On 2015-06-05 17:03, Yousong Zhou wrote: > > On Jun 5, 2015 3:14 PM, "Felix Fietkau" <[email protected] > <mailto:[email protected]>> wrote: >> >> On 2015-06-04 15:41, Yousong Zhou wrote: >> > - Return early on calloc() failure. >> > - Correct comment text for __calloc_a(). >> > >> > Signed-off-by: Yousong Zhou <[email protected] > <mailto:[email protected]>> >> > --- >> > utils.c | 4 ++++ >> > utils.h | 6 +++--- >> > 2 files changed, 7 insertions(+), 3 deletions(-) >> > >> > diff --git a/utils.c b/utils.c >> > index 8fd19f4..627b0f6 100644 >> > --- a/utils.c >> > +++ b/utils.c >> > @@ -43,6 +43,10 @@ void *__calloc_a(size_t len, ...) >> > va_end(ap1); >> > >> > ptr = calloc(1, alloc_len); >> > + if (!ptr) { >> > + va_end(ap); >> > + return NULL; >> > + } >> What's the point? The return value without this check will be NULL >> anyway, and optimizing a rare error case does not seem useful to me. >> > > rare case indeed, also a trivial change. null check is for > correctness. it also seems to me assigning invalid pointer values to > output arguments is not right. well, my 2 cents :) Well, the way you implemented it, the output argument values will be undefined, which is not significantly better :)
- Felix _______________________________________________ openwrt-devel mailing list [email protected] https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
