[haiku-development] Re: Maintainer of the ARM port
- From: "Adrien Destugues" <pulkomandy@xxxxxxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Mon, 11 May 2015 19:26:59 +0000
Why not add a new platform on newos.org ? this os are more smaller and
portable as Haiku.
Haiku have used this kernel too, so it is good when a new platform run
on newos.org to backporting to Haiku.
We forked from newos more than a decade ago. It is very unlikely that any work
done there can be backported to Haiku.
But now there is an exception starting the boot script when it prepares the
stack width the arguments for the new thread. (Exception: Data Abort)
Is this also common for the ARM platform, or is it special for the beaglexm
I used?
I think we never tried to get further than "cannot find boot partition" on
beagle-xm. The MMU code and the memory map are still not complete and not very
well tested, which can probably explain your "data abort". You can enable
tracing of the MMU code (in the bootloader at
http://cgit.haiku-os.org/haiku/tree/src/system/boot/arch/arm/arch_mmu.cpp by
enabling TRACE_MMU) for the early boot process, and use KDL commands to debug
the kernel, if it managed to load some of them before the crash. Otherwise you
will need to add some tracing there as well
(
http://cgit.haiku-os.org/haiku/tree/src/system/kernel/arch/arm would be a good
starting point).
KDL has a bt command to show the backtrace, which should give you an idea where
the crash happened. From there you can look in the sources and see what could
be wrong. The ARM code has a lot of missing code with a "#warning" and possibly
no panic(), which means the code will silently do nothing.
--
Adrien.
Other related posts:
