This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


[Xen-devel] RE: [RFC][PATCH][VTD][v3] consolidate VT-d quirks into a sin

>>> On 28.10.10 at 00:29, "Kay, Allen M" <allen.m.kay@xxxxxxxxx> wrote:
>> The issue is not with doing a mapping, but with where (in virtual address 
> space)
>> you map to: You're passing a *physical* address for what is to be a 
> *virtual* one
>> (first argument to map_pages_to_xen()), i.e. as soon as there is any domain 
> the
>> mapping will conflict with the domain's use of virtual addresses.
> Agree.  I fixed it in the attached v3 patch by using a fixmap entry for 
> doing ioremap.

Yes, this looks better now. However, map_igd_reg() now returns
"status" without ever initializing the variable. Didn't the compiler
warn (and the build fail because) of that? I think the function has
no need to return non-void anymore.

The non-void return value of cantiga_vtd_ops_preamble() also
looks bogus, btw.

One thing I may not have noticed in earlier versions is your use
of IGD_BAR_MASK - you define it as 0xFFFF0000 but then use
it to mask a 64-bit value (i.e. cutting of the top 32 bits).


Xen-devel mailing list