On Mon, 11 May 2026, Matt Roper <[email protected]> wrote:
> On Mon, May 11, 2026 at 01:30:59PM -0300, Gustavo Sousa wrote:
>> Just like it is done for the platform-specific bandwidth parameters, use
>> a separate function named get_display_bw_params() to return the display
>> IP-specific parameters. This simplifies intel_bw_init_hw() by having
>> just one call for each of the *_get_bw_info() functions.
>>
>> v2:
>> - Prefer to call get_display_bw_params() only once in
>> intel_bw_init_hw() instead of having multiple calls in each of the
>> affected *_get_bw_info() functions. (Jani)
>>
>> Cc: Jani Nikula <[email protected]>
>> Signed-off-by: Gustavo Sousa <[email protected]>
>> ---
>> drivers/gpu/drm/i915/display/intel_bw.c | 36
>> +++++++++++++++++++++------------
>> 1 file changed, 23 insertions(+), 13 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/i915/display/intel_bw.c
>> b/drivers/gpu/drm/i915/display/intel_bw.c
>> index c01356d38e64..acd1b6901b46 100644
>> --- a/drivers/gpu/drm/i915/display/intel_bw.c
>> +++ b/drivers/gpu/drm/i915/display/intel_bw.c
>> @@ -493,6 +493,26 @@ static const struct intel_display_bw_params
>> xelpdp_bw_params = {
>> .displayrtids = 256,
>> };
>>
>> +static const struct intel_display_bw_params *get_display_bw_params(struct
>> intel_display *display)
>> +{
>> + if (DISPLAY_VER(display) >= 14) {
>> + return &xelpdp_bw_params;
>> + } else if (DISPLAY_VER(display) >= 12) {
>> + /*
>> + * RKL's SoC was based on ICL and the display, even though being
>> + * gen12, had changes to the memory interface to match gen11's,
>> + * consequently inheriting gen11's display-specific bandwidth
>> + * parameters.
>> + */
>> + if (display->platform.rocketlake)
>> + return &gen11_bw_params;
>> + else
>> + return &gen12_bw_params;
>> + } else {
>> + return &gen11_bw_params;
>
> It doesn't really matter, but this is technically going to assign gen11
> parameters for all the pre-gen11 platforms that call through here on
> i915. If we never use the values it probably doesn't hurt anything, but
> it might be best to make this a condition on gen11 rather than an 'else'
> just to avoid any confusion.
>
>
> Matt
>
>> + }
>> +}
>> +
>> static int icl_get_bw_info(struct intel_display *display,
>> const struct dram_info *dram_info,
>> const struct intel_soc_bw_params *soc_bw_params,
>> @@ -843,6 +863,7 @@ void intel_bw_init_hw(struct intel_display *display)
>> {
>> const struct dram_info *dram_info = intel_dram_info(display);
>> const struct intel_soc_bw_params *soc_bw_params =
>> get_soc_bw_params(display);
>> + const struct intel_display_bw_params *display_bw_params =
>> get_display_bw_params(display);
Feels like it gets increasingly weird to call all these functions
unconditionally when we bail out for !display right below.
BR,
Jani.
>>
>> if (!HAS_DISPLAY(display))
>> return;
>> @@ -858,23 +879,12 @@ void intel_bw_init_hw(struct intel_display *display)
>>
>> if (DISPLAY_VERx100(display) >= 1401 && display->platform.dgfx) {
>> xe2_hpd_get_bw_info(display, dram_info, soc_bw_params);
>> - } else if (DISPLAY_VER(display) >= 14) {
>> - tgl_get_bw_info(display, dram_info, soc_bw_params,
>> &xelpdp_bw_params);
>> } else if (display->platform.dg2) {
>> dg2_get_bw_info(display);
>> } else if (DISPLAY_VER(display) >= 12) {
>> - /*
>> - * RKL's SoC was based on ICL and the display, even though being
>> - * gen12, had changes to the memory interface to match gen11's,
>> - * consequently inheriting gen11's display-specific bandwidth
>> - * parameters.
>> - */
>> - if (display->platform.rocketlake)
>> - tgl_get_bw_info(display, dram_info, soc_bw_params,
>> &gen11_bw_params);
>> - else
>> - tgl_get_bw_info(display, dram_info, soc_bw_params,
>> &gen12_bw_params);
>> + tgl_get_bw_info(display, dram_info, soc_bw_params,
>> display_bw_params);
>> } else if (DISPLAY_VER(display) == 11) {
>> - icl_get_bw_info(display, dram_info, soc_bw_params,
>> &gen11_bw_params);
>> + icl_get_bw_info(display, dram_info, soc_bw_params,
>> display_bw_params);
>> }
>> }
>>
>>
>> --
>> 2.53.0
>>
--
Jani Nikula, Intel
