Hi Dan,
2012/4/16 Dan Carpenter <dan.carpenter@xxxxxxxxxx>:
>
> Don't start messing in scheduler code if you have any doubts about
> what you are doing. My advise is don't send a patch for this
> warning. Most times people leave BUG() enabled so we don't really
> care about the warning.
Okey. Perhaps I picked the wrong example, there are a whole bunch of warnings
related to the same BUG() stuff, here's a template example:
void *p;
switch (something) {
case a:
p = foo();
case b:
p = bar();
default:
BUG();
}
dosomething(p);
--
This will trigger warning "p could be used uninitialized" or something.
I know most of us just leave BUG on, so you say these warnings aren't important?
>> ---
>> 2.
>> Another one is flat_set_persistent() which is only IMO properly defined in
>> arch/sh/. The current
>> definition produces a warning in other architectures if binary flat format
>> is used.
>>
>> So, I guess I should send a patch to each arch-specific list, right? using
>> arch-specific git trees?
>> I would like to have just one git tree, and not ten :)
>
> I'm not sure what you mean here. What is the warning? It sounds
> like something is wrong in a Kconfig file.
>
I'll explain better. flat_set_persistent() is used by no-one, except
from blackfin arch
where it has something in it. So it gets defined to nothing in every
architecture, except
from blackfin.
Now, arch/sh/include/asm/flat.h defines it like:
#define flag_set_persisten(relval,p) { (void)p; 0; })
while the rest of them just defines it to zero:
#define flag_set_persisten(relval,p) 0
This will cause a warning for unused variable p, in fs/binfmt_flat.c
and IMO the arch/sh is doing it right (just to avoid the warning).
I tested both definitions with bloat-o-metering and it showed no
extra code.
Do you see my point now?
Thanks for your help,
Ezequiel.
--
To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
