Re: [PATCH 7/8] percpu: add __percpu sparse annotations tohw_breakpoint

From: Al Viro
Date: Mon Jan 25 2010 - 22:57:59 EST


On Tue, Jan 26, 2010 at 12:10:51PM +0900, Tejun Heo wrote:
> Hello,
>
> On 01/26/2010 11:48 AM, Al Viro wrote:
> > On Tue, Jan 26, 2010 at 11:43:56AM +0900, Tejun Heo wrote:
> >
> >>> Eh... You are leaving that noderef in place in case of array. And _that_
> >>> is not an address space, so casts to AS 0 won't do you any good.
> >>
> >> Any ideas on how to fix it?
> >
> > BTW, before we go any further, which warnings are you getting from sparse
> > and which version of sparse are you using?
> >
> > noderef is one thing; address_space mess is a different story. The version
> > I have here steps into the former, but not the latter; what are you seeing?
>
> Oops, I too am seeing the noderef thing not the address space warning.

OK... So all messing around __kernel __force is actually a red herring.

Frankly, for now I'd keep it as in your patch. Yes, including workarounds
in these few places. Longer term... We probably want to implement
__attribute__((qualify(...)))/__attribute__((unqualify(...))), revert
__typeof__ for AS/noderef to what gcc is doing for normal qualifiers
(i.e. "if p is int const *, typeof(*p) v gives const int") go for explicit
__unqualify((address_space,noderef)) in there. Playing interesting games
with arrays for unqualify (i.e. creating parallel chains of type nodes
all way down to the place where original qualifier had been applied).
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/