Re: [Qemu-devel] Re: OMAP3 bug: disassembler disagreed with translator

[Peter Maydell]

On 18 February 2011 12:24,  <address@hidden> wrote:
> On Feb 18, 2011, at 14:08, ext Антон Кочков wrote:
>
>> Here also modified omap3_boot.c file. I'm dont know why it is
>> successfully load bootrom at 0x40014000, but cant do this for 0x14000
>> - it always fill by zeroes.
>> I'm only need copy memory from already loaded rom image from
>> 0x40014000 to 0x00014000 and nothing more.
>
> Our OMAP3 boot emulation is a very simplified model. For example,
> the boot ROM is never mapped to Q0 region, only Q1 region --
> the boot ROM execution is started at 0x40014000 rather than
> 0x00014000. There is no memory mapped at Q0 unless you attach
> something to the GPMC before reset.

However most boards do attach something to the GPMC;
the Beagle puts a NAND device on GPMC CS0, which means that
the GPMC will try to map NAND read/write functions on address
0 [*]. If you're trying to map a boot ROM there as well this will
clash and you need to do something clever to work however
the hardware does (presumably mapping the ROM there at startup
and unmapping the ROM later in response to some undocumented
event).

[*] Only true in the version of omap_gpmc.c in the meego
tree, but if you're doing omap3 stuff then I assume you're
using that, not upstream.

-- PMM