Cleanup byteswap.c a bit (#154)

* Remove word_swap_longword.

This was the same as `swapx` and not used except in one place which
now uses `swapx` instead.

* Remove assembly versions of swapx, byte_swap_word, word_swap_page.

`swapx` and `byte_swap_word` were defined as:

```
extern inline const unsigned int swapx (unsigned int word)
{
   asm("roll  $16,%0" : "=g" (word) : "0" (word));
   return(word);
}

extern inline const unsigned short byte_swap_word (unsigned short word)
{
   asm("rolw  $8,%0" : "=r" (word) : "0" (word));

   return(word);
}
```

But the generated code from the C version is:

```
(lldb) disassemble -n swapx
ldex`swapx:
ldex[0x10002e0d0] <+0>:  pushq  %rbp
ldex[0x10002e0d1] <+1>:  movq   %rsp, %rbp
ldex[0x10002e0d4] <+4>:  movl   %edi, %eax
ldex[0x10002e0d6] <+6>:  roll   $0x10, %eax
ldex[0x10002e0d9] <+9>:  popq   %rbp
ldex[0x10002e0da] <+10>: retq

(lldb) disassemble -n byte_swap_word
ldex`byte_swap_word:
ldex[0x10002e0e0] <+0>:  pushq  %rbp
ldex[0x10002e0e1] <+1>:  movq   %rsp, %rbp
ldex[0x10002e0e4] <+4>:  rolw   $0x8, %di
ldex[0x10002e0e8] <+8>:  movzwl %di, %eax
ldex[0x10002e0eb] <+11>: popq   %rbp
ldex[0x10002e0ec] <+12>: retq
```

So we don't really stand to gain by re-enabling this old assembler code.

We would gain from switching to C99 or C11 and improving our
inlining.

* Remove 386 asm version of bit_reverse_region.

This is implemented in C and the code isn't actually set up to
allow using the assembler version.

Author: waywardmonkeys

inc/inlnPS2.h

@@ -65,65 +65,6 @@
 /************************************************************************/
 
 
-#undef SWAP_WORDS
-#define SWAP_WORDS swapx
-
-
-extern inline const unsigned int swapx (unsigned int word)
- {
-    asm("roll	$16,%0" : "=g" (word) : "0" (word));
-    return(word);
- }
-
-
-
-extern inline const unsigned int word_swap_longword (unsigned int word)
- {
-    asm("roll	$16,%0" : "=r" (word) : "0" (word));
-
-    return(word);
-  }
-
-
-
-extern inline const unsigned short byte_swap_word (unsigned short word)
- {
-    asm("rolw	$8,%0" : "=r" (word) : "0" (word));
-
-    return(word);
-  }
-
-
-
-extern inline void word_swap_page(unsigned short * page, int count)
-  {
-   asm volatile("\
-	pushl %edi					\n\
-	pushl %esi					\n\
-	pushl %ecx					\n\
-	cld");
-    asm volatile("\
-	movl %0,%%esi	// word pointer.		\n\
-	movl %%esi,%%edi				\n\
-	movl %1,%%ecx	// count" : : "g" (page), "g" (count));
-    asm volatile("\
-	lodsl						\n\
-	rolw	$8,%ax					\n\
-	roll	$16,%eax				\n\
-	rolw	$8,%ax					\n\
-	stosl						\n\
-	loop	.-13					\n\
-							\n\
-	// epilogue.					\n\
-	popl %ecx					\n\
-	popl %esi					\n\
-	popl %edi					\
-	");
-
-  }
-
-
-
 /************************************************************************/
 /*									*/
 /*									*/

inc/inlndos.h

@@ -65,65 +65,6 @@
 /************************************************************************/
 
 
-#undef SWAP_WORDS
-#define SWAP_WORDS swapx
-
-
-extern inline  unsigned int swapx (unsigned int word)
- {
-    asm("rol	%0,16" : "=g" (word) : "0" (word));
-    return(word);
- }
-
-
-
-extern inline  unsigned int word_swap_longword (unsigned int word)
- {
-    asm("rol	%0,16" : "=r" (word) : "0" (word));
-
-    return(word);
-  }
-
-
-
-extern inline  unsigned short byte_swap_word (unsigned short word)
- {
-    asm("rol	%0,8" : "=r" (word) : "0" (word));
-
-    return(word);
-  }
-
-
-
-extern inline  void word_swap_page(unsigned short * page, int count)
-  {
-   asm volatile("\
-	pushl edi					\n\
-	pushl esi					\n\
-	pushl ecx					\n\
-	cld");
-    asm volatile("
-	movl esi,%0	// word pointer.		\n\
-	movl edi,esi				\n\
-	movl ecx,%1	// count" : : "g" (page), "g" (count));
-    asm volatile("			\
-	lodsl						\n\
-	rol	ax,8					\n\
-	rol	eax,16				\n\
-	rol	ax,8					\n\
-	stosl						\n\
-	loop	.-13					\n\
-							\n\
-	// epilogue.					\n\
-	popl ecx					\n\
-	popl esi					\n\
-	popl edi					\n\
-	");
-
-  }
-
-
-
 /************************************************************************/
 /*									*/
 /*									*/

src/byteswap.c

@@ -27,10 +27,6 @@
 
 #include "byteswapdefs.h"
 
