[06/12] hw/arm/virt: Fix property "gic-version" error handling

virt_set_gic_version() calls exit(1) when passed an invalid property
value.  Property setters are not supposed to do that.  Screwed up in
commit b92ad39.  Harmless, because the property belongs to a machine.
Set an error object instead.

Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: qemu-arm@nongnu.org
Signed-off-by: Markus Armbruster <armbru@redhat.com>
---
 hw/arm/virt.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

On 10 December 2015 at 10:29, Markus Armbruster <armbru@redhat.com> wrote:
> virt_set_gic_version() calls exit(1) when passed an invalid property
> value.  Property setters are not supposed to do that.  Screwed up in
> commit b92ad39.  Harmless, because the property belongs to a machine.
> Set an error object instead.
>
> Cc: Peter Maydell <peter.maydell@linaro.org>
> Cc: qemu-arm@nongnu.org
> Signed-off-by: Markus Armbruster <armbru@redhat.com>
> ---
>  hw/arm/virt.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/hw/arm/virt.c b/hw/arm/virt.c
> index 9c6792c..2a120dd 100644
> --- a/hw/arm/virt.c
> +++ b/hw/arm/virt.c
> @@ -1126,9 +1126,8 @@ static void virt_set_gic_version(Object *obj, const char *value, Error **errp)
>      } else if (!strcmp(value, "host")) {
>          vms->gic_version = 0; /* Will probe later */
>      } else {
> -        error_report("Invalid gic-version option value");
> -        error_printf("Allowed gic-version values are: 3, 2, host\n");
> -        exit(1);
> +        error_setg(errp, "Invalid gic-version value");
> +        error_append_hint(errp, "Valid values are: 3, 2, host\n");
>      }

Should hint strings have trailing newlines?

Otherwise
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Peter Maydell <peter.maydell@linaro.org> writes:

> On 10 December 2015 at 10:29, Markus Armbruster <armbru@redhat.com> wrote:
>> virt_set_gic_version() calls exit(1) when passed an invalid property
>> value.  Property setters are not supposed to do that.  Screwed up in
>> commit b92ad39.  Harmless, because the property belongs to a machine.
>> Set an error object instead.
>>
>> Cc: Peter Maydell <peter.maydell@linaro.org>
>> Cc: qemu-arm@nongnu.org
>> Signed-off-by: Markus Armbruster <armbru@redhat.com>
>> ---
>>  hw/arm/virt.c | 5 ++---
>>  1 file changed, 2 insertions(+), 3 deletions(-)
>>
>> diff --git a/hw/arm/virt.c b/hw/arm/virt.c
>> index 9c6792c..2a120dd 100644
>> --- a/hw/arm/virt.c
>> +++ b/hw/arm/virt.c
>> @@ -1126,9 +1126,8 @@ static void virt_set_gic_version(Object *obj, const char *value, Error **errp)
>>      } else if (!strcmp(value, "host")) {
>>          vms->gic_version = 0; /* Will probe later */
>>      } else {
>> -        error_report("Invalid gic-version option value");
>> -        error_printf("Allowed gic-version values are: 3, 2, host\n");
>> -        exit(1);
>> +        error_setg(errp, "Invalid gic-version value");
>> +        error_append_hint(errp, "Valid values are: 3, 2, host\n");
>>      }
>
> Should hint strings have trailing newlines?

Two answers:

1. No, because error_report_err() prints the hint followed by a newline.

2. Yes, because error_appent_hint() accumulates, and requiring its users
to know which call will be the final one is an awkward interface.

I'll change error_report_err() and improve the documentation.

> Otherwise
> Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

Thanks!