public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Biener <rguenther@suse.de>
To: Jason Merrill <jason@redhat.com>
Cc: Bernd Edlinger <bernd.edlinger@hotmail.de>,
	    Jakub Jelinek <jakub@redhat.com>,
	Jonathan Wakely <jwakely@redhat.com>,
	    Florian Weimer <fweimer@redhat.com>,
	GCC Patches <gcc-patches@gcc.gnu.org>,
	    Jeff Law <law@redhat.com>
Subject: Re: [PATCH] Add a new type attribute always_alias (PR79671)
Date: Tue, 11 Apr 2017 10:32:00 -0000	[thread overview]
Message-ID: <alpine.LSU.2.20.1704111226090.30715@zhemvz.fhfr.qr> (raw)
In-Reply-To: <CADzB+2n0YC+ykdbhOV4_4LHZDuPCzTSt5vbBN8Lva1n9Pjc1pg@mail.gmail.com>

On Mon, 10 Apr 2017, Jason Merrill wrote:

> On Mon, Apr 10, 2017 at 11:30 AM, Richard Biener <rguenther@suse.de> wrote:
> > On Mon, 10 Apr 2017, Jason Merrill wrote:
> >> On Mon, Apr 10, 2017 at 8:50 AM, Richard Biener <rguenther@suse.de> wrote:
> >> >         * tree.c (build_cplus_array_type): Set TYPE_TYPELESS_STORAGE
> >> >         for arrays of unsigned char or std::byte.
> >>
> >> I think it would be good to have a flag to select whether these
> >> semantics apply to any char variant and std::byte, only unsigned char
> >> and std::byte, or only std::byte.
> >
> > Any suggestion?  Not sure we really need it (I'm hesitant to write
> > all the testcases to verify it actually works).
> 
> Well, there's existing code that uses plain char (e.g. boost) that I
> want to support.  If there's a significant optimization penalty for
> allowing that, we probably also want to be able to limit the impact.
> If there isn't much penalty, let's just support all char variants.

I haven't assessed the penalty involved but it can't be worse than
the situation we had in GCC 6.  So I think it's reasonable to support
all char variants for now.  One could add some instrumenting to
alias_set_subset_of / alias_sets_conflict_p but it would only yield
an upper bound on the number of failed queries (TBAA is a quite early
out before PTA info is used for example).

The following variant -- added missing

Index: gcc/cp/tree.c
===================================================================
--- gcc/cp/tree.c       (revision 246832)
+++ gcc/cp/tree.c       (working copy)
@@ -972,6 +979,7 @@ build_cplus_array_type (tree elt_type, t
                 as it will overwrite alignment etc. of all variants.  */
              TYPE_SIZE (t) = TYPE_SIZE (m);
              TYPE_SIZE_UNIT (t) = TYPE_SIZE_UNIT (m);
+             TYPE_TYPELESS_STORAGE (t) = TYPE_TYPELESS_STORAGE (m);
            }
 
          TYPE_MAIN_VARIANT (t) = m;

that caused LTO bootstrap to fail and removed the tree-ssa-structalias.c
change (committed separately) [LTO] bootstrapped and tested ok on 
x86_64-unknown-linux-gnu.

I've tested some template examples and they seem to work fine.

Ok for trunk?

Disclaimer: there might still be an issue with cross-language LTO
support, but it might at most result in TYPE_TYPELESS_STORAGE
getting lost.  Trying to craft a testcase to verify my theory.

Thanks,
Richard.

2017-04-11  Richard Biener  <rguenther@suse.de>

	PR c++/79671
	cp/
	* tree.c (build_cplus_array_type): Set TYPE_TYPELESS_STORAGE
	for arrays of unsigned char or std::byte.

