Peter Maydell <peter.mayd...@linaro.org> writes:
> On 13 February 2013 12:57, Markus Armbruster <arm...@redhat.com> wrote:
>>> - "-d options activate log (default logfile=%s)\n"
>>> - "-D logfile override default logfile location\n"
>>> + "-d options activate log (default is to log to stderr)\n"
>>> + "-D logfile write logs to 'logfile' rather than stderr\n"
>>> "-p pagesize set the host page size to 'pagesize'\n"
>>> "-singlestep always run in singlestep mode\n"
>>> "-strace log system calls\n"
>>
>> No need to mention the default twice.
>>
>> Pointing to -d help would be nice.
>
> Yeah. I'll standardise the phrasing here bsd-user, linux-user
> and system mode all say
> "-d item1[,...] enable logging of specified items\n"
> " (use '-d help' for a list of log items)\n"
> "-D logfile write logs to 'logfile' (default stderr)\n"
Looks good.
>>> @@ -861,7 +858,9 @@ int main(int argc, char **argv)
>>> }
>>>
>>> /* init debug */
>>> - qemu_set_log_filename(log_file);
>>> + if (log_file) {
>>> + qemu_set_log_filename(log_file);
>>> + }
>>> if (log_mask) {
>>> int mask;
>>>
>>
>> Doesn't qemu_set_log_filename(NULL) do the right thing?
>
> Yes, it does. I'll drop this hunk.
>
>> bsd-user: if there's more than one -D, all but the last get silently
>> ignored. We create that log file when logging is enabled.
>>
>> linux-user: we create all the log files when logging is enabled.
>>
>> No biggie, but consistency would be nice. I prefer bsd-user's behavior.
>
> I agree, but I think changing the linux-user behaviour to
> match bsd-user should be a separate patch.
Makes sense.
