Date: Thu, 24 Feb 2000 20:39:47 -0500 (EST) From: David Gilbert <dgilbert@velocet.ca> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: Bruce Evans <bde@zeta.org.au>, David Gilbert <dgilbert@velocet.ca>, freebsd-current@FreeBSD.ORG Subject: Patch to try (was Re: Wierd AMD panics caused by VMWare?) Message-ID: <14517.56675.102218.408699@trooper.velocet.net> In-Reply-To: <200002250033.QAA43662@apollo.backplane.com> References: <Pine.BSF.4.21.0002250740030.7691-100000@alphplex.bde.org> <200002242105.NAA42378@apollo.backplane.com> <200002250033.QAA43662@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>>>>> "Matthew" == Matthew Dillon <dillon@apollo.backplane.com> writes: Matthew> Please try the following patch and tell me if the crashes Matthew> still occur. If this fixes the problem then I'm homing in on Matthew> the bug. Matthew> I am beginning to suspect that there is a case where a Matthew> pmap can get cleared without the tlb being flushed, causing Matthew> origpte to be 0 when the new pte is later created again at Matthew> the same spot. When the new page is allocated and added to Matthew> the pmap later on, it fails to flush the tlb entry in Matthew> pmap_enter() because it believes it does not have to. Matthew> This would explain why messing with the pmap_remove_all() Matthew> case does not do what we expect -- I suspect that routine is Matthew> not being called at all in certain cases. Matthew> I have not found the case where the pmap gets cleared Matthew> without being flushed yet but when I force the flush in Matthew> pmap_enter(), it seems to stop the crashes (but I can't be Matthew> 100% sure because I had to mess around with that file Matthew> descriptor program to get it to crash the first time). So... to get this straight, you'd like us to back out that other patch and apply this one? I have to be honest that the machine that we have is only doing this once a month because we kept adding RAM to it until it stopped exhibiting the symptoms often. If it helps, it appears to us that 3.0 didn't have the problem and 3.2 did. Dave. -- ============================================================================ |David Gilbert, Velocet Communications. | Two things can only be | |Mail: dgilbert@velocet.net | equal if and only if they | |http://www.velocet.net/~dgilbert | are precisely opposite. | =========================================================GLO================ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?14517.56675.102218.408699>