1. The display driver/controller. No idea, the stock NTX kernels aren't verbose enough. Feels low-level power-management related, as being plugged in or having wi-fi enabled usually prevents almost all of the hangs.
2. Not as of yet, as this is extremely use-case specific, and I feel like taking the Nickel approach of doing this in an input handler sounds slightly more sensible than asking fbink to force it before every refresh. (Which is what we did in KOReader, but, again, that's a specific use-case where we know we never really send ioctl storms, so it's mostly fine for us).
3.
Quote:
Originally Posted by NiLuJe
you might want to try throwing a
fbink_wakeup_epdc (which is only available on master right now) before every refresh attempt (or on every user input, which is what Nickel does with a ~5s debounce).
|
Since you're using the Python module, you don't actually have access to that function until I rebuild and update all the things (... soon(TM)).
EDIT: Also, as mentioned above, this isn't exactly aimed at the Mk. 7 issues. In fact, the Clara HD kernel might not even *provide* the sysfs knob this is poking!
EDITē: Depending on what exactly you're doing, you might also want to batch some draw and/or refresh calls together, too (c.f., the no_refresh config & fbink_refresh), to avoid sending ioctl storms.
TL;DR: So, yeah, check if you can repro while plugged in and monitoring over SSH via USBNet and/or over Wi-Fi, because if you can, that's not what I've been talking about, and might simply be PEBCAK
.