/* Basic IPA optimizations and utilities.
- Copyright (C) 2003, 2004, 2005, 2007, 2008, 2009, 2010
+ Copyright (C) 2003, 2004, 2005, 2007, 2008, 2009, 2010, 2011
Free Software Foundation, Inc.
This file is part of GCC.
#include "gimple.h"
#include "ggc.h"
#include "flags.h"
-
-/* Fill array order with all nodes with output flag set in the reverse
- topological order. */
-
-int
-cgraph_postorder (struct cgraph_node **order)
-{
- struct cgraph_node *node, *node2;
- int stack_size = 0;
- int order_pos = 0;
- struct cgraph_edge *edge, last;
- int pass;
-
- struct cgraph_node **stack =
- XCNEWVEC (struct cgraph_node *, cgraph_n_nodes);
-
- /* We have to deal with cycles nicely, so use a depth first traversal
- output algorithm. Ignore the fact that some functions won't need
- to be output and put them into order as well, so we get dependencies
- right through inline functions. */
- for (node = cgraph_nodes; node; node = node->next)
- node->aux = NULL;
- for (pass = 0; pass < 2; pass++)
- for (node = cgraph_nodes; node; node = node->next)
- if (!node->aux
- && (pass
- || (!cgraph_only_called_directly_p (node)
- && !node->address_taken)))
- {
- node2 = node;
- if (!node->callers)
- node->aux = &last;
- else
- node->aux = node->callers;
- while (node2)
- {
- while (node2->aux != &last)
- {
- edge = (struct cgraph_edge *) node2->aux;
- if (edge->next_caller)
- node2->aux = edge->next_caller;
- else
- node2->aux = &last;
- /* Break possible cycles involving always-inline
- functions by ignoring edges from always-inline
- functions to non-always-inline functions. */
- if (edge->caller->local.disregard_inline_limits
- && !edge->callee->local.disregard_inline_limits)
- continue;
- if (!edge->caller->aux)
- {
- if (!edge->caller->callers)
- edge->caller->aux = &last;
- else
- edge->caller->aux = edge->caller->callers;
- stack[stack_size++] = node2;
- node2 = edge->caller;
- break;
- }
- }
- if (node2->aux == &last)
- {
- order[order_pos++] = node2;
- if (stack_size)
- node2 = stack[--stack_size];
- else
- node2 = NULL;
- }
- }
- }
- free (stack);
- for (node = cgraph_nodes; node; node = node->next)
- node->aux = NULL;
- return order_pos;
-}
+#include "pointer-set.h"
+#include "target.h"
+#include "tree-iterator.h"
+#include "ipa-utils.h"
/* Look for all functions inlined to NODE and update their inlined_to pointers
to INLINED_TO. */
{
struct cgraph_node *node = ipa_ref_node (ref);
if (!node->reachable
+ && node->analyzed
&& (!DECL_EXTERNAL (node->decl)
|| before_inlining_p))
- {
- node->reachable = true;
- enqueue_cgraph_node (node, first);
- }
+ node->reachable = true;
+ enqueue_cgraph_node (node, first);
}
else
{
}
}
-/* Return true when function NODE can be removed from callgraph
- if all direct calls are eliminated. */
-static inline bool
-varpool_can_remove_if_no_refs (struct varpool_node *node)
+/* Return true when NODE can not be local. Worker for cgraph_local_node_p. */
+
+static bool
+cgraph_non_local_node_p_1 (struct cgraph_node *node, void *data ATTRIBUTE_UNUSED)
{
- return (!node->force_output && !node->used_from_other_partition
- && (DECL_COMDAT (node->decl) || !node->externally_visible));
+ /* FIXME: Aliases can be local, but i386 gets thunks wrong then. */
+ return !(cgraph_only_called_directly_or_aliased_p (node)
+ && !ipa_ref_has_aliases_p (&node->ref_list)
+ && node->analyzed
+ && !DECL_EXTERNAL (node->decl)
+ && !node->local.externally_visible
+ && !node->reachable_from_other_partition
+ && !node->in_other_partition);
}
/* Return true when function can be marked local. */
static bool
cgraph_local_node_p (struct cgraph_node *node)
{
- return (cgraph_only_called_directly_p (node)
- && node->analyzed
- && !DECL_EXTERNAL (node->decl)
- && !node->local.externally_visible
- && !node->reachable_from_other_partition
- && !node->in_other_partition);
+ struct cgraph_node *n = cgraph_function_or_thunk_node (node, NULL);
+
+ /* FIXME: thunks can be considered local, but we need prevent i386
+ from attempting to change calling convention of them. */
+ if (n->thunk.thunk_p)
+ return false;
+ return !cgraph_for_node_and_aliases (n,
+ cgraph_non_local_node_p_1, NULL, true);
+
+}
+
+/* Return true when NODE has ADDR reference. */
+
+static bool
+has_addr_references_p (struct cgraph_node *node,
+ void *data ATTRIBUTE_UNUSED)
+{
+ int i;
+ struct ipa_ref *ref;
+
+ for (i = 0; ipa_ref_list_refering_iterate (&node->ref_list, i, ref); i++)
+ if (ref->use == IPA_REF_ADDR)
+ return true;
+ return false;
}
/* Perform reachability analysis and reclaim all unreachable nodes.
gcc_assert (!vnode->aux);
#endif
varpool_reset_queue ();
+ /* Mark functions whose bodies are obviously needed.
