[PATCH 2/4] arm64: dts: qcom: sc7180: Add iommus property to ETR

[Bjorn Andersson]

On Mon 27 Jul 21:40 PDT 2020, Sai Prakash Ranjan wrote:

> On 2020-07-28 02:28, Bjorn Andersson wrote:
> > On Tue 23 Jun 23:56 PDT 2020, Sai Prakash Ranjan wrote:
> > 
> > > Hi Bjorn,
> > > 
> > > On 2020-06-21 13:39, Sai Prakash Ranjan wrote:
> > > > Hi Bjorn,
> > > >
> > > > On 2020-06-21 12:52, Bjorn Andersson wrote:
> > > > > On Tue 09 Jun 06:30 PDT 2020, Sai Prakash Ranjan wrote:
> > > > >
> > > > > > Define iommus property for Coresight ETR component in
> > > > > > SC7180 SoC with the SID and mask to enable SMMU
> > > > > > translation for this master.
> > > > > >
> > > > >
> > > > > We don't have &apps_smmu in linux-next, as we've yet to figure out how
> > > > > to disable the boot splash or support the stream mapping handover.
> > > > >
> > > > > So I'm not able to apply this.
> > > > >
> > > >
> > > > This is for SC7180 which has apps_smmu not SM8150.
> > > >
> > > 
> > > Please let me know if this needs further explanation.
> > > 
> > 
> > I must have commented on the wrong patch, sorry about that. The SM8150
> > patch in this series does not compile due to the lack of &apps_smmu.
> > 
> > I've picked the other 3 patches.
> > 
> 
> Thanks Bjorn, I can resend SM8150 coresight change when SMMU support lands
> for it
> since coresight ETR won't work without it on android bootloaders.
> 
> As for the other 3 patches, Patch 1 and Patch 2 will apply cleanly to the
> right coresight
> nodes but due to the missing unique context in Patch 3, it could be applied
> to some other node.
> We had to upload this change 3 times in chromium tree to get it applied to
> the right replicator node :)
> and this property in Patch 3 is important to fix a hard lockup. I'm not sure
> why this patch is missing
> the proper context :/
> 
> I couldn't find the changes yet in qcom/for-next or other branches to see if
> it is
> applied to right replicator node. In case you haven't applied it yet, Patch
> 3 change
> should be applied to "replicator at 6b06000" node.
> 

Thanks for pointing that out, I've fixed up the incorrectly applied
change. (Still not published the branch)


For the future I believe you can pass -U <n> to git format-patch to get
<n> number of lines of context. Making that bigger than the default 3
should help for the coresight patches.

Thanks,
Bjorn