Keir,
I''d like to restrict the address check in the code snipped below so
that addresses in the range 0xFED4 0000 to 0xFED4 5000 do not end up in
the domain_crash() function. I have a virtual device (qemu device model)
that uses addresses in this range. Can I just open up that window with a
check like this or can the window even be further open onto higher
addresses?
if ( !hvm_apic_support(d) &&
(((gpa >= 0xFEC00000) && (gpa < FED40000)) ||
(gpa >= 0xFED45000) ))
The device is a TPM TIS interface. Specs at:
https://www.trustedcomputinggroup.org/groups/pc_client/TCG_PCClientTPMSpecification_1-20_1-00_FINAL.pdf
Page 27 shows the address range.
from xen/arch/x86/mm/shadow/multi.c:
mmio:
perfc_incrc(shadow_fault_mmio);
if ( !hvm_apic_support(d) && (gpa >= 0xFEC00000) )
{
/* Need to deal with these disabled-APIC accesses, as
* handle_mmio() apparently does not currently do that. */
/* TJD: What about it, then? For now, I''m turning this BUG()
* into a domain_crash() since we don''t want to kill Xen. */
SHADOW_ERROR("disabled-APIC access: not supported\n.");
domain_crash(d);
}
Stefan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
On 19/10/06 8:37 pm, "Stefan Berger" <stefanb@us.ibm.com> wrote:> if ( !hvm_apic_support(d) && > (((gpa >= 0xFEC00000) && (gpa < FED40000)) || > (gpa >= 0xFED45000) ))The test should be removed entirely and handle_mmio() should be extended to deal appropriately with accesses to unused addresses, if it doesn''t already. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser <Keir.Fraser@cl.cam.ac.uk> wrote on 10/19/2006 04:47:21 PM:> On 19/10/06 8:37 pm, "Stefan Berger" <stefanb@us.ibm.com> wrote: > > > if ( !hvm_apic_support(d) && > > (((gpa >= 0xFEC00000) && (gpa < FED40000)) || > > (gpa >= 0xFED45000) )) > > The test should be removed entirely and handle_mmio() should be extendedto> deal appropriately with accesses to unused addresses, if it doesn''talready. It handles the MMIO addresses correctly - at least from what I can see. It''s probably easier for you to cut those few lines, so I won''t send a patch. Stefan> > -- Keir > >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel