Re: [PATCH] irqchip/irq-gic-v3-its.c: Return its_invall_cmd.col when building INVALL

[Shaokun Zhang]

Hi Marc,

On 2021/12/8 16:25, Marc Zyngier wrote:
> On 2021-12-08 01:54, Shaokun Zhang wrote:
>> From: Wudi Wang <wangwudi@xxxxxxxxxxxxx>
>>
>> INVALL CMD specifies that the ITS must ensure any caching associated with
>> the interrupt collection defined by ICID is consistent with the LPI
>> configuration tables held in memory for all Redistributors. SYNC is
>> required to ensure that INVALL is executed.
> 
> The patch title doesn't quite spell out the issue. It should say something
> like:
> 
> "Force synchronisation when issuing INVALL">

Make sense.

>>
>> Currently, LPI configuration data may be inconsistent with that in the
>> memory within a short period of time after the INVALL command is executed.
> 
> I'm curious: have you seen any issue with this on actual HW? In my
> experience, all implementations treat INVALL as a synchronous command,
> 
> Or was this solely done via inspection?
> 

It is noticed by checking the implementation of INVALL API function, not
by on actual HW.

>>
>> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
>> Cc: Marc Zyngier <maz@xxxxxxxxxx>
>> Signed-off-by: Wudi Wang <wangwudi@xxxxxxxxxxxxx>
>> Signed-off-by: Shaokun Zhang <zhangshaokun@xxxxxxxxxxxxx>
> 
> This needs:
> 
> Fixes: cc2d3216f53 ("irqchip: GICv3: ITS command queue")
> 

Oops, indeed, apologies that forget to add this tag.

>> ---
>>  drivers/irqchip/irq-gic-v3-its.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
>> index eb0882d15366..0cb584d9815b 100644
>> --- a/drivers/irqchip/irq-gic-v3-its.c
>> +++ b/drivers/irqchip/irq-gic-v3-its.c
>> @@ -742,7 +742,7 @@ static struct its_collection
>> *its_build_invall_cmd(struct its_node *its,
>>
>>      its_fixup_cmd(cmd);
>>
>> -    return NULL;
>> +    return desc->its_invall_cmd.col;
>>  }
>>
>>  static struct its_vpe *its_build_vinvall_cmd(struct its_node *its,
> 
> I'll fix the above locally, no need to resend.
> 

Thanks Marc's help.

> Thanks,
> 
>         M.