+ This is mostly when they can be referenced externally. Inline clones
+ are special since their declarations are shared with master clone and thus
+ cgraph_can_remove_if_no_direct_calls_and_refs_p should not be called on them. */
for (node = cgraph_nodes; node; node = node->next)
- if (!cgraph_can_remove_if_no_direct_calls_and_refs_p (node)
- && ((!DECL_EXTERNAL (node->decl))
- || before_inlining_p))
+ if (node->analyzed && !node->global.inlined_to
+ && (!cgraph_can_remove_if_no_direct_calls_and_refs_p (node)
+ /* Keep around virtual functions for possible devirtualization. */
+ || (before_inlining_p && DECL_VIRTUAL_P (node->decl))))
{
gcc_assert (!node->global.inlined_to);
enqueue_cgraph_node (node, &first);
gcc_assert (!node->aux);
node->reachable = false;
}
+
+ /* Mark variables that are obviously needed. */
for (vnode = varpool_nodes; vnode; vnode = vnode->next)
{
vnode->next_needed = NULL;
vnode->prev_needed = NULL;
- if (!varpool_can_remove_if_no_refs (vnode))
+ if ((vnode->analyzed || vnode->force_output)
+ && !varpool_can_remove_if_no_refs (vnode))
{
vnode->needed = false;
varpool_mark_needed_node (vnode);
if (node->reachable)
{
for (e = node->callees; e; e = e->next_callee)
- if (!e->callee->reachable
- && node->analyzed
- && (!e->inline_failed || !e->callee->analyzed
- || (!DECL_EXTERNAL (e->callee->decl))
- || before_inlining_p))
- {
+ {
+ if (!e->callee->reachable
+ && node->analyzed
+ && (!e->inline_failed
+ || !DECL_EXTERNAL (e->callee->decl)
+ || before_inlining_p))
e->callee->reachable = true;
- enqueue_cgraph_node (e->callee, &first);
- }
+ enqueue_cgraph_node (e->callee, &first);
+ }
process_references (&node->ref_list, &first, &first_varpool, before_inlining_p);
}
cgraph_node_remove_callees (node);
ipa_remove_all_references (&node->ref_list);
node->analyzed = false;
- node->local.inlinable = false;
}
if (!node->aux)
{
+ struct cgraph_edge *e;
+ bool found = false;
+ int i;
+ struct ipa_ref *ref;
+
node->global.inlined_to = NULL;
if (file)
fprintf (file, " %s", cgraph_node_name (node));
- if (!node->analyzed || !DECL_EXTERNAL (node->decl) || before_inlining_p)
- cgraph_remove_node (node);
- else
+ /* See if there is reachable caller. */
+ for (e = node->callers; e && !found; e = e->next_caller)
+ if (e->caller->reachable)
+ found = true;
+ for (i = 0; (ipa_ref_list_refering_iterate (&node->ref_list, i, ref)
+ && !found); i++)
+ if (ref->refering_type == IPA_REF_CGRAPH
+ && ipa_ref_refering_node (ref)->reachable)
+ found = true;
+ else if (ref->refering_type == IPA_REF_VARPOOL
+ && ipa_ref_refering_varpool_node (ref)->needed)
+ found = true;
+
+ /* If so, we need to keep node in the callgraph. */
+ if (found)
{
- struct cgraph_edge *e;
-
- /* See if there is reachable caller. */
- for (e = node->callers; e; e = e->next_caller)
- if (e->caller->reachable)
- break;
-
- /* If so, we need to keep node in the callgraph. */
- if (e || node->needed)
+ if (node->analyzed)
{
struct cgraph_node *clone;
if (!clone)
{
cgraph_release_function_body (node);
- node->analyzed = false;
- node->local.inlinable = false;
+ if (node->prev_sibling_clone)
+ node->prev_sibling_clone->next_sibling_clone = node->next_sibling_clone;
+ else if (node->clone_of)
+ node->clone_of->clones = node->next_sibling_clone;
+ if (node->next_sibling_clone)
+ node->next_sibling_clone->prev_sibling_clone = node->prev_sibling_clone;
+ if (node->clone_of)
+ node->former_clone_of = node->clone_of->decl;
+ node->clone_of = NULL;
+ node->next_sibling_clone = NULL;
+ node->prev_sibling_clone = NULL;
}
else
gcc_assert (!clone->in_other_partition);
+ node->analyzed = false;
+ changed = true;
cgraph_node_remove_callees (node);
ipa_remove_all_references (&node->ref_list);
- if (node->prev_sibling_clone)
- node->prev_sibling_clone->next_sibling_clone = node->next_sibling_clone;
- else if (node->clone_of)
- node->clone_of->clones = node->next_sibling_clone;
- if (node->next_sibling_clone)
- node->next_sibling_clone->prev_sibling_clone = node->prev_sibling_clone;
- node->clone_of = NULL;
- node->next_sibling_clone = NULL;
- node->prev_sibling_clone = NULL;
}
- else
- cgraph_remove_node (node);
}
- changed = true;
+ else
+ {
+ cgraph_remove_node (node);
+ changed = true;
+ }
}
}
for (node = cgraph_nodes; node; node = node->next)
if (node->address_taken
&& !node->reachable_from_other_partition)
{
- int i;
- struct ipa_ref *ref;
- bool found = false;
- for (i = 0; ipa_ref_list_refering_iterate (&node->ref_list, i, ref)
- && !found; i++)
- {
- gcc_assert (ref->use == IPA_REF_ADDR);
- found = true;
- }
- if (!found)
+ if (!cgraph_for_node_and_aliases (node, has_addr_references_p, NULL, true))
{
if (file)
fprintf (file, " %s", cgraph_node_name (node));
}
}
}
+ if (file)
+ fprintf (file, "\n");
#ifdef ENABLE_CHECKING
verify_cgraph ();
FIXME: This can not be done in between gimplify and omp_expand since
readonly flag plays role on what is shared and what is not. Currently we do
- this transformation as part of ipa-reference pass, but it would make sense
- to do it before early optimizations. */
+ this transformation as part of whole program visibility and re-do at
+ ipa-reference pass (to take into account clonning), but it would
+ make sense to do it before early optimizations. */
void
ipa_discover_readonly_nonaddressable_vars (void)
fprintf (dump_file, "\n");
}
+/* Return true when there is a reference to node and it is not vtable. */
+static bool
+cgraph_address_taken_from_non_vtable_p (struct cgraph_node *node)
+{
+ int i;
+ struct ipa_ref *ref;
+ for (i = 0; ipa_ref_list_refering_iterate (&node->ref_list, i, ref); i++)
+ if (ref->use == IPA_REF_ADDR)
+ {
+ struct varpool_node *node;
+ if (ref->refering_type == IPA_REF_CGRAPH)
+ return true;
+ node = ipa_ref_refering_varpool_node (ref);
+ if (!DECL_VIRTUAL_P (node->decl))
+ return true;
+ }
+ return false;
+}
+
+/* COMDAT functions must be shared only if they have address taken,
+ otherwise we can produce our own private implementation with
+ -fwhole-program.
