[PATCH] drm/amd/pm: setup APU dpm clock table in SMU HW initialization

Dirk Gouders dirk at gouders.net
Wed Sep 30 09:53:31 UTC 2020


Evan Quan <evan.quan at amd.com> writes:

> As the dpm clock table is needed during DC HW initialization.
> And that (DC HW initialization) comes before smu_late_init()
> where current APU dpm clock table setup is performed. So, NULL
> pointer dereference will be triggered. By moving APU dpm clock
> table setup to smu_hw_init(), this can be avoided.

Thanks for the quick response.  I tested the patch and it fixes the call
trace I initially reportet (#1 in the table below).  #2 is unaffected by
this patch.  I could try to bisect it as well bud did not do it, so far.

Probably, I caused some confusion in the original thread and I will try to
order it a bit.  What I noticed is:

         with amdgpu.discovery value    | noticed issue
         ===============================================================
1)                unset or "1"          | call trace because of
                                        | assert(0) in rn_clk_mgr_helper_populate_bw_params()
         -------------------------------+-------------------------------
2)                      0               | NULL pointer dereference in soc15_set_ip_blocks()

This patch fixes #1, i.e. avoids the assert() in following code in
drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr.c

	for (i = PP_SMU_NUM_FCLK_DPM_LEVELS - 1; i >= 0; i--) {
		if (clock_table->FClocks[i].Freq != 0 && clock_table->FClocks[i].Vol != 0) {
			j = i;
			break;
		}
	}

	if (j == -1) {
		/* clock table is all 0s, just use our own hardcode */
		ASSERT(0);
		return;
	}

To me, the commit message sounds as if the patch fixes #2 whereas it
really is #1 that gets fixed.  I also wonder if we probably want a
fixes-line for completeness:

Fixes: 02cf91c113ea (drm/amd/powerplay: postpone operations not required for hw setup to late_init)

Dirk

> Change-Id: I2bb1f9ba26f9c8820c08241da62f7be64ab75840
> Signed-off-by: Evan Quan <evan.quan at amd.com>
> Reported-by: Dirk Gouders <dirk at gouders.net>
> ---
>  drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 22 +++++++++++-----------
>  1 file changed, 11 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> index f46cf9ea355e..8f6045def272 100644
> --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> @@ -482,17 +482,6 @@ static int smu_late_init(void *handle)
>  		return ret;
>  	}
>  
> -	/*
> -	 * Set initialized values (get from vbios) to dpm tables context such as
> -	 * gfxclk, memclk, dcefclk, and etc. And enable the DPM feature for each
> -	 * type of clks.
> -	 */
> -	ret = smu_set_default_dpm_table(smu);
> -	if (ret) {
> -		dev_err(adev->dev, "Failed to setup default dpm clock tables!\n");
> -		return ret;
> -	}
> -
>  	ret = smu_populate_umd_state_clk(smu);
>  	if (ret) {
>  		dev_err(adev->dev, "Failed to populate UMD state clocks!\n");
> @@ -1021,6 +1010,17 @@ static int smu_smc_hw_setup(struct smu_context *smu)
>  		return ret;
>  	}
>  
> +	/*
> +	 * Set initialized values (get from vbios) to dpm tables context such as
> +	 * gfxclk, memclk, dcefclk, and etc. And enable the DPM feature for each
> +	 * type of clks.
> +	 */
> +	ret = smu_set_default_dpm_table(smu);
> +	if (ret) {
> +		dev_err(adev->dev, "Failed to setup default dpm clock tables!\n");
> +		return ret;
> +	}
> +
>  	ret = smu_notify_display_change(smu);
>  	if (ret)
>  		return ret;


More information about the amd-gfx mailing list