From mboxrd@z Thu Jan 1 00:00:00 1970 From: Gerd Knorr Subject: Re: bug: slab corruption (net backend?) Date: Mon, 11 Jul 2005 15:25:35 +0200 Message-ID: <20050711132535.GB13297@bytesex> References: <20050707094132.GA29339@bytesex> <20050707113514.GA30556@bytesex> <20050707122242.GA31229@bytesex> <87br5ehi15.fsf@bytesex.org> <42CD52F4.2010407@us.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <42CD52F4.2010407@us.ibm.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Nivedita Singhvi Cc: xen-devel@lists.xensource.com, Steven Hand List-Id: xen-devel@lists.xenproject.org > >As the tcpdump below shows the xen machine sends the same packet over > >and over again. I'd guess some kind of memory corruption which kills > >the packet checksum and makes eskarina drop the packet. Maybe just > >use-after-free as the slabdebug stuff will fill released memory blocks > >with some pattern. > > Can you disable tcp checksum offload (ethtool -K tx off) and see > if you can reproduce this problem? Never seen that again since I've turned off CONFIG_SLAB_DEBUG, which seems to be incompatible with the page mapping the xen backend drivers do. Gerd -- panic("it works"); /* avoid being flooded with debug messages */