public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] Avoid creating (const (reg ...)) [PR108603]
@ 2023-03-02 10:21 Richard Sandiford
  2023-03-02 10:26 ` Richard Biener
  0 siblings, 1 reply; 2+ messages in thread
From: Richard Sandiford @ 2023-03-02 10:21 UTC (permalink / raw)
  To: gcc-patches

convert_memory_address_addr_space_1 has two modes: one in which it
tries to create a self-contained RTL expression (which might fail)
and one in which it can emit new instructions where necessary.

When handling a CONST, the function recurses into the CONST's
operand and then constifies the result.  But that's only valid if
the result is still a self-contained expression.  If new instructions
have been emitted, the expression will refer to the (non-constant)
results of those instructions.

In the PR, this caused us to emit a nonsensical (const (reg ...))
REG_EQUAL note.

Tested on aarch64-linux-gnu & x86_64-linux-gnu.  OK to install?

Richard


gcc/
	PR tree-optimization/108603
	* explow.cc (convert_memory_address_addr_space_1): Only wrap
	the result of a recursive call in a CONST if no instructions
	were emitted.

gcc/testsuite/
	PR tree-optimization/108603
	* gcc.target/aarch64/sve/pr108603.c: New test.
---
 gcc/explow.cc                                   | 11 ++++++++---
 gcc/testsuite/gcc.target/aarch64/sve/pr108603.c |  8 ++++++++
 2 files changed, 16 insertions(+), 3 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/aarch64/sve/pr108603.c

diff --git a/gcc/explow.cc b/gcc/explow.cc
index 83439b32abe..32e9498ee07 100644
--- a/gcc/explow.cc
+++ b/gcc/explow.cc
@@ -349,9 +349,14 @@ convert_memory_address_addr_space_1 (scalar_int_mode to_mode ATTRIBUTE_UNUSED,
       return temp;
 
     case CONST:
-      temp = convert_memory_address_addr_space_1 (to_mode, XEXP (x, 0), as,
-						  true, no_emit);
-      return temp ? gen_rtx_CONST (to_mode, temp) : temp;
+      {
+	auto *last = no_emit ? nullptr : get_last_insn ();
+	temp = convert_memory_address_addr_space_1 (to_mode, XEXP (x, 0), as,
+						    true, no_emit);
+	if (temp && (no_emit || last == get_last_insn ()))
+	  return gen_rtx_CONST (to_mode, temp);
+	return temp;
+      }
 
     case PLUS:
     case MULT:
diff --git a/gcc/testsuite/gcc.target/aarch64/sve/pr108603.c b/gcc/testsuite/gcc.target/aarch64/sve/pr108603.c
new file mode 100644
index 00000000000..a2aea9f0b12
--- /dev/null
+++ b/gcc/testsuite/gcc.target/aarch64/sve/pr108603.c
@@ -0,0 +1,8 @@
+/* { dg-options "-O2 -mabi=ilp32 -fdata-sections" } */
+
+int a[128];
+long long *p;
+void f() {
+  for (long i = 0; i < sizeof(long); i++)
+    p[i] = a[i];
+}
-- 
2.25.1


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] Avoid creating (const (reg ...)) [PR108603]
  2023-03-02 10:21 [PATCH] Avoid creating (const (reg ...)) [PR108603] Richard Sandiford
@ 2023-03-02 10:26 ` Richard Biener
  0 siblings, 0 replies; 2+ messages in thread
From: Richard Biener @ 2023-03-02 10:26 UTC (permalink / raw)
  To: Richard Sandiford, gcc-patches

On Thu, Mar 2, 2023 at 11:21 AM Richard Sandiford via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> convert_memory_address_addr_space_1 has two modes: one in which it
> tries to create a self-contained RTL expression (which might fail)
> and one in which it can emit new instructions where necessary.
>
> When handling a CONST, the function recurses into the CONST's
> operand and then constifies the result.  But that's only valid if
> the result is still a self-contained expression.  If new instructions
> have been emitted, the expression will refer to the (non-constant)
> results of those instructions.
>
> In the PR, this caused us to emit a nonsensical (const (reg ...))
> REG_EQUAL note.
>
> Tested on aarch64-linux-gnu & x86_64-linux-gnu.  OK to install?

OK.

Thanks,
Richard.

> Richard
>
>
> gcc/
>         PR tree-optimization/108603
>         * explow.cc (convert_memory_address_addr_space_1): Only wrap
>         the result of a recursive call in a CONST if no instructions
>         were emitted.
>
> gcc/testsuite/
>         PR tree-optimization/108603
>         * gcc.target/aarch64/sve/pr108603.c: New test.
> ---
>  gcc/explow.cc                                   | 11 ++++++++---
>  gcc/testsuite/gcc.target/aarch64/sve/pr108603.c |  8 ++++++++
>  2 files changed, 16 insertions(+), 3 deletions(-)
>  create mode 100644 gcc/testsuite/gcc.target/aarch64/sve/pr108603.c
>
> diff --git a/gcc/explow.cc b/gcc/explow.cc
> index 83439b32abe..32e9498ee07 100644
> --- a/gcc/explow.cc
> +++ b/gcc/explow.cc
> @@ -349,9 +349,14 @@ convert_memory_address_addr_space_1 (scalar_int_mode to_mode ATTRIBUTE_UNUSED,
>        return temp;
>
>      case CONST:
> -      temp = convert_memory_address_addr_space_1 (to_mode, XEXP (x, 0), as,
> -                                                 true, no_emit);
> -      return temp ? gen_rtx_CONST (to_mode, temp) : temp;
> +      {
> +       auto *last = no_emit ? nullptr : get_last_insn ();
> +       temp = convert_memory_address_addr_space_1 (to_mode, XEXP (x, 0), as,
> +                                                   true, no_emit);
> +       if (temp && (no_emit || last == get_last_insn ()))
> +         return gen_rtx_CONST (to_mode, temp);
> +       return temp;
> +      }
>
>      case PLUS:
>      case MULT:
> diff --git a/gcc/testsuite/gcc.target/aarch64/sve/pr108603.c b/gcc/testsuite/gcc.target/aarch64/sve/pr108603.c
> new file mode 100644
> index 00000000000..a2aea9f0b12
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/aarch64/sve/pr108603.c
> @@ -0,0 +1,8 @@
> +/* { dg-options "-O2 -mabi=ilp32 -fdata-sections" } */
> +
> +int a[128];
> +long long *p;
> +void f() {
> +  for (long i = 0; i < sizeof(long); i++)
> +    p[i] = a[i];
> +}
> --
> 2.25.1
>

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2023-03-02 10:26 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-03-02 10:21 [PATCH] Avoid creating (const (reg ...)) [PR108603] Richard Sandiford
2023-03-02 10:26 ` Richard Biener

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).