	* tree-core.h (tree_type_common): Add typeless_storage flag.
	* tree.h (TYPE_TYPELESS_STORAGE): New macro.
	(canonical_type_used_p): Add arrays with TYPE_TYPELESS_STORAGE.
	* alias.c (alias_set_entry): Add has_typeless_storage member.
	(alias_set_subset_of): Handle it.
	(alias_sets_conflict_p): Likewise.
	(init_alias_set_entry): Initialize it.
	(get_alias_set): For ARRAY_TYPEs handle TYPE_TYPELESS_STORAGE.
	(record_alias_subset): Likewise.
	* lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
	* tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
	TYPE_TYPELESS_STORAGE.
	* tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.

	lto/
	* lto.c (compare_tree_sccs_1): Compare TYPE_TYPELESS_STORAGE.

	* g++.dg/torture/pr79671.C: New testcase.

Index: gcc/alias.c
===================================================================
--- gcc/alias.c	(revision 246832)
+++ gcc/alias.c	(working copy)
@@ -136,6 +136,9 @@ struct GTY(()) alias_set_entry {
   bool is_pointer;
   /* Nonzero if is_pointer or if one of childs have has_pointer set.  */
   bool has_pointer;
+  /* Nonzero if we have a child serving as typeless storage (or are
+     such storage ourselves).  */
+  bool has_typeless_storage;
 
   /* The children of the alias set.  These are not just the immediate
      children, but, in fact, all descendants.  So, if we have:
@@ -419,7 +422,8 @@ alias_set_subset_of (alias_set_type set1
   /* Check if set1 is a subset of set2.  */
   ase2 = get_alias_set_entry (set2);
   if (ase2 != 0
-      && (ase2->has_zero_child
+      && (ase2->has_typeless_storage
+	  || ase2->has_zero_child
 	  || (ase2->children && ase2->children->get (set1))))
     return true;
 
@@ -480,7 +484,8 @@ alias_sets_conflict_p (alias_set_type se
   /* See if the first alias set is a subset of the second.  */
   ase1 = get_alias_set_entry (set1);
   if (ase1 != 0
-      && ase1->children && ase1->children->get (set2))
+      && (ase1->has_typeless_storage
+	  || (ase1->children && ase1->children->get (set2))))
     {
       ++alias_stats.num_dag;
       return 1;
@@ -489,7 +494,8 @@ alias_sets_conflict_p (alias_set_type se
   /* Now do the same, but with the alias sets reversed.  */
   ase2 = get_alias_set_entry (set2);
   if (ase2 != 0
-      && ase2->children && ase2->children->get (set1))
+      && (ase2->has_typeless_storage
+	  || (ase2->children && ase2->children->get (set1))))
     {
       ++alias_stats.num_dag;
       return 1;
@@ -825,6 +831,7 @@ init_alias_set_entry (alias_set_type set
   ase->has_zero_child = false;
   ase->is_pointer = false;
   ase->has_pointer = false;
+  ase->has_typeless_storage = false;
   gcc_checking_assert (!get_alias_set_entry (set));
   (*alias_sets)[set] = ase;
   return ase;
@@ -955,6 +962,7 @@ get_alias_set (tree t)
      Just be pragmatic here and make sure the array and its element
      type get the same alias set assigned.  */
   else if (TREE_CODE (t) == ARRAY_TYPE
+	   && ! TYPE_TYPELESS_STORAGE (t)
 	   && (!TYPE_NONALIASED_COMPONENT (t)
 	       || TYPE_STRUCTURAL_EQUALITY_P (t)))
     set = get_alias_set (TREE_TYPE (t));
@@ -1094,6 +1102,15 @@ get_alias_set (tree t)
 
   TYPE_ALIAS_SET (t) = set;
 
