From: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
To: Marijn Suijten <marijn.suijten@somainline.org>,
Abhinav Kumar <quic_abhinavk@quicinc.com>
Cc: freedreno@lists.freedesktop.org,
kernel test robot <lkp@intel.com>,
David Airlie <airlied@linux.ie>,
linux-arm-msm@vger.kernel.org, dri-devel@lists.freedesktop.org,
Stephen Boyd <swboyd@chromium.org>,
Rob Clark <robdclark@gmail.com>, Daniel Vetter <daniel@ffwll.ch>,
Bjorn Andersson <bjorn.andersson@linaro.org>,
Sean Paul <sean@poorly.run>
Subject: Re: [Freedreno] [PATCH v2] drm/msm/dsi: use RMW cycles in dsi_update_dsc_timing
Date: Mon, 2 May 2022 13:02:09 +0300 [thread overview]
Message-ID: <f8986d6d-2c2b-a0a0-d63c-d4670f6864ff@linaro.org> (raw)
In-Reply-To: <20220502083442.t365cfmivlga557g@SoMainline.org>
On 02/05/2022 11:34, Marijn Suijten wrote:
> On 2022-05-01 16:56:45, Abhinav Kumar wrote:
>> [snip]
>> Wouln't this macro already make sure that 'reg' doesnt have anything in
>> the top 16 bits? Its doing a & with 0x00003f00
>
> Like I said, it is unlikely that this happens, only if someone starts
> changing the code that assigns to `reg` which is unlikely to pass review
> anyway.
>
>> [snip]
>> We can have a common bitfield layout for the two channels for command mode.
>>
>> So we can do something like below for common fields:
>>
>> -static inline uint32_t
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE(uint32_t val)
>> +static inline uint32_t
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM_DATATYPE(uint32_t val, uint32_t
>> stream_id)
>> {
>> - return ((val) <<
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__SHIFT) &
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__MASK;
>> + return ((val) <<
>> (DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__SHIFT + (stream_id
>> * 16)) & DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__MASK;
>> }
>>
>> Video mode can also use all of these except for WC as that field is not
>> present for cmd modes.
>>
>> This can go as a separate change .
>>
>> I can push this and perhaps get a Tested-by from Vinod as I dont have a
>> setup to re-validate this.
>
> How would you represent this in XML? I was hoping for a method that
> allows to construct the value in a generic way, without register names,
> and then simply have a "register macro" that moves (and perhaps masks)
> the preconstructed value into the right place. A bit like how `enum`s
> are currently set up in XML, but with bit ranges for the values and
> macros to set a value.
>
> I think I've _partially_ found what I was looking for: a `<bitset>`.
> However, I don't know if we can utilize this multiple times within a
> single `reg32`, once with an offset for stream1. Alas, it's just
> bikeshedding at this point.
Unfortunately the following naïve patch doesn't work, stream1 bits are
still defined in the 0:15 bit space. One would have to modify rnn tool
to make sure that it takes into account the low/high parts of the
bitfield when generating offsets/masks.
diff --git a/src/freedreno/registers/dsi/dsi.xml
b/src/freedreno/registers/dsi/dsi.xml
index f2eef4ff41ae..b0166628ad0a 100644
--- a/src/freedreno/registers/dsi/dsi.xml
+++ b/src/freedreno/registers/dsi/dsi.xml
@@ -361,22 +361,19 @@
xsi:schemaLocation="http://nouveau.freedesktop.org/ rules-ng.xsd">
<bitfield name="MAJOR" low="24" high="31" type="uint"/>
</reg32>
<reg32 offset="0x002d4" name="CPHY_MODE_CTRL"/>
- <reg32 offset="0x0029c" name="VIDEO_COMPRESSION_MODE_CTRL">
- <bitfield name="WC" low="16" high="31" type="uint"/>
+ <bitset name="COMPRESSION_MODE_CTRL" inline="true">
<bitfield name="DATATYPE" low="8" high="13" type="uint"/>
<bitfield name="PKT_PER_LINE" low="6" high="7"
type="uint"/>
<bitfield name="EOL_BYTE_NUM" low="4" high="5"
type="uint"/>
<bitfield name="EN" pos="0" type="boolean"/>
+ </bitset>
+ <reg32 offset="0x0029c" name="VIDEO_COMPRESSION_MODE_CTRL">
+ <bitfield name="WC" low="16" high="31" type="uint"/>
+ <bitfield name="STREAM0" low="0" high="15"
