Date: Tue, 27 Jul 2010 15:50:33 +0300 From: Alexander Motin <mav@FreeBSD.org> To: Poul-Henning Kamp <phk@phk.freebsd.dk> Cc: freebsd-geom@FreeBSD.org Subject: Re: Hyperactive g_event thread Message-ID: <4C4ED619.7050305@FreeBSD.org> In-Reply-To: <24822.1280220363@critter.freebsd.dk> References: <24822.1280220363@critter.freebsd.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
Poul-Henning Kamp wrote: > In message <4C4E919C.2060009@FreeBSD.org>, Alexander Motin writes: > >> The only reason for doing it I >> see in imperfect locking of g_run_events() loop and respective >> tsleep()/wakeup(). So timeout there may be needed for collection >> potentially lost wakeup() events. > > Back when GEOM was written, the kernels SMP facilites were rather > rudimentary and in a few cases downright buggy, so workarounds > like that timeout were employed. > > They can hopefully safely be removed now. I have already removed alike timeouts on up/down threads and it indeed was safe there. But are you really sure about this specific case? Cause I'm not. Up/down threads using msleep and checking lack of work after dropping/grabbing lock. Event thread instead does several tasks, drops lock few times between them and uses tsleep(). I would say there should be a bunch of race conditions. -- Alexander Motin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C4ED619.7050305>