Re: [PATCH v6 2/3] drm/i915: Lookup and attach ACPI device node for connectors
From: Rajat Jain
Date: Thu Mar 05 2020 - 22:28:13 EST
On Thu, Mar 5, 2020 at 1:41 AM Jani Nikula <jani.nikula@xxxxxxxxxxxxxxx> wrote:
>
> On Wed, 04 Mar 2020, Rajat Jain <rajatja@xxxxxxxxxx> wrote:
> > Lookup and attach ACPI nodes for intel connectors. The lookup is done
> > in compliance with ACPI Spec 6.3
> > https://uefi.org/sites/default/files/resources/ACPI_6_3_final_Jan30.pdf
> > (Ref: Pages 1119 - 1123).
> >
> > This can be useful for any connector specific platform properties. (This
> > will be used for privacy screen in next patch).
> >
> > Signed-off-by: Rajat Jain <rajatja@xxxxxxxxxx>
> > ---
> > v6: Addressed minor comments from Jani at
> > https://lkml.org/lkml/2020/1/24/1143
> > - local variable renamed.
> > - used drm_dbg_kms()
> > - used acpi_device_handle()
> > - Used opaque type acpi_handle instead of void*
> > v5: same as v4
> > v4: Same as v3
> > v3: fold the code into existing acpi_device_id_update() function
> > v2: formed by splitting the original patch into ACPI lookup, and privacy
> > screen property. Also move it into i915 now that I found existing code
> > in i915 that can be re-used.
> >
> > drivers/gpu/drm/i915/display/intel_acpi.c | 24 +++++++++++++++++++
> > .../drm/i915/display/intel_display_types.h | 5 ++++
> > drivers/gpu/drm/i915/display/intel_dp.c | 3 +++
> > 3 files changed, 32 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_acpi.c b/drivers/gpu/drm/i915/display/intel_acpi.c
> > index 3e6831cca4ac1..870c1ad98df92 100644
> > --- a/drivers/gpu/drm/i915/display/intel_acpi.c
> > +++ b/drivers/gpu/drm/i915/display/intel_acpi.c
> > @@ -222,11 +222,22 @@ static u32 acpi_display_type(struct intel_connector *connector)
> > return display_type;
> > }
> >
> > +/*
> > + * Ref: ACPI Spec 6.3
> > + * https://uefi.org/sites/default/files/resources/ACPI_6_3_final_Jan30.pdf
> > + * Pages 1119 - 1123 describe, what I believe, a standard way of
> > + * identifying / addressing "display panels" in the ACPI. It provides
> > + * a way for the ACPI to define devices for the display panels attached
> > + * to the system. It thus provides a way for the BIOS to export any panel
> > + * specific properties to the system via ACPI (like device trees).
> > + */
> > void intel_acpi_device_id_update(struct drm_i915_private *dev_priv)
> > {
> > struct drm_device *dev = &dev_priv->drm;
> > struct intel_connector *connector;
> > struct drm_connector_list_iter conn_iter;
> > + struct acpi_device *conn_dev;
> > + u64 conn_addr;
> > u8 display_index[16] = {};
> >
> > /* Populate the ACPI IDs for all connectors for a given drm_device */
> > @@ -242,6 +253,19 @@ void intel_acpi_device_id_update(struct drm_i915_private *dev_priv)
> > device_id |= display_index[type]++ << ACPI_DISPLAY_INDEX_SHIFT;
> >
> > connector->acpi_device_id = device_id;
> > +
> > + /* Build the _ADR to look for */
> > + conn_addr = device_id | ACPI_DEVICE_ID_SCHEME |
> > + ACPI_BIOS_CAN_DETECT;
> > +
> > + drm_dbg_kms(dev, "Checking connector ACPI node at _ADR=%llX\n",
> > + conn_addr);
> > +
> > + /* Look up the connector device, under the PCI device */
> > + conn_dev = acpi_find_child_device(
> > + ACPI_COMPANION(&dev->pdev->dev),
> > + conn_addr, false);
> > + connector->acpi_handle = acpi_device_handle(conn_dev);
> > }
> > drm_connector_list_iter_end(&conn_iter);
> > }
> > diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h
> > index 5e00e611f077f..d70612cc1ba2a 100644
> > --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> > +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> > @@ -411,9 +411,14 @@ struct intel_connector {
> > */
> > struct intel_encoder *encoder;
> >
> > +#ifdef CONFIG_ACPI
> > /* ACPI device id for ACPI and driver cooperation */
> > u32 acpi_device_id;
> >
> > + /* ACPI handle corresponding to this connector display, if found */
