Bug 460121

Summary:	RHEL4U7 GA cpu usage raise to 100% for 30s when plugin a usb1.1 device
Product:	Red Hat Enterprise Linux 4	Reporter:	Andiry <andiry.xu>
Component:	kernel	Assignee:	Pete Zaitcev <zaitcev>
Status:	CLOSED WONTFIX	QA Contact:	Martin Jenner <mjenner>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	4.7	CC:	andriusb, balkov, bhavna.sarathy, cward, emcnabb, joachim.deguara, libin.yang, peterm, rdoty, shane.huang, tao, tom.gao, vgoyal, zaitcev
Target Milestone:	rc
Target Release:	4.8
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2009-02-25 19:52:42 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	450283

Description Andiry 2008-08-26 09:52:24 UTC

Description of problem:
RHEL4U7 GA cpu usage raise to 100% for 30s when plugin a usb1.1 device

Version-Release number of selected component (if applicable):
RHEL4U7, 2.6.9-78.EL.  

How reproducible:
Always

Steps to Reproduce:
1. install RHEL4U7 GA 32bit or 64bit
2. boot to 2.6.9-78.EL kernel, NOT smp kernel
3. unplug a usb mouse and re- insert, system become slow, meanwhile cpu usage raised to 100%, for about 30s.
  
Actual results:
Cpu usage raised to 100% for 30s.

Expected results:
Cpu usage should return to normal in a short time.

Additional info:

Comment 1 Russell Doty 2008-08-26 14:10:39 UTC

What chipsets is this occurring with? Does it occur with other USB 1.1 devices? Do USB 2.0 devices behave properly?

Comment 2 Andiry 2008-08-27 01:21:48 UTC

We found this issue occur on AMD SB700 and Intel ICH6. 
It also occurs with other USB 1.1 devices like USB keyboard and USB1.1 speaker.
USB 2.0 devices behave properly.

Comment 3 Issue Tracker 2008-10-15 21:59:04 UTC

(sending this to the engineering bugzilla too)

I am able to reproduce this in my environment. Here are my observations so
far:

On RHEL4.7, I do see a CPU spike for around 5 - 10 seconds but not as long
as 30 seconds. This occurs on both the SB700 and BCM5785 systems at my
desk. This noticeably slows down the entire system - for example, when
plugging in USB speakers the mouse becomes quite choppy for several
seconds. This occurs whether X is running or not (regardless of runlevel),
and with kernel-smp there is a CPU spike but it doesn't seem noticeable.

Right before leaving today, I installed RHEL4.5 (to see if this was
possibly a regression) and it seems to handle device plugging much better.
The CPU spike only lasts for 1 or 2 seconds and isn't noticeable to the
user. AMD, could you try installing RHEL4.5 to verify my observation?


This event sent from IssueTracker by emcnabb 
 issue 215950

Comment 4 Andiry 2008-10-16 03:28:38 UTC

(Copy from issue-tracker)

Yeah, Evan, you are right.

I verified on RHEL 4.5 (2.6.9-55.EL), the symptom is much better compared to RHEL 4.6. 

I use a USB 1.1 speaker to veirfy. On RHEL 4.5, only the first time plug in the device will cause a CPU spike for about 2-3 seconds, and then plug-out and plug in the same device, there is no CPU spike anymore. While on RHEL 4.6, each time plug in the same device, the CPU spike lasts for about 10 seconds.

Comment 7 Pete Zaitcev 2009-01-13 18:45:24 UTC

I am unable to see it on 2.6.9-78.12.EL, but my userland is behind:

hotplug-2004_04_01-7.8
udev-039-10.22.el4
module-init-tools-3.1-0.pre5.3.10
hal-0.4.2-8.EL4
hal-0.4.2-8.EL4 (i386)

N.B., using speakers is not possible unless bug 460100 is fixed (or at
least my patch attached there is applied).

I exchanged e-mails with Evan regarding the circumstances, and he says
this is possible to trigger with a keyboard. His configuration that
produces a noticeable spike is a keybord directly attached to OHCI.