ChangeLog-gcc-4.1.txt 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189
  1. Changelog of the TIGCC-local changes to GCC (4.1 branch):
  2. 2006-04-30 Kevin Kofler <Kevin@tigcc.ticalc.org>
  3. * Import GCC 4.1 20060428 snapshot.
  4. * version.c (VERSUFFIX): Bump to "4.1.1-pre6".
  5. * gcc/c.opt (-Wpointer-sign): Don't change the initializer (now set to -1 upstream).
  6. * c-opts.c (c_common_handle_option): Don't enable -Wpointer-sign for -Wall.
  7. 2006-01-05 Kevin Kofler <Kevin@tigcc.ticalc.org>
  8. * version.c (VERSUFFIX): Bump to "4.1.0-pre5".
  9. * config/m68k/m68k.md (movsi): Don't allow writes of a label directly to memory under -mpcrel.
  10. (anonymous pattern for pcrel movsi): Split into case for address register and case for push.
  11. Only accept stack pushes, not arbitrary predecrements.
  12. 2005-12-24 Kevin Kofler <Kevin@tigcc.ticalc.org>
  13. * version.c (VERSUFFIX): Bump to "4.1.0-pre4".
  14. * toplev.c (process_options): Don't warn about -fomit-frame-pointer with unwind tables and
  15. don't force -fno-omit-frame-pointer.
  16. 2005-12-22 Kevin Kofler <Kevin@tigcc.ticalc.org>
  17. * version.c (VERSUFFIX): Bump to "4.1.0-pre3".
  18. * c-parser.c (c_parser_cast_expression): Allow all postfix operators, not just ++ and --.
  19. 2005-12-22 Kevin Kofler <Kevin@tigcc.ticalc.org>
  20. * version.c (VERSUFFIX): Bump to "4.1.0-pre2".
  21. * c-parser.c (c_parser_cast_expression): Rename to c_parser_cast_expression_1, don't handle
  22. Objective C here.
  23. (c_parser_cast_expression): New function. Handle Objective C and postfix operators.
  24. (c_parser_unary_expression): Add in_cast argument. Pass it to c_parser_postfix_expression.
  25. (c_parser_cast_expression_1, c_parser_sizeof_expression, c_parser_alignof_expression): Update
  26. calls to c_parser_unary_expression.
  27. (c_parser_postfix_expression): Add in_cast argument. Pass it to
  28. c_parser_postfix_expression_after_primary.
  29. (c_parser_postfix_expression_after_primary): Add in_cast argument. Only accept postincrement/
  30. postdecrement if it is false.
  31. (c_parser_cast_expression, c_parser_postfix_expression_after_paren_type): Update calls to
  32. postfix_expression_after_primary.
  33. 2005-12-22 Kevin Kofler <Kevin@tigcc.ticalc.org>
  34. * version.c (VERSUFFIX): Add "4.1.0-pre1" tag.
  35. 2005-12-18 - 2005-12-21 Kevin Kofler <Kevin@tigcc.ticalc.org>
  36. * Import GCC 4.1 20051216 snapshot.
  37. * Merge changes from GCC 4.0.2 (see ChangeLog-gcc-4.0.txt).
  38. * Drop obsolete backports:
  39. DWARF 2 patches/backports from Fedora GCC:
  40. - gcc4-dwarf2-pr20268.patch
  41. 2005-03-30 James E. Wilson <wilson@specifixinc.com>
  42. PR debug/20268
  43. * dwarf2out.c (add_high_low_attributes): New function, extracted from
  44. gen_lexical_block_die.
  45. (gen_lexical_block_die, gen_inlined_subroutine_die): Call it.
  46. - gcc4-dwarf2-inline-details.patch
  47. 2005-07-25 James E Wilson <wilson@specifixinc.com>
  48. * dwarf2out.c (add_call_src_coords_attributes): New.
  49. (gen_inlined_subroutine_die): Call it.
  50. (maybe_emit_file, init_file_table): Add comments.
  51. (prune_unused_types_walk_attribs): Pass DW_AT_call_file through
  52. maybe_emit_file.
  53. * tree-inline.c (remap_block): Copy BLOCK_SOURCE_LOCATION.
  54. (expand_call_inline): Set BLOCK_SOURCE_LOCATION.
  55. * tree.h (BLOCK_SOURCE_LOCATION): New.
  56. (struct tree_block): New field locus.
