Date: Tue, 27 Jul 2010 20:27:56 +0300 From: Alexander Motin <mav@FreeBSD.org> To: =?UTF-8?B?TWFyaXVzIE7DvG5uZXJpY2g=?= <marius@nuenneri.ch> Cc: Poul-Henning Kamp <phk@phk.freebsd.dk>, freebsd-geom@freebsd.org Subject: Re: Hyperactive g_event thread Message-ID: <4C4F171C.9010106@FreeBSD.org> In-Reply-To: <AANLkTikvV4oymBBA%2B_0zbzd_edS8dRfqqJRODE0989%2Bn@mail.gmail.com> References: <4C4ED619.7050305@FreeBSD.org> <27237.1280241532@critter.freebsd.dk> <AANLkTi=uRPV2T0=t_1s=Jc4PmBtai=__HqhCtYpiDdTQ@mail.gmail.com> <AANLkTikvV4oymBBA%2B_0zbzd_edS8dRfqqJRODE0989%2Bn@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Marius Nünnerich wrote: > On Tue, Jul 27, 2010 at 18:07, Marius Nünnerich <marius@nuenneri.ch> wrote: >> On Tue, Jul 27, 2010 at 16:38, Poul-Henning Kamp <phk@phk.freebsd.dk> wrote: >>> In message <4C4ED619.7050305@FreeBSD.org>, Alexander Motin writes: >>>> Poul-Henning Kamp wrote: >>>> 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. >>> Quite likely, I didn't say it would be a trivial thing to remove >>> that workaround :-) >> >> I was running with a patch that removed the timeout for a while like 2 >> years ago. Albeit not with high load. Worked fine at that time, I will >> search for the patch when I'm back home later today. >> > Here it is: > http://lists.freebsd.org/pipermail/freebsd-geom/2008-December/003200.html > The mail is quite old now, I don't know if the patch still applies. > Will check that if I can find the time. I would be happy to see this > committed after all this time :) Yes, I was thinking about something like that. Patch mostly applies now and even builds. May be I would just put there sx_sleep() instead if msleep(), as topology_lock is sx. Could somebody to review this and tell how correct is to not drop topology lock between events handling and could there be places where g_wait_event woken without holding topology_lock? -- Alexander Motin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C4F171C.9010106>