diff --git a/libavcodec/bitstream.h b/libavcodec/bitstream.h
index af25b6dcf5..22ec01069f 100644
--- a/libavcodec/bitstream.h
+++ b/libavcodec/bitstream.h
@@ -877,7 +877,7 @@ void free_vlc(VLC *vlc);
  *                  read the longest vlc code
  *                  = (max_vlc_length + bits - 1) / bits
  */
-static always_inline int get_vlc2(GetBitContext *s, VLC_TYPE (*table)[2],
+static av_always_inline int get_vlc2(GetBitContext *s, VLC_TYPE (*table)[2],
                                   int bits, int max_depth)
 {
     int code;
diff --git a/libavcodec/bytestream.h b/libavcodec/bytestream.h
index 25c457fe4c..a742fa1c14 100644
--- a/libavcodec/bytestream.h
+++ b/libavcodec/bytestream.h
@@ -22,32 +22,32 @@
 #ifndef FFMPEG_BYTESTREAM_H
 #define FFMPEG_BYTESTREAM_H
 
-static always_inline unsigned int bytestream_get_le32(uint8_t **b)
+static av_always_inline unsigned int bytestream_get_le32(uint8_t **b)
 {
     (*b) += 4;
     return LE_32(*b - 4);
 }
 
-static always_inline unsigned int bytestream_get_le16(uint8_t **b)
+static av_always_inline unsigned int bytestream_get_le16(uint8_t **b)
 {
     (*b) += 2;
     return LE_16(*b - 2);
 }
 
-static always_inline unsigned int bytestream_get_byte(uint8_t **b)
+static av_always_inline unsigned int bytestream_get_byte(uint8_t **b)
 {
     (*b)++;
     return (*b)[-1];
 }
 
-static always_inline unsigned int bytestream_get_buffer(uint8_t **b, uint8_t *dst, unsigned int size)
+static av_always_inline unsigned int bytestream_get_buffer(uint8_t **b, uint8_t *dst, unsigned int size)
 {
     memcpy(dst, *b, size);
     (*b) += size;
     return size;
 }
 
-static always_inline void bytestream_put_be32(uint8_t **b, const unsigned int value)
+static av_always_inline void bytestream_put_be32(uint8_t **b, const unsigned int value)
 {
     *(*b)++ = value >> 24;
     *(*b)++ = value >> 16;
@@ -55,13 +55,13 @@ static always_inline void bytestream_put_be32(uint8_t **b, const unsigned int va
     *(*b)++ = value;
 };
 
-static always_inline void bytestream_put_be16(uint8_t **b, const unsigned int value)
+static av_always_inline void bytestream_put_be16(uint8_t **b, const unsigned int value)
 {
     *(*b)++ = value >> 8;
     *(*b)++ = value;
 }
 
-static always_inline void bytestream_put_le32(uint8_t **b, const unsigned int value)
+static av_always_inline void bytestream_put_le32(uint8_t **b, const unsigned int value)
 {
     *(*b)++ = value;
     *(*b)++ = value >> 8;
@@ -69,18 +69,18 @@ static always_inline void bytestream_put_le32(uint8_t **b, const unsigned int va
     *(*b)++ = value >> 24;
 }
 
-static always_inline void bytestream_put_le16(uint8_t **b, const unsigned int value)
+static av_always_inline void bytestream_put_le16(uint8_t **b, const unsigned int value)
 {
     *(*b)++ = value;
     *(*b)++ = value >> 8;
 }
 
-static always_inline void bytestream_put_byte(uint8_t **b, const unsigned int value)
+static av_always_inline void bytestream_put_byte(uint8_t **b, const unsigned int value)
 {
     *(*b)++ = value;
 }
 
-static always_inline void bytestream_put_buffer(uint8_t **b, const uint8_t *src, unsigned int size)
+static av_always_inline void bytestream_put_buffer(uint8_t **b, const uint8_t *src, unsigned int size)
 {
     memcpy(*b, src, size);
     (*b) += size;
diff --git a/libavcodec/cabac.h b/libavcodec/cabac.h
index 43fe78e3bb..f47406a9ea 100644
--- a/libavcodec/cabac.h
+++ b/libavcodec/cabac.h
@@ -363,7 +363,7 @@ static inline void renorm_cabac_decoder_once(CABACContext *c){
         refill(c);
 }
 
-static int always_inline get_cabac_inline(CABACContext *c, uint8_t * const state){
+static int av_always_inline get_cabac_inline(CABACContext *c, uint8_t * const state){
     //FIXME gcc generates duplicate load/stores for c->low and c->range
 #define LOW          "0"
 #define RANGE        "4"
@@ -631,7 +631,7 @@ static int get_cabac_bypass(CABACContext *c){
 }
 
 
-static always_inline int get_cabac_bypass_sign(CABACContext *c, int val){
+static av_always_inline int get_cabac_bypass_sign(CABACContext *c, int val){
 #if defined(ARCH_X86) && !(defined(PIC) && defined(__GNUC__))
     asm volatile(
         "movl "RANGE    "(%1), %%ebx            \n\t"
diff --git a/libavcodec/dv.c b/libavcodec/dv.c
index 76095a4819..803d3502de 100644
--- a/libavcodec/dv.c
+++ b/libavcodec/dv.c
@@ -560,7 +560,7 @@ static inline void dv_decode_video_segment(DVVideoContext *s,
 
 #ifdef DV_CODEC_TINY_TARGET
 /* Converts run and level (where level != 0) pair into vlc, returning bit size */
-static always_inline int dv_rl2vlc(int run, int level, int sign, uint32_t* vlc)
+static av_always_inline int dv_rl2vlc(int run, int level, int sign, uint32_t* vlc)
 {
     int size;
     if (run < DV_VLC_MAP_RUN_SIZE && level < DV_VLC_MAP_LEV_SIZE) {
@@ -585,7 +585,7 @@ static always_inline int dv_rl2vlc(int run, int level, int sign, uint32_t* vlc)
     return size;
 }
 
-static always_inline int dv_rl2vlc_size(int run, int level)
+static av_always_inline int dv_rl2vlc_size(int run, int level)
 {
     int size;
 
@@ -601,13 +601,13 @@ static always_inline int dv_rl2vlc_size(int run, int level)
     return size;
 }
 #else
-static always_inline int dv_rl2vlc(int run, int l, int sign, uint32_t* vlc)
+static av_always_inline int dv_rl2vlc(int run, int l, int sign, uint32_t* vlc)
 {
     *vlc = dv_vlc_map[run][l].vlc | sign;
     return dv_vlc_map[run][l].size;
 }
 
-static always_inline int dv_rl2vlc_size(int run, int l)
+static av_always_inline int dv_rl2vlc_size(int run, int l)
 {
     return dv_vlc_map[run][l].size;
 }
@@ -627,7 +627,7 @@ typedef struct EncBlockInfo {
     uint32_t partial_bit_buffer; /* we can't use uint16_t here */
 } EncBlockInfo;
 
-static always_inline PutBitContext* dv_encode_ac(EncBlockInfo* bi, PutBitContext* pb_pool,
+static av_always_inline PutBitContext* dv_encode_ac(EncBlockInfo* bi, PutBitContext* pb_pool,
                                        PutBitContext* pb_end)
 {
     int prev;
@@ -670,7 +670,7 @@ static always_inline PutBitContext* dv_encode_ac(EncBlockInfo* bi, PutBitContext
     return pb;
 }
 
-static always_inline void dv_set_class_number(DCTELEM* blk, EncBlockInfo* bi,
+static av_always_inline void dv_set_class_number(DCTELEM* blk, EncBlockInfo* bi,
                                               const uint8_t* zigzag_scan, const int *weight, int bias)
 {
     int i, area;
@@ -742,7 +742,7 @@ static always_inline void dv_set_class_number(DCTELEM* blk, EncBlockInfo* bi,
 
 //FIXME replace this by dsputil
 #define SC(x, y) ((s[x] - s[y]) ^ ((s[x] - s[y]) >> 7))
-static always_inline int dv_guess_dct_mode(DCTELEM *blk) {
+static av_always_inline int dv_guess_dct_mode(DCTELEM *blk) {
     DCTELEM *s;
     int score88 = 0;
     int score248 = 0;
diff --git a/libavcodec/faandct.c b/libavcodec/faandct.c
index e3c0d84a2c..6f73ee5e97 100644
--- a/libavcodec/faandct.c
+++ b/libavcodec/faandct.c
@@ -70,7 +70,7 @@ B6*B0, B6*B1, B6*B2, B6*B3, B6*B4, B6*B5, B6*B6, B6*B7,
 B7*B0, B7*B1, B7*B2, B7*B3, B7*B4, B7*B5, B7*B6, B7*B7,
 };
 
-static always_inline void row_fdct(FLOAT temp[64], DCTELEM * data)
+static av_always_inline void row_fdct(FLOAT temp[64], DCTELEM * data)
 {
     FLOAT tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
     FLOAT tmp10, tmp11, tmp12, tmp13;
diff --git a/libavcodec/ffv1.c b/libavcodec/ffv1.c
index 62623e591b..1ca18a4e87 100644
--- a/libavcodec/ffv1.c
+++ b/libavcodec/ffv1.c
@@ -186,7 +186,7 @@ typedef struct FFV1Context{
     DSPContext dsp;
 }FFV1Context;
 
-static always_inline int fold(int diff, int bits){
+static av_always_inline int fold(int diff, int bits){
     if(bits==8)
         diff= (int8_t)diff;
     else{
diff --git a/libavcodec/h264.c b/libavcodec/h264.c
index 00ab7d7ff8..d7c48bd4a3 100644
--- a/libavcodec/h264.c
+++ b/libavcodec/h264.c
@@ -400,7 +400,7 @@ static void svq3_add_idct_c(uint8_t *dst, DCTELEM *block, int stride, int qp, in
 static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize);
 static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize);
 
