28.04.2014 14:19, Peter Maydell wrote:
> On 27 April 2014 17:26, Michael Tokarev <m...@tls.msk.ru> wrote:
>> -static QemuMutex vevent_queue_lock;
>> -static QemuCond vevent_queue_condition;
>> +static GMutex vevent_queue_lock;
>> +static GCond vevent_queue_condition;
>>
>> void vevent_queue_init(void)
>> {
>> - qemu_mutex_init(&vevent_queue_lock);
>> - qemu_cond_init(&vevent_queue_condition);
>> + g_mutex_init(&vevent_queue_lock);
>> + g_cond_init(&vevent_queue_condition);
>> vevent_queue_head = vevent_queue_tail = NULL;
>> }
>
> Just to record a conversation from IRC: this won't
> compile on older versions of glib, because those don't
> support g_cond_init(). Conversely, if you use g_cond_new()
> then newer glib will complain that it's deprecated.
> trace/simple.c has an example of how you have to work
> around this API churn...
I've added a tiny (but hackish and fun) wrapper header for all
this, and pushed whole thing into a branch on my site -- see
http://git.corpit.ru/?p=qemu.git;a=shortlog;h=refs/heads/libcacard-standalone
The wrapper is here:
http://git.corpit.ru/?p=qemu.git;a=blob;f=libcacard/glib-compat.h;h=47fd06a43572fc5cc639fb7d705e85f3965d6b91;hb=1aeefee2a4bb2694b2314cddd04e08675360e255
It allows to use the new primitives/interface with either new (unchanged)
or old (wrapped) glib cond and mutex API.
I think this is better than a way used in trace/simple.c, and can be
used there too, making stuff much more readable.
Please note: the wrapper redifines some core symbols which are used in
glib so it must come somewhere near the end of #include set. Most
interesting is that it redefines GCOnd into GCond* (only for old glib).
Thanks,
/mjt
