Red Hat Bugzilla – Bug 54499
Dual IDE on Via Apollo Pro 598 chipset hangs on heavy DMA
Last modified: 2008-08-01 12:22:51 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.3) Gecko/20010816
Description of problem:
Kernel 2.4.3-12 running on Gigabyte 6VXE+ (VIA Apollo Pro chipset), will
hang when doing heavy DMA used on dual IDE channels (e.g., Sofware RAID
configuration). This can only be worked around when hard disk is set to no
DMA at the expense of system performance.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1.Sofware RAID 1 config, master drives (hda, hdc) on each IDE channel
2.perform heavy I/O (simultaneous file copy in multiple consoles of large
(> 100 MB) files to RAID partitions, and 'sync' issued.
3.Disk access will hang on hdc. System will attempt to auto-recover via IDE
Reset, and disabling DMA on hdc.
Actual Results: System recovers most of the time using kernel 2.4.3-12
after significant delay and IDE Reset. Kernel 2.4.2 RH7.1 default) install
would usually lockup hard. Kernel 2.4.9-0.18 (Rawhide Oct 3) would lockup
hard, no error messages. If DMA is disabled on hdc, the problem is not
Expected Results: DMA on dual IDE (software RAID) to function properly on
RH 7.1 distribution, up to date wrt. current errata.
I tried the Rawhide Kernel 2.4.9-0.18 (Oct 3) with necessary RPM upgrades
to satisfy dependencies (modutils, mkinitrd, e2fsprogs, setup, filesystem,
bash). System would lockup when heavy DMA occurs, no error messages,
necessitating a reboot using alt-sysrq. I thought that fixes for VIA
chipsets should've gone in since 2.4.3 ?
Created attachment 33723 [details]
dmesg for 2.4.3-12 with DMA on hda enabled, no DMA on hdc
Created attachment 33724 [details]
Created attachment 33725 [details]
Output of lspci
Attachments indicate working system config with DMA disabled on hdc. Bug is
triggered when both IDE channels have DMA enabled.
I have seen a similar hang with RedHat 7.1, 7.2, and 8.0, attempting
to access hda on a VIA Apollo VT82C586 IDE controller, when it is paired
with a second disk as hdb (and a CDROM as hdc and a CDRW as hdd). Even
in Anaconda, the hang happens as soon as parted tries to access the drive.
Sometimes it even hangs at boot time, right after it reads the partition
table on hda. This same hardware works just fine with a 2.2 kernel.
Hmm, I've given up on this config (and VIA chipsets). Incidentally, I've also
retired that motherboard, and have not gone back to a Software RAID config
since then. Sorry, can't comment further.
Thanks for the bug report. However, Red Hat no longer maintains this version of
the product. Please upgrade to the latest version and open a new bug if the problem
The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases,
and if you believe this bug is interesting to them, please report the problem in
the bug tracker at: http://bugzilla.fedora.us/