2.6.28-3-rt fails to boot with dmraid on amd64

[Nicolas Joyard]

Binary package hint: dmraid

Using Jaunty beta

linux-rt 2.6.28.3.1
dmraid 1.0.0.rc15-6ubuntu1

Using a fake-raid0 on an nvidia chipset. Booting with 2.6.28-11-generic on this system works.

When booting with 2.6.28-3-rt, fails to boot most of the time :
- sometimes it hangs on "Mounting local filesystems..."
- sometimes it hangs on the very beginning of fsck
- sometimes it manages to boot (every 15-20 boots) !

When the system hangs, even sysrq commands don't work.

The following message can be seen in /var/log/messages when boot succeeds (it does not appear when using -generic) :
dmraid-activate: ERROR: Cannot retrieve RAID set information for nvidia_ibifiefd

The output from lspci -nnvv and dmraid -ay -vvv -d are attached, as well as kern.log from a successful -rt boot (I cannot find any log from failed boots).

[jrk (jrk+)]

I'm experiencing the same problem on a fakeraid 1 with a Via VT8237S chipset.

Boots fine with the 2.6.28-11-generic, but if I try 2.6.28-3-rt I get a kernel panic (not syncing VFS, unable to mount root fs on unknown block).

I was able to boot rt when using the i386 image, but not with the AMD64 image.

[Danny Wood (danwood76)]

Sounds like a kernel issue.

Make sure the the initramfs is built properly by running:
sudo update-initramfs -u -k all

Will test the rt kernels later.
I did build my own rt 2.6.29 and had no issues so its probably an issue with the rt kernel, I would post a bug against that.

[Nicolas Joyard (joyard-nicolas)]

I forgot to mention I was running with AMD64 architecture.

I tried updtating the initramfs, without any improvement. I also found out that when the system manages to boot to the desktop, it still hangs eventually.

Let me know if any other information would be useful.

[jrk (jrk+)]

Sorry, forgot to mention I already tried sudo update-initramfs -u -k all.

[Nicolas Joyard (joyard-nicolas)]

I tried booting with the "nosmp" kernel option. The system booted fine, and has been running without hanging for about 2 hours. But it seems a bit of a shame to run with nosmp on a quad-core...

[Alessio Igor Bogani (abogani)]

It seems unrelated to dmraid.

[Alessio Igor Bogani (abogani)]

Fix that bug require to analyze plus of 2000 line of code. No way to do this in and for Jaunty Release due SRU policies. So I postponed that bug to Karminc, sorry.

[Nicolas Joyard (joyard-nicolas)]

I think the issue is solved, as 2.6.31-5-rt (karmic alpha) now boots fine. Alessio, do you know about any changes that were made to the "plus of 2000 lines of code" you previously talked about in this way, or is this just a (nice) coincidence ?

[Alessio Igor Bogani (abogani)]

I talked about Jaunty. In Karmic Upstream do the magic!