From: Paul Cercueil <paul@crapouillou.net>
Date: Sat, 21 May 2022 16:52:47 +0000 (+0100)
Subject: dfxvideo: Restore support for big-endian arch
X-Git-Tag: r24l~454^2~1
X-Git-Url: https://notaz.gp2x.de/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=495d603c20ec58689bd7f73348d16e17f5491f5b;p=pcsx_rearmed.git

dfxvideo: Restore support for big-endian arch

The hardware registers, emulated RAM or emulated VRAM are all in
little-endian format. Therefore, each access must be done with one of
the byte-swap macros, for the plugin to work on big-endian systems.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
---

diff --git a/plugins/dfxvideo/gpu.c b/plugins/dfxvideo/gpu.c
index 3d20dfa4..60dcaa47 100644
--- a/plugins/dfxvideo/gpu.c
+++ b/plugins/dfxvideo/gpu.c
@@ -985,7 +985,7 @@ ENDVRAM:
          if((gpuDataC==254 && gpuDataP>=3) ||
             (gpuDataC==255 && gpuDataP>=4 && !(gpuDataP&1)))
           {
-           if((gpuDataM[gpuDataP] & 0xF000F000) == 0x50005000)
+           if((gpuDataM[gpuDataP] & HOST2LE32(0xF000F000)) == HOST2LE32(0x50005000))
             gpuDataP=gpuDataC-1;
           }
         }
diff --git a/plugins/dfxvideo/gpu.h b/plugins/dfxvideo/gpu.h
index 25fcc3ce..7b5aaef0 100644
--- a/plugins/dfxvideo/gpu.h
+++ b/plugins/dfxvideo/gpu.h
@@ -69,8 +69,8 @@
 
 // byteswappings
 
-#define SWAP16(x) ({ uint16_t y=(x); (((y)>>8 & 0xff) | ((y)<<8 & 0xff00)); })
-#define SWAP32(x) ({ uint32_t y=(x); (((y)>>24 & 0xfful) | ((y)>>8 & 0xff00ul) | ((y)<<8 & 0xff0000ul) | ((y)<<24 & 0xff000000ul)); })
+#define SWAP16(x) __builtin_bswap16(x)
+#define SWAP32(x) __builtin_bswap32(x)
 
 #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
 
@@ -107,7 +107,7 @@
 #define GETLE32_(X) LE2HOST32(*(uint32_t *)X)
 #define GETLE16D(X) ({uint32_t val = GETLE32(X); (val<<16 | val >> 16);})
 #define PUTLE16(X, Y) do{*((uint16_t *)X)=HOST2LE16((uint16_t)Y);}while(0)
-#define PUTLE32_(X, Y) do{*((uint32_t *)X)=HOST2LE16((uint32_t)Y);}while(0)
+#define PUTLE32_(X, Y) do{*((uint32_t *)X)=HOST2LE32((uint32_t)Y);}while(0)
 #ifdef __arm__
 #define GETLE32(X) (*(uint16_t *)(X)|(((uint16_t *)(X))[1]<<16))
 #define PUTLE32(X, Y) do{uint16_t *p_=(uint16_t *)(X);uint32_t y_=Y;p_[0]=y_;p_[1]=y_>>16;}while(0)
diff --git a/plugins/dfxvideo/gpulib_if.c b/plugins/dfxvideo/gpulib_if.c
index c1466ed8..c0066b5a 100644
--- a/plugins/dfxvideo/gpulib_if.c
+++ b/plugins/dfxvideo/gpulib_if.c
@@ -53,8 +53,8 @@
 
 // byteswappings
 
-#define SWAP16(x) ({ uint16_t y=(x); (((y)>>8 & 0xff) | ((y)<<8 & 0xff00)); })
-#define SWAP32(x) ({ uint32_t y=(x); (((y)>>24 & 0xfful) | ((y)>>8 & 0xff00ul) | ((y)<<8 & 0xff0000ul) | ((y)<<24 & 0xff000000ul)); })
+#define SWAP16(x) __builtin_bswap16(x)
+#define SWAP32(x) __builtin_bswap32(x)
 
 #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
 
@@ -91,7 +91,7 @@
 #define GETLE32_(X) LE2HOST32(*(uint32_t *)X)
 #define GETLE16D(X) ({uint32_t val = GETLE32(X); (val<<16 | val >> 16);})
 #define PUTLE16(X, Y) do{*((uint16_t *)X)=HOST2LE16((uint16_t)Y);}while(0)
-#define PUTLE32_(X, Y) do{*((uint32_t *)X)=HOST2LE16((uint32_t)Y);}while(0)
+#define PUTLE32_(X, Y) do{*((uint32_t *)X)=HOST2LE32((uint32_t)Y);}while(0)
 #ifdef __arm__
 #define GETLE32(X) (*(uint16_t *)(X)|(((uint16_t *)(X))[1]<<16))
 #define PUTLE32(X, Y) do{uint16_t *p_=(uint16_t *)(X);uint32_t y_=Y;p_[0]=y_;p_[1]=y_>>16;}while(0)
@@ -327,7 +327,7 @@ int do_cmd_list(uint32_t *list, int list_len, int *last_cmd)
 
   for (; list < list_end; list += 1 + len)
   {
-    cmd = *list >> 24;
+    cmd = GETLE32(list) >> 24;
     len = cmd_lengths[cmd];
     if (list + 1 + len > list_end) {
       cmd = -1;
@@ -338,7 +338,7 @@ int do_cmd_list(uint32_t *list, int list_len, int *last_cmd)
     if (cmd == 0xa0 || cmd == 0xc0)
       break; // image i/o, forward to upper layer
     else if ((cmd & 0xf8) == 0xe0)
-      gpu.ex_regs[cmd & 7] = list[0];
+      gpu.ex_regs[cmd & 7] = GETLE32(list);
 #endif
 
     primTableJ[cmd]((void *)list);
@@ -357,7 +357,7 @@ int do_cmd_list(uint32_t *list, int list_len, int *last_cmd)
             goto breakloop;
           }
 
-          if((*list_position & 0xf000f000) == 0x50005000)
+          if((*list_position & HOST2LE32(0xf000f000)) == HOST2LE32(0x50005000))
             break;
 
           list_position++;
@@ -380,7 +380,7 @@ int do_cmd_list(uint32_t *list, int list_len, int *last_cmd)
             goto breakloop;
           }
 
-          if((*list_position & 0xf000f000) == 0x50005000)
+          if((*list_position & HOST2LE32(0xf000f000)) == HOST2LE32(0x50005000))
             break;
 
           list_position += 2;
@@ -395,8 +395,8 @@ int do_cmd_list(uint32_t *list, int list_len, int *last_cmd)
       case 0xA0:          //  sys -> vid
       {
         short *slist = (void *)list;
-        u32 load_width = slist[4];
-        u32 load_height = slist[5];
+        u32 load_width = LE2HOST32(slist[4]);
+        u32 load_height = LE2HOST32(slist[5]);
         u32 load_size = load_width * load_height;
 
         len += load_size / 2;
diff --git a/plugins/dfxvideo/prim.c b/plugins/dfxvideo/prim.c
index 7a71d4c8..c872c284 100644
--- a/plugins/dfxvideo/prim.c
+++ b/plugins/dfxvideo/prim.c
@@ -1240,7 +1240,7 @@ static void primLineGEx(unsigned char *baseAddr)
    sly1=(short)(((int)sly1<<SIGNSHIFT)>>SIGNSHIFT);
   }
 
- lc1 = gpuData[0] & 0xffffff;
+ lc1 = GETLE32(&gpuData[0]) & 0xffffff;
 
  DrawSemiTrans = (SEMITRANSBIT(GETLE32(&gpuData[0]))) ? TRUE : FALSE;
 
