On Fri, Jun 24, 2016 at 6:35 AM, Jakub Jelinek <ja...@redhat.com> wrote: > On Thu, Jun 23, 2016 at 05:23:21PM +0200, Jakub Jelinek wrote: >> Thinking about this again, there could be another option - keep >> __atomic_compare_exchange_N in the IL, but under certain conditions (similar >> to what the patch uses in fold_builtin_atomic_compare_exchange) for these >> builtins ignore &var on the second argument, and if we actually turn var >> into non-addressable, convert the builtin call similarly to what >> fold_builtin_atomic_compare_exchange does in the patch (except the store >> would be non-conditional then; the gimple-fold.c part wouldn't be needed >> then). > > Here is this second approach implemented. Generally it gives slight code > size savings over the other patch on the testcases I've posted (and even the > other patch used to produce generally smaller code than vanilla). > > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? > > 2016-06-24 Jakub Jelinek <ja...@redhat.com> > > PR middle-end/66867 > * builtins.c (expand_ifn_atomic_compare_exchange_into_call, > expand_ifn_atomic_compare_exchange): New functions. > * internal-fn.c (expand_ATOMIC_COMPARE_EXCHANGE): New function. > * tree.h (build_call_expr_internal_loc): Rename to ... > (build_call_expr_internal_loc_array): ... this. Fix up type of > last argument. > * internal-fn.def (ATOMIC_COMPARE_EXCHANGE): New internal fn. > * predict.c (expr_expected_value_1): Handle IMAGPART_EXPR of > ATOMIC_COMPARE_EXCHANGE result. > * builtins.h (expand_ifn_atomic_compare_exchange): New prototype. > * gimple-fold.h (optimize_atomic_compare_exchange_p, > fold_builtin_atomic_compare_exchange): New prototypes. > * gimple-fold.c (optimize_atomic_compare_exchange_p, > fold_builtin_atomic_compare_exchange): New functions.. > * tree-ssa.c (execute_update_addresses_taken): If > optimize_atomic_compare_exchange_p, ignore &var in 2nd argument > of call when finding addressable vars, and if such var becomes > non-addressable, call fold_builtin_atomic_compare_exchange. >
This caused: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71716 H.J.
