Re: Delaying i8042 probe?

On Thu, 16 Sep 2021 11:22:03 +0200,
Takashi Iwai wrote:

On Sun, 12 Sep 2021 00:54:55 +0200,
Dmitry Torokhov wrote:

quoted

Hi Fabio,

On Sat, Sep 11, 2021 at 03:50:25PM -0300, Fabio Estevam wrote:

quoted

On Sat, Sep 11, 2021 at 3:43 PM Fabio Estevam [off-list ref] wrote:

quoted

On Sat, Sep 11, 2021 at 4:32 AM Takashi Iwai [off-list ref] wrote:

quoted

OK, I'll update and let the reporter testing it.

Sorry, platform_device_alloc() and platform_device_add() were missing
in the earlier patch.

New patch atached.

Dmitry, does this look correct?

Please consider this one instead.

This is unfortunately is a band-aid. I wonder what other driver pokes
the embedded controller on these devices for it to start responding to
8042 queries...

Does the laptop have the latest BIOS? I wonder what ACPI tables look
like.

ACPI dump is included in hwinfo output,
  https://bugzilla.suse.com/show_bug.cgi?id=1190256#c1

If other format is required, let me know.  I thought this could be a
typical pinctrl thing, alas it doesn't look so.  The pinctrl-amd is
also built-in, and it's initialized before the input stuff...

And about BIOS: I don't think we can expect every user updates BIOS.
This report is not alone; as I checked through net, there have been a
few other reports for other distros like Arch.  On Arch, they "fixed"
the problem by reverting the config from built-in to modules (the bug
surfaced after their kconfig changes).

That said, even if it's a band-aid, we need some fix.  Can the
deferred probe be applied in some restricted manner, e.g. only for the
known buggy devices (and/or module option) and cap the max repeats?

Dmitry, what is your take for fixing this bug?  

I find Fabio's deferred probe patch reasonable.  Maybe we may restrict
the application of the deferred probe only for known working devices
and an option, something like a patch below, just to be safer.
(There are devices exposing PnP for i8042 although there isn't really,
IIRC.)


thanks,

Takashi

-- 8< --
From: Takashi Iwai <redacted>
Subject: [PATCH] Input: i8042 - Perform deferred probe only with DMI list and
 option

Add an i8042.probe_defer option to enable the deferred probing
feature, otherwise it'll be disabled and done in a single shot like
before.  For the known devices that need the deferred probe, we
provide a DMI-based allow-list.  As of this patch, ASUS ZenBook
UX425UA is added there.

BugLink: https://bugzilla.suse.com/show_bug.cgi?id=1190256
Signed-off-by: Takashi Iwai <redacted>
---
 Documentation/admin-guide/kernel-parameters.txt |  2 ++
 drivers/input/serio/i8042-x86ia64io.h           | 14 ++++++++++++++
 drivers/input/serio/i8042.c                     |  6 +++++-
 3 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 91ba391f9b32..6e6622d8f4a0 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt

@@ -1690,6 +1690,8 @@
 			architectures force reset to be always executed
 	i8042.unlock	[HW] Unlock (ignore) the keylock
 	i8042.kbdreset	[HW] Reset device connected to KBD port
+	i8042.probe_defer
+			[HW] Allow deferred probing upon i8042 probe errors
 
 	i810=		[HW,DRM]
 

diff --git a/drivers/input/serio/i8042-x86ia64io.h b/drivers/input/serio/i8042-x86ia64io.h
index a5a003553646..41335a1d7001 100644
--- a/drivers/input/serio/i8042-x86ia64io.h
+++ b/drivers/input/serio/i8042-x86ia64io.h

@@ -981,6 +981,17 @@ static const struct dmi_system_id __initconst i8042_dmi_kbdreset_table[] = {
 	{ }
 };
 
+static const struct dmi_system_id i8042_dmi_probe_defer_table[] __initconst = {
+	{
+		/* ASUS ZenBook UX425UA */
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "ASUSTeK COMPUTER INC."),
+			DMI_MATCH(DMI_PRODUCT_NAME, "ZenBook UX425UA"),
+		},
+	},
+	{ }
+};
+
 #endif /* CONFIG_X86 */
 
 #ifdef CONFIG_PNP

@@ -1301,6 +1312,9 @@ static int __init i8042_platform_init(void)
 	if (dmi_check_system(i8042_dmi_kbdreset_table))
 		i8042_kbdreset = true;
 
+	if (dmi_check_system(i8042_dmi_probe_defer_table))
+		i8042_probe_defer = true;
+
 	/*
 	 * A20 was already enabled during early kernel init. But some buggy
 	 * BIOSes (in MSI Laptops) require A20 to be enabled using 8042 to

diff --git a/drivers/input/serio/i8042.c b/drivers/input/serio/i8042.c
index a72a8c538164..ea0c52ca95c4 100644
--- a/drivers/input/serio/i8042.c
+++ b/drivers/input/serio/i8042.c

@@ -45,6 +45,10 @@ static bool i8042_unlock;
 module_param_named(unlock, i8042_unlock, bool, 0);
 MODULE_PARM_DESC(unlock, "Ignore keyboard lock.");
 
+static bool i8042_probe_defer;
+module_param_named(probe_defer, i8042_probe_defer, bool, 0);
+MODULE_PARM_DESC(unlock, "Allow deferred probing.");
+
 enum i8042_controller_reset_mode {
 	I8042_RESET_NEVER,
 	I8042_RESET_ALWAYS,

@@ -1005,7 +1009,7 @@ static int i8042_controller_init(void)
 
 		if (i8042_command(&ctr[n++ % 2], I8042_CMD_CTL_RCTR)) {
 			pr_err("Can't read CTR while initializing i8042\n");
-			return -EPROBE_DEFER;
+			return i8042_probe_defer ? -EPROBE_DEFER : -EIO;
 		}
 
 	} while (n < 2 || ctr[0] != ctr[1]);

-- 
2.31.1