On Tue, Feb 27, 2007 at 04:07:00PM +0100, Tilman Schmidt wrote:
> Kyle McMartin schrieb:
> > On Tue, Feb 27, 2007 at 12:02:50PM +0100, Geert Uytterhoeven wrote:
> >> But we don't run sparse on every build. We do run gcc (a lot).
> >
> > We probably could, in theory...
>
> Not usefully. I try it o
Kyle McMartin wrote:
(Or perhaps gcc can be extended such that a string of allowable qualifiers
can be passed along in kernel.h, that would just be super...)
That would of course be the best. Something like an
"format_custom_size("r", resource_t)" attribute.
-hpa
-
To unsubscribe
Kyle McMartin schrieb:
> On Tue, Feb 27, 2007 at 12:02:50PM +0100, Geert Uytterhoeven wrote:
>> But we don't run sparse on every build. We do run gcc (a lot).
>
> We probably could, in theory...
Not usefully. I try it occasionally, but there's way too much spewage,
in particular of the "symbol wa
On Tue, Feb 27, 2007 at 12:02:50PM +0100, Geert Uytterhoeven wrote:
> But we don't run sparse on every build. We do run gcc (a lot).
>
We probably could, in theory...
> > --- a/lib/vsprintf.c
> > +++ b/lib/vsprintf.c
> > @@ -477,6 +478,8 @@ int vsnprintf(char *buf, size_t size, const char *fmt,
On Mon, 26 Feb 2007, Kyle McMartin wrote:
> I submitted this patch a while ago, but nobody commented on it.
> All these casts of resource_size_t to (unsigned long long) seem horribly
> wasteful to me...
>
> Of course, to add a new format qualifier, we lose out on checking the format
> string by gc
I submitted this patch a while ago, but nobody commented on it.
All these casts of resource_size_t to (unsigned long long) seem horribly
wasteful to me...
Of course, to add a new format qualifier, we lose out on checking the format
string by gcc... Nowadays, this role can be done by sparse.
(Or p
6 matches
Mail list logo
