Re: kdump: resolve sysctl numbers
On Wed, Jul 27, 2011, Otto Moerbeek wrote:
+#define SETNAME(name) do { names = (name); limit = nitems(name); } while (0)
userland is not supposed to use nitems I think? But it keeps sneaking
in because the kernel headers don't protect it.
Re: kdump: resolve sysctl numbers
On Wed, Jul 27, 2011 at 10:58:22AM -0400, Ted Unangst wrote:
On Wed, Jul 27, 2011, Otto Moerbeek wrote:
+#define SETNAME(name) do { names = (name); limit = nitems(name); } while
(0)
userland is not supposed to use nitems I think? But it keeps sneaking
in because the kernel headers don't protect it.
Well, I thought it's ok here since kdump already has it's
fingers into the kernel jar anyway.
-Otto
Re: kdump: resolve sysctl numbers
On Wed, Jul 27, 2011 at 10:58:22AM -0400, Ted Unangst wrote:
On Wed, Jul 27, 2011, Otto Moerbeek wrote:
+#define SETNAME(name) do { names = (name); limit = nitems(name); } while
(0)
userland is not supposed to use nitems I think? But it keeps sneaking
in because the kernel headers don't protect it.
That's right. It's used in some places like pcidump, npppd and tmux, but it's
locally defined as:
#ifndef nitems
#define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
#endif
--
Cheers,
Jasper
Capable, generous men do not create victims, they nurture them.
Re: kdump: resolve sysctl numbers
On 2011 Jul 27 (Wed) at 19:22:34 +0200 (+0200), Jasper Lievisse Adriaanse wrote:
On Wed, Jul 27, 2011 at 10:58:22AM -0400, Ted Unangst wrote:
On Wed, Jul 27, 2011, Otto Moerbeek wrote:
+#define SETNAME(name) do { names = (name); limit = nitems(name); } while
(0)
userland is not supposed to use nitems I think? But it keeps sneaking
in because the kernel headers don't protect it.
That's right. It's used in some places like pcidump, npppd and tmux, but it's
locally defined as:
#ifndef nitems
#define nitems(_a)(sizeof((_a)) / sizeof((_a)[0]))
#endif
What is the reason for this not being kosher yet,
and if it's not ment to be, why isn't it protected by _KERNEL ?
