Re: [PATCH](s): Use spinlocks instead of STI/CLI in SoundBlaster

2001-01-27 Thread Chris Rankin
> > --- linux-2.4.0/drivers/sound/sb.h.orig Fri Jan 26 13:57:40 2001 > > +++ linux-2.4.0/drivers/sound/sb.h Fri Jan 26 13:58:42 2001 > > @@ -137,6 +137,8 @@ > >void (*midi_input_intr) (int dev, unsigned char data); > >void *midi_irq_cookie; /* IRQ cookie for the

Re: [PATCH](s): Use spinlocks instead of STI/CLI in SoundBlaster

2001-01-27 Thread Pavel Machek
Hi! > I hear on the grapevine that 2.4 kernel modules should use spinlocks > in preference to cli() and sti(). Well I'm not sure how big a win it > is, particularly on a UP machine, but here's a patch for the > SoundBlaster. I've added a spinlock_t to the "struct b_devc" so that > multiple SoundB

Re: [PATCH](s): Use spinlocks instead of STI/CLI in SoundBlaster

2001-01-26 Thread Chris Rankin
> spin_lock_irqsave() and save_flags()+cli() are identical on > uniprocessor builds. OK, so a UP-build proves nothing ... > cli() is quite putrid on SMP and should be shot. This much I understand ... > You can test your patch on uniprocessor hardware - just build > an SMP kernel and run it. I

[PATCH](s): Use spinlocks instead of STI/CLI in SoundBlaster

2001-01-25 Thread Chris Rankin
Hi, I hear on the grapevine that 2.4 kernel modules should use spinlocks in preference to cli() and sti(). Well I'm not sure how big a win it is, particularly on a UP machine, but here's a patch for the SoundBlaster. I've added a spinlock_t to the "struct b_devc" so that multiple SoundBlasters ea