This is pretty hackish at the moment, but seems to work mostly well. Known quirk: if you disable host mode while some devices are still attached, they won't be disconnected - perhaps need to call for disconnect in this case. Change config: enable usb host support Certain usb devices enabled too, like HID, UVC, usb masstorage, usb serial. ipv6 disabled to keep the kernel size under 3M Added syfs output of charger status. Also ability of power draw control. Changed to usb host mode on Nook: Ability to request externally driven vbus (by battery or external source). Default to externally driven vbus when external power is connected at vbus request time. Don't try to do anything with vbus when a change comes for the same mode we are already in. Defconfig: More USB input device drivers
The adb driver enqueues the same reuest during read hoping that by the time it is doing next queue, the previous one completed. But if there was a signal (like what happens on suspend) this is not true and we need to dequeue the request first before bailing out.
This reverts commit 70c2604.
This is Con Kolivas' BFS patch applied to the encore kernel, v 401. Source is at http://ck.kolivas.org/patches/bfs/2.6.32/ Note that I took one liberty in the patch to sched.h, namely kernel/fork.c was choking on errors about missing "rt" so I allowed it in the structure.
Still experiencing a kp, but I don't think it's related. Checking this in for others (dalin) to try.
This should work, athough it's only been lightly tested. * Use with bts firmware from TI in /system/etc/firmware * Uses latest gb uim-sysfs * Also includes changes to defconfig, including ext4 support * brings the dead to life btwilink.c is the driver, making a guest appearance from .35.