-#if defined(ISC)
-#include "inlnPS2.h"
-#else
-
 /****************************************************************/
 /*                                                              */
 /*                 swap halves of a single 4-byte word          */
@@ -49,23 +45,6 @@ unsigned short byte_swap_word(unsigned short word) {
   return (((word >> 8) & 0xff) | ((word & 0xff) << 8));
 }
 
-/****************************************************************/
-/*                                                              */
-/*                   Word-swap a 2-word integer                 */
-/*           Does NOT byte-swap the words themselves.           */
-/*                                                              */
-/****************************************************************/
-/***
-unsigned int word_swap_longword(word)
-  unsigned int word;
-  {
-      return( ((word>>16)&0xffff)+((word&0xffff)<<16) );
-  } ***/
-#ifndef I386
-#define word_swap_longword(word) (((word >> 16) & 0xffff) | ((word & 0xffff) << 16))
-#endif
-#endif /* !ISC */
-
 /****************************************************************/
 /*                                                              */
 /*            Byte-swap a region wordcount words long           */
@@ -77,7 +56,6 @@ void byte_swap_page(unsigned short *page, int wordcount) {
   for (i = 0; i < wordcount; i++) { *(page + i) = byte_swap_word(*(page + i)); }
 }
 
-#ifndef GCC386
 /****************************************************************/
 /*                                                              */
 /*     Byte- & word-swap a region wordcount long-words long     */
@@ -90,9 +68,8 @@ void word_swap_page(unsigned short *page, int longwordcount) {
   for (i = 0; i < (longwordcount + longwordcount); i++) {
     *(page + i) = byte_swap_word(*(page + i));
   }
-  for (i = 0; i < longwordcount; i++) { *(longpage + i) = word_swap_longword(*(longpage + i)); }
+  for (i = 0; i < longwordcount; i++) { *(longpage + i) = swapx(*(longpage + i)); }
 }
-#endif /* GCC386 */
 
 /****************************************************************/
 /*                                                              */

src/dsp386.il

@@ -34,100 +34,6 @@
 /************************************************************************/
 
 
-
-/ swap the words in a long-word.
-.inline swapx,1
-	movl	0(%esp),%eax
-	rol	$16,%eax
-.end
-
-
-/ swap the words in a long-word
-.inline word_swap_longword,1
-	movl	0(%esp),%eax
-	rol	$16,%eax
-.end
-
-/swap the bytes in a 16-bit word
-.inline byte_swap_word,1
-	movl	0(%esp),%eax
-	rolw	$8,%ax
-.end
-
-
-.inline word_swap_page,8
-	movl	4(%esp),%ecx	/ word count into the loop counter
-	movl	0(%esp),%edx	/ address of the block to swap
-	movl	0(%edx),%eax
-	rolw	$8,%ax
-	roll	$16,%eax
-	rolw	$8,%ax
-	movl	%eax,0(%edx)
-	add	$4,%edx
-	loop	.-20
-.end
-
-.inline bit_reverse_region,16
-/args: (top width height rasterwidth)
-	.noopt
-	pushl	%edi
-	pushl	%esi
-	pushl	%ebx
-	movl	12(%esp),%edx	/top
-	movl	%edx,%eax
-	andl	$-2,%edx
-	movl	%edx,12(%esp)
-	andl	$1,%eax
-	shll	$4,%eax
-	addl	16(%esp),%eax	/width
-	movl	20(%esp),%edx	/height
-/		24(%esp)	/rasterwidth, in words
-	leal	31(%eax),%eax
-	sarl	$5,%eax
-	leal	(%eax,%eax),%eax
-	movl	%eax,16(%esp)	/ word wid now in width.
-	cld			/ so we move up thru memory
-	leal	reversedbits,%ebx / for xlateb
-	movl	12(%esp),%edi	/ starting init of word ptr
-	sub	$0,%edx
-	jle	.+63		
-/ ..4:
-	movl	%edi,%esi	/so both edi & esi are at start of line to swap
-
-	movl	16(%esp),%ecx	/ byte count to ecx
-	leal	(%ecx,%ecx),%ecx
-/ ..3:
-	lodsb
-	xlat
-	stosb
-	loop	.-3		/ ..3
-
-	movl	16(%esp),%ecx
-	incl	%ecx
-	sarl	$1,%ecx
-	movl	12(%esp),%esi	/ must be into esi, then edi, to
-/	andl	$-2,%edi	/ defeat the peephole optimizer.
-	movl	%esi,%edi
-/ ..5:
-	lodsl
-	rolw	$8,%ax
-	roll	$16,%eax
-	rolw	$8,%ax
-	stosl
-	loop	.-13		/ ..5
-	movl	12(%esp),%edi	/ word = word + rasterwidth
-	addl	24(%esp),%edi
-	addl	24(%esp),%edi
-	movl	%edi,12(%esp)
-	decl	%edx
-/ .L60:	/ at this point, starting word addr is in %edi
-	jg	.-59		/ ..4
-	popl	%ebx
-	popl	%esi
-	popl	%edi
-	.optim
-.end
-
 /////////////////////////////
 //
 //	Dispatch loop speedup functions for the 386i