Linux software RAID maintainers and developers,

Two months ago I wrote to the Linux kernel mailing list regarding a 
condition expressed as "BUG: soft lockup - CPU#3 stuck for 61s!".  I 
initially battled this recurring problem in both Fedora 10 and Fedora 
11.  Rafael J. Wysocki suggested that I update the kernel (to 2.6.31-rc4 
or later) and see if the problem resurfaced.  I then used kernel 
2.6.31-0.94.rc4.fc12.x86_64 and found that the problem still continued, 
but noticeably only when the md data-check process was run.

You can read the last post to the LKML thread (with links to the entire 
thread) here:

http://lkml.org/lkml/2009/8/6/387

The md data-check is being run in Fedora 11 by 
/etc/cron.weekly/raid-check which is a little shell script that looks 
like this:

------------------------------------------------------
#!/bin/bash

for dev in `grep "^md.*: active" /proc/mdstat | cut -f 1 -d ' '`; do
        [ -f /sys/block/$dev/md/sync_action ] && \
                echo "check" > /sys/block/$dev/md/sync_action
done
------------------------------------------------------

I have disabled this weekly data-check, and since doing so have not 
encountered any soft lockup (or any other problem, for that matter).

For reference, you can see in this thread here that I am not the only 
one to have this problem:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/212684

It may important to note that I am running a 4-disk RAID1 array on 500GB 
Western-Digial SATA drives.  I've attached the output of 'lspci -vv' as 
"lspci.out".  I've attached the output of 'hdparm -I /dev/sda' as 
"hdparm.out".  /dev/sda, /dev/sdb, /dev/sdc, and /dev/sdd are all 
identical drive types.  I've attached the output of 'for i in 0 1 2; do 
mdadm --detail /dev/md$i; done' as "mdadm.out".

Please let me know if this is a configuration problem, a kernel bug, or 
something else.  Please let me know how to fix this problem so that I 
can safely re-enable md data-check.

Thanks,

Lee.