Message-ID: <a6c0b929-383f-4541-b4f1-c2e54f547de4@ti.com>
Date: Wed, 25 Jun 2025 16:45:01 +0530
From: Jayesh Choudhary <j-choudhary@...com>
To: Tomi Valkeinen <tomi.valkeinen@...asonboard.com>, <jyri.sarha@....fi>,
        <dri-devel@...ts.freedesktop.org>, <devarsht@...com>,
        <mripard@...nel.org>
CC: <maarten.lankhorst@...ux.intel.com>, <tzimmermann@...e.de>,
        <airlied@...il.com>, <simona@...ll.ch>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] drm/tidss: Decouple max_pclk from tidss feats to
 remove clock dependency

Hello Tomi,

On 24/06/25 17:29, Tomi Valkeinen wrote:
> Hi,
> 
> On 18/06/2025 13:05, Jayesh Choudhary wrote:
>> TIDSS hardware by itself does not have variable max_pclk for each VP.
>> Each VP supports a fixed maximum pixel clock. K2 devices and AM62*
>> devices uses "ultra-light" version where each VP supports a max of
>> 300MHz whereas J7* devices uses TIDSS where all VP can support a
>> max pclk of 600MHz.
>> The limitation that has been modeled till now comes from the clock
>> (PLL can only be programmed to a particular max value). Due to this
>> we end up using different compatible for each SoC when the clocking
>> architecture changes for VPs, even when the hardware is essentially
>> the same.
>> max_pclk cannot be entirely removed since the display controller
>> should tell if a particular mode clock can be supported or not in crtc's
>> "mode_valid()" call. So remove "max_pclk_khz" from the static display
>> feat and add it to "tidss_device" structure which would be modified in
>> runtime. In mode_valid() call, check if a best frequency match for mode
>> clock can be found or not using "clk_round_rate()". Based on that,
>> propagate "max_pclk" and check max_clk again only if the requested mode
>> clock is greater than saved value. (As the preferred display mode is
>> usually the max resolution, driver ends up checking the maximum clock
>> the first time itself which is used in subsequent checks)
>> Since TIDSS display controller provides clock tolerance of 5%, we use
>> this while checking the max_pclk. Also, move up "dispc_pclk_diff()"
>> before it is called.
> 
> An empty line between paragraphs makes the desc easier to read.

Okay. Will add empty lines here.

> 
>> This will make the existing compatibles reusable.
>>
>> Signed-off-by: Jayesh Choudhary <j-choudhary@...com>
>> ---
>>
>> Changelog v1->v2:
>> - Rebase it on linux-next after OLDI support series[0] as all of its
>>    patches are reviewed and tested and it touches one of the functions
>>    used.
>>    
>> v1 patch link:
>> <https://lore.kernel.org/all/20250618075804.139844-1-j-choudhary@ti.com/>
>>
>> Test log on TI's J784S4 SoC with a couple of downstream patches
>> to integrate DSI support on one of the video ports:
>> <https://gist.github.com/Jayesh2000/ad4ab87028740efa60e5eb83fb892097>
>>
>>  From the logs, we can see that for CLK ID 218 (DSS), we do not have to
>> call sci_clk_determine_rate() multiple times. So there is very little
>> overhead of this call even with multiple mode_valid() called during
>> display run.
>>  From weston-simple-egl application, I have seen that there is no frame
>> drop or performance impact.
>>
>> Once this patch gets in, I will send patches for AM62P and J722S DSS
>> support.
>>
>> [0]: https://lore.kernel.org/all/20250528122544.817829-1-aradhya.bhatia@linux.dev/

One more observation here. This patch works absolutely fine in the
current tree (all SOCs that have TIDSS in upstream tree)
***This change is fully backward compatible.***

But after applying some downstream patches for OLDI support on
J722S/AM62P, I saw that clk_round_rate() causes issue in display
as the VP clock is not actually being handled by TIDSS, but actually it
is owned by OLDI and the clock request is for (7 * VP) value to
accound for the fixed clock divider present in case of OLDI panel.

In the OLDI support[0], I see that the serial clock is handled properly
and we also check the value after setting it in the driver. So we should
avoid performing clock operations for OLDI in TIDSS driver.

OLDI driver during tidss_oldi_init() stores the parent VP in 
oldi->parent_vp which can be used to avoid performing clk_round_rate()
in tidss driver, and in oldi, we can add atomic_check hook to propagate
max_pclk[] value for OLDI VP.

