> diff --git a/gcc/ChangeLog b/gcc/ChangeLog
> index fca665b..3b62bcc 100644
> --- a/gcc/ChangeLog
> +++ b/gcc/ChangeLog
> @@ -1,3 +1,31 @@
> +2013-10-29 Martin Liska <[email protected]>
> + Jan Hubicka <[email protected]>
> +
> + * cgraph.c (dump_cgraph_node): Profile dump added.
> + * cgraph.h (struct cgraph_node): New time profile variable added.
> + * cgraphclones.c (cgraph_clone_node): Time profile is cloned.
> + * gcov-io.h (gcov_type): New profiler type introduced.
> + * ipa-profile.c (lto_output_node): Streaming for time profile added.
> + (input_node): Time profiler is read from LTO stream.
> + * predict.c (maybe_hot_count_p): Hot prediction changed.
> + * profile.c (instrument_values): New case for time profiler added.
> + (compute_value_histograms): Read of time profile.
> + * tree-pretty-print.c (dump_function_header): Time profiler is dumped.
> + * tree-profile.c (init_ic_make_global_vars): Time profiler function
> added.
> + (gimple_init_edge_profiler): TP function instrumentation.
> + (gimple_gen_time_profiler): New.
> + * value-prof.c (gimple_add_histogram_value): Support for time profiler
> + added.
> + (dump_histogram_value): TP type added to dumps.
> + (visit_hist): More sensitive check that takes TP into account.
> + (gimple_find_values_to_profile): TP instrumentation.
> + * value-prof.h (hist_type): New histogram type added.
> + (struct histogram_value_t): Pointer to struct function added.
> + * libgcc/Makefile.in: New GCOV merge function for TP added.
> + * libgcov.c: function_counter variable introduced.
> + (_gcov_merge_time_profile): New.
> + (_gcov_time_profiler): New.
> +
> 2013-10-29 David Malcolm <[email protected]>
>
> * doc/gty.texi ("Inheritance and GTY"): Make it clear that
> diff --git a/gcc/cgraph.c b/gcc/cgraph.c
> index 52d9ab0..c95a54e 100644
> --- a/gcc/cgraph.c
> +++ b/gcc/cgraph.c
> @@ -1890,6 +1890,7 @@ dump_cgraph_node (FILE *f, struct cgraph_node *node)
> if (node->profile_id)
> fprintf (f, " Profile id: %i\n",
> node->profile_id);
> + fprintf (f, " First run: %i\n", node->tp_first_run);
> fprintf (f, " Function flags:");
> if (node->count)
> fprintf (f, " executed "HOST_WIDEST_INT_PRINT_DEC"x",
> diff --git a/gcc/cgraph.h b/gcc/cgraph.h
> index 7706419..479d49f 100644
> --- a/gcc/cgraph.h
> +++ b/gcc/cgraph.h
> @@ -247,7 +247,6 @@ struct GTY(()) cgraph_clone_info
> bitmap combined_args_to_skip;
> };
>
> -
> /* The cgraph data structure.
> Each function decl has assigned cgraph_node listing callees and callers.
> */
>
> @@ -324,6 +323,8 @@ struct GTY(()) cgraph_node {
> unsigned tm_clone : 1;
> /* True if this decl is a dispatcher for function versions. */
> unsigned dispatcher_function : 1;
> + /* Time profiler: first run of function. */
> + int tp_first_run;
Move this up after profile_id.
> --- a/gcc/gcov-io.c
> +++ b/gcc/gcov-io.c
> @@ -68,7 +68,7 @@ gcov_open (const char *name, int mode)
> #if IN_LIBGCOV
> const int mode = 0;
> #endif
> -#if GCOV_LOCKED
> +#if GCOV_LOCKED
> struct flock s_flock;
> int fd;
>
Accidental change?
> @@ -651,6 +658,9 @@ lto_symtab_prevailing_decl (tree decl)
> if (TREE_CODE (decl) == FUNCTION_DECL && DECL_BUILT_IN (decl))
> return decl;
>
> + if (!DECL_ASSEMBLER_NAME_SET_P (decl))
> + return decl;
> +
> /* Ensure DECL_ASSEMBLER_NAME will not set assembler name. */
> gcc_assert (DECL_ASSEMBLER_NAME_SET_P (decl));
>
Remove this change - it is unrelated hack from my old tree.
