[PATCH v2] drm/amdgpu/OLAND: clip the ref divider max value

Christian König christian.koenig at amd.com
Fri Aug 20 08:05:39 UTC 2021


Uff, I think that was SI but could be CIK as well.

We have a table for this somewhere, but I don't have it at hand.

Regards,
Christian.

Am 20.08.21 um 09:43 schrieb Sharma, Shashank:
> [AMD Official Use Only]
>
> Agree, on the similar note, which Gen is OLAND BTW 😊 ?
>
> Regards
> Shashank
> -----Original Message-----
> From: Koenig, Christian <Christian.Koenig at amd.com>
> Sent: Friday, August 20, 2021 12:08 PM
> To: Sharma, Shashank <Shashank.Sharma at amd.com>; amd-gfx at lists.freedesktop.org
> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>
> Subject: Re: [PATCH v2] drm/amdgpu/OLAND: clip the ref divider max value
>
> Sounds like a good idea to me, but I would limit this generally or at least for the whole generation and not just one particular chipset.
>
> Regards,
> Christian.
>
> Am 20.08.21 um 08:05 schrieb Sharma, Shashank:
>>  From 4841e5ba60e33ff798bde6cb69fbd7e137b6db9c Mon Sep 17 00:00:00 2001
>> From: Shashank Sharma <shashank.sharma at amd.com>
>> Date: Fri, 20 Aug 2021 10:20:02 +0530
>> Subject: [PATCH v2] drm/amdgpu/OLAND: clip the ref divider max value
>> MIME-Version: 1.0
>> Content-Type: text/plain; charset=UTF-8
>> Content-Transfer-Encoding: 8bit
>>
>> This patch limits the ref_div_max value to 100, during the calculation
>> of PLL feedback reference divider. With current value (128), the
>> produced fb_ref_div value generates unstable output at particular
>> frequencies. Radeon driver limits this value at 100.
>>
>> On Oland, when we try to setup mode 2048x1280 at 60 (a bit weird, I
>> know), it demands a clock of 221270 Khz. It's been observed that the
>> PLL calculations using values 128 and 100 are vastly different, and
>> look like this:
>>
>> +------------------------------------------+
>> |Parameter    |AMDGPU        |Radeon       |
>> |             |              |             |
>> +-------------+----------------------------+
>> |Clock feedback              |             | divider max  |  128
>> ||   100       | cap value    |              |             |
>> |             |              |             |
>> |             |              |             |
>> +------------------------------------------+
>> |ref_div_max  |              |             |
>> |             |  42          |  20         |
>> |             |              |             |
>> |             |              |             |
>> +------------------------------------------+
>> |ref_div      |  42          |  20         |
>> |             |              |             |
>> +------------------------------------------+
>> |fb_div       |  10326       |  8195       |
>> +------------------------------------------+
>> |fb_div       |  1024        |  163        |
>> +------------------------------------------+
>> |fb_dev_p     |  4           |  9          | frac fb_de^_p|
>> ||             |
>> +----------------------------+-------------+
>>
>> With ref_div_max value clipped at 100, AMDGPU driver can also drive
>> videmode 2048x1280 at 60 (221Mhz) and produce proper output without any
>> blanking and distortion on the screen.
>>
>> PS: This value was changed from 128 to 100 in Radeon driver also, here:
>> https://github.com/freedesktop/drm-tip/commit/4b21ce1b4b5d262e7d4656b8
>> ececc891fc3cb806
>>
>>
>> V1:
>> Got acks from:
>> Acked-by: Alex Deucher <alexander.deucher at amd.com>
>> Acked-by: Christian König <christian.koenig at amd.com>
>>
>> V2:
>> - Restricting the changes only for OLAND, just to avoid any regression
>>    for other cards.
>> - Changed unsigned -> unsigned int to make checkpatch quiet.
>>
>> Cc: Alex Deucher <Alexander.Deucher at amd.com>
>> Cc: Christian König <christian.koenig at amd.com>
>> Cc: Eddy Qin <Eddy.Qin at amd.com>
>> Signed-off-by: Shashank Sharma <shashank.sharma at amd.com>
>> ---
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c    | 20 +++++++++++++-------
>>   drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h    |  3 ++-
>>   drivers/gpu/drm/amd/amdgpu/atombios_crtc.c |  2 +-
