Re: [PATCH 0/2 v3] SGI RTC: add clocksource/clockevent driver and generic timer vector

[Dimitri Sivanich <sivanich@sgi.com>]

On Thu, Nov 20, 2008 at 05:44:22PM -0800, H. Peter Anvin wrote:
> Ingo Molnar wrote:
> > * Dimitri Sivanich <sivanich@sgi.com> wrote:
> > 
> >> The following patches provide a driver for synchronized RTC 
> >> clocksource and clockevents for SGI systems, as well as a generic 
> >> timer system interrupt.
> >>
> >> With these patches, a module can be installed that registers the 
> >> system-wide synchronized RTC clocksource and timers as both a 
> >> clocksource and clockevents device running in high resolution mode.
> >>
> >> [PATCH 1/2 v3] SGI RTC: add clocksource driver
> >> [PATCH 2/2 v3] SGI RTC: add generic timer system interrupt
> > 
> > Looks very clean and well-done to me.
> > 
> > I had to take a good look at the rtc_timer_head->expires[] construct - 
> > but i guess that's the best approach, as the max number of entries is 
> > hard to judge at build time. (and we wont get any real limit 
> > protection from gcc anyway)
> > 
> > Thomas, any objections?
> 
> I have *extremely* serious reservations about reserving even more
> hardware vectors for SGI only.  This affects all systems, and quite
> frankly should not be necessary at all.
> 
> The SGI UV people have pushed this at a number of points in the past,
> and we have told them to use an irqchip instead.  This patch tries to
> allocate yet another reserved vector, instead.

I labeled the system vector as generic to indicate that this is available
to other platforms.  See my other email to you for the reasons why I did
not use an irq instead.