+  if (TREE_CODE (t) == ARRAY_TYPE
+      && TYPE_TYPELESS_STORAGE (t))
+    {
+      alias_set_entry *ase = get_alias_set_entry (set);
+      if (!ase)
+	ase = init_alias_set_entry (set);
+      ase->has_typeless_storage = true;
+    }
+
   /* If this is an aggregate type or a complex type, we must record any
      component aliasing information.  */
   if (AGGREGATE_TYPE_P (t) || TREE_CODE (t) == COMPLEX_TYPE)
@@ -1173,6 +1190,8 @@ record_alias_subset (alias_set_type supe
 	    superset_entry->has_zero_child = true;
           if (subset_entry->has_pointer)
 	    superset_entry->has_pointer = true;
+	  if (subset_entry->has_typeless_storage)
+	    superset_entry->has_typeless_storage = true;
 
 	  if (subset_entry->children)
 	    {
Index: gcc/cp/tree.c
===================================================================
--- gcc/cp/tree.c	(revision 246832)
+++ gcc/cp/tree.c	(working copy)
@@ -949,6 +949,13 @@ build_cplus_array_type (tree elt_type, t
   else
     {
       t = build_array_type (elt_type, index_type);
+      if (elt_type == unsigned_char_type_node
+	  || elt_type == signed_char_type_node
+	  || elt_type == char_type_node
+	  || (TREE_CODE (elt_type) == ENUMERAL_TYPE
+	      && TYPE_CONTEXT (elt_type) == std_node
+	      && !strcmp ("byte", TYPE_NAME_STRING (elt_type))))
+	TYPE_TYPELESS_STORAGE (t) = 1;
     }
 
   /* Now check whether we already have this array variant.  */
@@ -972,6 +979,7 @@ build_cplus_array_type (tree elt_type, t
 		 as it will overwrite alignment etc. of all variants.  */
 	      TYPE_SIZE (t) = TYPE_SIZE (m);
 	      TYPE_SIZE_UNIT (t) = TYPE_SIZE_UNIT (m);
+	      TYPE_TYPELESS_STORAGE (t) = TYPE_TYPELESS_STORAGE (m);
 	    }
 