+ Return true when turning COMDAT functoin static can not lead to wrong
+ code when the resulting object links with a library defining same COMDAT.
+
+ Virtual functions do have their addresses taken from the vtables,
+ but in C++ there is no way to compare their addresses for equality. */
+
+bool
+cgraph_comdat_can_be_unshared_p (struct cgraph_node *node)
+{
+ if ((cgraph_address_taken_from_non_vtable_p (node)
+ && !DECL_VIRTUAL_P (node->decl))
+ || !node->analyzed)
+ return false;
+ if (node->same_comdat_group)
+ {
+ struct cgraph_node *next;
+
+ /* If more than one function is in the same COMDAT group, it must
+ be shared even if just one function in the comdat group has
+ address taken. */
+ for (next = node->same_comdat_group;
+ next != node; next = next->same_comdat_group)
+ if (cgraph_address_taken_from_non_vtable_p (next)
+ && !DECL_VIRTUAL_P (next->decl))
+ return false;
+ }
+ return true;
+}
+
/* Return true when function NODE should be considered externally visible. */
static bool
-cgraph_externally_visible_p (struct cgraph_node *node, bool whole_program)
+cgraph_externally_visible_p (struct cgraph_node *node,
+ bool whole_program, bool aliased)
{
if (!node->local.finalized)
return false;
if (!DECL_COMDAT (node->decl)
&& (!TREE_PUBLIC (node->decl) || DECL_EXTERNAL (node->decl)))
return false;
- if (!whole_program)
+
+ /* Do not even try to be smart about aliased nodes. Until we properly
+ represent everything by same body alias, these are just evil. */
+ if (aliased)
+ return true;
+
+ /* Do not try to localize built-in functions yet. One of problems is that we
+ end up mangling their asm for WHOPR that makes it impossible to call them
+ using the implicit built-in declarations anymore. Similarly this enables
+ us to remove them as unreachable before actual calls may appear during
+ expansion or folding. */
+ if (DECL_BUILT_IN (node->decl))
+ return true;
+
+ /* If linker counts on us, we must preserve the function. */
+ if (cgraph_used_from_object_file_p (node))
return true;
if (DECL_PRESERVE_P (node->decl))
return true;
- /* COMDAT functions must be shared only if they have address taken,
- otherwise we can produce our own private implementation with
- -fwhole-program. */
- if (DECL_COMDAT (node->decl))
- {
- if (node->address_taken || !node->analyzed)
- return true;
- if (node->same_comdat_group)
- {
- struct cgraph_node *next;
-
- /* If more than one function is in the same COMDAT group, it must
- be shared even if just one function in the comdat group has
- address taken. */
- for (next = node->same_comdat_group;
- next != node;
- next = next->same_comdat_group)
- if (next->address_taken || !next->analyzed)
- return true;
- }
- }
+ if (lookup_attribute ("externally_visible", DECL_ATTRIBUTES (node->decl)))
+ return true;
+ if (TARGET_DLLIMPORT_DECL_ATTRIBUTES
+ && lookup_attribute ("dllexport", DECL_ATTRIBUTES (node->decl)))
+ return true;
+ if (node->resolution == LDPR_PREVAILING_DEF_IRONLY)
+ return false;
+ /* When doing LTO or whole program, we can bring COMDAT functoins static.
+ This improves code quality and we know we will duplicate them at most twice
+ (in the case that we are not using plugin and link with object file
+ implementing same COMDAT) */
+ if ((in_lto_p || whole_program)
+ && DECL_COMDAT (node->decl)
+ && cgraph_comdat_can_be_unshared_p (node))
+ return false;
+
+ /* When doing link time optimizations, hidden symbols become local. */
+ if (in_lto_p
+ && (DECL_VISIBILITY (node->decl) == VISIBILITY_HIDDEN
+ || DECL_VISIBILITY (node->decl) == VISIBILITY_INTERNAL)
+ /* Be sure that node is defined in IR file, not in other object
+ file. In that case we don't set used_from_other_object_file. */
+ && node->analyzed)
+ ;
+ else if (!whole_program)
+ return true;
+
if (MAIN_NAME_P (DECL_NAME (node->decl)))
return true;
- if (lookup_attribute ("externally_visible", DECL_ATTRIBUTES (node->decl)))
+
+ return false;
+}
+
+/* Return true when variable VNODE should be considered externally visible. */
+
+bool
+varpool_externally_visible_p (struct varpool_node *vnode, bool aliased)
+{
+ if (!DECL_COMDAT (vnode->decl) && !TREE_PUBLIC (vnode->decl))
+ return false;
+
+ /* Do not even try to be smart about aliased nodes. Until we properly
+ represent everything by same body alias, these are just evil. */
+ if (aliased)
+ return true;
+
+ /* If linker counts on us, we must preserve the function. */
+ if (varpool_used_from_object_file_p (vnode))
+ return true;
+
+ if (DECL_HARD_REGISTER (vnode->decl))
+ return true;
+ if (DECL_PRESERVE_P (vnode->decl))
+ return true;
+ if (lookup_attribute ("externally_visible",
+ DECL_ATTRIBUTES (vnode->decl)))
+ return true;
+ if (TARGET_DLLIMPORT_DECL_ATTRIBUTES
+ && lookup_attribute ("dllexport",
+ DECL_ATTRIBUTES (vnode->decl)))
+ return true;
+
+ /* See if we have linker information about symbol not being used or
+ if we need to make guess based on the declaration.
+
+ Even if the linker clams the symbol is unused, never bring internal
+ symbols that are declared by user as used or externally visible.
+ This is needed for i.e. references from asm statements. */
+ if (varpool_used_from_object_file_p (vnode))
+ return true;
+ if (vnode->resolution == LDPR_PREVAILING_DEF_IRONLY)
+ return false;
+
+ /* As a special case, the COMDAT virutal tables can be unshared.
