Previously I've written a blog post commenting the behaviour of nandwrite with oob data, which empty pages have the hardware ECC bits set to all zeros, and render those pages not reusable unless the whole block is erased. Then in the previous days I try to use nandwrite to write U-Boot to the flash (after erasing those blocks, of course), but it appears that the OBM refuses to boot it, which is strange as writing in U-Boot succeeded.... read more
As I am migrating my development environment from my shared desktop to my (new) notebook, I have the first time performing git rebase.
Why are there so few people developing?? (I actually don't think 1 person is "a few people".)
Shall I continue developing?... What the hell...
If someone has a software emulator to fully emulate the iPAQ 212, I would be very glad.
-- post ended --
For those who've tied existig builds should feel peformance sluggish. i don't remember whether I've mentioned it before, the major speed is determined by the graphcal acceleration (as Android is very graphical intense), but some speed is also affected by the touchscreen as some of you might already noticed, touching the touchscreen makes screen animation eve slower. Let me explain a bit;
The touchscreen chip (wm9713) is responsible for the touchscreen. To get touch coordiates, there are two modes, namely polling mode and continuous mode.... read more
I tried to write my Windows Mobile backup rom back. When it writes an empty block (i.e. full 0xFF) with oob and ecc off, the last 24 bytes of the oob (ecc) gets written 0x00, corrupting the filesystem...
I made a hack in nanddump, ignoring all blocks filled with 0xFF so it doesn't write the wrong thing.
In order to create a partition layout similar to real Android devices, I decided to try to use loop devices.
I've been searching for this on Google, but many point out that the mount command in init.rc cannot handle loop (I think like losetup or mount -o loop <file>). Some even suggest adding a custom command to handle losetup command.
Finally, I decided to look into the source code of init. In the file system/core/init/builtins.c I was surprised to see that in the function do_mount (line 319) there is some code for handling loop device.... read more