Date: Tue, 03 Nov 2009 22:28:00 +0000 From: Tom Judge <tom@tomjudge.com> To: Gavin Atkinson <gavin.atkinson@ury.york.ac.uk> Cc: Pieter de Goeje <pieter@degoeje.nl>, Weldon S Godfrey 3 <weldon@excelsusphoto.com>, freebsd-current@freebsd.org, Ivan Voras <ivoras@freebsd.org> Subject: Re: FreeBSD 8.0 - network stack crashes? Message-ID: <4AF0AE70.7030004@tomjudge.com> In-Reply-To: <20091103221739.E75071@ury.york.ac.uk> References: <alpine.BSF.2.00.0911020747560.80499@emmett.excelsus.com> <hcnt4c$12i$1@ger.gmane.org> <alpine.BSF.2.00.0911021952220.1264@emmett.excelsus.com> <200911030937.11619.pieter@degoeje.nl> <alpine.BSF.2.00.0911030832320.36102@emmett.excelsus.com> <4AF05177.7030705@tomjudge.com> <4AF06017.6000505@tomjudge.com> <20091103221739.E75071@ury.york.ac.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
Gavin Atkinson wrote: > On Tue, 3 Nov 2009, Tom Judge wrote: >> Tom Judge wrote: >>> Weldon S Godfrey 3 wrote: >>>> we are using onboard NICs on the Dell using the bce driver. We did >>>> try several times to see if using an intel PCIexpress card using the >>>> em driver, and we had the same symptoms. >>>> >>>> Could the bce driver have the same leak? >>> >>> The bce driver does not have a memory leak, it does however have a >>> bug which causes memory fragmentation leading to denied mbuf allocation. >>> >>> There is a work around for this in current, you can get the patch >>> like this: >>> >>> http://svn.freebsd.org/viewvc/base/head/ >>> >> That should be: >> >> svn diff -r 198319:198320 http://svn.freebsd.org/base/head >> >>> You need to put >>> >>> options BCE_JUMBO_HDRSPLIT >>> >>> In your kernel to enable the work arround. > > Unless I'm missing something, these seem like they may be totally > different bugs. The symptoms that Weldon S Godfrey has show the number > of "mbuf clusters in use" rising to the point at which the limit is > reached, whereas the thread on -stable where BCE_JUMBO_HDRSPLIT is > recommended has the symptom of "requests for 9k jumbo clusters denied" > increasing and the mbuf clusters not being anywhere near to the maximum. > > So, I think there may be some confusion here. Jumbo frames are not in use what what I have read in the thread however there are denied requests for mbuf+clusters, so this could be the issue in the bce driver with standrad size frames. See: 0/201276/90662 requests for mbufs denied (mbufs/clusters/mbuf+clusters) I guess that there are 2 issues at work here, adding the bce patch should not cause any problems but may resolve the issue when using the bce driver. Is it not worth a try? Tom
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4AF0AE70.7030004>