Re: schedule() vs softirqs

[Keir Fraser <keir@xensource.com>]

On 15/12/06 19:09, "Hollis Blanchard" <hollisb@us.ibm.com> wrote:

> Ah OK, I see now how x86 is doing that. I don't think that code flow
> really makes sense: why would you jump out of do_softirq() into assembly
> just to call do_softirq() again?

Well, that's the way it works out on x86. It is a bit odd, but it works and
is unlikely to affect performance. I think returning from schedule() would
have its own problems (e.g., context switch from idle domain to guest would
return to the idle loop, which we'd need explicit code to bail from, etc).

> Also, that doesn't solve the lazy register saving problem.

I assume this is a PPC-specific issue?

> However, I think I see how we can implement our desired context_switch()
> scheme in arch-specific code. The context_switch() call in schedule()
> will return, so please don't add a BUG() after that. :)

We already support this mode of operation for IA64 which always returns from
schedule().

 -- Keir