diff --git a/ffmpeg.c b/ffmpeg.c
index 47a90daaf8..f8cf6b7701 100644
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -1394,7 +1394,6 @@ static void do_streamcopy(InputStream *ist, OutputStream *ost, const AVPacket *p
 
     write_frame(of->ctx, &opkt, ost);
     ost->st->codec->frame_number++;
-    av_free_packet(&opkt);
 }
 
 static void rate_emu_sleep(InputStream *ist)
diff --git a/libavcodec/x86/fft.asm b/libavcodec/x86/fft.asm
index 4603da7f9f..5071741d63 100644
--- a/libavcodec/x86/fft.asm
+++ b/libavcodec/x86/fft.asm
@@ -105,7 +105,8 @@ SECTION_TEXT
     pfadd    %5, %4 ; {t6,t5}
     pxor     %3, [ps_m1p1] ; {t8,t7}
     mova     %6, %1
-    PSWAPD   %3, %3
+    movd [r0+12], %3
+    punpckhdq %3, [r0+8]
     pfadd    %1, %5 ; {r0,i0}
     pfsub    %6, %5 ; {r2,i2}
     mova     %4, %2
@@ -499,19 +500,6 @@ fft8 %+ SUFFIX:
 %endmacro
 
 %if ARCH_X86_32
-%macro PSWAPD 2
-%if cpuflag(3dnowext)
-    pswapd %1, %2
-%elifidn %1, %2
-    movd [r0+12], %1
-    punpckhdq %1, [r0+8]
-%else
-    movq  %1, %2
-    psrlq %1, 32
-    punpckldq %1, %2
-%endif
-%endmacro
-
 INIT_MMX 3dnowext
 FFT48_3DNOW
 
diff --git a/libavcodec/x86/fmtconvert.asm b/libavcodec/x86/fmtconvert.asm
index 8313d261d9..4f2379b2e2 100644
--- a/libavcodec/x86/fmtconvert.asm
+++ b/libavcodec/x86/fmtconvert.asm
@@ -246,16 +246,6 @@ FLOAT_TO_INT16_INTERLEAVE2
 INIT_XMM sse2
 FLOAT_TO_INT16_INTERLEAVE2
 
-
-%macro PSWAPD_SSE 2
-    pshufw %1, %2, 0x4e
-%endmacro
-%macro PSWAPD_3DNOW 2
-    movq  %1, %2
-    psrlq %1, 32
-    punpckldq %1, %2
-%endmacro
-
 %macro FLOAT_TO_INT16_INTERLEAVE6 0
 ; void float_to_int16_interleave6_sse(int16_t *dst, const float **src, int len)
 cglobal float_to_int16_interleave6, 2, 8, 0, dst, src, src1, src2, src3, src4, src5, len
@@ -285,11 +275,11 @@ cglobal float_to_int16_interleave6, 2, 8, 0, dst, src, src1, src2, src3, src4, s
     packssdw   mm0, mm3
     packssdw   mm1, mm4
     packssdw   mm2, mm5
-    pswapd     mm3, mm0
+    PSWAPD     mm3, mm0
     punpcklwd  mm0, mm1
     punpckhwd  mm1, mm2
     punpcklwd  mm2, mm3
-    pswapd     mm3, mm0
+    PSWAPD     mm3, mm0
     punpckldq  mm0, mm2
     punpckhdq  mm2, mm1
     punpckldq  mm1, mm3
@@ -305,12 +295,9 @@ cglobal float_to_int16_interleave6, 2, 8, 0, dst, src, src1, src2, src3, src4, s
 %endmacro ; FLOAT_TO_INT16_INTERLEAVE6
 
 INIT_MMX sse
-%define pswapd PSWAPD_SSE
 FLOAT_TO_INT16_INTERLEAVE6
 INIT_MMX 3dnow
-%define pswapd PSWAPD_3DNOW
 FLOAT_TO_INT16_INTERLEAVE6
-%undef pswapd
 INIT_MMX 3dnowext
 FLOAT_TO_INT16_INTERLEAVE6
 
diff --git a/libavfilter/vf_drawtext.c b/libavfilter/vf_drawtext.c
index 7b672712dc..849bf1344e 100644
--- a/libavfilter/vf_drawtext.c
+++ b/libavfilter/vf_drawtext.c
@@ -277,7 +277,7 @@ static int load_glyph(AVFilterContext *ctx, Glyph **glyph_ptr, uint32_t code)
     FT_Glyph_Get_CBox(*glyph->glyph, ft_glyph_bbox_pixels, &glyph->bbox);
 
