Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 12 Apr 2003 01:07:44 +0000
From:      Nicolas Souchu <nsouch@free.fr>
To:        Bruce Evans <bde@zeta.org.au>
Cc:        hackers@FreeBSD.org
Subject:   Re: In case anybody's interested...
Message-ID:  <20030412010744.B99524@armor.free.fr>
In-Reply-To: <20030409162415.T2279@gamplex.bde.org>; from bde@zeta.org.au on Wed, Apr 09, 2003 at 04:41:29PM %2B1000
References:  <1049821320.14809.8.camel@builder02.qubesoft.com> <20030409021112.5FC1A2A7EA@canning.wemm.org> <20030409034708.GD30960@elvis.mu.org> <20030409162415.T2279@gamplex.bde.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Apr 09, 2003 at 04:41:29PM +1000, Bruce Evans wrote:
[...]
> Testing of lpt while debugging this showed other nice races: lpt
> (actually ppc) detaches and attaches the interrupt for every i/o.  It

Indeed for interrupt sharing. Actually, only lpt is used as a ppbus device.

> apparently doesn't wait long enough for the last (non-stray) interrupt
> from the previous i/o to arrive.  The interrupt sometimes arrives in
> the window between the detach and the attach.  It is then recorded as
> "stray".  It is stray all right, but "stray" was meant to mean stray
> in the hardware and not stray due to software bugs.

The interrupt certainly arrive between the the TEARDOWN and SETUP calls
in ppc_teardown_intr() of isa/ppc.c

-- 
Nicholas Souchu - nsouch@free.fr - nsouch@FreeBSD.org



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030412010744.B99524>