[PATCH -rt] disable building all staging drivers

[PATCH -rt] disable building all staging drivers

[Uwe Kleine-König]

I know that at least COMEDI and RT2870 are broken with -rt, but Thomas
requested disabling all of staging.  He's the maintainer.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Greg Kroah-Hartman <greg@kroah.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Frank Mori Hess <fmhess@users.sourceforge.net>
---
 drivers/staging/Kconfig |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/staging/Kconfig b/drivers/staging/Kconfig
index 211af86..3d1a966 100644
--- a/drivers/staging/Kconfig
+++ b/drivers/staging/Kconfig
@@ -1,6 +1,7 @@
 menuconfig STAGING
 	bool "Staging drivers"
 	default n
+	depends on BROKEN
 	---help---
 	  This option allows you to select a number of drivers that are
 	  not of the "normal" Linux kernel quality level.  These drivers
-- 
1.6.3.1

Re: [PATCH -rt] disable building all staging drivers

[Greg KH]

On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> I know that at least COMEDI and RT2870 are broken with -rt,

I didn't know this, has anyone sent me a bug report?  What in -rt causes
these drivers to break?

> but Thomas requested disabling all of staging.  He's the maintainer.

This seems like a patch to ensure that the staging drivers never get a
chance to be fixed for any potential -rt issues.  How about just sending
me bug reports instead?

thanks,

greg k-h

Re: [PATCH -rt] disable building all staging drivers

[Thomas Gleixner]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 666 bytes --]

On Sun, 14 Jun 2009, Greg KH wrote:

> On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > I know that at least COMEDI and RT2870 are broken with -rt,
> 
> I didn't know this, has anyone sent me a bug report?  What in -rt causes
> these drivers to break?
> 
> > but Thomas requested disabling all of staging.  He's the maintainer.
> 
> This seems like a patch to ensure that the staging drivers never get a
> chance to be fixed for any potential -rt issues.  How about just sending
> me bug reports instead?

Sorry, my bad. But at least the patch now brought your attention on
it. I provide you an useful bugreport in the next days.

Thanks,

	tglx

Re: [PATCH -rt] disable building all staging drivers

[Uwe Kleine-König]

Hi Greg,

On Sun, Jun 14, 2009 at 04:28:02AM -0700, Greg KH wrote:
> On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > I know that at least COMEDI and RT2870 are broken with -rt,
> 
> I didn't know this, has anyone sent me a bug report?  What in -rt causes
> these drivers to break?
> 
> > but Thomas requested disabling all of staging.  He's the maintainer.
> 
> This seems like a patch to ensure that the staging drivers never get a
> chance to be fixed for any potential -rt issues.  How about just sending
> me bug reports instead?
COMEDI has:

  CC [M]  drivers/staging/comedi/comedi_fops.o
In file included from drivers/staging/comedi/comedidev.h:40,
                 from drivers/staging/comedi/comedi_fops.c:45:
drivers/staging/comedi/interrupt.h:27: error: conflicting types for ‘irqreturn_t’
include/linux/irqreturn.h:14: error: previous declaration of ‘irqreturn_t’ was here
In file included from drivers/staging/comedi/comedidev.h:40,
                 from drivers/staging/comedi/comedi_fops.c:45:
drivers/staging/comedi/interrupt.h:30:1: warning: "IRQ_RETVAL" redefined
In file included from include/linux/interrupt.h:10,
                 from drivers/staging/comedi/interrupt.h:22,
                 from drivers/staging/comedi/comedidev.h:40,
                 from drivers/staging/comedi/comedi_fops.c:45:
include/linux/irqreturn.h:15:1: warning: this is the location of the previous definition
make[3]: *** [drivers/staging/comedi/comedi_fops.o] Error 1
make[2]: *** [drivers/staging/comedi] Error 2
make[1]: *** [drivers/staging] Error 2
make: *** [drivers] Error 2

commit

	http://git.kernel.org/?p=linux/kernel/git/tip/linux-2.6-tip.git;a=commitdiff;h=bedd30d986a05e32dc3eab874e4b9ed8a38058bb
	
removes the #define IRQ_NONE and drivers/staging/comedi/interrupt.h has

	#ifndef IRQ_NONE
	typedef void irqreturn_t;
	#define IRQ_NONE
	#define IRQ_HANDLED
	#define IRQ_RETVAL(x) (void)(x)
	...

