[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Anybody debugging PAE guest kernels?

> > Anyhow, the function to look at is map_domain_va_pae in
> > tools/libxc/xc_ptrace.c. The checks for HVM support seem to not always
> > happen correctly. HTH.
> Thanks for the tip.  However, note that I'm not running HVM and the
> error I'm getting comes from Xen proper:

Exactly. I've seen cases where the HVM check returns true when running
against a paravirtualized FreeBSD guest. I had my mentee ifdef out the
HVM check in his local copy of gdbserver. In your case, if the code is
not parsing the page tables correctly it can result in in trying to read
/ write memory it isn't allowed to. Tracing errors back from mm.c is
always a monumental PITA, but it might be worth looking at. At the very
least, check what type it expects the page to be - I've long since
forgotten what the numbers correspond to. If you're not already familiar
with the page typing code you should have a look, familiarity with it
will make tracking down errors much easier. One last thing to bear in
mind for future reference is that DOM0 can't map page tables, even for a
domU (at least last I tried), read/write.


> >> (XEN) DOM0: (file=mm.c, line=1750) Bad type (saw 2e000001 != exp
> >> e0000000)
> >> for mfn fa693 (pfn 572)
> >> (XEN) DOM0: (file=mm.c, line=618) Error getting mfn fa693 (pfn
> >> 572) from L1
> >> entry 00000000fa693027 for dom7
> -- Randy
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-devel

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.