Tweak Interface_field_reference_expression::do_get_backend to apply an
additional backend type conversion to the returned result. This is
needed due to the fact that the top level type of the expression is a
function descriptor, however the value being manufactured is a pointer
to <thunk-descriptor, value> struct.
Reviewed-on: https://go-review.googlesource.com/51712
From-SVN: r250686
We recently started walking through the hash table of pointer types to
finalize them. Unfortunately it is possible to create a new pointer
type while finalizing an existing one (test case: test/fixedbugs/issue5291)
and that breaks the iteration. So, instead, keep a list of
placeholder pointer types, and iterate through them while permitting
the list to be extended as we go.
Reviewed-on: https://go-review.googlesource.com/51771
From-SVN: r250683
For calls that return multiple results we used to create a temporary
of struct type to hold the results, and also create a separate
temporary for each result. Then the call expression would copy each
result out of the struct to the temporary, and Call_result_expression
would refer to the desired temporary.
Simplify this to just use a single temporary of struct type, and
change Call_result_expression to fetch a field of the struct.
This may reduce some incorrect tree sharing in the backend code.
Reviewed-on: https://go-review.googlesource.com/51770
From-SVN: r250682
* tree-predcom.c: (struct chain): Handle store-store chain in which
stores for elimination only store loop invariant values.
(execute_pred_commoning_chain): Ditto.
(prepare_initializers_chain_store_elim): Ditto.
(prepare_finalizers): Ditto.
(is_inv_store_elimination_chain): New function.
(initialize_root_vars_store_elim_1): New function.
From-SVN: r250670
* tree-predcom.c: Revise general description of the pass.
(enum chain_type): New enum type for store elimination.
(struct chain): New field supporting store elimination.
(struct component): Ditto.
(dump_chain): Dump store-stores chain.
(release_chain): Release resources.
(split_data_refs_to_components): Compute and create component
contains only stores for elimination.
(get_chain_last_ref_at): New function.
(make_invariant_chain): Initialization.
(make_rooted_chain): Specify chain type in parameter and record it.
(add_looparound_copies): Skip for store-stores chain.
(determine_roots_comp): Compute type of chain and pass it to
make_rooted_chain.
(initialize_root_vars_store_elim_2): New function.
(finalize_eliminated_stores): New function.
(remove_stmt): Handle store for elimination.
(execute_pred_commoning_chain): Execute predictive commoning on
store-store chains.
(determine_unroll_factor): Skip unroll for store-stores chain.
(prepare_initializers_chain_store_elim): New function.
(prepare_initializers_chain): Hanlde store-store chain.
(prepare_finalizers_chain, prepare_finalizers): New function.
(tree_predictive_commoning_loop): Return integer value indicating
if loop is unrolled or lcssa form is corrupted.
(tree_predictive_commoning): Rewrite for lcssa form if necessary.
gcc/testsuite
* gcc.dg/tree-ssa/predcom-dse-1.c: New test.
* gcc.dg/tree-ssa/predcom-dse-2.c: New test.
* gcc.dg/tree-ssa/predcom-dse-3.c: New test.
* gcc.dg/tree-ssa/predcom-dse-4.c: New test.
* gcc.dg/tree-ssa/predcom-dse-5.c: New test.
* gcc.dg/tree-ssa/predcom-dse-6.c: New test.
* gcc.dg/tree-ssa/predcom-dse-7.c: New test.
* gcc.dg/tree-ssa/predcom-dse-8.c: New test.
* gcc.dg/tree-ssa/predcom-dse-9.c: New test.
* gcc.dg/tree-ssa/predcom-dse-10.c: New test.
* gcc.dg/tree-ssa/predcom-dse-11.c: New test.
From-SVN: r250669
* tree-predcom.c (initialize_root): Delete.
(execute_pred_commoning_chain): Initialize root vars and replace
reference of non-combined chain directly, rather than call above
function.
From-SVN: r250668
* tree-predcom.c (struct chain): New field init_seq.
(release_chain): Release init_seq.
(prepare_initializers_chain): Record intialization stmts in above
field.
(insert_init_seqs): New function.
(tree_predictive_commoning_loop): Call insert_init_seqs.
From-SVN: r250666
2017-07-28 Richard Biener <rguenther@suse.de>
* match.pd: Remove superfluous :c.
* genmatch.c (simplify::id): Add member.
(lower_commutative, lower_opt_convert, lower_cond, lower_for):
Copy id.
(current_id): New global.
(dt_node::parent): Move from ...
(dt_operand::parent): ... here. Add for_id member.
(is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
(decision_tree::find_node): Relax order requirement when
merging DT_TRUE nodes to ones inbetween the current simplify
and the one we try to merge with. Add diagnostic whenever
we need to enforce pattern order by not merging.