I will post next revision on top of [0] soon accounting this as well.
(Some oldi driver changes)

[0]: 
https://lore.kernel.org/all/20250528122544.817829-1-aradhya.bhatia@linux.dev/


>>
>>   drivers/gpu/drm/tidss/tidss_dispc.c | 74 ++++++++++++-----------------
>>   drivers/gpu/drm/tidss/tidss_dispc.h |  1 -
>>   drivers/gpu/drm/tidss/tidss_drv.h   |  2 +
>>   3 files changed, 33 insertions(+), 44 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/tidss/tidss_dispc.c b/drivers/gpu/drm/tidss/tidss_dispc.c
>> index c0277fa36425..ad9ffc3685b4 100644
>> --- a/drivers/gpu/drm/tidss/tidss_dispc.c
>> +++ b/drivers/gpu/drm/tidss/tidss_dispc.c
>> @@ -58,10 +58,6 @@ static const u16 tidss_k2g_common_regs[DISPC_COMMON_REG_TABLE_LEN] = {
>>   const struct dispc_features dispc_k2g_feats = {
>>   	.min_pclk_khz = 4375,
>>   
>> -	.max_pclk_khz = {
>> -		[DISPC_VP_DPI] = 150000,
>> -	},
>> -
>>   	/*
>>   	 * XXX According TRM the RGB input buffer width up to 2560 should
>>   	 *     work on 3 taps, but in practice it only works up to 1280.
>> @@ -144,11 +140,6 @@ static const u16 tidss_am65x_common_regs[DISPC_COMMON_REG_TABLE_LEN] = {
>>   };
>>   
>>   const struct dispc_features dispc_am65x_feats = {
>> -	.max_pclk_khz = {
>> -		[DISPC_VP_DPI] = 165000,
>> -		[DISPC_VP_OLDI_AM65X] = 165000,
>> -	},
>> -
>>   	.scaling = {
>>   		.in_width_max_5tap_rgb = 1280,
>>   		.in_width_max_3tap_rgb = 2560,
>> @@ -244,11 +235,6 @@ static const u16 tidss_j721e_common_regs[DISPC_COMMON_REG_TABLE_LEN] = {
>>   };
>>   
>>   const struct dispc_features dispc_j721e_feats = {
>> -	.max_pclk_khz = {
>> -		[DISPC_VP_DPI] = 170000,
>> -		[DISPC_VP_INTERNAL] = 600000,
>> -	},
>> -
>>   	.scaling = {
>>   		.in_width_max_5tap_rgb = 2048,
>>   		.in_width_max_3tap_rgb = 4096,
>> @@ -315,11 +301,6 @@ const struct dispc_features dispc_j721e_feats = {
>>   };
>>   
>>   const struct dispc_features dispc_am625_feats = {
>> -	.max_pclk_khz = {
>> -		[DISPC_VP_DPI] = 165000,
>> -		[DISPC_VP_INTERNAL] = 170000,
>> -	},
>> -
>>   	.scaling = {
>>   		.in_width_max_5tap_rgb = 1280,
>>   		.in_width_max_3tap_rgb = 2560,
>> @@ -380,10 +361,6 @@ const struct dispc_features dispc_am62a7_feats = {
>>   	 * if the code reaches dispc_mode_valid with VP1,
>>   	 * it should return MODE_BAD.
>>   	 */
>> -	.max_pclk_khz = {
>> -		[DISPC_VP_TIED_OFF] = 0,
>> -		[DISPC_VP_DPI] = 165000,
>> -	},
>>   
>>   	.scaling = {
>>   		.in_width_max_5tap_rgb = 1280,
>> @@ -441,10 +418,6 @@ const struct dispc_features dispc_am62a7_feats = {
>>   };
>>   
>>   const struct dispc_features dispc_am62l_feats = {
>> -	.max_pclk_khz = {
>> -		[DISPC_VP_DPI] = 165000,
>> -	},
>> -
>>   	.subrev = DISPC_AM62L,
>>   
>>   	.common = "common",
>> @@ -1347,25 +1320,48 @@ static void dispc_vp_set_default_color(struct dispc_device *dispc,
>>   			DISPC_OVR_DEFAULT_COLOR2, (v >> 32) & 0xffff);
>>   }
>>   
>> +/*
>> + * Calculate the percentage difference between the requested pixel clock rate
>> + * and the effective rate resulting from calculating the clock divider value.
>> + */
>> +unsigned int dispc_pclk_diff(unsigned long rate, unsigned long real_rate)
>> +{
>> +	int r = rate / 100, rr = real_rate / 100;
>> +
>> +	return (unsigned int)(abs(((rr - r) * 100) / r));
>> +}
>> +
>> +static int check_max_pixel_clock(struct dispc_device *dispc,
>> +				 u32 hw_videoport, unsigned long clock)
>> +{
>> +	if (clock > dispc->tidss->max_pclk[hw_videoport]) {
>> +		unsigned long round_clock = clk_round_rate(dispc->vp_clk[hw_videoport], clock);
>> +
>> +		if (dispc_pclk_diff(clock, round_clock) > 5)
>> +			return -EINVAL;
>> +
>> +		dispc->tidss->max_pclk[hw_videoport] = round_clock;
>> +	}
>> +
>> +	return 0;
>> +}
>> +
>>   enum drm_mode_status dispc_vp_mode_valid(struct dispc_device *dispc,
>>   					 u32 hw_videoport,
>>   					 const struct drm_display_mode *mode)
>>   {
>>   	u32 hsw, hfp, hbp, vsw, vfp, vbp;
>>   	enum dispc_vp_bus_type bus_type;
>> -	int max_pclk;
>>   
>>   	bus_type = dispc->feat->vp_bus_type[hw_videoport];
>>   
>> -	max_pclk = dispc->feat->max_pclk_khz[bus_type];
>> -
>> -	if (WARN_ON(max_pclk == 0))
>> +	if (bus_type == DISPC_VP_TIED_OFF)
>>   		return MODE_BAD;
>>   
>>   	if (mode->clock < dispc->feat->min_pclk_khz)
>>   		return MODE_CLOCK_LOW;
>>   
>> -	if (mode->clock > max_pclk)
>> +	if (check_max_pixel_clock(dispc, hw_videoport, mode->clock * 1000))
>>   		return MODE_CLOCK_HIGH;
>>   
>>   	if (mode->hdisplay > 4096)
>> @@ -1437,17 +1433,6 @@ void dispc_vp_disable_clk(struct dispc_device *dispc, u32 hw_videoport)
>>   	clk_disable_unprepare(dispc->vp_clk[hw_videoport]);
>>   }
>>   
>> -/*
>> - * Calculate the percentage difference between the requested pixel clock rate
>> - * and the effective rate resulting from calculating the clock divider value.
>> - */
>> -unsigned int dispc_pclk_diff(unsigned long rate, unsigned long real_rate)
>> -{
>> -	int r = rate / 100, rr = real_rate / 100;
>> -
>> -	return (unsigned int)(abs(((rr - r) * 100) / r));
>> -}
>> -
>>   int dispc_vp_set_clk_rate(struct dispc_device *dispc, u32 hw_videoport,
>>   			  unsigned long rate)
>>   {
>> @@ -3087,6 +3072,9 @@ int dispc_init(struct tidss_device *tidss)
>>   	}
>>   	dev_dbg(dev, "DSS fclk %lu Hz\n", clk_get_rate(dispc->fclk));
>>   
>> +	for (i = 0; i < dispc->feat->num_vps; i++)
>> +		dispc->tidss->max_pclk[i] = 0;
> 
> I think this is not needed, the struct should be zero initialized at alloc.

