since the commit
a855724dc08b pinctrl: amd: Fix mistake in handling clearing pins at startup
after boot, pressing power button is detected just once and ignored afterwards.
product: IdeaPad 5 14ALC05
cpu: AMD Ryzen 5 5500U with Radeon Graphics
bios version: G5CN16WW(V1.04)
distro: Arch Linux
desktop environment: KDE Plasma 5.27.7
steps to reproduce:
boot the computer
log in
run sudo evtest
select event2
(on my computer the power button is always represented by /dev/input/event2,
i don't know if it's the same on others)
press the power button multiple times
(might have to close the log out dialog depending on the DE)
expected behavior:
all the power button presses are recorded
observed behavior:
only the first power button press is recorded
i also have a desktop computer with a ryzen 5 2600x processor, but that isn't affected
#regzbot introduced: a855724dc08b
On 8/26/2023 15:33, Luca Pigliacampo wrote:
since the commit
a855724dc08b pinctrl: amd: Fix mistake in handling clearing pins at startup
after boot, pressing power button is detected just once and ignored afterwards.
product: IdeaPad 5 14ALC05
cpu: AMD Ryzen 5 5500U with Radeon Graphics
bios version: G5CN16WW(V1.04)
distro: Arch Linux
desktop environment: KDE Plasma 5.27.7
steps to reproduce:
boot the computer
log in
run sudo evtest
select event2
(on my computer the power button is always represented by /dev/input/event2,
i don't know if it's the same on others)
press the power button multiple times
(might have to close the log out dialog depending on the DE)
expected behavior:
all the power button presses are recorded
observed behavior:
only the first power button press is recorded
i also have a desktop computer with a ryzen 5 2600x processor, but that isn't affected
#regzbot introduced: a855724dc08b
If it bisects down to that commit then it means that the debounce behavior wasn't working as intended for your system and is now adjusted to work properly. That commit fixes it so that the "Master" GPIO controller register isn't inadvertently changed.
In a number of Lenovo designs I've found that GPIO0 is connected to the EC and the EC uses this BOTH for lid and power button. I don't know if that's the case for your design, but if it is this might explain the difference.
Given it changes the debounce behavior, can you hold the power button a little longer to get the event you're looking for?
Also, can you please check if you can reproduce this with 6.5-rc8/6.5 (whatever happens tomorrow)?
If so; can you please open a kernel Bugzilla with the following details:
1. The contents of /sys/kernel/debug/gpio on a kernel that doesn't have that series included/backported (the entire series of fixes - not just that one commit). A good test point for this is kernel 6.4.5. 2. The contents of /sys/kernel/debug/gpio on kernel 6.5-rc8/6.5. 3. An acpidump for your system. 4. If you have a dual boot with Windows, I would like for your to capture some registers using RW Everything in Windows for comparison. I'll provide your more steps on that bug report when you file it if necessary.
You can CC me directly to the bug report.
linux-stable-mirror@lists.linaro.org