[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #17 from Jakub Jelinek --- Author: jakub Date: Mon May 20 21:33:46 2019 New Revision: 271440 URL: https://gcc.gnu.org/viewcvs?rev=271440=gcc=rev Log: PR c++/59813 PR target/90418 * function.h (struct function): Add calls_eh_return member. * gimplify.c (gimplify_call_expr): Set cfun->calls_eh_return when gimplifying __builtin_eh_return call. * tree-inline.c (initialize_cfun): Copy calls_eh_return from src_cfun to cfun. (expand_call_inline): Or in src_cfun->calls_eh_return into dst_cfun->calls_eh_return. * tree-tailcall.c (suitable_for_tail_call_opt_p): Return false if cfun->calls_eh_return. * lto-streamer-in.c (input_struct_function_base): Read calls_eh_return. * lto-streamer-out.c (output_struct_function_base): Write calls_eh_return. Modified: trunk/gcc/ChangeLog trunk/gcc/function.h trunk/gcc/gimplify.c trunk/gcc/lto-streamer-in.c trunk/gcc/lto-streamer-out.c trunk/gcc/tree-inline.c trunk/gcc/tree-tailcall.c
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #16 from Jakub Jelinek --- Author: jakub Date: Sat May 11 09:33:22 2019 New Revision: 271093 URL: https://gcc.gnu.org/viewcvs?rev=271093=gcc=rev Log: PR c++/59813 * config/aarch64/aarch64.c (aarch64_expand_epilogue): Don't add EH_RETURN_STACKADJ_RTX to sp in sibcall epilogues. Modified: trunk/gcc/ChangeLog trunk/gcc/config/aarch64/aarch64.c
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #15 from Jakub Jelinek --- Created attachment 46332 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46332=edit gcc10-pr59813-aarch64.patch Untested fix. The problem is that after adding sp addition back to the caller's sp in the sibcall epilogue the aarch64 epilogue appends add sp, sp, x4 instruction, which adds a value of a random register to stack pointer (in this case 1) and so the tail called function SIGBUSes. That addition must be done only in the eh epilogue (aarch64 seems to have only !for_sibcall and for_sibcall style epilogues, so one has to hope that a function with __builtin_eh_return doesn't have more than one normal epilogue, i.e. that they are all merged together).
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813
--- Comment #14 from Jakub Jelinek ---
The only difference the patch makes that matters for those tests is in
unwind-dw2.c, where in _Unwind_Resume_or_Rethrow function there is:
- _20 = _Unwind_RaiseException (exc_4(D));
+ _20 = _Unwind_RaiseException (exc_4(D)); [tail call]
change. This is in:
{
struct _Unwind_Context this_context, cur_context;
_Unwind_Reason_Code code;
unsigned long frames;
if (exc->private_1 == 0)
return _Unwind_RaiseException (exc);
...
}
where this_context and cur_context variables are indeed addressable variables,
but are only first touched in the ... code, so before my change we'd refuse to
make _Unwind_RaiseException call into a tail call, but now it is a tail call
and I don't see anything that would make it look wrong at the GIMPLE level.
If the aarch64 backend can't handle tail calls in functions doing
__builtin_eh_return or something similar, then it needs to punt in those
special cases.
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #13 from Jakub Jelinek --- Ah, seems it is libgcc_s.so.1 rather than libstdc++. Bisecting.
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #12 from Jakub Jelinek --- Are you sure about the bisection btw? I've just reverted those changes, rebuilt cc1plus and rebuilt libstdc++ with that, but get still the same failures.
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #11 from Jakub Jelinek --- Is something in libstdc++ miscompiled or something in the tests? Like, can you try those tests against libstdc++ built with that change reverted, but test with gcc with that revision in? If it is in libstdc++, can you bisect to a particular *.o file (built by r271012 vs. r271013)? Is it the tree-inline.[ch] change, or the tree-tailcall.c change? For both one needs the tree-ssa-live.[ch] changes, but otherwise the changes are independent (essentially, it could be committed as the tree-ssa-live.[ch] patch, and then either nothing, or tree-inline.[ch], or tree-tailcall.c, or both).
