Date: Fri, 16 Apr 2004 14:08:37 -0400 From: John Baldwin <jhb@FreeBSD.org> To: Robert Watson <rwatson@FreeBSD.org> Cc: cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/dev/random randomdev.c Message-ID: <200404161408.37851.jhb@FreeBSD.org> In-Reply-To: <Pine.NEB.3.96L.1040415145240.95950E-100000@fledge.watson.org> References: <Pine.NEB.3.96L.1040415145240.95950E-100000@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday 15 April 2004 02:54 pm, Robert Watson wrote: > On Thu, 15 Apr 2004, John Baldwin wrote: > > On Thursday 15 April 2004 03:32 am, Mike Silbersack wrote: > > > On Tue, 13 Apr 2004, John Baldwin wrote: > > > > You should drop the lock before uiomove(). If the uio can possibly > > > > reference user memory, then uiomove() will do a copyout() which can > > > > fault and thus sleep if the page is in swap. > > > > > > > > -- > > > > John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ > > > > > > I'm going to have a similar problem in sendfile when I try to > > > de-giantify it, and I'm sure read / write will have issues as well... > > > will WITNESS catch this potential problem? > > > > Not currently. One could add a WITNESS_WARN() check to uiomove() fairly > > easily though to force witness to catch this. > > We also need to generate sleep warnings it M_TRYWAIT is passed to the mbuf > allocator, which I believe we currently don't do. Also, did we ever > re-add the sleep warning to the process tracing code? To STOPEVENT()? No.. -- John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" = http://www.FreeBSD.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200404161408.37851.jhb>