diff mbox series

tree-inline: Fix a -fcompare-debug issue in the inliner [PR94167]

Message ID 20200314074356.GB2156@tucnak
State New
Headers show
Series tree-inline: Fix a -fcompare-debug issue in the inliner [PR94167] | expand

Commit Message

Li, Pan2 via Gcc-patches March 14, 2020, 7:43 a.m. UTC
Hi!

The following testcase fails with -fcompare-debug.  The problem is that
bar is marked as address_taken only with -g and not without.
I've tracked it down to insert_init_stmt calling gimple_regimplify_operands
even on DEBUG_STMTs.  That function will just insert normal stmts before
the DEBUG_STMT if the DEBUG_STMT operand isn't gimple val or invariant.
While DCE will turn those statements into debug temporaries, it can cause
differences in SSA_NAMEs and more importantly, the ipa references are
generated from those before the DCE happens.
On the testcase, the DEBUG_STMT value is (int)bar.

We could generate DEBUG_STMTs with debug temporaries instead, but I fail to
see the reason to do that, DEBUG_STMTs allow other expressions and all we
want to ensure is that the expressions aren't too large (arbitrarily
complex), but during inlining/function versioning I don't see why something
would queue a DEBUG_STMT with arbitrarily complex expressions in there.

So, this patch just doesn't regimplify DEBUG_STMTs.  Bootstrapped/regtested
on x86_64-linux and i686-linux, ok for trunk?

2020-03-14  Jakub Jelinek  <jakub@redhat.com>

	PR debug/94167
	* tree-inline.c (insert_init_stmt): Don't gimple_regimplify_operands
	DEBUG_STMTs.

	* gcc.dg/pr94167.c: New test.


	Jakub

Comments

Richard Biener March 16, 2020, 7:36 a.m. UTC | #1
On Sat, 14 Mar 2020, Jakub Jelinek wrote:

> Hi!
> 
> The following testcase fails with -fcompare-debug.  The problem is that
> bar is marked as address_taken only with -g and not without.
> I've tracked it down to insert_init_stmt calling gimple_regimplify_operands
> even on DEBUG_STMTs.  That function will just insert normal stmts before
> the DEBUG_STMT if the DEBUG_STMT operand isn't gimple val or invariant.
> While DCE will turn those statements into debug temporaries, it can cause
> differences in SSA_NAMEs and more importantly, the ipa references are
> generated from those before the DCE happens.
> On the testcase, the DEBUG_STMT value is (int)bar.
> 
> We could generate DEBUG_STMTs with debug temporaries instead, but I fail to
> see the reason to do that, DEBUG_STMTs allow other expressions and all we
> want to ensure is that the expressions aren't too large (arbitrarily
> complex), but during inlining/function versioning I don't see why something
> would queue a DEBUG_STMT with arbitrarily complex expressions in there.
> 
> So, this patch just doesn't regimplify DEBUG_STMTs.  Bootstrapped/regtested
> on x86_64-linux and i686-linux, ok for trunk?

OK.

Thanks,
Richard.

> 2020-03-14  Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR debug/94167
> 	* tree-inline.c (insert_init_stmt): Don't gimple_regimplify_operands
> 	DEBUG_STMTs.
> 
> 	* gcc.dg/pr94167.c: New test.
> 
> --- gcc/tree-inline.c.jj	2020-03-04 12:58:35.033528551 +0100
> +++ gcc/tree-inline.c	2020-03-13 16:19:04.129493354 +0100
> @@ -3361,10 +3361,10 @@ insert_init_stmt (copy_body_data *id, ba
>  	  gimple_assign_set_rhs1 (init_stmt, rhs);
>  	}
>        gsi_insert_after (&si, init_stmt, GSI_NEW_STMT);
> -      gimple_regimplify_operands (init_stmt, &si);
> -
>        if (!is_gimple_debug (init_stmt))
>  	{
> +	  gimple_regimplify_operands (init_stmt, &si);
> +
>  	  tree def = gimple_assign_lhs (init_stmt);
>  	  insert_init_debug_bind (id, bb, def, def, init_stmt);
>  	}
> --- gcc/testsuite/gcc.dg/pr94167.c.jj	2020-03-13 16:33:34.046632902 +0100
> +++ gcc/testsuite/gcc.dg/pr94167.c	2020-03-13 16:33:08.353012748 +0100
> @@ -0,0 +1,33 @@
> +/* PR debug/94167 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fcompare-debug" } */
> +
> +struct S { int g, h; signed char i; int j; signed char k; int l[4]; } a, c;
> +struct T { signed char g; } e;
> +int *b, d;
> +static void foo ();
> +
> +void
> +bar (void)
> +{
> +  while (d)
> +    {
> +      int k;
> +      struct T f[3];
> +      foo (bar, a);
> +      for (k = 0;; k++)
> +	f[k] = e;
> +    }
> +}
> +
> +static inline void
> +foo (int x, struct S y, struct T z)
> +{
> +  for (z.g = 2; z.g; z.g--)
> +    {
> +      c = a = y;
> +      *b |= 6;
> +      if (y.g)
> +	break;
> +    }
> +}
> 
> 	Jakub
> 
>
diff mbox series

Patch

--- gcc/tree-inline.c.jj	2020-03-04 12:58:35.033528551 +0100
+++ gcc/tree-inline.c	2020-03-13 16:19:04.129493354 +0100
@@ -3361,10 +3361,10 @@  insert_init_stmt (copy_body_data *id, ba
 	  gimple_assign_set_rhs1 (init_stmt, rhs);
 	}
       gsi_insert_after (&si, init_stmt, GSI_NEW_STMT);
-      gimple_regimplify_operands (init_stmt, &si);
-
       if (!is_gimple_debug (init_stmt))
 	{
+	  gimple_regimplify_operands (init_stmt, &si);
+
 	  tree def = gimple_assign_lhs (init_stmt);
 	  insert_init_debug_bind (id, bb, def, def, init_stmt);
 	}
--- gcc/testsuite/gcc.dg/pr94167.c.jj	2020-03-13 16:33:34.046632902 +0100
+++ gcc/testsuite/gcc.dg/pr94167.c	2020-03-13 16:33:08.353012748 +0100
@@ -0,0 +1,33 @@ 
+/* PR debug/94167 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -fcompare-debug" } */
+
+struct S { int g, h; signed char i; int j; signed char k; int l[4]; } a, c;
+struct T { signed char g; } e;
+int *b, d;
+static void foo ();
+
+void
+bar (void)
+{
+  while (d)
+    {
+      int k;
+      struct T f[3];
+      foo (bar, a);
+      for (k = 0;; k++)
+	f[k] = e;
+    }
+}
+
+static inline void
+foo (int x, struct S y, struct T z)
+{
+  for (z.g = 2; z.g; z.g--)
+    {
+      c = a = y;
+      *b |= 6;
+      if (y.g)
+	break;
+    }
+}