[coreboot] [commit] r5286 - ...
segher at kernel.crashing.org
Fri Mar 26 00:16:21 CET 2010
>> >> I am told that AmlCode is defined as array of (unsigned) char in
>> >> some other file. Declaring it as some other type here is not
>> >> valid C, and *will* break with GCC, with some options (-combine
>> >> or LTO at least) -- it will not compile.
>> > The biggest worry for me is incorrect execution. If it doesn't
>> > when
>> > it breaks, then that's a good thing.
>> It will probably _work_ with current GCC and no whole-program stuff,
>> but how do you *know* it does? Better to just fix it.
> I misunderstood you. I thought you were saying that it wouldn't compile
> you used those options, but that it would work otherwise.
If you use whole-program optimisation, it will not even compile, indeed.
If you do _not_ use whole-program optimisation, it will compile, and
likely even work -- but no guarantees. The "workaround" correct code
isn't very hard or expensive, best to use that.
It is a pity that GCC cannot detect all problems in your code, but that
is equivalent to solving the halting problem (_and_ reading your mind ;-) ).
More information about the coreboot