-static always_inline uint32_t pack16to32(int a, int b){
+static av_always_inline uint32_t pack16to32(int a, int b){
 #ifdef WORDS_BIGENDIAN
    return (b&0xFFFF) + (a<<16);
 #else
@@ -423,7 +423,7 @@ const uint8_t ff_div6[52]={
  * @param w width of the rectangle, should be a constant
  * @param size the size of val (1 or 4), should be a constant
  */
-static always_inline void fill_rectangle(void *vp, int w, int h, int stride, uint32_t val, int size){
+static av_always_inline void fill_rectangle(void *vp, int w, int h, int stride, uint32_t val, int size){
     uint8_t *p= (uint8_t*)vp;
     assert(size==1 || size==4);
     assert(w<=4);
diff --git a/libavcodec/h264idct.c b/libavcodec/h264idct.c
index 3506418ad6..a6a56d33a7 100644
--- a/libavcodec/h264idct.c
+++ b/libavcodec/h264idct.c
@@ -28,7 +28,7 @@
 
 #include "dsputil.h"
 
-static always_inline void idct_internal(uint8_t *dst, DCTELEM *block, int stride, int block_stride, int shift, int add){
+static av_always_inline void idct_internal(uint8_t *dst, DCTELEM *block, int stride, int block_stride, int shift, int add){
     int i;
     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
diff --git a/libavcodec/i386/fdct_mmx.c b/libavcodec/i386/fdct_mmx.c
index 2ffbfecf64..7e2682a4af 100644
--- a/libavcodec/i386/fdct_mmx.c
+++ b/libavcodec/i386/fdct_mmx.c
@@ -284,7 +284,7 @@ TABLE_SSE2
 }};
 
 
-static always_inline void fdct_col(const int16_t *in, int16_t *out, int offset)
+static av_always_inline void fdct_col(const int16_t *in, int16_t *out, int offset)
 {
     movq_m2r(*(in + offset + 1 * 8), mm0);
     movq_m2r(*(in + offset + 6 * 8), mm1);
@@ -364,7 +364,7 @@ static always_inline void fdct_col(const int16_t *in, int16_t *out, int offset)
 }
 
 
-static always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
+static av_always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
 {
     asm volatile(
 #define FDCT_ROW_SSE2_H1(i,t)                    \
@@ -426,7 +426,7 @@ static always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
     );
 }
 
-static always_inline void fdct_row_mmx2(const int16_t *in, int16_t *out, const int16_t *table)
+static av_always_inline void fdct_row_mmx2(const int16_t *in, int16_t *out, const int16_t *table)
 {
     pshufw_m2r(*(in + 4), mm5, 0x1B);
     movq_m2r(*(in + 0), mm0);
@@ -469,7 +469,7 @@ static always_inline void fdct_row_mmx2(const int16_t *in, int16_t *out, const i
     movq_r2m(mm7, *(out + 4));
 }
 
-static always_inline void fdct_row_mmx(const int16_t *in, int16_t *out, const int16_t *table)
+static av_always_inline void fdct_row_mmx(const int16_t *in, int16_t *out, const int16_t *table)
 {
 //FIXME reorder (i dont have a old mmx only cpu here to benchmark ...)
     movd_m2r(*(in + 6), mm1);
diff --git a/libavcodec/jfdctfst.c b/libavcodec/jfdctfst.c
index 38424563d1..a9dcfab825 100644
--- a/libavcodec/jfdctfst.c
+++ b/libavcodec/jfdctfst.c
@@ -145,7 +145,7 @@
 
 #define MULTIPLY(var,const)  ((DCTELEM) DESCALE((var) * (const), CONST_BITS))
 
-static always_inline void row_fdct(DCTELEM * data){
+static av_always_inline void row_fdct(DCTELEM * data){
   int_fast16_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int_fast16_t tmp10, tmp11, tmp12, tmp13;
   int_fast16_t z1, z2, z3, z4, z5, z11, z13;
diff --git a/libavcodec/jfdctint.c b/libavcodec/jfdctint.c
index 58f3a1446a..250312467f 100644
--- a/libavcodec/jfdctint.c
+++ b/libavcodec/jfdctint.c
@@ -181,7 +181,7 @@
 #endif
 
 
-static always_inline void row_fdct(DCTELEM * data){
+static av_always_inline void row_fdct(DCTELEM * data){
   int_fast32_t tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
   int_fast32_t tmp10, tmp11, tmp12, tmp13;
   int_fast32_t z1, z2, z3, z4, z5;
diff --git a/libavcodec/mathops.h b/libavcodec/mathops.h
index 9ae34d71b2..c6ec705977 100644
--- a/libavcodec/mathops.h
+++ b/libavcodec/mathops.h
@@ -46,7 +46,7 @@
 //gcc 3.4 creates an incredibly bloated mess out of this
 //#    define MULH(a,b) (((int64_t)(a) * (int64_t)(b))>>32)
 
-static always_inline int MULH(int a, int b){
+static av_always_inline int MULH(int a, int b){
     return ((int64_t)(a) * (int64_t)(b))>>32;
 }
 #endif
diff --git a/libavcodec/motion_est.c b/libavcodec/motion_est.c
index 2c5705c2ba..fc057712d4 100644
--- a/libavcodec/motion_est.c
+++ b/libavcodec/motion_est.c
@@ -103,7 +103,7 @@ static int get_flags(MotionEstContext *c, int direct, int chroma){
            + (chroma ? FLAG_CHROMA : 0);
 }
 
