gcc_assert (!(flags & (ECF_LOOPING_CONST_OR_PURE
| ECF_MAY_BE_ALLOCA
| ECF_SIBCALL
+ | ECF_LEAF
| ECF_NOVOPS)));
}
lto_output_bitpack (&bp);
lto_output_fn_decl_index (ob->decl_state, ob->main_stream, node->decl);
lto_output_sleb128_stream (ob->main_stream, node->count);
+ lto_output_sleb128_stream (ob->main_stream, node->count_materialization_scale);
if (tag == LTO_cgraph_analyzed_node)
{
{
if (profile_info)
{
- /* We do not output num, it is not terribly useful. */
+ /* We do not output num, sum_all and run_max, they are not used by
+ GCC profile feedback and they are difficult to merge from multiple
+ units. */
gcc_assert (profile_info->runs);
lto_output_uleb128_stream (ob->main_stream, profile_info->runs);
- lto_output_sleb128_stream (ob->main_stream, profile_info->sum_all);
- lto_output_sleb128_stream (ob->main_stream, profile_info->run_max);
- lto_output_sleb128_stream (ob->main_stream, profile_info->sum_max);
+ lto_output_uleb128_stream (ob->main_stream, profile_info->sum_max);
}
else
lto_output_uleb128_stream (ob->main_stream, 0);
node = cgraph_node (fn_decl);
node->count = lto_input_sleb128 (ib);
+ node->count_materialization_scale = lto_input_sleb128 (ib);
if (tag == LTO_cgraph_analyzed_node)
{
/* Input profile_info from IB. */
static void
-input_profile_summary (struct lto_input_block *ib)
+input_profile_summary (struct lto_input_block *ib,
+ struct lto_file_decl_data *file_data)
{
unsigned int runs = lto_input_uleb128 (ib);
if (runs)
{
- if (!profile_info)
- {
- profile_info = <o_gcov_summary;
- lto_gcov_summary.runs = runs;
- lto_gcov_summary.sum_all = lto_input_sleb128 (ib);
- lto_gcov_summary.run_max = lto_input_sleb128 (ib);
- lto_gcov_summary.sum_max = lto_input_sleb128 (ib);
- }
- /* We can support this by scaling all counts to nearest common multiple
- of all different runs, but it is perhaps not worth the effort. */
- else if (profile_info->runs != runs
- || profile_info->sum_all != lto_input_sleb128 (ib)
- || profile_info->run_max != lto_input_sleb128 (ib)
- || profile_info->sum_max != lto_input_sleb128 (ib))
- sorry ("combining units with different profiles is not supported");
- /* We allow some units to have profile and other to not have one. This will
- just make unprofiled units to be size optimized that is sane. */
+ file_data->profile_info.runs = runs;
+ file_data->profile_info.sum_max = lto_input_uleb128 (ib);
+ if (runs > file_data->profile_info.sum_max)
+ fatal_error ("Corrupted profile info in %s: sum_max is smaller than runs",
+ file_data->file_name);
}
}
+/* Rescale profile summaries to the same number of runs in the whole unit. */
+
+static void
+merge_profile_summaries (struct lto_file_decl_data **file_data_vec)
+{
+ struct lto_file_decl_data *file_data;
+ unsigned int j;
+ gcov_unsigned_t max_runs = 0;
+ struct cgraph_node *node;
+ struct cgraph_edge *edge;
+
+ /* Find unit with maximal number of runs. If we ever get serious about
+ roundoff errors, we might also consider computing smallest common
+ multiply. */
+ for (j = 0; (file_data = file_data_vec[j]) != NULL; j++)
+ if (max_runs < file_data->profile_info.runs)
+ max_runs = file_data->profile_info.runs;
+
+ if (!max_runs)
+ return;
+
+ /* Simple overflow check. We probably don't need to support that many train
+ runs. Such a large value probably imply data corruption anyway. */
+ if (max_runs > INT_MAX / REG_BR_PROB_BASE)
+ {
+ sorry ("At most %i profile runs is supported. Perhaps corrupted profile?",
+ INT_MAX / REG_BR_PROB_BASE);
+ return;
+ }
+
+ profile_info = <o_gcov_summary;
+ lto_gcov_summary.runs = max_runs;
+ lto_gcov_summary.sum_max = 0;
+
+ /* Rescale all units to the maximal number of runs.
+ sum_max can not be easily merged, as we have no idea what files come from
+ the same run. We do not use the info anyway, so leave it 0. */
+ for (j = 0; (file_data = file_data_vec[j]) != NULL; j++)
+ if (file_data->profile_info.runs)
+ {
+ int scale = ((REG_BR_PROB_BASE * max_runs
+ + file_data->profile_info.runs / 2)
+ / file_data->profile_info.runs);
+ lto_gcov_summary.sum_max = MAX (lto_gcov_summary.sum_max,
+ (file_data->profile_info.sum_max
+ * scale
+ + REG_BR_PROB_BASE / 2)
+ / REG_BR_PROB_BASE);
+ }
+
+ /* Watch roundoff errors. */
+ if (lto_gcov_summary.sum_max < max_runs)
+ lto_gcov_summary.sum_max = max_runs;
+
+ /* If merging already happent at WPA time, we are done. */
+ if (flag_ltrans)
+ return;
+
+ /* Now compute count_materialization_scale of each node.
