On 01/05/2016 06:40 PM, Stefano Stabellini wrote:
On Tue, 5 Jan 2016, Cao jin wrote:
[...]
This is true, however I think that relying on it is error prone: in a
couple of years from now somebody might change the call sequence without
updating the error handling (easy to forget), causing QEMU to crash on
error. I think it is safer not to rely on errp != NULL.
I see, sounds reasonable
[...]
I think we only need to use a local Error variable when we want to check
for the returned error, in cases such as:
if (*errp) {
In other cases, when we are not interested in *errp, we can simply
propagate the error, like you have done in your patches.
Ok, will fix it
.
--
Yours Sincerely,
Cao Jin
