Re: [PATCH] allow context attribute on variables

[Josh Triplett <josh@freedesktop.org>]

[-- Attachment #1: Type: text/plain, Size: 1328 bytes --]

Johannes Berg wrote:
> This patch makes it possible to add the context attribute on
> variables, to warn for example in this case:
> 
>     struct something {
>         int x __attribute__((context(L,1,1)));
>     };
> 
>     extern struct something *s;
> 
>     static void warn_access13(void)
>     {
>         s->x = 7;
>     }
> 
> This is achieved by translating the context attribute on
> variables that are loaded from/stored to into a context
> expression internally. A number of tests are included,
> including tests for a struct member (as above) and array
> accesses.
> 
> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
> ---
> I didn't see any need for distinguishing between rd/wr nor
> for a new attribute, so I just used the one we already have.
> 
> The exact_context() attribute is also allowed so we have the
> same max/no-max semantics as functions.
> 
> If read/write is necessary to distinguish (I don't see a use
> case) then we can do that fairly easily later.

I like the idea of matching the existing attributes as closely as
possible.  However, RCU provides a use case for distinguishing readers
and writers: you might have a variable that you can read if you hold
the RCU reader "lock", but you can only write if you hold a spinlock.

- Josh Triplett


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]