Add accessor functions for Neon_Overflow, such that its value can be

read/changed when compiling with GCC/ARM too.

26 files changed, 48 insertions, 34 deletions

diff --git a/README b/README
index b926924..bbb0d83 100644
--- a/README
+++ b/README
@@ -23,13 +23,12 @@ $ make check # to compare the results with the expected output
 
 Known issues:
 -------------
-The tests currently fail to build with GCC/ARM:
-- no support for Neon_Overflow/fpsrc register
-- ICE when compiling ref_vldX.c, ref_vldX_lane.c, ref_vstX_lane.c
-- fails to compile vst1_lane.c
+Some tests currently fail to build with GCC/ARM:
 - missing include files: dspfns.h, armdsp.h
 
-VS[LR]I.64 tests are disabled because QEmu aborts.
+As GCC/ARM provides no support for the Neon_Overflow/fpsrc register,
+auxiliary accessor functions have been implemented in
+stm-arm-neon-ref.h.
 
 Engineering:
 ------------
diff --git a/ref_v_binary_sat_op.c b/ref_v_binary_sat_op.c
index cafbcb7..921573e 100644
--- a/ref_v_binary_sat_op.c
+++ b/ref_v_binary_sat_op.c
@@ -39,8 +39,9 @@ THE SOFTWARE.
 FNNAME (INSN_NAME)
 {
   /* vector_res = OP(vector1,vector2), then store the result.  */
+
 #define TEST_BINARY_SAT_OP1(INSN, Q, T1, T2, W, N)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_##T2##W(VECT_VAR(vector1, T1, W, N),	\
 		      VECT_VAR(vector2, T1, W, N));	\
diff --git a/ref_v_unary_sat_op.c b/ref_v_unary_sat_op.c
index bd110fb..16c65ee 100644
--- a/ref_v_unary_sat_op.c
+++ b/ref_v_unary_sat_op.c
@@ -40,7 +40,7 @@ FNNAME (INSN_NAME)
 {
   /* Basic test: y=OP(x), then store the result.  */
 #define TEST_UNARY_SAT_OP1(INSN, Q, T1, T2, W, N)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_##T2##W(VECT_VAR(vector, T1, W, N));	\
   vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N),		\
diff --git a/ref_vqdmlal.c b/ref_vqdmlal.c
index fe6dd57..d2f229d 100644
--- a/ref_vqdmlal.c
+++ b/ref_vqdmlal.c
@@ -45,7 +45,7 @@ FNNAME (INSN_NAME)
   /* vector_res = OP(vector, vector3, vector4),
      then store the result.  */
 #define TEST_VQDMLXL1(INSN, T1, T2, W, W2, N)		\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##_##T2##W2(VECT_VAR(vector, T1, W, N),		\
 		    VECT_VAR(vector3, T1, W2, N),	\
diff --git a/ref_vqdmlal_lane.c b/ref_vqdmlal_lane.c
index 6f1064c..456473c 100644
--- a/ref_vqdmlal_lane.c
+++ b/ref_vqdmlal_lane.c
@@ -44,7 +44,7 @@ FNNAME (INSN_NAME)
   /* vector_res = vqdmlxl_lane(vector, vector3, vector4, lane),
      then store the result.  */
 #define TEST_VQDMLXL_LANE1(INSN, T1, T2, W, W2, N, V)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##_##T2##W2(VECT_VAR(vector, T1, W, N),		\
 		    VECT_VAR(vector3, T1, W2, N),	\
diff --git a/ref_vqdmlal_n.c b/ref_vqdmlal_n.c
index 5aba435..3f1a08b 100644
--- a/ref_vqdmlal_n.c
+++ b/ref_vqdmlal_n.c
@@ -44,7 +44,7 @@ FNNAME (INSN_NAME)
   /* vector_res = vqdmlxl_n(vector, vector3, val),
      then store the result.  */
 #define TEST_VQDMLXL_N1(INSN, T1, T2, W, W2, N, V)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##_##T2##W2(VECT_VAR(vector, T1, W, N),		\
 		    VECT_VAR(vector3, T1, W2, N),	\
diff --git a/ref_vqdmulh.c b/ref_vqdmulh.c
index 688ffba..479b780 100644
--- a/ref_vqdmulh.c
+++ b/ref_vqdmulh.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* vector_res = vqdmulh(vector,vector2,lane), then store the result.  */
 #define TEST_VQDMULH2(INSN, Q, T1, T2, W, N)		\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_##T2##W(VECT_VAR(vector, T1, W, N),	\
 		      VECT_VAR(vector2, T1, W, N));	\
diff --git a/ref_vqdmulh_lane.c b/ref_vqdmulh_lane.c
index 8b9f635..60b98af 100644
--- a/ref_vqdmulh_lane.c
+++ b/ref_vqdmulh_lane.c
@@ -40,7 +40,7 @@ FNNAME (INSN)
 {
   /* vector_res = vqdmulh_lane(vector,vector2,lane), then store the result.  */
 #define TEST_VQDMULH_LANE2(INSN, Q, T1, T2, W, N, N2, L)	\
-  Neon_Overflow = 0;						\
+  Set_Neon_Overflow(0);						\
   VECT_VAR(vector_res, T1, W, N) =				\
     INSN##Q##_lane_##T2##W(VECT_VAR(vector, T1, W, N),		\
 			   VECT_VAR(vector2, T1, W, N2),	\
diff --git a/ref_vqdmulh_n.c b/ref_vqdmulh_n.c
index ae9d24e..ca78483 100644
--- a/ref_vqdmulh_n.c
+++ b/ref_vqdmulh_n.c
@@ -42,7 +42,7 @@ FNNAME (INSN)
 
