Messages in this thread | ![/](/images/icornerl.gif) | | Date | Sun, 3 Aug 2003 22:54:27 -0700 (PDT) | From | Linus Torvalds <> | Subject | Re: do_div considered harmful |
| |
On Mon, 4 Aug 2003 Andries.Brouwer@cwi.nl wrote: > > Now that I compare, he wrote > nativeMb = do_div(nativeMb, 1000000); > to divide nativeMb by 1000000. > > So, it seems natural to expect that do_div() gives the quotient. > But it gives the remainder. > (Strange, Erik showed correct output.)
Actually, the above is "undefined behaviour", since it has _two_ assignments in the same thing. Exactly because "do_div()" modifies both the first argument _and_ returns a value. So depending on the implementation of do_div() (whether there are any sequence points etc) and on random compiler behaviour (if there are no sequence points in do_div() internally), in the example above "nativeMb" migth be totally undefined after the above.
And yes, as a special case, it might be the divisor.
I agree that "do_div()" has strange semantics and is very likely misnamed, but they are kind of forced upon us by the fact that C functions cannot return two values. Renaming do_div() at this point is just going to make it harder to write kernel- portable source, so I suspect we're better off just commenting it, and making people aware of how do_div() works.
Not very many people should use "do_div()" directly (and a quick grep shows that not very many people do). It's generally a mistake to do so, I suspect. The thing was originally written explicitly for "printk()" and nothing else.
Linus
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| ![\](/images/icornerr.gif) |