Thanks Chris. That was exactly the problem I was seeing; good to know that there is some sanity to why the MLO I picked from the latest cumulus image started working.

-Vikram

On Thu, Jan 26, 2012 at 9:29 AM, <gumstix-users-request@lists.sourceforge.net> wrote:
---------- Forwarded message ----------
From: Chris Whittenburg <whittenburg@gmail.com>
To: "General mailing list for gumstix users." <gumstix-users@lists.sourceforge.net>, vikram.godbole@gmail.com
Cc: 
Date: Thu, 26 Jan 2012 09:29:11 -0600
Subject: Re: [Gumstix-users] "current" Android image does not boot on WaterSTORM
Regarding the MLO issue, you might find some information in this thread:

http://gumstix.8.n6.nabble.com/x-load-version-1-5-0-from-Sept-16-td1173275.html#a1224581

Basically you need an updated MLO with the STORM devices.

-chris


On Thu, Jan 26, 2012 at 12:00 AM, Vikram Godbole
<vikram.godbole@gmail.com> wrote:
>
> Hi all,
>
> I am trying to test Android on my WaterSTORM COM using the image here:
>
> http://cumulus.gumstix.org/images/android/current/
>
> I followed the "Android Prebuilt" HOWTO from the wiki, and created the
> bootable microSD card using the mkandroidsd script. I'm running into two
> problems:
>
> 1. After I pop in the card and powerup the COM... nothing happens. Meaning
> there is no output on the console after the usual two rows of extended-ascii
> glyphs that appear following a reset. The glyphs should normally be followed
> by the X-Loader banner, followed by the U-Boot banner, followed by the the
> usual boot process output. Also, the red and blue LEDs on GPIO21 and GPIO22,
> which normally flash momentarily after a reset, now stay on. Funny thing is,
> once in about 10 attempts at resetting the COM, it does go beyond this hang
> point, and proceeds with the normal boot sequence.
>
> I initially suspected the SD card, so I copied the same "current" Android
> image to a known-good card- but I see the same problem. This suggests there
> is something wrong with the X-Loader. MLO and U-Boot versions when the COM
> boots fine are seen in the console as:
>
> Texas Instruments X-Loader 1.4.4ss (Mar 25 2011 - 15:35:41)
> U-Boot 2010.09 (Mar 25 2011 - 15:02:51)
>
> UPDATE: I just tried replacing the MLO from the android/current directory
> with that from the last-available angstrom factory image
> (images/angstrom/factory/2011-08-30-1058/), and that has fixed the hang
> issue.
>
>
> 2. On the rare occasions that the boot sequence does proceed correctly, the
> kernel appears to boot and I see the Android logo on the touchscreen LCD,
> but then I see the following messages on the console output:
>
> Warning: unable to open an initial console.
> PVR_K: (FAIL) SGXInit: Incompatible HW core rev (10205) and SW core rev
> (10201).
>
> After several minutes, the Android logo disappears, the LCD goes to (and
> stays) black, and I start getting "untracked pid exited" messages ad
> infinitum:
>
> request_suspend_state: wakeup (0->0) at 1483382721114 (2000-01-01
> 00:33:38.623962567 UTC)
> init: untracked pid 3336 exited
> binder: 1012: binder_alloc_buf, no vma
> binder: 3356:3356 transaction failed 29201, size 80-4
> binder: 1012: binder_alloc_buf, no vma
> binder: 3356:3356 transaction failed 29201, size 88-0
>
>
> My questions are as follows:
>
> 1. Why does a different MLO exist with every kernel image? It seems like all
> the MLO does is initialize the memory controller and peripherals, and load
> into memory a specific file with the name "u-boot.bin" from the first
> partition on the SD card. These functions are wholly independent of what
> kernel will eventually get loaded (or indeed, what the contents of
> u-boot.bin are). As long as you're not trying to do anything unusual (like
> load the X-Loader from MMC and U-Boot from UART), the MLO functionality can
> be set in stone..? That is, one known-good MLO ought to work with any
> u-boot.bin as long as you are booting entirely from the SD card?
>
> 2. The "Incompatible HW and SW" message suggests that the "current" android
> image was built before the STORM series with the DM37x cores (which
> evidently carry a later-rev SGX core). Is there a more current gumstix
> Android effort underway? Do drivers exist for the latest SGX core? If not,
> how can I coax the available pre-built image to boot on the WaterSTORM COM?
>
> 3. As a more general observation, Android-related posts seem to be
> relatively sparse on the list. I surmise that can only be because
> porting/customizing Android to Overo is either fairly straightforward or
> fiendishly difficult. :) I have not attempted building from source yet;
> would it work best to follow the gumstix wiki page on building Android from
> source? Are there any gotchas to be wary of?
>
> Thanks,
> Vikram
>