[NiLuJe]

Yup, that's a kernel issue. The kernel wakes up from suspend via panel/button interrupts, but the actual input events are lost (and/or mangled into the void: given how terrible the elan code is, I wouldn't exclude any possibilities ;p).

Firm presses *may* alleviate the issue, but that may just be a feeling. Or, if you wanna feel like an idiot, shaking or tilting the device first so that it wakes up from gyro interrupts first ;p.

(And, unfortunately, the full stream is lost, we don't get *any* events, so we have no recourse from userland).

EDIT: Err, I was talking about Elan because I was confusing this with the Elipsa lost taps issues, but that's a tad more complex, because it's broken outside of power-management states, too (and possibly more complex, because there's also another bunch of issues related to taps wrongly registered as 0-pressure). But, it also exhibits the same issue when it's the wakeup source, so the general point stands. Except on the Sage it happens with physical buttons, too, but only post-PM, no lost events when the kernel is awake ;p.

TL;DR: It's still an NTX board, I wouldn't have expected a kernel without weird quirks ;p.