swscale/ppc: Move VSX-using code to its own file

Passes fate on LE (with "lavc/jrevdct: Avoid an aliasing violation" applied).

Signed-off-by: Lauri Kasanen <cand@gmx.com>
Tested-by: Michael Kostylev on BE
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

1 files changed, 10 insertions, 281 deletions

diff --git a/libswscale/ppc/swscale_altivec.c b/libswscale/ppc/swscale_altivec.c
index 8c6056d841..1d2b2faae0 100644
--- a/libswscale/ppc/swscale_altivec.c
+++ b/libswscale/ppc/swscale_altivec.c
@@ -31,21 +31,14 @@
 #include "yuv2rgb_altivec.h"
 #include "libavutil/ppc/util_altivec.h"
 
-#if HAVE_ALTIVEC
+#if HAVE_ALTIVEC && HAVE_BIGENDIAN
 #define vzero vec_splat_s32(0)
 
-#if HAVE_BIGENDIAN
 #define  GET_LS(a,b,c,s) {\
         vector signed short l2  = vec_ld(((b) << 1) + 16, s);\
         ls  = vec_perm(a, l2, c);\
         a = l2;\
     }
-#else
-#define  GET_LS(a,b,c,s) {\
-        ls  = a;\
-        a = vec_vsx_ld(((b) << 1)  + 16, s);\
-    }
-#endif
 
 #define yuv2planeX_8(d1, d2, l1, src, x, perm, filter) do {\
         vector signed short ls;\
@@ -59,7 +52,6 @@
         d2 = vec_add(d2, vf2);\
     } while (0)
 
-#if HAVE_BIGENDIAN
 #define LOAD_FILTER(vf,f) {\
         vector unsigned char perm0 = vec_lvsl(joffset, f);\
         vf = vec_ld(joffset, f);\
@@ -69,89 +61,7 @@
         p = vec_lvsl(xoffset, s);\
         ll1   = vec_ld(xoffset, s);\
 }
-#else
-#define LOAD_FILTER(vf,f) {\
-        vf = vec_vsx_ld(joffset, f);\
-}
-#define LOAD_L1(ll1,s,p){\
-        ll1  = vec_vsx_ld(xoffset, s);\
-}
-#endif
-
-static void yuv2planeX_16_altivec(const int16_t *filter, int filterSize,
-                                  const int16_t **src, uint8_t *dest,
-                                  const uint8_t *dither, int offset, int x)
-{
-    register int i, j;
-    LOCAL_ALIGNED(16, int, val, [16]);
-    vector signed int vo1, vo2, vo3, vo4;
-    vector unsigned short vs1, vs2;
-    vector unsigned char vf;
-    vector unsigned int altivec_vectorShiftInt19 =
-        vec_add(vec_splat_u32(10), vec_splat_u32(9));
-
-    for (i = 0; i < 16; i++)
-        val[i] = dither[(x + i + offset) & 7] << 12;
-
-    vo1 = vec_ld(0,  val);
-    vo2 = vec_ld(16, val);
-    vo3 = vec_ld(32, val);
-    vo4 = vec_ld(48, val);
-
-    for (j = 0; j < filterSize; j++) {
-        unsigned int joffset=j<<1;
-        unsigned int xoffset=x<<1;
-        vector unsigned char perm;
-        vector signed short l1,vLumFilter;
-        LOAD_FILTER(vLumFilter,filter);
-        vLumFilter = vec_splat(vLumFilter, 0);
-        LOAD_L1(l1,src[j],perm);
-        yuv2planeX_8(vo1, vo2, l1, src[j], x,     perm, vLumFilter);
-        yuv2planeX_8(vo3, vo4, l1, src[j], x + 8, perm, vLumFilter);
-    }
-
-    vo1 = vec_sra(vo1, altivec_vectorShiftInt19);
-    vo2 = vec_sra(vo2, altivec_vectorShiftInt19);
-    vo3 = vec_sra(vo3, altivec_vectorShiftInt19);
-    vo4 = vec_sra(vo4, altivec_vectorShiftInt19);
-    vs1 = vec_packsu(vo1, vo2);
-    vs2 = vec_packsu(vo3, vo4);
-    vf  = vec_packsu(vs1, vs2);
-    VEC_ST(vf, 0, dest);
-}
-
-
-static inline void yuv2planeX_u(const int16_t *filter, int filterSize,
-                                const int16_t **src, uint8_t *dest, int dstW,
-                                const uint8_t *dither, int offset, int x)
-{
-    int i, j;
-
-    for (i = x; i < dstW; i++) {
-        int t = dither[(i + offset) & 7] << 12;
-        for (j = 0; j < filterSize; j++)
-            t += src[j][i] * filter[j];
-        dest[i] = av_clip_uint8(t >> 19);
-    }
-}
-
-static void yuv2planeX_altivec(const int16_t *filter, int filterSize,
-                               const int16_t **src, uint8_t *dest, int dstW,
-                               const uint8_t *dither, int offset)
-{
-    int dst_u = -(uintptr_t)dest & 15;
-    int i;
-
-    yuv2planeX_u(filter, filterSize, src, dest, dst_u, dither, offset, 0);
-
-    for (i = dst_u; i < dstW - 15; i += 16)
-        yuv2planeX_16_altivec(filter, filterSize, src, dest + i, dither,
-                              offset, i);
-
-    yuv2planeX_u(filter, filterSize, src, dest, dstW, dither, offset, i);
-}
 
