s390x: replace nop with "nop 0" in inline assembly

... as a bandaid for building with clang until LLVM's internal assembler
supports nops without operand.

Signed-off-by: Marius Hillenbrand <mhillen@linux.ibm.com>

kernel/zarch/dgemv_n_4.c

@@ -169,7 +169,7 @@ static void dgemv_kernel_4x4(BLASLONG n, FLOAT **ap, FLOAT *x, FLOAT *y,
     "agfi   %%r1,32\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) y)
     : [y] "a"(y), "m"(*(const struct { FLOAT x[n]; } *) ap0),[ap0] "a"(ap0),
        "m"(*(const struct { FLOAT x[n]; } *) ap1),[ap1] "a"(ap1),
@@ -274,7 +274,7 @@ static void dgemv_kernel_4x2(BLASLONG n, FLOAT **ap, FLOAT *x, FLOAT *y,
     "agfi   %%r1,32\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) y)
     : [y] "a"(y), "m"(*(const struct { FLOAT x[n]; } *) ap0),[ap0] "a"(ap0),
        "m"(*(const struct { FLOAT x[n]; } *) ap1),[ap1] "a"(ap1),
@@ -351,7 +351,7 @@ static void dgemv_kernel_4x1(BLASLONG n, FLOAT *a0, FLOAT *x, FLOAT *y,
     "agfi   %%r1,32\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) y)
     : [y] "a"(y), "m"(*(const struct { FLOAT x[n]; } *) a0),[a0] "a"(a0),
        "m"(*(const FLOAT (*)[1]) x),[x] "a"(x),[alpha] "Q"(*alpha),

kernel/zarch/dgemv_t_4.c

@@ -438,7 +438,7 @@ static void add_y_kernel_4(BLASLONG n, FLOAT da, FLOAT *src, FLOAT *dest) {
     "agfi   %%r1,32\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) dest)
     : [dest] "a"(dest),[da] "Q"(da), "m"(*(const struct { FLOAT x[n]; } *) src),
        [src] "a"(src),[n] "r"(n)

kernel/zarch/icamax.c

@@ -213,7 +213,7 @@ static BLASLONG icamax_kernel_32(BLASLONG n, FLOAT *x, FLOAT *amax) {
     "ste    %%f0,%[amax]\n\t"
     "vlgvg  %[iamax],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamax] "=r"(iamax),[amax] "=Q"(*amax),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n * 2]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8",

kernel/zarch/icamin.c

@@ -213,7 +213,7 @@ static BLASLONG icamin_kernel_32(BLASLONG n, FLOAT *x, FLOAT *amin) {
     "ste    %%f0,%[amin]\n\t"
     "vlgvg  %[iamin],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamin] "=r"(iamin),[amin] "=Q"(*amin),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n * 2]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v8",

kernel/zarch/idamax.c

@@ -160,7 +160,7 @@ static BLASLONG idamax_kernel_32(BLASLONG n, FLOAT *x, FLOAT *amax) {
     "std    %%f0,%[amax]\n\t"
     "vlgvg  %[iamax],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamax] "=r"(iamax),[amax] "=Q"(*amax),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v16",

kernel/zarch/idamin.c

@@ -160,7 +160,7 @@ static BLASLONG idamin_kernel_32(BLASLONG n, FLOAT *x, FLOAT *amin) {
     "std    %%f0,%[amin]\n\t"
     "vlgvg  %[iamin],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamin] "=r"(iamin),[amin] "=Q"(*amin),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v16",

kernel/zarch/idmax.c

@@ -140,7 +140,7 @@ static BLASLONG idmax_kernel_32(BLASLONG n, FLOAT *x, FLOAT *max) {
     "std    %%f0,%[max]\n\t"
     "vlgvg  %[imax],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [imax] "=r"(imax),[max] "=Q"(*max),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v16",

kernel/zarch/idmin.c

@@ -140,7 +140,7 @@ static BLASLONG idmin_kernel_32(BLASLONG n, FLOAT *x, FLOAT *min) {
     "std    %%f0,%[min]\n\t"
     "vlgvg  %[imin],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [imin] "=r"(imin),[min] "=Q"(*min),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v6", "v7", "v16",

kernel/zarch/isamax.c

@@ -204,7 +204,7 @@ static BLASLONG isamax_kernel_64(BLASLONG n, FLOAT *x, FLOAT *amax) {
     "ste    %%f0,%[amax]\n\t"
     "vlgvg  %[iamax],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamax] "=r"(iamax),[amax] "=Q"(*amax),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v4", "v5", "v6", "v7", "v8", "v16",

