Re: bit ops on unsigned long?

[Rusty Russell <rusty@rustcorp.com.au>]

In message <200203250621.g2P6LG023329@vindaloo.ras.ucalgary.ca> you write:
> > These changed are required because otherwise you try to do set_bit on
> > something not aligned as a long on all archs.
> 
> But of course. I'm not denying that. Naturally the type should be
> changed. I thought that was obvious so I didn't bother agreeing. But
> in fact, it already *is* aligned on a long boundary. Better, in
> fact. It's aligned on a 16 byte boundary. Even though the type was
> __u32.

I'm confused:

@@ -212,7 +212,7 @@
 struct minor_list
 {
     int major;
-    __u32 bits[8];
+    unsigned long bits[256 / BITS_PER_LONG];
     struct minor_list *next;
 };

How, exactly, did "bits" end up on a 16-bute boundary before this
patch?

Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.