-#if HAVE_BIGENDIAN
 // The 3 above is 2 (filterSize == 4) + 1 (sizeof(short) == 2).
 
 // The neat trick: We only care for half the elements,
@@ -187,191 +97,12 @@ static void yuv2planeX_altivec(const int16_t *filter, int filterSize,
     vf1 = vec_ld((a * 2 * filterSize) + (b * 2) + 16 + off, f);\
     vf  = vec_perm(vf0, vf1, per);\
 }
-#else /* else of #if HAVE_BIGENDIAN */
-#define GET_VF4(a, vf, f) {\
-    vf = (vector signed short)vec_vsx_ld(a << 3, f);\
-    vf = vec_mergeh(vf, (vector signed short)vzero);\
-}
-#define FIRST_LOAD(sv, pos, s, per) {}
-#define UPDATE_PTR(s0, d0, s1, d1) {}
-#define LOAD_SRCV(pos, a, s, per, v0, v1, vf) {\
-    vf = vec_vsx_ld(pos + a, s);\
-}
-#define LOAD_SRCV8(pos, a, s, per, v0, v1, vf) LOAD_SRCV(pos, a, s, per, v0, v1, vf)
-#define GET_VFD(a, b, f, vf0, vf1, per, vf, off) {\
-    vf  = vec_vsx_ld((a * 2 * filterSize) + (b * 2) + off, f);\
-}
-#endif /* end of #if HAVE_BIGENDIAN */
 
-static void hScale_altivec_real(SwsContext *c, int16_t *dst, int dstW,
-                                const uint8_t *src, const int16_t *filter,
-                                const int32_t *filterPos, int filterSize)
-{
-    register int i;
-    LOCAL_ALIGNED(16, int, tempo, [4]);
+#define FUNC(name) name ## _altivec
+#include "swscale_ppc_template.c"
+#undef FUNC
 