+ In LTO mode turn vtables into static variables. The variable is readonly,
+ so this does not enable more optimization, but referring static var
+ is faster for dynamic linking. Also this match logic hidding vtables
+ from LTO symbol tables. */
+ if ((in_lto_p || flag_whole_program)
+ && !vnode->force_output
+ && DECL_COMDAT (vnode->decl) && DECL_VIRTUAL_P (vnode->decl))
+ return false;
+
+ /* When doing link time optimizations, hidden symbols become local. */
+ if (in_lto_p
+ && (DECL_VISIBILITY (vnode->decl) == VISIBILITY_HIDDEN
+ || DECL_VISIBILITY (vnode->decl) == VISIBILITY_INTERNAL)
+ /* Be sure that node is defined in IR file, not in other object
+ file. In that case we don't set used_from_other_object_file. */
+ && vnode->finalized)
+ ;
+ else if (!flag_whole_program)
+ return true;
+
+ /* Do not attempt to privatize COMDATS by default.
+ This would break linking with C++ libraries sharing
+ inline definitions.
+
+ FIXME: We can do so for readonly vars with no address taken and
+ possibly also for vtables since no direct pointer comparsion is done.
+ It might be interesting to do so to reduce linking overhead. */
+ if (DECL_COMDAT (vnode->decl) || DECL_WEAK (vnode->decl))
return true;
return false;
}
{
struct cgraph_node *node;
struct varpool_node *vnode;
+ struct pointer_set_t *aliased_nodes = pointer_set_create ();
+ struct pointer_set_t *aliased_vnodes = pointer_set_create ();
+ unsigned i;
+ alias_pair *p;
+
+ /* Discover aliased nodes. */
+ FOR_EACH_VEC_ELT (alias_pair, alias_pairs, i, p)
+ {
+ if (dump_file)
+ fprintf (dump_file, "Alias %s->%s",
+ IDENTIFIER_POINTER (DECL_ASSEMBLER_NAME (p->decl)),
+ IDENTIFIER_POINTER (p->target));
+
+ if ((node = cgraph_node_for_asm (p->target)) != NULL
+ && !DECL_EXTERNAL (node->decl))
+ {
+ if (!node->analyzed)
+ continue;
+ cgraph_mark_needed_node (node);
+ gcc_assert (node->needed);
+ pointer_set_insert (aliased_nodes, node);
+ if (dump_file)
+ fprintf (dump_file, " node %s/%i",
+ cgraph_node_name (node), node->uid);
+ }
+ else if ((vnode = varpool_node_for_asm (p->target)) != NULL
+ && !DECL_EXTERNAL (vnode->decl))
+ {
+ varpool_mark_needed_node (vnode);
+ gcc_assert (vnode->needed);
+ pointer_set_insert (aliased_vnodes, vnode);
+ if (dump_file)
+ fprintf (dump_file, " varpool node %s",
+ varpool_node_name (vnode));
+ }
+ if (dump_file)
+ fprintf (dump_file, "\n");
+ }
for (node = cgraph_nodes; node; node = node->next)
{
+ int flags = flags_from_decl_or_type (node->decl);
+
+ /* Optimize away PURE and CONST constructors and destructors. */
+ if (optimize
+ && (flags & (ECF_CONST | ECF_PURE))
+ && !(flags & ECF_LOOPING_CONST_OR_PURE))
+ {
+ DECL_STATIC_CONSTRUCTOR (node->decl) = 0;
+ DECL_STATIC_DESTRUCTOR (node->decl) = 0;
+ }
+
+ /* Frontends and alias code marks nodes as needed before parsing is finished.
+ We may end up marking as node external nodes where this flag is meaningless
+ strip it. */
+ if (node->needed
+ && (DECL_EXTERNAL (node->decl) || !node->analyzed))
+ node->needed = 0;
+
/* C++ FE on lack of COMDAT support create local COMDAT functions
(that ought to be shared but can not due to object format
limitations). It is neccesary to keep the flag to make rest of C++ FE
}
gcc_assert ((!DECL_WEAK (node->decl) && !DECL_COMDAT (node->decl))
|| TREE_PUBLIC (node->decl) || DECL_EXTERNAL (node->decl));
- if (cgraph_externally_visible_p (node, whole_program))
+ if (cgraph_externally_visible_p (node, whole_program,
+ pointer_set_contains (aliased_nodes,
+ node)))
{
gcc_assert (!node->global.inlined_to);
node->local.externally_visible = true;
if (!node->local.externally_visible && node->analyzed
&& !DECL_EXTERNAL (node->decl))
{
- struct cgraph_node *alias;
- gcc_assert (whole_program || !TREE_PUBLIC (node->decl));
+ gcc_assert (whole_program || in_lto_p || !TREE_PUBLIC (node->decl));
cgraph_make_decl_local (node->decl);
- for (alias = node->same_body; alias; alias = alias->next)
- cgraph_make_decl_local (alias->decl);
+ node->resolution = LDPR_PREVAILING_DEF_IRONLY;
if (node->same_comdat_group)
/* cgraph_externally_visible_p has already checked all other nodes
in the group and they will all be made local. We need to
segfault though. */
dissolve_same_comdat_group_list (node);
}
- node->local.local = cgraph_local_node_p (node);
+
+ if (node->thunk.thunk_p
+ && TREE_PUBLIC (node->decl))
+ {
+ struct cgraph_node *decl_node = node;
+
+ decl_node = cgraph_function_node (decl_node->callees->callee, NULL);
+
+ /* Thunks have the same visibility as function they are attached to.