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #10 from Christophe Lyon --- And some regressions in g++ too: g++.dg/compat/eh/unexpected1 cp_compat_x_tst.o-cp_compat_y_tst.o execute g++.dg/cpp0x/lambda/lambda-eh2.C -std=gnu++14 execution test g++.dg/cpp0x/nullptr35.C -std=c++14 execution test g++.dg/cpp0x/nullptr35.C -std=c++17 execution test g++.dg/eh/registers1.C -std=gnu++14 execution test g++.dg/eh/registers1.C -std=gnu++17 execution test g++.dg/eh/registers1.C -std=gnu++98 execution test g++.dg/eh/uncaught1.C -std=gnu++14 execution test g++.dg/eh/uncaught1.C -std=gnu++17 execution test g++.dg/eh/uncaught1.C -std=gnu++98 execution test g++.dg/eh/uncaught4.C -std=gnu++14 execution test g++.dg/eh/uncaught4.C -std=gnu++17 execution test g++.dg/eh/uncaught4.C -std=gnu++98 execution test g++.dg/eh/unexpected1.C -std=c++14 execution test g++.dg/eh/unexpected1.C -std=c++98 execution test g++.old-deja/g++.abi/cxa_vec.C -std=gnu++14 execution test g++.old-deja/g++.abi/cxa_vec.C -std=gnu++17 execution test g++.old-deja/g++.abi/cxa_vec.C -std=gnu++98 execution test g++.old-deja/g++.eh/badalloc1.C -std=c++14 execution test g++.old-deja/g++.eh/badalloc1.C -std=c++17 execution test g++.old-deja/g++.eh/badalloc1.C -std=c++98 execution test g++.old-deja/g++.eh/fntry1.C -std=c++14 execution test g++.old-deja/g++.eh/fntry1.C -std=c++17 execution test g++.old-deja/g++.eh/fntry1.C -std=c++98 execution test g++.old-deja/g++.eh/rethrow1.C -std=c++14 execution test g++.old-deja/g++.eh/rethrow1.C -std=c++17 execution test g++.old-deja/g++.eh/rethrow1.C -std=c++98 execution test g++.old-deja/g++.eh/rethrow2.C -std=c++14 execution test g++.old-deja/g++.eh/rethrow2.C -std=c++17 execution test g++.old-deja/g++.eh/rethrow2.C -std=c++98 execution test g++.old-deja/g++.eh/rethrow3.C -std=c++14 execution test g++.old-deja/g++.eh/rethrow3.C -std=c++17 execution test g++.old-deja/g++.eh/rethrow3.C -std=c++98 execution test g++.old-deja/g++.eh/rethrow4.C -std=c++14 execution test g++.old-deja/g++.eh/rethrow4.C -std=c++17 execution test g++.old-deja/g++.eh/rethrow4.C -std=c++98 execution test g++.old-deja/g++.eh/rethrow5.C -std=c++14 execution test g++.old-deja/g++.eh/rethrow5.C -std=c++17 execution test g++.old-deja/g++.eh/rethrow5.C -std=c++98 execution test g++.old-deja/g++.eh/rethrow6.C -std=c++14 execution test g++.old-deja/g++.eh/rethrow6.C -std=c++17 execution test g++.old-deja/g++.eh/rethrow6.C -std=c++98 execution test g++.old-deja/g++.eh/spec2.C -std=c++14 execution test g++.old-deja/g++.eh/spec2.C -std=c++98 execution test g++.old-deja/g++.mike/eh23.C -std=gnu++14 execution test g++.old-deja/g++.mike/eh23.C -std=gnu++17 execution test g++.old-deja/g++.mike/eh23.C -std=gnu++98 execution test g++.old-deja/g++.mike/eh33.C -std=gnu++14 execution test g++.old-deja/g++.mike/eh33.C -std=gnu++98 execution test g++.old-deja/g++.mike/eh39.C -std=gnu++14 execution test g++.old-deja/g++.mike/eh39.C -std=gnu++17 execution test g++.old-deja/g++.mike/eh39.C -std=gnu++98 execution test g++.old-deja/g++.mike/eh40.C -std=gnu++14 execution test g++.old-deja/g++.mike/eh40.C -std=gnu++17 execution test g++.old-deja/g++.mike/eh40.C -std=gnu++98 execution test g++.old-deja/g++.mike/eh50.C -std=gnu++14 execution test g++.old-deja/g++.mike/eh50.C -std=gnu++98 execution test g++.old-deja/g++.mike/eh51.C -std=gnu++14 execution test g++.old-deja/g++.mike/eh51.C -std=gnu++98 execution test g++.old-deja/g++.robertl/eb31.C -std=c++14 execution test g++.old-deja/g++.robertl/eb31.C -std=c++17 execution test g++.old-deja/g++.robertl/eb31.C -std=c++98 execution test