Note that Linus merged bedd30d986a05e32dc3eab874e4b9ed8a38058bb already
so that problem should exist on mainline now, too.  (I havn't tried
though.)

I cannot reproduce the failure I saw for RT2870 at the moment.

Best regards
Uwe

-- 
Pengutronix e.K.                              | Uwe Kleine-König            |
Industrial Linux Solutions                    | http://www.pengutronix.de/  |

Re: [PATCH -rt] disable building all staging drivers

[Greg KH]

On Mon, Jun 15, 2009 at 01:00:44AM +0200, Uwe Kleine-König wrote:
> Hi Greg,
> 
> On Sun, Jun 14, 2009 at 04:28:02AM -0700, Greg KH wrote:
> > On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > > I know that at least COMEDI and RT2870 are broken with -rt,
> > 
> > I didn't know this, has anyone sent me a bug report?  What in -rt causes
> > these drivers to break?
> > 
> > > but Thomas requested disabling all of staging.  He's the maintainer.
> > 
> > This seems like a patch to ensure that the staging drivers never get a
> > chance to be fixed for any potential -rt issues.  How about just sending
> > me bug reports instead?
> COMEDI has:
> 
>   CC [M]  drivers/staging/comedi/comedi_fops.o
> In file included from drivers/staging/comedi/comedidev.h:40,
>                  from drivers/staging/comedi/comedi_fops.c:45:
> drivers/staging/comedi/interrupt.h:27: error: conflicting types for ‘irqreturn_t’
> include/linux/irqreturn.h:14: error: previous declaration of ‘irqreturn_t’ was here
> In file included from drivers/staging/comedi/comedidev.h:40,
>                  from drivers/staging/comedi/comedi_fops.c:45:
> drivers/staging/comedi/interrupt.h:30:1: warning: "IRQ_RETVAL" redefined
> In file included from include/linux/interrupt.h:10,
>                  from drivers/staging/comedi/interrupt.h:22,
>                  from drivers/staging/comedi/comedidev.h:40,
>                  from drivers/staging/comedi/comedi_fops.c:45:
> include/linux/irqreturn.h:15:1: warning: this is the location of the previous definition
> make[3]: *** [drivers/staging/comedi/comedi_fops.o] Error 1
> make[2]: *** [drivers/staging/comedi] Error 2
> make[1]: *** [drivers/staging] Error 2
> make: *** [drivers] Error 2

This should be resolved with the linux-next tree, right?  There are a
few hundred staging patches in there that fix a lot of issues.  They
will all go to Linus in a day or so.

thanks,

greg k-h

Re: [PATCH -rt] disable building all staging drivers

[Uwe Kleine-König]

Hi Greg,

On Mon, Jun 15, 2009 at 01:00:44AM +0200, Uwe Kleine-König wrote:
> On Sun, Jun 14, 2009 at 04:28:02AM -0700, Greg KH wrote:
> > On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > > I know that at least COMEDI and RT2870 are broken with -rt,
> > 
> > I didn't know this, has anyone sent me a bug report?  What in -rt causes
> > these drivers to break?
> > 
> > > but Thomas requested disabling all of staging.  He's the maintainer.
> > 
> > This seems like a patch to ensure that the staging drivers never get a
> > chance to be fixed for any potential -rt issues.  How about just sending
> > me bug reports instead?
> I cannot reproduce the failure I saw for RT2870 at the moment.
OK, hit again:

17:02:43 E: ERROR: "there_is_no_init_MUTEX_LOCKED_for_RT_semaphores" [drivers/staging/rt2870/rt2870sta.ko] undefined!

