On Wed, 22 May 2019, Philipp Reisner wrote:
> Am Mittwoch, 22. Mai 2019, 11:37:40 CEST schrieb Julia Lawall:
> > On Wed, 22 May 2019, Philipp Reisner wrote:
> > > Hi,
> > >
> > > [...]
> > >
> > > I love the whole idea, but gave up working on such a things myself.
> > >
> > > > > Because clang analysis wants it to be global!
> > > >
> > > > A __protected_by() annotation sounds like a good idea. I don't really
> > > > care about the format too much. If the information were in a comment
> > > > and we could parse with a perl script that would be fine. Or we could
> > > >
> > > > do:
> > > > struct foo {
> > > >
> > > > struct mutex lock;
> > > > __start_protected(lock);
> > > > int a, b, c;
> > > > __end_protected(lock);
> > > >
> > > > };
> > >
> > > Regarding the syntax I vote for a __protected_by(lock) instead of
> > > __start_protected(lock) / __end_protected(lock).
> >
> > What do you mean exactly, eg in the above example?
> >
> struct foo {
> struct mutex lock;
> int a __protected_by_(lock);
> int b __protected_by_(lock);
> };
OK, thanks. Might the lock be in another structure (eg struct upperfoo)?
So
int a __protected_by_(upperfoo.lock);
julia
