Merge pull request #4736 from XiWeiGu/loongarch_issue4728

LoongArch: Fixed issue 4728

kernel/loongarch64/scal.S

@@ -56,80 +56,6 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
    MTC  a1, $r0
    slli.d INCX, INCX, BASE_SHIFT
    bge $r0,    N, .L999
-   CMPEQ   $fcc0, ALPHA, a1
-   bceqz   $fcc0, .L50
-   srai.d I, N, 3
-   bne INCX, TEMP, .L20
-   bge $r0,    I, .L15
-   .align 3
-
-.L12:
-   ST a1,  X,   0 * SIZE
-   ST a1,  X,   1 * SIZE
-   ST a1,  X,   2 * SIZE
-   ST a1,  X,   3 * SIZE
-   ST a1,  X,   4 * SIZE
-   ST a1,  X,   5 * SIZE
-   ST a1,  X,   6 * SIZE
-   ST a1,  X,   7 * SIZE
-   addi.w  I, I, -1
-   addi.d X, X, 8 * SIZE
-   blt $r0,    I, .L12
-   .align 3
-
-.L15:
-   andi    I,  N, 7
-   bge $r0,    I, .L999
-   .align  3
-.L16:
-   ST a1,  X,   0 * SIZE
-   addi.d  I, I, -1
-   addi.d X, X, SIZE
-   blt $r0,    I, .L16
-   move $r4, $r17
-   fmov.d $f0, $f22
-   jirl    $r0, $r1, 0x0
-   .align 3
-
-.L20:
-   srai.d  I, N, 3
-   bge $r0,    I, .L25
-   .align 3
-
-.L22:
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   add.d   X, X, INCX
-   ST a1,  X,   0 * SIZE
-   addi.d  I, I, -1
-   add.d  X, X, INCX
-   blt $r0,    I, .L22
-   .align 3
-
-.L25:
-   andi    I,  N, 7
-   bge $r0,    I, .L999
-   .align  3
-.L26:
-   addi.d  I, I, -1
-   ST a1,  X,   0 * SIZE
-   add.d  X, X, INCX
-   blt $r0,    I, .L26
-   move $r4, $r17
-   fmov.d $f0, $f22
-   jirl    $r0, $r1, 0x0
-   .align 3
 
 .L50:
    srai.d I, N, 3

kernel/loongarch64/scal_lasx.S

@@ -58,12 +58,11 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     FFINT  a2, a2
     slli.d TEMP, TEMP, BASE_SHIFT
     slli.d INCX, INCX, BASE_SHIFT
-    CMPEQ  $fcc0, ALPHA, a1
-    bcnez  $fcc0, .L20   //ALPHA==0
     CMPEQ  $fcc0, ALPHA, a2
     bcnez $fcc0, .L999  //ALPHA==1 return
+
     srai.d I, N, 3
-    beq INCX, TEMP, .L30    //ALPHA！=0|1 and INCX==1
+    beq INCX, TEMP, .L30    //ALPHA!=1 and INCX==1
     MTG  TEMP, ALPHA
 #ifdef DOUBLE
     xvreplgr2vr.d VALPHA, TEMP
@@ -73,7 +72,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     move XX, X
     .align 3
 
