loading 'VIRT' kernel in qemu

[Jason Harmening]

On Fri, Dec 30, 2016 at 12:52 PM, Oleksandr Tymoshenko <gonzo at bluezbox.com>
wrote:

>
> On Dec 30, 2016, at 2:55 AM, Michael Tuexen <tuexen at fh-muenster.de> wrote:
>
> On 30 Dec 2016, at 09:14, Oleksandr Tymoshenko <gonzo at bluezbox.com> wrote:
>
>
> On Dec 29, 2016, at 5:25 PM, Oleksandr Tymoshenko <gonzo at bluezbox.com>
> wrote:
>
>
> On Dec 29, 2016, at 5:12 AM, Jason Harmening <jason.harmening at gmail.com>
> wrote:
>
>
> .. skipped ..
>
>
> Ah, thanks!  That should make things a good deal easier.
> Using the nanobsd-generated images and the qemu invocation from
> qemu-armv7.cfg, I get a bit further but end up with a repeating prefetch
> abort:
> Prefetch Abort Exception PC at 0xFF176790  CPSR 0x80000113
>
> This is either in UEFI or very early in kernel bootstrap, since the
> copyright hasn't printed yet...haven't poked at it to figure out where the
> pc comes from yet.
>
>
> Looks like it's boot1.efi. After upgrading to clang 3.9.0 -fPIC is
> required to  build loader. In addition to this problem there seems to be
> some error when calculating value for relocation. Looks like it’s going to
> take a bit of time to track down, I can’t identify it immediately.
>
>
>
> Hi Jason,
>
> I think I fixed the PIC problem along with FDT-related one:
> https://people.freebsd.org/~gonzo/arm/patches/arm-uefi.diff
>
> Any plan to get this into head?
>
>
> Yes, I’ll submit this patch to Phabricator and it’ll be committed to
> HEAD when/if it passes the review.
>

To close the loop (sorry, I was traveling for the last week & couldn't test
until now): applying this patch & passing -m 256M fixes things and I can
boot reliably.  Thanks for all your help!