linux-kernel - Re: [PATCH v2 2/2] drm/msm/dpu: Add SC8180x to hw catalog

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 15 Feb 2022 20:03:14 -0600
From:   Bjorn Andersson <bjorn.andersson@...aro.org>
To:     Abhinav Kumar <quic_abhinavk@...cinc.com>
Cc:     Dmitry Baryshkov <dmitry.baryshkov@...aro.org>,
        linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
        dri-devel@...ts.freedesktop.org, freedreno@...ts.freedesktop.org
Subject: Re: [PATCH v2 2/2] drm/msm/dpu: Add SC8180x to hw catalog

On Tue 15 Feb 19:34 CST 2022, Abhinav Kumar wrote:

> 
> 
> On 2/15/2022 4:20 PM, Dmitry Baryshkov wrote:
> > On Tue, 15 Feb 2022 at 23:21, Abhinav Kumar <quic_abhinavk@...cinc.com> wrote:
> > > On 2/15/2022 10:42 AM, Dmitry Baryshkov wrote:
> > > > On Tue, 15 Feb 2022 at 20:42, Abhinav Kumar <quic_abhinavk@...cinc.com> wrote:
> > > > > On 2/15/2022 9:28 AM, Bjorn Andersson wrote:
> > > > > > On Tue 15 Feb 11:14 CST 2022, Abhinav Kumar wrote:
> > > > > > > On 2/14/2022 8:33 PM, Bjorn Andersson wrote:
> > > > > > > > From: Rob Clark <robdclark@...omium.org>
[..]
> > > > (thus leading us to cases when someone would forget to add INTF_EDP
> > > > next to INTF_DP)
> > > > 
> > > > Also, if we are switching from INTF_DP to INTF_EDP, should we stop
> > > > using end-to-end numbering (like MSM_DP_CONTROLLER_2 for INTF_5) and
> > > > add a separate numbering scheme for INTF_EDP?
> > > > 
> > > We should change the controller ID to match what it actually is.
> > > 
> > > Now that you pointed this out, this looks even more confusing to me to
> > > say that  MSM_DP_CONTROLLER_2 is actually a EDP controller because
> > > fundamentally and even hardware block wise they are different.
> > 
> > So, do we split msm_priv->dp too? It's indexed using
> > MSM_DP_CONTROLLER_n entries.
> > Do we want to teach drm/msm/dp code that there are priv->dp[] and
> > priv->edp arrays?
> 
> ok so now priv->dp and priv->edp arrays are also in the picture here :)
> 
> Actually all these questions should have probably come when we were figuring
> out how best to re-use eDP and DP driver.
> 
> Either way atleast, its good we are documenting all these questions on this
> thread so that anyone can refer this to know what all was missed out :)
> 
> priv->dp is of type msm_dp. When re-using DP driver for eDP and since
> struct msm_dp is the shared struct between dpu and the msm/dp, I get your
> point of re-using MSM_DP_CONTROLLER_* as thats being use to index.
> 
> So MSM_DP_CONTROLLER_* is more of an index into the DP driver and not really
> a hardware indexing scheme.
> 
> If we split into two arrays, we need more changes to dpu_encoder too.
> 
> Too instrusive a change at this point, even though probably correct.
> 

I'm sorry, but performing such a split would create a whole bunch of
duplication and I don't see the reasons yet. Can you please give me an
example of when the DPU _code_ would benefit from being specifically
written for EDP vs DP?

Things where it doesn't make sense to enable certain features in
runtime - but really have different implementation for the two interface
types.

> But are you seeing more changes required even if we just change INTF_DP to
> INTF_eDP for the eDP entries? What are the challenges there?
> 

What are the benefits?

Regards,
Bjorn