[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 3 Mar 2009 19:52:25 +0100
From: Ingo Molnar <mingo@...e.hu>
To: Dimitri Sivanich <sivanich@....com>
Cc: Thomas Gleixner <tglx@...utronix.de>,
"H. Peter Anvin" <hpa@...or.com>,
Andrew Morton <akpm@...ux-foundation.org>,
john stultz <johnstul@...ibm.com>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/6 v6] SGI RTC: add generic system interrupt
* Dimitri Sivanich <sivanich@....com> wrote:
> On Tue, Mar 03, 2009 at 04:34:33PM +0100, Ingo Molnar wrote:
> >
> > * Dimitri Sivanich <sivanich@....com> wrote:
> >
> > > This patch allocates a system interrupt vector for various
> > > platform specific uses.
> >
> > this is really ugly.
> >
> > Also, why are all these symbols exported? There's no need to
> > build the UV RTC driver as a module. It's either built-in or not
> > built-in - it's small enough.
>
> OK.
>
> >
> > this stuff:
> >
> > > +/* Function pointer for generic interrupt vector handling */
> > > +static void (*generic_interrupt_extension)(void);
> > > +static char generic_show_string[28];
> > > +static char generic_show_prefix[6];
> > > +
> > > +int is_generic_interrupt_registered()
> > > +{
> > > + if (generic_interrupt_extension)
> > > + return 1;
> > > + else
> > > + return 0;
> > > +}
> > > +
> > > +char *generic_interrupt_string(void)
> > > +{
> > > + return generic_show_string;
> > > +}
> > > +
> > > +char *generic_interrupt_prefix(void)
> > > +{
> > > + return generic_show_prefix;
> > > +}
> >
> > is SMP unsafe, etc. etc. - not something we should ever call
> > from a module. We just shouldnt do it in this form. What
> > necessiates it?
>
> I wanted a way to show specific strings when displaying irq statistics. We could show these in a more generic way I suppose.
>
> Any given platform should just be changing these once, hence the lack of need for locking.
>
> >
> > All we need is:
> >
> > > + /* generic IPI for platform specific use */
> > > + alloc_intr_gate(GENERIC_INTERRUPT_VECTOR, generic_interrupt);
> >
> > plus one trivial callback function - and then the UV platform
> > uses it for its own purpose. It's not like two platforms will be
> > running at once so there's no locking needed, etc.
>
> Right. But still register the callback function as I have it
> now?
i'd suggest to just override that global function pointer from
the UV detection routines.
this way we have it in a minimalistically generic fashion, but
with a minimum amount of fuss around it.
Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists