usb/iLO problem on DL320 g4

[Igor Dobrovitski]

Greetings!
We recently bought 2 DL320 g4 boxes, and installed FC4 Linux on them.
The following messages are observed in the logs, they happen every second or so:
Mar 27 08:22:29 ldap02 kernel: usb 5-2: new full speed USB device using uhci_hcd and address 115
Mar 27 08:22:32 ldap02 kernel: hub 5-2:1.0: USB hub found
Mar 27 08:22:32 ldap02 kernel: hub 5-2:1.0: 7 ports detected
Mar 27 08:22:33 ldap02 kernel: usb 5-1: USB disconnect, address 114
Mar 27 08:22:34 ldap02 kernel: usb 5-1: new full speed USB device using uhci_hcd and address 116
Mar 27 08:22:38 ldap02 kernel: input: HP Virtual Keyboard as /class/input/input39554
Mar 27 08:22:38 ldap02 kernel: input: USB HID v1.01 Keyboard [HP Virtual Keyboard] on usb-0000:01:04.4-1
Mar 27 08:22:40 ldap02 kernel: input: HP Virtual Keyboard as /class/input/input39555
Mar 27 08:22:40 ldap02 kernel: input: USB HID v1.01 Mouse [HP Virtual Keyboard] on usb-0000:01:04.4-1
Mar 27 08:22:41 ldap02 kernel: usb 5-2: USB disconnect, address 115

This takes up quite a bit of CPU, and only booting Linux with usb support turned off could get rid of these messages. We have done everything possible attempting to rectify the situation, all to no avail:
1. Install PSP-740
2. Upgrade the kernel to the latest, and install PSP-7.40
3. Upgrade iLO2 firmware to the latest version. The BIOS is at the latest version.
4. Disabling both iLO and USB doesn't get rid of the messages - although I feel that it should have.

The cause of the trouble seems to be the iLO2's virtual mouse and keyboard continuously re-connecting for some reason. The 2 servers that we bought exhibit this behaviour, so it's not a single faulty box.
Has anyone else installed FC4 on a Proliant DL320, and did they have the same problems?

Thanks in advance.

[Hynes, C.]

Hi Igor,

I am experiencing the same problems with FC5. In addition, the system hangs on boot while loading Bluetooth services.

Bye,
Claudio

[Jef_4]

Hi,

We see this problem with both RHEL AS 4, as well as Debian Etch. These aren't optimal workarounds, but they currently work for our needs while evaluate them.

First, the offending kernel module that causes this problem is uhci_hcd. A simple rmmod of that mod will stop the messages. (took a while to figure out, as the messages never stop scrolling -- was typing blind for a while :) )

If you build your own kernels, skipping the UHCI USB driver also solves the problem. (Keeping the USB 2.0 driver, EHCI enabled, is fine)

I am still interested in knowing what might be going on with the hardware that this keeps happening, so if anyone has insight, please let me know.

I do intend to follow up with Redhat on this, as they actually certify RHEL 3 and 4 on this hardware. I also intend to follow up with HP support as well.

jef

[Erik Dykema]

Hi-
I also experience the same problem, on a DL 320 G4. With both FC 5 and Debian Etch, the "Virtual Keyboard' message keeps reappearing over and over again, and the server hangs at boot for FC5. With Debian, if you press ctrl-c on a physical keyboard, you can get to a place to login (typing blind as previously described).
For the guy who said that it was hanging at bluetooth, I noticed that too, but it's not actually hanging on the bluetooth driver, it's hanging right after trying the bluetooth driver, with this same thing. Any resolution you get from Redhat / Debian / HP, please share!!

thanks,
Erik

[Jef_4]

I've been working with HP support on this issue, but it doesn't appear to really be an HP problem. After our initial install with the looping problem, 2 more RHEL AS 4 installs behaved normally. The uhci_hcd driver loads, and then ends with this bit of output:

May 8 13:08:47 michigan kernel: uhci_hcd 0000:01:04.4: UHCI Host Controller
May 8 13:08:47 michigan kernel: uhci_hcd 0000:01:04.4: port count
misdetected? forcing to 2 ports

This is where the Debian Etch Beta 2 install continues on and goes loopy (literally).

The only thing I can think that I did different with the initial RedHat install was that I didn't use the X installer. The other 2 I left plain vanilla. I'm going to continue a few more tests, to see if there is some bug in the RedHat installer that gets tickled.

I'm guessing something in the actual kernel itself is the problem. More research to be done.....

[Olivier Bozetie]

I get the same message on Proliant ML310 G3 with latest BIOS trying to install Debian etch (latest build).
By the way, no success in disabling iLO2 and/or USB+USB2. On the other hand i am not sure to completely disable iLO2.
It seems like the behaviour is similar with FC5... how am i going to install a Linux on this server, that's a good question !
Mister HP, any advice ?
Is there a way to disable iLO2 and/or an Debian install option not to probe these virtual usb devices ?
Thx
Olivier

[Jef_4]

Olivier,

The messages, other than being annoying, appear to be mostly harmless. However, I'm not making any use of any major USB features at the moment, either, so it may depend on your situation. But as a short-term way to do some testing, you can install an init script that will do an rmmod uhci_hcd.