-static always_inline int cmp(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
+static av_always_inline int cmp(MpegEncContext *s, const int x, const int y, const int subx, const int suby,
                       const int size, const int h, int ref_index, int src_index,
                       me_cmp_func cmp_func, me_cmp_func chroma_cmp_func, const int flags){
     MotionEstContext * const c= &s->me;
diff --git a/libavcodec/motion_est_template.c b/libavcodec/motion_est_template.c
index 980727d5a8..5a01ef6222 100644
--- a/libavcodec/motion_est_template.c
+++ b/libavcodec/motion_est_template.c
@@ -555,7 +555,7 @@ if( (y)>(ymax<<(S)) ) printf("%d %d %d %d %d ymax" #v, ymax, (x), (y), s->mb_x,
     const int qpel= flags&FLAG_QPEL;\
     const int shift= 1+qpel;\
 
-static always_inline int small_diamond_search(MpegEncContext * s, int *best, int dmin,
+static av_always_inline int small_diamond_search(MpegEncContext * s, int *best, int dmin,
                                        int src_index, int ref_index, int const penalty_factor,
                                        int size, int h, int flags)
 {
@@ -964,7 +964,7 @@ if(256*256*256*64 % (stats[0]+1)==0){
     return dmin;
 }
 
-static always_inline int diamond_search(MpegEncContext * s, int *best, int dmin,
+static av_always_inline int diamond_search(MpegEncContext * s, int *best, int dmin,
                                        int src_index, int ref_index, int const penalty_factor,
                                        int size, int h, int flags){
     MotionEstContext * const c= &s->me;
@@ -984,7 +984,7 @@ static always_inline int diamond_search(MpegEncContext * s, int *best, int dmin,
         return   var_diamond_search(s, best, dmin, src_index, ref_index, penalty_factor, size, h, flags);
 }
 
-static always_inline int epzs_motion_search_internal(MpegEncContext * s, int *mx_ptr, int *my_ptr,
+static av_always_inline int epzs_motion_search_internal(MpegEncContext * s, int *mx_ptr, int *my_ptr,
                              int P[10][2], int src_index, int ref_index, int16_t (*last_mv)[2],
                              int ref_mv_scale, int flags, int size, int h)
 {
diff --git a/libavcodec/mpeg12.c b/libavcodec/mpeg12.c
index c392e17afa..36d7153519 100644
--- a/libavcodec/mpeg12.c
+++ b/libavcodec/mpeg12.c
@@ -508,7 +508,7 @@ static inline void put_mb_modes(MpegEncContext *s, int n, int bits,
     }
 }
 
-static always_inline void mpeg1_encode_mb_internal(MpegEncContext *s,
+static av_always_inline void mpeg1_encode_mb_internal(MpegEncContext *s,
                                                    DCTELEM block[6][64],
                                                    int motion_x, int motion_y,
                                                    int mb_block_count)
diff --git a/libavcodec/mpegvideo.c b/libavcodec/mpegvideo.c
index 4f286fb2f9..41127013c6 100644
--- a/libavcodec/mpegvideo.c
+++ b/libavcodec/mpegvideo.c
@@ -2946,7 +2946,7 @@ static inline int hpel_motion_lowres(MpegEncContext *s,
 }
 
 /* apply one mpeg motion vector to the three components */
-static always_inline void mpeg_motion(MpegEncContext *s,
+static av_always_inline void mpeg_motion(MpegEncContext *s,
                                uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                                int field_based, int bottom_field, int field_select,
                                uint8_t **ref_picture, op_pixels_func (*pix_op)[4],
@@ -3063,7 +3063,7 @@ if(s->quarter_sample)
 }
 
 /* apply one mpeg motion vector to the three components */
-static always_inline void mpeg_motion_lowres(MpegEncContext *s,
+static av_always_inline void mpeg_motion_lowres(MpegEncContext *s,
                                uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr,
                                int field_based, int bottom_field, int field_select,
                                uint8_t **ref_picture, h264_chroma_mc_func *pix_op,
@@ -3895,7 +3895,7 @@ void ff_clean_intra_table_entries(MpegEncContext *s)
    s->mv       : motion vector
    s->interlaced_dct : true if interlaced dct used (mpeg2)
  */
-static always_inline void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64], int lowres_flag)
+static av_always_inline void MPV_decode_mb_internal(MpegEncContext *s, DCTELEM block[12][64], int lowres_flag)
 {
     int mb_x, mb_y;
     const int mb_xy = s->mb_y * s->mb_stride + s->mb_x;
@@ -4318,7 +4318,7 @@ static void get_vissual_weight(int16_t *weight, uint8_t *ptr, int stride){
     }
 }
 
-static always_inline void encode_mb_internal(MpegEncContext *s, int motion_x, int motion_y, int mb_block_height, int mb_block_count)
+static av_always_inline void encode_mb_internal(MpegEncContext *s, int motion_x, int motion_y, int mb_block_height, int mb_block_count)
 {
     int16_t weight[8][64];
     DCTELEM orig[8][64];
@@ -4600,7 +4600,7 @@ static always_inline void encode_mb_internal(MpegEncContext *s, int motion_x, in
     }
 }
 
