On Thu, May 2, 2019, at 3:06 AM, Peter Maydell wrote: > On Tue, 30 Apr 2019 at 21:29, Alistair Francis <alistai...@gmail.com> wrote: > > > > On Tue, Apr 30, 2019 at 9:02 AM Peter Maydell <peter.mayd...@linaro.org> > > wrote: > > > Can you explain the purpose of the reset code? None of the other > > > v7m boards seem to need to do a manual qemu_register_reset(). > > > > The reset code allows the machine to work with the -kernel option. > > Without the reset override using -kernel results in the guest starting > > at the wrong address. We can use the -device loader option without the > > reset code though. > > That sounds in line with how -kernel works on the other armv7m > boards -- the expectation is that your image file includes a > full vector table and the CPU will read the PC and SP from it > when it resets. If you want "honour the entry point" you can > use -device loader, as you say. > > Ignoring the entry point for -kernel ELF files is certainly > a bit confusing, but I think if we want to change this we should > do it globally, rather than having one board which behaves > differently to the rest. Changing it does have some awkwardness:
Hmm... That is a good point. It is confusing having something just for one board. I'll drop this part and we can re-evaluate later. > * possibility of breaking previously working images I have no way to test the other boards, so this might be difficult to change. > * we can get the initial PC from the ELF entrypoint, but if > we do this what do we do about the initial SP value ? Not sure about this one either. I'm guessing it changes between the different M cores. Alistair > > thanks > -- PMM > >
