Date: Sun, 7 Mar 2004 09:59:04 -0500 (EST) From: Andrew Gallatin <gallatin@cs.duke.edu> To: Sam Leffler <sam@errno.com> Cc: freebsd-current@freebsd.org Subject: Re: reproducable "panic: pmap_enter: attempted pmap_enter on 4MB page" Message-ID: <16459.14520.982776.823223@grasshopper.cs.duke.edu> In-Reply-To: <B5B15702-6FB1-11D8-92E0-000A95AD0668@errno.com> References: <16453.61095.940884.597029@grasshopper.cs.duke.edu> <16456.43337.389382.661597@grasshopper.cs.duke.edu> <B5B15702-6FB1-11D8-92E0-000A95AD0668@errno.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Sam Leffler writes: > On Mar 5, 2004, at 8:22 AM, Andrew Gallatin wrote: > > I've been getting this panic off+on for >9 months but haven't been able > to characterize it and/or narrow down the cause. I mostly see it on an > NFS server. For me it seemed to always happen when a particular kva was used. (0xc86ce000). It did not matter if I used my real driver, or a contrived 20-line test case that just malloc'ed a bunch of memory at module load. As soon as pmap_enter() was called on that kva, kaboom! But this behaviour stopped when: a) I changed from a non-WITNESS, non-INVARIANTS kernel to a WITNESS and INVARIANTS kernel. a) I upgraded kernel sources from Weds -> Fri It almost smells like something size related. Eg, something is overflowing and corrupting the page tables when the kernel has a particular size or alignment. I'm just curious -- when this happens to you, is it the same KVA that I saw? There is a printf just before the panic which spits out the kva.. Drew
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?16459.14520.982776.823223>