gcc/ChangeLog

   1 2013-01-03  Richard Henderson  <rth@redhat.com>
   2
   3         * config/i386/i386.c (ix86_expand_move): Always assign to op1
   4         after eliminating TLS symbols.
   5
   6 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
   7
   8         PR bootstrap/50167
   9         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
  10         * graphite-poly.c (debug_gmp_value): Likewise.
  11
  12 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
  13
  14         PR target/55712
  15         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
  16         selected code model, define __code_mode_small__, __code_model_medium__,
  17         __code_model_large__, __code_model_32__ or __code_model_kernel__.
  18         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
  19         xchg temporary register with %k.  Declare temporary register as
  20         early clobbered.
  21         [__x86_64__]: For medium and large code models, preserve %rbx register.
  22
  23 2013-01-03  Richard Biener  <rguenther@suse.de>
  24
  25         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
  26         (dump_subscript): Adjust.
  27         (finalize_ddr_dependent): Do not dump redundant info.
  28         (analyze_siv_subscript): Adjust.
  29         (subscript_dependence_tester): Likewise.
  30         (compute_affine_dependence): Likewise.
  31
  32 2013-01-03  Richard Biener  <rguenther@suse.de>
  33
  34         Revert
  35         2013-01-03  Richard Biener  <rguenther@suse.de>
  36
  37         PR tree-optimization/55857
  38         * tree-vect-stmts.c (vectorizable_load): Do not setup
  39         re-alignment for invariant loads.
  40
  41         2013-01-02  Richard Biener  <rguenther@suse.de>
  42
  43         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
  44         invariant load do not generate a vector load from the scalar location.
  45
  46 2013-01-03  Richard Biener  <rguenther@suse.de>
  47
  48         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
  49         for not vectorizing.
  50         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
  51         not build INDIRECT_REFs, call get_name once only.
  52         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
  53         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
  54
  55 2013-01-03  Richard Biener  <rguenther@suse.de>
  56
  57         PR tree-optimization/55857
  58         * tree-vect-stmts.c (vectorizable_load): Do not setup
  59         re-alignment for invariant loads.
  60
  61 2013-01-03  Richard Biener  <rguenther@suse.de>
  62
  63         PR lto/55848
  64         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
  65         prefer a built-in decl.
  66
  67 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
  68
  69         * gcc.c (process_command): Update copyright notice dates.
  70         * gcov.c (print_version): Likewise.
  71         * gcov-dump.c (print_version): Likewise.
  72
  73         PR rtl-optimization/55838
  74         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
  75         iv0.step, iv1.step and step.
  76
  77 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
  78             Marc Glisse  <marc.glisse@inria.fr>
  79
  80         PR tree-optimization/55832
  81         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
  82         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
  83         integer_{one,zero}_node.
  84
  85 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
  86
  87         PR debug/54402
  88         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
  89         * var-tracking.c (reverse_op): Don't add reverse ops to
  90         VALUEs that have already
  91         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
  92
  93 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
  94
  95         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
  96
  97 2013-01-02  Teresa Johnson  <tejohnson@google.com>
  98
  99         * dumpfile.c (dump_loc): Print filename with location.
 100         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
 101         new location_t parameter to emit complete unroll message with
 102         new dump framework.
 103         (canonicalize_loop_induction_variables): Compute loops location
 104         and pass to try_unroll_loop_completely.
 105         * loop-unroll.c (report_unroll_peel): New function.
 106         (peel_loops_completely): Use new dump format with location
 107         for main dumpfile message, and invoke report_unroll_peel on success.
 108         (decide_unrolling_and_peeling): Ditto.
 109         (decide_peel_once_rolling): Remove old dumpfile message subsumed
 110         by report_unroll_peel.
 111         (decide_peel_completely): Ditto.
 112         (decide_unroll_constant_iterations): Ditto.
 113         (decide_unroll_runtime_iterations): Ditto.
 114         (decide_peel_simple): Ditto.
 115         (decide_unroll_stupid): Ditto.
 116         * cfgloop.c (get_loop_location): New function.
 117         * cfgloop.h (get_loop_location): Declare.
 118
 119 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
 120
 121         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
 122         NULL.
 123
 124 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
 125
 126         PR middle-end/55198
 127         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
 128         BLKmode objects when EXPAND_MEMORY is specified.
 129
 130 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
 131
 132         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
 133         in loop predicate.
 134         (fold_builtin_cpu): Do not share cpu model decls across statements.
 135
 136 2013-01-02  Jason Merrill  <jason@redhat.com>
 137
 138         PR c++/55804
 139         * tree.c (build_array_type_1): Revert earlier change.
 140
 141 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
 142
 143         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
 144         "cortex-a57".
 145         * config/aarch64/aarch64-tune.md: Re-generate.
 146
 147 2013-01-02  Richard Biener  <rguenther@suse.de>
 148
 149         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
 150         invariant load do not generate a vector load from the scalar location.
 151
 152 2013-01-02  Richard Biener  <rguenther@suse.de>
 153
 154         PR bootstrap/55784
 155         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
 156         * configure: Regenerate.
 157
 158 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
 159
 160         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
 161         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
 162         (expand_builtin_int_roundingfn_2): Keep the original target around
 163         for the fallback case.
 164
 165 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
 166
 167         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
 168         to be clear for sign changes.
 169
 170 2013-01-01  Jan Hubicka  <jh@suse.cz>
 171
 172         * ipa-inline-analysis.c: Fix formatting.
 173
 174 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
 175
 176         PR tree-optimization/55831
 177         * tree-vect-loop.c (get_initial_def_for_induction): Use
 178         gsi_after_labels instead of gsi_start_bb.
 179 \f
 180 Copyright (C) 2013 Free Software Foundation, Inc.
 181
 182 Copying and distribution of this file, with or without modification,
 183 are permitted in any medium without royalty provided the copyright
 184 notice and this notice are preserved.