Hi, I did have some problems with spurious I2C timeouts when trying to boot mainline Linux 3.17 but after finding an OMAP5 clock divider patch in linux-next and applying, the problem appears to be gone (at least almost - it started once after 4700 seconds with the board in idle mode).
This is very good news since we now have a shared kernel between the (OMAP3 based) GTA04 and the (OMAP5 based) Pyra - which makes driver development and upstreaming much simpler for me because we already do that for the GTA04. It is just sufficient to provide different device tree files through U-Boot for both devices. Well, it could even be possible to develop some OpenPandora device tree file...
So you can get it through any of: http://git.goldelico.com/?p=gta04-kernel.git;a=shortlog;h=refs/heads/3.17-rc... https://github.com/goldelico/gta04-kernel/tree/3.17-rc3 http://boards.openpandora.org/page/git.html <--- it is no longer listed by GitWeb there although the git repo exists
A kernel binary can be created through a makesd script here:
http://download.goldelico.com/gta04/unstable/pyra/
Anyways the kernel works quite well on the OMA5432EVM, except: * I don't find frame buffers * HDMI is missing (although configured) * there is no HDMI i2c to inspect by i2cdetect -r 5 * DSS does not try to load a lcd panel driver * there are spurious EHCI timeouts after power-on (on second boot they are gone and the USB host ports and Ethernet are working) * the usb3-otg port now reports "palmas: cable attached/detached" but I still don't see it working
But there is also a good aspect: I have the impression that the OMAP5 is not getting as hot as it was with the 3.8 kernel.
So I think I can now finally go into hardware tests again (except for LCD panels) and tweak the kernel to get access to e.g. charger or fuel gauge circuits.
If anyone has interesting patches or indications what we are doing wrong (DSS + Display? USB/OTG?), please submit to this mailing list for inclusion.
BR, Nikolaus