> > + acpi_handle acpi_handle;
> > +#endif
> > +
> > /* Reads out the current hw, returning true if the connector is enabled
> > * and active (i.e. dpms ON state). */
> > bool (*get_hw_state)(struct intel_connector *);
> > diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c
> > index 0a417cd2af2bc..171821113d362 100644
> > --- a/drivers/gpu/drm/i915/display/intel_dp.c
> > +++ b/drivers/gpu/drm/i915/display/intel_dp.c
> > @@ -44,6 +44,7 @@
> > #include "i915_debugfs.h"
> > #include "i915_drv.h"
> > #include "i915_trace.h"
> > +#include "intel_acpi.h"
> > #include "intel_atomic.h"
> > #include "intel_audio.h"
> > #include "intel_connector.h"
> > @@ -6868,6 +6869,8 @@ intel_dp_add_properties(struct intel_dp *intel_dp, struct drm_connector *connect
> >
> > connector->state->scaling_mode = DRM_MODE_SCALE_ASPECT;
> >
> > + /* Lookup the ACPI node corresponding to the connector */
> > + intel_acpi_device_id_update(dev_priv);
>
> I find this part problematic.
>
> Normally, we call the function at probe via i915_driver_register() ->
> intel_opregion_register() -> intel_opregion_resume() ->
> intel_didl_outputs() -> intel_acpi_device_id_update(). It gets called
> *once* at probe, after we have all the outputs (and thus connectors)
> figured out.
>
> This in turn calls it for every DP connector, before we even have all
> connectors registered. But it also re-iterates the previously handled
> connectors again and again.
>
> The problem, of course, is that you'll need connector->acpi_handle to
> figure out whether the feature is present and whether the property is
> needed. Figuring out acpi_handle also requires
> connector->acpi_device_id.
>
> It's a bit of a catch-22.
>
> I think the options are:
>
> 1) See if we can postpone creating and attaching properties to connector
> ->late_register hook. (I didn't have the time to look into it yet, at
> all.)
Apparently not. The drm core doesn't like to add properties in
late_register() callback. I just tried it and get this warning:
[ 1.223163] WARNING: CPU: 2 PID: 1 at
drivers/gpu/drm/drm_mode_object.c:45 __drm_mode_object_add+0xab/0xaf
....
<snip>
.....
[ 1.223540] Call Trace:
[ 1.223540] drm_property_create+0xcc/0x155
[ 1.223540] drm_property_create_enum+0x26/0x79
[ 1.223540] intel_attach_privacy_screen_property+0x3a/0x5d
[ 1.223540] intel_dp_connector_register+0x6a/0xa8
[ 1.223540] drm_connector_register+0x61/0xaa
[ 1.223540] drm_connector_register_all+0x46/0x8b
[ 1.223540] drm_modeset_register_all+0x3e/0x67
[ 1.223540] drm_dev_register+0x124/0x187
[ 1.223540] i915_driver_probe+0xa18/0xda0
[ 1.223540] i915_pci_probe+0x145/0x168
>
> 2) Provide a way to populate connector->acpi_device_id and
> connector->acpi_handle on a per-connector basis. At least the device id
> remains constant for the lifetime of the drm_device
Are you confirming that the connector->acpi_device_id remains constant
for the lifetime of the drm_device, as calculated in
intel_acpi_device_id_update()? Even in the face of external displays
(monitors) being connected and disconnected during the lifetime of the
system? If so, then I think we can have a solution.
> (why do we keep
> updating it at every resume?!) but can we be sure ->acpi_handle does
> too? (I don't really know my way around ACPI.)
I don't understand why this was being updated on every resume in that
case (this existed even before my patchset). I believe we do not need
it. Yes, the ->acpi_handle will not change if the ->acpi_device_id
does not change. I believe the way forward should then be to populate
connector->acpi_device_id and connector->acpi_handle ONE TIME at the
time of connector init (and not update it on every resume). Does this
sound ok?
I can write the code up and send as my next patch iteration.
Thanks,
Rajat
>
> > }
> > }
>
> --
> Jani Nikula, Intel Open Source Graphics Center