   /* vector_res = vqdmulh_n(vector,val), then store the result.  */
 #define TEST_VQDMULH_N2(INSN, Q, T1, T2, W, N, L)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_n_##T2##W(VECT_VAR(vector, T1, W, N),	\
 			L);				\
diff --git a/ref_vqdmull.c b/ref_vqdmull.c
index 35bf6f7..ba98f9c 100644
--- a/ref_vqdmull.c
+++ b/ref_vqdmull.c
@@ -40,7 +40,7 @@ FNNAME (INSN)
 {
   /* Basic test: y=vqdmull(x,x), then store the result.  */
 #define TEST_VQDMULL2(INSN, T1, T2, W, W2, N)		\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W2, N) =			\
     INSN##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		   VECT_VAR(vector2, T1, W, N));	\
diff --git a/ref_vqdmull_lane.c b/ref_vqdmull_lane.c
index 504b61b..6d4fb14 100644
--- a/ref_vqdmull_lane.c
+++ b/ref_vqdmull_lane.c
@@ -42,7 +42,7 @@ FNNAME (INSN)
 
   /* vector_res = vqdmull_lane(vector,vector2,lane), then store the result.  */
 #define TEST_VQDMULL_LANE2(INSN, T1, T2, W, W2, N, L)			\
-  Neon_Overflow = 0;							\
+  Set_Neon_Overflow(0);							\
   VECT_VAR(vector_res, T1, W2, N) =					\
     INSN##_lane_##T2##W(VECT_VAR(vector, T1, W, N),			\
 			VECT_VAR(vector2, T1, W, N),			\
diff --git a/ref_vqdmull_n.c b/ref_vqdmull_n.c
index 28557f4..a4465e2 100644
--- a/ref_vqdmull_n.c
+++ b/ref_vqdmull_n.c
@@ -42,7 +42,7 @@ FNNAME (INSN)
 