  57. - gcc4-dwarf2-usefbreg.patch
  58. 2005-08-10 Jakub Jelinek <jakub@redhat.com>
  59. * dwarf2out.c (concat_loc_descriptor): Add can_use_fbreg argument,
  60. pass it down to loc_descriptor.
  61. (loc_descriptor): Pass can_use_fbreg to concat_loc_descriptor.
  62. (containing_function_has_frame_base): Move earlier in the file.
  63. (loc_descriptor_from_tree_1): Use containing_function_has_frame_base
  64. instead of always assuming fbreg can't be used.
  65. - gcc4-dwarf2-prefer-1elt-vartracking.patch
  66. 2005-08-10 Jakub Jelinek <jakub@redhat.com>
  67. * dwarf2out.c (add_location_or_const_value_attribute): Prefer locations
  68. gathered by var-tracking in single entry loc_list over loc_descriptor_from_tree.
  69. - gcc4-dwarf2-frame_base.patch
  70. 2005-08-22 Richard Henderson <rth@redhat.com>
  71. * function.c (ARG_POINTER_CFA_OFFSET): Move ...
  72. * defaults.h (ARG_POINTER_CFA_OFFSET): ... here.
  73. (INCOMING_FRAME_SP_OFFSET): Moved from ...
  74. * dwarf2out.c (INCOMING_FRAME_SP_OFFSET): ... here.
  75. (struct cfa_loc): Change reg to unsigned int,
  76. rearrange for better packing.
  77. (lookup_cfa_1): Remove inline marker.
  78. (cfa_equal_p): Split out of ...
  79. (def_cfa_1): ... here. Use INVALID_REGNUM.
  80. (build_cfa_loc): Handle !cfa->indirect.
  81. (frame_pointer_cfa_offset): New.
  82. (dbx_reg_number): Assert register elimination performed; do
  83. leaf register remapping.
  84. (reg_loc_descriptor): Avoid calling dbx_reg_number when unused.
  85. (eliminate_reg_to_offset): New.
  86. (based_loc_descr): Remove can_use_fbreg argument. Use fbreg only
  87. for verifiably local stack frame addresses; re-base to CFA.
  88. (mem_loc_descriptor): Remove can_use_fbreg argument.
  89. (concat_loc_descriptor, loc_descriptor): Likewise.
  90. (containing_function_has_frame_base): Remove.
  91. (rtl_for_decl_location): Don't do register elimination or
  92. leaf register remapping here.
  93. (secname_for_decl): Split out from ..
  94. (add_location_or_const_value_attribute): ... here.
  95. (convert_cfa_to_loc_list): New.
  96. (compute_frame_pointer_to_cfa_displacement): New.
  97. (gen_subprogram_die): Use them.
  98. * tree.h (frame_base_decl): Remove.
  99. * var-tracking.c (frame_base_decl, frame_stack_adjust): Remove.
  100. (prologue_stack_adjust): Remove.
  101. (vt_stack_adjustments): Use INCOMING_FRAME_SP_OFFSET.
  102. (adjust_stack_reference): Re-base memories to arg_pointer_rtx.
  103. (set_frame_base_location): Remove.
  104. (compute_bb_dataflow, emit_notes_in_bb): Don't call it.
  105. (dump_attrs_list, dump_dataflow_set): Use string concatenation.
  106. (vt_add_function_parameters): Don't eliminate_regs.
  107. (vt_initialize): Don't create frame_base_decl.
  108. 2005-09-21 Andreas Schwab <schwab@suse.de>
  109. PR rtl-optimization/23043
  110. * postreload-gcse.c (eliminate_partially_redundant_load): Fix typo
  111. when allocating a struct unoccr.
  112. 2005-09-29 Jakub Jelinek <jakub@redhat.com>
  113. * c-decl.c (c_write_global_declarations_1): If any
  114. wrapup_global_declaration_2 call returned true, restart the loop.
  115. * Drop no longer needed reversion of (part of larger patch):
  116. 2004-01-17 Joseph S. Myers <jsm@polyomino.org.uk>
  117. c-typeck.c (build_conditional_expr): Call non_lvalue instead of pedantic_non_lvalue.
  118. (The check has moved to fold-const.c and does call pedantic_non_lvalue now.)
  119. * Drop no longer needed patches:
  120. * collect2.c: Applied patch to allow collect2 to build on MinGW by Lev Serebryakov
  121. (2002-10-25).
  122. * expr.c (emit_push_insn): Don't take a modulus by 0. Set offset to constant 0 instead.
