Bug 431312 - 2.6.24: After a random amount of time, SATA drive speed is reduced to 20MB/s
Summary: 2.6.24: After a random amount of time, SATA drive speed is reduced to 20MB/s
Keywords:
Status: CLOSED UPSTREAM
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: rawhide
Hardware: All
OS: Linux
low
low
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2008-02-02 15:33 UTC by Stewart Adam
Modified: 2008-02-09 00:15 UTC (History)
0 users

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2008-02-09 00:15:35 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Gzip'd test data (25.20 KB, application/x-gzip)
2008-02-02 15:33 UTC, Stewart Adam
no flags Details

Description Stewart Adam 2008-02-02 15:33:49 UTC
Description of problem:
I have three SATA drives, a Raptor connected to Intel ICH8 and 2x WD Caviar
drives connected to a JMicron  20360/20363. Kernels 2.6.22 and 2.6.23 are fine,
however booting into 2.6.24 results in all SATA drive speeds being reduced to
20MB/s from the usal 65-80MB/s range I get with "hdparm -t /dev/sdX"

What is weird about this problem is I discovered that the correct drivers are
being used, and booting into single user mode and testing with hdparm does
result in about 75MB/s on all drives. However, after a bit of disk activity (eg
changing runlevels or starting X) the drive speed is suddenly reduced.

Version-Release number of selected component (if applicable):
kernel-2.6.24-9.fc9

How reproducible:
Always

Steps to Reproduce:
1. Boot in single user mode, hdparm -t /dev/sda
2. Boot in 3rd runlevel, hdparm -t /dev/sda
3. Boot in 5th runlevel, hdparm -t /dev/sda
4. Boot in single user mode and start service one by one, after each start run
hdparm -t /dev/sda
  
Actual results:
Results from hdparm on each of the steps above:
1 --> 75MB/s
2 --> 20MB/s
3 --> 20MB/s
4 --> 75MB/s, however at one point the drives speed will drop to 20MB/s (it
isn't a specific service, it will change each time a trial is run)

Expected results:
Drive speed isn't reduced to 20MB/s


Additional info:
I've attached logs from various kernels:
* kernel-2.6.24-9.fc9.x86_64 (before drive speed was reduced)
* kernel-2.6.24-9.fc9.x86_64 (after drive speed was reduced)
* kernel-2.6.24-0.138.rc7.fc9.x86_64
* kernel-2.6.22.9-91.fc7.x86_64
The logs include dmesg, a timed run of dd reading a non-cached 4.6GB file, the
`vmstat 1` output while the file was being copied and the hdparm -t tests on all
three SATA drives.

Comment 1 Stewart Adam 2008-02-02 15:33:49 UTC
Created attachment 293797 [details]
Gzip'd test data

Comment 2 Stewart Adam 2008-02-09 00:15:35 UTC
I saw the commit for 2.6.24.1 and rebuilt it locally, problem solved.

$ sudo hdparm -t /dev/sda /dev/sdb /dev/sdc

/dev/sda:
 Timing buffered disk reads:  192 MB in  3.01 seconds =  63.72 MB/sec

/dev/sdb:
 Timing buffered disk reads:  184 MB in  3.03 seconds =  60.70 MB/sec

/dev/sdc:
 Timing buffered disk reads:  224 MB in  3.01 seconds =  74.34 MB/sec
 


Note You need to log in before you can comment on or make changes to this bug.