2015-03-26 16:12:29

by George Moutsopoulos

[permalink] [raw]
Subject: mediatek mt7630e and ksoftirqd CPU load

Dear developers,

if this is not the right place to post this, sorry for the spam. I
haven't filed a bug report yet. Tell me if I should. I am not sure if
the problem is something kernel developers can solve.

On a laptop ASUS TP500LN, I have a wireless pcie card mt7630e from
mediatek. It is a wifi/bluetooth combo.

The driver is not in the kernel, but mediatek provides it for linux
from the download site
http://www.mediatek.com/en/downloads/

The driver does not support kernel 3.16 and above. I followed the
instructions at
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1220146

The module compiles and loads succesfully. I followed the post by
tobias bora from the ubuntu bug. It compiles on kernel version 3.16
and above (tried up to 3.19).

Like the rest on the ubuntu bug list, the wifi works and I connect
succesfully, but ksoftirqd gets 100% cpu on one of the cores.

Shall I open a bug report? I have contacted mediatek twice but they
haven't replied. I hope it is something you can look at.

Many regards,
George


2015-03-26 16:56:14

by George Moutsopoulos

[permalink] [raw]
Subject: Re: mediatek mt7630e and ksoftirqd CPU load

Thank you Larry for the prompt reply. This is very clear.
Regards,
George

On Thu, Mar 26, 2015 at 6:44 PM, Larry Finger <[email protected]> wrote:
> On 03/26/2015 11:12 AM, George Moutsopoulos wrote:
>>
>> Dear developers,
>>
>> if this is not the right place to post this, sorry for the spam. I
>> haven't filed a bug report yet. Tell me if I should. I am not sure if
>> the problem is something kernel developers can solve.
>>
>> On a laptop ASUS TP500LN, I have a wireless pcie card mt7630e from
>> mediatek. It is a wifi/bluetooth combo.
>>
>> The driver is not in the kernel, but mediatek provides it for linux
>> from the download site
>> http://www.mediatek.com/en/downloads/
>>
>> The driver does not support kernel 3.16 and above. I followed the
>> instructions at
>> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1220146
>>
>> The module compiles and loads succesfully. I followed the post by
>> tobias bora from the ubuntu bug. It compiles on kernel version 3.16
>> and above (tried up to 3.19).
>>
>> Like the rest on the ubuntu bug list, the wifi works and I connect
>> succesfully, but ksoftirqd gets 100% cpu on one of the cores.
>>
>> Shall I open a bug report? I have contacted mediatek twice but they
>> haven't replied. I hope it is something you can look at.
>
>
> Due to the fact that the driver is not in the kernel, it is not possible to
> file a bug with anyone but the vendor.
>
> It is possible that someone on this list with Mediatek experience may be
> able to help you, but there is no certainty. That is the downside of using a
> vendor driver.
>
> As the specs for that computer specify "integrated 802.11 b/g/n", it is
> probably not possible to change that device. I suggest that you purchase an
> inexpensive USB wifi stick to provide wifi on this computer. If you do so,
> you will be aware enough to check for Linux compatibility.
>
> Larry
>

2015-03-26 16:44:47

by Larry Finger

[permalink] [raw]
Subject: Re: mediatek mt7630e and ksoftirqd CPU load

On 03/26/2015 11:12 AM, George Moutsopoulos wrote:
> Dear developers,
>
> if this is not the right place to post this, sorry for the spam. I
> haven't filed a bug report yet. Tell me if I should. I am not sure if
> the problem is something kernel developers can solve.
>
> On a laptop ASUS TP500LN, I have a wireless pcie card mt7630e from
> mediatek. It is a wifi/bluetooth combo.
>
> The driver is not in the kernel, but mediatek provides it for linux
> from the download site
> http://www.mediatek.com/en/downloads/
>
> The driver does not support kernel 3.16 and above. I followed the
> instructions at
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1220146
>
> The module compiles and loads succesfully. I followed the post by
> tobias bora from the ubuntu bug. It compiles on kernel version 3.16
> and above (tried up to 3.19).
>
> Like the rest on the ubuntu bug list, the wifi works and I connect
> succesfully, but ksoftirqd gets 100% cpu on one of the cores.
>
> Shall I open a bug report? I have contacted mediatek twice but they
> haven't replied. I hope it is something you can look at.

Due to the fact that the driver is not in the kernel, it is not possible to file
a bug with anyone but the vendor.

It is possible that someone on this list with Mediatek experience may be able to
help you, but there is no certainty. That is the downside of using a vendor driver.

As the specs for that computer specify "integrated 802.11 b/g/n", it is probably
not possible to change that device. I suggest that you purchase an inexpensive
USB wifi stick to provide wifi on this computer. If you do so, you will be aware
enough to check for Linux compatibility.

Larry


2015-04-01 16:05:49

by Jakub Kicinski

[permalink] [raw]
Subject: Re: mediatek mt7630e and ksoftirqd CPU load

On Thu, 26 Mar 2015 11:44:44 -0500, Larry Finger wrote:
> On 03/26/2015 11:12 AM, George Moutsopoulos wrote:
> > Dear developers,
> >
> > if this is not the right place to post this, sorry for the spam. I
> > haven't filed a bug report yet. Tell me if I should. I am not sure if
> > the problem is something kernel developers can solve.
> >
> > On a laptop ASUS TP500LN, I have a wireless pcie card mt7630e from
> > mediatek. It is a wifi/bluetooth combo.
> >
> > The driver is not in the kernel, but mediatek provides it for linux
> > from the download site
> > http://www.mediatek.com/en/downloads/
> >
> > The driver does not support kernel 3.16 and above. I followed the
> > instructions at
> > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1220146
> >
> > The module compiles and loads succesfully. I followed the post by
> > tobias bora from the ubuntu bug. It compiles on kernel version 3.16
> > and above (tried up to 3.19).
> >
> > Like the rest on the ubuntu bug list, the wifi works and I connect
> > succesfully, but ksoftirqd gets 100% cpu on one of the cores.
> >
> > Shall I open a bug report? I have contacted mediatek twice but they
> > haven't replied. I hope it is something you can look at.
>
> Due to the fact that the driver is not in the kernel, it is not possible to file
> a bug with anyone but the vendor.

Larry, did you notice that rt2x00/mt_linux.c does not have a GPL
header? Do you think this can cause some legal problems if this code
or anything based on it is merged upstream?

> It is possible that someone on this list with Mediatek experience may be able to
> help you, but there is no certainty. That is the downside of using a vendor driver.

I can try to help, sorry for a late response.

George, can you try https://github.com/kuba-moo/mt7630e ? This is a
slightly cleaned up version of the vendor driver V3.14_20140625_v2.

I run it for a few hours and I cannot reproduce the issue you describe.

2015-04-01 20:01:36

by Jakub Kicinski

[permalink] [raw]
Subject: Re: mediatek mt7630e and ksoftirqd CPU load

On Wed, 1 Apr 2015 19:33:13 +0000, [email protected] wrote:
> Thank you very much for your reply.
>
> https://github.com/kuba-moo/mt7630e solves my 100% CPU problem.
>
> I will publicize your repository on the Ubuntu bug
>
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1220146
>
> If you don’t mind. You will make many people like me happy!

Glad to hear that it works for you.