PROBLEM: cdrecord, ide-scsi and all 2.4.x kernels

From: Ricardo Ferreira (stormlabs@gmx.net)
Date: Fri Nov 02 2001 - 22:31:34 EST

If this message appears twice, its caused by the crappy smtp server of my ISP. Sorry.

I CCed this mail to the people in the MAINTAINERS file that i thought might have more knowledge of what
might be causing this. Gadi Oxman i fetched from ide-scsi.c

Hardware:
-----------
Abit VP6
2x PIII-1Ghz
1GB SDRAM

IBM 40GB UDMA100 7200rpm connected to 1st channel of onboard HPT370
Yamaha 8824 CDRW connected to 2nd channel of onboard HPT370
Seagate 10GB UDMA66 connected to 1st channel of onboard VIA controller
ASUS 50x CDROM connected to 2nd channel of onboard VIA controller
Empty Promise ULTRA66 card

Description:
-------------
Everytime i try to burn a CD i get these messages from the kernel:

ide-scsi: CoD != 0 in idescsi_pc_intr
hdk: ATAPI reset complete
ide-scsi: CoD != 0 in idescsi_pc_intr
hdk: ATAPI reset complete
ide-scsi: scatter gather table too small, padding with zeros
ide-scsi: CoD != 0 in idescsi_pc_intr
hdk: ATAPI reset complete
ide-scsi: scatter gather table too small, padding with zeros
ide-scsi: scatter gather table too small, padding with zeros
ide-scsi: CoD != 0 in idescsi_pc_intr
hdk: ATAPI reset complete

... and then (after a variable number of those messages) the burn fails. I tried connecting the CDR to other
controllers and the only one that doesn't have this problem is the VIA controller.
Even the promise fails. So this isn't the fault of the HPT370 driver.

I tested kernels 2.4.1 , 2.4.8 & 2.4.12, 2.4.13, 2.4.13-ac5 & ac6 and all have this problem. I then tested 2.2.18
(when i still had openlinux installed. i since installed RH 7.2) with the ide patches and it runs cleanly.
Not a single message from the kernel under the same load and using the same cdrecord binary (1.10)
and the same (or equivalent) drivers loaded. So it isn't a hardware problem either.

The only thing i can find in commom between the bug ocorrences in the HPT and the Promise is that they
both share an IRQ for their 2 IDE channels. The VIA, which does not misbehave, doesn't share interrupts.
I even tried removing the Promise and disabling USB and that does not fix it. And the fact that 2.2.18 works
proves the devices support IRQ sharing. I also found out that this problem also happens when reading from
the CDR. I did not test if it causes data corruption. Also, booting with nosmp does not fix the problem either.

Any patches i could test ? Ideas ? Something ? Is this a known problem ? 

-- 
[------------------------------------------------][-------------------------]
|"One World, One Web, One Program" - Microsoft Ad||    stormlabs@gmx.net    |
|"Ein Volk, Ein Reich, Ein Fuhrer" - Adolf Hitler||http://storm.superzip.net|
[------------------------------------------------][-------------------------]
       --> thor up 2 days | sentinel up 59 days | loki up 59 days <--
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Wed Nov 07 2001 - 21:00:20 EST