Synchro Digital WM8650 Win CE to Android

Okay, so I'm visiting relatives for post-Christmas and New year's in France. Let us just say that someone bought a "Synchro Digital WM8650 Win CE". Anyway, after they saw all of Windows CE's (affectionately known as Crippled Edition) wonders or lack thereof they were jealous of the Android device(s) I was simultaneously showing off, they asking if I could look into installing Android on their WM8650.

To my surprise this Netbook has quite a large community behind, presumably because of it's lo-cost. WonderMedia has also released the kernel sources here. There is at least one (android) ROM specifically targeting the ARM 8650 SoC chipset; namely Uberoid by HcH. However, I discovered the best candidate ROM was Modroid v11 (download : release page) by Wicknix (I found it after reading this article which is mainly discussing Linux ports such as debian/ubuntu/gentoo to WM8650) so I went with it. These devices don't boot in the same way as an Intel/AMD so it's not so straight forward to dual-boot them (but I believe it is possible if you boot the second OS off the SD card, but I don't know how to do that), and, it wasn't necessary for me as the goal was to completely replace WinCE. Obviously, there is no CD drive so booting from one of those is out of the question. After doing a little research it turns out that WonderMedia (and others) can boot from the SD card so far as there is a file called "wmt_scriptcmd" in the root of the target boot volume. Any usable ROM should include this file. Everything else is as straight forward as this:

1. Download modroid.
2. Format an SD card (FAT32 should be fine, i.e. it worked for me) and unzip modroid to the SD-card.
3. !!!WARNING!!! This step will erase WinCE and any data you have saved completely. Plug the SD-card into the WM8650 and watch modroid install itself. It will ask you to eject the SD-card and reboot once it is finished. 
4. The first boot should take a little while so maybe make yourself a cup of tea.
5. Modroid should boot into a fresh system. All done!!!

Those of you that are interested should note that the linux bootstrap (and maybe it is the same for CE) is U-boot and I believe it is possible to create custom boot scripts (I think these are the same kinds of scripts as the one shipped (wmt_scriptcmd) with the ROMs).

FINALLY, there is one last twist to my tale. My target device's keyboard was in French (skip to the end if you're not into reading all the text). Changing the language/keyboard in the Android settings simply changes the language and not the keyboard. According to some posts around on the net, this could be because these ROMs kernels have hardcoded keymaps/layouts (as opposed to linux modules) not sure how credible this is (however for those interested the solution suggested was to deflate the image add the kernel params "keypad="azerty"" and repack it). I did a little digging and it seems that libui.so is associated with the installed keylayouts (under /System/usr/keylayout) and with one of the libwmt*.so. 

Finally, I went in with the solution of making a backup copy of /system/usr/keylayout/qwerty.kl and replacing it with the keylayout "french.kl". This at least means the alphabetic characters are mapped correctly, however some of the symbols need to be refined ... I will post if and when I refine them...

**edit**

The ".kl" seem to be tab-delimited, the characters that map to keycodes are uppercase (e.g. COMMA). I'm having difficultly finding the mapping between these uppercase variables and the actual character/code which is output, I believe part of it lies in "linux/input.h" but that's not the whole story. If only I knew where the libui source was...

**edit**

Might be able to figure out which of the number keys are supposed to be mapped in french.kl. Using https://github.com/asedeno/unkcm

**edit**

ref: http://lxr.post-tech.com/source/frameworks/base/include/ui/KeycodeLabels.h
ref: http://lxr.linux.no

**edit**

spent the last couple of hours with the answer staring me in the face. rename qwerty.kcm.bin to qwerty-original.kcm.bin, and, place a copy of french.kcm.bin in place of qwerty.kcm.bin.

UK Internet Blackhole

I have family in Motherwell (for those that don't know it's in scotland). After seeing this article about the town's lack of facebook users (the town has always been a bit delipidated) so I decided to keep my techno-eyes peeled for technology mishaps or goodies on my recent visit to the town. I stayed at two hotels, obviously not at the same time, the first hotel had no WiFi it wasn't a surprise. Second hotel had WiFi it was unsecured, I wasn't surprised. It seemed to take a seriously long time to obtain a DHCP lease from the network so I did a little digging. It seems that the hotel's network was being managed remotely, probably has part of a block with other hotels within the same ADSL ISP. This surprised me a little; I would have thought, given the setup, that the ISP obviously providing a service to business would have the knowledge and sense to secure the AP. I encrypted anything mildly sensitive through an SSH tunnel, SSL would probably cover my email login, but I was feeling paranoid.

