lkml.org 
[lkml]   [1996]   [Sep]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: SCSI CD-ROM driver bug in 2.0.14?
  From: "Ulrich Windl" <ulrich.windl@rz.uni-regensburg.de>
Date: Fri, 6 Sep 1996 08:25:09 +0200

I donŽt know whether it was a bad mastered CD-R, or a bug in the
SCSI CD-ROM driver, but I got a media error followed by the attempt to
read past the end of the device. This was kernel 2.0.14:

Sep 5 21:05:08 elf kernel: Max size:313558 Log zone size:2048
Sep 5 21:05:08 elf kernel: First datazone:28 Root inode number 57344
Sep 5 21:05:08 elf kernel: ISO9660 Extensions: RRIP_1991A
Sep 5 21:18:49 elf kernel: scsi0: MEDIUM ERROR on channel 0, id 2, lun 0, CDB: Read (6) 04 c8 d5 02 00
Sep 5 21:18:49 elf kernel: Current error sr0b:00: sense key Medium Error
Sep 5 21:18:49 elf kernel: Additional sense indicates Circ unrecovered error
Sep 5 21:18:49 elf kernel: CD-ROM I/O error: dev 0b:00, sector 1254232
Sep 5 21:18:49 elf kernel: attempt to access beyond end of device
Sep 5 21:18:49 elf kernel: 0b:00: rw=0, want=627117, limit=627116
Sep 5 21:18:49 elf kernel: attempt to access beyond end of device
Sep 5 21:18:49 elf kernel: 0b:00: rw=0, want=627118, limit=627116
Sep 5 21:18:49 elf kernel: attempt to access beyond end of device
Sep 5 21:18:49 elf kernel: 0b:00: rw=0, want=627117, limit=627116
Sep 5 21:18:49 elf kernel: attempt to access beyond end of device
Sep 5 21:18:49 elf kernel: 0b:00: rw=0, want=627118, limit=627116

Did the error get reflected to user space or not? If not, then you've hit the
result of some strangeness in the SCSI specification. Specifically, the SCSI
specification allows for the value returned by READ CAPACITY to be up to 75 2K
sectors past the last readable block. So we cannot know at the time we make
the request whether it is in fact past the end or not. The present strategy is
to see if we hit an error, and then handle it without returning an error to the
user program. It does still result in these kernel messages, however. This
also interacts with read-ahead; we could stop all read-ahead when near the
expected end of the CD-ROM, but that could yield bad performance.

BTW: ShouldnŽt there be a blank between the "sr" and the "0b:00" in the
error message?

For better or worse, that's the format used by all these SCSI "print_sense"
messages. It could certainly be improved.

Leonard

\
 
 \ /
  Last update: 2005-03-22 13:38    [W:0.033 / U:0.740 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site