On 11/04/2011 02:14 PM, Jeff Law wrote:
These are the changes to libstdc++ to make use of the new atomics.
I changed the files to use the new atomics, and bkoz did a
shuffling of the include file layout to better suit the new c++
approach.
previously, libstdc++ provided a locked implementation in
atomic_0.h with the theory that eventually it would be used. The
new scheme involves leaving non-lock-free implementations to an
external library. This involved removing the old lock
implementation and restructuring things now that multiple
implementation dont have to be supported. SO a lot fo this is
churn... 2 include files deleted and one merged into another one..
Is there an ABI/API impact to all this churn? Or was the atomic_0.h
stuff never used? Was it an exposed API or strictly internal? I get
the impression that is_lock_free wasn't ever defined, which is
probably good in this context.
Given these are primarily Benjamin's changes, I'm going to defer to
his wisdom :-) If he doesn't chime in, ping him directly.
well, actually, the way it was setup, I don't think it was ever used.
Things defaulted to using the lockfree __sync routines, and if you tried
to do something that didn't map to one of those, you got either a
compiler error (in the case of non-integral classes), or external
__sync calls. It was in a transitional state, and never fully flushed
out. At least thats was my impression when I started trying to work with
it to get full support. I dont think you could trigger the code in
atomic_0.h without changing the compiler/headerfiles yourself. At
least I dont think so. Well, It may have been possible by explicitly
specifying the __atomic0 namespace with baseclasses... Im not sure.
I'll let bkoz deal with answering that :-) Since support was
experimental, I dont think anyone was suppose to count on it anyway :-P
I believe that any atomic integral that worked before will work today
with these changes and without an external library... With the addition
of the forthcoming atomic library (which is not a part of gcc) you
should be able to compile and run anything, including arbitrary user
PODs. WIth any luck at all, the atomic library ABI I set up is defined
enough now to continue to function as is in future releases.
Andrew