+ Make sure the C++ front end set this up properly. */
+ if (DECL_ONE_ONLY (decl_node->decl))
+ {
+ gcc_checking_assert (DECL_COMDAT (node->decl)
+ == DECL_COMDAT (decl_node->decl));
+ gcc_checking_assert (DECL_COMDAT_GROUP (node->decl)
+ == DECL_COMDAT_GROUP (decl_node->decl));
+ gcc_checking_assert (node->same_comdat_group);
+ }
+ if (DECL_EXTERNAL (decl_node->decl))
+ DECL_EXTERNAL (node->decl) = 1;
+ }
}
+ for (node = cgraph_nodes; node; node = node->next)
+ node->local.local = cgraph_local_node_p (node);
for (vnode = varpool_nodes; vnode; vnode = vnode->next)
{
/* weak flag makes no sense on local variables. */
if (!vnode->finalized)
continue;
if (vnode->needed
- && (DECL_COMDAT (vnode->decl) || TREE_PUBLIC (vnode->decl))
- && (!whole_program
- /* We can privatize comdat readonly variables whose address is not taken,
- but doing so is not going to bring us optimization oppurtunities until
- we start reordering datastructures. */
- || DECL_COMDAT (vnode->decl)
- || DECL_WEAK (vnode->decl)
- || lookup_attribute ("externally_visible",
- DECL_ATTRIBUTES (vnode->decl))))
+ && varpool_externally_visible_p
+ (vnode,
+ pointer_set_contains (aliased_vnodes, vnode)))
vnode->externally_visible = true;
else
vnode->externally_visible = false;
if (!vnode->externally_visible)
{
- gcc_assert (whole_program || !TREE_PUBLIC (vnode->decl));
+ gcc_assert (in_lto_p || whole_program || !TREE_PUBLIC (vnode->decl));
cgraph_make_decl_local (vnode->decl);
+ vnode->resolution = LDPR_PREVAILING_DEF_IRONLY;
}
gcc_assert (TREE_STATIC (vnode->decl));
}
+ pointer_set_destroy (aliased_nodes);
+ pointer_set_destroy (aliased_vnodes);
if (dump_file)
{
static unsigned int
local_function_and_variable_visibility (void)
{
- return function_and_variable_visibility (flag_whole_program && !flag_lto && !flag_whopr);
+ return function_and_variable_visibility (flag_whole_program && !flag_lto);
}
struct simple_ipa_opt_pass pass_ipa_function_and_variable_visibility =
fprintf (dump_file, " %s", varpool_node_name (vnode));
fprintf (dump_file, "\n\n");
}
+ if (optimize)
+ ipa_discover_readonly_nonaddressable_vars ();
return 0;
}
NULL, /* variable_transform */
};
-/* Hash a cgraph node set element. */
-
-static hashval_t
-hash_cgraph_node_set_element (const void *p)
-{
- const_cgraph_node_set_element element = (const_cgraph_node_set_element) p;
- return htab_hash_pointer (element->node);
-}
-
-/* Compare two cgraph node set elements. */
-static int
-eq_cgraph_node_set_element (const void *p1, const void *p2)
-{
- const_cgraph_node_set_element e1 = (const_cgraph_node_set_element) p1;
- const_cgraph_node_set_element e2 = (const_cgraph_node_set_element) p2;
-
- return e1->node == e2->node;
-}
-
-/* Create a new cgraph node set. */
-
-cgraph_node_set
-cgraph_node_set_new (void)
-{
- cgraph_node_set new_node_set;
-
- new_node_set = GGC_NEW (struct cgraph_node_set_def);
- new_node_set->hashtab = htab_create_ggc (10,
- hash_cgraph_node_set_element,
- eq_cgraph_node_set_element,
- NULL);
- new_node_set->nodes = NULL;
- return new_node_set;
-}
-
-/* Add cgraph_node NODE to cgraph_node_set SET. */
+/* Simple ipa profile pass propagating frequencies across the callgraph. */
-void
-cgraph_node_set_add (cgraph_node_set set, struct cgraph_node *node)
+static unsigned int
+ipa_profile (void)
{
- void **slot;
- cgraph_node_set_element element;
- struct cgraph_node_set_element_def dummy;
-
- dummy.node = node;
- slot = htab_find_slot (set->hashtab, &dummy, INSERT);
+ struct cgraph_node **order = XCNEWVEC (struct cgraph_node *, cgraph_n_nodes);
+ struct cgraph_edge *e;
+ int order_pos;
+ bool something_changed = false;
+ int i;
- if (*slot != HTAB_EMPTY_ENTRY)
+ order_pos = ipa_reverse_postorder (order);
+ for (i = order_pos - 1; i >= 0; i--)
{
- element = (cgraph_node_set_element) *slot;
- gcc_assert (node == element->node
- && (VEC_index (cgraph_node_ptr, set->nodes, element->index)
- == node));
- return;
+ if (order[i]->local.local && cgraph_propagate_frequency (order[i]))
+ {
+ for (e = order[i]->callees; e; e = e->next_callee)
+ if (e->callee->local.local && !e->callee->aux)
+ {
+ something_changed = true;
+ e->callee->aux = (void *)1;
+ }
+ }
+ order[i]->aux = NULL;
}
- /* Insert node into hash table. */
- element =
- (cgraph_node_set_element) GGC_NEW (struct cgraph_node_set_element_def);
- element->node = node;
- element->index = VEC_length (cgraph_node_ptr, set->nodes);
- *slot = element;
-
- /* Insert into node vector. */
- VEC_safe_push (cgraph_node_ptr, gc, set->nodes, node);
+ while (something_changed)
+ {
+ something_changed = false;
+ for (i = order_pos - 1; i >= 0; i--)
+ {
+ if (order[i]->aux && cgraph_propagate_frequency (order[i]))
+ {
+ for (e = order[i]->callees; e; e = e->next_callee)
+ if (e->callee->local.local && !e->callee->aux)
+ {
+ something_changed = true;
+ e->callee->aux = (void *)1;
+ }
+ }
+ order[i]->aux = NULL;
+ }
+ }
+ free (order);
+ return 0;
}
-/* Remove cgraph_node NODE from cgraph_node_set SET. */
-
-void
-cgraph_node_set_remove (cgraph_node_set set, struct cgraph_node *node)
+static bool
+gate_ipa_profile (void)
{
- void **slot, **last_slot;
- cgraph_node_set_element element, last_element;
- struct cgraph_node *last_node;
- struct cgraph_node_set_element_def dummy;
-
- dummy.node = node;
- slot = htab_find_slot (set->hashtab, &dummy, NO_INSERT);
- if (slot == NULL)
- return;
-
- element = (cgraph_node_set_element) *slot;
- gcc_assert (VEC_index (cgraph_node_ptr, set->nodes, element->index)
- == node);
+ return flag_ipa_profile;
+}
- /* Remove from vector. We do this by swapping node with the last element
- of the vector. */
- last_node = VEC_pop (cgraph_node_ptr, set->nodes);
- if (last_node != node)
- {
- dummy.node = last_node;
- last_slot = htab_find_slot (set->hashtab, &dummy, NO_INSERT);
- last_element = (cgraph_node_set_element) *last_slot;
- gcc_assert (last_element);
-
- /* Move the last element to the original spot of NODE. */
- last_element->index = element->index;
- VEC_replace (cgraph_node_ptr, set->nodes, last_element->index,
- last_node);
- }
+struct ipa_opt_pass_d pass_ipa_profile =
+{
+ {
+ IPA_PASS,
+ "profile_estimate", /* name */
+ gate_ipa_profile, /* gate */
+ ipa_profile, /* execute */
+ NULL, /* sub */
+ NULL, /* next */
+ 0, /* static_pass_number */
+ TV_IPA_PROFILE, /* tv_id */
+ 0, /* properties_required */
+ 0, /* properties_provided */
+ 0, /* properties_destroyed */
+ 0, /* todo_flags_start */
+ 0 /* todo_flags_finish */
+ },
+ NULL, /* generate_summary */
+ NULL, /* write_summary */
+ NULL, /* read_summary */
+ NULL, /* write_optimization_summary */
+ NULL, /* read_optimization_summary */
+ NULL, /* stmt_fixup */
+ 0, /* TODOs */
+ NULL, /* function_transform */
+ NULL /* variable_transform */
+};
- /* Remove element from hash table. */
- htab_clear_slot (set->hashtab, slot);
- ggc_free (element);
-}
+/* Generate and emit a static constructor or destructor. WHICH must
+ be one of 'I' (for a constructor) or 'D' (for a destructor). BODY
+ is a STATEMENT_LIST containing GENERIC statements. PRIORITY is the
+ initialization priority for this constructor or destructor.