Next stop, the local library. Unsurprisingly their access points were unsecured, very well you say as it's meant to be public. Yes, and, no. A public network should be accessible, but shouldn't each separate link (wifi-user) be encrypted from other users to avoid sniffing? Well I that's my take, it seems to be a question few people asking  when it comes to security of intentional-open public networks accessible to everyone (e.g. libraries, major transport hubs, etc., etc.). Whilst SSL, will in most cases, help a user protect credit card details, email logins etc., all other internet bound traffic can be sniffed easily. With the advent of 4G I'm expected the ISPs and Mobile telecoms to converge somewhat, and, hopefully develop viable solutions to this problem. The mobile telecoms already did this for GSM (and presumably 3G and HPSA are also encrypted but I don't know enough about them to say for sure), albeit weak encrytion by today's standards but they did achieve it. Now the same technology is required for WiFi points.

In summary, in the past we sent information via emails and MSN messenger that we perhaps might not wish others to get their ends on, and, we (or rather those of us who are tech-savvy) did this with the knowledge it was sent in clear-text. In those days, it was an acceptable risk (as long as you weren't given your credit card details to a mate) you could see the cable to your internet modem and you were happy that no one was siphoning your data from there. Then you were happy that no-one at your ISP was that interested in your data (they should be earning enough from the $£ you send them once a month). Once it was there, you were reasonably happy that fred-next-door doesn't know you called him a c**t  to your sister on MSN. The difference now-a-days is that the node closest to you is no longer secure, so those in your immediate area can snoop were they couldn't before wifi came along. Particularly because, attackers can now be yards away from you, there is an enhanced social-security issue. Vulnerable individuals, such as children, could be put at greater risk. For example given the library scenario, a child may message a friend using an IM such as MSN, if they have a photo of themselves and their IM alias is their real name, a wifi sniffer could easy intercept this through an unsecured wifi point. Once a stranger can put a name to a child's face, the child is at a much greater risk. This example is one of many, I'm sure this is one of the more convoluted examples but I hope it is understandable to most.

As for Motherwell, I know that most of my, or at least most of the younger, relatives are using Facebook. In fact, a 15yo relative posts at least three public fb messages a day, complete with scottish accent. So, I can't pass too much judgement on the town with regards to social media, but as wifi points go I think hotels should restrict wifi to patrons only, and, libraries should restrict theirs to members only. To be honest, I wouldn't be surprised if Motherwell was the UK's most gravitating internet blackhole.

Or you can ignore me completely and join the campaign beneath:
http://www.dansdata.com/gz080.htm

last.fm rewinds

I really like last.fm; like many others I was a little sceptical about giving £3 a month so I could listen to music, but I was pleasantly surprised by the recommendations and a-radio-station-per-artist on-demand playlists and, of course, your own playlist based on tracks you have "loved". So my horror began today when I was greeted with an error message which pointed at a grovelling explanation on this page. 

Starting November 17th, 2010, Last.fm is discontinuing a few subscriber-only radio stations.

...

Licensing music is a complex and labour intensive process.  By discontinuing a few stations, we're able to focus our energy on improving our most popular features, developing new and innovative stations, and offering the best music discovery service to our global audience.

and

The following subscriber-only radio stations will no longer be available:

Loved Tracks Radio: streaming your list of loved tracks

Playlists: streaming a list of tracks you've chosen for a playlist

Personal Tag Radio: streaming a list of artists, albums or tracks that you have tagged

Yes you got it, last.fm no longer allows users to listen to the tracks they've chosen in effect forcing user's to listen to recommendation playlists. Granted the recommendation playlists are good, but everyone is going to want to listen to "just their music" once in a while. Here's what won't change:

Access to your list of loved tracks

Access to your playlists and the list of tracks within them

Access to your personal tags and the list of tracks within each

Your ability to listen to global tag stations

So basically you can access the names of the tracks you like but you can't play them and only them on-demand. Already, last.fm orders the sequence in which specific tracks in the playlist are up next, but now we can't listen to a specific music, instead music will HAVE to be played from a "recommended" playlist entirely derived by some formula at last.fm HQ. This "formula" (for want of a better word) is now very likely to have a very heavily weighted function on the music tracks legal licencing, I don't know about you but I couldn't give a damn about the licencing if I'm listening to "recommended" tracks. The last.fm "forced-sequencing" was tolerable given that this is a service last.fm have to make money out of, but with the onset of the "forced-playlists" (recommended playlists or artists radios) I don't think the £3 a month fee is justified.

The reason they cite is due to the complexity of music licencing. Yes, that's right, apparently it is more complicated to provide on-demand music from a variety of artists (from recommendations) than it is to play *one* specific track on demand. That doesn't smell right to me. Is this because "the recommendations" will be easier to licence on-demand? Implying that last.fm will group tracks for recommendations by the legal mumbojumbo in the track's licencing agreement. Surely this is the last thing on a user's mind when subscribing to a "recommendation". Go figure.


I've not yet decided whether I'll leave last.fm because of this, but I'm seriously considering it. 

a first blog

This is my first blog. I am 27 years old and I work as a Software Engineer for the UK Government. I've started blogging for a number of reasons (1) To discuss stuff with like-minded people and maybe become friends :-D (2) So I have somewhere to go when I feel like I can rant freely/anonymously (3) To keep track/share my technical exploits. 


I think that's pretty much it.