Merge remote-tracking branch 'qatar/master'

* qatar/master: (36 commits)
  ARM: allow unaligned buffer in fixed-point NEON FFT4
  fate: test more FFT etc sizes
  dca: set AVCodecContext frame_size for DTS audio
  YASM: Shut up unused variable compiler warning with --disable-yasm.
  x86_32: Fix build on x86_32 with --disable-yasm.
  iirfilter: add fate test
  doxygen: Add qmul docs.
  ogg: propagate return values and return more meaningful error values
  H.264: fix overreads of qscale_table
  Remove unused static tables and static inline functions.
  eval: clear Parser instances before using
  dct-test: remove 'ref' function pointer from tables
  build: Remove deleted 'check' target from .PHONY list.
  oggdec: Abort Ogg header parsing when encountering a data packet.
  Add LGPL license boilerplate to files lacking it.
  mxfenc: small typo fix
  doxygen: Fix documentation for some VP8 functions.
  sha: use AV_RB32() instead of assuming buffer can be cast to uint32_t*
  des: allow unaligned input and output buffers
  aes: allow unaligned input and output buffers
  ...

Conflicts:
	libavcodec/dct-test.c
	libavcodec/libvpxenc.c
	libavcodec/x86/dsputil_mmx.c
	libavcodec/x86/h264_qpel_mmx.c
	libavfilter/x86/gradfun.c
	libavformat/oggdec.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

1 files changed, 26 insertions, 15 deletions

diff --git a/libavutil/aes.c b/libavutil/aes.c
index 7c92a2757f..49093efc53 100644
--- a/libavutil/aes.c
+++ b/libavutil/aes.c
@@ -22,6 +22,7 @@
 
 #include "common.h"
 #include "aes.h"
+#include "intreadwrite.h"
 
 typedef union {
     uint64_t u64[2];
@@ -67,6 +68,20 @@ static inline void addkey(av_aes_block *dst, const av_aes_block *src,
     dst->u64[1] = src->u64[1] ^ round_key->u64[1];
 }
 
+static inline void addkey_s(av_aes_block *dst, const uint8_t *src,
+                            const av_aes_block *round_key)
+{
+    dst->u64[0] = AV_RN64(src)     ^ round_key->u64[0];
+    dst->u64[1] = AV_RN64(src + 8) ^ round_key->u64[1];
+}
+
+static inline void addkey_d(uint8_t *dst, const av_aes_block *src,
+                            const av_aes_block *round_key)
+{
+    AV_WN64(dst,     src->u64[0] ^ round_key->u64[0]);
+    AV_WN64(dst + 8, src->u64[1] ^ round_key->u64[1]);
+}
+
 static void subshift(av_aes_block s0[2], int s, const uint8_t *box)
 {
     av_aes_block *s1 = (av_aes_block *) (s0[0].u8 - s);
@@ -119,32 +134,28 @@ static inline void crypt(AVAES *a, int s, const uint8_t *sbox,
     subshift(&a->state[0], s, sbox);
 }
 
-void av_aes_crypt(AVAES *a, uint8_t *dst_, const uint8_t *src_,
-                  int count, uint8_t *iv_, int decrypt)
+void av_aes_crypt(AVAES *a, uint8_t *dst, const uint8_t *src,
+                  int count, uint8_t *iv, int decrypt)
 {
-    av_aes_block       *dst = (av_aes_block *) dst_;
-    const av_aes_block *src = (const av_aes_block *) src_;
-    av_aes_block        *iv = (av_aes_block *) iv_;
-
     while (count--) {
-        addkey(&a->state[1], src, &a->round_key[a->rounds]);
+        addkey_s(&a->state[1], src, &a->round_key[a->rounds]);
         if (decrypt) {
             crypt(a, 0, inv_sbox, dec_multbl);
             if (iv) {
-                addkey(&a->state[0], &a->state[0], iv);
-                *iv = *src;
+                addkey_s(&a->state[0], iv, &a->state[0]);
+                memcpy(iv, src, 16);
             }
-            addkey(dst, &a->state[0], &a->round_key[0]);
+            addkey_d(dst, &a->state[0], &a->round_key[0]);
         } else {
             if (iv)
-                addkey(&a->state[1], &a->state[1], iv);
+                addkey_s(&a->state[1], iv, &a->state[1]);
             crypt(a, 2, sbox, enc_multbl);
-            addkey(dst, &a->state[0], &a->round_key[0]);
+            addkey_d(dst, &a->state[0], &a->round_key[0]);
             if (iv)
-                *iv = *dst;
+                memcpy(iv, dst, 16);
         }
-        src++;
-        dst++;
+        src += 16;
+        dst += 16;
     }
 }