  123. 2004-07-25 Kevin Kofler <Kevin@tigcc.ticalc.org>
  124. * varasm.c (enum section_category): Move declaration up.
  125. (categorize_decl_for_section): Move prototype up.
  126. 2005-05-05 Kevin Kofler <Kevin@tigcc.ticalc.org>
  127. * dwarf2out.c (dwarf2out_start_source_file): Use the correct file number.
  128. * version.c: Don't hand-maintain version numbers, use (TIGCC) suffix instead.
  129. * Moved AMS-floating-point-related patch from config/m68k/m68k.c to config/m68k/predicates.md:
  130. (const_uint32_operand): Don't handle CONST_DOUBLE.
  131. Check for "INTVAL (op) >= 0" (possibly wrong?).
  132. * config/m68k/m68k.c (extended_address_operand): Move to config/m68k/predicates.md.
  133. * config/m68k/m68k.h: Don't add extended_address_operand to PREDICATE_CODES anymore.
  134. Don't define MASK_* and TARGET_* (except MASK_REG_RELATIVE,
  135. TARGET_REG_RELATIVE) anymore.
  136. (TARGET_DIRECTFLOAT): Define to always 0 (not supported yet).
  137. (TARGET_SWITCHES, TARGET_OPTIONS): Move options to...
  138. * config/m68k/m68k.opt: ... here. Remove "-mnoalign-int", "-mnopcrel", "-mnotios", "-mnoregparm",
  139. "-mnomerge-sections", "-mnomerge-to-data", "-mnorodata-to-text" (use
  140. "-mno-*" with the dash instead). Enable negative for "-mcoff-abslines".
  141. * Revert broken part of PR debug/24070 fix:
  142. 2005-10-06 Richard Henderson <rth@redhat.com>
  143. PR debug/24070
  144. * dwarf2out.c (lookup_filename): Return the result of maybe_emit_file.
  145. * builtins.c: Also disable expand_builtin_int_roundingfn, fold_builtin_int_roundingfn and
  146. stack-smashing protection builtins.
  147. (fold_builtins_1): Disable stack-smashing protection builtins.
  148. * builtins.def: Likewise.
  149. * Drop changes to c-parse.in (replaced by c-parser.c).
  150. * c-parser.c (c_parser_parameter_declaration): Handle an optional asmspec before the attributes
  151. to support register parameters. Pass it to
  152. build_c_parm (or NULL if there is none).
  153. (c_parser_compound_statement_nostart): Don't error about labels at end of compound statements,
  154. only pedwarn if pedantic.
  155. * config/m68k/m68k.c (output_move_double, floating_exact_log2): Use gcc_unreachable instead of
  156. abort.
  157. * config/smapbcd.h (REAL_VALUE_TO_TARGET_SINGLE, REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
  158. (UNSIGNED_INF, POSITIVE_INF, NEGATIVE_INF, NAN, REAL_VALUE_LDEXP, REAL_VALUE_UNSIGNED_FIX,
  159. REAL_VALUE_RNDZINT, REAL_VALUE_TO_TARGET_LONG_DOUBLE, REAL_VALUE_TO_TARGET_SMAP_BCD):
  160. Add ull suffix to big numbers to get rid of warnings.
  161. * c-decl.c (c_insert_default_attributes): Use ISO C90 prototype.
  162. * c-typeck.c (build_modify_expr): Update parameters for default_function_array_conversion call.
  163. * expr.c (emit_block_move_via_libcall, clear_storage_via_libcall): Fix use of uninitialized
  164. variable.
  165. * real.c (real_value_dtof): Add ull suffix to big number to get rid of warning.
  166. (real_value_htof): Delete unused variable "exp2".
  167. (real_from_integer): Delete unused variable "i".
  168. * simplify-rtx.c (simplify_const_binary_operation): Declare tmp1 pointer const.
  169. (simplify_immed_subreg): Declare tmp pointer const.
  170. * tree-object-size.c (alloc_object_size): Don't handle BUILT_IN_CALLOC.
  171. (pass_through_call): Don't handle BUILT_IN_*_CHK.
  172. * optabs.c (lowpart_subreg_maybe_copy, expand_copysign_absneg): Disable.
  173. * config/m68k/m68k.c (m68k_handle_stkparm_attribute, m68k_handle_regparm_attribute): Update
  174. calls to warning().
  175. (m68k_save_reg, m68k_output_function_prologue): Fix pointer signedness mismatch.