, i simply put it down to operator error!
Quoted 4 lines
It doesn't crash, only file corruption..
Quoted 1 line
The only gripe i have is that the scsi driver breaks when i attach
external
Quoted 3 lines
Quoted 1 line
devices, any ideas?
I have file corruption without any external devices. It looks like the
problems
Quoted 1 line
appear when more than one SCSI devices are present. You only have one
internal
Quoted 1 line
SCSI drive, right?
2 drives - see below
If i attempt to attach an external device the system just hangs during scsi initialisation. I think this is fixed in later kernels but i don't have cvs up yet so i can't move past 2.2.1.
Quoted 2 lines
Do you mind telling us more exactly what type of file corruption you
suffer
Quoted 1 line
from? It would also help if you could attach an output from hinv, or just
tell
Quoted 1 line
us what hardware you have.
CPU: MIPS-R4400 FPU<MIPS-R4400FPC> ICACHE DCACHE SCACHE Loading R4000 MMU routines.
CPU revision is: 00000450
Primary instruction cache 16kb, linesize 16 bytes) Primary data cache 16kb, linesize 16 bytes) Secondary cache sized at 1024K linesize 128 8-bit graphics
128MB RAM
sda 540MB with base IRIX
sdb 2GB with Linux.and swap
an interesting point here is that /proc/cpuinfo tells me i only have an r4000!
Andy
Quoted 4 lines
Regards,
Ulf
Quoted 1 line · Andrew Linfoot
CPU: MIPS-R4400 FPU<MIPS-R4400FPC> ICACHE DCACHE SCACHE
This is what the firmware detected.
Quoted 1 line
Loading R4000 MMU routines.
This will always say R4000 if it's a R4000 class CPU.
Quoted 4 lines
CPU revision is: 00000450
Primary instruction cache 16kb, linesize 16 bytes) Primary data cache 16kb, linesize 16 bytes) Secondary cache sized at 1024K linesize 128
Which says that you have an R4400SC V5.0.
Quoted 2 lines
an interesting point here is that
/proc/cpuinfotells me i only have an r4000!
Good spotting, the CPU detection code has a bug causes only V4 CPUs to be detected as R4400. The difference is just cosmetics; R4000 and R4400 are almost the same. Especially it's unrelated to the file corruption problem.
Ralf
Quoted 8 lines · Andrew Linfoot
I have been running 2.2.1 on my indy for about 3 weeks now and have encountered problems like this,
in fact i am surprised at how stable the box is, i have been doing some quite large builds of qt, kde and the like too. If this is reproducible then i would be prepared to experiment on one of my boxes.The only gripe i have is that the scsi driver breaks when i attach external devices, any ideas?
There is nothings special with external devices, so if they result in crashes you should double check that cabeling and termination are ok.
Ralf