Bug 967 - Onstream DI-30 fails to respond
Summary: Onstream DI-30 fails to respond
Status: CLOSED CODE_FIX
Alias: None
Product: IO/Storage
Classification: Unclassified
Component: IDE (show other bugs)
Hardware: i386 Linux
: P2 normal
Assignee: Bartlomiej Zolnierkiewicz
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2003-07-20 02:51 UTC by Stef van der Made
Modified: 2003-11-08 13:01 UTC (History)
2 users (show)

See Also:
Kernel Version: 2.6.0-test1-2.6.0.test9
Subsystem:
Regression: ---
Bisected commit-id:


Attachments
dmesg output (9.59 KB, text/plain)
2003-07-20 02:52 UTC, Stef van der Made
Details
Dmesg for shuttle ak32e (6.68 KB, text/plain)
2003-09-25 12:50 UTC, Stef van der Made
Details
screen shot for strace mt -f /dev/ht0 status (183.45 KB, image/pdf)
2003-10-17 12:45 UTC, Stef van der Made
Details
Proposed fix (16.56 KB, patch)
2003-11-01 14:25 UTC, Stef van der Made
Details | Diff
Strace mt -f /dev/hto status (1.72 KB, text/plain)
2003-11-01 15:03 UTC, Stef van der Made
Details
dmesg for shuttle ak32e (7.56 KB, text/plain)
2003-11-01 15:08 UTC, Stef van der Made
Details
Strace mt -f /dev/ht0 status (2.45 KB, text/plain)
2003-11-02 03:54 UTC, Stef van der Made
Details
proposed fix (17.46 KB, patch)
2003-11-02 03:55 UTC, Stef van der Made
Details | Diff
dmesg for shuttle ak32e (7.65 KB, text/plain)
2003-11-02 04:02 UTC, Stef van der Made
Details
second part of the proposed patch (523 bytes, patch)
2003-11-02 12:34 UTC, Stef van der Made
Details | Diff

Description Stef van der Made 2003-07-20 02:51:36 UTC
Distribution: Slackware 8.1
Hardware Environment: MSI k7tmaster
Software Environment: glibc 2.3.1, gcc3.3
Problem Description: Onstream DI-30 is not seen as a device (ht0) as it did in
2.4 . I've also tried to use hdd which is the ide name of the drive to see if it
would repsond like that. 

Steps to reproduce:

bash-2.05#  tar -xvb 64 -f /dev/ht0 /usr/download/restore/
tar: Cannot open /dev/ht0: No such device
tar: Error is not recoverable: exiting now
bash-2.05#
Comment 1 Stef van der Made 2003-07-20 02:52:49 UTC
Created attachment 582 [details]
dmesg output

This is the output created by dmesg
Comment 2 Bartlomiej Zolnierkiewicz 2003-08-24 12:56:45 UTC
Please try 2.6.0-test4 (compile in or as module ide-tape support).
IDE tapes should work now (or not)...
Comment 3 Stef van der Made 2003-09-01 12:35:42 UTC
I'm not in the opertunity to test this as my mainboard has crapped out. As soon 
as my pc is back alive and kicking I will do some testing on this fix. Sorry 
for the delay.
Comment 4 Stef van der Made 2003-09-22 14:37:19 UTC
I'll try this next weekend. My pc is back alive again I only I'm not in the
opertunity earlier to try this.
Comment 5 Stef van der Made 2003-09-25 12:50:42 UTC
Created attachment 934 [details]
Dmesg for shuttle ak32e

When using the mt status command on ht0 the pc freezes. The same goes for tar
-xf /dev/ht0 /usr/download. I've included a dmesg from the boot sequence.
Please advise how I can provide usefull (debug) information
Comment 6 Stef van der Made 2003-10-11 10:50:11 UTC
Changed kernel version to 2.6.0test1 to 2.6.0-test7
Comment 7 Bartlomiej Zolnierkiewicz 2003-10-15 13:43:55 UTC
Please run mt status command using strace.
It will give info about what ioctls are being called.
Thanks.
Comment 8 Stef van der Made 2003-10-16 12:27:55 UTC
running strace and mt -f /dev/ht0 status freezes the machine imediatly. I do see
some screen output. Is it usefull if I copy this info into the bug report ? Ans
how many of the lines do you need the last 5 or everything. I haven't tried any
of the mgic sys req keys. Might those help to get out of the freeze ?
Comment 9 Stef van der Made 2003-10-17 12:45:08 UTC
Created attachment 1070 [details]
screen shot for strace mt -f /dev/ht0 status

