Messages in this thread |  | | From | Linus Torvalds <> | Date | Fri, 26 Oct 2012 12:41:13 -0700 | Subject | Re: [PATCH 4/5] IPMI: Fix some uninitialized warning |
| |
On Fri, Oct 26, 2012 at 12:35 PM, Corey Minyard <cminyard@mvista.com> wrote: > > I'll do a patch to switch it to uninitialized_var() for the next set of > things I send. Thanks.
Actually, I'm seriously considering getting rid of uninitialized_var().
The thing is an abomination, and we've already had one bug due to it (real uninitialized variable that was incorrectly hidden by uninitialized_var()).
The thing is moronic. The whole thing is almost entirely due to compiler bugs (*stupid* gcc behavior), and we would have been better off with an explicit (unnecessary) initialization that at least doesn't cause random crashes etc if it turns out to be wrong.
So please don't use that abortion. Just initialize things to 0/NULL/whatever if gcc complains and you want to get rid of them. Better than the crap that is uninitialized_var(), which just causes silent random behavior if it is wrong.
Linus
|  |