On 21 February 2012 09:15, Peter Crosthwaite <peter.crosthwa...@petalogix.com> wrote: > On Tue, Feb 21, 2012 at 5:56 AM, Peter Maydell <peter.mayd...@linaro.org> > wrote: >> On 20 February 2012 19:51, Andreas Färber <afaer...@suse.de> wrote: >>> Am 20.02.2012 20:43, schrieb Peter Maydell: >>>> I don't particularly care how we QOMify arm_boot (it's not exactly at >>>> the top of my priority list demanding attention), I do care that (a) >>>> we have a sensible user-facing interface [ie command line options] and >>>> (b) vl.c can usefully just pass the information from those options >>>> straight to the boot loader code. > > So when I trialled this patch I used the -device argument to > instantiate the bootloader on the command line: > > qemu-system-arm -M versatilepb --device > arm_linux_loader,kernel_filename=kernel.foo,initrd=image.bar... > > With this approach I was able to add command line args (if you will) > to arm_boot without touching vl.c, qemu_opts or any of the arm machine > models.
I think this is wrong, because it means the user has to know what the name of the arm_boot device is (as well as because it means that the machine model can't set the arm_boot properties that it needs to, as Paul and Andreas point out). vl.c should search for "some device, any device, which provides a kernel loading interface" and set the kernel/initrd/cmdline/dtb properties on it, so that the same kind of command line works on different machines even if they implement different boot loader devices, subclass arm_boot or whatever. As you say, the infrastructure to do it properly is kind of missing at the moment; that's why I don't want to get basic DTB tied up with QOMification. -- PMM