> diff --git a/gcc/predict.c b/gcc/predict.c
> index cc9a053..4b655d3 100644
> --- a/gcc/predict.c
> +++ b/gcc/predict.c
> @@ -170,7 +170,7 @@ maybe_hot_count_p (struct function *fun, gcov_type count)
> if (fun && profile_status_for_function (fun) != PROFILE_READ)
> return true;
> /* Code executed at most once is not hot. */
> - if (profile_info->runs >= count)
> + if (count <= 1)
> return false;
> return (count >= get_hot_bb_threshold ());
> }
And also this change.
> @@ -895,9 +907,19 @@ compute_value_histograms (histogram_values values,
> unsigned cfg_checksum,
> hist->hvalue.counters = XNEWVEC (gcov_type, hist->n_counters);
> for (j = 0; j < hist->n_counters; j++)
> if (aact_count)
> - hist->hvalue.counters[j] = aact_count[j];
> - else
> - hist->hvalue.counters[j] = 0;
> + hist->hvalue.counters[j] = aact_count[j];
> + else
> + hist->hvalue.counters[j] = 0;
> +
> + if (hist->type == HIST_TYPE_TIME_PROFILE)
> + {
> + node = cgraph_get_node (hist->fun->decl);
> +
> + node->tp_first_run = hist->hvalue.counters[0];
> +
> + if (dump_file)
> + fprintf (dump_file, "Read tp_first_run: %d\n",
> node->tp_first_run);
> + }
Probably add a comment why you need to annotate counter here.
> diff --git a/gcc/tree-pretty-print.c b/gcc/tree-pretty-print.c
> index b2c5411..0fe262c 100644
> --- a/gcc/tree-pretty-print.c
> +++ b/gcc/tree-pretty-print.c
> @@ -3390,7 +3390,9 @@ dump_function_header (FILE *dump_file, tree fdecl, int
> flags)
> fprintf (dump_file, ", decl_uid=%d", DECL_UID (fdecl));
> if (node)
> {
> - fprintf (dump_file, ", symbol_order=%d)%s\n\n", node->symbol.order,
> + fprintf (dump_file, ", tp_first_run=%d, symbol_order=%d)%s\n\n",
> + node->tp_first_run,
> + node->symbol.order,
> node->frequency == NODE_FREQUENCY_HOT
> ? " (hot)"
> : node->frequency == NODE_FREQUENCY_UNLIKELY_EXECUTED
I would skip this change for now - we do not want to increase verbosity too
much. Pehraps this can
go with -details or so.
> static void
> init_ic_make_global_vars (void)
> {
> - tree gcov_type_ptr;
> + tree gcov_type_ptr;
>
> - ptr_void = build_pointer_type (void_type_node);
> + ptr_void = build_pointer_type (void_type_node);
Be cureful about accidental whitespace changes.
>
> /* Workaround for binutils bug 14342. Once it is fixed, remove lto path.
> */
> if (flag_lto)
> @@ -102,7 +104,7 @@ init_ic_make_global_vars (void)
>
> varpool_finalize_decl (ic_void_ptr_var);
>
> - gcov_type_ptr = build_pointer_type (get_gcov_type ());
> + gcov_type_ptr = build_pointer_type (get_gcov_type ());
> /* Workaround for binutils bug 14342. Once it is fixed, remove lto path.
> */
> if (flag_lto)
> {
Here too.
> @@ -455,6 +471,20 @@ gimple_gen_ic_func_profiler (void)
> gsi_insert_before (&gsi, stmt2, GSI_SAME_STMT);
> }
>
> +
> +void
> +gimple_gen_time_profiler (unsigned tag, unsigned base,
> + gimple_stmt_iterator &gsi)
Add block comment before function.
> @@ -545,7 +575,7 @@ tree_profiling (void)
>
> if (! flag_branch_probabilities
> && flag_profile_values)
> - gimple_gen_ic_func_profiler ();
> + gimple_gen_ic_func_profiler ();
>
> if (flag_branch_probabilities
> && flag_profile_values
Maybe also accidental?
> @@ -692,13 +702,16 @@ gcov_exit (void)
>
> if (gi_ptr->merge[t_ix])
> {
> - if (!cs_prg->runs++)
> + int first_run = !cs_prg->runs;
> +
> + cs_prg->runs++;
> + if (first_run)
> cs_prg->num = cs_tprg->num;
> cs_prg->sum_all += cs_tprg->sum_all;
> if (cs_prg->run_max < cs_tprg->run_max)
> cs_prg->run_max = cs_tprg->run_max;
> cs_prg->sum_max += cs_tprg->run_max;
> - if (cs_prg->runs == 1)
> + if (first_run)
> memcpy (cs_prg->histogram, cs_tprg->histogram,
> sizeof (gcov_bucket_type) * GCOV_HISTOGRAM_SIZE);
> else
This change is probably part of the patch getting fork working right and
unrelated to this patch, right?
