Re: status of the Altix mmtimer driver

2017-05-26 Thread Dimitri Sivanich
Note that the mmtimer struct can probably be removed from the k_itimer struct
in include/linux/posix-timers.h, just for completeness.

--- a/include/linux/posix-timers.h
+++ b/include/linux/posix-timers.h
@@ -73,12 +73,6 @@ struct k_itimer {
} real;
struct cpu_timer_list cpu;
struct {
-   unsigned int clock;
-   unsigned int node;
-   unsigned long incr;
-   unsigned long expires;
-   } mmtimer;
-   struct {
struct alarm alarmtimer;
ktime_t interval;
} alarm;

On Tue, May 23, 2017 at 07:36:01PM -0500, Dimitri Sivanich wrote:
> Since this is SN2 specific, this can be removed.
> 
> On Tue, May 23, 2017 at 05:27:16PM +0200, Thomas Gleixner wrote:
> > On Tue, 23 May 2017, Christoph Hellwig wrote:
> > 
> > Cc: + Mike
> > 
> > > Hi folks,
> > > 
> > > what's the status of the SGI support for the IA64 SN2-specific
> > > altic mmtimer (drivers/char/mmtimer.c)?  It's the only driver that
> > > register a k_lock outside the core kernel, so if we could get rid of
> > > it because the remaining IA64 user are all on somewhat old distros
> > > anyway it would make our life a whole lot easier.
> > > 


Re: status of the Altix mmtimer driver

2017-05-23 Thread Dimitri Sivanich
Since this is SN2 specific, this can be removed.

On Tue, May 23, 2017 at 05:27:16PM +0200, Thomas Gleixner wrote:
> On Tue, 23 May 2017, Christoph Hellwig wrote:
> 
> Cc: + Mike
> 
> > Hi folks,
> > 
> > what's the status of the SGI support for the IA64 SN2-specific
> > altic mmtimer (drivers/char/mmtimer.c)?  It's the only driver that
> > register a k_lock outside the core kernel, so if we could get rid of
> > it because the remaining IA64 user are all on somewhat old distros
> > anyway it would make our life a whole lot easier.
> > 


Re: status of the Altix mmtimer driver

2017-05-23 Thread Thomas Gleixner
On Tue, 23 May 2017, Christoph Hellwig wrote:

Cc: + Mike

> Hi folks,
> 
> what's the status of the SGI support for the IA64 SN2-specific
> altic mmtimer (drivers/char/mmtimer.c)?  It's the only driver that
> register a k_lock outside the core kernel, so if we could get rid of
> it because the remaining IA64 user are all on somewhat old distros
> anyway it would make our life a whole lot easier.
>