Messages in this thread | | | Date | Thu, 2 Jul 2015 17:06:29 +0300 | From | Sergei Zviagintsev <> | Subject | Re: [PATCH RFC 3/5] kdbus: do explicit overflow check in kdbus_conn_quota_inc() |
| |
Hi,
On Thu, Jul 02, 2015 at 10:35:30AM +0200, David Herrmann wrote: > Hi > > On Sun, Jun 28, 2015 at 3:17 PM, Sergei Zviagintsev <sergei@s15v.net> wrote: > > Replace the use of max() with explicit and obvious overflow check. > > > > Signed-off-by: Sergei Zviagintsev <sergei@s15v.net> > > --- > > ipc/kdbus/connection.c | 6 ++++-- > > 1 file changed, 4 insertions(+), 2 deletions(-) > > > > diff --git a/ipc/kdbus/connection.c b/ipc/kdbus/connection.c > > index af044f93c14f..1d44e280eff0 100644 > > --- a/ipc/kdbus/connection.c > > +++ b/ipc/kdbus/connection.c > > @@ -668,9 +668,11 @@ int kdbus_conn_quota_inc(struct kdbus_conn *c, struct kdbus_user *u, > > > > id = u ? u->id : KDBUS_USER_KERNEL_ID; > > if (id >= c->n_quota) { > > - unsigned int users; > > + unsigned int users = KDBUS_ALIGN8(id) + 8; > > + > > + if (users < id) /* overflow */ > > + users = id; > > > > - users = max(KDBUS_ALIGN8(id) + 8, id); > > To me, the max() looks fine. I mean, it should be obvious that it > checks for an overflow, right? Otherwise, I'd prefer adding a comment > instead of the explicit conditional.
Well, at first sight I assumed that there was some clever algorithm which later had been quick-fixed and so on...
I am not sure about commenting current variant, as CodingStyle tells us to write obvious code instead of explanations. And if it is already obvious, then it doesn't need a comment, right? :)
In another letter of this series I mentioned commits af8e2f750985 and ac5c385cc67a from out-of-tree kdbus repo, which show that not all of these overflow things are obvious for maintainer, not talking about someone who isn't familiar with the code.
> > Thanks > David > > > quota = krealloc(c->quota, users * sizeof(*quota), > > GFP_KERNEL | __GFP_ZERO); > > if (!quota) > > -- > > 1.8.3.1 > >
| |