On 9.08.2024 8:09 PM, Maximilian Luz wrote:
Hi,
Thanks for taking this up! A couple of comments below:
[...]
Doc needs updating, this is just the one copied from
ssam_controller_caps_load_acpi().
Oops.
[...]
quoted
};
-/* -- ACPI based device setup. ---------------------------------------------- */
+/* -- Serial device setup. ---------------------------------------------- */
Nitpick, but could we maybe keep that at 80 columns please? :)
Sure
quoted
static acpi_status ssam_serdev_setup_via_acpi_crs(struct acpi_resource *rsc,
void *ctx)
@@ -352,13 +355,28 @@ static acpi_status ssam_serdev_setup_via_acpi_crs(struct acpi_resource *rsc,
return AE_CTRL_TERMINATE;
}
-static acpi_status ssam_serdev_setup_via_acpi(acpi_handle handle,
- struct serdev_device *serdev)
+static int ssam_serdev_setup_via_acpi(struct serdev_device *serdev, acpi_handle handle)
{
- return acpi_walk_resources(handle, METHOD_NAME__CRS,
- ssam_serdev_setup_via_acpi_crs, serdev);
+ acpi_status status;
+
+ status = acpi_walk_resources(handle, METHOD_NAME__CRS,
+ ssam_serdev_setup_via_acpi_crs, serdev);
+
+ return status ? -ENXIO : 0;
}
+static int ssam_serdev_setup(struct acpi_device *ssh, struct serdev_device *serdev)
+{
+ if (ssh)
+ return ssam_serdev_setup_via_acpi(serdev, ssh->handle);
+
+ /* TODO: these values may differ per board/implementation */
+ serdev_device_set_baudrate(serdev, 4 * HZ_PER_MHZ);
Isn't this defined in the DT spec that you're adding as "current-speed"
in patch 2? Why not load it from there?
Yeah and it's not under `required:`.. i added it for future proofing
quoted
+ serdev_device_set_flow_control(serdev, true);
+ serdev_device_set_parity(serdev, SERDEV_PARITY_NONE);
+
+ return 0;
+}
/* -- Power management. ----------------------------------------------------- */
@@ -624,13 +642,15 @@ static int ssam_serial_hub_probe(struct serdev_device *serdev)
acpi_status astatus;
This can be removed, see below.
quoted
int status;
- status = gpiod_count(dev, NULL);
- if (status < 0)
- return dev_err_probe(dev, status, "no GPIO found\n");
+ if (ssh) {
+ status = gpiod_count(dev, NULL);
+ if (status < 0)
+ return dev_err_probe(dev, status, "no GPIO found\n");
- status = devm_acpi_dev_add_driver_gpios(dev, ssam_acpi_gpios);
- if (status)
- return status;
+ status = devm_acpi_dev_add_driver_gpios(dev, ssam_acpi_gpios);
+ if (status)
+ return status;
+ }
/* Allocate controller. */
ctrl = kzalloc(sizeof(*ctrl), GFP_KERNEL);@@ -655,7 +675,7 @@ static int ssam_serial_hub_probe(struct serdev_device *serdev)
goto err_devopen;
}
- astatus = ssam_serdev_setup_via_acpi(ssh->handle, serdev);
+ astatus = ssam_serdev_setup(ssh, serdev);> if (ACPI_FAILURE(astatus)) {
ssam_serdev_setup() returns an int, so this should now just use
"status".
Right
quoted
status = dev_err_probe(dev, -ENXIO, "failed to setup serdev\n");
goto err_devinit;
@@ -717,10 +737,31 @@ static int ssam_serial_hub_probe(struct serdev_device *serdev)
* For now let's thus default power/wakeup to false.
*/
device_set_wakeup_capable(dev, true);
+
+ /*
+ * When using DT, we have to register the platform hub driver manually,
+ * as it can't be matched based on top-level board compatible (like it
+ * does the ACPI case).
+ */
+ if (!ssh) {
+ struct platform_device *ph_pdev =
+ platform_device_register_simple("surface_aggregator_platform_hub",
+ 0, NULL, 0);
+ if (IS_ERR(ph_pdev))
+ return dev_err_probe(dev, PTR_ERR(ph_pdev),
+ "Failed to register the platform hub driver\n");
+ }
+
+ status = ssam_register_clients(&serdev->dev, ctrl);
+ if (status)
+ goto err_clients;
Is the ssam_register_clients() call required or is it a remnant from a
previous version? We're now not adding any children to the controller
itself but model ACPI and do all of that with the platform hub. So
unless I'm missing something, I think this should not be necessary.
Yeah, the platform hub driver calls it anyway
[...]
quoted
+static const struct software_node *ssam_node_group_sl7[] = {
+ &ssam_node_root,
+ &ssam_node_bat_ac,
+ &ssam_node_bat_main,
+ &ssam_node_tmp_perf_profile_with_fan,
+ &ssam_node_fan_speed,
+ &ssam_node_hid_sam_keyboard,
Did you check if there are any other HID devices connected? In the past,
keyboard and touchpad have been split into two separate devices, so is
it a combo keyboard + touchpad device this time? Some models also had
HID-based sensor and other devices.
No, touchpad is wired directly to the SoC via QSPI, same for the touch
panel
Would just be good to know if this can be assumed to be complete or if
we're maybe missing something here.
quoted
+ /* TODO: evaluate thermal sensors devices when we get a driver for that */
FYI I've posted the driver at [1]. It needs a small Kbuild dependency
fix but apart from that I think it should be final, if you want to give
that a try.
[1]: https://lore.kernel.org/lkml/20240804230832.247852-1-luzmaximilian@gmail.com/T/ (local)
I'll give it a shot, thanks
The rest looks fine. I'll try to find some time to update my SPX branch
this weekend and give it a spin.
About time that thing lands upstream ;)
Konrad