linux-kernel - Re: [Freedreno] [PATCH v2 04/17] drm/msm/dpu: Fix PP_BLK

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, 25 Apr 2023 14:37:21 -0700
From:   Abhinav Kumar <quic_abhinavk@...cinc.com>
To:     Marijn Suijten <marijn.suijten@...ainline.org>
CC:     <dri-devel@...ts.freedesktop.org>,
        Jordan Crouse <jordan@...micpenguin.net>,
        AngeloGioacchino Del Regno 
        <angelogioacchino.delregno@...ainline.org>,
        David Airlie <airlied@...il.com>,
        Chandan Uddaraju <chandanu@...eaurora.org>,
        Archit Taneja <architt@...eaurora.org>,
        Robert Foss <rfoss@...nel.org>, Vinod Koul <vkoul@...nel.org>,
        Kuogee Hsieh <quic_khsieh@...cinc.com>,
        Rajesh Yadav <ryadav@...eaurora.org>,
        <linux-arm-msm@...r.kernel.org>,
        Adam Skladowski <a39.skl@...il.com>,
        Martin Botka <martin.botka@...ainline.org>,
        <~postmarketos/upstreaming@...ts.sr.ht>,
        Jeykumar Sankaran <jsanka@...eaurora.org>,
        Sean Paul <sean@...rly.run>,
        Neil Armstrong <neil.armstrong@...aro.org>,
        Loic Poulain <loic.poulain@...aro.org>,
        "Jami Kettunen" <jami.kettunen@...ainline.org>,
        Bjorn Andersson <andersson@...nel.org>,
        <linux-kernel@...r.kernel.org>,
        Konrad Dybcio <konrad.dybcio@...aro.org>,
        Rob Clark <robdclark@...il.com>,
        Daniel Vetter <daniel@...ll.ch>,
        Dmitry Baryshkov <dmitry.baryshkov@...aro.org>,
        <freedreno@...ts.freedesktop.org>,
        Sravanthi Kollukuduru <skolluku@...eaurora.org>
Subject: Re: [Freedreno] [PATCH v2 04/17] drm/msm/dpu: Fix PP_BLK_DIPHER ->
 DITHER typo



On 4/25/2023 1:43 PM, Marijn Suijten wrote:
> On 2023-04-25 09:47:30, Abhinav Kumar wrote:
>>
>>
>> On 4/25/2023 9:33 AM, Marijn Suijten wrote:
>>> On 2023-04-25 09:18:58, Abhinav Kumar wrote:
>>>>
>>>>
>>>> On 4/24/2023 11:54 PM, Marijn Suijten wrote:
>>>>> On 2023-04-24 16:09:45, Abhinav Kumar wrote:
>>>>> <snip>
>>>>>>>> dither block should be present on many other chipsets too but looks like
>>>>>>>> on sm8550 was enabling it. Not sure how it was validated there. But we
>>>>>>>> are enabling dither, even other chipsets have this block.
>>>>>>>
>>>>>>> Correct, they all seem to have it starting at sdm845.  My patch message
>>>>>>> seems to lack the word "exclusively" as the PP on sm8550 appears to
>>>>>>> exclusively contain a DITHER subblock (unless other blocks are available
>>>>>>> that simply aren't supported within this driver yet) and no other
>>>>>>> registers.  Hence this aptly named macro exist to emit just the feature
>>>>>>> bitflag for that and a .len of zero.
>>>>>>>
>>>>>>
>>>>>> I think after the TE blocks were moved to INTF, dither is the only
>>>>>> sub-block for all Ping-Pongs not just in sm8550.
>>>>>
>>>>> So you are asking / leaving context to make all >= 5.0.0 pingpong blocks
>>>>> use this macro with only a single DITHER sblk in PP?
>>>>>
>>>>> As far as I recall SM8550 is the first SoC to use zero registers in PP,
>>>>> which is specifically what this macro takes care of too.  Then, there
>>>>> are only a few SoCs downstream still (erroneously?) referencing TE2 as
>>>>> the only other sub-blk, those SoCs still use sdm845_pp_sblk_te.
>>>>>
>>>>
>>>> So, what I didnt follow is why should sm8450 use PP_BLK_TE Vs sm8550
>>>> should use PP_BLK_DIPHER?
>>>>
>>>> Atleast for those two, both should be using PP_BLK_DIPHER.
>>>>
>>>> Thats what I was trying to note here.
>>>>
>>>> This isnt even right as there is no PP_BLK_TE in sm8450.
>>>
>>> SM8450 doesn't use PP_BLK_TE (TE2) anymore since the second patch in
>>> this series.  If you think it should use the DITHER (not DIPHER!) macro
>>> instead of the regular PP_BLK with a size of 0xd4, we can do that in
>>> another patch as that's not strictly related to this series.
>>>
>>
>> Yes, thanks for pointing the TE2 was removed in the prev patch of this
>> series for sm8450. I was just focusing too much on this patch.
>>
>> And Yes, I think we should use the DIPHER ..... oh sorry .... DITHER ;)
>>
>> Yes, it can go as a different series, like I already wrote many times in
>> this.
> 
> Thanks, that'd be great.  I wasn't sure at this point what you wanted to
> be changed here, after commenting on a typo fix rather than i.e. patch 2
> that deals with the TE2 sub-block of PP :)
> 

The reason I commented on this patch is because all the discussion so 
far was surrounding the PP_BLK_DITHER macro which was being touched in 
this patch.

So even now, we found out about sm8450 and sm8550 because of the 
question that why sm8550 alone should use PP_BLK_DITHER and not sm8450.

This patch led to all the discussion about PP_BLK_DITHER.

Even though it was just a typo fix patch, it uncovered deeper issues in 
catalog about why PP_BLK_DITHER wasnt used more often.

>> But atleast now, someone will remember to do it.
>>
>>> Note that that's the only difference between these macros.  The size
>>> becomes 0 but the .features mask is the same (SM8450 uses
>>> PINGPONG_SM8150_MASK).
>>>
>>> These patches are anyway already distracting from my series, but were
>>> easier to do in one go as I was touching the PP and INTF catalog blocks
>>> regardless.
>>>
>>> While at it, perhaps we should check if the version and offset for the
>>> DITHER block are correct?  SM8450 uses SDM845 sblk definitions.
>>>
>>
>> Yes I already checked. the version and offset of dither are same between
>> sm8450 and sm8550.
> 
> Thanks for checking, so then sm8450 is wrong on multiple occasions.
> Let's check all other SoCs that use sdm845_pp_sblk whether they should
> have used sc7280_pp_sblk instead.
> 
> - Marijn