Re: [Bug Report] intel/vbtn: Dell Inspiron 7352 has unreliable tablet-mode switch

[Hans de Goede <hdegoede@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 3638 bytes --]

Hi Arnold,

Thank you for reporting this.

Unfortunately removing the Dell Inspiron 7352 from
the dmi_switches_allow_list will not help.

The intel-vbtn driver now a days also creates an input-device
with a tablet-switch upon receiving the first tablet-switch
related event, to avoid needing to maintain an ever growing
list of devices on the allow-list.

And since the Dell Inspiron 7352 does have a somewhat working
tablet-mode-switch it does send such events. So removing it
from the allow-list will only result in the creation of
an input-device for the tablet-mode-switch being delayed till
the first event.

Now we could add a dmi_switches_deny_list for this purpose,
but first lets see if we can fix things.

On 10/29/23 20:52, Arnold Gozum wrote:
> Hi, sorry for the delayed reply. Your patch doesn't seem to work, I
> still have the issue where the switch is in the wrong state after
> suspend/resume.

Ok, so this does sound like the issue where the switch completely
stops reporting state-changes is fixed with the addition of
the extra "VBDL" call ?

I think that the wrong mode after suspend/resume is just a matter
of manually checking the mode after a suspend/resume.

Can you give the attached patch a try and see if that fixes things ?

Regards,

Hans




> And yes, it's been a while, and I believe the issues did exist during
> that time however it was easy to ignore/forget since I'm on X11 where
> libinput doesn't respond to tablet mode switches, so I neglected to
> report the issue for a while. Also, about the BIOS, I'm a little
> hesistant to update it since I don't have a battery. I have version A11
> and the newest is A15, but Dell's update notes only mention security
> fixes, so maybe it doesn't matter.
> 
> On 2023-10-17 22:05, AceLan Kao wrote:
>> Arnold Gozum <arngozum@gmail.com> 於 2023年10月18日 週三 上午8:53寫道：
>>>
>>> Hi,
>>>
>>> In Linux 5.11, Dell Inspiron 7352 was added to the
>>> dmi_switches_allow_list as it is a 2-in-1 which reports a chassis type
>>> 10 (actually it was me who submitted the patch).
>>>
>>> However, the tablet mode switch can be unreliable. Randomly, switch
>>> events stop being reported and SW_TABLET_MODE will by stuck at 1 or 0,
>>> which I have tested by running evtest while flipping the device to and
>>> from tablet mode. This is fixed after a reboot, or a suspend followed by
>>> unloading and reloading the intel-vbtn module. It can also sometimes be
>>> the case that upon resume, SW_TABLET_MODE does not reflect the actual
>>> state of the device, which is fixed by physically flipping the screen
>>> back and forth to update the state.
>>>
>>> Because of these issues, I think this model should be removed from the
>>> allow list, unless more investigation should be done.
>> Hi Arnold,
>>
>> It's been a long time since you submitted the patch. Did those issues
>> not occur during that time?
>> Have you tried updating the BIOS to see if it helps?
>>
>> From your description, I think calling VBDL might reset the status.
>> You might want to try it below.
>>
>> diff --git a/drivers/platform/x86/intel/vbtn.c
>> b/drivers/platform/x86/intel/vbtn.c
>> index 6fa1735ad7a49..681650f52ff22 100644
>> --- a/drivers/platform/x86/intel/vbtn.c
>> +++ b/drivers/platform/x86/intel/vbtn.c
>> @@ -198,6 +198,8 @@ static void notify_handler(acpi_handle handle, u32
>> event, void *context)
>>        autorelease = val && (!ke_rel || ke_rel->type == KE_IGNORE);
>>
>>        sparse_keymap_report_event(input_dev, event, val, autorelease);
>> +
>> +       acpi_evaluate_object(handle, "VBDL", NULL, NULL);
>> }
>>
>> /*
>>
>>>
>>> Thanks,
>>> Arnold
> 

[-- Attachment #2: 0001-platform-x86-intel-vbtn-Fix-missing-tablet-mode-swit.patch --]
[-- Type: text/x-patch, Size: 2764 bytes --]

From d3f88dc009af6b15d8130b917ed73f2aadbf66d6 Mon Sep 17 00:00:00 2001
From: Hans de Goede <hdegoede@redhat.com>
Date: Mon, 20 Nov 2023 14:59:06 +0100
Subject: [PATCH] platform/x86: intel-vbtn: Fix missing tablet-mode-switch
 events

2 issues have been reported on the Dell Inspiron 7352:

1. Sometimes the tablet-mode-switch stops reporting tablet-mode
   change events.

   Add a "VBDL" call to notify_handler() to work around this.

2. Sometimes the tablet-mode is incorrect after suspend/resume

   Add a detect_tablet_mode() to resume() to fix this.

Reported-by: Arnold Gozum <arngozum@gmail.com>
Closes: https://lore.kernel.org/platform-driver-x86/87271a74-c831-4eec-b7a4-1371d0e42471@gmail.com/
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
---
 drivers/platform/x86/intel/vbtn.c | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/platform/x86/intel/vbtn.c b/drivers/platform/x86/intel/vbtn.c
index 6fa1735ad7a4..210b0a81b7ec 100644
--- a/drivers/platform/x86/intel/vbtn.c
+++ b/drivers/platform/x86/intel/vbtn.c
@@ -73,10 +73,10 @@ struct intel_vbtn_priv {
 	bool wakeup_mode;
 };
 
-static void detect_tablet_mode(struct platform_device *device)
+static void detect_tablet_mode(struct device *dev)
 {
-	struct intel_vbtn_priv *priv = dev_get_drvdata(&device->dev);
-	acpi_handle handle = ACPI_HANDLE(&device->dev);
+	struct intel_vbtn_priv *priv = dev_get_drvdata(dev);
+	acpi_handle handle = ACPI_HANDLE(dev);
 	unsigned long long vgbs;
 	acpi_status status;
 	int m;
@@ -89,6 +89,8 @@ static void detect_tablet_mode(struct platform_device *device)
 	input_report_switch(priv->switches_dev, SW_TABLET_MODE, m);
 	m = (vgbs & VGBS_DOCK_MODE_FLAG) ? 1 : 0;
 	input_report_switch(priv->switches_dev, SW_DOCK, m);
+
+	input_sync(priv->switches_dev);
 }
 
 /*
@@ -134,7 +136,7 @@ static int intel_vbtn_input_setup(struct platform_device *device)
 	priv->switches_dev->id.bustype = BUS_HOST;
 
 	if (priv->has_switches) {
-		detect_tablet_mode(device);
+		detect_tablet_mode(&device->dev);
 
 		ret = input_register_device(priv->switches_dev);
 		if (ret)
@@ -198,6 +200,9 @@ static void notify_handler(acpi_handle handle, u32 event, void *context)
 	autorelease = val && (!ke_rel || ke_rel->type == KE_IGNORE);
 
 	sparse_keymap_report_event(input_dev, event, val, autorelease);
+
+	/* Some devices need this to report further events */
+	acpi_evaluate_object(handle, "VBDL", NULL, NULL);
 }
 
 /*
@@ -352,7 +357,13 @@ static void intel_vbtn_pm_complete(struct device *dev)
 
 static int intel_vbtn_pm_resume(struct device *dev)
 {
+	struct intel_vbtn_priv *priv = dev_get_drvdata(dev);
+
 	intel_vbtn_pm_complete(dev);
+
+	if (priv->has_switches)
+		detect_tablet_mode(dev);
+
 	return 0;
 }
 
-- 
2.41.0