[Public]
-----Original Message-----
From: Chen, Guchun <guchun.c...@amd.com>
Sent: Friday, March 24, 2023 3:26 PM
To: Huang, Tim <tim.hu...@amd.com>; amd-gfx@lists.freedesktop.org
Cc: Deucher, Alexander <alexander.deuc...@amd.com>; Zhang, Yifan
<yifan1.zh...@amd.com>; Ma, Li <li...@amd.com>; Du, Xiaojian
<xiaojian...@amd.com>; Huang, Tim <tim.hu...@amd.com>
Subject: RE: [PATCH] drm/amd/pm: re-enable the gfx imu when smu resume
> -----Original Message-----
> From: amd-gfx <amd-gfx-boun...@lists.freedesktop.org> On Behalf Of Tim
> Huang
> Sent: Friday, March 24, 2023 3:08 PM
> To: amd-gfx@lists.freedesktop.org
> Cc: Deucher, Alexander <alexander.deuc...@amd.com>; Zhang, Yifan
> <yifan1.zh...@amd.com>; Ma, Li <li...@amd.com>; Du, Xiaojian
> <xiaojian...@amd.com>; Huang, Tim <tim.hu...@amd.com>
> Subject: [PATCH] drm/amd/pm: re-enable the gfx imu when smu resume
>
> If the gfx imu is poweroff when suspend, then it need to be re-enabled
> when resume.
>
> Signed-off-by: Tim Huang <tim.hu...@amd.com>
> ---
> drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c | 40 ++++++++++++++++-
> ------
> 1 file changed, 28 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> index b5d64749990e..94fe8593444a 100644
> --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
> @@ -162,10 +162,15 @@ int smu_get_dpm_freq_range(struct smu_context
> *smu,
>
> int smu_set_gfx_power_up_by_imu(struct smu_context *smu) {
> - if (!smu->ppt_funcs || !smu->ppt_funcs->set_gfx_power_up_by_imu)
> - return -EOPNOTSUPP;
> + int ret = 0;
> + struct amdgpu_device *adev = smu->adev;
>
> - return smu->ppt_funcs->set_gfx_power_up_by_imu(smu);
> + if (smu->ppt_funcs->set_gfx_power_up_by_imu) {
> + ret = smu->ppt_funcs->set_gfx_power_up_by_imu(smu);
> + if (ret)
> + dev_err(adev->dev, "Failed to enable gfx imu!\n");
> + }
> + return ret;
> }
>
> static u32 smu_get_mclk(void *handle, bool low) @@ -196,6 +201,19 @@
> static u32 smu_get_sclk(void *handle, bool low)
> return clk_freq * 100;
> }
>
> +static int smu_set_gfx_imu_enable(struct smu_context *smu) {
> + struct amdgpu_device *adev = smu->adev;
> +
> + if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
> + return 0;
> +
> + if (amdgpu_in_reset(smu->adev) || adev->in_s0ix)
> + return 0;
> +
> + return smu_set_gfx_power_up_by_imu(smu); }
> +
> static int smu_dpm_set_vcn_enable(struct smu_context *smu,
> bool enable)
> {
> @@ -1396,15 +1414,9 @@ static int smu_hw_init(void *handle)
> }
>
> if (smu->is_apu) {
> - if ((smu->ppt_funcs->set_gfx_power_up_by_imu) &&
> - likely(adev->firmware.load_type ==
> AMDGPU_FW_LOAD_PSP)) {
> - ret = smu->ppt_funcs-
> >set_gfx_power_up_by_imu(smu);
> - if (ret) {
> - dev_err(adev->dev, "Failed to Enable gfx
> imu!\n");
> - return ret;
> - }
> - }
> -
> + ret = smu_set_gfx_imu_enable(smu);
> + if (ret)
> + return ret;
> smu_dpm_set_vcn_enable(smu, true);
> smu_dpm_set_jpeg_enable(smu, true);
> smu_set_gfx_cgpg(smu, true);
> @@ -1681,6 +1693,10 @@ static int smu_resume(void *handle)
> return ret;
> }
>
> + ret = smu_set_gfx_imu_enable(smu);
> + if (ret)
> + return ret;
>> smu_set_gfx_imu_enable in smu_hw_init is valid when sum->is_apu is true. So
>> such check is still necessary in smu_resume?
>>Regards,
>>Guchun
For both hw_init or resume, it should be ok if not have the smu->is_apu
checking. Because we will check the valid of
smu->ppt_funcs->set_gfx_power_up_by_imu in the final call path,
so, for dGPU or APUs without IMU that did not implement this interface will
return 0. Thanks.
Tim
> smu_set_gfx_cgpg(smu, true);
>
> smu->disable_uclk_switch = 0;
> --
> 2.25.1