Okay will remove this.

> 
>>   	of_property_read_u32(dispc->dev->of_node, "max-memory-bandwidth",
>>   			     &dispc->memory_bandwidth_limit);
>>   
>> diff --git a/drivers/gpu/drm/tidss/tidss_dispc.h b/drivers/gpu/drm/tidss/tidss_dispc.h
>> index b8614f62186c..45b1a8aa9089 100644
>> --- a/drivers/gpu/drm/tidss/tidss_dispc.h
>> +++ b/drivers/gpu/drm/tidss/tidss_dispc.h
>> @@ -75,7 +75,6 @@ enum dispc_dss_subrevision {
>>   
>>   struct dispc_features {
>>   	int min_pclk_khz;
>> -	int max_pclk_khz[DISPC_VP_MAX_BUS_TYPE];
>>   
>>   	struct dispc_features_scaling scaling;
>>   
>> diff --git a/drivers/gpu/drm/tidss/tidss_drv.h b/drivers/gpu/drm/tidss/tidss_drv.h
>> index d14d5d28f0a3..59c67ae8e721 100644
>> --- a/drivers/gpu/drm/tidss/tidss_drv.h
>> +++ b/drivers/gpu/drm/tidss/tidss_drv.h
>> @@ -22,6 +22,8 @@ struct tidss_device {
>>   
>>   	const struct dispc_features *feat;
>>   	struct dispc_device *dispc;
>> +	long max_pclk[TIDSS_MAX_PORTS];
>> +
>>   
>>   	unsigned int num_crtcs;
>>   	struct drm_crtc *crtcs[TIDSS_MAX_PORTS];
> 
> One thing to keep in mind is that if we ever change the source clk, we
> need to also clear the max_pclk for that VP.

So I will clear them (mark as 0) in tidss_remove?

> 
> Shouldn't we still have a check for the DSS internal max pclk somewhere?

Makes sense. So instead of moving max_pclk_khz from dispc_features to
tidss_device, I will have it in both. We can have int max_pclk_khz
similar to min_pclk_khz which would be static for a tidss device.

And then I will rename max_pclk[] in tidss_device structure to
max_pclk_vp[].

> 
>   Tomi
> 

Warm Regards,
Jayesh

Powered by blists - more mailing lists