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 Christophe Lyon changed: What|Removed |Added CC||clyon at gcc dot gnu.org --- Comment #9 from Christophe Lyon --- (In reply to Jakub Jelinek from comment #8) > Author: jakub > Date: Wed May 8 17:06:46 2019 > New Revision: 271013 > > URL: https://gcc.gnu.org/viewcvs?rev=271013=gcc=rev > Log: > PR c++/59813 > PR tree-optimization/89060 > * tree-ssa-live.h (live_vars_map): New typedef. > (compute_live_vars, live_vars_at_stmt, destroy_live_vars): Declare. > * tree-ssa-live.c: Include gimple-walk.h and cfganal.h. > (struct compute_live_vars_data): New type. > (compute_live_vars_visit, compute_live_vars_1, compute_live_vars, > live_vars_at_stmt, destroy_live_vars): New functions. > * tree-tailcall.c: Include tree-ssa-live.h. > (live_vars, live_vars_vec): New global variables. > (find_tail_calls): Perform variable life analysis before punting. > (tree_optimize_tail_calls_1): Clean up live_vars and live_vars_vec. > * tree-inline.h (struct copy_body_data): Add eh_landing_pad_dest > member. > * tree-inline.c (add_clobbers_to_eh_landing_pad): Remove BB argument. > Perform variable life analysis to select variables that really need > clobbers added. > (copy_edges_for_bb): Don't call add_clobbers_to_eh_landing_pad here, > instead set id->eh_landing_pad_dest and assert it is the same. > (copy_cfg_body): Call it here if id->eh_landing_pad_dest is non-NULL. > > * gcc.dg/tree-ssa/pr89060.c: New test. > > Added: > trunk/gcc/testsuite/gcc.dg/tree-ssa/pr89060.c > Modified: > trunk/gcc/ChangeLog > trunk/gcc/testsuite/ChangeLog > trunk/gcc/tree-inline.c > trunk/gcc/tree-inline.h > trunk/gcc/tree-ssa-live.c > trunk/gcc/tree-ssa-live.h > trunk/gcc/tree-tailcall.c This patch introduced regressions in libstdc++ on aarch64, I'm now seeing these failures: 18_support/exception_ptr/lifespan.cc execution test 18_support/nested_exception/throw_with_nested.cc execution test 18_support/uncaught_exception/14026.cc execution test 20_util/specialized_algorithms/memory_management_tools/1.cc execution test 20_util/unsynchronized_pool_resource/allocate.cc execution test 20_util/variant/exception_safety.cc execution test 20_util/variant/run.cc execution test 21_strings/basic_string/cons/char/69092.cc execution test 22_locale/locale/cons/12438.cc execution test 22_locale/numpunct/members/pod/2.cc execution test 23_containers/deque/cons/2.cc execution test 23_containers/deque/requirements/exception/basic.cc execution test 23_containers/deque/requirements/exception/propagation_consistent.cc execution test 23_containers/forward_list/requirements/exception/basic.cc execution test 23_containers/forward_list/requirements/exception/propagation_consistent.cc execution test 23_containers/list/operations/78389.cc execution test 23_containers/list/requirements/exception/basic.cc execution test 23_containers/list/requirements/exception/propagation_consistent.cc execution test 23_containers/map/requirements/exception/basic.cc execution test 23_containers/map/requirements/exception/propagation_consistent.cc execution test 23_containers/multimap/requirements/exception/basic.cc execution test 