whitespace cosmetics, prettyprinting

Originally committed as revision 19751 to svn://svn.ffmpeg.org/ffmpeg/trunk
release/0.6
Diego Biurrun 16 years ago
parent f4258b1391
commit ee1550116b
  1. 70
      libavutil/internal.h

@ -56,7 +56,7 @@
#endif
#ifndef INT16_MIN
#define INT16_MIN (-0x7fff-1)
#define INT16_MIN (-0x7fff - 1)
#endif
#ifndef INT16_MAX
@ -64,7 +64,7 @@
#endif
#ifndef INT32_MIN
#define INT32_MIN (-0x7fffffff-1)
#define INT32_MIN (-0x7fffffff - 1)
#endif
#ifndef INT32_MAX
@ -76,7 +76,7 @@
#endif
#ifndef INT64_MIN
#define INT64_MIN (-0x7fffffffffffffffLL-1)
#define INT64_MIN (-0x7fffffffffffffffLL - 1)
#endif
#ifndef INT64_MAX
@ -96,7 +96,7 @@
#endif
#ifndef offsetof
# define offsetof(T,F) ((unsigned int)((char *)&((T *)0)->F))
# define offsetof(T, F) ((unsigned int)((char *)&((T *)0)->F))
#endif
/* Use to export labels from asm. */
@ -129,11 +129,11 @@ extern const uint32_t ff_inverse[256];
#if ARCH_X86
# define FASTDIV(a,b) \
({\
int ret,dmy;\
int ret, dmy;\
__asm__ volatile(\
"mull %3"\
:"=d"(ret),"=a"(dmy)\
:"1"(a),"g"(ff_inverse[b])\
:"=d"(ret), "=a"(dmy)\
:"1"(a), "g"(ff_inverse[b])\
);\
ret;\
})
@ -152,14 +152,14 @@ static inline av_const int FASTDIV(int a, int b)
static inline av_const int FASTDIV(int a, int b)
{
int r, t;
__asm__ volatile ("umull %1, %0, %2, %3"
: "=&r"(r), "=&r"(t) : "r"(a), "r"(ff_inverse[b]));
__asm__ volatile("umull %1, %0, %2, %3"
: "=&r"(r), "=&r"(t) : "r"(a), "r"(ff_inverse[b]));
return r;
}
#elif CONFIG_FASTDIV
# define FASTDIV(a,b) ((uint32_t)((((uint64_t)a)*ff_inverse[b])>>32))
# define FASTDIV(a,b) ((uint32_t)((((uint64_t)a) * ff_inverse[b]) >> 32))
#else
# define FASTDIV(a,b) ((a)/(b))
# define FASTDIV(a,b) ((a) / (b))
#endif
extern const uint8_t ff_sqrt_tab[256];
@ -168,20 +168,20 @@ static inline av_const unsigned int ff_sqrt(unsigned int a)
{
unsigned int b;
if(a<255) return (ff_sqrt_tab[a+1]-1)>>4;
else if(a<(1<<12)) b= ff_sqrt_tab[a>>4 ]>>2;
if (a < 255) return (ff_sqrt_tab[a + 1] - 1) >> 4;
else if (a < (1 << 12)) b = ff_sqrt_tab[a >> 4] >> 2;
#if !CONFIG_SMALL
else if(a<(1<<14)) b= ff_sqrt_tab[a>>6 ]>>1;
else if(a<(1<<16)) b= ff_sqrt_tab[a>>8 ] ;
else if (a < (1 << 14)) b = ff_sqrt_tab[a >> 6] >> 1;
else if (a < (1 << 16)) b = ff_sqrt_tab[a >> 8] ;
#endif
else{
int s= av_log2_16bit(a>>16)>>1;
unsigned int c= a>>(s+2);
b= ff_sqrt_tab[c>>(s+8)];
b= FASTDIV(c,b) + (b<<s);
else {
int s = av_log2_16bit(a >> 16) >> 1;
unsigned int c = a >> (s + 2);
b = ff_sqrt_tab[c >> (s + 8)];
b = FASTDIV(c,b) + (b << s);
}
return b - (a<b*b);
return b - (a < b * b);
}
#if ARCH_X86
@ -194,14 +194,14 @@ static inline av_const unsigned int ff_sqrt(unsigned int a)
);
#else
#define MASK_ABS(mask, level)\
mask= level>>31;\
level= (level^mask)-mask;
mask = level >> 31;\
level = (level ^ mask) - mask;
#endif
#if HAVE_CMOV
#define COPY3_IF_LT(x,y,a,b,c,d)\
__asm__ volatile (\
"cmpl %0, %3 \n\t"\
#define COPY3_IF_LT(x, y, a, b, c, d)\
__asm__ volatile(\
"cmpl %0, %3 \n\t"\
"cmovl %3, %0 \n\t"\
"cmovl %4, %1 \n\t"\
"cmovl %5, %2 \n\t"\
@ -209,11 +209,11 @@ __asm__ volatile (\
: "r" (y), "r" (b), "r" (d)\
);
#else
#define COPY3_IF_LT(x,y,a,b,c,d)\
if((y)<(x)){\
(x)=(y);\
(a)=(b);\
(c)=(d);\
#define COPY3_IF_LT(x, y, a, b, c, d)\
if ((y) < (x)) {\
(x) = (y);\
(a) = (b);\
(c) = (d);\
}
#endif
@ -251,8 +251,8 @@ if((y)<(x)){\
#define CHECKED_ALLOC(p, size)\
{\
p= av_malloc(size);\
if(p==NULL && (size)!=0){\
p = av_malloc(size);\
if (p == NULL && (size) != 0) {\
av_log(NULL, AV_LOG_ERROR, "Cannot allocate memory.");\
goto fail;\
}\
@ -260,8 +260,8 @@ if((y)<(x)){\
#define CHECKED_ALLOCZ(p, size)\
{\
p= av_mallocz(size);\
if(p==NULL && (size)!=0){\
p = av_mallocz(size);\
if (p == NULL && (size) != 0) {\
av_log(NULL, AV_LOG_ERROR, "Cannot allocate memory.");\
goto fail;\
}\

Loading…
Cancel
Save