Re: [PATCH v2] usb: typec: ucsi: Stop a power level change notification storm

[David Vornholt]

Hi Mark,

On Mon, Oct 26, 2026 at 04:01:32PM -0400, Mark Pearson wrote:
> I need to read through the patch properly - but the X1 Carbon is
> quite popular and used by lots of people. Have we been able to rule
> out it being the E273m on another machine?

No, I haven't tried the E273m on another machine.

On this X1, though, I've only seen it with the E273m. The X1 has been
running v2 since the 10th, and up to the 16th its warning fired 14
times, every time with the E273m connected, on either port. The
monitor hasn't been connected since the 16th, and the warning hasn't
fired since, although the X1 has been charging over USB-C on most
days.

Right now the same port has a 20 V PD contract (the source offers up
to 20 V/5 A and PPS) with the notification enabled. The warning
hasn't fired, and GPE 0x6E stays at a few per second.

So it seems to take this EC and the E273m together. From the UCSI
side I can't tell what the EC is reacting to.

> I've created internal ticket LO-4728 so we can see if we can
> reproduce and get input from the FW team on why so many
> notifications would be generated - generated so many seems quite
> wrong.

Thank you. If it would help the FW team, I can capture the ucsi trace
events or an acpidump with the monitor connected.

> I've only skim read the proposed patch, but I'm also hesitant to say
> this is X1 C13 specific.

Agreed. This X1 is the only machine I've tested. If LO-4728 shows
that other platforms have the same problem, their board names could
go into the quirk too.

> As anote, if you do end doing a quirk then that field is bad as
> there are two different variants of the X1 Carbon 13 - ARL and LNL.
> Better to use DMI checks on the BOARD_NAME - there are two per
> platform, and I believe you're on Lunarlake so would be 21NS and
> 21NT.

Thanks. Yes, the board name here is 21NTS4EE00. If we go with the
quirk, v3 will match "21NS" and "21NT" on DMI_BOARD_NAME.

Thanks,
David