-/* Find NODE in SET and return an iterator to it if found. A null iterator
- is returned if NODE is not in SET. */
+ FINAL specify whether the externally visible name for collect2 should
+ be produced. */
-cgraph_node_set_iterator
-cgraph_node_set_find (cgraph_node_set set, struct cgraph_node *node)
+static void
+cgraph_build_static_cdtor_1 (char which, tree body, int priority, bool final)
{
- void **slot;
- struct cgraph_node_set_element_def dummy;
- cgraph_node_set_element element;
- cgraph_node_set_iterator csi;
-
- dummy.node = node;
- slot = htab_find_slot (set->hashtab, &dummy, NO_INSERT);
- if (slot == NULL)
- csi.index = (unsigned) ~0;
+ static int counter = 0;
+ char which_buf[16];
+ tree decl, name, resdecl;
+
+ /* The priority is encoded in the constructor or destructor name.
+ collect2 will sort the names and arrange that they are called at
+ program startup. */
+ if (final)
+ sprintf (which_buf, "%c_%.5d_%d", which, priority, counter++);
else
+ /* Proudce sane name but one not recognizable by collect2, just for the
+ case we fail to inline the function. */
+ sprintf (which_buf, "sub_%c_%.5d_%d", which, priority, counter++);
+ name = get_file_function_name (which_buf);
+
+ decl = build_decl (input_location, FUNCTION_DECL, name,
+ build_function_type_list (void_type_node, NULL_TREE));
+ current_function_decl = decl;
+
+ resdecl = build_decl (input_location,
+ RESULT_DECL, NULL_TREE, void_type_node);
+ DECL_ARTIFICIAL (resdecl) = 1;
+ DECL_RESULT (decl) = resdecl;
+ DECL_CONTEXT (resdecl) = decl;
+
+ allocate_struct_function (decl, false);
+
+ TREE_STATIC (decl) = 1;
+ TREE_USED (decl) = 1;
+ DECL_ARTIFICIAL (decl) = 1;
+ DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT (decl) = 1;
+ DECL_SAVED_TREE (decl) = body;
+ if (!targetm.have_ctors_dtors && final)
{
- element = (cgraph_node_set_element) *slot;
- gcc_assert (VEC_index (cgraph_node_ptr, set->nodes, element->index)
- == node);
- csi.index = element->index;
+ TREE_PUBLIC (decl) = 1;
+ DECL_PRESERVE_P (decl) = 1;
}
- csi.set = set;
-
- return csi;
-}
+ DECL_UNINLINABLE (decl) = 1;
-/* Dump content of SET to file F. */
+ DECL_INITIAL (decl) = make_node (BLOCK);
+ TREE_USED (DECL_INITIAL (decl)) = 1;
-void
-dump_cgraph_node_set (FILE *f, cgraph_node_set set)
-{
- cgraph_node_set_iterator iter;
+ DECL_SOURCE_LOCATION (decl) = input_location;
+ cfun->function_end_locus = input_location;
- for (iter = csi_start (set); !csi_end_p (iter); csi_next (&iter))
+ switch (which)
{
- struct cgraph_node *node = csi_node (iter);
- fprintf (f, " %s/%i", cgraph_node_name (node), node->uid);
+ case 'I':
+ DECL_STATIC_CONSTRUCTOR (decl) = 1;
+ decl_init_priority_insert (decl, priority);
+ break;
+ case 'D':
+ DECL_STATIC_DESTRUCTOR (decl) = 1;
+ decl_fini_priority_insert (decl, priority);
+ break;
+ default:
+ gcc_unreachable ();
}
- fprintf (f, "\n");
-}
-/* Dump content of SET to stderr. */
+ gimplify_function_tree (decl);
-DEBUG_FUNCTION void
-debug_cgraph_node_set (cgraph_node_set set)
-{
- dump_cgraph_node_set (stderr, set);
+ cgraph_add_new_function (decl, false);
+
+ set_cfun (NULL);
+ current_function_decl = NULL;
}
-/* Hash a varpool node set element. */
+/* Generate and emit a static constructor or destructor. WHICH must
+ be one of 'I' (for a constructor) or 'D' (for a destructor). BODY
+ is a STATEMENT_LIST containing GENERIC statements. PRIORITY is the
+ initialization priority for this constructor or destructor. */
-static hashval_t
-hash_varpool_node_set_element (const void *p)
+void
+cgraph_build_static_cdtor (char which, tree body, int priority)
{
- const_varpool_node_set_element element = (const_varpool_node_set_element) p;
- return htab_hash_pointer (element->node);
+ cgraph_build_static_cdtor_1 (which, body, priority, false);
}
-/* Compare two varpool node set elements. */
-
-static int
-eq_varpool_node_set_element (const void *p1, const void *p2)
-{
- const_varpool_node_set_element e1 = (const_varpool_node_set_element) p1;
- const_varpool_node_set_element e2 = (const_varpool_node_set_element) p2;
+/* A vector of FUNCTION_DECLs declared as static constructors. */
+static VEC(tree, heap) *static_ctors;
+/* A vector of FUNCTION_DECLs declared as static destructors. */
+static VEC(tree, heap) *static_dtors;
- return e1->node == e2->node;
-}
+/* When target does not have ctors and dtors, we call all constructor
+ and destructor by special initialization/destruction function
+ recognized by collect2.
