Re: [PATCH 1/3] make sparse keep its promise about context tracking

[Johannes Berg <johannes@sipsolutions.net>]

[-- Attachment #1: Type: text/plain, Size: 874 bytes --]


> > and you'd end up with one context named "dbl" and another one named
> > "&dev_base_lock".
> 
> That might still work, depending on how consistently kernel code uses
> the same argument.  If you suggest explicitly changing calls to
> read_lock to call __acquire__ with the appropriate context, it might
> prove equally easy to make the argument of read_lock that context.

True.

> That doesn't make sense to me; after preprocessing, x no longer
> exists, so I don't see how you could pick up the identifier "x".  I
> can understand that you might pick up two different expressions in
> place of x which you can't easily compare, though.

Oops, yes, preprocessing will remove it, of course. But yeah, you can
end up with two different expressions.

> And as for how to solve it: I think alias analysis might work.

How would I do that?

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]