Re: CONFIG_PREEMPT and server workloads

[Chris Mason]

On Fri, 2004-03-19 at 06:23, Takashi Iwai wrote:
> At Thu, 18 Mar 2004 22:07:41 -0500,
> Eric St-Laurent wrote:
> > 
> > On Thu, 2004-03-18 at 10:28, Takashi Iwai wrote:
> > 
> > > in my case with reiserfs, i got 0.4ms RT-latency with my test suite
> > > (with athlon 2200+).
> > > 
> > > there is another point to be fixed in the reiserfs journal
> > > transaction.  then you'll get 0.1ms RT-latency without preemption.
> > 
> > Are you talking about the following patch recently merged in Linus tree?
> > 
> > [PATCH] resierfs: scheduling latency improvements
> > http://linus.bkbits.net:8080/linux-2.5/cset@40571b49jtE7PzOtsXjBx65-GoDijg
> 
> i've tested the suse kernel, so the patch above was already in it.
> i'm not sure whether mm tree already includes the all relevant
> fixes... Chris?
> 
The whole reiserfs patcheset from -suse is now at:

ftp.suse.com/pub/people/mason/patches/data-logging/experimental/2.6.4

The mm tree does already have the most important reiserfs latency fix,
which is named reiserfs-lock-lat.  Andrew sent along to linus as well,
so it should be in mainline.

> what i wrote above about is loops in do_journal_end().  but i can't
> tell you surely unless i test the mm kernel again.
> 
The spot you found in do_journal_end still needs the cond_resched.  The
patches above don't include that one yet (I'll upload it today).

-chris


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/