type="COMPRESSION_MODE_CTRL"/>
</reg32>
<reg32 offset="0x002a4" name="COMMAND_COMPRESSION_MODE_CTRL">
- <bitfield name="STREAM1_DATATYPE" low="24" high="29"
type="uint"/>
- <bitfield name="STREAM1_PKT_PER_LINE" low="22" high="23"
type="uint"/>
- <bitfield name="STREAM1_EOL_BYTE_NUM" low="20" high="21"
type="uint"/>
- <bitfield name="STREAM1_EN" pos="16" type="boolean"/>
- <bitfield name="STREAM0_DATATYPE" low="8" high="13"
type="uint"/>
- <bitfield name="STREAM0_PKT_PER_LINE" low="6" high="7"
type="uint"/>
- <bitfield name="STREAM0_EOL_BYTE_NUM" low="4" high="5"
type="uint"/>
- <bitfield name="STREAM0_EN" pos="0" type="boolean"/>
+ <bitfield name="STREAM1" low="16" high="31"
type="COMPRESSION_MODE_CTRL"/>
+ <bitfield name="STREAM0" low="0" high="15"
type="COMPRESSION_MODE_CTRL"/>
</reg32>
<reg32 offset="0x002a8" name="COMMAND_COMPRESSION_MODE_CTRL2">
<bitfield name="STREAM1_SLICE_WIDTH" low="16" high="31"
type="uint"/>
--
With best wishes
Dmitry
WARNING: multiple messages have this Message-ID (diff)
From: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
To: Marijn Suijten <marijn.suijten@somainline.org>,
Abhinav Kumar <quic_abhinavk@quicinc.com>
Cc: Sean Paul <sean@poorly.run>, kernel test robot <lkp@intel.com>,
David Airlie <airlied@linux.ie>,
linux-arm-msm@vger.kernel.org, dri-devel@lists.freedesktop.org,
Stephen Boyd <swboyd@chromium.org>,
Bjorn Andersson <bjorn.andersson@linaro.org>,
freedreno@lists.freedesktop.org
Subject: Re: [Freedreno] [PATCH v2] drm/msm/dsi: use RMW cycles in dsi_update_dsc_timing
Date: Mon, 2 May 2022 13:02:09 +0300 [thread overview]
Message-ID: <f8986d6d-2c2b-a0a0-d63c-d4670f6864ff@linaro.org> (raw)
In-Reply-To: <20220502083442.t365cfmivlga557g@SoMainline.org>
On 02/05/2022 11:34, Marijn Suijten wrote:
> On 2022-05-01 16:56:45, Abhinav Kumar wrote:
>> [snip]
>> Wouln't this macro already make sure that 'reg' doesnt have anything in
>> the top 16 bits? Its doing a & with 0x00003f00
>
> Like I said, it is unlikely that this happens, only if someone starts
> changing the code that assigns to `reg` which is unlikely to pass review
> anyway.
>
>> [snip]
>> We can have a common bitfield layout for the two channels for command mode.
>>
>> So we can do something like below for common fields:
>>
>> -static inline uint32_t
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE(uint32_t val)
>> +static inline uint32_t
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM_DATATYPE(uint32_t val, uint32_t
>> stream_id)
>> {
>> - return ((val) <<
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__SHIFT) &
>> DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__MASK;
>> + return ((val) <<
>> (DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__SHIFT + (stream_id
>> * 16)) & DSI_COMMAND_COMPRESSION_MODE_CTRL_STREAM0_DATATYPE__MASK;
>> }
>>
>> Video mode can also use all of these except for WC as that field is not
>> present for cmd modes.
>>
>> This can go as a separate change .
>>
>> I can push this and perhaps get a Tested-by from Vinod as I dont have a
>> setup to re-validate this.
>
> How would you represent this in XML? I was hoping for a method that
> allows to construct the value in a generic way, without register names,
> and then simply have a "register macro" that moves (and perhaps masks)
> the preconstructed value into the right place. A bit like how `enum`s
> are currently set up in XML, but with bit ranges for the values and
> macros to set a value.
>
> I think I've _partially_ found what I was looking for: a `<bitset>`.
> However, I don't know if we can utilize this multiple times within a
> single `reg32`, once with an offset for stream1. Alas, it's just
> bikeshedding at this point.
Unfortunately the following naïve patch doesn't work, stream1 bits are
still defined in the 0:15 bit space. One would have to modify rnn tool
to make sure that it takes into account the low/high parts of the
bitfield when generating offsets/masks.
