diff mbox

Fill bitregion_{start,end} in store_constructor (PR, tree-optimization/78428).

Message ID f06f62fa-1f69-5554-461b-42186d43e2f9@suse.cz
State New
Headers show

Commit Message

Martin Liška Dec. 9, 2016, 10:19 a.m. UTC
On 11/23/2016 11:49 AM, Richard Biener wrote:
> On Wed, Nov 23, 2016 at 11:26 AM, Martin Liška <mliska@suse.cz> wrote:

>> Following patch fixes situation where we do a store to a bitfield which

>> is at boundary of a record. This leads to usage of wider store, leading

>> to overwriting a following memory location.

>>

>> Patch can bootstrap on ppc64le-redhat-linux and survives regression tests.

>> Apart from that, the reported test-case in PR works on x86_64-linux-gnu.

>>

>> Ready to be installed?

> 

> +  HOST_WIDE_INT bitregion_end

> +    = exp_size == -1 ? 0 : exp_size * BITS_PER_UNIT - 1;

> 

> I don't think looking at the CONSTRUCTOR to determine bitregion_end is

> a good idea.

> The function gets 'size' as argument which is documented as "number of

> bytes we are

> allowed to modify" - so better use that.

> 

> @@ -6308,7 +6314,8 @@ store_constructor (tree exp, rtx target, int

> cleared, HOST_WIDE_INT size,

>                 MEM_KEEP_ALIAS_SET_P (to_rtx) = 1;

>               }

> 

> -           store_constructor_field (to_rtx, bitsize, bitpos, mode,

> +           store_constructor_field (to_rtx, bitsize, bitpos,

> +                                    0, bitregion_end, mode,

>                                      value, cleared,

>                                      get_alias_set (TREE_TYPE (field)),

>                                      reverse);

> 

> this stores to to_rtx which may be offsetted from target this means in this case

> bitregion_end is not conservative enough - you'd need to resort to the

> field width

> in that case I guess (and for variable field size not specify any end

> -- I suppose

> the 'size' store_constructor gets might also be "unknown"?).  But maybe all

> the non-constant offset / size cases are "dead code" now that we are in GIMPLE?

> Note they likely can only appear from Ada code anyway -- CCing Eric.

> 

> I suppose a "safe" thing to do would be to give up on the first

> variable offset/size

> and re-set bitregion_end to zero for this and all following fields.


Ok. I'm sending a patch that put gcc_unreachable to places where either size
or (and) offset is a non-constant. This survives regression tests (including ada)
on x86_64-linux-gnu. Apart from that normal bootstrap + regression tests works fine
on ppc64le-redhat-linux.

I'm not sure whether there can be a code that has variable size and offset?

Martin

> 

> The other cases look fine to me.

> 

> Thanks,

> Richard.

> 

>> Martin

Comments

Eric Botcazou Dec. 12, 2016, 11:10 a.m. UTC | #1
> Ok. I'm sending a patch that put gcc_unreachable to places where either size

> or (and) offset is a non-constant. This survives regression tests

> (including ada) on x86_64-linux-gnu. Apart from that normal bootstrap +

> regression tests works fine on ppc64le-redhat-linux.


I didn't manage to break it so it is OK by me.

-- 
Eric Botcazou
Martin Liška Dec. 13, 2016, 9:05 a.m. UTC | #2
On 12/12/2016 12:10 PM, Eric Botcazou wrote:
>> Ok. I'm sending a patch that put gcc_unreachable to places where either size

>> or (and) offset is a non-constant. This survives regression tests

>> (including ada) on x86_64-linux-gnu. Apart from that normal bootstrap +

>> regression tests works fine on ppc64le-redhat-linux.

> 

> I didn't manage to break it so it is OK by me.

> 


Out of curiosity, I blame the commit which added the 

if (offset)
  {
    machine_mode address_mode;
    rtx offset_rtx;

    offset
      = SUBSTITUTE_PLACEHOLDER_IN_EXPR (offset,
					make_tree (TREE_TYPE (exp),
						   target));

and it comes to commit done in 1993:

commit c869557a9ccc1bd3e5474b144bcb84065db23549
Author: kenner <kenner@138bc75d-0d04-0410-961f-82ee72b054a4>
Date:   Mon Oct 4 01:48:03 1993 +0000

    (store_expr): Use expr_size value, not size_int.
    (store_constructor): Handle case of variable position and allow it to contain
    a PLACEHOLDER_EXPR.
    (get_inner_reference): Make a WITH_RECORD_EXPR if required.
    (expand_expr, case PLACEHOLDER_EXPR, WITH_RECORD_EXPR): New cases.
    (expand_expr, case ARRAY_REF): Make WITH_RECORD_EXPR expressions when needed.
    
    
    git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@5584 138bc75d-0d04-0410-961f-82ee72b054a4

May I install the patch?
Martin
Richard Biener Dec. 13, 2016, 2:35 p.m. UTC | #3
On Tue, Dec 13, 2016 at 10:05 AM, Martin Liška <mliska@suse.cz> wrote:
> On 12/12/2016 12:10 PM, Eric Botcazou wrote:

>>> Ok. I'm sending a patch that put gcc_unreachable to places where either size

>>> or (and) offset is a non-constant. This survives regression tests

>>> (including ada) on x86_64-linux-gnu. Apart from that normal bootstrap +

>>> regression tests works fine on ppc64le-redhat-linux.