-/* Create a new varpool node set. */
+ When we are going to build this function, collect all constructors and
+ destructors and turn them into normal functions. */
-varpool_node_set
-varpool_node_set_new (void)
+static void
+record_cdtor_fn (struct cgraph_node *node)
{
- varpool_node_set new_node_set;
-
- new_node_set = GGC_NEW (struct varpool_node_set_def);
- new_node_set->hashtab = htab_create_ggc (10,
- hash_varpool_node_set_element,
- eq_varpool_node_set_element,
- NULL);
- new_node_set->nodes = NULL;
- return new_node_set;
+ if (DECL_STATIC_CONSTRUCTOR (node->decl))
+ VEC_safe_push (tree, heap, static_ctors, node->decl);
+ if (DECL_STATIC_DESTRUCTOR (node->decl))
+ VEC_safe_push (tree, heap, static_dtors, node->decl);
+ node = cgraph_get_node (node->decl);
+ DECL_DISREGARD_INLINE_LIMITS (node->decl) = 1;
}
-/* Add varpool_node NODE to varpool_node_set SET. */
+/* Define global constructors/destructor functions for the CDTORS, of
+ which they are LEN. The CDTORS are sorted by initialization
+ priority. If CTOR_P is true, these are constructors; otherwise,
+ they are destructors. */
-void
-varpool_node_set_add (varpool_node_set set, struct varpool_node *node)
+static void
+build_cdtor (bool ctor_p, VEC (tree, heap) *cdtors)
{
- void **slot;
- varpool_node_set_element element;
- struct varpool_node_set_element_def dummy;
-
- dummy.node = node;
- slot = htab_find_slot (set->hashtab, &dummy, INSERT);
+ size_t i,j;
+ size_t len = VEC_length (tree, cdtors);
- if (*slot != HTAB_EMPTY_ENTRY)
+ i = 0;
+ while (i < len)
{
- element = (varpool_node_set_element) *slot;
- gcc_assert (node == element->node
- && (VEC_index (varpool_node_ptr, set->nodes, element->index)
- == node));
- return;
- }
-
- /* Insert node into hash table. */
- element =
- (varpool_node_set_element) GGC_NEW (struct varpool_node_set_element_def);
- element->node = node;
- element->index = VEC_length (varpool_node_ptr, set->nodes);
- *slot = element;
+ tree body;
+ tree fn;
+ priority_type priority;
+
+ priority = 0;
+ body = NULL_TREE;
+ j = i;
+ do
+ {
+ priority_type p;
+ fn = VEC_index (tree, cdtors, j);
+ p = ctor_p ? DECL_INIT_PRIORITY (fn) : DECL_FINI_PRIORITY (fn);
+ if (j == i)
+ priority = p;
+ else if (p != priority)
+ break;
+ j++;
+ }
+ while (j < len);
- /* Insert into node vector. */
- VEC_safe_push (varpool_node_ptr, gc, set->nodes, node);
+ /* When there is only one cdtor and target supports them, do nothing. */
+ if (j == i + 1
+ && targetm.have_ctors_dtors)
+ {
+ i++;
+ continue;
+ }
+ /* Find the next batch of constructors/destructors with the same
+ initialization priority. */
+ for (;i < j; i++)
+ {
+ tree call;
+ fn = VEC_index (tree, cdtors, i);
+ call = build_call_expr (fn, 0);
+ if (ctor_p)
+ DECL_STATIC_CONSTRUCTOR (fn) = 0;
+ else
+ DECL_STATIC_DESTRUCTOR (fn) = 0;
+ /* We do not want to optimize away pure/const calls here.
+ When optimizing, these should be already removed, when not
+ optimizing, we want user to be able to breakpoint in them. */
+ TREE_SIDE_EFFECTS (call) = 1;
+ append_to_statement_list (call, &body);
+ }
+ gcc_assert (body != NULL_TREE);
+ /* Generate a function to call all the function of like
+ priority. */
+ cgraph_build_static_cdtor_1 (ctor_p ? 'I' : 'D', body, priority, true);
+ }
}
-/* Remove varpool_node NODE from varpool_node_set SET. */
+/* Comparison function for qsort. P1 and P2 are actually of type
+ "tree *" and point to static constructors. DECL_INIT_PRIORITY is
+ used to determine the sort order. */
-void
-varpool_node_set_remove (varpool_node_set set, struct varpool_node *node)
+static int
+compare_ctor (const void *p1, const void *p2)
{
- void **slot, **last_slot;
- varpool_node_set_element element, last_element;
- struct varpool_node *last_node;
- struct varpool_node_set_element_def dummy;
-
- dummy.node = node;
- slot = htab_find_slot (set->hashtab, &dummy, NO_INSERT);
- if (slot == NULL)
- return;
-
- element = (varpool_node_set_element) *slot;
- gcc_assert (VEC_index (varpool_node_ptr, set->nodes, element->index)
- == node);
-
- /* Remove from vector. We do this by swapping node with the last element
- of the vector. */
- last_node = VEC_pop (varpool_node_ptr, set->nodes);
- if (last_node != node)
- {
- dummy.node = last_node;
- last_slot = htab_find_slot (set->hashtab, &dummy, NO_INSERT);
- last_element = (varpool_node_set_element) *last_slot;
- gcc_assert (last_element);
-
- /* Move the last element to the original spot of NODE. */
- last_element->index = element->index;
- VEC_replace (varpool_node_ptr, set->nodes, last_element->index,
- last_node);
- }
-
- /* Remove element from hash table. */
- htab_clear_slot (set->hashtab, slot);
- ggc_free (element);
+ tree f1;
+ tree f2;
+ int priority1;
+ int priority2;
+
+ f1 = *(const tree *)p1;
+ f2 = *(const tree *)p2;
+ priority1 = DECL_INIT_PRIORITY (f1);
+ priority2 = DECL_INIT_PRIORITY (f2);
+
+ if (priority1 < priority2)
+ return -1;
+ else if (priority1 > priority2)
+ return 1;
+ else
