Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 20 Feb 2013 07:31:07 -0800
From:      Marcel Moolenaar <marcel@xcllnt.net>
To:        mexas@bristol.ac.uk
Cc:        marcel@FreeBSD.org, freebsd-ia64@FreeBSD.org
Subject:   Re: ia64/147501: [ia64] options PREEMPTION causes instability
Message-ID:  <E57126C6-D82B-4FCA-B7AD-F635255E242D@xcllnt.net>
In-Reply-To: <201302200916.r1K9GV8L029432@mech-cluster241.men.bris.ac.uk>
References:  <201302200916.r1K9GV8L029432@mech-cluster241.men.bris.ac.uk>

next in thread | previous in thread | raw e-mail | index | archive | help

On Feb 20, 2013, at 1:16 AM, Anton Shterenlikht <mexas@bristol.ac.uk> wrote:

> 	Date: Wed, 20 Feb 2013 04:03:21 GMT
> 	From: marcel@FreeBSD.org
> 	Subject: Re: ia64/147501: [ia64] options PREEMPTION causes instability
> 
> 	Synopsis: [ia64] options PREEMPTION causes instability
> 
> 	State-Changed-From-To: patched->closed
> 	State-Changed-By: marcel
> 	State-Changed-When: Wed Feb 20 04:02:27 UTC 2013
> 	State-Changed-Why: 
> 	Fix r246890 & r246715 have been MFC'd.
> 
> 	http://www.freebsd.org/cgi/query-pr.cgi?pr=147501
> 
> So do you recommend including this option in
> the kernel now?

Yes. I've changed all machines to include PREEMPTION.
I'll change the GENERIC kernel shortly.

> What about IPI_PREEMPTION?

IPI_PREEMPTION is only needed for the 4BSD scheduler.
It's not applicable with you have the ULE scheduler.
If you use the 4BSD scheduler, I *think* it's safe to
enable IPI_PREEMPTION as well, but have not tested
this myself. The functionality is shared with that of
the ULE scheduler (i.e. sending IPI_PREEMPT to other
CPUs).

Only i386 and amd64 has LINT built with IPI_PREEMPTION.
No other architecture has it in their NOTES.

HTH,

-- 
Marcel Moolenaar
marcel@xcllnt.net





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E57126C6-D82B-4FCA-B7AD-F635255E242D>