I would skip it along with the whitespace changes so Rons' patch to break up
libgcov have fewer rejects.
> @@ -795,17 +808,23 @@ gcov_exit (void)
> gcov_write_unsigned (gfi_ptr->cfg_checksum);
>
> ci_ptr = gfi_ptr->ctrs;
> +
> for (t_ix = 0; t_ix < GCOV_COUNTERS; t_ix++)
> {
> - if (!gi_ptr->merge[t_ix])
> + gcov_merge_fn merge = gi_ptr->merge[t_ix];
> +
> + if (!merge)
> continue;
>
> n_counts = ci_ptr->num;
> gcov_write_tag_length (GCOV_TAG_FOR_COUNTER (t_ix),
> GCOV_TAG_COUNTER_LENGTH (n_counts));
> +
> gcov_type *c_ptr = ci_ptr->values;
> + gcov_type value;
Whitespace looks wrong.
> while (n_counts--)
> - gcov_write_counter (*c_ptr++);
> + gcov_write_counter (*c_ptr++);
> +
> ci_ptr++;
> }
> if (buffered)
> @@ -824,6 +843,8 @@ gcov_exit (void)
> "profiling:%s:Error writing\n",
> gi_filename);
> }
> +
> + gcov_clear ();
> }
>
> /* Reset all counters to zero. */
I believe this is also unrelated (part of fork/vfork changes)
> @@ -851,6 +872,7 @@ gcov_clear (void)
> continue;
>
> memset (ci_ptr->values, 0, sizeof (gcov_type) * ci_ptr->num);
> +
> ci_ptr++;
> }
> }
> @@ -912,7 +934,6 @@ __gcov_flush (void)
> __gthread_mutex_lock (&__gcov_flush_mx);
>
> gcov_exit ();
> - gcov_clear ();
>
> __gthread_mutex_unlock (&__gcov_flush_mx);
> }
Also unrelated.
> @@ -974,6 +995,24 @@ __gcov_merge_ior (gcov_type *counters, unsigned
> n_counters)
> }
> #endif
>
> +#ifdef L_gcov_merge_time_profile
> +void
> +__gcov_merge_time_profile (gcov_type *counters, unsigned n_counters)
> +{
> + unsigned int i;
> + gcov_type value;
> +
> + for (i = 0; i < n_counters; i++)
> + {
> + value = gcov_read_counter ();
> +
> + if (value && (!counters[i] || value < counters[i]))
> + counters[i] = value;
> + }
> +}
> +#endif /* L_gcov_merge_time_profile */
Add comments for this function explaining how merging goes and also for
L_gcov_time_profiler.
> +
> +
> #ifdef L_gcov_merge_single
> /* The profile merging function for choosing the most common value.
> It is given an array COUNTERS of N_COUNTERS old counters and it
> @@ -1171,6 +1210,7 @@ __thread
> #endif
> gcov_type * __gcov_indirect_call_counters;
>
> +
> /* By default, the C++ compiler will use function addresses in the
> vtable entries. Setting TARGET_VTABLE_USES_DESCRIPTORS to nonzero
> tells the compiler to use function descriptors instead. The value
> @@ -1202,6 +1242,16 @@ __gcov_indirect_call_profiler_v2 (gcov_type value,
> void* cur_func)
> }
> #endif
>
> +#ifdef L_gcov_time_profiler
> +
> +void
> +__gcov_time_profiler (gcov_type* counters)
> +{
> + if (!counters[0])
> + counters[0] = ++function_counter;
> +}
> +#endif
> +
> #ifdef L_gcov_average_profiler
> /* Increase corresponding COUNTER by VALUE. FIXME: Perhaps we want
> to saturate up. */
> @@ -1233,10 +1283,12 @@ __gcov_fork (void)
> {
> pid_t pid;
> extern __gthread_mutex_t __gcov_flush_mx;
> +
> __gcov_flush ();
> pid = fork ();
> if (pid == 0)
> __GTHREAD_MUTEX_INIT_FUNCTION (&__gcov_flush_mx);
> +
> return pid;
> }
> #endif
Please post updated patch. Basically it looks fine with these changes.
Try to also prepare a testcase (in testsuite/gcc.dg/tree-prof you can
see existing examples how to profile and grep dump files for first run
counters).
Honza
