Mitigation of license for the symbol __rt_mutex_init and more

Mitigation of license for the symbol __rt_mutex_init and more

[Andrey Gursky]

Hi!

Since coming of the linux-3.0-rt patch I've decided to try to switch
to that kernel, especially the goal of the RT_PREEMPT claims to make a
desktop kernel become a realtime one. Thanks to all rt-developers for
this opportunity!

I'm testing it on 3 platforms:
1) notebook Core2Duo T7250,
2) old Pentium 4 3.2 GHz with an old passive ISDN hisax-card,
3) even older AthlonXP 2.8 GHz with TV-tuner.

The first desktop problem is:
the symbol __rt_mutex_init is exported as GPL-only. This make the
usage of proprietary nvidia driver impossible, while the normal
__mutex_init is not only GPL. Could you change the license also for
the rt_mutex_init (...and maybe few more if they being linked with
nvidia/ati(?) driver just as with the vanilla "desktop" kernel)?

The next issue is more complicated. On the P4 (platform 2) I use an
ISDN-hisax modem card and per-second internet account. After a timeout
the connection is being hung up. In firefox I have a lot of tabs. Some
web pages are "active". They try to call back on their own. While this
is happening my gnome freezes sometimes. SysRq-R helps sometimes to
switch back to the console. But the gdm doesn't start anymore. Is this
a relevant issue for the rt-project? What'd be the first place to look
at to debug this behavior?

Regards,
Andrey

Re: Mitigation of license for the symbol __rt_mutex_init and more

[Uwe Kleine-König]

On Mon, Aug 08, 2011 at 01:10:22AM +0200, Andrey Gursky wrote:
> Hi!
> 
> Since coming of the linux-3.0-rt patch I've decided to try to switch
> to that kernel, especially the goal of the RT_PREEMPT claims to make a
> desktop kernel become a realtime one. Thanks to all rt-developers for
> this opportunity!
> 
> I'm testing it on 3 platforms:
> 1) notebook Core2Duo T7250,
> 2) old Pentium 4 3.2 GHz with an old passive ISDN hisax-card,
> 3) even older AthlonXP 2.8 GHz with TV-tuner.
> 
> The first desktop problem is:
> the symbol __rt_mutex_init is exported as GPL-only. This make the
> usage of proprietary nvidia driver impossible, while the normal
> __mutex_init is not only GPL. Could you change the license also for
> the rt_mutex_init (...and maybe few more if they being linked with
> nvidia/ati(?) driver just as with the vanilla "desktop" kernel)?
> 
> The next issue is more complicated. On the P4 (platform 2) I use an
> ISDN-hisax modem card and per-second internet account. After a timeout
> the connection is being hung up. In firefox I have a lot of tabs. Some
> web pages are "active". They try to call back on their own. While this
> is happening my gnome freezes sometimes. SysRq-R helps sometimes to
> switch back to the console. But the gdm doesn't start anymore. Is this
> a relevant issue for the rt-project? What'd be the first place to look
> at to debug this behavior?
enable PROVE_LOCKING and see if that claims to find something. You might
have to setup netconsole for that to work.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Mitigation of license for the symbol __rt_mutex_init and more

[Andrey Gursky]

Uwe Kleine-König <u.kleine-koenig <at> pengutronix.de> writes:

> 
> On Mon, Aug 08, 2011 at 01:10:22AM +0200, Andrey Gursky wrote:
> > Hi!
> > 
> > Since coming of the linux-3.0-rt patch I've decided to try to switch
> > to that kernel, especially the goal of the RT_PREEMPT claims to make a
> > desktop kernel become a realtime one. Thanks to all rt-developers for
> > this opportunity!
> > 
> > I'm testing it on 3 platforms:
> > 1) notebook Core2Duo T7250,
> > 2) old Pentium 4 3.2 GHz with an old passive ISDN hisax-card,
> > 3) even older AthlonXP 2.8 GHz with TV-tuner.
> > 
> > The first desktop problem is:
> > the symbol __rt_mutex_init is exported as GPL-only. This make the
> > usage of proprietary nvidia driver impossible, while the normal
> > __mutex_init is not only GPL. Could you change the license also for
> > the rt_mutex_init (...and maybe few more if they being linked with
> > nvidia/ati(?) driver just as with the vanilla "desktop" kernel)?
> > 
> > The next issue is more complicated. On the P4 (platform 2) I use an
> > ISDN-hisax modem card and per-second internet account. After a timeout
> > the connection is being hung up. In firefox I have a lot of tabs. Some
> > web pages are "active". They try to call back on their own. While this
> > is happening my gnome freezes sometimes. SysRq-R helps sometimes to
> > switch back to the console. But the gdm doesn't start anymore. Is this
> > a relevant issue for the rt-project? What'd be the first place to look
> > at to debug this behavior?
> enable PROVE_LOCKING and see if that claims to find something. You might
> have to setup netconsole for that to work.

As of 3.0-rt3 the NETCONSOLE depends on !PREEMPT_RT_FULL [=y], so I couldn't
enable it. I've tried to satisfy the dependencies of NETCONSOLE_DYNAMIC:
Depends on: NETDEVICES [=y] && NETCONSOLE [=n] && SYSFS [=y] && CONFIGFS_FS [=y]
&& (NETCONSOLE [=n]!=y || CONFIGFS_FS [=y]!=m)
Despite I have NETDEVICES=y, SYSFS=y, CONFIGFS_FS=y and NETCONSOLE doesn't
appear in .config, it is impossible to turn on the NETCONSOLE_DYNAMIC.

Regards,
Andrey

--
To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html