Re: [bitmap memory]: Allocation part 2

[Jeffrey A Law <law at redhat dot com>]

On Fri, 2004-11-26 at 10:27 +0000, Nathan Sidwell wrote:
> Jeffrey A Law wrote:
> > On Wed, 2004-11-24 at 10:54 +0000, Nathan Sidwell wrote:
> > 
> >>Jeffrey A Law wrote:
> >>
> >>>That patch is, err, long.  Is there any way you can break it into
> >>>distinct functional hunks to make review easier?
> >>
> >>This 3rd part removes the heap allocation scheme by forwarding
> >>to the obstack allocation scheme.  I'll post the obvious sed-generated
> >>patch if/when all these three parts are approved.
> 
> This is the final part of the patch series.  It removes BITMAP_XMALLOC
> and BITMAP_XFREE, which are now only aliases for BITMAP_ALLOC (NULL)
> and BITMAP_FREE.
> 
> I've booted and tested earlier incarnations of this patch
> (on i686-pc-linux-gnu), but last night's run of this part only dies with
> unrelated libjava problems. I was able to run the c testsuite with no
> problems.
> 
> ok?
> 
> nathan
> 
> plain text document attachment (bitmap-memory-5a.log)
> 2004-11-26  Nathan Sidwell  <nathan@codesourcery.com>
> 
> 	* bitmap.h (BITMAP_XMALLOC, BITMAP_XFREE): Remove.
> 	* bt-load.c (btr_def_live_range, combine_btr_defs,
> 	migrate_btr_def, migrate_btr_defs): Adjust.
> 	* cfgcleanup.c (thread_jump): Adjust.
> 	* cfgloop.c (get_loop_body_in_bfs_order):
> 	* df.c (df_insn_table_realloc, df_bitmaps_alloc, df_bitmaps_free,
> 	df_alloc, df_free, df_du_chain_create, df_bb_rd_local_compute,
> 	df_rd_local_compute, df_reg_info_compute): Adjust.
> 	* dominance.c (init_dom_info, free_dom_info): Adjust.
> 	* flow.c (init_propagate_block_info, free_propagate_block_info):
> 	* gcse.c (alloc_gcse_mem, free_gcse_mem): Adjust.
> 	* global.c (allocate_bb_info, free_bb_info, modify_reg_pav):
> 	* loop-invariant.c (find_defs, find_invariant_insn,
> 	find_invariants, free_inv_motion_data): Adjust.
> 	* predict.c (tree_predict_by_opcode, estimate_bb_frequencies):
> 	* ra-build.c (get_sub_conflicts, union_web_part_roots,
> 	livethrough_conflicts_bb, init_bb_info, free_bb_info,
> 	init_one_web, reset_conflicts, conflicts_between_webs,
> 	detect_spill_temps, ra_build_free, ra_build_free_all): Adjust.
> 	* ra-colorize.c (reset_lists, merge_moves, conservative): Adjust.
> 	* ra-rewrite.c (rewrite_program, rewrite_program2,
> 	detect_web_parts_to_rebuild, actual_spill, emit_colors,
> 	remove_suspicious_death_notes): Adjust.
> 	* ra.c (alloc_mem, free_all_mem, check_df): Adjust.
> 	* stmt.c (expand_case): Adjust.
> 	* tree-cfg.c (tree_duplicate_sese_region): Adjust.
> 	* tree-dfa.c (mark_new_vars_to_rename): Adjust.
> 	* tree-if-conv.c (get_loop_body_in_if_conv_order): Adjust.
> 	* tree-into-ssa.c (insert_phi_nodes_for, def_blocks_free,
> 	get_def_blocks_for, rewrite_into_ssa, rewrite_ssa_into_ssa): Adjust.
> 	* tree-optimize.c (tree_rest_of_compilation): Adjust.
> 	* tree-outof-ssa.c (new_temp_expr_table, free_temp_expr_table,
> 	analyze_edges_for_bb, perform_edge_inserts): Adjust.
> 	* tree-scalar-evolution.c (scev_initialize, scev_finalize): Adjust.
> 	* tree-sra.c (tree_sra): Adjust.
> 	* tree-ssa-alias.c (init_alias_info, delete_alias_info): Adjust.
> 	* tree-ssa-ccp.c (ccp_fold_builtin): Adjust.
> 	* tree-ssa-dce.c (tree_dce_init, tree_dce_done): Adjust.
> 	* tree-ssa-dom.c (tree_ssa_dominator_optimize): Adjust.
> 	* tree-ssa-dse.c (tree_ssa_dse): Adjust.
> 	* tree-ssa-forwprop.c (tree_ssa_forward_propagate_single_use_var):
> 	Adjust.
> 	* tree-ssa-live.c (new_tree_live_info, delete_tree_live_info,
> 	calculate_live_on_entry, calculate_live_on_exit,
> 	build_tree_conflict_graph): Adjust.
> 	* tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_init, record_use,
> 	record_important_candidates, set_use_iv_cost, find_depends,
> 	determine_use_iv_costs, iv_ca_new, iv_ca_free, free_loop_data,
> 	tree_ssa_iv_optimize_finalize): Adjust.
> 	* tree-ssa-loop-manip.c (add_exit_phis_var, get_loop_exits,
> 	add_exit_phis, find_uses_to_rename_use,
> 	rewrite_into_loop_closed_ssa,
> 	tree_duplicate_loop_to_header_edge): Adjust.
> 	* tree-ssa-pre.c (init_pre, fini_pre): Adjust.
> 	* tree-ssa.c (verify_flow_insensitive_alias_info, verify_ssa,
> 	init_tree_ssa, delete_tree_ssa): Adjust.
> 	* tree-ssanames.c (marked_ssa_names, init_ssanames,
> 	fini_ssanames): Adjust.
> 	* tree-vectorizer.c (slpeel_tree_peel_loop_to_edge): Adjust.
I think this got delayed when we went into regression fix mode only.
Now is probably a reasonable time to resubmit and get this last bit of
the bitmap memory allocation cleanup finished.

Can you please resubmit?

Sorry for the delay,
jeff