[PATCH v2 2/2] iommu/arm-smmu: Do not access non-existing SMR registers

On 8/7/2014 2:22 AM, Will Deacon wrote:

On Thu, Aug 07, 2014 at 12:34:09AM +0100, Olav Haugan wrote:

quoted

On 8/6/2014 10:35 AM, Will Deacon wrote:

quoted

Hmm, I'm checking this with the architects because the TRMs aren't exactly
clear. The NUMSMRG works for stream-indexing (i.e. reports the number of
S2CRs), then all we have to do is change the above ~SMR_VALID to 0x0, since
those registers will be SBZP.

You don't agree that we should avoid doing register writes if not
necessarily? In general I like to avoid trying to write registers that
is not needed so that we don't do more work than needed.

If this was a fast-path then I'd certainly agree. However, device reset is
a pretty rare event and I'd be inclined to err on the side of code clarity
in preference to performance.

Anyway, it turns out that we can't use NUMSMRG for this so all that's moot.
We instead need to do 1 << SMMU_IDR0.NUMSIDB, which will give us the number
of S2CRs.

quoted

quoted

quoted

As far as I can tell there are no other register telling us how many S2CR
registers exist. That also brings up another point that there is no check
in the code to ensure we are not trying to program more than the available
S2CR registers when we use stream indexing.

On an SMMU using stream-indexing, if the StreamID goes off the end of the
S2CRs that's a fairly serious hardware configuration issue which I don't
think Linux is in a position to handle. I agree that a warning wouldn't
hurt on device add/attach though.

Yes, I meant to protect against programming errors where someone
specified more Stream IDs than available S2CRs.

Sure. Could you have a go at hacking this up, please?

Yes, I'll update the patches. Thanks for clarifying how to determine the
# of S2CRs.

Thanks,

Olav

-- 
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation