I managed to successfully port the android 2.6.23 kernel changes over to a TI Davinci TMS320DM6446 EVM Board - its an arm926ejs core with a DSP processor as well in the die. The changes were patched over the davinci-git kernel (2.6.23 as well).
The board boots up fine over NFS - i get the red blob moving around, and then the final UI comes up fine with no error messages on the console. I connected a QWERTY keyboard as a USB HID, and it worked absolutely fine the first time i connected it to the board; on subsequent boots however, a single keystroke started getting mis-interpreted as multiple keystrokes - typing "H" would result in 3 Hs for example; navigating the UI was really hard as it would skip over many of the icons. On further boots, this became even more problematic as it would duplicate them 5 times. A "getevent" showed that in addition to the event0 device being registered, some other weird .nfs0020932092302 device in the /dev/ input/ folder was being registered as well. If i deleted that device (an "ls" in the folder didn't show it), i would get newer .nfsxxxxxxxxx keystrokes being logged. Any ideas why this is happening? I'm using the ramdisk provided in the m5-rc14 SDK version. Thank you kindly, Jerry --~--~---------~--~----~------------~-------~--~----~ unsubscribe: [email protected] website: http://groups.google.com/group/android-porting -~----------~----~----~----~------~----~------~--~---