>>

>> I didn't manage to break it so it is OK by me.

>>

>

> Out of curiosity, I blame the commit which added the

>

> if (offset)

>   {

>     machine_mode address_mode;

>     rtx offset_rtx;

>

>     offset

>       = SUBSTITUTE_PLACEHOLDER_IN_EXPR (offset,

>                                         make_tree (TREE_TYPE (exp),

>                                                    target));

>

> and it comes to commit done in 1993:

>

> commit c869557a9ccc1bd3e5474b144bcb84065db23549

> Author: kenner <kenner@138bc75d-0d04-0410-961f-82ee72b054a4>

> Date:   Mon Oct 4 01:48:03 1993 +0000

>

>     (store_expr): Use expr_size value, not size_int.

>     (store_constructor): Handle case of variable position and allow it to contain

>     a PLACEHOLDER_EXPR.

>     (get_inner_reference): Make a WITH_RECORD_EXPR if required.

>     (expand_expr, case PLACEHOLDER_EXPR, WITH_RECORD_EXPR): New cases.

>     (expand_expr, case ARRAY_REF): Make WITH_RECORD_EXPR expressions when needed.


Yeah, I expect that we got rid of the cases at the time we introduced
GIMPLE (and gimplification).

>

>     git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@5584 138bc75d-0d04-0410-961f-82ee72b054a4

>

> May I install the patch?


OK from my POV.

Thanks,
Richard.

> Martin

>
Martin Liška Dec. 15, 2016, 11 a.m. UTC | #4
On 12/13/2016 03:35 PM, Richard Biener wrote:
> OK from my POV.

> 

> Thanks,

> Richard.


Hi.

I prepared backport (and tested) for GCC 5 and 6 branches,
may I install the patch after 6.3 will be released?

Thanks,
Martin
Richard Biener Dec. 16, 2016, 1:02 p.m. UTC | #5
On Thu, Dec 15, 2016 at 12:00 PM, Martin Liška <mliska@suse.cz> wrote:
> On 12/13/2016 03:35 PM, Richard Biener wrote:

>> OK from my POV.

>>

>> Thanks,

>> Richard.

>

> Hi.

>

> I prepared backport (and tested) for GCC 5 and 6 branches,

> may I install the patch after 6.3 will be released?


Ok for early next year.

Thanks,
Richard.

> Thanks,

> Martin
diff mbox

Patch

From c8d3e7c33c22bfdaf9e285e9e747701b8646cdf2 Mon Sep 17 00:00:00 2001
From: marxin <mliska@suse.cz>
Date: Tue, 22 Nov 2016 10:15:42 +0100
Subject: [PATCH] Fill bitregion_{start,end} in store_constructor (PR
 tree-optimization/78428).

gcc/ChangeLog:

2016-12-08  Martin Liska  <mliska@suse.cz>

	* expr.c (store_constructor_field): Add new arguments to the
	function.
	(store_constructor): Set up bitregion_end and add
	gcc_unreachable to fields that have either non-constant size
	or (and) offset.

gcc/testsuite/ChangeLog:

2016-12-08  Martin Liska  <mliska@suse.cz>

	* gcc.dg/tree-ssa/pr78428.c: New test.
---
 gcc/expr.c                              | 52 +++++++++++++--------------------
 gcc/testsuite/gcc.dg/tree-ssa/pr78428.c | 27 +++++++++++++++++
 2 files changed, 47 insertions(+), 32 deletions(-)
 create mode 100644 gcc/testsuite/gcc.dg/tree-ssa/pr78428.c

diff --git a/gcc/expr.c b/gcc/expr.c
index fe752fb..88da8dd 100644
--- a/gcc/expr.c
+++ b/gcc/expr.c
@@ -80,7 +80,8 @@  static void clear_by_pieces (rtx, unsigned HOST_WIDE_INT, unsigned int);
 static rtx_insn *compress_float_constant (rtx, rtx);
 static rtx get_subtarget (rtx);
 static void store_constructor_field (rtx, unsigned HOST_WIDE_INT,
-				     HOST_WIDE_INT, machine_mode,
+				     HOST_WIDE_INT, unsigned HOST_WIDE_INT,
+				     unsigned HOST_WIDE_INT, machine_mode,
 				     tree, int, alias_set_type, bool);
 static void store_constructor (tree, rtx, int, HOST_WIDE_INT, bool);
 static rtx store_field (rtx, HOST_WIDE_INT, HOST_WIDE_INT,
@@ -6077,7 +6078,10 @@  all_zeros_p (const_tree exp)
 
 static void
 store_constructor_field (rtx target, unsigned HOST_WIDE_INT bitsize,
-			 HOST_WIDE_INT bitpos, machine_mode mode,
+			 HOST_WIDE_INT bitpos,
+			 unsigned HOST_WIDE_INT bitregion_start,
+			 unsigned HOST_WIDE_INT bitregion_end,
+			 machine_mode mode,
 			 tree exp, int cleared,
 			 alias_set_type alias_set, bool reverse)
 {
@@ -6112,8 +6116,8 @@  store_constructor_field (rtx target, unsigned HOST_WIDE_INT bitsize,
 			 reverse);
     }
   else
-    store_field (target, bitsize, bitpos, 0, 0, mode, exp, alias_set, false,
-		 reverse);
+    store_field (target, bitsize, bitpos, bitregion_start, bitregion_end, mode,
+		 exp, alias_set, false, reverse);
 }
 