(decision_tree::insert): Set current_id.
(decision_tree::print_node): Dump parent node and for_id.
(parser::last_id): Add member.
(parser::push_simplify): Assign unique id.
(parser::parser): Initialize last_id.
From-SVN: r250664
2017-07-28 Martin Liska <mliska@suse.cz>
PR sanitizer/81340
* sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
gimple_build_debug_bind.
2017-07-28 Martin Liska <mliska@suse.cz>
PR sanitizer/81340
* g++.dg/asan/pr81340.C: New test.
From-SVN: r250660
2017-07-28 Martin Liska <mliska@suse.cz>
PR sanitizer/81460
* sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
parameters that are of a variable-length.
2017-07-28 Martin Liska <mliska@suse.cz>
PR sanitizer/81460
* gcc.dg/asan/pr81460.c: New test.
From-SVN: r250655
PR tree-optimization/81578
* tree-parloops.c (build_new_reduction): Bail out if
reduction_code isn't one of the standard OpenMP reductions.
Move the details printing after that decision.
* gcc.dg/pr81578.c: New test.
From-SVN: r250651
PR c/45784
* c-omp.c (c_finish_omp_for): If the condition is wrapped in
rhs of COMPOUND_EXPR(s), skip them and readd their lhs into
new COMPOUND_EXPRs around the rhs of the comparison.
* testsuite/libgomp.c/pr45784.c: New test.
* testsuite/libgomp.c++/pr45784.C: New test.
From-SVN: r250635
During combine GCC tries to merge CMP (with zero) and AND into a TST. However,
in cases where an ANDS operand is not compatible, this was being missed. Adding
a define_split where this operand was moved to a register seems to help out.
Committed on behalf of Sudi Das
---
gcc/
2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
Sudakshina Das <sudi.das@arm.com>
* config/aarch64/aarch64.md
(define_split for and<mode>3nr_compare): Move
non aarch64_logical_operand to a register.
(define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
register immediate operand to a register.
* config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
gcc/testsuite
2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
Sudakshina Das <sudi.das@arm.com>
* gcc.target/aarch64/tst_imm_split_1.c: New Test.
Co-Authored-By: Sudakshina Das <sudi.das@arm.com>
From-SVN: r250631
2017-07-27 Richard Biener <rguenther@suse.de>
PR tree-optimization/81502
* tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
with incompatible but same sized type.
(execute_update_addresses_taken): Likewise.
* gcc.target/i386/vect-insert-1.c: New testcase.
From-SVN: r250620
While answering a user question on the equivalence of
-ftree-loop-vectorize + -ftree-slp-vectorize and -ftree-vectorize I
spotted one case which broke the equivalence. pass_ch::process_loop_p
was guarded on flag_tree_vectorize, meaning you would get it for
-ftree-vectorize, but not for -ftree-loop-vectorize/-ftree-slp-vectorize.
This patch fixes that, getting rid of the only use of flag_tree_vectorize
in the code base.
gcc/
* tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
flag_tree_loop_vectorize rather than flag_tree_vectorize.
From-SVN: r250619
The HI/QI atomic_fetch_<atomic><mode>" expander accepted symbolic
references and emitted CAS patterns whose insn predicates rejected them.
Fixed by allowing symbolic references there as well. Reload will get
rid of them due to the constraint letter.
Regression tested on s390x.
gcc/ChangeLog:
2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
PR target/81534
* config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
Change s_operand to memory_operand.
gcc/testsuite/ChangeLog:
2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
PR target/81534
* gcc.target/s390/pr81534.c: New test.
From-SVN: r250617
The little-endian VSX code uses rotates to swap the two 64-bit halves of
128-bit scalar modes. This is fine for TImode and V1TImode, but it
isn't really valid to use RTL rotates on floating-point modes like
KFmode and TFmode, and doing that triggered an assert added by the
SVE series. This patch uses bit-casts to V1TImode instead.
2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
gcc/
* config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
* config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
(rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
Emit instructions rather than returning an expression. Handle TFmode
and KFmode by casting to TImode.
(rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
(rs6000_emit_le_vsx_store): Likewise.
* config/rs6000/vsx.md (VSX_TI): New iterator.
(*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
(*vsx_le_undo_permute_<mode>): Likewise.
(*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
emit the split sequence.
(*vsx_le_perm_store_<mode>): Likewise.
From-SVN: r250615
Trevor Saunders
GCC Administrator
Peter Bergner
Ian Lance Taylor
Tony Reix
Tamar Christina
Bin Cheng
Rainer Orth
Richard Biener
Martin Liska
Jakub Jelinek
Sebastian Huber
Kyrylo Tkachov
Marek Polacek
Eric Botcazou
James Greenhalgh
Andreas Krebbel
Richard Sandiford