+ During LTRANS we already have values of count_materialization_scale
+ computed, so just update them. */
+ for (node = cgraph_nodes; node; node = node->next)
+ if (node->local.lto_file_data->profile_info.run_max)
+ {
+ int scale;
+ if (node->local.lto_file_data->profile_info.runs)
+ scale =
+ ((node->count_materialization_scale * max_runs
+ + node->local.lto_file_data->profile_info.run_max / 2)
+ / node->local.lto_file_data->profile_info.run_max);
+ else
+ scale = node->count_materialization_scale;
+ node->count_materialization_scale = scale;
+ if (scale < 0)
+ fatal_error ("Profile information in %s corrupted",
+ file_data->file_name);
+
+ if (scale == REG_BR_PROB_BASE)
+ continue;
+ for (edge = node->callees; edge; edge = edge->next_callee)
+ edge->count = ((edge->count * scale + REG_BR_PROB_BASE / 2)
+ / REG_BR_PROB_BASE);
+ node->count = ((node->count * scale + REG_BR_PROB_BASE / 2)
+ / REG_BR_PROB_BASE);
+ }
+}
+
/* Input and merge the cgraph from each of the .o files passed to
lto1. */
&data, &len);
if (!ib)
fatal_error ("cannot find LTO cgraph in %s", file_data->file_name);
- input_profile_summary (ib);
+ input_profile_summary (ib, file_data);
file_data->cgraph_node_encoder = lto_cgraph_encoder_new ();
nodes = input_cgraph_1 (file_data, ib);
lto_destroy_simple_input_block (file_data, LTO_section_cgraph,
VEC_free (cgraph_node_ptr, heap, nodes);
VEC_free (varpool_node_ptr, heap, varpool);
}
+ merge_profile_summaries (file_data_vec);
+
/* Clear out the aux field that was used to store enough state to
tell which nodes should be overwritten. */
/* Read the CFG for function FN from input block IB. */
static void
-input_cfg (struct lto_input_block *ib, struct function *fn)
+input_cfg (struct lto_input_block *ib, struct function *fn,
+ int count_materialization_scale)
{
unsigned int bb_count;
basic_block p_bb;
if (bb == NULL)
bb = make_new_block (fn, index);
- edge_count = lto_input_uleb128 (ib);
+ edge_count = (lto_input_uleb128 (ib) * count_materialization_scale
+ + REG_BR_PROB_BASE / 2) / REG_BR_PROB_BASE;
/* Connect up the CFG. */
for (i = 0; i < edge_count; i++)
static void
input_bb (struct lto_input_block *ib, enum LTO_tags tag,
- struct data_in *data_in, struct function *fn)
+ struct data_in *data_in, struct function *fn,
+ int count_materialization_scale)
{
unsigned int index;
basic_block bb;
index = lto_input_uleb128 (ib);
bb = BASIC_BLOCK_FOR_FUNCTION (fn, index);
- bb->count = lto_input_sleb128 (ib);
+ bb->count = (lto_input_sleb128 (ib) * count_materialization_scale
+ + REG_BR_PROB_BASE / 2) / REG_BR_PROB_BASE;
bb->loop_depth = lto_input_sleb128 (ib);
bb->frequency = lto_input_sleb128 (ib);
bb->flags = lto_input_sleb128 (ib);
DECL_INITIAL (fn_decl) = lto_input_tree (ib, data_in);
gcc_assert (DECL_INITIAL (fn_decl));
DECL_SAVED_TREE (fn_decl) = NULL_TREE;
+ node = cgraph_node (fn_decl);
/* Read all the basic blocks. */
tag = input_record_start (ib);
while (tag)
{
- input_bb (ib, tag, data_in, fn);
+ input_bb (ib, tag, data_in, fn,
+ node->count_materialization_scale);
tag = input_record_start (ib);
}
gimple_set_body (fn_decl, bb_seq (ei_edge (ei)->dest));
}
- node = cgraph_node (fn_decl);
fixup_call_stmt_edges (node, stmts);
execute_all_ipa_stmt_fixups (node, stmts);
{
struct function *fn = DECL_STRUCT_FUNCTION (fn_decl);
struct lto_in_decl_state *decl_state;
+ struct cgraph_node *node = cgraph_node (fn_decl);
push_cfun (fn);
init_tree_ssa (fn);
gcc_assert (decl_state);
file_data->current_decl_state = decl_state;
- input_cfg (&ib_cfg, fn);
+ input_cfg (&ib_cfg, fn, node->count_materialization_scale);
/* Set up the struct function. */
input_function (fn_decl, data_in, &ib_main);
OSDN Git Service