 	  TYPE_MAIN_VARIANT (t) = m;
Index: gcc/lto/lto.c
===================================================================
--- gcc/lto/lto.c	(revision 246832)
+++ gcc/lto/lto.c	(working copy)
@@ -1161,7 +1161,10 @@ compare_tree_sccs_1 (tree t1, tree t2, t
 	  compare_values (TYPE_FINAL_P);
 	}
       else if (code == ARRAY_TYPE)
-	compare_values (TYPE_NONALIASED_COMPONENT);
+	{
+	  compare_values (TYPE_NONALIASED_COMPONENT);
+	  compare_values (TYPE_TYPELESS_STORAGE);
+	}
       compare_values (TYPE_PACKED);
       compare_values (TYPE_RESTRICT);
       compare_values (TYPE_USER_ALIGN);
Index: gcc/lto-streamer-out.c
===================================================================
--- gcc/lto-streamer-out.c	(revision 246832)
+++ gcc/lto-streamer-out.c	(working copy)
@@ -1142,7 +1142,10 @@ hash_tree (struct streamer_tree_cache_d
 	  hstate.add_flag (TYPE_FINAL_P (t));
 	}
       else if (code == ARRAY_TYPE)
-	hstate.add_flag (TYPE_NONALIASED_COMPONENT (t));
+	{
+	  hstate.add_flag (TYPE_NONALIASED_COMPONENT (t));
+	  hstate.add_flag (TYPE_TYPELESS_STORAGE (t));
+	}
       hstate.commit_flag ();
       hstate.add_int (TYPE_PRECISION (t));
       hstate.add_int (TYPE_ALIGN (t));
Index: gcc/testsuite/g++.dg/torture/pr79671.C
===================================================================
--- gcc/testsuite/g++.dg/torture/pr79671.C	(nonexistent)
+++ gcc/testsuite/g++.dg/torture/pr79671.C	(working copy)
@@ -0,0 +1,25 @@
+// { dg-do run }
+
+void *operator new(__SIZE_TYPE__, void *p2) { return p2; }
+struct B { B(int i_) : i(i_) {} int i; };
+struct X
+{
+  unsigned char buf[sizeof (B)];
+};
+
+int __attribute__((noinline)) foo()
+{
+  X x alignas(B), y alignas(B);
+  new (&x) B (0);
+  y = x;
+  B *q = reinterpret_cast <B *>(&y);
+  asm volatile ("" : "=r" (q) : "r" (q));
+  return q->i;
+}
+
+int main()
+{
+  if (foo() != 0)
+    __builtin_abort ();
+  return 0;
+}
Index: gcc/tree-core.h
===================================================================
--- gcc/tree-core.h	(revision 246832)
+++ gcc/tree-core.h	(working copy)
@@ -1511,7 +1511,9 @@ struct GTY(()) tree_type_common {
      so we need to store the value 32 (not 31, as we need the zero
      as well), hence six bits.  */
   unsigned align : 6;
-  unsigned spare : 25;
+  unsigned typeless_storage : 1;
+  unsigned spare : 24;
+
   alias_set_type alias_set;
   tree pointer_to;
   tree reference_to;
Index: gcc/tree-streamer-in.c
===================================================================
--- gcc/tree-streamer-in.c	(revision 246832)
+++ gcc/tree-streamer-in.c	(working copy)
@@ -375,7 +375,10 @@ unpack_ts_type_common_value_fields (stru
       TYPE_FINAL_P (expr) = (unsigned) bp_unpack_value (bp, 1);
     }
   else if (TREE_CODE (expr) == ARRAY_TYPE)
-    TYPE_NONALIASED_COMPONENT (expr) = (unsigned) bp_unpack_value (bp, 1);
+    {
+      TYPE_NONALIASED_COMPONENT (expr) = (unsigned) bp_unpack_value (bp, 1);
+      TYPE_TYPELESS_STORAGE (expr) = (unsigned) bp_unpack_value (bp, 1);
+    }
   TYPE_PRECISION (expr) = bp_unpack_var_len_unsigned (bp);
   SET_TYPE_ALIGN (expr, bp_unpack_var_len_unsigned (bp));
 #ifdef ACCEL_COMPILER
Index: gcc/tree-streamer-out.c
===================================================================
--- gcc/tree-streamer-out.c	(revision 246832)
+++ gcc/tree-streamer-out.c	(working copy)
@@ -327,7 +327,10 @@ pack_ts_type_common_value_fields (struct
       bp_pack_value (bp, TYPE_FINAL_P (expr), 1);
     }
   else if (TREE_CODE (expr) == ARRAY_TYPE)
-    bp_pack_value (bp, TYPE_NONALIASED_COMPONENT (expr), 1);
+    {
+      bp_pack_value (bp, TYPE_NONALIASED_COMPONENT (expr), 1);
+      bp_pack_value (bp, TYPE_TYPELESS_STORAGE (expr), 1);
+    }
   bp_pack_var_len_unsigned (bp, TYPE_PRECISION (expr));
   bp_pack_var_len_unsigned (bp, TYPE_ALIGN (expr));
 }
Index: gcc/tree.h
===================================================================
--- gcc/tree.h	(revision 246832)
+++ gcc/tree.h	(working copy)
@@ -2035,6 +2035,9 @@ extern machine_mode element_mode (const_
 #define TYPE_NONALIASED_COMPONENT(NODE) \
   (ARRAY_TYPE_CHECK (NODE)->type_common.transparent_aggr_flag)
 
+#define TYPE_TYPELESS_STORAGE(NODE) \
+  (ARRAY_TYPE_CHECK (NODE)->type_common.typeless_storage)
+
 /* Indicated that objects of this type should be laid out in as
    compact a way as possible.  */
 #define TYPE_PACKED(NODE) (TYPE_CHECK (NODE)->base.u.bits.packed_flag)
@@ -4914,7 +4917,7 @@ inline bool
 canonical_type_used_p (const_tree t)
 {
   return !(POINTER_TYPE_P (t)
-	   || TREE_CODE (t) == ARRAY_TYPE
+	   || (TREE_CODE (t) == ARRAY_TYPE && ! TYPE_TYPELESS_STORAGE (t))
 	   || TREE_CODE (t) == VECTOR_TYPE);
 }
 

  reply	other threads:[~2017-04-11 10:32 UTC|newest]

