[haiku-development] Re: Raspberry Pi 2

  • From: LECOQ Vincent <vincent.lecoq@xxxxxxxxx>
  • To: haiku-development@xxxxxxxxxxxxx
  • Date: Tue, 13 Oct 2015 15:27:59 +0200

Hello,

have somebody difficulties to build u-boot (for mkimage) on haiku ?

I experiment a double incompatible definition of umode_t like as following :

/Haiku system/home/Desktop/u-boot-2015.07> export
CROSS_COMPILE=/Haiku-dev/Workspace/haiku/generated.armpi/cross-tools-arm/bin/arm-un
known-haiku-
/Haiku system/home/Desktop/u-boot-2015.07> make distclean rpi_2_defconfig
all
[...]
HOSTCC tools/gen_eth_addr
In file included from ././include/linux/types.h:5:0,
from ././include/libfdt_env.h:13,
from <command-line>:0:
./arch/arm/include/asm/types.h:4:24: error: conflicting types for 'umode_t'
typedef unsigned short umode_t;
^
In file included from /boot/system/develop/headers/posix/alloca.h:9:0,
from /boot/system/develop/headers/posix/stdlib.h:9,
from /boot/system/develop/headers/bsd/stdlib.h:9,
from ././include/compiler.h:23,
from ././include/libfdt_env.h:12,
from <command-line>:0:
/boot/system/develop/headers/posix/sys/types.h:42:28: note: previous
declaration of 'umode_t' was here
typedef __haiku_std_uint32 umode_t;
^
make[2]: *** [tools/gen_eth_addr] Error 1
make[1]: *** [tools] Error 2
make: *** [__build_one_by_one] Error 2
/Haiku system/home/Desktop/u-boot-2015.07>



Thanks.


2015-10-12 9:50 GMT+02:00 LECOQ Vincent <vincent.lecoq@xxxxxxxxx>:

Hello,

Build tools and most of compilation works now, only the last step fail due
to mkimage is missing. I m searching how to compile it for haiku.
I think the SD/mmc is important to have too, i will ride on this way first
as soon as possible.

2015-10-10 23:09 GMT+02:00 Alexander von Gluck IV <kallisti5@xxxxxxxxxxx>:

October 10 2015 4:04 PM, "Adrien Destugues" <pulkomandy@xxxxxxxxx> wrote:
On Sat, Oct 10, 2015 at 10:40:49PM +0200, LECOQ Vincent wrote:

Thanks for the up to date build steps, I m currently follow them.
In the Arm port status, I see kernel work still in work, what are the
points need help first ? Have we a more detailled status ?

On the BeagleBoard we have the kernel booting far enough to complain it
can't mount the root filesystem (so it can't load drivers from there).

On the Pi, some work on the drivers will be required first to reach the
same state (setting up the serial port is done, not sure about the
framebuffer which is not really required but nice to have).

Then, the next big step is getting some kind of mass storage working. On
the beagleboard this would be either USB or SD/MMC. I think on the Pi
USB is the only choice. Our current USB stack was designed to plug to
the PCI bus manager, and it will need to be reworked so it can also work
with FDT, which replaces the PCI bus for our ARM port.

An alternative, if you don't want to mess with the USB drivers, is to
use the "NOR Flash" driver. This allows to put a filesystem in RAM and
can be used for experimenting. In that case, the block device will be
easier to setup, and some steps can be taken to build a root filesystem
with a few more drivers (and KDL commands) than we currently have. And
then try mounting and using it and see how far the boot process gets.


We really need an SD/mmc bus manager and driver long-term... I started
digging into it at one point but it ate up too much free time pretty
quickly.

I likely won't put any more work into it for the foreseeable future:

http://unixzen.com/patchwork/mmc-bus.diff

-- Alex




--
LECOQ Vincent
vincent.lecoq@xxxxxxxxx




--
LECOQ Vincent
vincent.lecoq@xxxxxxxxx
06 09 32 18 84
09 50 24 05 58
«La faculté de citer est un substitut commode à l'intelligence» — Sommerset
Maugham
«L'ennui avec les citations sur Internet c'est qu'il est difficile de
savoir si elles sont authentiques.» — Napoléon Bonaparte.

Other related posts: