Re: [Qemu-devel] [PULL 5/5] bootdevice: add Error **errp argument for QEMUBootSetHandler

[Peter Maydell <peter.maydell@linaro.org>]

On 16 December 2014 at 13:04, Gonglei <arei.gonglei@huawei.com> wrote:
> On 2014/12/16 20:42, Peter Maydell wrote:
>
>> On 16 December 2014 at 09:22,  <arei.gonglei@huawei.com> wrote:
>>> @@ -412,9 +411,7 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
>>>      object_property_set_link(OBJECT(machine), OBJECT(s),
>>>                               "rtc_state", &error_abort);
>>>
>>> -    if (set_boot_dev(s, boot_device)) {
>>> -        exit(1);
>>> -    }
>>> +    set_boot_dev(s, boot_device, &error_abort);
>>
>> This turns a "print error message and exit" path into
>> an abort(), which doesn't seem right (this can be triggered
>> by bad user input arguments, yes?). error_abort should
>> only be used in cases where you would assert() if there
>> was an error (ie where it would be a QEMU bug if it
>> happened).
>>
>
> Yes, agree. How does use a incremental patch fix this, Peter?
>
> diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> index 99deba6..d7822b8 100644
> --- a/hw/i386/pc.c
> +++ b/hw/i386/pc.c
> @@ -364,6 +364,7 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
>      FDriveType fd_type[2] = { FDRIVE_DRV_NONE, FDRIVE_DRV_NONE };
>      static pc_cmos_init_late_arg arg;
>      PCMachineState *pc_machine = PC_MACHINE(machine);
> +    Error *local_err = NULL;
>
>      /* various important CMOS locations needed by PC/Bochs bios */
>
> @@ -411,7 +412,10 @@ void pc_cmos_init(ram_addr_t ram_size, ram_addr_t above_4g_mem_size,
>      object_property_set_link(OBJECT(machine), OBJECT(s),
>                               "rtc_state", &error_abort);
>
> -    set_boot_dev(s, boot_device, &error_abort);
> +    set_boot_dev(s, boot_device, &local_err);
> +    if (local_err) {
> +        exit(1);
> +    }

That won't print the error message at all...

-- PMM