preferred working environment - advice from exp developers reg.

[Joel Sherrill]

On 07/21/2012 11:31 AM, Vidh wrote:
> Hi,
>
> I am a beginner at RTEMS.
> I prefer using RTEMS through simulator since I will be contributing 
> remotely.
>
Simulators are great!!!

Right now, I think everything can be done targetting mips/jmr3904. The 
simulator
is built into gdb and trivial to use with the script in rtems-testing.

Until we need a NIC other than "loopback", it can do everything we need. 
It has the
nice feature that the address range from 0x0 to 0x87ffffff is invalid so 
just about
any "normal" dereference of a NULL pointer gives you a nice fault and 
"bt" in
gdb will show you where you were.

I posted yesterday about the current status and what comes next:

http://www.rtems.org/pipermail/rtems-devel/2012-July/001415.html

I have tried to post every few days as progress is made. Once you can
duplicate where we are locally, we can pick you a task. We are on the
edge of this working enough where we will be bringing over telnetd,
httpd, etc. It is a bit single threaded now but there at least 4 fronts
to work on which are largely independent.

+ basic functionality going forward - next step is IPV4 only loopback
    socket read/write test
+ enable IPV6 and address interface configuration - currently core dumps.
     We work around this by disabling IPV6. :)
+ some files in freebsd-user/ are disabled because they do not compile yet.
+ On Qemu pc386, the rtl8139 NIC does not successfully initialize. As best
    I can tell, this is because the PCI code takes a path which results 
in NOT
    probing whether the card is I/O or memory space. We work around this
    by focusing on lo0 and fxp (e.g. Intel Ether Express Pro).

Once basic lo0 socket stuff works on IPV4, we will begin to bring up the
fxp0 driver. That is when we will learn if interrupts from NICs work in
our port. We can work around this with DEVICE_POLLED.

Then we bring up the various services.  These could be made to compile
on the new stack before we can actually test them.

So lots of potential activities.
> I am interested and trying to contribute to libbsd network stack 
> porting activity that is in progress.
> I am an ubuntu user but the wikis do not promise much for ubuntu.
There are active core RTEMS developers on many OSes including RPM based
Linux distros, Ubuntu, MacOS, and FreeBSD.

The RPM based distros are quick to get started with because there are RPMs
and a repo.

The others require building the tools from source but there are scripts and
instructions to do that.
>
> So I would like to hear some advice from senior developers or 
> developers who recently got started
> passing through this decision making hurdle on what OS would be more 
> suited.
>
> thanks! excuse me if this has been asked already and discussed 
> elaborately.
> if that is the case, please let me know where I can find those info,
>
We are open to help. This has been a large and challenging project. But 
when we
get the underlying RTEMS versions of BSD core kernel routines working, code
just falls into place. :)

--joel
> thanks & regards,
>
> -- 
> Vidhoon Viswanathan
> Design Engineer,
> Texas Instruments
> +91 7760711773
> <http://www.google.com/url?sa=t&source=web&cd=1&ved=0CCUQFjAA&url=http%3A%2F%2Fwww.ti.com%2Fww%2Fin%2F&rct=j&q=texas%20instruments%20%20india&ei=YL9ITrKwIJHJrQeQrMiLCA&usg=AFQjCNGKHhcMqfR9NNNDa2ZFTmEjN2IORQ&sig2=LzDgYQhcI0xWX_hbJ7t75Q&cad=rja> 
> <http://focus.ti.com/general/docs/gencontent.tsp?contentId=46946>
>
> <http://www.facebook.com/people/Vidhoon-Vishwanath/1212699428> 
> <http://twitter.com/#%21/vidhoon> <http://vidhoon.wordpress.com>
>