-static always_inline void encode_mb(MpegEncContext *s, int motion_x, int motion_y)
+static av_always_inline void encode_mb(MpegEncContext *s, int motion_x, int motion_y)
 {
     if (s->chroma_format == CHROMA_420) encode_mb_internal(s, motion_x, motion_y,  8, 6);
     else                                encode_mb_internal(s, motion_x, motion_y, 16, 8);
diff --git a/libavcodec/pthread.c b/libavcodec/pthread.c
index 4737211cbc..f2c636bbe9 100644
--- a/libavcodec/pthread.c
+++ b/libavcodec/pthread.c
@@ -76,7 +76,7 @@ static void* worker(void *v)
     }
 }
 
-static always_inline void avcodec_thread_park_workers(ThreadContext *c, int thread_count)
+static av_always_inline void avcodec_thread_park_workers(ThreadContext *c, int thread_count)
 {
     pthread_cond_wait(&c->last_job_cond, &c->current_job_lock);
     pthread_mutex_unlock(&c->current_job_lock);
diff --git a/libavcodec/smacker.c b/libavcodec/smacker.c
index 2f21858488..2e1784075d 100644
--- a/libavcodec/smacker.c
+++ b/libavcodec/smacker.c
@@ -320,12 +320,12 @@ static int decode_header_trees(SmackVContext *smk) {
     return 0;
 }
 
-static always_inline void last_reset(int *recode, int *last) {
+static av_always_inline void last_reset(int *recode, int *last) {
     recode[last[0]] = recode[last[1]] = recode[last[2]] = 0;
 }
 
 /* get code and update history */
-static always_inline int smk_get_code(GetBitContext *gb, int *recode, int *last) {
+static av_always_inline int smk_get_code(GetBitContext *gb, int *recode, int *last) {
     register int *table = recode;
     int v, b;
 
diff --git a/libavcodec/snow.c b/libavcodec/snow.c
index 6bc9a8f1ae..ae6608bf31 100644
--- a/libavcodec/snow.c
+++ b/libavcodec/snow.c
@@ -709,7 +709,7 @@ static inline int get_symbol2(RangeCoder *c, uint8_t *state, int log2){
     return v;
 }
 
-static always_inline void lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
+static av_always_inline void lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
     const int mirror_left= !highpass;
     const int mirror_right= (width&1) ^ highpass;
     const int w= (width>>1) - 1 + (highpass & width);
@@ -732,7 +732,7 @@ static always_inline void lift(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst
 }
 
 #ifndef lift5
-static always_inline void lift5(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
+static av_always_inline void lift5(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
     const int mirror_left= !highpass;
     const int mirror_right= (width&1) ^ highpass;
     const int w= (width>>1) - 1 + (highpass & width);
@@ -764,7 +764,7 @@ static always_inline void lift5(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int ds
 #endif
 
 #ifndef liftS
-static always_inline void liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
+static av_always_inline void liftS(DWTELEM *dst, DWTELEM *src, DWTELEM *ref, int dst_step, int src_step, int ref_step, int width, int mul, int add, int shift, int highpass, int inverse){
     const int mirror_left= !highpass;
     const int mirror_right= (width&1) ^ highpass;
     const int w= (width>>1) - 1 + (highpass & width);
@@ -2206,7 +2206,7 @@ static int encode_q_branch(SnowContext *s, int level, int x, int y){
     }
 }
 
-static always_inline int same_block(BlockNode *a, BlockNode *b){
+static av_always_inline int same_block(BlockNode *a, BlockNode *b){
     if((a->type&BLOCK_INTRA) && (b->type&BLOCK_INTRA)){
         return !((a->color[0] - b->color[0]) | (a->color[1] - b->color[1]) | (a->color[2] - b->color[2]));
     }else{
@@ -2553,7 +2553,7 @@ void ff_snow_inner_add_yblock(uint8_t *obmc, const int obmc_stride, uint8_t * *
 }
 
 //FIXME name clenup (b_w, block_w, b_width stuff)
-static always_inline void add_yblock(SnowContext *s, int sliced, slice_buffer *sb, DWTELEM *dst, uint8_t *dst8, const uint8_t *obmc, int src_x, int src_y, int b_w, int b_h, int w, int h, int dst_stride, int src_stride, int obmc_stride, int b_x, int b_y, int add, int offset_dst, int plane_index){
+static av_always_inline void add_yblock(SnowContext *s, int sliced, slice_buffer *sb, DWTELEM *dst, uint8_t *dst8, const uint8_t *obmc, int src_x, int src_y, int b_w, int b_h, int w, int h, int dst_stride, int src_stride, int obmc_stride, int b_x, int b_y, int add, int offset_dst, int plane_index){
     const int b_width = s->b_width  << s->block_max_depth;
     const int b_height= s->b_height << s->block_max_depth;
     const int b_stride= b_width;
@@ -2712,7 +2712,7 @@ assert(src_stride > 2*MB_SIZE + 5);
 #endif
 }
 
