Re: Hopefully, kmalloc() will always succeed, but if it doesn't then....

[Roland Dreier]

 >         struct mixart_enum_connector_resp *connector;
 >         struct mixart_audio_info_req  *audio_info_req;
 >         struct mixart_audio_info_resp *audio_info;
 > 
 >         connector = kmalloc(sizeof(*connector), GFP_KERNEL);
 >         audio_info_req = kmalloc(sizeof(*audio_info_req), GFP_KERNEL);
 >         audio_info = kmalloc(sizeof(*audio_info), GFP_KERNEL);
 >         if (! connector || ! audio_info_req || ! audio_info) {
 >                 err = -ENOMEM;
 >                 goto __error;
 >         }

This is not a bug.  All of the pointers are initialized, and if
kmalloc() fails, then one of them will be set to NULL.  However,
kfree(NULL) is a perfectly fine thing to do (kfree just returns
immediately in this case).

So this is just a way of saving some tests and optimizing for the
common case when all allocations succeed.  In other words, this is
good code -- although the spacing is slightly bogus: it should be

        if (!connector || !audio_info_req || !audio_info) {

and also using __error as a label is slightly silly -- why not just
make it "error"?

 - R.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/