This is the output I got from strace when running strace mt -f /dev/ht0 status.
The system request key was not usable after the crash the system was totaly
locked up only the reset key still did the trick to get response.
Comment 10 Stef van der Made 2003-10-25 14:38:56 UTC
Updated kernel version
Comment 11 Stef van der Made 2003-10-25 15:57:17 UTC
When testing with test9 I can see this text 'part' flying over the screen

'command in request queue' and 'media_acccess_finished' keep flashing over the
screen. Again hard reboot is the only way out
Comment 12 Stef van der Made 2003-11-01 14:25:03 UTC
Created attachment 1311 [details]
Proposed fix

Suggested patch by  B.Zolnierkiewicz@elka.pw.edu.pl
Comment 13 Stef van der Made 2003-11-01 15:03:54 UTC
Created attachment 1312 [details]
Strace mt -f /dev/hto status

The system no longer locks up with the patch. I'm hoping that the new strace
tells in more detail what is happening. When I tried the mt -f /dev/ht0 status
command the drive had a tape inserted. The comand hower came back with the
messge that there was no tape inserted.
Comment 14 Stef van der Made 2003-11-01 15:08:09 UTC
Created attachment 1313 [details]
dmesg for shuttle ak32e

New dmesg. I'm 100% sure that the drive has the 1.09 firmware version. It looks
as if the drive is no longer properly detected
Comment 15 Stef van der Made 2003-11-02 03:54:14 UTC
Created attachment 1320 [details]
Strace mt -f /dev/ht0 status

Drive does go to work but gets back with this strace and is after using the
command in use or busy. e.g. any new command to the drive fails.
Comment 16 Stef van der Made 2003-11-02 03:55:26 UTC
Created attachment 1321 [details]
proposed fix
Comment 17 Stef van der Made 2003-11-02 04:02:22 UTC
Created attachment 1322 [details]
dmesg for shuttle ak32e
Comment 18 Stef van der Made 2003-11-02 11:11:32 UTC
Oops on mt -f /dev/ht0 status

root@made0120:/home/stef#  mt -f /dev/ht0 status
Unable to handle kernel NULL pointer dereference at virtual address 00000020
*pde = 00000000
Oops: 0000 [#1]
CPU:    0
EIP:    0060:[<c0233302>]    Not tainted
EFLAGS: 00010292
EIP is at idetape_chrdev_ioctl+0x12/0x370
eax: dc3062c0   ebx: 801c6d02   ecx: bffff7fc   edx: c037fcc0
esi: 00000000   edi: 801c6d02   ebp: dba97f90   esp: dba97f38
ds: 007b   es: 007b   ss: 0068
Process mt (pid: 129, threadinfo=dba96000 task=dc3d00c0)
Stack: dbf29438 dc3062c0 dbea2140 00000800 0804b4b8 dfae2000 dba97f9c c014e9fb
       dbea2140 dffe4780 00000800 dba97f68 dbea2140 dffe4780 00000603 fffffff4
       00000003 00000101 00000001 801c6d02 dc3062c0 ffffffe7 dba97fbc c0161544
Call Trace:
 [<c014e9fb>] filp_open+0x5b/0x60
 [<c0161544>] sys_ioctl+0xf4/0x2b0
 [<c01093ef>] syscall_call+0x7/0xb
 
Code: 8b 5e 20 c7 45 c8 00 00 00 00 8b 83 0c 18 00 00 89 45 c4 83
 Segmentation fault


Comment 19 Stef van der Made 2003-11-02 12:34:47 UTC
Created attachment 1323 [details]
second part of the proposed patch
Comment 20 Stef van der Made 2003-11-02 12:35:36 UTC
Problems solved with the 2 patches attached. mt and tar now can communicate with
the drive without any troubles
Comment 21 Stef van der Made 2003-11-08 13:01:09 UTC
Patch included in 2.6.0-test9-bk13

Note You need to log in before you can comment on or make changes to this bug.