2.2.X lockups -> myfix -> nodma

Scott Marlowe (smarlowe@ihs.com)
Wed, 6 Oct 1999 09:17:55 -0600 (MDT)


I've experienced lockups with the 2.2.5, 2.2.9, and 2.2.12 kernel under
testing with very heavy disk activity.

Hardware:

AMD K6-2 300MHz with 64 Meg PC100 DIMM
FIC VA503+ mobo (apollo MVP3 chipset)
1 x 4G Fujitsu Ultra DMA 33 hard drive
1 x 17G Maxtor Ultra DMA 66 hard drive
1 x 8.5G Maxtor Ultra DMA 66 hard drive

All of the 8.5 and part of the 17 form a RAID-1 set. Under testing or
mkraid or mke2fs the system would hang, no oops, no messages, no nothing,
solid lockup. CTRL-ALT-DEL had no effect, I had to hit the reset button for
the first time in my life with Linux.

I had the kernel compiled with RAID-1 et. al. included (not modules) and DMA
enabled. I recompiled the kernel with DMA disabled, and it seems rock solid
stable now. I replaced the cables, and the drives being accessed in RAID-1
will lock up the machine. Removing the RAID-1 and running the drives
individually it does not lock up, but accessing both at the same time (dd
if=/dev/hdxn | md5sum) causes lockups with DMA on.

note that without DMA I'm xferring at 6MBytes/sec with hdparm -t, with
50-70% CPU utilization. I'd really like to be able to run DMA, I've got a
server that's the same exact hardware running 2.0.36 with DMA on with no
lockups, under heavy simultaneous drive load. I'll test some more and see
if I can isolate it to the 2.2.x series kernels and DMA.

Scott Marlowe email: smarlowe@ihs.com (work)
Systems Engineer phone: (303)858-6106
Curriculum Dev pager: (888)473-5549
Web Developer IHS Enigineering

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/