Best regards
Uwe

-- 
Pengutronix e.K.                              | Uwe Kleine-König            |
Industrial Linux Solutions                    | http://www.pengutronix.de/  |

Re: [PATCH -rt] disable building all staging drivers

[Greg KH]

On Mon, Jun 22, 2009 at 05:15:15PM +0200, Uwe Kleine-König wrote:
> Hi Greg,
> 
> On Mon, Jun 15, 2009 at 01:00:44AM +0200, Uwe Kleine-König wrote:
> > On Sun, Jun 14, 2009 at 04:28:02AM -0700, Greg KH wrote:
> > > On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > > > I know that at least COMEDI and RT2870 are broken with -rt,
> > > 
> > > I didn't know this, has anyone sent me a bug report?  What in -rt causes
> > > these drivers to break?
> > > 
> > > > but Thomas requested disabling all of staging.  He's the maintainer.
> > > 
> > > This seems like a patch to ensure that the staging drivers never get a
> > > chance to be fixed for any potential -rt issues.  How about just sending
> > > me bug reports instead?
> > I cannot reproduce the failure I saw for RT2870 at the moment.
> OK, hit again:
> 
> 17:02:43 E: ERROR: "there_is_no_init_MUTEX_LOCKED_for_RT_semaphores" [drivers/staging/rt2870/rt2870sta.ko] undefined!

What needs to be done in the rt2870 driver to resolve this issue?

thanks,

greg k-h

Re: [PATCH -rt] disable building all staging drivers

[Daniel Walker]

On Mon, 2009-06-22 at 10:19 -0700, Greg KH wrote:
> On Mon, Jun 22, 2009 at 05:15:15PM +0200, Uwe Kleine-König wrote:
> > Hi Greg,
> > 
> > On Mon, Jun 15, 2009 at 01:00:44AM +0200, Uwe Kleine-König wrote:
> > > On Sun, Jun 14, 2009 at 04:28:02AM -0700, Greg KH wrote:
> > > > On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > > > > I know that at least COMEDI and RT2870 are broken with -rt,
> > > > 
> > > > I didn't know this, has anyone sent me a bug report?  What in -rt causes
> > > > these drivers to break?
> > > > 
> > > > > but Thomas requested disabling all of staging.  He's the maintainer.
> > > > 
> > > > This seems like a patch to ensure that the staging drivers never get a
> > > > chance to be fixed for any potential -rt issues.  How about just sending
> > > > me bug reports instead?
> > > I cannot reproduce the failure I saw for RT2870 at the moment.
> > OK, hit again:
> > 
> > 17:02:43 E: ERROR: "there_is_no_init_MUTEX_LOCKED_for_RT_semaphores" [drivers/staging/rt2870/rt2870sta.ko] undefined!
> 
> What needs to be done in the rt2870 driver to resolve this issue?

The semaphore needs to be converted to a completion. Or the code needs
to be re-worked so it can use a mutex.

Daniel

Re: [PATCH -rt] disable building all staging drivers

[Greg KH]

On Mon, Jun 22, 2009 at 11:06:04AM -0700, Daniel Walker wrote:
> On Mon, 2009-06-22 at 10:19 -0700, Greg KH wrote:
> > On Mon, Jun 22, 2009 at 05:15:15PM +0200, Uwe Kleine-König wrote:
> > > Hi Greg,
> > > 
> > > On Mon, Jun 15, 2009 at 01:00:44AM +0200, Uwe Kleine-König wrote:
> > > > On Sun, Jun 14, 2009 at 04:28:02AM -0700, Greg KH wrote:
> > > > > On Sat, Jun 13, 2009 at 11:38:38PM +0200, Uwe Kleine-König wrote:
> > > > > > I know that at least COMEDI and RT2870 are broken with -rt,
> > > > > 
> > > > > I didn't know this, has anyone sent me a bug report?  What in -rt causes
> > > > > these drivers to break?
> > > > > 
> > > > > > but Thomas requested disabling all of staging.  He's the maintainer.
> > > > > 
> > > > > This seems like a patch to ensure that the staging drivers never get a
> > > > > chance to be fixed for any potential -rt issues.  How about just sending
> > > > > me bug reports instead?
> > > > I cannot reproduce the failure I saw for RT2870 at the moment.
> > > OK, hit again:
> > > 
> > > 17:02:43 E: ERROR: "there_is_no_init_MUTEX_LOCKED_for_RT_semaphores" [drivers/staging/rt2870/rt2870sta.ko] undefined!
> > 
> > What needs to be done in the rt2870 driver to resolve this issue?
> 
> The semaphore needs to be converted to a completion. Or the code needs
> to be re-worked so it can use a mutex.

