[v2,1/4] drm/msm/dsi: Adjust pclk rate for compression

Adjust the pclk rate to divide hdisplay by the compression ratio when DSC
is enabled.

Changes in v2:
- Adjusted pclk_rate math to divide only the hdisplay value by
  compression ratio

Signed-off-by: Jessica Zhang <quic_jesszhan@quicinc.com>
---
 drivers/gpu/drm/msm/dsi/dsi_host.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

On 2023-05-05 14:49:08, Jessica Zhang wrote:
> On 5/5/2023 2:23 PM, Jessica Zhang wrote:
> > Adjust the pclk rate to divide hdisplay by the compression ratio when DSC
> > is enabled.
> > 
> > Changes in v2:
> > - Adjusted pclk_rate math to divide only the hdisplay value by
> >    compression ratio
> > 
> > Signed-off-by: Jessica Zhang <quic_jesszhan@quicinc.com>
> > ---
> >   drivers/gpu/drm/msm/dsi/dsi_host.c | 17 +++++++++++++----
> >   1 file changed, 13 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
> > index 43a5ec33eee8..0e5778e8091f 100644
> > --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
> > +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
> > @@ -561,7 +561,8 @@ void dsi_link_clk_disable_v2(struct msm_dsi_host *msm_host)
> >   	clk_disable_unprepare(msm_host->byte_clk);
> >   }
> >   
> > -static unsigned long dsi_get_pclk_rate(const struct drm_display_mode *mode, bool is_bonded_dsi)
> > +static unsigned long dsi_get_pclk_rate(const struct drm_display_mode *mode,
> > +		struct drm_dsc_config *dsc, bool is_bonded_dsi)
> >   {
> >   	unsigned long pclk_rate;
> >   
> > @@ -576,6 +577,14 @@ static unsigned long dsi_get_pclk_rate(const struct drm_display_mode *mode, bool
> >   	if (is_bonded_dsi)
> >   		pclk_rate /= 2;
> >   
> > +	/* If DSC is enabled, divide hdisplay by compression ratio */
> > +	if (dsc) {
> > +		int new_hdisplay = DIV_ROUND_UP(mode->hdisplay * msm_dsc_get_bpp_int(dsc),
> > +				dsc->bits_per_component * 3);
> > +		int fps = DIV_ROUND_UP(pclk_rate, mode->htotal * mode->vtotal);
> 
> Should've used drm_mode_vrefresh() here... Will spin a v3 with that 
> change (along with any additional comments)

Perhaps unsigned long on some of these?  Overall the computations and
multi-lines look rather cluttered, perhaps (parts of) this is/are a
prime candidate to go into the new helpers?

Note that I cannot get the 4k mode working at 60Hz on one of my panels
(30Hz works with minor corruption), regardless of this patch.  See also:
https://gitlab.freedesktop.org/drm/msm/-/issues/24#note_1900031

> > +		pclk_rate = (new_hdisplay + (mode->htotal - mode->hdisplay)) * mode->vtotal * fps;
> > +	}
> > +
> >   	return pclk_rate;
> >   }
> >   
> > @@ -585,7 +594,7 @@ unsigned long dsi_byte_clk_get_rate(struct mipi_dsi_host *host, bool is_bonded_d
> >   	struct msm_dsi_host *msm_host = to_msm_dsi_host(host);
> >   	u8 lanes = msm_host->lanes;
> >   	u32 bpp = dsi_get_bpp(msm_host->format);
> > -	unsigned long pclk_rate = dsi_get_pclk_rate(mode, is_bonded_dsi);
> > +	unsigned long pclk_rate = dsi_get_pclk_rate(mode, msm_host->dsc, is_bonded_dsi);
> >   	u64 pclk_bpp = (u64)pclk_rate * bpp;
> >   
> >   	if (lanes == 0) {
> > @@ -604,7 +613,7 @@ unsigned long dsi_byte_clk_get_rate(struct mipi_dsi_host *host, bool is_bonded_d
> >   
> >   static void dsi_calc_pclk(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
> >   {
> > -	msm_host->pixel_clk_rate = dsi_get_pclk_rate(msm_host->mode, is_bonded_dsi);
> > +	msm_host->pixel_clk_rate = dsi_get_pclk_rate(msm_host->mode, msm_host->dsc, is_bonded_dsi);
> >   	msm_host->byte_clk_rate = dsi_byte_clk_get_rate(&msm_host->base, is_bonded_dsi,
> >   							msm_host->mode);
> >   
> > @@ -634,7 +643,7 @@ int dsi_calc_clk_rate_v2(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
> >   
> >   	dsi_calc_pclk(msm_host, is_bonded_dsi);
> >   
> > -	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, is_bonded_dsi) * bpp;
> > +	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, msm_host->dsc, is_bonded_dsi) * bpp;

Let's rebase on top of "drm/msm/dsi: simplify pixel clk rate handling"
[1] to clean this up.

[1]: https://lore.kernel.org/linux-arm-msm/20230118130031.2345941-1-dmitry.baryshkov@linaro.org/

- Marijn

> >   	do_div(pclk_bpp, 8);
> >   	msm_host->src_clk_rate = pclk_bpp;
> >   
> > 
> > -- 
> > 2.40.1
> >

On 5/8/2023 2:56 PM, Marijn Suijten wrote:
> On 2023-05-05 14:49:08, Jessica Zhang wrote:
>> On 5/5/2023 2:23 PM, Jessica Zhang wrote:
>>> Adjust the pclk rate to divide hdisplay by the compression ratio when DSC
>>> is enabled.
>>>
>>> Changes in v2:
>>> - Adjusted pclk_rate math to divide only the hdisplay value by
>>>     compression ratio
>>>
>>> Signed-off-by: Jessica Zhang <quic_jesszhan@quicinc.com>
>>> ---
>>>    drivers/gpu/drm/msm/dsi/dsi_host.c | 17 +++++++++++++----
>>>    1 file changed, 13 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/msm/dsi/dsi_host.c b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>> index 43a5ec33eee8..0e5778e8091f 100644
>>> --- a/drivers/gpu/drm/msm/dsi/dsi_host.c
>>> +++ b/drivers/gpu/drm/msm/dsi/dsi_host.c
>>> @@ -561,7 +561,8 @@ void dsi_link_clk_disable_v2(struct msm_dsi_host *msm_host)
>>>    	clk_disable_unprepare(msm_host->byte_clk);
>>>    }
>>>    
>>> -static unsigned long dsi_get_pclk_rate(const struct drm_display_mode *mode, bool is_bonded_dsi)
>>> +static unsigned long dsi_get_pclk_rate(const struct drm_display_mode *mode,
>>> +		struct drm_dsc_config *dsc, bool is_bonded_dsi)
>>>    {
>>>    	unsigned long pclk_rate;
>>>    
>>> @@ -576,6 +577,14 @@ static unsigned long dsi_get_pclk_rate(const struct drm_display_mode *mode, bool
>>>    	if (is_bonded_dsi)
>>>    		pclk_rate /= 2;
>>>    
>>> +	/* If DSC is enabled, divide hdisplay by compression ratio */
>>> +	if (dsc) {
>>> +		int new_hdisplay = DIV_ROUND_UP(mode->hdisplay * msm_dsc_get_bpp_int(dsc),
>>> +				dsc->bits_per_component * 3);
>>> +		int fps = DIV_ROUND_UP(pclk_rate, mode->htotal * mode->vtotal);
>>
>> Should've used drm_mode_vrefresh() here... Will spin a v3 with that
>> change (along with any additional comments)
> 
> Perhaps unsigned long on some of these?  Overall the computations and
> multi-lines look rather cluttered, perhaps (parts of) this is/are a
> prime candidate to go into the new helpers?

Hi Marijn,

Sorry for the late reply, wanted to get the MSM DSC helpers series 
settled first before addressing these changes.

Sounds good, I'll put these calculations in a separate 
dsi_adjust_compressed_pclk() helper.

> 
> Note that I cannot get the 4k mode working at 60Hz on one of my panels
> (30Hz works with minor corruption), regardless of this patch.  See also:
> https://gitlab.freedesktop.org/drm/msm/-/issues/24#note_1900031
As discussed elsewhere, we suspect that this is unrelated to DSC 
specifically and might be an issue with the upstream driver not taking 
transfer time into account with calculating pclk_rate.

We will look into this as a separate issue.

> 
>>> +		pclk_rate = (new_hdisplay + (mode->htotal - mode->hdisplay)) * mode->vtotal * fps;
>>> +	}
>>> +
>>>    	return pclk_rate;
>>>    }
>>>    
>>> @@ -585,7 +594,7 @@ unsigned long dsi_byte_clk_get_rate(struct mipi_dsi_host *host, bool is_bonded_d
>>>    	struct msm_dsi_host *msm_host = to_msm_dsi_host(host);
>>>    	u8 lanes = msm_host->lanes;
>>>    	u32 bpp = dsi_get_bpp(msm_host->format);
>>> -	unsigned long pclk_rate = dsi_get_pclk_rate(mode, is_bonded_dsi);
>>> +	unsigned long pclk_rate = dsi_get_pclk_rate(mode, msm_host->dsc, is_bonded_dsi);
>>>    	u64 pclk_bpp = (u64)pclk_rate * bpp;
>>>    
>>>    	if (lanes == 0) {
>>> @@ -604,7 +613,7 @@ unsigned long dsi_byte_clk_get_rate(struct mipi_dsi_host *host, bool is_bonded_d
>>>    
>>>    static void dsi_calc_pclk(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
>>>    {
>>> -	msm_host->pixel_clk_rate = dsi_get_pclk_rate(msm_host->mode, is_bonded_dsi);
>>> +	msm_host->pixel_clk_rate = dsi_get_pclk_rate(msm_host->mode, msm_host->dsc, is_bonded_dsi);
>>>    	msm_host->byte_clk_rate = dsi_byte_clk_get_rate(&msm_host->base, is_bonded_dsi,
>>>    							msm_host->mode);
>>>    
>>> @@ -634,7 +643,7 @@ int dsi_calc_clk_rate_v2(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
>>>    
>>>    	dsi_calc_pclk(msm_host, is_bonded_dsi);
>>>    
>>> -	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, is_bonded_dsi) * bpp;
>>> +	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, msm_host->dsc, is_bonded_dsi) * bpp;
> 
> Let's rebase on top of "drm/msm/dsi: simplify pixel clk rate handling"
> [1] to clean this up.
> 
> [1]: https://lore.kernel.org/linux-arm-msm/20230118130031.2345941-1-dmitry.baryshkov@linaro.org/

