On Tue, Dec 31, 2019 at 2:33 AM Jani Nikula <jani.nikula@xxxxxxxxxxxxxxx> wrote:
>
> On Mon, 23 Dec 2019, Lucas De Marchi <lucas.demarchi@xxxxxxxxx> wrote:
> > Identify 3 possible cases in which the index numbers can be different
> > from the "port" and add them to the description-based ddi initialization
> > table. This can be used in place of additional functions mapping from
> > on to the other. Right now we already cover part of this by creating kind of
> > virtual phy numbering, but that comes with downsides:
> >
> > a) there's not really a "phy numbering" in the spec, this is purely a
> > software thing; hardware uses whatever they want thinking mapping from
> > one to the other arbitrarily is easy in software.
> >
> > b) currently the mapping occurs on "leaf" functions, making the decision
> > based on the platform.
> >
> > With this new table the approach will be: the port as defined by the
> > enum port is purely a driver convention and won't be used anymore to
> > define the register offset or register bits. For that we have the other
> > 3 indexes, identified as being possibly different from the current usage
> > of register bits: ddi, vbt and phy. The phy type is also added here,
> > meant to replace the checks for combo vs tc (although the helper
> > functions can remain so we may differentiate between, e.g. Dekel and MG
> > phys).
>
> I'm not sure how the vbt_idx is supposed to be used (because it's
> actually not used anywhere here). I would like to reduce the amount of
> VBT info spread around in the driver, so I'd really need to know more.
The index the VBT uses to refer to a DDI may not match the index used by
the display engine, that is why I'm adding it here.
Lucas De Marchi
>
> > While at it, also give names to the ports so they can be easily
> > identified.
> >
> > Signed-off-by: Lucas De Marchi <lucas.demarchi@xxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/display/intel_display.c | 54 +++++++++----------
> > drivers/gpu/drm/i915/display/intel_display.h | 7 +++
> > .../drm/i915/display/intel_display_types.h | 5 ++
> > 3 files changed, 39 insertions(+), 27 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c
> > index ad85cf75c815..219f180fa395 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.c
> > +++ b/drivers/gpu/drm/i915/display/intel_display.c
> > @@ -16277,14 +16277,14 @@ static bool ddi_is_port_present(struct drm_i915_private *i915,
> > static const struct intel_output tgl_output = {
> > .dsi_init = icl_dsi_init,
> > .ddi_ports = {
> > - { .port = PORT_A },
> > - { .port = PORT_B },
> > - { .port = PORT_D },
> > - { .port = PORT_E },
> > - { .port = PORT_F },
> > - { .port = PORT_G },
> > - { .port = PORT_H },
> > - { .port = PORT_I },
> > + { .name = "DDI A", .port = PORT_A, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x0, .phy_idx = 0x0, .vbt_idx = 0x0, },
> > + { .name = "DDI B", .port = PORT_B, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x1, .phy_idx = 0x1, .vbt_idx = 0x1, },
> > + { .name = "DDI TC1", .port = PORT_D, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x3, .phy_idx = 0x0, .vbt_idx = 0x2, },
> > + { .name = "DDI TC2", .port = PORT_E, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x4, .phy_idx = 0x1, .vbt_idx = 0x3, },
> > + { .name = "DDI TC3", .port = PORT_F, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x5, .phy_idx = 0x2, .vbt_idx = 0x4, },
> > + { .name = "DDI TC4", .port = PORT_G, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x6, .phy_idx = 0x3, .vbt_idx = 0x5, },
> > + { .name = "DDI TC5", .port = PORT_H, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x7, .phy_idx = 0x4, .vbt_idx = 0x6, },
> > + { .name = "DDI TC6", .port = PORT_I, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x8, .phy_idx = 0x5, .vbt_idx = 0x7, },
> > { .port = PORT_NONE }
>
> Makes you wonder if this sort of info should be linked from the
> intel_device_info instead of adding new arrays. Not sure.
Yeah, but I think it may be too much info there. Otherwise we should
migrate the powerwell definitions, hpd, etc. Just
like for the powerwells, I think the table is only relevant for the
init function. Every other user should just go through the
port_info from the port it's working with instead of looping through
all of ports in the intel_device_info. By keeping the array
private here we at least prevent that I think.
Lucas De Marchi
>
> BR,
> Jani.
