Re: [PATCH] ACPI: Avoid infinite recursion when dump-vmstate

[Igor Mammedov]

On Mon, 19 Oct 2020 17:31:56 +0800
Peng Liang <liangpeng10@huawei.com> wrote:

> There is a field with vmstate_ghes_state as vmsd in vmstate_ghes_state,
> which will lead to infinite recursion in dump_vmstate_vmsd.
> 
> Fixes: a08a64627b ("ACPI: Record the Generic Error Status Block address")
> Reported-by: Euler Robot <euler.robot@huawei.com>
> Signed-off-by: Peng Liang <liangpeng10@huawei.com>
> ---
>  hw/acpi/generic_event_device.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/hw/acpi/generic_event_device.c b/hw/acpi/generic_event_device.c
> index 6df400e1ee16..4b6867300a55 100644
> --- a/hw/acpi/generic_event_device.c
> +++ b/hw/acpi/generic_event_device.c
> @@ -334,8 +334,7 @@ static const VMStateDescription vmstate_ghes_state = {
>      .minimum_version_id = 1,
>      .needed = ghes_needed,
>      .fields      = (VMStateField[]) {
> -        VMSTATE_STRUCT(ghes_state, AcpiGedState, 1,
> -                       vmstate_ghes_state, AcpiGhesState),
> +        VMSTATE_UINT64(ghes_state.ghes_addr_le, AcpiGedState),

not sure its' ok handle it this way,

see how it is done with another structure:

static const VMStateDescription vmstate_ged_state = {                           
 
    .name = "acpi-ged-state",                                                   
 
    .version_id = 1,                                                            
 
    .minimum_version_id = 1,                                                    
 
    .fields      = (VMStateField[]) {                                           
 
        VMSTATE_UINT32(sel, GEDState),                                          
 
        VMSTATE_END_OF_LIST()                                                   
 
    }                                                                           
 
}; 

...

VMSTATE_STRUCT(ged_state, AcpiGedState, 1, vmstate_ged_state, GEDState),

i.e. it looks like we are missing structure definition for AcpiGhesState

CCing David,
 to help with migration magic in case I'm wrong or missed something

>          VMSTATE_END_OF_LIST()
>      }
>  };