23_containers/multimap/requirements/exception/propagation_consistent.cc execution test 23_containers/multiset/requirements/exception/basic.cc execution test 23_containers/multiset/requirements/exception/propagation_consistent.cc execution test 23_containers/set/requirements/exception/basic.cc execution test 23_containers/set/requirements/exception/propagation_consistent.cc execution test 23_containers/unordered_map/requirements/exception/basic.cc execution test 23_containers/unordered_map/requirements/exception/propagation_consistent.cc execution test 23_containers/unordered_multimap/requirements/exception/basic.cc execution test 23_containers/unordered_multimap/requirements/exception/propagation_consistent.cc execution test 23_containers/unordered_multiset/insert/hash_policy.cc execution test 23_containers/unordered_multiset/requirements/exception/basic.cc execution test 23_containers/unordered_multiset/requirements/exception/propagation_consistent.cc execution test 23_containers/unordered_set/insert/hash_policy.cc execution test 23_containers/unordered_set/max_load_factor/robustness.cc execution test 23_containers/unordered_set/requirements/exception/basic.cc execution test 23_containers/unordered_set/requirements/exception/propagation_consistent.cc execution test 23_containers/vector/capacity/2.cc execution test
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 --- Comment #8 from Jakub Jelinek --- Author: jakub Date: Wed May 8 17:06:46 2019 New Revision: 271013 URL: https://gcc.gnu.org/viewcvs?rev=271013=gcc=rev Log: PR c++/59813 PR tree-optimization/89060 * tree-ssa-live.h (live_vars_map): New typedef. (compute_live_vars, live_vars_at_stmt, destroy_live_vars): Declare. * tree-ssa-live.c: Include gimple-walk.h and cfganal.h. (struct compute_live_vars_data): New type. (compute_live_vars_visit, compute_live_vars_1, compute_live_vars, live_vars_at_stmt, destroy_live_vars): New functions. * tree-tailcall.c: Include tree-ssa-live.h. (live_vars, live_vars_vec): New global variables. (find_tail_calls): Perform variable life analysis before punting. (tree_optimize_tail_calls_1): Clean up live_vars and live_vars_vec. * tree-inline.h (struct copy_body_data): Add eh_landing_pad_dest member. * tree-inline.c (add_clobbers_to_eh_landing_pad): Remove BB argument. Perform variable life analysis to select variables that really need clobbers added. (copy_edges_for_bb): Don't call add_clobbers_to_eh_landing_pad here, instead set id->eh_landing_pad_dest and assert it is the same. (copy_cfg_body): Call it here if id->eh_landing_pad_dest is non-NULL. * gcc.dg/tree-ssa/pr89060.c: New test. Added: trunk/gcc/testsuite/gcc.dg/tree-ssa/pr89060.c Modified: trunk/gcc/ChangeLog trunk/gcc/testsuite/ChangeLog trunk/gcc/tree-inline.c trunk/gcc/tree-inline.h trunk/gcc/tree-ssa-live.c trunk/gcc/tree-ssa-live.h trunk/gcc/tree-tailcall.c
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 Jakub Jelinek changed: What|Removed |Added CC||vanyacpp at gmail dot com --- Comment #7 from Jakub Jelinek --- *** Bug 77938 has been marked as a duplicate of this bug. ***
[Bug c++/59813] tail-call elimination didn't fire for left-shift of char to cout
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59813 Jakub Jelinek changed: What|Removed |Added CC||jakub at gcc dot gnu.org --- Comment #6 from Jakub Jelinek --- *** Bug 89060 has been marked as a duplicate of this bug. ***
