Am 31.05.2011 12:05, schrieb Hans de Goede:
> Hi,
>
> On 05/31/2011 11:56 AM, Kevin Wolf wrote:
>> Am 31.05.2011 11:51, schrieb Hans de Goede:
>>> Hi,
>>>
>>> On 05/31/2011 11:42 AM, Michael Tokarev wrote:
>>>> 31.05.2011 13:35, Hans de Goede wrote:
>>>>> ---
>>>>> hw/usb-bus.c | 23 ++++++++++++-----------
>>>>> hw/usb-msd.c | 5 +++--
>>>>> usb-linux.c | 6 +++++-
>>>>> 3 files changed, 20 insertions(+), 14 deletions(-)
>>>>>
>>>>> diff --git a/hw/usb-bus.c b/hw/usb-bus.c
>>>>> index 0a49921..2ae2678 100644
>>>>> --- a/hw/usb-bus.c
>>>>> +++ b/hw/usb-bus.c
>>>>
>>>>> if (dev->attached) {
>>>>> - fprintf(stderr, "Warning: tried to attach usb device %s twice\n",
>>>>> + fprintf(stderr, "Error: tried to attach usb device %s twice\n",
>>>>> dev->product_desc);
>>>>
>>>> qemu_error() maybe, while we're at it?
>>>> Here and in a few other places.
>>>
>>> That does not seem to exist, do you perhaps mean error_printf() ?
>>
>> error_report() is what you should use, so that messages go to the
>> monitor if the function is called from a monitor command. error_printf()
>> is used by it internally, but usually isn't used directly.
>>
>
> I've looked at error_report, but IMHO it is made of crazy, I'm not going
> to construct a json dict every time I need to log some simple error message
> (and the existing ones are not suitable for many error messages).
error_report() works with plain strings. Maybe you confuse it with the
QMP error reporting function?
Kevin
