Re: Fix ipa-prop wrt volatile memory accesses

2022-06-12 Thread Martin Jambor
Hi,

On Fri, Jun 10 2022, Jan Hubicka wrote:
> Hi,
> this patch prevents ipa-prop from propagating aggregates when load is
> volatile.  Martin, does this look OK?  It seem to me that ipa-prop may
> need some additional volatile flag checks.

load_from_unmodified_param_or_agg checks for it, although at a bit weird
place (quite late and it looks like volatile reads from an aggregate
parameter passed by value might still be considered OK).  Building of
IPA_JF_LOAD_AGG jump functions also checks for it.
compute_complex_assign_jump_func and get_ancestor_addr_info look at
mem_refs in addr_exprs, so that should be safe.

>From a quick scan of ipa-prop, it seems that only ipa_load_from_parm_agg
missed it (modulo the parenthesis above).

>
> Bootstrapped/regtested x86_64-linux, OK?

Sure, OK, thanks!

Martin


> gcc/ChangeLog:
>
> 2022-06-10  Jan Hubicka  
>
>   PR ipa/105739
>   * ipa-prop.cc (ipa_load_from_parm_agg): Disqualify volatile memory
>   accesses.
>
> gcc/testsuite/ChangeLog:
>
> 2022-06-10  Jan Hubicka  
>
>   * gcc.dg/ipa/pr105739.c: New test.
>
> diff --git a/gcc/ipa-prop.cc b/gcc/ipa-prop.cc
> index afd9222b5a2..c037668e7d8 100644
> --- a/gcc/ipa-prop.cc
> +++ b/gcc/ipa-prop.cc
> @@ -1112,6 +1112,10 @@ ipa_load_from_parm_agg (struct ipa_func_body_info *fbi,
>if (!base)
>  return false;
>  
> +  /* We can not propagate across volatile loads.  */
> +  if (TREE_THIS_VOLATILE (op))
> +return false;
> +
>if (DECL_P (base))
>  {
>int index = ipa_get_param_decl_index_1 (descriptors, base);
> diff --git a/gcc/testsuite/gcc.dg/ipa/pr105739.c 
> b/gcc/testsuite/gcc.dg/ipa/pr105739.c
> new file mode 100644
> index 000..8dbe8fc2494
> --- /dev/null
> +++ b/gcc/testsuite/gcc.dg/ipa/pr105739.c
> @@ -0,0 +1,30 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fdump-tree-optimized" } */
> +
> +
> +__attribute__((noinline))
> +static int
> +test2(int a)
> +{
> +if (__builtin_constant_p (a))
> +__builtin_abort ();
> +return a;
> +}
> +static int
> +test(int *a)
> +{
> +int val = *(volatile int *)a;
> +if (__builtin_constant_p (val))
> +__builtin_abort ();
> +if (val)
> +  return test2(val);
> +return 0;
> +}
> +int a;
> +int
> +main()
> +{
> +a = 0;
> +return test ();
> +}
> +/* { dg-final { scan-tree-dump "test2" "optimized" } } */


Fix ipa-prop wrt volatile memory accesses

2022-06-10 Thread Jan Hubicka via Gcc-patches
Hi,
this patch prevents ipa-prop from propagating aggregates when load is
volatile.  Martin, does this look OK?  It seem to me that ipa-prop may
need some additional volatile flag checks.

Bootstrapped/regtested x86_64-linux, OK?

Honza

gcc/ChangeLog:

2022-06-10  Jan Hubicka  

PR ipa/105739
* ipa-prop.cc (ipa_load_from_parm_agg): Disqualify volatile memory
accesses.

gcc/testsuite/ChangeLog:

2022-06-10  Jan Hubicka  

* gcc.dg/ipa/pr105739.c: New test.

diff --git a/gcc/ipa-prop.cc b/gcc/ipa-prop.cc
index afd9222b5a2..c037668e7d8 100644
--- a/gcc/ipa-prop.cc
+++ b/gcc/ipa-prop.cc
@@ -1112,6 +1112,10 @@ ipa_load_from_parm_agg (struct ipa_func_body_info *fbi,
   if (!base)
 return false;
 
+  /* We can not propagate across volatile loads.  */
+  if (TREE_THIS_VOLATILE (op))
+return false;
+
   if (DECL_P (base))
 {
   int index = ipa_get_param_decl_index_1 (descriptors, base);
diff --git a/gcc/testsuite/gcc.dg/ipa/pr105739.c 
b/gcc/testsuite/gcc.dg/ipa/pr105739.c
new file mode 100644
index 000..8dbe8fc2494
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/ipa/pr105739.c
@@ -0,0 +1,30 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fdump-tree-optimized" } */
+
+
+__attribute__((noinline))
+static int
+test2(int a)
+{
+if (__builtin_constant_p (a))
+__builtin_abort ();
+return a;
+}
+static int
+test(int *a)
+{
+int val = *(volatile int *)a;
+if (__builtin_constant_p (val))
+__builtin_abort ();
+if (val)
+  return test2(val);
+return 0;
+}
+int a;
+int
+main()
+{
+a = 0;
+return test ();
+}
+/* { dg-final { scan-tree-dump "test2" "optimized" } } */