Red Hat Bugzilla – Bug 137273
kernel oops in raid5_unplug_device using disk druid under anaconda
Last modified: 2015-01-04 17:11:11 EST
Description of problem:
Installing the FC3 rc (both rc1 with .640 and now rc2 with .643),
there is a kernel oops when I go to assign a mount point to my
existing/formatted RAID 5 array in disk druid. Happens in both
graphical and text install. I'm not installing on the raid array
itself (have an existing rawhide install there) so I'm not sure if
this oops affects ability to access the RAID from that point on; the
install continues on fine to sda2 in my case.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
0. Have an existing RAID 5 array. Mine was originally created
somewhere around RHL 7-9 time, not exactly sure.
1. Boot rescue CD, start install
2. select HTTP install
3. do manual partition using disk druid; assign mount point to RAID
steps 1-2 probably aren't relevant, but that's the process I've been
doing. Actually the whole install thing may not be relevant, maybe
raidstart and then raidstop in rescue mode from one of these .64x
discs is enough to show it? I should try that, duh.
in text mode, you actually see the oops pop up on the background; in
graphical, you don't see it. In either case go to the shell on
console 2 (or wherever it is) and dmesg and you'll get the full text
of the oops.
Attached are three sample dmesg dumps:
1) 'dmesg' command output (thus only recent bits) from rc1 with .640,
in 'linux text' install, showing the oops
2) 'dmesg' from rc2 with .643, graphical install, showing oops again
3) /var/log/dmesg showing complete boot under .643 from fully
installed system so you can see what hardware is all there.
Created attachment 105824 [details]
dmesg w/oops under .640, FC3-rc1 in text install
Created attachment 105825 [details]
dmesg w/oops under .643, FC3-rc2 in graphical install
Created attachment 105826 [details]
full dmesg from .643 boot to fresh FC3-rc2 install
*** This bug has been marked as a duplicate of 127862 ***