Patches gladly accepted :)

thanks,

greg k-h

Re: [PATCH -rt] disable building all staging drivers

[Thomas Gleixner]

On Mon, 22 Jun 2009, Greg KH wrote:
> > 17:02:43 E: ERROR: "there_is_no_init_MUTEX_LOCKED_for_RT_semaphores" [drivers/staging/rt2870/rt2870sta.ko] undefined!
> 
> What needs to be done in the rt2870 driver to resolve this issue?

It looks like the semaphores are used to wait for an event. There are
actually 3 of those:

	 init_MUTEX_LOCKED(&(pAd->mlme_semaphore));
	 init_MUTEX_LOCKED(&(pAd->RTUSBCmd_semaphore));
	 init_MUTEX_LOCKED(&(pAd->RTUSBTimer_semaphore));

All are used in kernel threads which wait with down_interruptible() on
them. The up() calls are in some completely non obvious places.

I had a look into the timer one. The semaphore is the least thing you
need to worry about there. The open coded timer list management and
the thread which does timer polling for some obscure reason are the
things which need to be fixed in the first place. Once this is cleaned
up the semaphore will be gone anyway.

But, after looking into that code I really have to ask you whether you
are serious about fixing it instead of rewriting it from scratch and
just keep the code as is for reference to the hardware details.

Thanks,

	tglx

Re: [PATCH -rt] disable building all staging drivers

[Greg KH]

On Tue, Jun 23, 2009 at 01:24:19PM +0200, Thomas Gleixner wrote:
> On Mon, 22 Jun 2009, Greg KH wrote:
> > > 17:02:43 E: ERROR: "there_is_no_init_MUTEX_LOCKED_for_RT_semaphores" [drivers/staging/rt2870/rt2870sta.ko] undefined!
> > 
> > What needs to be done in the rt2870 driver to resolve this issue?
> 
> It looks like the semaphores are used to wait for an event. There are
> actually 3 of those:
> 
> 	 init_MUTEX_LOCKED(&(pAd->mlme_semaphore));
> 	 init_MUTEX_LOCKED(&(pAd->RTUSBCmd_semaphore));
> 	 init_MUTEX_LOCKED(&(pAd->RTUSBTimer_semaphore));
> 
> All are used in kernel threads which wait with down_interruptible() on
> them. The up() calls are in some completely non obvious places.
> 
> I had a look into the timer one. The semaphore is the least thing you
> need to worry about there. The open coded timer list management and
> the thread which does timer polling for some obscure reason are the
> things which need to be fixed in the first place. Once this is cleaned
> up the semaphore will be gone anyway.
> 
> But, after looking into that code I really have to ask you whether you
> are serious about fixing it instead of rewriting it from scratch and
> just keep the code as is for reference to the hardware details.

That is exactly what the code is there for.  There are other groups
rewriting these drivers from scratch, and hopefully they will be done in
a kernel release or two and then these drivers can go away.

If, in the mean time, you don't want them to be built for -rt kernels,
just send me a Kconfig patch for the individual drivers, not for the
whole of the staging tree.

thanks,

greg k-h

Re: [PATCH -rt] disable building all staging drivers

[Thomas Gleixner]

On Tue, 23 Jun 2009, Greg KH wrote:
> > But, after looking into that code I really have to ask you whether you
> > are serious about fixing it instead of rewriting it from scratch and
> > just keep the code as is for reference to the hardware details.
> 
> That is exactly what the code is there for.  There are other groups
> rewriting these drivers from scratch, and hopefully they will be done in
> a kernel release or two and then these drivers can go away.

Ok. I just wondered :)

> If, in the mean time, you don't want them to be built for -rt kernels,
> just send me a Kconfig patch for the individual drivers, not for the
> whole of the staging tree.

Sure, I can mark them individually broken on RT in the -rt patch
itself if we stomp over such issues.

