Xen devel - Nov 2010 - [PATCH]: Allow HVM guests to use entire address space

HVM guests are allowed to use the entire virtual address space for
whatever they want, without any Xen hole.  __addr_ok() is therefore
always true for such guests.

Signed-off-by: Steven Smith <steven.smith@eu.citrix.com>

diff -r 7188d1e4b0e1 xen/include/asm-x86/x86_64/uaccess.h
--- a/xen/include/asm-x86/x86_64/uaccess.h	Tue Nov 09 12:00:05 2010 +0000
+++ b/xen/include/asm-x86/x86_64/uaccess.h	Wed Nov 10 13:34:52 2010 +0000
@@ -20,7 +20,8 @@ void free_compat_arg_xlat(unsigned int c
  * non-canonical address (and thus fault) before ever reaching VIRT_START.
  */
 #define __addr_ok(addr) \
-    (((unsigned long)(addr) < (1UL<<48)) || \
+    (is_hvm_vcpu(current) || \
+     ((unsigned long)(addr) < (1UL<<48)) || \
      ((unsigned long)(addr) >= HYPERVISOR_VIRT_END))
 
 #define access_ok(addr, size) \
@@ -30,6 +31,7 @@ void free_compat_arg_xlat(unsigned int c
     (access_ok(addr, (count)*(size)))
 
 #define __compat_addr_ok(d, addr) \
+    (is_hvm_vcpu(current) || \
     ((unsigned long)(addr) < HYPERVISOR_COMPAT_VIRT_START(d))
 
 #define __compat_access_ok(d, addr, size) \



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 10/11/2010 13:40, "Gianni Tedesco"
<gianni.tedesco@citrix.com> wrote:
> HVM guests are allowed to use the entire virtual address space for
> whatever they want, without any Xen hole.  __addr_ok() is therefore
> always true for such guests.
This one got nacked and the bug was instead fixed by xen-unstable:21456 aka
xen-4.0-testing:21169. So we don''t need this patch, thank you. ;-)

 -- Keir
> Signed-off-by: Steven Smith <steven.smith@eu.citrix.com>
> 
> diff -r 7188d1e4b0e1 xen/include/asm-x86/x86_64/uaccess.h
> --- a/xen/include/asm-x86/x86_64/uaccess.h Tue Nov 09 12:00:05 2010 +0000
> +++ b/xen/include/asm-x86/x86_64/uaccess.h Wed Nov 10 13:34:52 2010 +0000
> @@ -20,7 +20,8 @@ void free_compat_arg_xlat(unsigned int c
>   * non-canonical address (and thus fault) before ever reaching VIRT_START.
>   */
>  #define __addr_ok(addr) \
> -    (((unsigned long)(addr) < (1UL<<48)) || \
> +    (is_hvm_vcpu(current) || \
> +     ((unsigned long)(addr) < (1UL<<48)) || \
>       ((unsigned long)(addr) >= HYPERVISOR_VIRT_END))
>  
>  #define access_ok(addr, size) \
> @@ -30,6 +31,7 @@ void free_compat_arg_xlat(unsigned int c
>      (access_ok(addr, (count)*(size)))
>  
>  #define __compat_addr_ok(d, addr) \
> +    (is_hvm_vcpu(current) || \
>      ((unsigned long)(addr) < HYPERVISOR_COMPAT_VIRT_START(d))
>  
>  #define __compat_access_ok(d, addr, size) \
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On Wed, 2010-11-10 at 13:46 +0000, Keir Fraser wrote:
> On 10/11/2010 13:40, "Gianni Tedesco"
<gianni.tedesco@citrix.com> wrote:
> 
> > HVM guests are allowed to use the entire virtual address space for
> > whatever they want, without any Xen hole.  __addr_ok() is therefore
> > always true for such guests.
> 
> This one got nacked and the bug was instead fixed by xen-unstable:21456 aka
> xen-4.0-testing:21169. So we don''t need this patch, thank you. ;-)
> 
>  -- Keir
Aha, good to know, I will make appropriate changes our end.

Gianni


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel