On 14-Dec-98 Kai Makisara wrote:
> On Sun, 13 Dec 1998, Peter Waltenberg wrote:
> ...
>> Attached this time is the Unixware generated tar file that when untarred from tape
>here seems to
>> lock up both 2.0.32/aic7xxx
>> and 2.1.131/aha1542
>>
>> The problem with tar mishandling these archives should have been fixed in
>"paxutils", however
>> tar shouldn't be taking out the scsi system when it has problems.
>>
> I have tried this with two 2.1.131ac10 systems but failed to have any
> problems. One of the systems had ncr53c875 to drive HP DAT and three
> disks, the other one had aha1542b to drive a Wangtek 5525ES and aha2940 to
> drive one disk. I tried both variable and fixed block modes. I looked at
> the commands the tape driver sends to the middle level SCSI driver when
> untarring but there was nothing suspicious. GNU tar failed to read the
> last 10 kB block but that was a tar problem.
>
> Kai
>
Thanks, I recompiled the aha1542 with debugging when it happened to me, the only thing
I saw was the last command being started (and no interrupt coming back) and an
eventual bus reset which left scsi in an unusable state.
I'd have guessed "random interrupt loss", but it's 100% repeatable here
(I lost the tape drive before I could test with that archive, but with a full backup
containing that
path, the tar xvf always locked up in the same place, and on two different Linux
systems.)
Solaris reads the same tape just fine.
Interestingly, using the newer tar (from paxutils) we don't have problems either.
I had hoped I could generate a simple test case, but it looks like I'll have to wait
until I can use tape drive here again.
Thanks
Peter
---------------------------------
E-Mail: Peter Waltenberg <[EMAIL PROTECTED]>
Date: 15-Dec-98
Time: 07:18:48
This message was sent by XFMail
----------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
