[Bug ipa/67811] [TM] ICE with try-block in transaction

2015-12-23 Thread rth at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67811

Richard Henderson  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |FIXED
   Assignee|unassigned at gcc dot gnu.org  |rth at gcc dot gnu.org

--- Comment #4 from Richard Henderson  ---
Fixed.

[Bug ipa/67811] [TM] ICE with try-block in transaction

2015-12-23 Thread rth at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67811

--- Comment #5 from Richard Henderson  ---
Author: rth
Date: Thu Dec 24 00:45:15 2015
New Revision: 231943

URL: https://gcc.gnu.org/viewcvs?rev=231943=gcc=rev
Log:
PR ipa/67811

 * tree-cfg.c (make_edges_bb): Add abort edge for outer transactions.

Modified:
trunk/gcc/ChangeLog
trunk/gcc/tree-cfg.c

[Bug ipa/67811] [TM] ICE with try-block in transaction

2015-12-22 Thread rth at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67811

--- Comment #3 from Richard Henderson  ---
Author: rth
Date: Tue Dec 22 19:42:24 2015
New Revision: 231907

URL: https://gcc.gnu.org/viewcvs?rev=231907=gcc=rev
Log:
PR ipa/67811

* gimple.h (struct gtransaction): Add label_norm, label_uninst;
replace label with label_over.
(gimple_build_transaction): Remove label parameter.
(gimple_transaction_label_norm): New.
(gimple_transaction_label_uninst): New.
(gimple_transaction_label_over): Rename from gimple_transaction_label.
(gimple_transaction_label_norm_ptr): New.
(gimple_transaction_label_uninst_ptr): New.
(gimple_transaction_label_over_ptr): Rename from
gimple_transaction_label_ptr.
(gimple_transaction_set_label_norm): New.
(gimple_transaction_set_label_uninst): New.
(gimple_transaction_set_label_over): Rename from
gimple_transaction_set_label.
* gimple-pretty-print.c (dump_gimple_transaction): Update.
* gimple-streamer-in.c (input_gimple_stmt) [GIMPLE_TRANSACTION]: Same.
* gimple-streamer-out.c (output_gimple_stmt) [GIMPLE_TRANSACTION]:
Same.
* gimple-walk.c (walk_gimple_op) [GIMPLE_TRANSACTION]: Same.
* tree-cfg.c (make_edges_bb) [GIMPLE_TRANSACTION]: Same.
(cleanup_dead_labels) [GIMPLE_TRANSACTION]: Same.
(verify_gimple_transaction): Same.
(gimple_redirect_edge_and_branch) [GIMPLE_TRANSACTION]: Same.
* tree-inline.c (remap_gimple_stmt) [GIMPLE_TRANSACTION]: Same.
* gimple.c (gimple_build_transaction): Remove label parameter;
initialize all three label memebers.
* gimplify.c (gimplify_transaction): Update call
to gimple_build_transaction.
* trans-mem.c (make_tm_uninst): New.
(lower_transaction): Create uninstrumented code path here...
(ipa_tm_scan_calls_transaction): ... not here.
(ipa_uninstrument_transaction): Remove.

testsuite/
* g++.dg/tm/noexcept-1.C: Update expected must_not_throw count.
* g++.dg/tm/noexcept-4.C: Likewise.
* g++.dg/tm/noexcept-5.C: Likewise.
* g++.dg/tm/pr67811.C: New.

Added:
trunk/gcc/testsuite/g++.dg/tm/pr67811.C
Modified:
trunk/gcc/ChangeLog
trunk/gcc/gimple-pretty-print.c
trunk/gcc/gimple-streamer-in.c
trunk/gcc/gimple-streamer-out.c
trunk/gcc/gimple-walk.c
trunk/gcc/gimple.c
trunk/gcc/gimple.h
trunk/gcc/gimplify.c
trunk/gcc/testsuite/ChangeLog
trunk/gcc/testsuite/g++.dg/tm/noexcept-1.C
trunk/gcc/testsuite/g++.dg/tm/noexcept-4.C
trunk/gcc/testsuite/g++.dg/tm/noexcept-5.C
trunk/gcc/trans-mem.c
trunk/gcc/tree-cfg.c
trunk/gcc/tree-inline.c

[Bug ipa/67811] [TM] ICE with try-block in transaction

2015-11-09 Thread rth at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67811

Richard Henderson  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2015-11-09
 Ever confirmed|0   |1

--- Comment #1 from Richard Henderson  ---
There's no change from .tmipa til .inline, so it's simply a
matter of not having diagnosed the problem earlier.

That said, the output looks right at a glance, so I'm not
sure what verify_flow_info is complaining about...

[Bug ipa/67811] [TM] ICE with try-block in transaction

2015-11-09 Thread rth at gcc dot gnu.org
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67811

--- Comment #2 from Richard Henderson  ---
copy_bbs fails to duplicate the eh regions included within
the requested region.  This no doubt causes all sorts of 
bad follow-on effects, heretofore unobserved.