| Message ID | 20240104183118.779778-2-wse@tuxedocomputers.com |
|---|---|
| State | New |
| Headers | show |
| Series | i8042: Add forcenorestore quirk to leave controller | expand |
Hi Werner, On Thu, Jan 04, 2024 at 07:31:17PM +0100, Werner Sembach wrote: > On s3 resume the i8042 driver tries to restore the controller to a known > state by reinitializing things, however this can confuse the controller > with different effects. Mostly occasionally unresponsive keyboards after > resume. > > These issues do not rise on s0ix resume as here the controller is assumed > to preserved its state from before suspend. > > This patch adds a quirk for devices where the reinitialization on s3 resume > is not needed and might be harmful as described above. It does this by > using the s0ix resume code path at selected locations. > > This new quirk goes beyond what the preexisting reset=never quirk does, > which only skips some reinitialization steps. I think the original change mentioned not only issues on resume, but also after boot, which this one does not address, at least directly, so I am not sure if this patch is the proper replacement. I would also like to understand better what exact step is troublesome, as I would be surprised if any interaction with the keyboard controller while suspending causes the issue to manifest. Is it enough, by chance, to skip restoring MUX mode and reset? Also, shoudl this system use s2idle by chance? Thanks.
Am 09.01.24 um 22:48 schrieb Dmitry Torokhov: > Hi Werner, > > On Thu, Jan 04, 2024 at 07:31:17PM +0100, Werner Sembach wrote: >> On s3 resume the i8042 driver tries to restore the controller to a known >> state by reinitializing things, however this can confuse the controller >> with different effects. Mostly occasionally unresponsive keyboards after >> resume. >> >> These issues do not rise on s0ix resume as here the controller is assumed >> to preserved its state from before suspend. >> >> This patch adds a quirk for devices where the reinitialization on s3 resume >> is not needed and might be harmful as described above. It does this by >> using the s0ix resume code path at selected locations. >> >> This new quirk goes beyond what the preexisting reset=never quirk does, >> which only skips some reinitialization steps. > I think the original change mentioned not only issues on resume, but > also after boot, which this one does not address, at least directly, so > I am not sure if this patch is the proper replacement. The original change introduced issues after boot and fixed issues after resume. The new quirk fixes the issues after resume without introducing issues after boot. The issues after boot where only affecting the NHxxRZQ and the N1xxCU iirc. > > I would also like to understand better what exact step is troublesome, > as I would be surprised if any interaction with the keyboard > controller while suspending causes the issue to manifest. Is it enough, > by chance, to skip restoring MUX mode and reset? SERIO_QUIRK_NOMUX and SERIO_QUIRK_RESET_ALWAYS are required in the old fix for the resume issues to go away (I don't know if SERIO_QUIRK_NOLOOP and SERIO_QUIRK_NOPNP are required, I threw them in just in case because we were running them already on the device, so they where somewhat "field proven" to not break things). However while SERIO_QUIRK_NOMUX and SERIO_QUIRK_RESET_ALWAYS for themself did not individually introduce the boot problem, in combination they did. That was all I was able to find out back when I tested the old quirks. > > Also, shoudl this system use s2idle by chance? N1xxCU is s3 only (intel 10th gen clevo) > > Thanks. >
diff --git a/drivers/input/serio/i8042-acpipnpio.h b/drivers/input/serio/i8042-acpipnpio.h index b585b1dab870e..10ec4534e5e14 100644 --- a/drivers/input/serio/i8042-acpipnpio.h +++ b/drivers/input/serio/i8042-acpipnpio.h @@ -83,6 +83,7 @@ static inline void i8042_write_command(int val) #define SERIO_QUIRK_KBDRESET BIT(12) #define SERIO_QUIRK_DRITEK BIT(13) #define SERIO_QUIRK_NOPNP BIT(14) +#define SERIO_QUIRK_FORCENORESTORE BIT(15) /* Quirk table for different mainboards. Options similar or identical to i8042 * module parameters. @@ -1657,6 +1658,8 @@ static void __init i8042_check_quirks(void) if (quirks & SERIO_QUIRK_NOPNP) i8042_nopnp = true; #endif + if (quirks & SERIO_QUIRK_FORCENORESTORE) + i8042_forcenorestore = true; } #else static inline void i8042_check_quirks(void) {} @@ -1690,7 +1693,7 @@ static int __init i8042_platform_init(void) i8042_check_quirks(); - pr_debug("Active quirks (empty means none):%s%s%s%s%s%s%s%s%s%s%s%s%s\n", + pr_debug("Active quirks (empty means none):%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", i8042_nokbd ? " nokbd" : "", i8042_noaux ? " noaux" : "", i8042_nomux ? " nomux" : "", @@ -1710,10 +1713,11 @@ static int __init i8042_platform_init(void) "", #endif #ifdef CONFIG_PNP - i8042_nopnp ? " nopnp" : ""); + i8042_nopnp ? " nopnp" : "", #else - ""); + "", #endif + i8042_forcenorestore ? " forcenorestore" : ""); retval = i8042_pnp_init(); if (retval) diff --git a/drivers/input/serio/i8042.c b/drivers/input/serio/i8042.c index 9fbb8d31575ae..2233d93f90e81 100644 --- a/drivers/input/serio/i8042.c +++ b/drivers/input/serio/i8042.c @@ -115,6 +115,10 @@ module_param_named(nopnp, i8042_nopnp, bool, 0); MODULE_PARM_DESC(nopnp, "Do not use PNP to detect controller settings"); #endif +static bool i8042_forcenorestore; +module_param_named(forcenorestore, i8042_forcenorestore, bool, 0); +MODULE_PARM_DESC(forcenorestore, "Force no restore on s3 resume, copying s2idle behaviour"); + #define DEBUG #ifdef DEBUG static bool i8042_debug; @@ -1232,7 +1236,7 @@ static int i8042_pm_suspend(struct device *dev) { int i; - if (pm_suspend_via_firmware()) + if (!i8042_forcenorestore && pm_suspend_via_firmware()) i8042_controller_reset(true); /* Set up serio interrupts for system wakeup. */ @@ -1248,7 +1252,7 @@ static int i8042_pm_suspend(struct device *dev) static int i8042_pm_resume_noirq(struct device *dev) { - if (!pm_resume_via_firmware()) + if (i8042_forcenorestore || !pm_resume_via_firmware()) i8042_interrupt(0, NULL); return 0; @@ -1271,7 +1275,7 @@ static int i8042_pm_resume(struct device *dev) * not restore the controller state to whatever it had been at boot * time, so we do not need to do anything. */ - if (!pm_suspend_via_firmware()) + if (i8042_forcenorestore || !pm_suspend_via_firmware()) return 0; /*
On s3 resume the i8042 driver tries to restore the controller to a known state by reinitializing things, however this can confuse the controller with different effects. Mostly occasionally unresponsive keyboards after resume. These issues do not rise on s0ix resume as here the controller is assumed to preserved its state from before suspend. This patch adds a quirk for devices where the reinitialization on s3 resume is not needed and might be harmful as described above. It does this by using the s0ix resume code path at selected locations. This new quirk goes beyond what the preexisting reset=never quirk does, which only skips some reinitialization steps. Signed-off-by: Werner Sembach <wse@tuxedocomputers.com> Cc: stable@vger.kernel.org --- drivers/input/serio/i8042-acpipnpio.h | 10 +++++++--- drivers/input/serio/i8042.c | 10 +++++++--- 2 files changed, 14 insertions(+), 6 deletions(-)