Date: Sat, 15 Apr 2000 09:26:39 -0700 From: Alfred Perlstein <bright@wintelcom.net> To: Brian Fundakowski Feldman <green@FreeBSD.ORG> Cc: Michael Reifenberger <root@nihil.plaut.de>, "dillon @ freebsd . org FreeBSD-Current" <current@FreeBSD.ORG>, alc@FreeBSD.ORG Subject: Re: panic: vm_object_shadow: source object has OBJ_ONEMAPPING set. Message-ID: <20000415092639.Q4381@fw.wintelcom.net> In-Reply-To: <Pine.BSF.4.21.0004150932460.16247-100000@green.dyndns.org>; from green@FreeBSD.ORG on Sat, Apr 15, 2000 at 09:38:34AM -0400 References: <Pine.BSF.4.21.0004131747510.492-300000@nihil.plaut.de> <Pine.BSF.4.21.0004150932460.16247-100000@green.dyndns.org>
next in thread | previous in thread | raw e-mail | index | archive | help
* Brian Fundakowski Feldman <green@FreeBSD.ORG> [000415 07:07] wrote: > On Thu, 13 Apr 2000, Michael Reifenberger wrote: > > > Hi, > > when using a linux java app (SAP PlatinGUI 46Cb2) I get the above panic. > > FreeBSD -current. Kernel+mods in sync. > > Linux from ports. Linux-Java-JDK 1.2.2 from blackdown as of yesterday. > > Backtrace see crash.txt. Kernelconfig see nihil. > > Any thoughts anyone? > > Yes, I've gotten these too. I really believe the assumptions the code > there makes are wrong, and I've got a patch to correct them to what I > think they are supposed to be. You've got the standard disclaimer on > the patch, though I assure you it has shown no ill effects to me, and I > noticed this bug through WINE. > > I've asked Poul-Henning Kamp, who seems to also think that the code makes > wrong assumptions. I've asked Matt Dillon and gotten no reply (a month > now, at least). I've asked Alan Cox, and he'd help if we could get him > a test case so he can watch it happen himself and debug it himself. > > Do you think you can find a specific set of steps for Alan to reproduce > it? Yes, find all places where source->ref_count is incremented and check for OBJ_ONEMAPPING as well as where OBJ_ONEMAPPING is set. Then add some printfs to find the snippet that's incrementing it to complain when the OBJ_ONEMAPPING bit is set, and complain if setting OBJ_ONEMAPPING when the refcount is too high. Blotting out a KASSERT isn't the right thing to do. -- -Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org] "I have the heart of a child; I keep it in a jar on my desk." 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?20000415092639.Q4381>