Towards an ARM system-building script
Ganbold Tsagaankhuu
ganbold at gmail.com
Tue Oct 9 11:02:47 UTC 2012
On Tue, Oct 9, 2012 at 6:48 PM, Ganbold Tsagaankhuu <ganbold at gmail.com> wrote:
> On Tue, Oct 9, 2012 at 5:10 PM, Ganbold Tsagaankhuu <ganbold at gmail.com> wrote:
>> On Tue, Oct 9, 2012 at 12:42 PM, Tim Kientzle <tim at kientzle.com> wrote:
>>>
>>> On Oct 8, 2012, at 9:07 PM, Ganbold Tsagaankhuu wrote:
>>>
>>>> On Mon, Sep 24, 2012 at 2:55 AM, Tim Kientzle <tim at kientzle.com> wrote:
>>>>> On Sep 4, 2012, at 9:52 PM, Tim Kientzle wrote:
>>>>>> On Sep 4, 2012, at 9:33 PM, Dave Cheney wrote:
>>>>>>
>>>>>>> Sorry to butt in on this discussion, but how feasible would it be to adapt this build script to the pandaboard. I understand there may be a config in svn similar to the beaglebone which may be applicable.
>>>
>>>>> I've made significant progress and would
>>>>> appreciate any feedback:
>>>>>
>>>>> github.com/kientzle/freebsd-beaglebone
>>>>>
>>>>> * PandaBoard: I have a completely untested
>>>>> sketch. Someone with a PandaBoard will need
>>>>> to work through the U-Boot configuration to make
>>>>> this work.
>>>>
>>>> I tried your script on PandaBoard ES.
>>>> Basically it works with some modifications/considerations:
>>>
>>> Wonderful! Thanks for trying it!
>>>
>>> Please send me any patches you used; I'll try
>>> to incorporate them.
>>>
>>>> 1. PANDABOARD kernel config should be modified to use mmcsd card as root.
>>>
>>> Would anyone object to adding this line to FreeBSD SVN?
>>>
>>> # Boot device is 2nd slice on MMC/SD card
>>> options ROOTDEVNAME=\"ufs:mmcsd0s2\"
>>>
>>>
>>>> 2. u-boot build complains about the board was configured to support THUMB mode
>>>
>>> Hmmm… Not sure I understand this.
>>>
>>>> 3. Maybe there is need some more patches to have correct u-boot.bin,
>>>> since it looks for uImage for instance
>>>
>>> There will certainly need to be some U-Boot patches.
>>>
>>> Perhaps Oleksandr can point us to the source he
>>> used for that u-boot.bin? If so, I can help fill in
>>> the options so it can correctly chain ubldr.
>>>
>>>> 4. I didn't dig into ubldr/u-boot much, instead modified uEnv.txt to
>>>> load kernel.bin from first partition which is FAT. I tried 8MB of FAT.
>>
>>
>> Progress, seems like ubldr is working after patching u-boot.
>> Here are the patches, uEnv.txt etc:
>>
>> http://people.freebsd.org/~ganbold/pandaboard/
>>
>> MLO is used from http://people.freebsd.org/~gonzo/pandaboard/
>
> Strange, with ubldr Ethernet is not found, although kernel config is
> the same as before.
dmesg are at the same place: http://people.freebsd.org/~ganbold/pandaboard/
Ganbold
>
> Ganbold
>
>>
>> Ganbold
>>
>>
>>
>>>> 5. I used MLO, u-boot.bin from http://people.freebsd.org/~gonzo/pandaboard/
>>>>
>>>> So how does ubldr work? Can it load kernel.bin from second partition
>>>> in case of PandaBoard?
>>>
>>> Yes, it can. At heart, ubldr is the same as the FreeBSD
>>> loader(8) that's used for i386/amd64 booting. In particular,
>>> it can load the kernel from UFS.
>>>
>>> The main difference between ubldr and loader(8)
>>> is in how they access hardware: loader(8) uses
>>> the PC BIOS and ubldr talks to U-Boot to do low-level
>>> device access.
>>>
>>> I think ubldr makes a lot of sense for BeagleBone and
>>> probably PandaBoard. I'm not sure it makes sense
>>> for Raspberry Pi, though that's a different discussion.
>>>
>>> But ubldr requires that U-Boot be built with a few
>>> non-standard options, so again we'll certainly have
>>> to patch U-Boot to enable those options.
>>>
>>>
>>> Tim
>>>
More information about the freebsd-arm
mailing list