Xen devel - Jul 2012 - [PATCH] x86: fix off-by-one in nr_irqs_gsi calculation

highest_gsi() returns the last valid GSI, not a count.

Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/xen/arch/x86/io_apic.c
+++ b/xen/arch/x86/io_apic.c
@@ -2531,7 +2531,9 @@ void __init init_ioapic_mappings(void)
         }
     }
 
-    nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi());
+    i = highest_gsi();
+    if ( i >= nr_irqs_gsi )
+        nr_irqs_gsi = i + 1;
 
     if ( max_gsi_irqs == 0 )
         max_gsi_irqs = nr_irqs ? nr_irqs / 8 : PAGE_SIZE;





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

On 26/07/2012 16:06, "Jan Beulich" <JBeulich@suse.com> wrote:
> highest_gsi() returns the last valid GSI, not a count.
> 
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
Why not "nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi()+1)"?
> --- a/xen/arch/x86/io_apic.c
> +++ b/xen/arch/x86/io_apic.c
> @@ -2531,7 +2531,9 @@ void __init init_ioapic_mappings(void)
>          }
>      }
>  
> -    nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi());
> +    i = highest_gsi();
> +    if ( i >= nr_irqs_gsi )
> +        nr_irqs_gsi = i + 1;
>  
>      if ( max_gsi_irqs == 0 )
>          max_gsi_irqs = nr_irqs ? nr_irqs / 8 : PAGE_SIZE;
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel

>>> On 26.07.12 at 17:31, Keir Fraser <keir@xen.org> wrote:
> On 26/07/2012 16:06, "Jan Beulich" <JBeulich@suse.com>
wrote:
> 
>> highest_gsi() returns the last valid GSI, not a count.
>> 
>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> 
> Why not "nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi()+1)"?
While I think x = max(y, z) is fine, I generally find x = max(x, y)
pretty useless - what''s the point of assigning a value to itself?
This is __init code, so code size considerations probably don''t
matter, but I''d still like to not encourage inefficient code like this
to be used elsewhere by giving a bad example...

Jan
>> --- a/xen/arch/x86/io_apic.c
>> +++ b/xen/arch/x86/io_apic.c
>> @@ -2531,7 +2531,9 @@ void __init init_ioapic_mappings(void)
>>          }
>>      }
>>  
>> -    nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi());
>> +    i = highest_gsi();
>> +    if ( i >= nr_irqs_gsi )
>> +        nr_irqs_gsi = i + 1;
>>  
>>      if ( max_gsi_irqs == 0 )
>>          max_gsi_irqs = nr_irqs ? nr_irqs / 8 : PAGE_SIZE;
>> 
>> 
>> 
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xen.org 
>> http://lists.xen.org/xen-devel

On 26/07/2012 16:43, "Jan Beulich" <JBeulich@suse.com> wrote:
>>>> On 26.07.12 at 17:31, Keir Fraser <keir@xen.org> wrote:
>> On 26/07/2012 16:06, "Jan Beulich" <JBeulich@suse.com>
wrote:
>> 
>>> highest_gsi() returns the last valid GSI, not a count.
>>> 
>>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
>> 
>> Why not "nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi()+1)"?
> 
> While I think x = max(y, z) is fine, I generally find x = max(x, y)
> pretty useless - what''s the point of assigning a value to itself?
> This is __init code, so code size considerations probably don''t
> matter, but I''d still like to not encourage inefficient code like
this
> to be used elsewhere by giving a bad example...
I think the code is clearer with max(). An even stronger argument is that it
makes the intent of the patch much more obvious too. I think you should
leave the max() construct in place in this patch.

 -- Keir
> Jan
> 
>>> --- a/xen/arch/x86/io_apic.c
>>> +++ b/xen/arch/x86/io_apic.c
>>> @@ -2531,7 +2531,9 @@ void __init init_ioapic_mappings(void)
>>>          }
>>>      }
>>>  
>>> -    nr_irqs_gsi = max(nr_irqs_gsi, highest_gsi());
>>> +    i = highest_gsi();
>>> +    if ( i >= nr_irqs_gsi )
>>> +        nr_irqs_gsi = i + 1;
>>>  
>>>      if ( max_gsi_irqs == 0 )
>>>          max_gsi_irqs = nr_irqs ? nr_irqs / 8 : PAGE_SIZE;
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> Xen-devel mailing list
>>> Xen-devel@lists.xen.org
>>> http://lists.xen.org/xen-devel
> 
> 
>