DragonFly kernel List (threaded) for 2004-09
Re: callout patch - make callouts per-cpu and MP safe.
On Mon, Sep 13, 2004 at 10:53:13PM -0700, Matthew Dillon wrote:
> This patch makes the kernel callout_*() interface per-cpu and MP safe.
> Individual callouts may also be registered as MP safe or not via a second
> argument to callout_reset() (using the same API a FreeBSD-5).
I never liked the additional argument in FreeBSD-5, but I guess it can't be
avoided for SMP safety. Does it imply that a MP-safe callout always runs on
the CPU it was initially registered on? Have you looked at the timeout API
from NetBSD/OpenBSD, they bind e.g. the function at init time. This might
be more appropiate on this case.
> I would appreciate more widespread testing before I commit it. I did
> some simple testing on a UP and SMP box but the API is used all over the
> system so prudence is required :-)
> fetch http://leaf.dragonflybsd.org/~dillon/callout01.patch
I'll check this patch, of course :)
> There is plenty of further work on the timeout_*() interface that can
> be done. I would dearly love to replace the old timeout() and
> untimeout() functions with the newer callout_init()/callout_reset() API,
> because that will free up 800KB+ of wired physical memory and allow us
> to remove a ton of cruft from kern_timeout.c. The old routines are used
> all over the kernel but any developer who wants to have a go at it please
> do! Submit patches one device at a time, though, and test that the
> kernel compiles and boots before you submit (The patches are also going
> to require very careful review once submitted).
> Any takers?
I've been working on this already for dev/netif whenever I did greater changes
to a driver. I've been working on other parts of the code base. If someone
wants to submit patches, I'd like to coordinate this effort off-list.
> Matthew Dillon