On 10 February 2017 at 15:19, Alex Bennée <alex.ben...@linaro.org> wrote: > Peter Maydell <peter.mayd...@linaro.org> writes: >> I guess a malloc-and-free is OK since the guest isn't going to be >> bouncing CPUs through reset very often, though it's a bit ugly to >> see in device code. > > Previous patches had expanded the run_on_cpu code to have things like > CPUState and a single field to avoid malloc where we can. However I need > the IMX6SRCState and I don't know if I can get that in the work > function. Will there only ever be one on the system?
In practice there will be only one but I don't think we should rely on that. malloc/free is probably better than overly convoluted code at this point. thanks -- PMM
