help offered
asic2fpga
I bought a samsung jet on November 2009. For the moment I was very
happy and excited since it was my first Smartphone. A week later a
friend of mine bought a galaxy with Android...
When I saw the OS performances I was quite annoyed with my Swamsung's
os.
Yesterday I saw this project and said it's for me...as i'm a quite a
geek during my free time:)
I'm a FPGA/RTL designer (PhD graduate in microelectronics) with a
background on computer sciences (engineering school). How can I help?
Is there any tutorial to start with? What is the project status update?
How far are we on the port?
Sincerely your,
Mike
Rectifier
Where we are (as far as I can see):
- The (original) bootloader has been modified slightly to allow
booting Qi from the SD card. However, this is still not finished, all
it does now is point to the address Qi should be at, when the right
keys are pressed.
- We'd like to have a display output in this phase, to debug the
bootloader, but that still requires a lot of disassembly
- The kernel needs to be compiled with all relevant "drivers". The
I5700 uses the wrong kernel (.29), so either the source would have to
be recompiled with a newer kernel (.31/.32) or Samsung's release of
the android 2.1 source of the i5700 will save us the trouble.
If you would like to help - set up a linux environment with all
available sources, as is explained here :
http://code.google.com/p/jetdroid/wiki/JetDroidBuildEnvironment, and
have a stroll around the github's, to get a more accurate status
update.
Welcome to the club! Let's hope we can save other's from making the
same mistake we did when buying this "smarter than smart" device ;-)
asic2fpga
Thanks for your message. I'm very excited to help to promote Jet.
Coupled to Android, it's might be a real iphone killer::)
Thanks again for the project status. I'll setup the dev' environment
this weekend.
By the way, if you guy's have any questions about FPGA (I saw some
topics about libero IDE) don't hesitate!
See ya!
asic2fpga
I saw in comments feeds that there was a video demo on youtube. Could
anyone send me the link?
Thanks in advance!
Mike
Peter Jaško
Aleksey Cherepanov
I have some questions about fpga.
I saw that on some dev boards it is possible to access jtag interface
through usb.
Is it possible on jet?
It would be great if you explain how jet connected internally. I mean
it is a black hole for me how two jet's processors (application and
baseband) co-operate.
Regards,
Aleksey Cherepanov
2010/3/5, asic2fpga <asic...@gmail.com>:
N K
The current kernel in the GitHub repository is based on the I5700 (Spica) and M900 (Moment/InstinctQ) kernel that is available from the Samsung opensource website. It is a kernel based on linux version 2.6.27. This version corresponds to Android 1.5 and 1.6. Unfortunatly to my knowledge there is no driver for the AmoLED display available for this kernel.
In public Samsung kernel repository I found a driver for the AmoLED display. It is intended for another processor but this driver should be easiliy portable to the S3C6410 architecture. Unfortunatly this driver is inside a 2.6.29 linux kernel that does not contain any phone specific stuff like in the I5700/M900 kernels. The kernel version 2.6.29 corresponds to Android 2.0 and 2.1.
Thus there are several options available to get to a kernel for the Jet:
- Backport the 2.6.29 AmoLED driver to the 2.6.27 kernel (this is what I started in the testing branch of the GitHub kernel repository. without much success, yet)
- Port the phone specific stuff (e.g. PMU, keypad, sonsor, compass drivers) to the plain 2.6.29 kernel from the Samsung repo.
- Wait for the source of the new I5700 Android 2.1 firmware to be released.
What do you think about this problem? Any comments?
Cheers,
Dopi
Peter Jaško
--
Sent from my mobile device
Rectifier
serveral JTAG connectors.
And they usually feature some form of chip connected to the USB port,
and the JTAG specific pins of the FPGA device. (at least - evaluation
boards do).
In case of the Jet, this is still unclear. The processor itself has
JTAG pins, and they are accessible, but I'm not sure whether there's a
primary (hardcoded) bootloader in either device that can decode some
form of USB communication to virtual JTAG signals.
This could be in the datasheet of the S3C though, will have to get
back to you on that.
In case of the I5700, the FSA9480 Usb switch does not seem to have any
direct link to the JTAG ports, so all will have to reside in the main
processor.
On 5 mrt, 14:29, Aleksey Cherepanov <aleksey.4erepa...@gmail.com>
wrote:
> Hi Mike!
>
> I have some questions about fpga.
> I saw that on some dev boards it is possible to access jtag interface
> through usb.
> Is it possible on jet?
>
> It would be great if you explain how jet connected internally. I mean
> it is a black hole for me how two jet's processors (application and
> baseband) co-operate.
>
> Regards,
> Aleksey Cherepanov
>
> 2010/3/5, asic2fpga <asic2f...@gmail.com>:
Rectifier
time, and quite possibly more than the wait will be for the release.
But all options are viable and could be worthwhile to gain some
experience with the device and android.
On 5 mrt, 14:38, N K <dopi...@googlemail.com> wrote:
> Thanks to rectifier for posting this update. I just have some minor remarks
> regarding the kernel versions
>
> The current kernel in the GitHub repository is based on the I5700 (Spica)
> and M900 (Moment/InstinctQ) kernel that is available from the Samsung
> opensource website. It is a kernel based on linux version 2.6.27. This
> version corresponds to Android 1.5 and 1.6. Unfortunatly to my knowledge
> there is no driver for the AmoLED display available for this kernel.
>
> In public Samsung kernel repository I found a driver for the AmoLED display.