>
>
> > }
> > };
> > @@ -16293,12 +16293,12 @@ static const struct intel_output icl_output = {
> > .dsi_init = icl_dsi_init,
> > .is_port_present = icl_is_port_present,
> > .ddi_ports = {
> > - { .port = PORT_A },
> > - { .port = PORT_B },
> > - { .port = PORT_C },
> > - { .port = PORT_D },
> > - { .port = PORT_E },
> > - { .port = PORT_F },
> > + { .name = "DDI A", .port = PORT_A, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x0, .phy_idx = 0x0, .vbt_idx = 0x0, },
> > + { .name = "DDI B", .port = PORT_B, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x1, .phy_idx = 0x1, .vbt_idx = 0x1, },
> > + { .name = "DDI TC1", .port = PORT_C, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x2, .phy_idx = 0x0, .vbt_idx = 0x2, },
> > + { .name = "DDI TC2", .port = PORT_D, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x3, .phy_idx = 0x1, .vbt_idx = 0x3, },
> > + { .name = "DDI TC3", .port = PORT_E, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x4, .phy_idx = 0x2, .vbt_idx = 0x4, },
> > + { .name = "DDI TC4", .port = PORT_F, .phy_type = PHY_TYPE_TC, .ddi_idx = 0x5, .phy_idx = 0x3, .vbt_idx = 0x5, },
> > { .port = PORT_NONE }
> > }
> > };
> > @@ -16306,10 +16306,10 @@ static const struct intel_output icl_output = {
> > static const struct intel_output ehl_output = {
> > .dsi_init = icl_dsi_init,
> > .ddi_ports = {
> > - { .port = PORT_A },
> > - { .port = PORT_B },
> > - { .port = PORT_C },
> > - { .port = PORT_D },
> > + { .name = "DDI A", .port = PORT_A, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x0, .phy_idx = 0x0, .vbt_idx = 0x0, },
> > + { .name = "DDI B", .port = PORT_B, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x1, .phy_idx = 0x1, .vbt_idx = 0x1, },
> > + { .name = "DDI C", .port = PORT_C, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x2, .phy_idx = 0x2, .vbt_idx = 0x2, },
> > + { .name = "DDI D", .port = PORT_D, .phy_type = PHY_TYPE_COMBO, .ddi_idx = 0x3, .phy_idx = 0x0, .vbt_idx = 0x3, },
> > { .port = PORT_NONE }
> > }
> > };
> > @@ -16317,9 +16317,9 @@ static const struct intel_output ehl_output = {
> > static const struct intel_output gen9lp_output = {
> > .dsi_init = vlv_dsi_init,
> > .ddi_ports = {
> > - { .port = PORT_A },
> > - { .port = PORT_B },
> > - { .port = PORT_C },
> > + { .name = "DDI A", .port = PORT_A, .ddi_idx = 0x0, .phy_idx = 0x0, .vbt_idx = 0x0 },
> > + { .name = "DDI B", .port = PORT_B, .ddi_idx = 0x1, .phy_idx = 0x1, .vbt_idx = 0x1 },
> > + { .name = "DDI C", .port = PORT_C, .ddi_idx = 0x2, .phy_idx = 0x2, .vbt_idx = 0x2 },
> > { .port = PORT_NONE }
> > },
> > };
> > @@ -16327,12 +16327,12 @@ static const struct intel_output gen9lp_output = {
> > static const struct intel_output ddi_output = {
> > .is_port_present = ddi_is_port_present,
> > .ddi_ports = {
> > - { .port = PORT_A },
> > - { .port = PORT_B },
> > - { .port = PORT_C },
> > - { .port = PORT_D },
> > - { .port = PORT_E },
> > - { .port = PORT_F },
> > + { .name = "DDI A", .port = PORT_A, .ddi_idx = 0x0, .phy_idx = 0x0, .vbt_idx = 0x0 },
> > + { .name = "DDI B", .port = PORT_B, .ddi_idx = 0x1, .phy_idx = 0x1, .vbt_idx = 0x1 },
> > + { .name = "DDI C", .port = PORT_C, .ddi_idx = 0x2, .phy_idx = 0x2, .vbt_idx = 0x2 },
> > + { .name = "DDI D", .port = PORT_D, .ddi_idx = 0x3, .phy_idx = 0x3, .vbt_idx = 0x3 },
> > + { .name = "DDI E", .port = PORT_E, .ddi_idx = 0x4, .phy_idx = 0x4, .vbt_idx = 0x4 },
> > + { .name = "DDI F", .port = PORT_F, .ddi_idx = 0x5, .phy_idx = 0x5, .vbt_idx = 0x5 },
> > { .port = PORT_NONE }
> > }
> > };
> > diff --git a/drivers/gpu/drm/i915/display/intel_display.h b/drivers/gpu/drm/i915/display/intel_display.h
> > index 05d68bd393dc..db19e5eee248 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display.h
> > @@ -303,6 +303,13 @@ enum phy {
> > I915_MAX_PHYS
> > };
> >
> > +enum phy_type {
> > + PHY_TYPE_NONE = 0,
> > +
> > + PHY_TYPE_COMBO,
> > + PHY_TYPE_TC,
> > +};
> > +
> > #define phy_name(a) ((a) + 'A')
> >
> > enum phy_fia {
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h
> > index 4d2f4ee35812..23a885895803 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> > @@ -1377,7 +1377,12 @@ struct intel_dp_mst_encoder {
> > };
> >
> > struct intel_ddi_port_info {
> > + const char *name;
> > enum port port;
> > + s8 phy_type;
> > + u8 ddi_idx;
> > + u8 phy_idx;
> > + u8 vbt_idx;
> > };
> >
> > static inline enum dpio_channel
>
> --
> Jani Nikula, Intel Open Source Graphics Center
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx
--
Lucas De Marchi
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
