xserver-multidpi/hw/xfree86/os-support/misc/BUSmemcpy.c
2006-07-21 17:56:00 -04:00

413 lines
10 KiB
C

/****************************************************************************
For Alpha Linux, BusToMem() and MemToBus() can be simply memcpy(), BUT:
we need to prevent unaligned operations when accessing DENSE space on the BUS,
as the video memory is mmap'd that way. The below code does this.
NOTE: we could simply use the "memcpy()" from LIBC here, but that, currently, is
not as fast.
Thanks to Linus Torvalds for contributing this code.
****************************************************************************/
#ifdef HAVE_XORG_CONFIG_H
#include <xorg-config.h>
#endif
#include <X11/X.h>
#include "xf86.h"
#include "xf86Priv.h"
#include "xf86_OSlib.h"
#ifdef __alpha__
#include "compiler.h"
/*
* The Jensen lacks dense memory, thus we have to address the bus via
* the sparse addressing scheme. These routines are only used in s3im.c
* Non time critical code uses SlowBCopy_{from/to} bus.
*
* Martin Ostermann (ost@comnets.rwth-aachen.de) - Apr.-Sep. 1996
*/
#ifdef TEST_JENSEN_CODE /* define to test the Sparse addressing on a non-Jensen */
#define LWORD_CODING (0x18)
#define SPARSE (5)
#else
#define LWORD_CODING (0x60)
#define SPARSE (7)
#endif
void
xf86JensenMemToBus(char *Base, long dst, long src, int count)
{
if( ((long)src^((long)dst)) & 3) {
/* src & dst are NOT aligned to each other */
unsigned long addr;
unsigned long low_word, high_word,last_read;
long rm,loop;
unsigned long tmp,org,org2,mask,src_org,count_org;
src_org=src;
count_org=count;
/* add EISA longword coding and round off*/
addr = (long)(Base+(dst<<SPARSE) + LWORD_CODING) & ~(3<<SPARSE);
rm = (long)dst & 3;
count += rm;
count = count_org + rm;
org = *(volatile unsigned int *)addr;
__asm__("ldq_u %0,%1"
:"=r" (low_word):"m" (*(unsigned long *)(src_org)));
src = src_org - rm;
if( count > 4 ) {
last_read = src_org+count_org - 1;
__asm__("ldq_u %0,%1"
:"=r" (high_word):"m" (*(unsigned long *)(src+4)));
__asm__("extll %1,%2,%0"
:"=r" (low_word)
:"r" (low_word), "r" ((unsigned long)(src)));
__asm__("extlh %1,%2,%0"
:"=r" (tmp)
:"r" (high_word), "r" ((unsigned long)(src)));
tmp |= low_word;
src += 4;
__asm__("mskqh %1,%2,%0"
:"=r" (tmp)
:"r" (tmp), "r" (rm));
__asm__("mskql %1,%2,%0"
:"=r" (org2)
:"r" (org), "r" (rm));
tmp |= org2;
loop = (count-4) >> 2; /* loop eqv. count>=4 ; count -= 4 */
while (loop) {
/* tmp to be stored completly -- need to read next word*/
low_word = high_word;
*(volatile unsigned int *) (addr) = tmp;
__asm__("ldq_u %0,%1"
:"=r" (high_word):"m" (*(unsigned long*)(src+4)));
loop --;
__asm__("extll %1,%2,%0"
:"=r" (low_word)
:"r" (low_word), "r" ((unsigned long)src));
__asm__("extlh %1,%2,%0"
:"=r" (tmp)
:"r" (high_word), "r" ((unsigned long)src));
src += 4;
tmp |= low_word;
addr += 4<<SPARSE;
}
if ( count & 3 ) {
/* Store tmp completly, and possibly read one more word.*/
*(volatile unsigned int *) (addr) = tmp;
__asm__("ldq_u %0,%1"
:"=r" (tmp):"m" (*((unsigned long *)(last_read)) ));
addr += 4<<SPARSE;
__asm__("extll %1,%2,%0"
:"=r" (low_word)
:"r" (high_word), "r" ((unsigned long)src));
__asm__("extlh %1,%2,%0"
:"=r" (tmp)
:"r" (tmp), "r" ((unsigned long)src));
tmp |= low_word;
org = *(volatile unsigned int *)addr;
__asm__("mskql %1,%2,%0"
:"=r" (tmp)