+ /* Ensure a stable sort. Constructors are executed in backwarding
+ order to make LTO initialize braries first. */
+ return DECL_UID (f2) - DECL_UID (f1);
}
-/* Find NODE in SET and return an iterator to it if found. A null iterator
- is returned if NODE is not in SET. */
+/* Comparison function for qsort. P1 and P2 are actually of type
+ "tree *" and point to static destructors. DECL_FINI_PRIORITY is
+ used to determine the sort order. */
-varpool_node_set_iterator
-varpool_node_set_find (varpool_node_set set, struct varpool_node *node)
+static int
+compare_dtor (const void *p1, const void *p2)
{
- void **slot;
- struct varpool_node_set_element_def dummy;
- varpool_node_set_element element;
- varpool_node_set_iterator vsi;
-
- dummy.node = node;
- slot = htab_find_slot (set->hashtab, &dummy, NO_INSERT);
- if (slot == NULL)
- vsi.index = (unsigned) ~0;
+ tree f1;
+ tree f2;
+ int priority1;
+ int priority2;
+
+ f1 = *(const tree *)p1;
+ f2 = *(const tree *)p2;
+ priority1 = DECL_FINI_PRIORITY (f1);
+ priority2 = DECL_FINI_PRIORITY (f2);
+
+ if (priority1 < priority2)
+ return -1;
+ else if (priority1 > priority2)
+ return 1;
else
- {
- element = (varpool_node_set_element) *slot;
- gcc_assert (VEC_index (varpool_node_ptr, set->nodes, element->index)
- == node);
- vsi.index = element->index;
- }
- vsi.set = set;
-
- return vsi;
+ /* Ensure a stable sort. */
+ return DECL_UID (f1) - DECL_UID (f2);
}
-/* Dump content of SET to file F. */
+/* Generate functions to call static constructors and destructors
+ for targets that do not support .ctors/.dtors sections. These
+ functions have magic names which are detected by collect2. */
-void
-dump_varpool_node_set (FILE *f, varpool_node_set set)
+static void
+build_cdtor_fns (void)
{
- varpool_node_set_iterator iter;
-
- for (iter = vsi_start (set); !vsi_end_p (iter); vsi_next (&iter))
+ if (!VEC_empty (tree, static_ctors))
{
- struct varpool_node *node = vsi_node (iter);
- fprintf (f, " %s", varpool_node_name (node));
+ gcc_assert (!targetm.have_ctors_dtors || in_lto_p);
+ VEC_qsort (tree, static_ctors, compare_ctor);
+ build_cdtor (/*ctor_p=*/true, static_ctors);
}
- fprintf (f, "\n");
-}
-
-/* Dump content of SET to stderr. */
-DEBUG_FUNCTION void
-debug_varpool_node_set (varpool_node_set set)
-{
- dump_varpool_node_set (stderr, set);
+ if (!VEC_empty (tree, static_dtors))
+ {
+ gcc_assert (!targetm.have_ctors_dtors || in_lto_p);
+ VEC_qsort (tree, static_dtors, compare_dtor);
+ build_cdtor (/*ctor_p=*/false, static_dtors);
+ }
}
-
-/* Simple ipa profile pass propagating frequencies across the callgraph. */
+/* Look for constructors and destructors and produce function calling them.
+ This is needed for targets not supporting ctors or dtors, but we perform the
+ transformation also at linktime to merge possibly numberous
+ constructors/destructors into single function to improve code locality and
+ reduce size. */
static unsigned int
-ipa_profile (void)
+ipa_cdtor_merge (void)
{
- struct cgraph_node **order = XCNEWVEC (struct cgraph_node *, cgraph_n_nodes);
- struct cgraph_edge *e;
- int order_pos;
- bool something_changed = false;
- int i;
-
- order_pos = cgraph_postorder (order);
- for (i = order_pos - 1; i >= 0; i--)
- {
- if (order[i]->local.local && cgraph_propagate_frequency (order[i]))
- {
- for (e = order[i]->callees; e; e = e->next_callee)
- if (e->callee->local.local && !e->callee->aux)
- {
- something_changed = true;
- e->callee->aux = (void *)1;
- }
- }
- order[i]->aux = NULL;
- }
-
- while (something_changed)
- {
- something_changed = false;
- for (i = order_pos - 1; i >= 0; i--)
- {
- if (order[i]->aux && cgraph_propagate_frequency (order[i]))
- {
- for (e = order[i]->callees; e; e = e->next_callee)
- if (e->callee->local.local && !e->callee->aux)
- {
- something_changed = true;
- e->callee->aux = (void *)1;
- }
- }
- order[i]->aux = NULL;
- }
- }
- free (order);
+ struct cgraph_node *node;
+ for (node = cgraph_nodes; node; node = node->next)
+ if (node->analyzed
+ && (DECL_STATIC_CONSTRUCTOR (node->decl)
+ || DECL_STATIC_DESTRUCTOR (node->decl)))
+ record_cdtor_fn (node);
+ build_cdtor_fns ();
+ VEC_free (tree, heap, static_ctors);
+ VEC_free (tree, heap, static_dtors);
return 0;
}
+/* Perform the pass when we have no ctors/dtors support
+ or at LTO time to merge multiple constructors into single
+ function. */
+
static bool
-gate_ipa_profile (void)
+gate_ipa_cdtor_merge (void)
{
- return flag_ipa_profile;
+ return !targetm.have_ctors_dtors || (optimize && in_lto_p);
}
-struct ipa_opt_pass_d pass_ipa_profile =
+struct ipa_opt_pass_d pass_ipa_cdtor_merge =
{
{
IPA_PASS,
- "ipa-profile", /* name */
- gate_ipa_profile, /* gate */
- ipa_profile, /* execute */
+ "cdtor", /* name */
+ gate_ipa_cdtor_merge, /* gate */
+ ipa_cdtor_merge, /* execute */
NULL, /* sub */
NULL, /* next */
0, /* static_pass_number */
- TV_IPA_PROFILE, /* tv_id */
+ TV_CGRAPHOPT, /* tv_id */
0, /* properties_required */
0, /* properties_provided */
0, /* properties_destroyed */
OSDN Git Service
