| Message ID | 582D77CB.9090606@foss.arm.com |
|---|---|
| State | New |
| Headers | show |
On Thu, Nov 17, 2016 at 09:26:35AM +0000, Kyrill Tkachov wrote: > Hi all, > > Testing with RTL checking enabled showed another failure in gcc.dg/torture/float32-builtin.c. > aarch64_float_const_zero_rtx_p was passed down a complex subreg expression during combine/recog. > The pattern was supposed to reject that expression but it made its way to aarch64_float_const_zero_rtx_p > which is only supposed to handle CONST_DOUBLEs. > > The culprit is the aarch64_reg_or_fp_zero predicate that restricts the accepted codes to reg,subreg,const_double but > passes down anything that is not a reg into aarch64_float_const_zero_rtx_p. That includes SUBREGs that do not > match register_operand, which we want to reject. > This patch checks that the argument is a CONST_DOUBLE before passing it to aarch64_float_const_zero_rtx_p > > > Bootstrapped and tested on aarch64-none-linux-gnu with RTL checking enabled. > > Ok for trunk? OK. Thanks, James > Thanks, > Kyrill > > P.S. This predicate is only used in the store-pair family of patterns. > > 2016-11-17 Kyrylo Tkachov <kyrylo.tkachov@arm.com> > > * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): Check for > const_double code before calling aarch64_float_const_zero_rtx_p.
commit 950f8638a894aa699d1743fc3277b0aee3009fb7
Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com>
Date: Tue Nov 15 16:04:39 2016 +0000
[AArch64] Fix gcc.dg/torture/float32-builtin.c with RTL checking
diff --git a/gcc/config/aarch64/predicates.md b/gcc/config/aarch64/predicates.md
index ebda6d8..0671cc8 100644
--- a/gcc/config/aarch64/predicates.md
+++ b/gcc/config/aarch64/predicates.md
@@ -54,9 +54,9 @@ (define_predicate "aarch64_reg_or_zero"
(match_test "op == const0_rtx"))))
(define_predicate "aarch64_reg_or_fp_zero"
- (and (match_code "reg,subreg,const_double")
- (ior (match_operand 0 "register_operand")
- (match_test "aarch64_float_const_zero_rtx_p (op)"))))
+ (ior (match_operand 0 "register_operand")
+ (and (match_code "const_double")
+ (match_test "aarch64_float_const_zero_rtx_p (op)"))))
(define_predicate "aarch64_reg_zero_or_m1_or_1"
(and (match_code "reg,subreg,const_int")
Hi all, Testing with RTL checking enabled showed another failure in gcc.dg/torture/float32-builtin.c. aarch64_float_const_zero_rtx_p was passed down a complex subreg expression during combine/recog. The pattern was supposed to reject that expression but it made its way to aarch64_float_const_zero_rtx_p which is only supposed to handle CONST_DOUBLEs. The culprit is the aarch64_reg_or_fp_zero predicate that restricts the accepted codes to reg,subreg,const_double but passes down anything that is not a reg into aarch64_float_const_zero_rtx_p. That includes SUBREGs that do not match register_operand, which we want to reject. This patch checks that the argument is a CONST_DOUBLE before passing it to aarch64_float_const_zero_rtx_p Bootstrapped and tested on aarch64-none-linux-gnu with RTL checking enabled. Ok for trunk? Thanks, Kyrill P.S. This predicate is only used in the store-pair family of patterns. 2016-11-17 Kyrylo Tkachov <kyrylo.tkachov@arm.com> * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): Check for const_double code before calling aarch64_float_const_zero_rtx_p.