AMD General > -----Original Message----- > From: Lazar, Lijo <[email protected]> > Sent: Friday, May 8, 2026 1:33 PM > To: Wang, Yang(Kevin) <[email protected]>; amd- > [email protected] > Cc: Deucher, Alexander <[email protected]>; Zhang, Hawking > <[email protected]>; Feng, Kenneth <[email protected]> > Subject: Re: [PATCH] drm/amd/pm: update dpm clock pm attributes for > aldebaran (gc 9.4.2) > > > > On 08-May-26 10:21 AM, Wang, Yang(Kevin) wrote: > > AMD General > > > > > > > > ---------------------------------------------------------------------- > > -- > > *From:* Lazar, Lijo <[email protected]> > > *Sent:* Friday, May 8, 2026 12:35 PM > > *To:* Wang, Yang(Kevin) <[email protected]>; amd- > > [email protected] <[email protected]> > > *Cc:* Deucher, Alexander <[email protected]>; Zhang, > Hawking > > <[email protected]>; Feng, Kenneth <[email protected]> > > *Subject:* Re: [PATCH] drm/amd/pm: update dpm clock pm attributes for > > aldebaran (gc 9.4.2) > > > > > > On 08-May-26 8:50 AM, Yang Wang wrote: > > > Split and correct DPM clock attribute constraints for GFX 9.4.1 > > (Arcturus) > > > and 9.4.2 (Aldebaran) ASICs: > > > > > > - Arcturus (9.4.1): Keep restrictions for mclk/socclk/fclk, update > > comment > - Aldebaran (9.4.2): Add dedicated per-clock constraints: > > > * mclk/socclk: Only support voltage control, disable level > > setting > * fclk: Mark as fully unsupported > * pcie: Move > > multi-AID check into 9.4.2 case, disable level setting > > > > Signed-off-by: Yang Wang <[email protected]> > --- > > > drivers/gpu/drm/amd/pm/amdgpu_pm.c | 26 ++++++++++++++++++++------ > > > > 1 file changed, 20 insertions(+), 6 deletions(-) > > diff --git > > a/drivers/gpu/drm/amd/pm/amdgpu_pm.c b/drivers/gpu/drm/ > > amd/pm/amdgpu_pm.c > index fd2e63530e8c..9ee3053973cf 100644 > --- > > a/drivers/gpu/drm/amd/pm/amdgpu_pm.c > > > +++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c > > > @@ -2033,16 +2033,11 @@ static int > > pp_dpm_clk_default_attr_update(struct amdgpu_device *adev, struct amd > > > gc_ver == IP_VERSION(11, 0, 2) || > > > gc_ver == IP_VERSION(11, 0, 3)) && adev- >vcn.num_vcn_inst >= > 2)) > > > *states = ATTR_STATE_UNSUPPORTED; > - } else if > > (DEVICE_ATTR_IS(pp_dpm_pcie)) { > - if (gc_ver == > > IP_VERSION(9, 4, 2) || > - amdgpu_is_multi_aid(adev)) > > > - *states = ATTR_STATE_UNSUPPORTED; > } > > > > switch (gc_ver) { > case IP_VERSION(9, 4, 1): > > > - case IP_VERSION(9, 4, 2): > > > - /* the Mi series card does not support standalone > > mclk/ socclk/fclk level setting */ > + /* Arcturus does > > not support standalone mclk/socclk/ fclk level setting */ > > > if (DEVICE_ATTR_IS(pp_dpm_mclk) || > > > DEVICE_ATTR_IS(pp_dpm_socclk) || > > > DEVICE_ATTR_IS(pp_dpm_fclk)) { > @@ -2050,6 +2045,25 @@ static int > > pp_dpm_clk_default_attr_update(struct amdgpu_device *adev, struct amd > > > dev_attr->store = NULL; > } > > > break; > + case IP_VERSION(9, 4, 2): > > > + if (DEVICE_ATTR_IS(pp_dpm_mclk) || > + > > DEVICE_ATTR_IS(pp_dpm_socclk)) { > + /* Aldebaran > > mclk/socclk DPM only supports voltage control, > + > > * not allow to set dpm level directly */ > + > > dev_attr->attr.mode &= ~S_IWUGO; > + > > dev_attr->store = NULL; > + } else if > > (DEVICE_ATTR_IS(pp_dpm_pcie)) { > + if > > (amdgpu_is_multi_aid(adev)) { > + *states > > = ATTR_STATE_UNSUPPORTED; > + } else { > + > > dev_attr->attr.mode &= ~S_IWUGO; > + > > dev_attr->store = NULL; > + } > + } > > else if (DEVICE_ATTR_IS(pp_dpm_fclk)) { > + /* > > Aldebaran does not support fclk dpm */ > + > > *states = ATTR_STATE_UNSUPPORTED; > > > > I see FCLK reporting is supported in aldebaran - > > aldebaran_get_current_clk_freq_by_table. > > > > That may also be kept as read-only then. > > [kevin] > > No, it should be disabled, because the fclk is always return 0 by > > metrics table if fclk dpm feature is disabled. > > > > In that case, aldebaran_emit_clk_levels etc. be modified as well to reflect > this > since FCLK DPM is always disabled?
Since we already mark FCLK as unsupported in sysfs attribute, there is no need to modify aldebaran_emit_clk_levels and related functions. Those paths won't be invoked anyway, and extra changes are unnecessary for this fix. So, I prefer keep it, and only keep changes in amdgpu_pm.c source file for this issue. Best Regards, Kevin > > Thanks, > Lijo > > Best Regards, > > Kevin > > > > Thanks, > > Lijo> + } > > > + break; > > > default: > > > break; > > > } > >
