[PATCH for-4.12] x86/p2m: fix p2m_finish_type_change()

[PATCH for-4.12] x86/p2m: fix p2m_finish_type_change()

[Razvan Cojocaru]

finish_type_change() returns a negative int on error, but the
current code checks if ( !rc ). Also properly indent the out:
label while at it.

Signed-off-by: Razvan Cojocaru <rcojocaru@bitdefender.com>
---
 xen/arch/x86/mm/p2m.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
index 5451f16..e08f6b1 100644
--- a/xen/arch/x86/mm/p2m.c
+++ b/xen/arch/x86/mm/p2m.c
@@ -1176,7 +1176,7 @@ int p2m_finish_type_change(struct domain *d,
 
     rc = finish_type_change(hostp2m, first_gfn, max_nr);
 
-    if ( !rc )
+    if ( rc < 0 )
         goto out;
 
 #ifdef CONFIG_HVM
@@ -1193,13 +1193,13 @@ int p2m_finish_type_change(struct domain *d,
                 rc = finish_type_change(altp2m, first_gfn, max_nr);
                 p2m_unlock(altp2m);
 
-                if ( !rc )
+                if ( rc < 0 )
                     goto out;
             }
     }
 #endif
 
-out:
+ out:
     p2m_unlock(hostp2m);
 
     return rc;
-- 
2.7.4


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [PATCH for-4.12] x86/p2m: fix p2m_finish_type_change()

[Jan Beulich]

>>> On 09.01.19 at 11:41, <rcojocaru@bitdefender.com> wrote:
> finish_type_change() returns a negative int on error, but the
> current code checks if ( !rc ).

For the purpose of determining the backporting scope (none here)
it would be nice if in such a case you could point out the commit
introducing the breakage.

> --- a/xen/arch/x86/mm/p2m.c
> +++ b/xen/arch/x86/mm/p2m.c
> @@ -1176,7 +1176,7 @@ int p2m_finish_type_change(struct domain *d,
>  
>      rc = finish_type_change(hostp2m, first_gfn, max_nr);
>  
> -    if ( !rc )
> +    if ( rc < 0 )
>          goto out;
>  
>  #ifdef CONFIG_HVM
> @@ -1193,13 +1193,13 @@ int p2m_finish_type_change(struct domain *d,
>                  rc = finish_type_change(altp2m, first_gfn, max_nr);
>                  p2m_unlock(altp2m);
>  
> -                if ( !rc )
> +                if ( rc < 0 )
>                      goto out;
>              }
>      }
>  #endif

May I ask that you also fix the other return value issue here: Either
we mean to indicate to the caller the 0/1 distinction (in which case
it should be correct, i.e. cumulative across all finish_type_change()
calls made here), which was the behavior before 0fb4b58c8b
("x86/altp2m: fix display frozen when switching to a new view early"),
or we mean to signal success to the caller only by handing back zero.

Jan



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [PATCH for-4.12] x86/p2m: fix p2m_finish_type_change()

[George Dunlap]

> On Jan 9, 2019, at 8:41 PM, Razvan Cojocaru <rcojocaru@bitdefender.com> wrote:
> 
> finish_type_change() returns a negative int on error, but the
> current code checks if ( !rc ). Also properly indent the out:
> label while at it.
> 
> Signed-off-by: Razvan Cojocaru <rcojocaru@bitdefender.com>

Acked-by: George Dunlap <george.dunlap@citrix.com>


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [PATCH for-4.12] x86/p2m: fix p2m_finish_type_change()

[George Dunlap]

> On Jan 9, 2019, at 9:16 PM, Jan Beulich <jbeulich@suse.com> wrote:
> 
>>>> On 09.01.19 at 11:41, <rcojocaru@bitdefender.com> wrote:
>> finish_type_change() returns a negative int on error, but the
>> current code checks if ( !rc ).
> 
> For the purpose of determining the backporting scope (none here)
> it would be nice if in such a case you could point out the commit
> introducing the breakage.
> 
>> --- a/xen/arch/x86/mm/p2m.c
>> +++ b/xen/arch/x86/mm/p2m.c
>> @@ -1176,7 +1176,7 @@ int p2m_finish_type_change(struct domain *d,
>> 
>>     rc = finish_type_change(hostp2m, first_gfn, max_nr);
>> 
>> -    if ( !rc )
>> +    if ( rc < 0 )
>>         goto out;
>> 
>> #ifdef CONFIG_HVM
>> @@ -1193,13 +1193,13 @@ int p2m_finish_type_change(struct domain *d,
>>                 rc = finish_type_change(altp2m, first_gfn, max_nr);
>>                 p2m_unlock(altp2m);
>> 
>> -                if ( !rc )
>> +                if ( rc < 0 )
>>                     goto out;
>>             }
>>     }
>> #endif
> 
> May I ask that you also fix the other return value issue here: Either
> we mean to indicate to the caller the 0/1 distinction (in which case
> it should be correct, i.e. cumulative across all finish_type_change()
> calls made here), which was the behavior before 0fb4b58c8b
> ("x86/altp2m: fix display frozen when switching to a new view early"),
> or we mean to signal success to the caller only by handing back zero.

The “root” caller of p2m_finish_type_change() is XEN_DMOP_map_mem_type_to_ioreq_server; It doesn’t look like it does anything useful with positive values, and it’s not clear from an interface perspective that returning a positive value from that hypercall makes any sense.

Perhaps it would be best to just discard the positive values for now.

 -George
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel