Faster blit colorkey or not, applied to bpp: 3->4 and 4->3

===== BlitNtoNKey ========
ABGR8888 -> BGR24 :  faster x3   (2168709 -> 562738)
ABGR8888 -> RGB24 :  faster x3   (2165055 -> 567458)

ARGB8888 -> BGR24 :  faster x3   (2169109 -> 564338)
ARGB8888 -> RGB24 :  faster x3   (2165266 -> 567081)

BGR24 -> ABGR8888 :  faster x3   (2997675 -> 891636)
BGR24 -> ARGB8888 :  faster x3   (2985449 -> 892028)
BGR24 -> BGR888 :  faster x3   (2961611 -> 891913)
BGR24 -> BGRA8888 :  faster x3   (3116305 -> 891534)
BGR24 -> BGRX8888 :  faster x3   (3179654 -> 896978)
BGR24 -> RGB888 :  faster x3   (2968191 -> 895112)
BGR24 -> RGBA8888 :  faster x3   (2998428 -> 893147)
BGR24 -> RGBX8888 :  faster x3   (2976529 -> 914853)

BGR888 -> BGR24 :  faster x3   (2161906 -> 563921)
BGR888 -> RGB24 :  faster x3   (2168228 -> 566634)

BGRA8888 -> BGR24 :  faster x4   (2270501 -> 561873)
BGRA8888 -> RGB24 :  faster x3   (2163179 -> 567330)

BGRX8888 -> BGR24 :  faster x3   (2162911 -> 562322)
BGRX8888 -> RGB24 :  faster x3   (2169617 -> 570927)

RGB24 -> ABGR8888 :  faster x3   (2977061 -> 925975)
RGB24 -> ARGB8888 :  faster x3   (2978148 -> 923680)
RGB24 -> BGR888 :  faster x3   (3001413 -> 935074)
RGB24 -> BGRA8888 :  faster x3   (2959003 -> 924096)
RGB24 -> BGRX8888 :  faster x3   (2965240 -> 927100)
RGB24 -> RGB888 :  faster x3   (2983921 -> 926063)
RGB24 -> RGBA8888 :  faster x3   (2963908 -> 925457)
RGB24 -> RGBX8888 :  faster x3   (2967957 -> 931700)

RGB888 -> BGR24 :  faster x3   (2173299 -> 563226)
RGB888 -> RGB24 :  faster x3   (2218374 -> 566164)

RGBA8888 -> BGR24 :  faster x3   (2166355 -> 561381)
RGBA8888 -> RGB24 :  faster x3   (2170322 -> 566729)

RGBX8888 -> BGR24 :  faster x3   (2168524 -> 564072)
RGBX8888 -> RGB24 :  faster x3   (2163680 -> 566956)

===== BlitNtoN ========

BGR24 -> BGRA8888 :  faster x3   (2458958 -> 797557)
BGR24 -> BGRX8888 :  faster x3   (2486085 -> 797745)
BGR24 -> RGBA8888 :  faster x3   (2422116 -> 797637)
BGR24 -> RGBX8888 :  faster x3   (2454426 -> 799085)

BGRA8888 -> BGR24 :  faster x4   (2468206 -> 524486)
BGRA8888 -> RGB24 :  faster x4   (2463581 -> 525561)

BGRX8888 -> BGR24 :  faster x4   (2583355 -> 524468)
BGRX8888 -> RGB24 :  faster x4   (2477242 -> 524284)

RGB24 -> BGRA8888 :  faster x2   (2453414 -> 818415)
RGB24 -> BGRX8888 :  faster x3   (2414915 -> 800863)
RGB24 -> RGBA8888 :  faster x3   (2461114 -> 798148)
RGB24 -> RGBX8888 :  faster x3   (2400922 -> 799203)

RGBA8888 -> BGR24 :  faster x4   (2494472 -> 526428)
RGBA8888 -> RGB24 :  faster x4   (2462260 -> 526791)

RGBX8888 -> BGR24 :  faster x4   (2541115 -> 524390)
RGBX8888 -> RGB24 :  faster x4   (2469059 -> 525416)

src/video/SDL_blit_N.c

@@ -2219,7 +2219,7 @@ BlitNtoN(SDL_BlitInfo * info)
     int dstbpp = dstfmt->BytesPerPixel;
     unsigned alpha = dstfmt->Amask ? info->a : 0;
 
-    /* Any src/dst 8888, not CopyAlpha, no ARGB2101010 */
+    /* Blit with permutation: 4->4 */
     if (srcbpp == 4 && dstbpp == 4 &&
         srcfmt->format != SDL_PIXELFORMAT_ARGB2101010 &&
         dstfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
@@ -2252,6 +2252,64 @@ BlitNtoN(SDL_BlitInfo * info)
         return;
     }
 