-static always_inline void predict_slice_buffered(SnowContext *s, slice_buffer * sb, DWTELEM * old_buffer, int plane_index, int add, int mb_y){
+static av_always_inline void predict_slice_buffered(SnowContext *s, slice_buffer * sb, DWTELEM * old_buffer, int plane_index, int add, int mb_y){
     Plane *p= &s->plane[plane_index];
     const int mb_w= s->b_width  << s->block_max_depth;
     const int mb_h= s->b_height << s->block_max_depth;
@@ -2779,7 +2779,7 @@ static always_inline void predict_slice_buffered(SnowContext *s, slice_buffer *
     STOP_TIMER("predict_slice")
 }
 
-static always_inline void predict_slice(SnowContext *s, DWTELEM *buf, int plane_index, int add, int mb_y){
+static av_always_inline void predict_slice(SnowContext *s, DWTELEM *buf, int plane_index, int add, int mb_y){
     Plane *p= &s->plane[plane_index];
     const int mb_w= s->b_width  << s->block_max_depth;
     const int mb_h= s->b_height << s->block_max_depth;
@@ -2836,7 +2836,7 @@ static always_inline void predict_slice(SnowContext *s, DWTELEM *buf, int plane_
     STOP_TIMER("predict_slice")
 }
 
-static always_inline void predict_plane(SnowContext *s, DWTELEM *buf, int plane_index, int add){
+static av_always_inline void predict_plane(SnowContext *s, DWTELEM *buf, int plane_index, int add){
     const int mb_h= s->b_height << s->block_max_depth;
     int mb_y;
     for(mb_y=0; mb_y<=mb_h; mb_y++)
@@ -3094,7 +3094,7 @@ static int get_4block_rd(SnowContext *s, int mb_x, int mb_y, int plane_index){
     return distortion + rate*penalty_factor;
 }
 
-static always_inline int check_block(SnowContext *s, int mb_x, int mb_y, int p[3], int intra, const uint8_t *obmc_edged, int *best_rd){
+static av_always_inline int check_block(SnowContext *s, int mb_x, int mb_y, int p[3], int intra, const uint8_t *obmc_edged, int *best_rd){
     const int b_stride= s->b_width << s->block_max_depth;
     BlockNode *block= &s->block[mb_x + mb_y * b_stride];
     BlockNode backup= *block;
@@ -3133,12 +3133,12 @@ static always_inline int check_block(SnowContext *s, int mb_x, int mb_y, int p[3
 }
 
 /* special case for int[2] args we discard afterward, fixes compilation prob with gcc 2.95 */
-static always_inline int check_block_inter(SnowContext *s, int mb_x, int mb_y, int p0, int p1, const uint8_t *obmc_edged, int *best_rd){
+static av_always_inline int check_block_inter(SnowContext *s, int mb_x, int mb_y, int p0, int p1, const uint8_t *obmc_edged, int *best_rd){
     int p[2] = {p0, p1};
     return check_block(s, mb_x, mb_y, p, 0, obmc_edged, best_rd);
 }
 
-static always_inline int check_4block_inter(SnowContext *s, int mb_x, int mb_y, int p0, int p1, int ref, int *best_rd){
+static av_always_inline int check_4block_inter(SnowContext *s, int mb_x, int mb_y, int p0, int p1, int ref, int *best_rd){
     const int b_stride= s->b_width << s->block_max_depth;
     BlockNode *block= &s->block[mb_x + mb_y * b_stride];
     BlockNode backup[4]= {block[0], block[1], block[b_stride], block[b_stride+1]};
diff --git a/libavcodec/snow.h b/libavcodec/snow.h
index f7cee131af..6794d2c5a4 100644
--- a/libavcodec/snow.h
+++ b/libavcodec/snow.h
@@ -137,7 +137,7 @@ static int w97_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int
 
 /* C bits used by mmx/sse2/altivec */
 
-static always_inline void snow_interleave_line_header(int * i, int width, DWTELEM * low, DWTELEM * high){
+static av_always_inline void snow_interleave_line_header(int * i, int width, DWTELEM * low, DWTELEM * high){
     (*i) = (width) - 2;
 
     if (width & 1){
@@ -146,14 +146,14 @@ static always_inline void snow_interleave_line_header(int * i, int width, DWTELE
     }
 }
 
-static always_inline void snow_interleave_line_footer(int * i, DWTELEM * low, DWTELEM * high){
+static av_always_inline void snow_interleave_line_footer(int * i, DWTELEM * low, DWTELEM * high){
     for (; (*i)>=0; (*i)-=2){
         low[(*i)+1] = high[(*i)>>1];
         low[*i] = low[(*i)>>1];
     }
 }
 
-static always_inline void snow_horizontal_compose_lift_lead_out(int i, DWTELEM * dst, DWTELEM * src, DWTELEM * ref, int width, int w, int lift_high, int mul, int add, int shift){
+static av_always_inline void snow_horizontal_compose_lift_lead_out(int i, DWTELEM * dst, DWTELEM * src, DWTELEM * ref, int width, int w, int lift_high, int mul, int add, int shift){
     for(; i<w; i++){
         dst[i] = src[i] - ((mul * (ref[i] + ref[i + 1]) + add) >> shift);
     }
@@ -163,7 +163,7 @@ static always_inline void snow_horizontal_compose_lift_lead_out(int i, DWTELEM *
     }
 }
 
