At Sat, 27 Jun 2015 12:50:44 -0300, Henrique de Moraes Holschuh wrote:
On Sat, 27 Jun 2015, Jan Kiszka wrote:
Please explain _in detail_ what you mean with "changing a power button to a reset button by acessing the SSMS ACPI method in a X121e".
Are we trigering a bug somewhere that crashes the x121e and causes it to reboot?
Well, there aren't much details to describe in this case: When this patch is applied and I press the power button, the device performs a hard reset. It doesn't reveal if the kernel crashes catastrophically, causing a triple fault or so, or if the firmware decides to reset the platform. Reverting the patch or preventing its effect via a quirk filter like I posted [1] makes the issue go away.
Ok, so we crashing the firmware, and on a box that is not likely to get further updates. Well, this is a pretty clear case: blacklist.
Since calling SSMS in the X121e is unsafe, it needs to be blacklisted in thinkpad-acpi. I will need dmidecode output for that (please XXX-out UUIDs and serial numbers in the dmidecode output before you send it to me, but do keep model numbers intact).
We can also have a blacklist in ALSA to avoid the whole symbol_request() for something that will have to fail, of course. I leave that for the ALSA maintainers to decide.
IMO, it's simpler to just leave the decision in thinkpad_acpi.
That said, you blacklisted in ALSA by full codec->subsystem_id. Which thinkpads have the 0x17aa21ec subsystem ID on the codec? Just the x121e?
It's not guaranteed, but Lenovo doesn't share the same PCI SSID for multiple models, AFAIK.
thanks,
Takashi