Re: [Xen-devel] [PATCH v3 3/3] AMD IOMMU: Check io_handler before registering mmio handler

[Paul Durrant <Paul.Durrant@xxxxxxxxxx>]

> -----Original Message-----
> From: Suravee Suthikulanit [mailto:suravee.suthikulpanit@xxxxxxx]
> Sent: 25 May 2016 19:52
> To: Paul Durrant; xen-devel@xxxxxxxxxxxxx; jbeulich@xxxxxxxx; George
> Dunlap
> Cc: Keir (Xen.org)
> Subject: Re: [PATCH v3 3/3] AMD IOMMU: Check io_handler before
> registering mmio handler
> 
> On 5/23/2016 3:23 AM, Paul Durrant wrote:
> >> -----Original Message-----
> >> > From: suravee.suthikulpanit@xxxxxxx
> >> > [mailto:suravee.suthikulpanit@xxxxxxx]
> >> > Sent: 22 May 2016 00:43
> >> > To: xen-devel@xxxxxxxxxxxxx; Paul Durrant; jbeulich@xxxxxxxx; George
> >> > Dunlap
> >> > Cc: Keir (Xen.org); Suravee Suthikulpanit; Suravee Suthikulapanit
> >> > Subject: [PATCH v3 3/3] AMD IOMMU: Check io_handler before
> registering
> >> > mmio handler
> >> >
> >> > From: Suravee Suthikulpanit <Suravee.Suthikulpanit@xxxxxxx>
> >> >
> >> > guest_iommu_init tries to register mmio handler before HVM domain
> >> > is initialized. This cause registration to silently failing.
> >> > This patch adds a sanitiy check and puts out error message.
> >> >
> >> > Signed-off-by: Suravee Suthikulapanit
> <suravee.suthikulpanit@xxxxxxx>
> > This patch is now defunct isn't it?
> >
> >   Paul
> >
> 
> It is no longer required, but I think this is a good sanity check in
> case something changes in the future and causing this to be called
> before the HVM I/O handler initialized.

Maybe, but shouldn't it result in a domain_crash()?

  Paul

> 
> Thanks,
> Suravee

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel