Re: [Freedreno] [PATCH v3 12/13] drm/msm/dsi: Add support for DSC configuration

[Vinod Koul <vkoul@kernel.org>]

On 17-02-22, 16:11, Marijn Suijten wrote:
> Hi Vinod,
> 
> Thanks for taking time to go through this review, please find some
> clarifications below.
> 
> On 2022-02-17 16:44:04, Vinod Koul wrote:
> > Hi Marijn,
> > 
> > On 11-12-21, 01:03, Marijn Suijten wrote:
> > 
> > > > +static int dsi_dsc_update_pic_dim(struct msm_display_dsc_config *dsc,
> > > > +				  int pic_width, int pic_height)
> > > 
> > > This function - adopted from downstream - does not seem to perform a
> > > whole lot, especially without the modulo checks against the slice size.
> > > Perhaps it can be inlined?
> > 
> > Most of the code here is :)
> > 
> > This was split from downstream code to check and update dimension. We
> > can inline this, or should we leave that to compiler. I am not a very
> > big fan of inlining...
> 
> It doesn't seem beneficial to code readability to have this function,
> which is only called just once and also has the same struct members read
> in a `DBG()` directly, abstracted away to a function.  Not really
> concerned about generated code/performance FWIW.
> 
> Also note that the caller isn't checking the `-EINVAL` result...

I have made this void inline.

> > > 
> > > > +{
> > > > +	if (!dsc || !pic_width || !pic_height) {
> > > > +		pr_err("DSI: invalid input: pic_width: %d pic_height: %d\n", pic_width, pic_height);
> > > > +		return -EINVAL;
> > > > +	}
> > > > +
> > > > +	dsc->drm->pic_width = pic_width;
> > > > +	dsc->drm->pic_height = pic_height;
> > > > +
> > > > +	return 0;
> > > > +}
> > > > +
> > > >  static void dsi_timing_setup(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
> > > >  {
> > > >  	struct drm_display_mode *mode = msm_host->mode;
> > > > @@ -940,7 +954,68 @@ static void dsi_timing_setup(struct msm_dsi_host *msm_host, bool is_bonded_dsi)
> > > >  		hdisplay /= 2;
> > > >  	}
> > > >  
> > > > +	if (msm_host->dsc) {
> > > > +		struct msm_display_dsc_config *dsc = msm_host->dsc;
> > > > +
> > > > +		/* update dsc params with timing params */
> > > > +		dsi_dsc_update_pic_dim(dsc, mode->hdisplay, mode->vdisplay);
> 
> That is, the result code here should be checked (or function inlined).

This function return void, so no point in checking

> > > > +
> > > > +		/* we do the calculations for dsc parameters here so that
> > > > +		 * panel can use these parameters
> > > > +		 */
> > > > +		dsi_populate_dsc_params(dsc);
> > > > +
> > > > +		/* Divide the display by 3 but keep back/font porch and
> > > > +		 * pulse width same
> > > > +		 */
> > > 
> > > A more general nit on the comments in this patch series: it is
> > > appreciated if comments explain the rationale rather than - or in
> > > addition to - merely paraphrasing the code that follows.
> > 
> > Yes it might be the case here, but in this case I wanted to explicitly
> > point out hat we need to divide display by 3...
> 
> The main point here is justifying _why_ there's a division by 3 for the
> active portion of the signal, I presume that's the compression ratio
> (having not read into the DSC compression standard yet at all)?

I have updated this comment

> > > > +		if (msm_host->dsc) {
> > > > +			struct msm_display_dsc_config *dsc = msm_host->dsc;
> > > > +			u32 reg, reg_ctrl, reg_ctrl2;
> > > > +			u32 slice_per_intf, bytes_in_slice, total_bytes_per_intf;
> > > > +
> > > > +			reg_ctrl = dsi_read(msm_host, REG_DSI_COMMAND_COMPRESSION_MODE_CTRL);
> > > > +			reg_ctrl2 = dsi_read(msm_host, REG_DSI_COMMAND_COMPRESSION_MODE_CTRL2);
> > > 
> > > Shouldn't old values be masked out first, before writing new bits or
> > > values below?  The video-mode variant doesn't read back old register
> > > values.
> > 
> > This follows downstream where the registers are read, modified and
> > written back
> 
> Are you sure about this?  The register values are never cleared, meaning
> that only bits get added through the `|=` below but never unset - unless
> downstream clears these registers elsewhere before ending up in (their
> equivalent of) dsi_timing_setup.

I have modified video mode to write and not read now. For command mode
all bits are set to some value so no need to mask old values for that

> Thanks.  I forgot to mention: there seem to be a lot of similarities
> between the video and commandmode computations, can those possibly be
> factored out of the if-else to save on duplication and accidental
> mismatches like these?

Thanks, this was a good suggestion and am happy to report that I have
incorporated this and indeed code looks better

-- 
~Vinod