Bug 21022 - Kernel 2.6.36 starts and works VERY slow, with long delays
Summary: Kernel 2.6.36 starts and works VERY slow, with long delays
Status: RESOLVED INVALID
Alias: None
Product: Other
Classification: Unclassified
Component: Other (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: other_other
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-10-23 23:38 UTC by Ruslan
Modified: 2012-08-14 11:23 UTC (History)
2 users (show)

See Also:
Kernel Version: 2.6.36
Subsystem:
Regression: Yes
Bisected commit-id:

Attachments
dmesg log for 2.6.36 (36.46 KB, text/x-log)
2010-10-23 23:38 UTC, Ruslan 		Details
Log from 2.6.34 (47.09 KB, text/x-log)
2010-10-23 23:39 UTC, Ruslan 		Details
Config file after make menuconfig (72.81 KB, application/xml)
2010-10-23 23:41 UTC, Ruslan 		Details
Add an attachment (proposed patch, testcase, etc.)

Description Ruslan 2010-10-23 23:38:36 UTC 
Created attachment 34602 [details]
dmesg log for 2.6.36

First what i noticed after upgrade from 2.6.34 to 2.6.36 is very long delays between some startup messages. I used the same .config file from 2.6.34, just made 'make menuconfig'; Esc-Esc, Enter to update the configuration.
I attach the dmesg logs for 2.6.36 and 2.6.34 for comparison. Note the timestamps.
Comment 1 Ruslan 2010-10-23 23:39:55 UTC 
Created attachment 34612 [details]
Log from 2.6.34
Comment 2 Ruslan 2010-10-23 23:41:19 UTC 
Created attachment 34622 [details]
Config file after make menuconfig
Comment 3 Andrew Morton 2010-10-25 23:01:24 UTC 
Your 2.6.36 build has

+ kmemcheck: Limiting number of CPUs to 1.
+ kmemcheck: Initialized
+ Brought up 1 CPUs
+ Total of 1 processors activated (3733.22 BogoMIPS).

Can you try disabling kmemcheck?
Comment 4 Ruslan 2010-10-26 11:14:50 UTC 
Without kmemcheck it works OK.
Comment 5 Ruslan 2010-10-26 11:43:52 UTC 
Changing kmemcheck on boot state from "one-shot" to "disabled" removes the delays still preserving "Limiting number of CPUs to 1".
Comment 6 Ruslan 2010-11-13 23:45:46 UTC 
So... is this normal for kmemcheck, so that this bug would be invalid, or is it still a real bug?
Note You need to log in before you can comment on or make changes to this bug.