tcg/sparc: Use the proper compilation flags for 32-bit
authorRichard Henderson <rth@twiddle.net>
Wed, 26 Apr 2017 17:39:08 +0000 (10:39 -0700)
committerRichard Henderson <rth@twiddle.net>
Mon, 5 Jun 2017 16:25:42 +0000 (09:25 -0700)
We have required a v9 cpu since 9b9c37c36439ee0452632253dac7a31897f27f70.
However, the flags we were using did not reliably enable v8plus, which
meant that the compiler didn't know it could inline 64-bit atomics.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
configure

index fbb6a93c9928d66ccbec1a1d2c2595048aa8403f..13e040d28c1e1830db50837afc39c91bd44b65d3 100755 (executable)
--- a/configure
+++ b/configure
@@ -1213,12 +1213,12 @@ case "$cpu" in
            LDFLAGS="-m64 $LDFLAGS"
            ;;
     sparc)
-           LDFLAGS="-m32 $LDFLAGS"
-           CPU_CFLAGS="-m32 -mcpu=ultrasparc"
+           CPU_CFLAGS="-m32 -mv8plus -mcpu=ultrasparc"
+           LDFLAGS="-m32 -mv8plus $LDFLAGS"
            ;;
     sparc64)
-           LDFLAGS="-m64 $LDFLAGS"
            CPU_CFLAGS="-m64 -mcpu=ultrasparc"
+           LDFLAGS="-m64 $LDFLAGS"
            ;;
     s390)
            CPU_CFLAGS="-m31"