lkml.org 
[lkml]   [1998]   [Apr]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectCorrupt file-system(s) leads to crash

To whomever manages the SCSI system:

I am making another Linux File-Server. It uses all new stuff including
a TYAN S1692D(L) Tiger a ATX board with two high-speed Pentiums, etc.

While trying to get the screen card running under X, after spending
several days copying zillions of files from the server it will replace,
the usual happened. The machine hung. I had to hit the reset switch.
Actually, since it wasn't connected, I had to cycle the power.

The root file-system wasn't damaged so it `fsck`ed fine. The next
4.3 Gb file-system also wasn't damaged. The third, however, was
damaged. This caused seeks beyond the end of the media, lots of
SCSI bus resets, then eventually a crash which left all disk drives
destroyed. Yes, the READ CAPACITY command wouldn't even work from
the Adaptec BIOS menu. I had to low-level format all three drives to
get them back on-line.

Now... Please put a check within scsi.c for any attempt to access
beyond the capacity of the media and return an error without actually
trying to see if you can get away with it!. The read-capacity command
occurs early in the startup. These values should be saved and used during
SCSI read/write access.

The last time I asked for this, several years ago, and even
submitted a patch, I was informed that this would "take 20zillion
machine-cycles per SCSI access, and slow the machine to a crawl -- reboot
with DOS..."

The problem with attempting to access beyond the media is that the
Disk drives are dumb and will try to do just that. Something, somewhere,
must be smart enough to prevent this from happening. Presently, if
a file-system is slightly corrupt, one bad offset, you can destroy
every file-system on the controller. Especially if it's an Adaptec
running the aic7xxx driver.

Cheers,
Dick Johnson
***** FILE SYSTEM MODIFIED *****
Penguin : Linux version 2.1.92 on an i586 machine (66.15 BogoMips).
Warning : It's hard to remain at the trailing edge of technology.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:42    [W:0.024 / U:0.120 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site