   /* vector_res = vqdmull_n(vector,val), then store the result.  */
 #define TEST_VQDMULL_N2(INSN, T1, T2, W, W2, N, L)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W2, N) =			\
     INSN##_n_##T2##W(VECT_VAR(vector, T1, W, N),	\
 		     L);				\
diff --git a/ref_vqmovn.c b/ref_vqmovn.c
index 9b91fe6..8b523f0 100644
--- a/ref_vqmovn.c
+++ b/ref_vqmovn.c
@@ -41,7 +41,7 @@ FNNAME (INSN_NAME)
 {
   /* Basic test: y=OP(x), then store the result.  */
 #define TEST_UNARY_OP1(INSN, T1, T2, W, W2, N)				\
-  Neon_Overflow = 0;							\
+  Set_Neon_Overflow(0);							\
   VECT_VAR(vector_res, T1, W, N) =					\
     INSN##_##T2##W2(VECT_VAR(vector, T1, W2, N));			\
   vst1##_##T2##W(VECT_VAR(result, T1, W, N),				\
diff --git a/ref_vqmovun.c b/ref_vqmovun.c
index 199cdb4..6b2547c 100644
--- a/ref_vqmovun.c
+++ b/ref_vqmovun.c
@@ -41,7 +41,7 @@ FNNAME (INSN_NAME)
 {
   /* Basic test: y=OP(x), then store the result.  */
 #define TEST_UNARY_OP1(INSN, T1, T2, W, W2, N)				\
-  Neon_Overflow = 0;							\
+  Set_Neon_Overflow(0);							\
   VECT_VAR(vector_res, T1, W, N) =					\
     INSN##_s##W2(VECT_VAR(vector, int, W2, N));				\
   vst1##_##T2##W(VECT_VAR(result, T1, W, N),				\
diff --git a/ref_vqrdmulh.c b/ref_vqrdmulh.c
index 19a7bf3..d62c87b 100644
--- a/ref_vqrdmulh.c
+++ b/ref_vqrdmulh.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* vector_res = vqrdmulh(vector,vector2), then store the result.  */
 #define TEST_VQRDMULH2(INSN, Q, T1, T2, W, N)		\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_##T2##W(VECT_VAR(vector, T1, W, N),	\
 		      VECT_VAR(vector2, T1, W, N));	\
diff --git a/ref_vqrdmulh_lane.c b/ref_vqrdmulh_lane.c
index 85ce55b..d5e225f 100644
--- a/ref_vqrdmulh_lane.c
+++ b/ref_vqrdmulh_lane.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* vector_res = vqrdmulh_lane(vector,vector2,lane), then store the result.  */
 #define TEST_VQRDMULH_LANE2(INSN, Q, T1, T2, W, N, N2, L)	\
-  Neon_Overflow = 0;						\
+  Set_Neon_Overflow(0);						\
   VECT_VAR(vector_res, T1, W, N) =				\
     INSN##Q##_lane_##T2##W(VECT_VAR(vector, T1, W, N),		\
 			   VECT_VAR(vector2, T1, W, N2),	\
diff --git a/ref_vqrdmulh_n.c b/ref_vqrdmulh_n.c
index ac6255b..4822b21 100644
--- a/ref_vqrdmulh_n.c
+++ b/ref_vqrdmulh_n.c
@@ -43,7 +43,7 @@ FNNAME (INSN)
 