     /* cache the newly created glyph */
-    if (!(node = av_mallocz(av_tree_node_size))) {
+    if (!(node = av_tree_node_alloc())) {
         ret = AVERROR(ENOMEM);
         goto error;
     }
diff --git a/libavformat/avformat.h b/libavformat/avformat.h
index 3d99518990..61bc4a6327 100644
--- a/libavformat/avformat.h
+++ b/libavformat/avformat.h
@@ -158,9 +158,15 @@
  * information will be in AVStream.time_base units, i.e. it has to be
  * multiplied by the timebase to convert them to seconds.
  *
- * The packet data belongs to the demuxer and is invalid after the next call to
- * av_read_frame(). The user must free the packet with av_free_packet() before
- * calling av_read_frame() again or closing the file.
+ * If AVPacket.destruct is set on the returned packet, then the packet is
+ * allocated dynamically and the user may keep it indefinitely.
+ * Otherwise, if AVPacket.destruct is NULL, the packet data is backed by a
+ * static storage somewhere inside the demuxer and the packet is only valid
+ * until the next av_read_frame() call or closing the file. If the caller
+ * requires a longer lifetime, av_dup_packet() will make an av_malloc()ed copy
+ * of it.
+ * In both cases, the packet must be freed with av_free_packet() when it is no
+ * longer needed.
  *
  * @section lavf_decoding_seek Seeking
  * @}
@@ -1550,13 +1556,13 @@ int av_read_packet(AVFormatContext *s, AVPacket *pkt);
  * omit invalid data between valid frames so as to give the decoder the maximum
  * information possible for decoding.
  *
- * The returned packet is valid
- * until the next av_read_frame() or until av_close_input_file() and
- * must be freed with av_free_packet. For video, the packet contains
- * exactly one frame. For audio, it contains an integer number of
- * frames if each frame has a known fixed size (e.g. PCM or ADPCM
- * data). If the audio frames have a variable size (e.g. MPEG audio),
- * then it contains one frame.
+ * If pkt->destruct is NULL, then the packet is valid until the next
+ * av_read_frame() or until av_close_input_file(). Otherwise the packet is valid
+ * indefinitely. In both cases the packet must be freed with
+ * av_free_packet when it is no longer needed. For video, the packet contains
+ * exactly one frame. For audio, it contains an integer number of frames if each
+ * frame has a known fixed size (e.g. PCM or ADPCM data). If the audio frames
+ * have a variable size (e.g. MPEG audio), then it contains one frame.
  *
  * pkt->pts, pkt->dts and pkt->duration are always set to correct
  * values in AVStream.time_base units (and guessed if the format cannot
@@ -1724,10 +1730,10 @@ int av_write_frame(AVFormatContext *s, AVPacket *pkt);
  * @param s media file handle
  * @param pkt The packet containing the data to be written. Libavformat takes
  * ownership of the data and will free it when it sees fit using the packet's
- * This can be NULL (at any time, not just at the end), to flush the
- * interleaving queues.
  * @ref AVPacket.destruct "destruct" field. The caller must not access the data
  * after this function returns, as it may already be freed.
+ * This can be NULL (at any time, not just at the end), to flush the
+ * interleaving queues.
  * Packet's @ref AVPacket.stream_index "stream_index" field must be set to the
  * index of the corresponding stream in @ref AVFormatContext.streams
  * "s.streams".
diff --git a/libavutil/crc.c b/libavutil/crc.c
index d640184876..bba8cb10a5 100644
--- a/libavutil/crc.c
+++ b/libavutil/crc.c
@@ -40,22 +40,6 @@ static struct {
 static AVCRC av_crc_table[AV_CRC_MAX][257];
 #endif
 
-/**
- * Initialize a CRC table.
- * @param ctx must be an array of size sizeof(AVCRC)*257 or sizeof(AVCRC)*1024
- * @param le If 1, the lowest bit represents the coefficient for the highest
- *           exponent of the corresponding polynomial (both for poly and
- *           actual CRC).
- *           If 0, you must swap the CRC parameter and the result of av_crc
- *           if you need the standard representation (can be simplified in
- *           most cases to e.g. bswap16):
- *           av_bswap32(crc << (32-bits))
- * @param bits number of bits for the CRC
- * @param poly generator polynomial without the x**bits coefficient, in the
- *             representation as specified by le
- * @param ctx_size size of ctx in bytes
- * @return <0 on failure
- */
 int av_crc_init(AVCRC *ctx, int le, int bits, uint32_t poly, int ctx_size)
 {
     unsigned i, j;
@@ -89,11 +73,6 @@ int av_crc_init(AVCRC *ctx, int le, int bits, uint32_t poly, int ctx_size)
     return 0;
 }
 
-/**
- * Get an initialized standard CRC table.
- * @param crc_id ID of a standard CRC
- * @return a pointer to the CRC table or NULL on failure
- */
 const AVCRC *av_crc_get_table(AVCRCId crc_id)
 {
 #if !CONFIG_HARDCODED_TABLES
@@ -108,13 +87,6 @@ const AVCRC *av_crc_get_table(AVCRCId crc_id)
     return av_crc_table[crc_id];
 }
 
-/**
- * Calculate the CRC of a block.
- * @param crc CRC of previous blocks if any or initial value for CRC
- * @return CRC updated with the data from the given block
- *
- * @see av_crc_init() "le" parameter
- */
 uint32_t av_crc(const AVCRC *ctx, uint32_t crc,
                 const uint8_t *buffer, size_t length)
 {
diff --git a/libavutil/crc.h b/libavutil/crc.h
index 12650542ba..9095ce6c62 100644
--- a/libavutil/crc.h
+++ b/libavutil/crc.h
@@ -36,8 +36,38 @@ typedef enum {
     AV_CRC_MAX,         /*< Not part of public API! Do not use outside libavutil. */
 }AVCRCId;
 
+/**
+ * Initialize a CRC table.
+ * @param ctx must be an array of size sizeof(AVCRC)*257 or sizeof(AVCRC)*1024
+ * @param le If 1, the lowest bit represents the coefficient for the highest
+ *           exponent of the corresponding polynomial (both for poly and
+ *           actual CRC).
+ *           If 0, you must swap the CRC parameter and the result of av_crc
+ *           if you need the standard representation (can be simplified in
+ *           most cases to e.g. bswap16):
+ *           av_bswap32(crc << (32-bits))
+ * @param bits number of bits for the CRC
+ * @param poly generator polynomial without the x**bits coefficient, in the
+ *             representation as specified by le
+ * @param ctx_size size of ctx in bytes
+ * @return <0 on failure
+ */
 int av_crc_init(AVCRC *ctx, int le, int bits, uint32_t poly, int ctx_size);
+
+/**
+ * Get an initialized standard CRC table.
+ * @param crc_id ID of a standard CRC
+ * @return a pointer to the CRC table or NULL on failure
+ */
 const AVCRC *av_crc_get_table(AVCRCId crc_id);
+
+/**
+ * Calculate the CRC of a block.
+ * @param crc CRC of previous blocks if any or initial value for CRC
+ * @return CRC updated with the data from the given block
+ *
+ * @see av_crc_init() "le" parameter
+ */
 uint32_t av_crc(const AVCRC *ctx, uint32_t start_crc, const uint8_t *buffer, size_t length) av_pure;
 
 #endif /* AVUTIL_CRC_H */
diff --git a/libavutil/x86/x86util.asm b/libavutil/x86/x86util.asm
index ade21c68bb..df2ceccdce 100644
--- a/libavutil/x86/x86util.asm
+++ b/libavutil/x86/x86util.asm
@@ -319,6 +319,18 @@
     %endif
 %endmacro
 
+%macro PSWAPD 2
+%if cpuflag(mmxext)
+    pshufw    %1, %2, q1032
+%elif cpuflag(3dnowext)
+    pswapd    %1, %2
+%elif cpuflag(3dnow)
+    movq      %1, %2
+    psrlq     %1, 32
+    punpckldq %1, %2
+%endif
+%endmacro
+
 %macro DEINTB 5 ; mask, reg1, mask, reg2, optional src to fill masks from
 %ifnum %5
     pand   m%3, m%5, m%4 ; src .. y6 .. y4