Re: [patch] softirq performance fixes, cleanups, 2.4.10.

[Andrea Arcangeli <andrea@suse.de>]

On Fri, Sep 28, 2001 at 08:18:20PM +0400, kuznet@ms2.inr.ac.ru wrote:
> Hello!
> 
> >  - removed 'mask' handling from do_softirq() - it's unnecessery due to the
> >    restarts. this further simplifies the code.
> 
> Ingo, but this means that only the first softirq is handled.
> "mask" implements round-robin and this is necessary.

he's allowing to repeat the loop more than once to hide it, to do the
"mask" with repetition correctly we'd need a per-softirq counter, not
just a bitmask so it wouldn't be handy to allocate on the stack, but
it's nothing unfixable.

However I also preferred the previous behaviour, I think it was much
nicer for general purpose (non specweb99 gigabit like scenarios).

> >  - '[ksoftirqd_CPU0]' is confusing on UP systems, changed it to
> >    '[ksoftirqd]' instead.
> 
> It is useless to argue about preferences, but universal naming scheme
> looks as less confusing yet. :-)

Agreed.

> Generally, I dislike this patch. It is utterly ugly.

I also dislike it overall but I can see why it improves performance, and
the deschedule thing makes sense for the flooding case.

I would be very confortable in only merging the deschedule part and this
is why I asked Ingo if he could measure the difference.

Andrea