Date: Fri, 2 Mar 2007 18:53:37 -0500 From: Kris Kennaway <kris@obsecurity.org> To: Scott Long <scottl@samsco.org> Cc: cvs-src@freebsd.org, src-committers@freebsd.org, cvs-all@freebsd.org, John Baldwin <jhb@freebsd.org>, Kris Kennaway <kris@obsecurity.org> Subject: Re: cvs commit: src/sys/kern kern_intr.c Message-ID: <20070302235337.GB33387@xor.obsecurity.org> In-Reply-To: <45E8A117.6070501@samsco.org> References: <200703021701.l22H1k3M071057@repoman.freebsd.org> <200703021233.28346.jhb@freebsd.org> <45E870B9.4040403@samsco.org> <200703021502.23689.jhb@freebsd.org> <45E8A117.6070501@samsco.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--ZoaI/ZTpAVc4A5k6 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Fri, Mar 02, 2007 at 03:11:35PM -0700, Scott Long wrote: > John Baldwin wrote: > >On Friday 02 March 2007 13:45, Scott Long wrote: > >>John Baldwin wrote: > >>>On Friday 02 March 2007 12:28, Kris Kennaway wrote: > >>>>On Fri, Mar 02, 2007 at 10:13:54AM -0700, Scott Long wrote: > >>>>>John Baldwin wrote: > >>>>>>jhb 2007-03-02 17:01:45 UTC > >>>>>> > >>>>>> FreeBSD src repository > >>>>>> > >>>>>> Modified files: > >>>>>> sys/kern kern_intr.c=20 > >>>>>> Log: > >>>>>> - Don't do the interrupt storm protection stuff for software=20 > >>>>>> interrupt > >>>>>> handlers. > >>>>>> - Use pause() when throtting during an interrupt storm. > >>>>>>=20 > >>>>>> Reported by: kris (1) > >>>>>>=20 > >>>>>> Revision Changes Path > >>>>>> 1.142 +3 -2 src/sys/kern/kern_intr.c > >>>>>Actually, having this helped detect a problem with a CAM driver. > >>>>> > >>>>>Scott > >>>>I saw it under normal operation when doing a lot of loopback traffic > >>>>on an 8-core. > >>>Also, if it ever kicked in for softclock, the system would deadlock. > >>> > >>Can it be selectively enabled? > > > >We could do that, sure. The original intent was to prevent a storming= =20 > >level-triggered PCI interrupt to hang the machine though. It was never= =20 > >really intended for swi's and having it be on for swi's was an oversight= =20 > >on my part. > > >=20 > It could be argued that "storming" on the swi is also a bug. It=20 > certainly was for a misbehaving driver recently. In my case the only effect was to throttle legitimate traffic (postgresql via localhost tcp). That is definitely incorrect behaviour (it ruined the throughput and forced me to reboot to reset it). Kris --ZoaI/ZTpAVc4A5k6 Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (FreeBSD) iD8DBQFF6LkBWry0BWjoQKURAgwFAKDXl3a2obraQoHsPfu7uSALXdwEygCgxSLa vg6719VQnJ9Yd4iwkHhodMc= =havn -----END PGP SIGNATURE----- --ZoaI/ZTpAVc4A5k6--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070302235337.GB33387>