MobileRead Forums - View Single Post

hawhill · 06-27-2012, 06:50 AM

Nah, we're not nearly there yet. And if the device is bricked down completely (I doubt so, but it's very well possible), there is no fix. Otherwise, the referenced thread has its value for referring to a few pitfalls for serial recovery.

For the OP: Recovery via serial interface is still the only known method. You need a 1.8V ttl level serial interface, and some basic knowledge about using a serial console. Plus, there is no guarantee it will work since it still needs a tiny bit of non-broken data on the flash drive (Uboot plus the kernel including initramfs).

For fellow developers interested in creating a USB recovery path: I'm planning to open a new thread for digging deeper into the USB recovery thing. I've played a bit with it lately, and the sources of ATK knc1 referenced in the other thread were helpful (although awful to read, very clearly written by drunk monkeys, reviewed by a sleeping panda). There is a "ROM kernel" (Freescale name), and ATK pushes a "RAM kernel" for accessing the different flash types. The latter is not fit for the Kindle, and the idea is to push uboot and/or a linux kernel instead. It's *not* easy to get there. However, luckily the engineers that built the Kindle PCB didn't mess it up - which is easy to do. The thing is the crossbar switch in the SoC, which needs to be correctly initialized to toggle the right pins. However, they didn't went far from the initial default, AFAICS. Currently I'm really missing a simple test point toggled by a GPIO, preferably, to have a simple test for uploading and running my own code in USB ROM kernel (i.e. SE Ringo) mode. Again: ATK will *not* work as it is for writing/reading data. The RAM kernel it is equipped with is designed for their own evaluation boards - different enough compared with the Kindle.

06-27-2012, 06:50 AM	#4
hawhill Wizard Posts: 1,379 Karma: 2155307 Join Date: Nov 2010 Location: Goettingen, Germany Device: Kindle Paperwhite, Kobo Mini	Nah, we're not nearly there yet. And if the device is bricked down completely (I doubt so, but it's very well possible), there is no fix. Otherwise, the referenced thread has its value for referring to a few pitfalls for serial recovery. For the OP: Recovery via serial interface is still the only known method. You need a 1.8V ttl level serial interface, and some basic knowledge about using a serial console. Plus, there is no guarantee it will work since it still needs a tiny bit of non-broken data on the flash drive (Uboot plus the kernel including initramfs). For fellow developers interested in creating a USB recovery path: I'm planning to open a new thread for digging deeper into the USB recovery thing. I've played a bit with it lately, and the sources of ATK knc1 referenced in the other thread were helpful (although awful to read, very clearly written by drunk monkeys, reviewed by a sleeping panda). There is a "ROM kernel" (Freescale name), and ATK pushes a "RAM kernel" for accessing the different flash types. The latter is not fit for the Kindle, and the idea is to push uboot and/or a linux kernel instead. It's not easy to get there. However, luckily the engineers that built the Kindle PCB didn't mess it up - which is easy to do. The thing is the crossbar switch in the SoC, which needs to be correctly initialized to toggle the right pins. However, they didn't went far from the initial default, AFAICS. Currently I'm really missing a simple test point toggled by a GPIO, preferably, to have a simple test for uploading and running my own code in USB ROM kernel (i.e. SE Ringo) mode. Again: ATK will not work as it is for writing/reading data. The RAM kernel it is equipped with is designed for their own evaluation boards - different enough compared with the Kindle.