+    /* Blit with permutation: 4->3 */
+    if (srcbpp == 4 && dstbpp == 3 &&
+        srcfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
+
+        Uint32 *src32 = (Uint32*)src;
+
+        /* Find the appropriate permutation */
+        int r, g, b, a;
+        get_permutation(srcfmt, dstfmt, &r, &g, &b, &a, NULL);
+
+        while (height--) {
+            /* *INDENT-OFF* */
+            DUFFS_LOOP(
+            {
+                Uint8 *s8 = (Uint8 *)src32;
+                dst[0] = s8[r];
+                dst[1] = s8[g];
+                dst[2] = s8[b];
+                ++src32;
+                dst += 3;
+            }, width);
+            /* *INDENT-ON* */
+            src32 = (Uint32 *)((Uint8 *)src32 + srcskip);
+            dst += dstskip;
+        }
+        return;
+    }
+
+    /* Blit with permutation: 3->4 */
+    if (srcbpp == 3 && dstbpp == 4 &&
+        dstfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
+
+        Uint32 *dst32 = (Uint32*)dst;
+
+        /* Find the appropriate permutation */
+        int missing = 0, r, g, b, a;
+        get_permutation(srcfmt, dstfmt, &r, &g, &b, &a, &missing);
+
+        while (height--) {
+            /* *INDENT-OFF* */
+            DUFFS_LOOP(
+            {
+                Uint8 *d8 = (Uint8 *)dst32;
+                d8[0] = src[r];
+                d8[1] = src[g];
+                d8[2] = src[b];
+                d8[3] = src[a];
+                d8[missing] = alpha;
+                src += 3;
+                ++dst32;
+            }, width);
+            /* *INDENT-ON* */
+            src += srcskip;
+            dst32 = (Uint32 *)((Uint8 *)dst32 + dstskip);
+        }
+        return;
+    }
+
     while (height--) {
         /* *INDENT-OFF* */
         DUFFS_LOOP(
@@ -2287,7 +2345,7 @@ BlitNtoNCopyAlpha(SDL_BlitInfo * info)
     int dstbpp = dstfmt->BytesPerPixel;
     int c;
 
-    /* Any src/dst 8888 for CopyAlpha, no ARGB2101010 */
+    /* Blit with permutation: 4->4 */
     if (srcbpp == 4 && dstbpp == 4 &&
         srcfmt->format != SDL_PIXELFORMAT_ARGB2101010 &&
         dstfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
@@ -2499,7 +2557,7 @@ BlitNtoNKey(SDL_BlitInfo * info)
         }
     }
 
-    /* Any src/dst 8888, not CopyAlpha, no ARGB2101010 */
+    /* Blit with permutation: 4->4 */
     if (srcbpp == 4 && dstbpp == 4 &&
         srcfmt->format != SDL_PIXELFORMAT_ARGB2101010 &&
         dstfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
@@ -2598,6 +2656,77 @@ BlitNtoNKey(SDL_BlitInfo * info)
         return;
     }
 
+    /* Blit with permutation: 4->3 */
+    if (srcbpp == 4 && dstbpp == 3 &&
+        srcfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
+
+        Uint32 *src32 = (Uint32*)src;
+
+        /* Find the appropriate permutation */
+        int r, g, b, a;
+        get_permutation(srcfmt, dstfmt, &r, &g, &b, &a, NULL);
+
+        while (height--) {
+            /* *INDENT-OFF* */
+            DUFFS_LOOP(
+            {
+                if ((*src32 & rgbmask) != ckey) {
+                    Uint8 *s8 = (Uint8 *)src32;
+                    dst[0] = s8[r];
+                    dst[1] = s8[g];
+                    dst[2] = s8[b];
+                }
+                ++src32;
+                dst += 3;
+            }, width);
+            /* *INDENT-ON* */
+            src32 = (Uint32 *)((Uint8 *)src32 + srcskip);
+            dst += dstskip;
+        }
+        return;
+    }
+
+    /* Blit with permutation: 3->4 */
+    if (srcbpp == 3 && dstbpp == 4 &&
+        dstfmt->format != SDL_PIXELFORMAT_ARGB2101010) {
+
+        Uint32 *dst32 = (Uint32*)dst;
+
+        Uint8 k0 = ckey & 0xFF;
+        Uint8 k1 = (ckey >> 8)  & 0xFF;
+        Uint8 k2 = (ckey >> 16) & 0xFF;
+
+        /* Find the appropriate permutation */
+        int missing = 0, r, g, b, a;
+        get_permutation(srcfmt, dstfmt, &r, &g, &b, &a, &missing);
+
+        while (height--) {
+            /* *INDENT-OFF* */
+            DUFFS_LOOP(
+            {
+                Uint8 s0 = src[0];
+                Uint8 s1 = src[1];
+                Uint8 s2 = src[2];
+
+                if (k0 != s0 || k1 != s1 || k2 != s2) {
+                    Uint8 *d8 = (Uint8 *)dst32;
+                    d8[0] = src[r];
+                    d8[1] = src[g];
+                    d8[2] = src[b];
+                    d8[3] = src[a];
+                    d8[missing] = alpha;
+                }
+                src += 3;
+                ++dst32;
+            }, width);
+            /* *INDENT-ON* */
+            src += srcskip;
+            dst32 = (Uint32 *)((Uint8 *)dst32 + dstskip);
+
+        }
+        return;
+    }
+
     while (height--) {
         /* *INDENT-OFF* */
         DUFFS_LOOP(
@@ -2674,7 +2803,7 @@ BlitNtoNKeyCopyAlpha(SDL_BlitInfo * info)
         return;
     }
 
-    /* Any src/dst 8888 for CopyAlpha, no ARGB2101010 */
+    /* Blit with permutation: 4->4 */
     if (srcbpp == 4 && dstbpp == 4 &&
         srcfmt->format != SDL_PIXELFORMAT_ARGB2101010 &&
         dstfmt->format != SDL_PIXELFORMAT_ARGB2101010) {