Re: [PATCH] m68k: virt: correctly set the initial PC

[Philippe Mathieu-Daudé]

On 5/4/21 4:20 PM, Laurent Vivier wrote:
> Le 04/05/2021 à 16:13, Philippe Mathieu-Daudé a écrit :
>> On 5/4/21 3:18 PM, Laurent Vivier wrote:
>>> Set initial PC to the entry of the loaded kernel.
>>>
>>> This fixes kernel reboot with "-kernel" parameter.
>>>
>>> Signed-off-by: Laurent Vivier <laurent@vivier.eu>
>>> ---
>>>  hw/m68k/virt.c | 22 +++++++++++++++++-----
>>>  1 file changed, 17 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/hw/m68k/virt.c b/hw/m68k/virt.c
>>> index e9a5d4c69b97..4fb3a7ebf0f2 100644
>>> --- a/hw/m68k/virt.c
>>> +++ b/hw/m68k/virt.c
>>> @@ -88,14 +88,21 @@
>>>  #define VIRT_VIRTIO_MMIO_BASE 0xff010000     /* MMIO: 0xff010000 - 
>>> 0xff01ffff */
>>>  #define VIRT_VIRTIO_IRQ_BASE  PIC_IRQ(2, 1)  /* PIC: 2, 3, 4, 5, IRQ: ALL 
>>> */
>>>  
>>> +typedef struct {
>>> +    M68kCPU *cpu;
>>> +    hwaddr initial_pc;
>>> +    hwaddr initial_stack;
>>> +} ResetInfo;
>>> +
>>>  static void main_cpu_reset(void *opaque)
>>>  {
>>> -    M68kCPU *cpu = opaque;
>>> +    ResetInfo *reset_info = opaque;
>>> +    M68kCPU *cpu = reset_info->cpu;
>>>      CPUState *cs = CPU(cpu);
>>>  
>>>      cpu_reset(cs);
>>> -    cpu->env.aregs[7] = ldl_phys(cs->as, 0);
>>> -    cpu->env.pc = ldl_phys(cs->as, 4);
>>> +    cpu->env.aregs[7] = reset_info->initial_stack;
>>> +    cpu->env.pc = reset_info->initial_pc;
>>>  }
>>>  
>>>  static void virt_init(MachineState *machine)
>>> @@ -116,6 +123,7 @@ static void virt_init(MachineState *machine)
>>>      SysBusDevice *sysbus;
>>>      hwaddr io_base;
>>>      int i;
>>> +    ResetInfo *reset_info;
>>>  
>>>      if (ram_size > 3399672 * KiB) {
>>>          /*
>>> @@ -127,9 +135,13 @@ static void virt_init(MachineState *machine)
>>>          exit(1);
>>>      }
>>>  
>>> +    reset_info = g_malloc0(sizeof(ResetInfo));
>>> +
>>>      /* init CPUs */
>>>      cpu = M68K_CPU(cpu_create(machine->cpu_type));
>>> -    qemu_register_reset(main_cpu_reset, cpu);
>>> +
>>> +    reset_info->cpu = cpu;
>>> +    qemu_register_reset(main_cpu_reset, reset_info);
>>>  
>>>      /* RAM */
>>>      memory_region_add_subregion(get_system_memory(), 0, machine->ram);
>>> @@ -209,7 +221,7 @@ static void virt_init(MachineState *machine)
>>>              error_report("could not load kernel '%s'", kernel_filename);
>>>              exit(1);
>>>          }
>>> -        stl_phys(cs->as, 4, elf_entry); /* reset initial PC */
>>> +        reset_info->initial_pc = elf_entry;
>> Missing the stack?
>>
>>            reset_info->initial_stack = ldl_phys(cs->as, 0);
> 
> No, as it's a g_malloc0(), initial_stack is set to 0, so aregs[7] (SP) will 
> be reset to 0 on reboot.
> We could also set it to  ram_size if we want something usable, but it's not 
> needed for the kernel
> entry point.

Indeed I missed g_malloc*0*.

It would be clearer to use g_malloc() and explicitly set the initial
stack IMHO.

> 
> Thanks,
> 
> Laurent
> 
>