I've looked into this patch and have made a comment on it. Just have 
some reservations about it as it changes the functionality of a clk 
handler op.

I will hold off on rebasing and wait for that thread to resolve first.

Thanks,

Jessica Zhang

> 
> - Marijn
> 
>>>    	do_div(pclk_bpp, 8);
>>>    	msm_host->src_clk_rate = pclk_bpp;
>>>    
>>>
>>> -- 
>>> 2.40.1
>>>

On 2023-05-19 12:04:00, Jessica Zhang wrote:
<snip>
> >>> +	/* If DSC is enabled, divide hdisplay by compression ratio */
> >>> +	if (dsc) {
> >>> +		int new_hdisplay = DIV_ROUND_UP(mode->hdisplay * msm_dsc_get_bpp_int(dsc),
> >>> +				dsc->bits_per_component * 3);
> >>> +		int fps = DIV_ROUND_UP(pclk_rate, mode->htotal * mode->vtotal);
> >>
> >> Should've used drm_mode_vrefresh() here... Will spin a v3 with that
> >> change (along with any additional comments)
> > 
> > Perhaps unsigned long on some of these?  Overall the computations and
> > multi-lines look rather cluttered, perhaps (parts of) this is/are a
> > prime candidate to go into the new helpers?
> 
> Hi Marijn,
> 
> Sorry for the late reply, wanted to get the MSM DSC helpers series 
> settled first before addressing these changes.

No hurry and no worry, that is exactly why I requested this to be split
across multiple series so that we can make progress on that in isolation
(or rather, make progress on the first series in the chain before
iterating on the next).  On the other hand Dmitry made the right remark
that it does cause contention for some patches that only become relevant
in future series... but that's mostly down to how the patches are
distributed across series.

> Sounds good, I'll put these calculations in a separate 
> dsi_adjust_compressed_pclk() helper.

Not sure if "adjust" carries the meaning, but I'll leave it to you to
come up with an initial revision and then we can discuss.  I am mostly
curious if there are generic (DSI) timing rules that apply DRM-wide, or
if these would be MSM-specific.

Otherwise assigning them to properly named local variables is the
perfect way to self-document the code.

> > Note that I cannot get the 4k mode working at 60Hz on one of my panels
> > (30Hz works with minor corruption), regardless of this patch.  See also:
> > https://gitlab.freedesktop.org/drm/msm/-/issues/24#note_1900031
> As discussed elsewhere, we suspect that this is unrelated to DSC 
> specifically and might be an issue with the upstream driver not taking 
> transfer time into account with calculating pclk_rate.
> 
> We will look into this as a separate issue.

Yes that is very likely, but it is still a good idea to take into
account when looking into adjusting DSC timing: can we do that in any
sensible way without first accounting for transfer time?

<snip>
> >>>    	dsi_calc_pclk(msm_host, is_bonded_dsi);
> >>>    
> >>> -	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, is_bonded_dsi) * bpp;
> >>> +	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, msm_host->dsc, is_bonded_dsi) * bpp;
> > 
> > Let's rebase on top of "drm/msm/dsi: simplify pixel clk rate handling"
> > [1] to clean this up.
> > 
> > [1]: https://lore.kernel.org/linux-arm-msm/20230118130031.2345941-1-dmitry.baryshkov@linaro.org/
> 
> I've looked into this patch and have made a comment on it. Just have 
> some reservations about it as it changes the functionality of a clk 
> handler op.
> 
> I will hold off on rebasing and wait for that thread to resolve first.

Looks like the resolution was to drop it, but we should still first
apply the following hunk from it so that this line in your patch can be
skipped entirely.

    -	pclk_bpp = (u64)dsi_get_pclk_rate(msm_host->mode, is_bonded_dsi) * bpp;
    +	pclk_bpp = msm_host->pixel_clk_rate * bpp;

- Marijn