ARM64: Fix fdlibm compilation flags.android-cts-5.1_r9android-cts-5.1_r8android-cts-5.1_r7android-cts-5.1_r6android-cts-5.1_r5android-cts-5.1_r4android-cts-5.1_r3android-cts-5.1_r28android-cts-5.1_r27android-cts-5.1_r26android-cts-5.1_r25android-cts-5.1_r24android-cts-5.1_r23android-cts-5.1_r22android-cts-5.1_r21android-cts-5.1_r20android-cts-5.1_r2android-cts-5.1_r19android-cts-5.1_r18android-cts-5.1_r17android-cts-5.1_r16android-cts-5.1_r15android-cts-5.1_r14android-cts-5.1_r13android-cts-5.1_r10android-cts-5.1_r1android-cts-5.0_r9android-cts-5.0_r8android-cts-5.0_r7android-cts-5.0_r6android-cts-5.0_r5android-cts-5.0_r4android-cts-5.0_r3android-5.1.1_r9android-5.1.1_r8android-5.1.1_r7android-5.1.1_r6android-5.1.1_r5android-5.1.1_r4android-5.1.1_r38android-5.1.1_r37android-5.1.1_r36android-5.1.1_r35android-5.1.1_r34android-5.1.1_r33android-5.1.1_r30android-5.1.1_r3android-5.1.1_r29android-5.1.1_r28android-5.1.1_r26android-5.1.1_r25android-5.1.1_r24android-5.1.1_r23android-5.1.1_r22android-5.1.1_r20android-5.1.1_r2android-5.1.1_r19android-5.1.1_r18android-5.1.1_r17android-5.1.1_r16android-5.1.1_r15android-5.1.1_r14android-5.1.1_r13android-5.1.1_r12android-5.1.1_r10android-5.1.1_r1android-5.1.0_r5android-5.1.0_r4android-5.1.0_r3android-5.1.0_r1android-5.0.2_r3android-5.0.2_r1android-5.0.1_r1android-5.0.0_r7android-5.0.0_r6android-5.0.0_r5.1android-5.0.0_r5android-5.0.0_r4android-5.0.0_r3android-5.0.0_r2android-5.0.0_r1lollipop-releaselollipop-mr1-wfc-releaselollipop-mr1-releaselollipop-mr1-fi-releaselollipop-mr1-devlollipop-mr1-cts-releaselollipop-devlollipop-cts-release

Currently fdlibm is compiled using the default C standard (gnu99).
This specifies a relaxed floating point model that enables fused
operations such as fused multiply-add (-ffp-contract=fast).

Since fdlibm is used to compute StrictMath operations this is too
permissive on targets that have support for fused operations such as
ARM64.

bug: 17663117
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>

(cherry picked from commit 0eb593375b6e890846eedbf08ee6901e1d73beba)

Change-Id: I713acb6b273787ae70ce1d664bea012b0a347ccb

1 files changed, 3 insertions, 0 deletions

diff --git a/Android.mk b/Android.mk
index 734bfae..5cf8648 100644
--- a/Android.mk
+++ b/Android.mk
@@ -54,6 +54,9 @@ cflags += "-D__LITTLE_ENDIAN"
 cflags += "-fno-strict-aliasing"
 cflags += "-ffloat-store"
 
+# c99 specifies a less relaxed floating point model that does not enable
+# floating point expession contraction (e.g: fused multiply-add operations).
+cflags += "-std=c99"
 
 #
 # Build for the target (device).