-static always_inline void snow_horizontal_compose_liftS_lead_out(int i, DWTELEM * dst, DWTELEM * src, DWTELEM * ref, int width, int w){
+static av_always_inline void snow_horizontal_compose_liftS_lead_out(int i, DWTELEM * dst, DWTELEM * src, DWTELEM * ref, int width, int w){
         for(; i<w; i++){
             dst[i] = src[i] - (((-(ref[i] + ref[(i+1)])+W_BO) - 4 * src[i]) >> W_BS);
         }
diff --git a/libavcodec/vc1.c b/libavcodec/vc1.c
index 7b385ca47f..cceeccfa5c 100644
--- a/libavcodec/vc1.c
+++ b/libavcodec/vc1.c
@@ -2140,7 +2140,7 @@ static void vc1_interp_mc(VC1Context *v)
     dsp->avg_h264_chroma_pixels_tab[0](s->dest[2], srcV, s->uvlinesize, 8, uvmx, uvmy);
 }
 
-static always_inline int scale_mv(int value, int bfrac, int inv, int qs)
+static av_always_inline int scale_mv(int value, int bfrac, int inv, int qs)
 {
     int n = bfrac;
 
diff --git a/libavcodec/vc1dsp.c b/libavcodec/vc1dsp.c
index 9139ffb284..f19f266d1a 100644
--- a/libavcodec/vc1dsp.c
+++ b/libavcodec/vc1dsp.c
@@ -326,7 +326,7 @@ static void vc1_inv_trans_4x4_c(DCTELEM block[64], int n)
 
 /** Filter used to interpolate fractional pel values
  */
-static always_inline int vc1_mspel_filter(const uint8_t *src, int stride, int mode, int r)
+static av_always_inline int vc1_mspel_filter(const uint8_t *src, int stride, int mode, int r)
 {
     switch(mode){
     case 0: //no shift
diff --git a/libavcodec/vmnc.c b/libavcodec/vmnc.c
index 7b5f567e77..49210158ff 100644
--- a/libavcodec/vmnc.c
+++ b/libavcodec/vmnc.c
@@ -72,7 +72,7 @@ typedef struct VmncContext {
 } VmncContext;
 
 /* read pixel value from stream */
-static always_inline int vmnc_get_pixel(uint8_t* buf, int bpp, int be) {
+static av_always_inline int vmnc_get_pixel(uint8_t* buf, int bpp, int be) {
     switch(bpp * 2 + be) {
     case 2:
     case 3: return *buf;
@@ -172,7 +172,7 @@ static void put_cursor(uint8_t *dst, int stride, VmncContext *c, int dx, int dy)
 }
 
 /* fill rectangle with given colour */
-static always_inline void paint_rect(uint8_t *dst, int dx, int dy, int w, int h, int color, int bpp, int stride)
+static av_always_inline void paint_rect(uint8_t *dst, int dx, int dy, int w, int h, int color, int bpp, int stride)
 {
     int i, j;
     dst += dx * bpp + dy * stride;
@@ -202,7 +202,7 @@ static always_inline void paint_rect(uint8_t *dst, int dx, int dy, int w, int h,
     }
 }
 
-static always_inline void paint_raw(uint8_t *dst, int w, int h, uint8_t* src, int bpp, int be, int stride)
+static av_always_inline void paint_raw(uint8_t *dst, int w, int h, uint8_t* src, int bpp, int be, int stride)
 {
     int i, j, p;
     for(j = 0; j < h; j++) {
diff --git a/libavcodec/vp3dsp.c b/libavcodec/vp3dsp.c
index a48515a5e5..bb9fed091f 100644
--- a/libavcodec/vp3dsp.c
+++ b/libavcodec/vp3dsp.c
@@ -39,7 +39,7 @@
 
 #define M(a,b) (((a) * (b))>>16)
 
-static always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
+static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
 {
     int16_t *ip = input;
     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
diff --git a/libavcodec/wavpack.c b/libavcodec/wavpack.c
index 5a54f7d0e2..18544831e8 100644
--- a/libavcodec/wavpack.c
+++ b/libavcodec/wavpack.c
@@ -98,7 +98,7 @@ static const uint8_t wp_exp2_table [256] = {
     0xea, 0xec, 0xed, 0xee, 0xf0, 0xf1, 0xf2, 0xf4, 0xf5, 0xf6, 0xf8, 0xf9, 0xfa, 0xfc, 0xfd, 0xff
 };
 
-static always_inline int wp_exp2(int16_t val)
+static av_always_inline int wp_exp2(int16_t val)
 {
     int res, neg = 0;
 
@@ -137,7 +137,7 @@ static inline int get_unary(GetBitContext *gb){
         }
 
 
-static always_inline int get_tail(GetBitContext *gb, int k)
+static av_always_inline int get_tail(GetBitContext *gb, int k)
 {
     int p, e, res;
 
diff --git a/libavutil/bswap.h b/libavutil/bswap.h
index 4614c90458..03d613db2d 100644
--- a/libavutil/bswap.h
+++ b/libavutil/bswap.h
@@ -37,7 +37,7 @@
 #endif
 
 #if defined(ARCH_X86)
-static always_inline uint16_t bswap_16(uint16_t x)
+static av_always_inline uint16_t bswap_16(uint16_t x)
 {
   __asm("rorw $8, %0"   :
         LEGACY_REGS (x) :
@@ -45,7 +45,7 @@ static always_inline uint16_t bswap_16(uint16_t x)
     return x;
 }
 
