When investigating a builtin-arith-overflow-12.c test failure I found
the cause to be improper placement of a cbranchsi4 instruction.
In openrisc a cbranchsi4 instruction will be split to something like:
(set (reg:BI ?sr_f)
(cmp:BI (reg:SI x) (reg:SI y)))
(set (pc)
(if_then_else (eq (reg:BI ?sr_f)
(const_int 0 [0]))
(label_ref:SI z)
(pc)))
Between combine and late_combine1 optimization passes I found a case
where a cbranchsi4 was getting injected between the "set sr_f" and
"jump if sr_f" instructions. The middle-end thought this was ok as it
could not see that the cbranchsi4 pattern will clobber the sr_f register
after splitting.
Add a clause to the cbranchsi4 pattern to indicate that it will clobber
sr_f. This allows the middle-end to avoid placing the cbranchsi4
pattern incorrectly.
After this patch these tests pass.
make check-gcc RUNTESTFLAGS='dg-torture.exp=builtin-arith-overflow-12.c'
gcc/ChangeLog:
* config/or1k/or1k.md (cbranchsi4): Add clobber clause.
Signed-off-by: Stafford Horne <[email protected]>
---
gcc/config/or1k/or1k.md | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/gcc/config/or1k/or1k.md b/gcc/config/or1k/or1k.md
index 85209a63f20..1af29f247d3 100644
--- a/gcc/config/or1k/or1k.md
+++ b/gcc/config/or1k/or1k.md
@@ -616,7 +616,8 @@
[(match_operand:SI 1 "reg_or_0_operand" "")
(match_operand:SI 2 "reg_or_s16_operand" "")])
(label_ref (match_operand 3 "" ""))
- (pc)))]
+ (pc)))
+ (clobber (reg:BI SR_F_REGNUM))]
""
"#"
"&& 1"
--
2.53.0
