Re: [PATCH v3 04/18] x86/hvm: make sure translated MMIO reads or writes fall within a page

From: "Jan Beulich" <JBeulich@suse.com>
To: Paul Durrant <paul.durrant@citrix.com>
Cc: Andrew Cooper <andrew.cooper3@citrix.com>,
	Keir Fraser <keir@xen.org>,
	xen-devel@lists.xenproject.org
Subject: Re: [PATCH v3 04/18] x86/hvm: make sure translated MMIO reads or writes fall within a page
Date: Tue, 23 Jun 2015 16:52:43 +0100	[thread overview]
Message-ID: <55899CEB0200007800088748@mail.emea.novell.com> (raw)
In-Reply-To: <1435055997-30017-5-git-send-email-paul.durrant@citrix.com>

>>> On 23.06.15 at 12:39, <paul.durrant@citrix.com> wrote:
> ...otherwise they will simply carry on to the next page using a normal
> linear-to-phys translation.

And what's wrong about this?

> --- a/xen/arch/x86/hvm/emulate.c
> +++ b/xen/arch/x86/hvm/emulate.c
> @@ -586,7 +586,6 @@ static int __hvmemul_read(
>                                          p_data);
>              if ( rc != X86EMUL_OKAY || bytes == chunk )
>                  return rc;
> -            addr += chunk;
>              off += chunk;
>              gpa += chunk;
>              p_data += chunk;
> @@ -594,6 +593,8 @@ static int __hvmemul_read(
>              if ( bytes < chunk )
>                  chunk = bytes;
>          }
> +
> +        return X86EMUL_UNHANDLEABLE;
>      }

All the loop above does is leverage that we don't need to do a
translation, as we already know the physical address. Continuing
if the access crosses a page boundary is not wrong at all afaict.

Jan