Re: [PATCH] Fix PR 103288, ICE after PHI-OPT, move an assigment when still in use for another bb

Richard Biener richard.guenther@gmail.com
Wed Nov 17 07:58:16 GMT 2021


On November 17, 2021 8:46:54 AM GMT+01:00, apinski--- via Gcc-patches <gcc-patches@gcc.gnu.org> wrote:
>From: Andrew Pinski <apinski@marvell.com>
>
>The problem is r12-5300-gf98f373dd822b35c allows phiopt to recognize more basic blocks
>but missed one location where phiopt could move an assignment from the middle block
>to the non-middle one.  This patch fixes that.
>
>OK? Bootstrapped and tested on x86_64-linux-gnu with no regressions.

OK. 
Richard. 

>	PR 103288
>
>gcc/ChangeLog:
>
>	* tree-ssa-phiopt.c (value_replacement): Return early if middle
>	block has more than one pred.
>
>gcc/testsuite/ChangeLog:
>
>	* gcc.c-torture/compile/pr103288-1.c: New test.
>---
> gcc/testsuite/gcc.c-torture/compile/pr103288-1.c | 6 ++++++
> gcc/tree-ssa-phiopt.c                            | 3 +++
> 2 files changed, 9 insertions(+)
> create mode 100644 gcc/testsuite/gcc.c-torture/compile/pr103288-1.c
>
>diff --git a/gcc/testsuite/gcc.c-torture/compile/pr103288-1.c b/gcc/testsuite/gcc.c-torture/compile/pr103288-1.c
>new file mode 100644
>index 00000000000..88d1c675599
>--- /dev/null
>+++ b/gcc/testsuite/gcc.c-torture/compile/pr103288-1.c
>@@ -0,0 +1,6 @@
>+
>+int ui_5;
>+long func_14_uli_8;
>+void func_14() {
>+        ui_5 &= (func_14_uli_8 ? 60 : ui_5) ? 5 : 0;
>+}
>diff --git a/gcc/tree-ssa-phiopt.c b/gcc/tree-ssa-phiopt.c
>index 6b22f6bedd4..8984a5e15ab 100644
>--- a/gcc/tree-ssa-phiopt.c
>+++ b/gcc/tree-ssa-phiopt.c
>@@ -1381,6 +1381,9 @@ value_replacement (basic_block cond_bb, basic_block middle_bb,
> 	}
>     }
> 
>+  if (!single_pred_p (middle_bb))
>+    return 0;
>+
>   /* Now optimize (x != 0) ? x + y : y to just x + y.  */
>   gsi = gsi_last_nondebug_bb (middle_bb);
>   if (gsi_end_p (gsi))



More information about the Gcc-patches mailing list