Re: [PATCH 7/9] iommu/arm-smmu-v3: Directly encode CMDQ_OP_ATC_INV

[Pranjal Shrivastava <praan@google.com>]

On Sat, May 09, 2026 at 01:54:01PM -0300, Jason Gunthorpe wrote:
> On Fri, May 08, 2026 at 11:46:39AM +0000, Pranjal Shrivastava wrote:
> > On Fri, May 01, 2026 at 11:29:16AM -0300, Jason Gunthorpe wrote:
> > > Add a new command make function and convert all the places using
> > > ATC_INV.
> > > 
> > > Split out full invalidation to directly make the cmd instead of
> > > overloading size=0 to mean full invalidation.
> > > 
> > > Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> > > ---
> > 
> > Nit: I guess it's worth mentioning that we remove CMDQ_ATC_0_GLOBAL as
> > we don't set ent->atc.global = true anywhere in the driver anymore.
> 
> I added this:
> 
> In section "3.9.1 ATS Interface" of F.b the specification says:
> 
>   When the SMMU returns an ATS Translation Completion for a request that
>   had a PASID, the Global bit of the Translation Completion Data Entry
>   must be zero.
> 
> Even though it faithfully forwards the G bit through to the ATS
> invalidation command there is no way to create G mappings so there is
> never any need to send a G invalidation. Thus don't expose global in the
> new helpers and leave CMDQ_ATC_0_GLOBAL unused.

Ack. Sounds good. Should also add that the only other place we might need
G=1 is for a full ATC wipe. However, we don't since the PCIe Spec 10.3.8,
mentions issuing an invalidation without a PASID prefix (SSV=0) already
forces the device to invalidate all ATC entries for any PASID at all 
addresses, e.g. arm_smmu_atc_inv_master(master, IOMMU_NO_PASID).

Although the comment in arm_smmu_enable_ats should suffice too.

Thanks,
Praan