-.L10:   //ALPHA！=0|1 and INCX!=1
+.L10:   //ALPHA!=1 and INCX!=1
     bge $r0, I, .L32
     .align 3
 .L11:
@@ -166,74 +165,6 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     blt $r0, I, .L11
     b .L32
     .align 3
-
-.L20:
-    srai.d I, N, 3
-    beq INCX, TEMP, .L24
-    bge $r0, I, .L22
-    .align 3
-
-.L21:
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    addi.d  I, I, -1
-    blt $r0, I, .L21
-    .align 3
-
-.L22:
-    andi I, N, 7
-    bge $r0, I, .L999
-    .align 3
-
-.L23:
-    ST  a1, X, 0 * SIZE
-    addi.d I, I, -1
-    add.d X, X, INCX
-    blt $r0, I, .L23
-    jirl $r0, $r1, 0
-    .align 3
-
-.L24:
-    bge $r0, I, .L26 /*N<8 INCX==1*/
-    .align 3
-.L25:
-    xvxor.v VX0, VX0, VX0
-    xvst VX0, X, 0 * SIZE
-#ifdef DOUBLE
-    xvst VX0, X, 4 * SIZE
-#endif
-    addi.d I, I, -1
-    addi.d X, X, 8 * SIZE
-    blt $r0, I, .L25
-    .align 3
-
-.L26:
-    andi I, N, 7
-    bge $r0, I, .L999
-    .align 3
-
-.L27:
-    ST  a1, X, 0 * SIZE
-    addi.d I, I, -1
-    addi.d X, X, SIZE
-    blt $r0, I, .L27
-    jirl $r0, $r1, 0
-    .align 3
-
 .L30:
     bge $r0, I, .L32/*N<8 INCX==1*/
     MTG TEMP, ALPHA

kernel/loongarch64/scal_lsx.S

@@ -58,12 +58,10 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     FFINT  a2, a2
     slli.d TEMP, TEMP, BASE_SHIFT
     slli.d INCX, INCX, BASE_SHIFT
-    CMPEQ  $fcc0, ALPHA, a1
-    bcnez  $fcc0, .L20   //ALPHA==0
     CMPEQ  $fcc0, ALPHA, a2
     bcnez $fcc0, .L999  //ALPHA==1 return
     srai.d I, N, 3
-    beq INCX, TEMP, .L30    //ALPHA！=0|1 and INCX==1
+    beq INCX, TEMP, .L30    //ALPHA!=1 and INCX==1
     MTG  TEMP, ALPHA
 #ifdef DOUBLE
     vreplgr2vr.d VALPHA, TEMP
@@ -73,7 +71,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     move XX, X
     .align 3
 
-.L10:   //ALPHA！=0|1 and INCX!=1
+.L10:   //ALPHA!=1 and INCX!=1
     bge $r0, I, .L32
     .align 3
 
@@ -171,78 +169,6 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
     b .L32
     .align 3
 
-.L20:
-    srai.d I, N, 3
-    beq INCX, TEMP, .L24
-    bge $r0, I, .L22
-    .align 3
-
-.L21:
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    ST a1, X, 0
-    add.d X, X, INCX
-    addi.d  I, I, -1
-    blt $r0, I, .L21
-    .align 3
-
-.L22:
-    andi I, N, 7
-    bge $r0, I, .L999
-    .align 3
-
-.L23:
-    ST  a1, X, 0 * SIZE
-    addi.d I, I, -1
-    add.d X, X, INCX
-    blt $r0, I, .L23
-    jirl $r0, $r1, 0
-    .align 3
-
-.L24:
-    bge $r0, I, .L26 /*N<8 INCX==1*/
-    .align 3
-
-.L25:
-    vxor.v VX0, VX0, VX0
-    vst VX0, X, 0 * SIZE
-#ifdef DOUBLE
-    vst VX0, X, 2 * SIZE
-    vst VX0, X, 4 * SIZE
-    vst VX0, X, 6 * SIZE
-#else
-    vst VX0, X, 4 * SIZE
-#endif
-    addi.d I, I, -1
-    addi.d X, X, 8 * SIZE
-    blt $r0, I, .L25
-    .align 3
-
-.L26:
-    andi I, N, 7
-    bge $r0, I, .L999
-    .align 3
-
-.L27:
-    ST  a1, X, 0 * SIZE
-    addi.d I, I, -1
-    addi.d X, X, SIZE
-    blt $r0, I, .L27
-    jirl $r0, $r1, 0
-    .align 3
-
 .L30:
     bge $r0, I, .L32/*N<8 INCX==1*/
     MTG  TEMP, ALPHA