Bug 47791

Summary: AMD SB950 USB 2 regression first with kernel 2.6.37
Product: Drivers Reporter: AMD Linux User (dagecko)
Component: USBAssignee: Greg Kroah-Hartman (greg)
Status: RESOLVED DOCUMENTED    
Severity: high CC: dagecko, jrnieder, ray.huang, shane.huang
Priority: P1    
Hardware: x86-64   
OS: Linux   
Kernel Version: 3.6-rc6 Subsystem:
Regression: Yes Bisected commit-id:
Bug Depends on:    
Bug Blocks: 21782    
Attachments: output from various commands and log files

Description AMD Linux User 2012-09-22 07:50:11 UTC
Created attachment 80721 [details]
output from various commands and log files

Tested with several kernels:

* 3.0
* most 3.2.x from debian wheezy
* 3.3.0
* 3.4
* 3.6RC6
* 2.6.30 - 2.6.32

USB 2 ports on the mainboard MSI 990XA-GD55 work with kernels 2.6 but stop to work with kernel 3.x.
USB 3 ports work well on both 2.6 and 3.x kernels.

Chipset on the mainboard is an AMD SB950.

I attached a file that reports outputs from: dmesg, lspci -vvv, lsusb -vvv, /var/log/syslog and /var/log/messages.

The main output error when we plug an USB device into the USB 2 port is:

"usb 4-4: device descriptor read/64, error -32"

Tested with:

MSI 990XA-GD55 mainboard, AMD FX 4170 CPU, on Debian Wheezy.
Comment 1 Greg Kroah-Hartman 2012-09-22 14:46:40 UTC
On Sat, Sep 22, 2012 at 07:50:11AM +0000, bugzilla-daemon@bugzilla.kernel.org wrote:
>            Summary: AMD SB950 USB 2 don't work on kernel 3.x but worked on
>                     2.6.x

Please post this on the linux-usb@vger.kernel.org mailing list.
Comment 2 AMD Linux User 2012-09-22 15:08:37 UTC
I just did it as requested.
Comment 3 AMD Linux User 2012-12-02 15:06:39 UTC
Added first version where the bug happens (2.6.37).
Renamed the bug title to reflect more the regression.
Comment 4 AMD Linux User 2012-12-03 13:29:08 UTC
First git commit that introduced the regression: b5153163ed580e00c67bdfecb02b2e3843817b3e
Comment 5 Jonathan Nieder 2012-12-08 08:37:35 UTC
References: http://thread.gmane.org/gmane.linux.usb.general/71551
Comment 6 AMD Linux User 2012-12-08 11:40:10 UTC
With a bisect from the root linux source tree, the commit that is the cause of the regression looks less suspicious.

Kernel version is 2.6.36-rc4

And the commit is this one:

5c80cc78de46aef6cd5e714208da05c3f7f548f8 is the first bad commit
commit 5c80cc78de46aef6cd5e714208da05c3f7f548f8

See references as written above for more information.
Comment 7 AMD Linux User 2013-01-11 15:37:36 UTC
As said here: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/859137 updated to the latest BIOS resolved the bug despite of the fact that the mainboard service explicitly said that it is impossible that the bug could come from them.