[RESPIN PATCH] irqchip/gic-v3-its:Fix GICv4.1 needless VSYNC after unmap VPE

Tangnianyao tangnianyao at huawei.com
Fri Apr 5 18:55:18 PDT 2024



On 4/3/2024 18:09, Marc Zyngier wrote:
> Thanks for respinning this.
>
> A few remarks:
>
> The subject line could be improved. Something like:
>
> "irqchip/gic-v4: Don't issue a VSYNC after VMAPP with V=0"
>
> On Wed, 03 Apr 2024 09:35:56 +0100,
> t00849498 <tangnianyao at huawei.com> wrote:
>> From: Nianyao Tang <tangnianyao at huawei.com>
>>
>> Quote from GIC spec 5.3.19, a VMAPP with {V, Alloc}=={0, x}
>> is self-synchronizing, This means the ITS command queue does not
>> show the command as consumed until all of its effects are completed.
> Since this is a direct quote, make it clear that it is so.
>
>> We don't need VSYNC to guarantee unmap finish. And VSYNC after unmap VPE
>> will reach an invalid vpe table entry, which may trigger exception
>> like SError or RAS. Let's fix it.
> This should be much stronger. It's not that we don't need VSYNC. It is
> that VSYNC is actively wrong. I suggest that you rewrite the commit
> message along these lines:
>
> <msg>
> As per the GICv4.1 spec (Arm IHI 0069H, 5.3.19):
>
> "A VMAPP with {V, Alloc}=={0, x} is self-synchronizing. This means the
>  ITS command queue does not show the command as consumed until all of
>  its effects are completed."
>
> Furthermore, VSYNC is allowed to deliver an SError when referencing a
> non existent VPE.
>
> By these definitions, a VMAPP followed by a VSYNC is a bug, as the
> later references a VPE that has been unmapped by the former.
>
> Fix it by eliding the VSYNC in this scenario.
> </msg>

Thanks for the above comments, I will resend later.
>
>> Signed-off-by: Nianyao Tang <tangnianyao at huawei.com>
> Please also add:
>
> Fixes: 64edfaa9a234 ("irqchip/gic-v4.1: Implement the v4.1 flavour of VMAPP")
>
> With the above fixed:
>
> Reviewed-by: Marc Zyngier <maz at kernel.org>
>
> Thanks,
>
> 	M.
>




More information about the linux-arm-kernel mailing list