:"r" (tmp), "r" (count&3));
__asm__("mskqh %1,%2,%0"
:"=r" (org)
:"r" (org), "r" (count&3));
tmp |= org;
}
*(volatile unsigned int *) (addr) = tmp;
return;
} else { /* count > 4 */
__asm__("ldq_u %0,%1"
:"=r" (high_word):"m" (*(unsigned long *)(src+4)));
__asm__("extll %1,%2,%0"
:"=r" (low_word)
:"r" (low_word), "r" ((unsigned long)(src)));
__asm__("extlh %1,%2,%0"
:"=r" (tmp)
:"r" (high_word), "r" ((unsigned long)(src)));
tmp |= low_word;
if( count < 4 ) {
mask = -1;
__asm__("mskqh %1,%2,%0"
:"=r" (mask)
:"r" (mask), "r" (rm));
__asm__("mskql %1,%2,%0"
:"=r" (mask)
:"r" (mask), "r" (count));
tmp = (tmp & mask) | (org & ~mask);
*(volatile unsigned int *) (addr) = tmp;
return;
} else {
__asm__("mskqh %1,%2,%0"
:"=r" (tmp)
:"r" (tmp), "r" (rm));
__asm__("mskql %1,%2,%0"
:"=r" (org2)
:"r" (org), "r" (rm));
tmp |= org2;
*(volatile unsigned int *) (addr) = tmp;
return;
}
}
} else { /* src & dst are aligned to each other */
unsigned long addr;
unsigned int tmp,org,rm;
unsigned int *src_r;
/* add EISA longword coding and round off*/
addr = (long)(Base+(dst<<SPARSE) + LWORD_CODING) & ~(3<<SPARSE);
src_r = (unsigned int*)((long)src & ~3L);
rm=(long)src & 3;
count += rm;
tmp = *src_r;
org = *(volatile unsigned int *)addr;
__asm__("mskqh %1,%2,%0"
:"=r" (tmp)
:"r" (tmp), "r" (rm));
__asm__("mskql %1,%2,%0"
:"=r" (org)
:"r" (org), "r" (rm));
tmp |= org;
while (count > 4) {
*(volatile unsigned int *) addr = tmp;
addr += 4<<SPARSE;
src_r += 1;
tmp = *src_r;
count -= 4;
}
org = *(volatile unsigned int *)addr;
__asm__("mskql %1,%2,%0"
:"=r" (tmp)
:"r" (tmp), "r" (count));
__asm__("mskqh %1,%2,%0"
:"=r" (org)
:"r" (org), "r" (count));
tmp |= org;
*(volatile unsigned int *) (addr) = tmp;
}
}
void
xf86JensenBusToMem(char *Base, char *dst, unsigned long src, int count)
{
#if 0
/* Optimization of BusToMem() is left as an exercise to the reader ;-)
* Consider that ldq_u/extlh/extll won't work because of the bus being
* only 4 bytes wide!
*/
#else
unsigned long addr;
long result;
addr = (unsigned long)(Base+(src<<SPARSE)) ;
while( addr & (3<<SPARSE) ){
if(count <= 0) return;
result = *(volatile int *) addr;
result >>= ((addr>>SPARSE) & 3) * 8;
*dst++ = (char) result;
addr += 1<<SPARSE;
count--;
}
count -=4;
while(count >= 0){
int i;
result = *(volatile int *) (addr+LWORD_CODING);
for(i=4;i--;) {
*dst++ = (char) result;
result >>= 8;
}
addr += 4<<SPARSE;
count -= 4;
}
count +=4;
while( count ){
result = *(volatile int *) addr;
result >>= ((addr>>SPARSE) & 3) * 8;
*dst++ = (char) result;
addr += 1<<SPARSE;
count--;
}
#endif
}
static unsigned long __memcpy(unsigned long dest, unsigned long src, int n);
_X_EXPORT void
xf86BusToMem(unsigned char *dst, unsigned char *src, int len)
{
__memcpy((unsigned long)dst, (unsigned long)src, len);
}
_X_EXPORT void
xf86MemToBus(unsigned char *dst, unsigned char *src, int len)
{
if (len == sizeof(int))
if (!(((long)src | (long)dst) & 3))
*((unsigned int*)dst) = *((unsigned int*)(src));
else {
int i;
if (((long)src) & 3)
i = ldl_u((unsigned int*)src);
else
i = *(unsigned int*)src;
if (((long)dst) & 3)
stl_u(i,(unsigned int*)dst);
else
*(unsigned int*)dst = i;
}
else
__memcpy((unsigned long)dst, (unsigned long)src, len);
}
/*
* linux/arch/alpha/lib/memcpy.c
*
* Copyright (C) 1995 Linus Torvalds, used with his permission.