>>   3 files changed, 16 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
>> index f2e20666c9c1..6d04c1d25bfb 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.c
>> @@ -80,12 +80,17 @@ static void amdgpu_pll_reduce_ratio(unsigned *nom,
>> unsigned *den,
>>    * Calculate feedback and reference divider for a given post divider.
>> Makes
>>    * sure we stay within the limits.
>>    */
>> -static void amdgpu_pll_get_fb_ref_div(unsigned nom, unsigned den,
>> unsigned post_div,
>> -                      unsigned fb_div_max, unsigned ref_div_max,
>> -                      unsigned *fb_div, unsigned *ref_div)
>> +static void amdgpu_pll_get_fb_ref_div(struct amdgpu_device *adev,
>> unsigned int nom,
>> +                      unsigned int den, unsigned int post_div,
>> +                      unsigned int fb_div_max, unsigned int
>> +ref_div_max,
>> +                      unsigned int *fb_div, unsigned int *ref_div)
>>   {
>> +
>>       /* limit reference * post divider to a maximum */
>> -    ref_div_max = min(128 / post_div, ref_div_max);
>> +    if (adev->asic_type == CHIP_OLAND)
>> +        ref_div_max = min(100 / post_div, ref_div_max);
>> +    else
>> +        ref_div_max = min(128 / post_div, ref_div_max);
>>
>>       /* get matching reference and feedback divider */
>>       *ref_div = min(max(DIV_ROUND_CLOSEST(den, post_div), 1u),
>> ref_div_max); @@ -112,7 +117,8 @@ static void
>> amdgpu_pll_get_fb_ref_div(unsigned
>> nom, unsigned den, unsigned post_
>>    * Try to calculate the PLL parameters to generate the given frequency:
>>    * dot_clock = (ref_freq * feedback_div) / (ref_div * post_div)
>>    */
>> -void amdgpu_pll_compute(struct amdgpu_pll *pll,
>> +void amdgpu_pll_compute(struct amdgpu_device *adev,
>> +            struct amdgpu_pll *pll,
>>               u32 freq,
>>               u32 *dot_clock_p,
>>               u32 *fb_div_p,
>> @@ -199,7 +205,7 @@ void amdgpu_pll_compute(struct amdgpu_pll *pll,
>>
>>       for (post_div = post_div_min; post_div <= post_div_max;
>> ++post_div) {
>>           unsigned diff;
>> -        amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max,
>> +        amdgpu_pll_get_fb_ref_div(adev, nom, den, post_div,
>> +fb_div_max,
>>                         ref_div_max, &fb_div, &ref_div);
>>           diff = abs(target_clock - (pll->reference_freq * fb_div) /
>>               (ref_div * post_div));
>> @@ -214,7 +220,7 @@ void amdgpu_pll_compute(struct amdgpu_pll *pll,
>>       post_div = post_div_best;
>>
>>       /* get the feedback and reference divider for the optimal value
>> */
>> -    amdgpu_pll_get_fb_ref_div(nom, den, post_div, fb_div_max,
>> ref_div_max,
>> +    amdgpu_pll_get_fb_ref_div(adev, nom, den, post_div, fb_div_max,
>> ref_div_max,
>>                     &fb_div, &ref_div);
>>
>>       /* reduce the numbers to a simpler ratio once more */ diff --git
>> a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
>> index db6136f68b82..44a583d6c9b4 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_pll.h
>> @@ -24,7 +24,8 @@
>>   #ifndef __AMDGPU_PLL_H__
>>   #define __AMDGPU_PLL_H__
>>
>> -void amdgpu_pll_compute(struct amdgpu_pll *pll,
>> +void amdgpu_pll_compute(struct amdgpu_device *adev,
>> +             struct amdgpu_pll *pll,
>>                u32 freq,
>>                u32 *dot_clock_p,
>>                u32 *fb_div_p,
>> diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
>> b/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
>> index 159a2a4385a1..afad094f84c2 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/atombios_crtc.c
>> @@ -851,7 +851,7 @@ void amdgpu_atombios_crtc_set_pll(struct drm_crtc
>> *crtc, struct drm_display_mode
>>       pll->reference_div = amdgpu_crtc->pll_reference_div;
>>       pll->post_div = amdgpu_crtc->pll_post_div;
>>
>> -    amdgpu_pll_compute(pll, amdgpu_crtc->adjusted_clock, &pll_clock,
>> +    amdgpu_pll_compute(adev, pll, amdgpu_crtc->adjusted_clock,
>> &pll_clock,
>>                   &fb_div, &frac_fb_div, &ref_div, &post_div);
>>
>>       amdgpu_atombios_crtc_program_ss(adev, ATOM_DISABLE,
>> amdgpu_crtc->pll_id,



More information about the amd-gfx mailing list