-    if (filterSize % 4) {
-        for (i = 0; i < dstW; i++) {
-            register int j;
-            register int srcPos = filterPos[i];
-            register int val    = 0;
-            for (j = 0; j < filterSize; j++)
-                val += ((int)src[srcPos + j]) * filter[filterSize * i + j];
-            dst[i] = FFMIN(val >> 7, (1 << 15) - 1);
-        }
-    } else
-        switch (filterSize) {
-        case 4:
-            for (i = 0; i < dstW; i++) {
-                register int srcPos = filterPos[i];
-
-                vector unsigned char src_vF = unaligned_load(srcPos, src);
-                vector signed short src_v, filter_v;
-                vector signed int val_vEven, val_s;
-                src_v = // vec_unpackh sign-extends...
-                        (vector signed short)(VEC_MERGEH((vector unsigned char)vzero, src_vF));
-                // now put our elements in the even slots
-                src_v = vec_mergeh(src_v, (vector signed short)vzero);
-                GET_VF4(i, filter_v, filter);
-                val_vEven = vec_mule(src_v, filter_v);
-                val_s     = vec_sums(val_vEven, vzero);
-                vec_st(val_s, 0, tempo);
-                dst[i] = FFMIN(tempo[3] >> 7, (1 << 15) - 1);
-            }
-        break;
-        case 8:
-            for (i = 0; i < dstW; i++) {
-                register int srcPos = filterPos[i];
-                vector unsigned char src_vF, src_v0, src_v1;
-                vector unsigned char permS;
-                vector signed short src_v, filter_v;
-                vector signed int val_v, val_s;
-                FIRST_LOAD(src_v0, srcPos, src, permS);
-                LOAD_SRCV8(srcPos, 0, src, permS, src_v0, src_v1, src_vF);
-                src_v = // vec_unpackh sign-extends...
-                        (vector signed short)(VEC_MERGEH((vector unsigned char)vzero, src_vF));
-                filter_v = vec_ld(i << 4, filter);
-                val_v = vec_msums(src_v, filter_v, (vector signed int)vzero);
-                val_s = vec_sums(val_v, vzero);
-                vec_st(val_s, 0, tempo);
-                dst[i] = FFMIN(tempo[3] >> 7, (1 << 15) - 1);
-            }
-        break;
-
-        case 16:
-            for (i = 0; i < dstW; i++) {
-                register int srcPos = filterPos[i];
-
-                vector unsigned char src_vF = unaligned_load(srcPos, src);
-                vector signed short src_vA = // vec_unpackh sign-extends...
-                                             (vector signed short)(VEC_MERGEH((vector unsigned char)vzero, src_vF));
-                vector signed short src_vB = // vec_unpackh sign-extends...
-                                             (vector signed short)(VEC_MERGEL((vector unsigned char)vzero, src_vF));
-                vector signed short filter_v0 = vec_ld(i << 5, filter);
-                vector signed short filter_v1 = vec_ld((i << 5) + 16, filter);
-
-                vector signed int val_acc = vec_msums(src_vA, filter_v0, (vector signed int)vzero);
-                vector signed int val_v   = vec_msums(src_vB, filter_v1, val_acc);
-
-                vector signed int val_s = vec_sums(val_v, vzero);
-
-                VEC_ST(val_s, 0, tempo);
-                dst[i] = FFMIN(tempo[3] >> 7, (1 << 15) - 1);
-            }
-        break;
-
-        default:
-            for (i = 0; i < dstW; i++) {
-                register int j, offset = i * 2 * filterSize;
-                register int srcPos = filterPos[i];
-
-                vector signed int val_s, val_v = (vector signed int)vzero;
-                vector signed short filter_v0R;
-                vector unsigned char permF, src_v0, permS;
-                FIRST_LOAD(filter_v0R, offset, filter, permF);
-                FIRST_LOAD(src_v0, srcPos, src, permS);
-
-                for (j = 0; j < filterSize - 15; j += 16) {
-                    vector unsigned char src_v1, src_vF;
-                    vector signed short filter_v1R, filter_v2R, filter_v0, filter_v1;
-                    LOAD_SRCV(srcPos, j, src, permS, src_v0, src_v1, src_vF);
-                    vector signed short src_vA = // vec_unpackh sign-extends...
-                                                 (vector signed short)(VEC_MERGEH((vector unsigned char)vzero, src_vF));
-                    vector signed short src_vB = // vec_unpackh sign-extends...
-                                                 (vector signed short)(VEC_MERGEL((vector unsigned char)vzero, src_vF));
-                    GET_VFD(i, j, filter, filter_v0R, filter_v1R, permF, filter_v0, 0);
-                    GET_VFD(i, j, filter, filter_v1R, filter_v2R, permF, filter_v1, 16);
-
-                    vector signed int val_acc = vec_msums(src_vA, filter_v0, val_v);
-                    val_v = vec_msums(src_vB, filter_v1, val_acc);
-                    UPDATE_PTR(filter_v2R, filter_v0R, src_v1, src_v0);
-                }
-
-                if (j < filterSize - 7) {
-                    // loading src_v0 is useless, it's already done above
-                    vector unsigned char src_v1, src_vF;
-                    vector signed short src_v, filter_v1R, filter_v;
-                    LOAD_SRCV8(srcPos, j, src, permS, src_v0, src_v1, src_vF);
-                    src_v = // vec_unpackh sign-extends...
-                            (vector signed short)(VEC_MERGEH((vector unsigned char)vzero, src_vF));
-                    GET_VFD(i, j, filter, filter_v0R, filter_v1R, permF, filter_v, 0);
-                    val_v = vec_msums(src_v, filter_v, val_v);
-                }
-                val_s = vec_sums(val_v, vzero);
-
-                VEC_ST(val_s, 0, tempo);
-                dst[i] = FFMIN(tempo[3] >> 7, (1 << 15) - 1);
-            }
-        }
-}
-
-static void yuv2plane1_8_u(const int16_t *src, uint8_t *dest, int dstW,
-                           const uint8_t *dither, int offset, int start)
-{
-    int i;
-    for (i = start; i < dstW; i++) {
-        int val = (src[i] + dither[(i + offset) & 7]) >> 7;
-        dest[i] = av_clip_uint8(val);
-    }
-}
-
-static void yuv2plane1_8_altivec(const int16_t *src, uint8_t *dest, int dstW,
-                           const uint8_t *dither, int offset)
-{
-    const int dst_u = -(uintptr_t)dest & 15;
-    int i, j;
-    LOCAL_ALIGNED(16, int16_t, val, [16]);
-    const vector uint16_t shifts = (vector uint16_t) {7, 7, 7, 7, 7, 7, 7, 7};
-    vector int16_t vi, vileft, ditherleft, ditherright;
-    vector uint8_t vd;
-
-    for (j = 0; j < 16; j++) {
-        val[j] = dither[(dst_u + offset + j) & 7];
-    }
-
-    ditherleft = vec_ld(0, val);
-    ditherright = vec_ld(0, &val[8]);
-
-    yuv2plane1_8_u(src, dest, dst_u, dither, offset, 0);
-
-    for (i = dst_u; i < dstW - 15; i += 16) {
-
-        vi = vec_vsx_ld(0, &src[i]);
-        vi = vec_adds(ditherleft, vi);
-        vileft = vec_sra(vi, shifts);
-
-        vi = vec_vsx_ld(0, &src[i + 8]);
-        vi = vec_adds(ditherright, vi);
-        vi = vec_sra(vi, shifts);
-
-        vd = vec_packsu(vileft, vi);
-        vec_st(vd, 0, &dest[i]);
-    }
-
-    yuv2plane1_8_u(src, dest, dstW, dither, offset, i);
-}
-
-#endif /* HAVE_ALTIVEC */
+#endif /* HAVE_ALTIVEC && HAVE_BIGENDIAN */
 
