Il 21/08/2014 11:18, Marc Marí ha scritto:
> El Thu, 21 Aug 2014 11:06:51 +0200
> Paolo Bonzini <pbonz...@redhat.com> escribió:
>> Il 20/08/2014 19:46, Marc Marí ha scritto:
>>> Add command query-mtree to get the memory tree of the guest.
>>>
>>> As we were looking for a flexible solution on accessing the guest
>>> memory from qtests, Stefan came with the idea to implement this new
>>> qmp command.
>>>
>>> This way, the result can be parsed, and the RAM direction
>>> extracted, so only a generic qtest malloc is necessary and not one
>>> per machine, as it is implemented at the moment (malloc-pc uses
>>> fw_cfg).
>>>
>>> The actual output is this: http://pastebin.com/nHAH9Jie
>>> Which corresponds to this info mtree: http://pastebin.com/B5vw8DDf
>>
>> I don't like this idea very much. libqos should be using the real
>> memory map information from the machine. In the case of x86, that
>> means fw_cfg; in the case of ARM, that would mean using the device
>> tree. Getting the information from an out-of-band channel (such as
>> QMP) is basically cheating. :)
>
> As we were looking at how to access the device tree, we found that the
> device tree is saved in memory with the bootloader or the kernel. So
> tests should be using a kernel every time a ARM machine is booted
> (and /dev/null, at least in virt machine, does not work). Do you have
> any better idea on how to do it?
It works for me:
$ qemu-system-arm -M virt -S -kernel /dev/null -initrd /dev/null \
-qtest stdio -qtest-log /dev/null
readl 0x40000014
OK 0x0000000044000000
read 0x44000084 20
OK 0x6c696e75782c64756d6d792d7669727400000000
where this string is "linux,dummy-virt". QMP could be used to provide
the program counter (via "query-cpus"), but it currently doesn't on ARM.
Perhaps you could make a simple patch to cpus.c for that?
Paolo
