Support and General Use > Hardware
AGPTEK Rocker
Pingouin:
--- Quote from: wodz on June 12, 2019, 06:42:45 AM ---Finally I found some time to mess with framebuffer and I have bad news...
--- End quote ---
Doh! Well, thanks anyway for trying.
wodz:
After some digging and reverse engineering kernel running on Agptek I am able to revive lcd after blanking. Lcd driver is terribly broken as it doesn't reenable DMA when unblanking. There is undocumented ioctl however which enables DMA transfers to lcd as side effect. I'll try to post build for testing within few days.
Chronon:
Nice work, wodz!
Pingouin:
Thanks wodz! ;D
The question will then be: does all your hard work actually saves any battery?
And the other question: why didn't AGPtek try to fix their LCD driver rather than implement a software dirty hack to mitigate it! Although it doesn't seem it's actually AGPtek soft/hardware anyway, but someone must have been spending some time doing the same thing you did and would have had the opportunity to point out the driver problem to get it fixed. Unless they decided that the quick and dirty hack was more cost effective.
wodz:
Driver comes from Ingenic itself. See https://github.com/Ingenic-community/kernel-xburst-bsp for reference. Software for Agptek was developed by Fiio I believe. Based on my RE work on original software I believe they do not blank lcd and hence they don't trigger this bug (I may be wrong of course but I can't trace any ioctl or direct write to sysfs associated with locking screen in OF). I think OF writes all black screen, turnes off backlight and then stops writing to lcd.
The only way to know if blanking saves any energy is to banchmark it.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version