Hi,
while evaluating another patch that introduces more lvalue paradoxical
subregs I ran into an ICE in combine at
wide_int o = wi::insert (rtx_mode_t (outer, temp_mode),
rtx_mode_t (inner, dest_mode),
offset, width);
because width (=GET_MODE_PRECISION (dest_mode)) > GET_MODE_PRECISION
(temp_mode).
From the comments and the code it looks like we do not want to handle a
paradoxical subreg here so I quickly added a check for it which prevents
the ICE. The check could also be added to reg_subword_p () I guess.
Is this the right thing to do or am I missing something, i.e. my other
patch might be doing something it should not?
Bootstrap on s390x was successful and testsuite is looking good so far.
Regards
Robin
--
gcc/ChangeLog:
* combine.c (try_combine): Check for paradoxical subreg.
>From 2b74c66f8d97c30c45e99336c9871cccd8cf94e1 Mon Sep 17 00:00:00 2001
From: Robin Dapp <rd...@linux.ibm.com>
Date: Tue, 22 Jun 2021 17:35:37 +0200
Subject: [PATCH 11/11] combine paradoxical subreg fix.
---
gcc/combine.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/gcc/combine.c b/gcc/combine.c
index 6476812a212..de04560db21 100644
--- a/gcc/combine.c
+++ b/gcc/combine.c
@@ -2782,7 +2782,8 @@ try_combine (rtx_insn *i3, rtx_insn *i2, rtx_insn *i1, rtx_insn *i0,
&& CONST_SCALAR_INT_P (SET_SRC (temp_expr))
&& GET_CODE (PATTERN (i3)) == SET
&& CONST_SCALAR_INT_P (SET_SRC (PATTERN (i3)))
- && reg_subword_p (SET_DEST (PATTERN (i3)), SET_DEST (temp_expr)))
+ && reg_subword_p (SET_DEST (PATTERN (i3)), SET_DEST (temp_expr))
+ && !paradoxical_subreg_p (SET_DEST (PATTERN (i3))))
{
rtx dest = SET_DEST (PATTERN (i3));
rtx temp_dest = SET_DEST (temp_expr);
--
2.31.1