Thread overview: 65+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-04-05  9:46 Bernd Edlinger
2017-04-05 13:28 ` Jakub Jelinek
2017-04-05 15:20   ` Richard Biener
2017-04-05 17:41     ` Bernd Edlinger
2017-04-05 20:18       ` Jason Merrill
2017-04-05 20:46         ` Bernd Edlinger
2017-04-05 22:54           ` Pedro Alves
2017-04-06 10:08           ` Jonathan Wakely
2017-04-06  7:23         ` Richard Biener
2017-04-05 15:27   ` Bernd Edlinger
2017-04-05 15:29     ` Jakub Jelinek
2017-04-05 14:50 ` Florian Weimer
2017-04-05 15:23   ` Richard Biener
2017-04-05 15:38     ` Bernd Edlinger
2017-04-05 16:03       ` Jonathan Wakely
2017-04-05 16:08         ` Jakub Jelinek
2017-04-05 17:23           ` Bernd Edlinger
2017-04-05 21:02             ` Bernd Edlinger
2017-04-05 23:17               ` Sandra Loosemore
2017-04-06  5:40                 ` Jakub Jelinek
2017-04-06  7:47               ` Richard Biener
2017-04-06  7:51                 ` Jakub Jelinek
2017-04-06  7:55                   ` Richard Biener
2017-04-06 14:11                     ` Bernd Edlinger
2017-04-06 14:17                       ` Florian Weimer
2017-04-06 14:23                         ` Richard Biener
2017-04-06 14:43                           ` Jonathan Wakely
2017-04-06 14:51                             ` Florian Weimer
2017-04-06 15:05                               ` Jakub Jelinek
2017-04-06 15:10                                 ` Florian Weimer
2017-04-06 19:13                               ` Richard Biener
2017-04-11 10:43                                 ` Florian Weimer
2017-04-11 10:48                                   ` Richard Biener
2017-04-06 17:39                         ` Bernd Edlinger
2017-04-06 17:48                           ` Florian Weimer
2017-04-06 18:12                             ` Bernd Edlinger
2017-04-06 18:19                               ` Florian Weimer
2017-04-06 18:49                                 ` Bernd Edlinger
2017-04-06 19:05                                   ` Florian Weimer
2017-04-06 19:20                                     ` Bernd Edlinger
2017-04-07  6:47                                       ` Richard Biener
2017-04-07 12:58                                         ` Bernd Edlinger
2017-04-06 19:16                               ` Richard Biener
2017-04-07  6:56                                 ` Florian Weimer
2017-04-07  8:01                                   ` Richard Biener
2017-04-06 19:14                           ` Richard Biener
2017-04-06 19:51                             ` Bernd Edlinger
2017-04-06 14:22                       ` Richard Biener
2017-04-06 21:00                 ` Bernd Edlinger
2017-04-07  6:54                   ` Richard Biener
2017-04-07 13:37                     ` Bernd Edlinger
2017-04-07 15:10                       ` Richard Biener
2017-04-07 15:33                         ` Bernd Edlinger
2017-04-07 20:22                           ` Jason Merrill
2017-04-10 12:50                             ` Richard Biener
2017-04-10 14:41                               ` Jason Merrill
2017-04-10 15:31                                 ` Richard Biener
2017-04-10 16:35                                   ` Jason Merrill
2017-04-11 10:32                                     ` Richard Biener [this message]
2017-04-11 11:53                                       ` Richard Biener
2017-04-11 13:35                                         ` Richard Biener
2017-04-11 18:47                                           ` Jason Merrill
2017-04-10 21:40                               ` Pedro Alves
2017-04-11  7:37                                 ` Richard Biener
2017-04-06 20:20               ` Bernd Edlinger

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=alpine.LSU.2.20.1704111226090.30715@zhemvz.fhfr.qr \
    --to=rguenther@suse.de \
    --cc=bernd.edlinger@hotmail.de \
    --cc=fweimer@redhat.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jakub@redhat.com \
    --cc=jason@redhat.com \
    --cc=jwakely@redhat.com \
    --cc=law@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).