Philipp Reisner wrote: > The sematics of what I implemented is: > You simply use the __attribute__((context(ctx,in,out))) to > annotate the context changes of functions. > > to annotate a variable, struct/union member or a function that > a certain locking primitive is required for accessing it, you > do it by > > __attribute__((require_context(ctx,min,max,"type"))) > > ctx ... the expression that describes the locking primitive > min ... the minimum of locks required of that locking primitive > max ... the maximum allowed of locks of that locking primitive. > type .. read, write, rdwr or call for the access type. > > So you can express you need to hold this and that locking > primitive to write to something. But an other locking > primitive might be sufficient for reading that something. > > The annotation for a variable foo protected by a recursive lock > bar would be: > > int foo __attribute__((require_context(bar,1,99999,"rdwr"))) I would *love* to see this work applied on top of the patches I applied from Johannes. I really want the ability to mark a variable or struct field as requiring a context to access. I would probably call this new attribute something like "data_context". Also, I'd prefer an explicit way of saying "no maximum context", rather than the hack of using a large number. I hestiate to introduce a third semantic for a pair of numbers describing a context. The context attribute uses in and out contexts, __context__ uses a delta and a required minimum, and this proposed attribute uses a minimum and a maximum. Similar to GCC's format attribute using the identifier printf rather than the string "printf", I think I'd prefer the last argument as an identifier. Also, a minor nit: could you please use "readwrite" rather than "rdwr"? - Josh Triplett
Attachment:
signature.asc
Description: OpenPGP digital signature