   /* vector_res = vqrdmulh_n(vector,val), then store the result.  */
 #define TEST_VQRDMULH_N2(INSN, Q, T1, T2, W, N, L)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_n_##T2##W(VECT_VAR(vector, T1, W, N),	\
 			L);				\
diff --git a/ref_vqrshl.c b/ref_vqrshl.c
index 975d7aa..c91a31b 100644
--- a/ref_vqrshl.c
+++ b/ref_vqrshl.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: v3=vqrshl(v1,v2), then store the result.  */
 #define TEST_VQRSHL2(INSN, T3, Q, T1, T2, W, N)			\
-  Neon_Overflow = 0;						\
+  Set_Neon_Overflow(0);						\
   VECT_VAR(vector_res, T1, W, N) =				\
     INSN##Q##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		      VECT_VAR(vector_shift, T3, W, N));	\
diff --git a/ref_vqrshrn_n.c b/ref_vqrshrn_n.c
index 0d6a295..21765e5 100644
--- a/ref_vqrshrn_n.c
+++ b/ref_vqrshrn_n.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: y=vqrshrn_n(x,v), then store the result.  */
 #define TEST_VQRSHRN_N2(INSN, T1, T2, W, W2, N, V)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W2, N) =			\
     INSN##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		   V);					\
diff --git a/ref_vqrshrun_n.c b/ref_vqrshrun_n.c
index 785c487..a80b396 100644
--- a/ref_vqrshrun_n.c
+++ b/ref_vqrshrun_n.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: y=vqrshrun_n(x,v), then store the result.  */
 #define TEST_VQRSHRUN_N2(INSN, T1, T2, W, W2, N, V)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, uint, W2, N) =			\
     INSN##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		   V);					\
diff --git a/ref_vqshl.c b/ref_vqshl.c
index d843755..277c2f4 100644
--- a/ref_vqshl.c
+++ b/ref_vqshl.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: v3=vqshl(v1,v2), then store the result.  */
 #define TEST_VQSHL2(INSN, T3, Q, T1, T2, W, N)			\
-  Neon_Overflow = 0;						\
+  Set_Neon_Overflow(0);						\
   VECT_VAR(vector_res, T1, W, N) =				\
     INSN##Q##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		      VECT_VAR(vector_shift, T3, W, N));	\
diff --git a/ref_vqshl_n.c b/ref_vqshl_n.c
index 74ec3c6..af25203 100644
--- a/ref_vqshl_n.c
+++ b/ref_vqshl_n.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: v2=vqshl_n(v1,v), then store the result.  */
 #define TEST_VQSHL_N2(INSN, Q, T1, T2, W, N, V)		\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W, N) =			\
     INSN##Q##_n_##T2##W(VECT_VAR(vector, T1, W, N),	\
 			V);				\
diff --git a/ref_vqshlu_n.c b/ref_vqshlu_n.c
index 5090708..3437e0f 100644
--- a/ref_vqshlu_n.c
+++ b/ref_vqshlu_n.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: v2=vqshlu_n(v1,v), then store the result.  */
 #define TEST_VQSHLU_N2(INSN, Q, T1, T2, T3, T4, W, N, V)	\
-  Neon_Overflow = 0;						\
+  Set_Neon_Overflow(0);						\
   VECT_VAR(vector_res, T3, W, N) =				\
     INSN##Q##_n_##T2##W(VECT_VAR(vector, T1, W, N),		\
 			V);					\
diff --git a/ref_vqshrn_n.c b/ref_vqshrn_n.c
index 6b6f047..f464563 100644
--- a/ref_vqshrn_n.c
+++ b/ref_vqshrn_n.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: y=vqshrn_n(x,v), then store the result.  */
 #define TEST_VQSHRN_N2(INSN, T1, T2, W, W2, N, V)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, T1, W2, N) =			\
     INSN##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		   V);					\
diff --git a/ref_vqshrun_n.c b/ref_vqshrun_n.c
index 4dc3bd6..cf9d4a5 100644
--- a/ref_vqshrun_n.c
+++ b/ref_vqshrun_n.c
@@ -41,7 +41,7 @@ FNNAME (INSN)
 {
   /* Basic test: y=vqshrun_n(x,v), then store the result.  */
 #define TEST_VQSHRUN_N2(INSN, T1, T2, W, W2, N, V)	\
-  Neon_Overflow = 0;					\
+  Set_Neon_Overflow(0);					\
   VECT_VAR(vector_res, uint, W2, N) =			\
     INSN##_##T2##W(VECT_VAR(vector, T1, W, N),		\
 		   V);					\
diff --git a/stm-arm-neon-ref.h b/stm-arm-neon-ref.h
index e4d17d3..7d09a38 100644
--- a/stm-arm-neon-ref.h
+++ b/stm-arm-neon-ref.h
@@ -238,7 +238,7 @@ static void dump_results_hex (const char *test_name)
 typedef union {
   struct {
     int _xxx:27;
-    int QC:1;
+    unsigned int QC:1;
     int V:1;
     int C:1;
     int Z:1;
@@ -255,7 +255,7 @@ typedef union {
     int Z:1;
     int C:1;
     int V:1;
-    int QC:1;
+    unsigned int QC:1;
     int _dnm:27;
   } b;
   unsigned int word;
@@ -266,10 +266,24 @@ typedef union {
 #ifdef __ARMCC_VERSION
 register _ARM_FPSCR _afpscr_for_qc __asm("fpscr");
 #define Neon_Overflow _afpscr_for_qc.b.QC
+#define Set_Neon_Overflow(x)  {Neon_Overflow = (x);}
 #else
-/* Fake declaration because GCC/ARM does not know this register */
- int myerrno;
-#define Neon_Overflow myerrno
+/* GCC/ARM does not know this register */
+#define Neon_Overflow  __read_neon_overflow()
+static int __read_neon_overflow() {
+    _ARM_FPSCR _afpscr_for_qc;
+    asm("vmrs %0,fpscr" : "=r" (_afpscr_for_qc));
+    return _afpscr_for_qc.b.QC;
+}
+#define Set_Neon_Overflow(x)  __set_neon_overflow((x))
+static void __set_neon_overflow(int x) {
+    _ARM_FPSCR _afpscr_for_qc;
+    asm("vmrs %0,fpscr" : "=r" (_afpscr_for_qc));
+    _afpscr_for_qc.b.QC = x;
+    asm("vmsr fpscr,%0" : : "r" (_afpscr_for_qc));
+    return;
+}
+
 #endif
 
 #endif /* STM_ARM_NEON_MODELS */