Messages in this thread |  | | Subject | Re: AHA2940UW vs. MSDOS/VFAT file systems | Date | Tue, 17 Sep 1996 13:45:22 +0200 | From | Herbert Wengatz <> |
| |
+> On Mon, 16 Sep 1996 Randy Gobbel wrote +> +> >I've had similar problems with MSDOS and VFAT file systems in previous 2,0 +> >kernels, but a) things were just weirdly slow, the system eventually +> >recovered, and b) I didn't know where to send kernel bug reports. Now I'm +> >getting a hard failure, with kernel 2.0.19 or 2.0.20, with or without +> >VFAT/MSDOS as modules: +> +> We have found similar problems, in fact with the same error output, when +> performing gzip -t file_of_500Mb, the HD was a 4Gb Quantum Atlas. +> With Seagate Barracuda ST15150W the system does not hang so easily but +> still fails under a more demanding program. The problems seem to disappear +> when the clock speed is set to 6.8Mhz but not higher! +> +> We have tested several kernels from 1.3.98 to 2.0.20, although the error +> output changes, the problems persist.
Well, then it's perhaps a hardware problem ? - Especially, because you wrote, that the problem gets better when you lower the speed on the SCSI-Bus...
First: Try making the cable shorter and the distances between the devices longer! And use active bus-termination!
And, judging from the Error-Messages the kernel gives you (I just looked up the posting 'bout this...), you really _have_ troubles with your cabling, not with Linux !
READ IT LOUD : (1 try from your SCSI-Controler...) SCSI host 0 abort (pid 22377) timed out - resetting ^^^^^^^^^ SCSI bus is being reset for host 0 channel 0. ^^^ ^^^^^ ^^^ ^^^^^^^^^^^^^^^^^ and (2nd try): SCSI host 0 channel 0 reset (pid 22377) timed out - trying harder SCSI bus is being reset for host 0 channel 0
and now: SCSI host 0 reset (pid 22377) timed out again - probably an unrecoverable SCSI bus or device hang. ^^^^^^^^^^^^^ ^^^ ^^^^
So what does the last message tells you ? You *HAVE* Problems on the SCSI-Bus. - Bad/old/too long/wrong terminated cable!
The Linux-driver for the Adaptec Controller already tells you, what he tries... And he tries hard and harder. - Three times... OK. Adaptec Controllers a "Primadonnas", but this is really a bad cable! +> +> Antonio Munoz
Regards,
Herbert
_____________________________________________________________________ Herbert Wengatz, 81375 Munich |Disclaim: This Mail is my own opinion, Office :hwe@uebemc.siemens.de |not that of my company. Private:hwe@rtfact.muc.de | http://www.muc.de/~hwe/rtfact
|  |