kernel/zarch/isamin.c

@@ -204,7 +204,7 @@ static BLASLONG isamin_kernel_64(BLASLONG n, FLOAT *x, FLOAT *amin) {
     "ste    %%f0,%[amin]\n\t"
     "vlgvg  %[iamin],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamin] "=r"(iamin),[amin] "=Q"(*amin),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v4", "v5", "v6", "v7", "v8", "v16",

kernel/zarch/ismax.c

@@ -184,7 +184,7 @@ static BLASLONG ismax_kernel_64(BLASLONG n, FLOAT *x, FLOAT *max) {
     "ste    %%f0,%[max]\n\t"
     "vlgvg  %[imax],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [imax] "=r"(imax),[max] "=Q"(*max),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v4", "v5", "v6", "v7", "v8", "v16",

kernel/zarch/ismin.c

@@ -184,7 +184,7 @@ static BLASLONG ismin_kernel_64(BLASLONG n, FLOAT *x, FLOAT *min) {
     "ste    %%f0,%[min]\n\t"
     "vlgvg  %[imin],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [imin] "=r"(imin),[min] "=Q"(*min),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v4", "v5", "v6", "v7", "v8", "v16",

kernel/zarch/izamax.c

@@ -157,7 +157,7 @@ static BLASLONG izamax_kernel_16(BLASLONG n, FLOAT *x, FLOAT *amax) {
     "std    %%f0,%[amax]\n\t"
     "vlgvg  %[iamax],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamax] "=r"(iamax),[amax] "=Q"(*amax),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n * 2]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v16", "v17", "v18",

kernel/zarch/izamin.c

@@ -157,7 +157,7 @@ static BLASLONG izamin_kernel_16(BLASLONG n, FLOAT *x, FLOAT *amin) {
     "std    %%f0,%[amin]\n\t"
     "vlgvg  %[iamin],%%v1,0\n\t"
     "2:\n\t"
-    "nop"
+    "nop 0"
     : [iamin] "=r"(iamin),[amin] "=Q"(*amin),[n] "+&r"(n)
     : "m"(*(const struct { FLOAT x[n * 2]; } *) x),[x] "a"(x)
     : "cc", "r1", "v0", "v1", "v2", "v3", "v4", "v5", "v16", "v17", "v18",

kernel/zarch/sgemv_n_4.c

@@ -159,7 +159,7 @@ static void sgemv_kernel_4x4(BLASLONG n, FLOAT **ap, FLOAT *x, FLOAT *y,
     "agfi   %%r1,16\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) y)
     : [y] "a"(y), "m"(*(const struct { FLOAT x[n]; } *) ap0),[ap0] "a"(ap0),
        "m"(*(const struct { FLOAT x[n]; } *) ap1),[ap1] "a"(ap1),
@@ -258,7 +258,7 @@ static void sgemv_kernel_4x2(BLASLONG n, FLOAT **ap, FLOAT *x, FLOAT *y,
     "agfi   %%r1,16\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) y)
     : [y] "a"(y), "m"(*(const struct { FLOAT x[n]; } *) ap0),[ap0] "a"(ap0),
        "m"(*(const struct { FLOAT x[n]; } *) ap1),[ap1] "a"(ap1),
@@ -331,7 +331,7 @@ static void sgemv_kernel_4x1(BLASLONG n, FLOAT *a0, FLOAT *x, FLOAT *y,
     "agfi   %%r1,16\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) y)
     : [y] "a"(y), "m"(*(const struct { FLOAT x[n]; } *) a0),[a0] "a"(a0),
        "m"(*(const FLOAT (*)[1]) x),[x] "a"(x),[alpha] "Q"(*alpha),

kernel/zarch/sgemv_t_4.c

@@ -431,7 +431,7 @@ static void add_y_kernel_4(BLASLONG n, FLOAT da, FLOAT *src, FLOAT *dest) {
     "agfi   %%r1,16\n\t"
     "brctg  %%r0,2b\n\t"
     "3:\n\t"
-    "nop"
+    "nop 0"
     : "+m"(*(struct { FLOAT x[n]; } *) dest)
     : [dest] "a"(dest),[da] "Q"(da), "m"(*(const struct { FLOAT x[n]; } *) src),
        [src] "a"(src),[n] "r"(n)