Date: Mon, 8 Aug 2005 17:30:36 +0100 (BST) From: Robert Watson <rwatson@FreeBSD.org> To: Rong-En Fan <grafan@gmail.com> Cc: stable@freebsd.org, A.S.Usov@kvi.nl Subject: Re: panic: sbflush_locked on 5.4-p5/i386 Message-ID: <20050808171707.E34546@fledge.watson.org> In-Reply-To: <6eb82e050808085132807d80@mail.gmail.com> References: <6eb82e0507241001615f7490@mail.gmail.com> <20050729001511.O60522@fledge.watson.org> <6eb82e050808085132807d80@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 8 Aug 2005, Rong-En Fan wrote: > After upgrading to 5-STABLE (about Aug 6), it works very good. With > mpsafenet=1, it can work more than one day without panic. For 5.4-p5, it > will panic at most half day or so. This bug seems fixed after 5.4 is > released. I'll keep watching this machine. Will let you know if it still > have similar panics ;-) If you can confirm that the problem really is fixed in -STABLE, there are two sets of changes that may be relevant: - There are a number of SACK-related bug fixes in RELENG_5 that are not in the RELENG_5_4 branch. - There are a number of TCP/inpcb-related locking fixes (one in particular for IPv6) that are in RELENG_5 but not RELENG_5_4. We can then try and figure out which change it is that's helping, and get that merged to RELENG_5_4 as an errata item. If you have IPv6 traffic, there are some fixes I made a month or two ago that might be particularly relevant during connection tear-down. Robert N M Watson
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050808171707.E34546>