GSOC - LwIP and Jailhouse hypervisor as possible tasks

[Pavel Pisa]

Hello Joel,

On Sunday 20 of March 2016 02:06:48 Joel Sherrill wrote:
> Thanks. The one in my directory is definitely ancient and should
> be ignored. It was sent to me WAY TOO LONG after it was done
> to be merged.

I have moved that link after newer attempts list.

> Does the page mention that there is an RSB package for LWIP now?

Links to RSB configuration files added.
I am little curious about LPC LwIP. It seems to build
complete RTEMS with LwIP networking but where is LPC driver
located? Has it be tested on real hardware?
It would worth to unite the work somehow in some repository.

Only TMS570 driver from our lwip-omk has been tested
with RTEMS till now. But our LPC LwIP  driver includes
IGMP support, we have tested it with mDNS on our systemless
targets and I think that code is reasonably well assembled.

The core work on build tree and LPC driver has been done
by Roman Bartosinski and partially paid by PiKRON company.

RTEMS and TMS570 port has been implemented above that base
by Premysl Houdek (CTU) as RTEMS GSoC followup and thesis.

It would worth to integrate BeagleBone driver.
May it be, that even TMS570 driver can be used on AM3xxx
because Ti EMAC controllers are similar.

> LWIP needs to be a first class citizen and the old stack needs to
> be moved out of the rtems tree and into its own build system.

I think that until there is done stacks porting and separation
which allows to build RTEMS included telnet and other services
against both new networking ports (new BSD and LwIP), the old
stack should be left in the three to allow check in single
RTEMS build pass that code is not dead/broken.

But may it be that other better or easier approach
can be taken.

Best wishes,

             Pavel