Req: about Polaris with RKL platform

[Alex Deucher]

On Tue, Aug 10, 2021 at 9:14 PM Koba Ko <koba.ko at canonical.com> wrote:
>
> On Tue, Aug 10, 2021 at 11:09 PM Alex Deucher <alexdeucher at gmail.com> wrote:
> >
> > On Tue, Aug 10, 2021 at 12:57 AM Koba Ko <koba.ko at canonical.com> wrote:
> > >
> > > On Tue, Aug 10, 2021 at 12:45 PM Mario Limonciello <superm1 at gmail.com> wrote:
> > > >
> > > >
> > > >
> > > > On Mon, Aug 9, 2021 at 9:37 AM Alex Deucher <alexdeucher at gmail.com> wrote:
> > > >>
> > > >> On Mon, Aug 9, 2021 at 9:59 AM Koba Ko <koba.ko at canonical.com> wrote:
> > > >> >
> > > >> > Previously, AMD had an issue about noise  with AMD-DG on the RKL platform
> > > >> > AMD provided a parameter.
> > > >> > #modprobe amdgpu ppfeaturemask=0xfff7bffb
> > > >> >
> > > >> >  I thought it's better to check and assign values in amd gpu.
> > > >> > Have a trouble determining the type of pch(RKL or else),
> > > >> > search in amd drm driver and can't find any about this.
> > > >> > Would someone please guide me? if there's an existing function.
> > > >> >
> > > >> > here's a proposal, check RKL PCH in amd driver,
> > > >> > the pch definitions must be splitted off from intel_pch.h in i915
> > > >> > folder to include/drm/intel_pch_definition.h
> > > >>
> > > >> Yes, something like that would work.
> > > >
> > > >
> > > > Can the issue that prompted this also happen with other ASIC with the
> > > > newer SMU families?  If so, should it probably be added to all of them
> > > > or further up in the code where the mask normally gets set from module
> > > > parameters to add the extra check there.
> > >
> > > Would amd guys please clarify this?
> > >
> > > Currently as i known,
> > > for smu series, amd upstream this commit only for smu7 and also
> > > provide modue parameters.
> > > 1.https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9d03730ecbc5afabfda26d4dbb014310bc4ea4d9
> > > 2. #modprobe amdgpu ppfeaturemask=0xfff7bffb
> >
> > SMU7-based and vega10/12/20 asics require this.  Newer parts were
> > fixed in SMU firmware:
> > navi10:
> > https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/commit/amdgpu/navi10_smc.bin?id=4fe6e53b96095101eebe4639cd2e2b6ecd84650d
> > navi14:
> > https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/commit/amdgpu/navi14_smc.bin?id=8ab7abaf63e95c29e04e5811cb24730a81486096
> > for newer asics, 21.10 and newer firmwares.  E.g.,
> > https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/commit/amdgpu/sienna_cichlid_smc.bin?id=ef5ea5d1d3f0a72a92e0a09f6cff253560374a39
> >
> With thie commit,
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9d03730ecbc5afabfda26d4dbb014310bc4ea4d9
> For SMU7-based, the module parameter is still necessary, right?

Yes, correct.

Alex


>
> > Alex
> >
> >
> > >
> > > >
> > > >>
> > > >> Alex
> > > >>
> > > >>
> > > >> >
> > > >> > > --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c
> > > >> > > +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c
> > > >> > > @@ -1629,7 +1629,7 @@ static void smu7_init_dpm_defaults(struct pp_hwmgr *hwmgr)
> > > >> > >
> > > >> > >         data->mclk_dpm_key_disabled = hwmgr->feature_mask & PP_MCLK_DPM_MASK ? false : true;
> > > >> > >         data->sclk_dpm_key_disabled = hwmgr->feature_mask & PP_SCLK_DPM_MASK ? false : true;
> > > >> > > -       data->pcie_dpm_key_disabled = hwmgr->feature_mask & PP_PCIE_DPM_MASK ? false : true;
> > > >> > > +       data->pcie_dpm_key_disabled = is_rkl_pch() || !(hwmgr->feature_mask & PP_PCIE_DPM_MASK);
> > > >> > >         /* need to set voltage control types before EVV patching */
> > > >> > >         data->voltage_control = SMU7_VOLTAGE_CONTROL_NONE;
> > > >> > >         data->vddci_control = SMU7_VOLTAGE_CONTROL_NONE;
> > > >
> > > >
> > > >
> > > > --
> > > > Mario Limonciello
> > > > superm1 at gmail.com