 
@@ -6148,6 +6152,7 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 {
   tree type = TREE_TYPE (exp);
   HOST_WIDE_INT exp_size = int_size_in_bytes (type);
+  HOST_WIDE_INT bitregion_end = size > 0 ? size * BITS_PER_UNIT - 1 : 0;
 
   switch (TREE_CODE (type))
     {
@@ -6225,7 +6230,7 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 	    if (tree_fits_uhwi_p (DECL_SIZE (field)))
 	      bitsize = tree_to_uhwi (DECL_SIZE (field));
 	    else
-	      bitsize = -1;
+	      gcc_unreachable ();
 
 	    mode = DECL_MODE (field);
 	    if (DECL_BIT_FIELD (field))
@@ -6236,31 +6241,10 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 		&& tree_fits_shwi_p (bit_position (field)))
 	      {
 		bitpos = int_bit_position (field);
-		offset = 0;
+		offset = NULL_TREE;
 	      }
 	    else
-	      bitpos = tree_to_shwi (DECL_FIELD_BIT_OFFSET (field));
-
-	    if (offset)
-	      {
-	        machine_mode address_mode;
-		rtx offset_rtx;
-
-		offset
-		  = SUBSTITUTE_PLACEHOLDER_IN_EXPR (offset,
-						    make_tree (TREE_TYPE (exp),
-							       target));
-
-		offset_rtx = expand_normal (offset);
-		gcc_assert (MEM_P (to_rtx));
-
-		address_mode = get_address_mode (to_rtx);
-		if (GET_MODE (offset_rtx) != address_mode)
-		  offset_rtx = convert_to_mode (address_mode, offset_rtx, 0);
-
-		to_rtx = offset_address (to_rtx, offset_rtx,
-					 highest_pow2_factor (offset));
-	      }
+	      gcc_unreachable ();
 
 	    /* If this initializes a field that is smaller than a
 	       word, at the start of a word, try to widen it to a full
@@ -6308,7 +6292,8 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 		MEM_KEEP_ALIAS_SET_P (to_rtx) = 1;
 	      }
 
-	    store_constructor_field (to_rtx, bitsize, bitpos, mode,
+	    store_constructor_field (to_rtx, bitsize, bitpos,
+				     0, bitregion_end, mode,
 				     value, cleared,
 				     get_alias_set (TREE_TYPE (field)),
 				     reverse);
@@ -6468,7 +6453,8 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 			  }
 
 			store_constructor_field
-			  (target, bitsize, bitpos, mode, value, cleared,
+			  (target, bitsize, bitpos, 0, bitregion_end,
+			   mode, value, cleared,
 			   get_alias_set (elttype), reverse);
 		      }
 		  }
@@ -6571,7 +6557,8 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 		    target = copy_rtx (target);
 		    MEM_KEEP_ALIAS_SET_P (target) = 1;
 		  }
-		store_constructor_field (target, bitsize, bitpos, mode, value,
+		store_constructor_field (target, bitsize, bitpos, 0,
+					 bitregion_end, mode, value,
 					 cleared, get_alias_set (elttype),
 					 reverse);
 	      }
@@ -6705,7 +6692,8 @@  store_constructor (tree exp, rtx target, int cleared, HOST_WIDE_INT size,
 		  ? TYPE_MODE (TREE_TYPE (value))
 		  : eltmode;
 		bitpos = eltpos * elt_size;
-		store_constructor_field (target, bitsize, bitpos, value_mode,
+		store_constructor_field (target, bitsize, bitpos, 0,
+					 bitregion_end, value_mode,
 					 value, cleared, alias, reverse);
 	      }
 	  }
diff --git a/gcc/testsuite/gcc.dg/tree-ssa/pr78428.c b/gcc/testsuite/gcc.dg/tree-ssa/pr78428.c
new file mode 100644
index 0000000..3a9b99c
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/pr78428.c
@@ -0,0 +1,27 @@ 
+/* PR tree-optimization/78428.  */
+/* { dg-options "-O2" } */
+/* { dg-do run } */
+
+struct S0
+{
+  int f2;
+  int f3:16;
+  int f4:18;
+} ;
+
+int a = 5;
+struct S0 b = { 3, 0, 0 };
+static struct S0 global[2] = { { 77, 0, 78 }, { 77, 0, 78 } };
+
+int main ()
+{
+  volatile struct S0 *j;
+  for (; a;)
+    {
+      __builtin_printf ("", b.f2);
+      j = &b;
+      *j = global[1];
+      a--;
+    }
+  return 0;
+}
-- 
2.10.2