-static always_inline uint32_t bswap_32(uint32_t x)
+static av_always_inline uint32_t bswap_32(uint32_t x)
 {
 #if __CPU__ != 386
  __asm("bswap   %0":
@@ -82,12 +82,12 @@ static inline uint64_t bswap_64(uint64_t x)
 
 #elif defined(ARCH_SH4)
 
-static always_inline uint16_t bswap_16(uint16_t x) {
+static av_always_inline uint16_t bswap_16(uint16_t x) {
         __asm__("swap.b %0,%0":"=r"(x):"0"(x));
         return x;
 }
 
-static always_inline uint32_t bswap_32(uint32_t x) {
+static av_always_inline uint32_t bswap_32(uint32_t x) {
         __asm__(
         "swap.b %0,%0\n"
         "swap.w %0,%0\n"
@@ -110,12 +110,12 @@ static inline uint64_t bswap_64(uint64_t x)
 }
 #else
 
-static always_inline uint16_t bswap_16(uint16_t x){
+static av_always_inline uint16_t bswap_16(uint16_t x){
     return (x>>8) | (x<<8);
 }
 
 #ifdef ARCH_ARM
-static always_inline uint32_t bswap_32(uint32_t x){
+static av_always_inline uint32_t bswap_32(uint32_t x){
     uint32_t t;
     __asm__ (
       "eor %1, %0, %0, ror #16 \n\t"
@@ -126,7 +126,7 @@ static always_inline uint32_t bswap_32(uint32_t x){
     return x;
 }
 #else
-static always_inline uint32_t bswap_32(uint32_t x){
+static av_always_inline uint32_t bswap_32(uint32_t x){
     x= ((x<<8)&0xFF00FF00) | ((x>>8)&0x00FF00FF);
     return (x>>16) | (x<<16);
 }
diff --git a/libavutil/common.h b/libavutil/common.h
index d8baf12379..ca6147a802 100644
--- a/libavutil/common.h
+++ b/libavutil/common.h
@@ -43,6 +43,17 @@
 #        include "berrno.h"
 #    endif
 #    include <math.h>
+#endif /* HAVE_AV_CONFIG_H */
+
+#ifndef av_always_inline
+#if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 0)
+#    define av_always_inline __attribute__((always_inline)) inline
+#else
+#    define av_always_inline inline
+#endif
+#endif
+
+#ifdef HAVE_AV_CONFIG_H
 #    include "internal.h"
 #endif /* HAVE_AV_CONFIG_H */
 
diff --git a/libavutil/internal.h b/libavutil/internal.h
index 6e3e47af0b..d498f99de6 100644
--- a/libavutil/internal.h
+++ b/libavutil/internal.h
@@ -26,14 +26,6 @@
 #ifndef INTERNAL_H
 #define INTERNAL_H
 
-#ifndef always_inline
-#if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 0)
-#    define always_inline __attribute__((always_inline)) inline
-#else
-#    define always_inline inline
-#endif
-#endif
-
 #ifndef attribute_used
 #if defined(__GNUC__) && (__GNUC__ > 3 || __GNUC__ == 3 && __GNUC_MINOR__ > 0)
 #    define attribute_used __attribute__((used))
@@ -309,7 +301,7 @@ if((y)<(x)){\
 /* XXX: add ISOC specific test to avoid specific BSD testing. */
 /* better than nothing implementation. */
 /* btw, rintf() is existing on fbsd too -- alex */
-static always_inline long int lrintf(float x)
+static av_always_inline long int lrintf(float x)
 {
 #ifdef __MINGW32__
 #  ifdef ARCH_X86_32
diff --git a/libpostproc/postprocess.c b/libpostproc/postprocess.c
index c9f2893e3c..f68c8a7232 100644
--- a/libpostproc/postprocess.c
+++ b/libpostproc/postprocess.c
@@ -466,7 +466,7 @@ static inline void horizX1Filter(uint8_t *src, int stride, int QP)
 /**
  * accurate deblock filter
  */
-static always_inline void do_a_deblock_C(uint8_t *src, int step, int stride, PPContext *c){
+static av_always_inline void do_a_deblock_C(uint8_t *src, int step, int stride, PPContext *c){
         int y;
         const int QP= c->QP;
         const int dcOffset= ((c->nonBQP*c->ppMode.baseDcDiff)>>8) + 1;
diff --git a/libpostproc/postprocess_template.c b/libpostproc/postprocess_template.c
index c22d5d1b6d..55602bbdf5 100644
--- a/libpostproc/postprocess_template.c
+++ b/libpostproc/postprocess_template.c
@@ -2645,7 +2645,7 @@ Switch between
 /**
  * accurate deblock filter
  */
-static always_inline void RENAME(do_a_deblock)(uint8_t *src, int step, int stride, PPContext *c){
+static av_always_inline void RENAME(do_a_deblock)(uint8_t *src, int step, int stride, PPContext *c){
         int64_t dc_mask, eq_mask, both_masks;
         int64_t sums[10*8*2];
         src+= step*3; // src points to begin of the 8x8 Block