*/
/*
* This is a reasonably optimized memcpy() routine.
*/
/*
* Note that the C code is written to be optimized into good assembly. However,
* at this point gcc is unable to sanely compile "if (n >= 0)", resulting in a
* explicit compare against 0 (instead of just using the proper "blt reg, xx" or
* "bge reg, xx"). I hope alpha-gcc will be fixed to notice this eventually..
*/
/*
* This should be done in one go with ldq_u*2/mask/stq_u. Do it
* with a macro so that we can fix it up later..
*/
#define ALIGN_DEST_TO8(d,s,n) \
while (d & 7) { \
if (n <= 0) return; \
n--; \
*(char *) d = *(char *) s; \
d++; s++; \
}
/*
* This should similarly be done with ldq_u*2/mask/stq. The destination
* is aligned, but we don't fill in a full quad-word
*/
#define DO_REST(d,s,n) \
while (n > 0) { \
n--; \
*(char *) d = *(char *) s; \
d++; s++; \
}
/*
* This should be done with ldq/mask/stq. The source and destination are
* aligned, but we don't fill in a full quad-word
*/
#define DO_REST_ALIGNED(d,s,n) DO_REST(d,s,n)
/*
* This does unaligned memory copies. We want to avoid storing to
* an unaligned address, as that would do a read-modify-write cycle.
* We also want to avoid double-reading the unaligned reads.
*
* Note the ordering to try to avoid load (and address generation) latencies.
*/
static __inline__ void __memcpy_unaligned(unsigned long d, unsigned long s, long n)
{
ALIGN_DEST_TO8(d,s,n);
n -= 8; /* to avoid compare against 8 in the loop */
if (n >= 0) {
unsigned long low_word, high_word;
__asm__("ldq_u %0,%1":"=r" (low_word):"m" (*(unsigned long *) s));
do {
unsigned long tmp;
__asm__("ldq_u %0,%1":"=r" (high_word):"m" (*(unsigned long *)(s+8)));
n -= 8;
__asm__("extql %1,%2,%0"
:"=r" (low_word)
:"r" (low_word), "r" (s));
__asm__("extqh %1,%2,%0"
:"=r" (tmp)
:"r" (high_word), "r" (s));
s += 8;
*(unsigned long *) d = low_word | tmp;
d += 8;
low_word = high_word;
} while (n >= 0);
}
n += 8;
DO_REST(d,s,n);
}
/*
* Hmm.. Strange. The __asm__ here is there to make gcc use a integer register
* for the load-store. I don't know why, but it would seem that using a floating
* point register for the move seems to slow things down (very small difference,
* though).
*
* Note the ordering to try to avoid load (and address generation) latencies.
*/
static __inline__ void __memcpy_aligned(unsigned long d, unsigned long s, long n)
{
ALIGN_DEST_TO8(d,s,n);
n -= 8;
while (n >= 0) {
unsigned long tmp;
__asm__("ldq %0,%1":"=r" (tmp):"m" (*(unsigned long *) s));
n -= 8;
s += 8;
*(unsigned long *) d = tmp;
d += 8;
}
n += 8;
DO_REST_ALIGNED(d,s,n);
}
static unsigned long __memcpy(unsigned long dest, unsigned long src, int n)
{
if (!((dest ^ src) & 7)) {
__memcpy_aligned(dest, src, n);
return dest;
}
__memcpy_unaligned(dest, src, n);
return dest;
}
#else /* __alpha__ */
void
xf86BusToMem(unsigned char *dst, unsigned char *src, int len)
{
memcpy(dst, src, len);
}
void
xf86MemToBus(unsigned char *dst, unsigned char *src, int len)
{
memcpy(dst, src, len);
}
#endif /* __alpha__ */