Thanks,

	tglx

[RT + staging] buildfailures

[Uwe Kleine-König]

Hello,

I just tried to compile 2.6.31-rc8-rt9 on amd64 and got the following
compiler errors in drivers/staging:

 - drivers/staging/comedi:

	  CC [M]  drivers/staging/comedi/drivers/quatech_daqp_cs.o
	drivers/staging/comedi/drivers/quatech_daqp_cs.c: In function ‘daqp_interrupt’:
	drivers/staging/comedi/drivers/quatech_daqp_cs.c:308: error: implicit declaration of function ‘up’
	drivers/staging/comedi/drivers/quatech_daqp_cs.c: In function ‘daqp_ai_insn_read’:
	drivers/staging/comedi/drivers/quatech_daqp_cs.c:421: error: implicit declaration of function ‘sema_init’
	drivers/staging/comedi/drivers/quatech_daqp_cs.c:434: error: implicit declaration of function ‘down_interruptible’

   Possible fix: #include <linux/semaphore.h> in
   drivers/staging/comedi/comedidev.h

 - drivers/staging/rt2870:
	  CC [M]  drivers/staging/rt2870/common/2870_rtmp_init.o
	drivers/staging/rt2870/common/2870_rtmp_init.c: In function ‘CreateThreads’:
	drivers/staging/rt2870/common/2870_rtmp_init.c:754: error: implicit declaration of function ‘semaphore_init_locked’

   with PREEMPT_RT semaphore_init_locked doesn't exist.  The driver
   needs to be converted to use mutexes instead.

   Hhhhm, tglx introduced these in

	041e313 (staging: Bulk convert the semaphore mess)

   !?

   drivers/staging/rt3070/common/2870_rtmp_init.c includes
   ../../rt2870/common/2870_rtmp_init.c, so RT3070 has the same problem.

Just to let you know ...

Best regards
Uwe

-- 
Pengutronix e.K.                              | Uwe Kleine-König            |
Industrial Linux Solutions                    | http://www.pengutronix.de/  |

Re: [RT + staging] buildfailures

[Greg KH]

On Mon, Sep 07, 2009 at 06:20:29PM +0200, Uwe Kleine-König wrote:
> Hello,
> 
> I just tried to compile 2.6.31-rc8-rt9 on amd64 and got the following
> compiler errors in drivers/staging:
> 
>  - drivers/staging/comedi:
> 
> 	  CC [M]  drivers/staging/comedi/drivers/quatech_daqp_cs.o
> 	drivers/staging/comedi/drivers/quatech_daqp_cs.c: In function ‘daqp_interrupt’:
> 	drivers/staging/comedi/drivers/quatech_daqp_cs.c:308: error: implicit declaration of function ‘up’
> 	drivers/staging/comedi/drivers/quatech_daqp_cs.c: In function ‘daqp_ai_insn_read’:
> 	drivers/staging/comedi/drivers/quatech_daqp_cs.c:421: error: implicit declaration of function ‘sema_init’
> 	drivers/staging/comedi/drivers/quatech_daqp_cs.c:434: error: implicit declaration of function ‘down_interruptible’
> 
>    Possible fix: #include <linux/semaphore.h> in
>    drivers/staging/comedi/comedidev.h

Patches welcome :)

>  - drivers/staging/rt2870:
> 	  CC [M]  drivers/staging/rt2870/common/2870_rtmp_init.o
> 	drivers/staging/rt2870/common/2870_rtmp_init.c: In function ‘CreateThreads’:
> 	drivers/staging/rt2870/common/2870_rtmp_init.c:754: error: implicit declaration of function ‘semaphore_init_locked’
> 
>    with PREEMPT_RT semaphore_init_locked doesn't exist.  The driver
>    needs to be converted to use mutexes instead.
> 
>    Hhhhm, tglx introduced these in
> 
> 	041e313 (staging: Bulk convert the semaphore mess)
> 
>    !?
> 
>    drivers/staging/rt3070/common/2870_rtmp_init.c includes
>    ../../rt2870/common/2870_rtmp_init.c, so RT3070 has the same problem.

Then he might want to resolve them, as this code has changed a _lot_ in
the linux-next tree :)

thanks,

greg k-h