diff --git a/src/freedreno/registers/dsi/dsi.xml
b/src/freedreno/registers/dsi/dsi.xml
index f2eef4ff41ae..b0166628ad0a 100644
--- a/src/freedreno/registers/dsi/dsi.xml
+++ b/src/freedreno/registers/dsi/dsi.xml
@@ -361,22 +361,19 @@
xsi:schemaLocation="http://nouveau.freedesktop.org/ rules-ng.xsd">
<bitfield name="MAJOR" low="24" high="31" type="uint"/>
</reg32>
<reg32 offset="0x002d4" name="CPHY_MODE_CTRL"/>
- <reg32 offset="0x0029c" name="VIDEO_COMPRESSION_MODE_CTRL">
- <bitfield name="WC" low="16" high="31" type="uint"/>
+ <bitset name="COMPRESSION_MODE_CTRL" inline="true">
<bitfield name="DATATYPE" low="8" high="13" type="uint"/>
<bitfield name="PKT_PER_LINE" low="6" high="7"
type="uint"/>
<bitfield name="EOL_BYTE_NUM" low="4" high="5"
type="uint"/>
<bitfield name="EN" pos="0" type="boolean"/>
+ </bitset>
+ <reg32 offset="0x0029c" name="VIDEO_COMPRESSION_MODE_CTRL">
+ <bitfield name="WC" low="16" high="31" type="uint"/>
+ <bitfield name="STREAM0" low="0" high="15"
type="COMPRESSION_MODE_CTRL"/>
</reg32>
<reg32 offset="0x002a4" name="COMMAND_COMPRESSION_MODE_CTRL">
- <bitfield name="STREAM1_DATATYPE" low="24" high="29"
type="uint"/>
- <bitfield name="STREAM1_PKT_PER_LINE" low="22" high="23"
type="uint"/>
- <bitfield name="STREAM1_EOL_BYTE_NUM" low="20" high="21"
type="uint"/>
- <bitfield name="STREAM1_EN" pos="16" type="boolean"/>
- <bitfield name="STREAM0_DATATYPE" low="8" high="13"
type="uint"/>
- <bitfield name="STREAM0_PKT_PER_LINE" low="6" high="7"
type="uint"/>
- <bitfield name="STREAM0_EOL_BYTE_NUM" low="4" high="5"
type="uint"/>
- <bitfield name="STREAM0_EN" pos="0" type="boolean"/>
+ <bitfield name="STREAM1" low="16" high="31"
type="COMPRESSION_MODE_CTRL"/>
+ <bitfield name="STREAM0" low="0" high="15"
type="COMPRESSION_MODE_CTRL"/>
</reg32>
<reg32 offset="0x002a8" name="COMMAND_COMPRESSION_MODE_CTRL2">
<bitfield name="STREAM1_SLICE_WIDTH" low="16" high="31"
type="uint"/>
--
With best wishes
Dmitry
next prev parent reply other threads:[~2022-05-02 10:02 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-30 17:55 [PATCH v2] drm/msm/dsi: use RMW cycles in dsi_update_dsc_timing Dmitry Baryshkov
2022-04-30 17:55 ` Dmitry Baryshkov
2022-04-30 18:42 ` Abhinav Kumar
2022-04-30 18:42 ` Abhinav Kumar
2022-04-30 18:58 ` Marijn Suijten
2022-04-30 18:58 ` Marijn Suijten
2022-04-30 19:25 ` [Freedreno] " Abhinav Kumar
2022-04-30 19:25 ` Abhinav Kumar
2022-05-01 20:06 ` Marijn Suijten
2022-05-01 20:06 ` Marijn Suijten
2022-05-01 23:56 ` Abhinav Kumar
2022-05-01 23:56 ` Abhinav Kumar
2022-05-02 8:02 ` Dmitry Baryshkov
2022-05-02 8:02 ` Dmitry Baryshkov
2022-05-02 8:34 ` Marijn Suijten
2022-05-02 8:34 ` Marijn Suijten
2022-05-02 10:02 ` Dmitry Baryshkov [this message]
2022-05-02 10:02 ` Dmitry Baryshkov
2022-05-02 21:56 ` Marijn Suijten
2022-05-02 21:56 ` Marijn Suijten
2022-04-30 19:28 ` Dmitry Baryshkov
2022-04-30 19:28 ` Dmitry Baryshkov
2022-05-01 20:41 ` Marijn Suijten
2022-05-01 20:41 ` Marijn Suijten
2022-05-01 22:44 ` Dmitry Baryshkov
2022-05-01 22:44 ` Dmitry Baryshkov
2022-05-02 8:43 ` Marijn Suijten
2022-05-02 8:43 ` Marijn Suijten
2022-05-02 9:41 ` Dmitry Baryshkov
2022-05-02 9:41 ` Dmitry Baryshkov
2022-05-02 21:53 ` Marijn Suijten
2022-05-02 21:53 ` Marijn Suijten
2022-05-04 13:41 ` [Freedreno] " Vinod Koul
2022-05-04 13:41 ` Vinod Koul
2022-05-04 13:38 ` Vinod Koul
2022-05-04 13:38 ` Vinod Koul
2022-05-04 13:35 ` Vinod Koul
2022-05-04 13:35 ` Vinod Koul
2022-05-04 13:31 ` [Freedreno] " Vinod Koul
2022-05-04 13:31 ` Vinod Koul
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=f8986d6d-2c2b-a0a0-d63c-d4670f6864ff@linaro.org \
--to=dmitry.baryshkov@linaro.org \
--cc=airlied@linux.ie \
--cc=bjorn.andersson@linaro.org \
--cc=daniel@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=freedreno@lists.freedesktop.org \
--cc=linux-arm-msm@vger.kernel.org \
--cc=lkp@intel.com \
--cc=marijn.suijten@somainline.org \
--cc=quic_abhinavk@quicinc.com \
--cc=robdclark@gmail.com \
--cc=sean@poorly.run \
--cc=swboyd@chromium.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.