> It is intended for another processor but this driver should be easiliy
> portable to the S3C6410 architecture. Unfortunatly this driver is inside a
> 2.6.29 linux kernel that does not contain any phone specific stuff like in
> the I5700/M900 kernels. The kernel version 2.6.29 corresponds to Android 2.0
> and 2.1.
>
> Thus there are several options available to get to a kernel for the Jet:
>
> 1. Backport the 2.6.29 AmoLED driver to the 2.6.27 kernel (this is what I
> started in the testing branch of the GitHub kernel repository. without much
> success, yet)
> 2. Port the phone specific stuff (e.g. PMU, keypad, sonsor, compass
> drivers) to the plain 2.6.29 kernel from the Samsung repo.
> 3. Wait for the source of the new I5700 Android 2.1 firmware to be
> released.
>
> I have not yet decided which option is the best to persue. I somehow pushed
> this decision out until the bootloader is actually demanding a working
> kernel.
>
> What do you think about this problem? Any comments?
>
> Cheers,
>
> Dopi
>
Aleksey Cherepanov
I think there is yet another option.
It is to wait for the sources of shw-m100s's firmware.
But i am not sure that it is realy useful.
Regards,
Aleksey Cherepanov
2010/3/6, Rectifier <s.eng...@gmail.com>:
N K
Until any of these sources is availble (either the M100S or the new I5700 Android 2.1 source) we could start with the 2.6.29 kernel from the Samsung repository. I will try to create a new branch with this kernel in the repsoitory. However, this raises the question if it makes sense to persue both ways (backporting the AmoLED driver to the current I5700 2.6.27 kernel and starting with the 2.6.29 kernel.
I am starting to like the idea of getting more familiar with the 2.6.29 kernel. Android 2.1 might be unavoidable for the Jet (somewhere I read that only Android >=2.0 supports 480x800 pixel resolutions). What do you think?
Cheers,
Dopi
edwin Fong
N K
unfortunately the M100S source is not yet available. Thus we are
discussing how to proceed without it. Wait or work out our own
solution . . .
Cheers,
Dopi
2010/3/7, edwin Fong <edw...@gmail.com>:
--
Gesendet von meinem Mobilgerät
Aleksey Cherepanov
In case when we will not have all requirements for making android we
will may try gnu/linux which we will try some drivers on. But there is
one bad thing as i know. Android works through framebuffer device. And
gnu/linux often uses xserver for graphics. Only few programs are
ported to fb. If i am not wrong then firefox is such program. So if
other options are not applicable then we may try this one.
Regards,
Aleksey Cherepanov
2010/3/7, N K <dop...@googlemail.com>:
Rectifier
In that case - a 'simple' kernel compilation could allow for hardware
compatibility, even though the applications are still slightly off.
I think for development purposes no one would complain if their
firefox only stretched half the display; as long as it works right?
In my opinion, and my experience, people will want the "latest and
greatest" anyway, so it's probably best to prepare for it as well...
On 7 mrt, 11:00, N K <dopi...@googlemail.com> wrote:
> Hi Edwin,
> unfortunately the M100S source is not yet available. Thus we are
> discussing how to proceed without it. Wait or work out our own
> solution . . .
>
> Cheers,
> Dopi
>
> 2010/3/7, edwin Fong <edwu...@gmail.com>:
>
>
>
> > Ya, only Android 2.0 or newer supports 800x480 res. but is the M100S source
> > code available now?
>
> >>> 2010/3/6, Rectifier <s.engel...@gmail.com>:
Aleksey Cherepanov
Why cannot we build android 2.0+ system with 2.6.27 kernel?
Such solution needs backporting of amoled driver.
Regards,
Aleksey Cherepanov
2010/3/7, Rectifier <s.eng...@gmail.com>:
Rectifier
The 6410 can boot out of :
- its secure internal boot memory
- Its NAND memory (except for the PoP type D device, which could be
the one used in the I5700/JET), but aparently only after initial
initialisation by the iROM memory.
- Out of the baseband processor (so-called modem boot)
What it can't do is boot out of the USB connection, nor does it appear
to feature something akin to a JTAG-over-USB option.
The thing is - its secure boot memory might contain some debug-boot
loader or be completely empty. I'm not sure whether it is actually
used at the moment. It would require a hardware connection to select
pins/balls to enable it though. Check out page 593 and more in the
application notes of the S3C. They also contain info on how to boot
out of SD cards (memory addresses, organisation etc)
So, unless we load a custom bootloader, no JTAG over USB will be
possible, which is a pity.
Aleksey Cherepanov
jtag over usb should be a feature of usb controller on the board. I
mean this is a hardware thing and is not related with a bootloader.
Regards,
Aleksey Cherepanov
2010/3/7, Rectifier <s.eng...@gmail.com>:
Rectifier
It is possible to include debug functionality to code as well, *IF*
the processor architecture allows for it.
It doesn't happen often enough though, and I don't believe the S3C to
be one of the lucky few.
Either way, it's not present, so we're left with trying to get the
bootloader to do what we want it to do. (i.e. boot stuff off of SD
cards etc)
On 7 mrt, 14:06, Aleksey Cherepanov <aleksey.4erepa...@gmail.com>
wrote:
> Hi Rectifier!
>
> jtag over usb should be a feature of usb controller on the board. I
> mean this is a hardware thing and is not related with a bootloader.
>
> Regards,
> Aleksey Cherepanov
>
> 2010/3/7, Rectifier <s.engel...@gmail.com>:
N K
Cheers,
Dopi