 av_cold void ff_sws_init_swscale_ppc(SwsContext *c)
 {
@@ -381,8 +112,9 @@ av_cold void ff_sws_init_swscale_ppc(SwsContext *c)
     if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
         return;
 
+#if HAVE_BIGENDIAN
     if (c->srcBpc == 8 && c->dstBpc <= 14) {
-        c->hyScale = c->hcScale = hScale_altivec_real;
+        c->hyScale = c->hcScale = hScale_real_altivec;
     }
     if (!is16BPS(dstFormat) && !isNBPS(dstFormat) &&
         dstFormat != AV_PIX_FMT_NV12 && dstFormat != AV_PIX_FMT_NV21 &&
@@ -390,6 +122,7 @@ av_cold void ff_sws_init_swscale_ppc(SwsContext *c)
         !c->needAlpha) {
         c->yuv2planeX = yuv2planeX_altivec;
     }
+#endif
 
     /* The following list of supported dstFormat values should
      * match what's found in the body of ff_yuv2packedX_altivec() */
@@ -414,12 +147,8 @@ av_cold void ff_sws_init_swscale_ppc(SwsContext *c)
             c->yuv2packedX = ff_yuv2rgb24_X_altivec;
             break;
         }
-
-        switch (c->dstBpc) {
-        case 8:
-            c->yuv2plane1 = yuv2plane1_8_altivec;
-            break;
-        }
     }
 #endif /* HAVE_ALTIVEC */
+
+    ff_sws_init_swscale_vsx(c);
 }