diff --git a/plugins/dfxvideo/soft.c b/plugins/dfxvideo/soft.c
index 1b22ed50..285500ea 100644
--- a/plugins/dfxvideo/soft.c
+++ b/plugins/dfxvideo/soft.c
@@ -292,7 +292,7 @@ static inline void GetShadeTransCol32(uint32_t * pdest,uint32_t color)
     {
      int32_t sr,sb,sg,src,sbc,sgc,c;
      src=XCOL1(color);sbc=XCOL2(color);sgc=XCOL3(color);
-     c=GETLE32(pdest)>>16;
+     c=HIWORD(GETLE32(pdest));
      sr=(XCOL1(c))-src;   if(sr&0x8000) sr=0;
      sb=(XCOL2(c))-sbc;  if(sb&0x8000) sb=0;
      sg=(XCOL3(c))-sgc; if(sg&0x8000) sg=0;
@@ -327,8 +327,8 @@ static inline void GetShadeTransCol32(uint32_t * pdest,uint32_t color)
     {
      uint32_t ma=GETLE32(pdest);
      PUTLE32(pdest, (X32PSXCOL(r,g,b))|lSetMask);//0x80008000;
-     if(ma&0x80000000) PUTLE32(pdest, (ma&0xFFFF0000)|(*pdest&0xFFFF));
-     if(ma&0x00008000) PUTLE32(pdest, (ma&0xFFFF)    |(*pdest&0xFFFF0000));
+     if(ma&0x80000000) PUTLE32(pdest, (ma&0xFFFF0000)|(GETLE32(pdest)&0xFFFF));
+     if(ma&0x00008000) PUTLE32(pdest, (ma&0xFFFF)    |(GETLE32(pdest)&0xFFFF0000));
      return;
     }
    PUTLE32(pdest, (X32PSXCOL(r,g,b))|lSetMask);//0x80008000;
@@ -950,7 +950,7 @@ static void FillSoftwareAreaTrans(short x0,short y0,short x1, // FILL AREA TRANS
   {
    static int iCheat=0;
    col+=iCheat;
-   if(iCheat==1) iCheat=0; else iCheat=1;
+   iCheat ^= 1;
   }
 
 
@@ -971,7 +971,7 @@ static void FillSoftwareAreaTrans(short x0,short y0,short x1, // FILL AREA TRANS
   {
    uint32_t *DSTPtr;
    unsigned short LineOffset;
-   uint32_t lcol=lSetMask|(((uint32_t)(col))<<16)|col;
+   uint32_t lcol = HOST2LE32(lSetMask | (((uint32_t)(col)) << 16) | col);
    dx>>=1;
    DSTPtr = (uint32_t *)(psxVuw + (1024*y0) + x0);
    LineOffset = 512 - dx;
@@ -980,7 +980,7 @@ static void FillSoftwareAreaTrans(short x0,short y0,short x1, // FILL AREA TRANS
     {
      for(i=0;i<dy;i++)
       {
-       for(j=0;j<dx;j++) { PUTLE32(DSTPtr, lcol); DSTPtr++; }
+       for(j=0;j<dx;j++) { *DSTPtr++ = lcol; }
        DSTPtr += LineOffset;
       }
     }
@@ -1035,14 +1035,14 @@ static void FillSoftwareArea(short x0,short y0,short x1,      // FILL AREA (BLK
   {
    uint32_t *DSTPtr;
    unsigned short LineOffset;
-   uint32_t lcol=(((int32_t)col)<<16)|col;
+   uint32_t lcol = HOST2LE32((((uint32_t)(col)) << 16) | col);
    dx>>=1;
    DSTPtr = (uint32_t *)(psxVuw + (1024*y0) + x0);
    LineOffset = 512 - dx;
 
    for(i=0;i<dy;i++)
     {
-     for(j=0;j<dx;j++) { PUTLE32(DSTPtr, lcol); DSTPtr++; }
+     for(j=0;j<dx;j++) { *DSTPtr++ = lcol; }
      DSTPtr += LineOffset;
     } 
   }
diff --git a/plugins/gpulib/gpu.c b/plugins/gpulib/gpu.c
index bde6da7a..d7290916 100644
--- a/plugins/gpulib/gpu.c
+++ b/plugins/gpulib/gpu.c
@@ -438,7 +438,7 @@ static void start_vram_transfer(uint32_t pos_word, uint32_t size_word, int is_re
   if (is_read) {
     gpu.status |= PSX_GPU_STATUS_IMG;
     // XXX: wrong for width 1
-    memcpy(&gpu.gp0, VRAM_MEM_XY(gpu.dma.x, gpu.dma.y), 4);
+    gpu.gp0 = LE32TOH(*(uint32_t *) VRAM_MEM_XY(gpu.dma.x, gpu.dma.y));
     gpu.state.last_vram_read_frame = *gpu.state.frame_count;
   }
 
@@ -690,8 +690,11 @@ uint32_t GPUreadData(void)
     flush_cmd_buffer();
 
   ret = gpu.gp0;
-  if (gpu.dma.h)
+  if (gpu.dma.h) {
+    ret = HTOLE32(ret);
     do_vram_io(&ret, 1, 1);
+    ret = LE32TOH(ret);
+  }
 
   log_io("gpu_read %08x\n", ret);
   return ret;