Subject: Re: kern_proc.c patch
To: Jason Thorpe <email@example.com>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
Date: 01/22/2000 14:08:27
In message <200001222018.MAA07131@lestat.nas.nasa.gov>Jason Thorpe writes
>By popular demand, here is the the patch to kern_proc.c.
> -- Jason R. Thorpe <firstname.lastname@example.org>
Thanks for eposting that, Jason.
For anyone (like me) running a less-than-fresh -current: looks like
1.35 is OK, since it uses splstatclock(). Is that correct?
On a completely different note... Wearning an NTP time-vulture hat,
locking out clock interrupts is Really Vicious. Just out of
curiosity, what'd it take to _not_ do splclock() here? Deferring any
and all proclist (scheduler activity) called from within hardclock(),
so they are marked to be done, but dont get done until the hardclock
interrupt returns from splclock down to what it was before?