Re: [PATCH] Realtime preempt support for PPC

[Frank Rowand]

Ingo Molnar wrote:
> * Frank Rowand <frowand@xxxxxxxxxx> wrote:
>
>
> > I have attached a patch to add realtime support for PowerPC (32 bit
> > only). [...]
>
>
> two comments:
>
> - why is us_to_tb needed? It just seems to add alot of unnecessary
>   clutter to the patch, while it's not used anywhere.

Sorry, the usage of us_to_tb got dropped from the patch when I moved
it from one of my development trees to another.  It gets used in
usecs_to_cycles().  The patch to add that is attached below.

Even with the usage of us_to_tb by usecs_to_cycles(), there is a lot
of added clutter for what is accomplished.  I considered a nasty hack
to try to derive the proper value from other sources, but adding
us_to_tb seemed a lot easier to understand.

This patch also adds the dreaded #ifdefs, so I suspect it will not be
the correct implementation long term.  I'm not sure if Manish's patch
included the ARM implementation of usecs_to_cycles() - it is a third
case in the #ifdef.

Performance instrumentation is also going to need mcount() and
_mcount(), which is currently a work in progress.

> - could you submit the drivers/net/ibm_emac/ibm_emac_core.c patch
>   upstream as well?

Yes, I will do that.


> otherwise it looks pretty clean and straightforward.
>
> 	Ingo


Source: MontaVista Software, Inc.
Signed-off-by: Frank Rowand <frowand@xxxxxxxxxx>

Index: linux-2.6.10/kernel/latency.c
===================================================================
--- linux-2.6.10.orig/kernel/latency.c
+++ linux-2.6.10/kernel/latency.c
@@ -192,7 +192,11 @@ static unsigned long notrace cycles_to_u

 static cycles_t notrace usecs_to_cycles(unsigned long delta)
 {
+#if defined CONFIG_X86
 	return (cycles_t) delta * (cycles_t) (cpu_khz/1000+1);
+#elif defined(CONFIG_PPC)
+	return delta * us_to_tb;
+#endif
 }

 #ifdef CONFIG_LATENCY_TRACE



Thanks,

-Frank (off to vacation for a couple days)
--
Frank Rowand <frank_rowand@xxxxxxxxxx>
MontaVista Software, Inc

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/