Commit 0d1262a41e019e4511071e339bb8aa018596a1fd
2017-09-24T17:36:47
[sfnt] Fix `premultiply_data' (#52092). * src/sfnt/pngshim.c (premultiply_data): Don't use vector extension if we have less than 16 bytes of data.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92
diff --git a/ChangeLog b/ChangeLog
index d8bd675..169cf05 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2017-09-24 Jonathan Kew <jfkthame@gmail.com>
+
+ [sfnt] Fix `premultiply_data' (#52092).
+
+ * src/sfnt/pngshim.c (premultiply_data): Don't use vector extension
+ if we have less than 16 bytes of data.
+
2017-09-24 Werner Lemberg <wl@gnu.org>
[otvalid] Fix handling of ValueRecords.
diff --git a/src/sfnt/pngshim.c b/src/sfnt/pngshim.c
index 560db48..4c0d007 100644
--- a/src/sfnt/pngshim.c
+++ b/src/sfnt/pngshim.c
@@ -82,42 +82,45 @@
typedef unsigned short v82 __attribute__(( vector_size( 16 ) ));
- /* process blocks of 16 bytes in one rush, which gives a nice speed-up */
- limit = row_info->rowbytes - 16 + 1;
- for ( ; i < limit; i += 16 )
+ if ( row_info->rowbytes > 15 )
{
- unsigned char* base = &data[i];
+ /* process blocks of 16 bytes in one rush, which gives a nice speed-up */
+ limit = row_info->rowbytes - 16 + 1;
+ for ( ; i < limit; i += 16 )
+ {
+ unsigned char* base = &data[i];
- v82 s, s0, s1, a;
+ v82 s, s0, s1, a;
- /* clang <= 3.9 can't apply scalar values to vectors */
- /* (or rather, it needs a different syntax) */
- v82 n0x80 = { 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80 };
- v82 n0xFF = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF };
- v82 n8 = { 8, 8, 8, 8, 8, 8, 8, 8 };
+ /* clang <= 3.9 can't apply scalar values to vectors */
+ /* (or rather, it needs a different syntax) */
+ v82 n0x80 = { 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80, 0x80 };
+ v82 n0xFF = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF };
+ v82 n8 = { 8, 8, 8, 8, 8, 8, 8, 8 };
- v82 ma = { 1, 1, 3, 3, 5, 5, 7, 7 };
- v82 o1 = { 0, 0xFF, 0, 0xFF, 0, 0xFF, 0, 0xFF };
- v82 m0 = { 1, 0, 3, 2, 5, 4, 7, 6 };
+ v82 ma = { 1, 1, 3, 3, 5, 5, 7, 7 };
+ v82 o1 = { 0, 0xFF, 0, 0xFF, 0, 0xFF, 0, 0xFF };
+ v82 m0 = { 1, 0, 3, 2, 5, 4, 7, 6 };
- memcpy( &s, base, 16 ); /* RGBA RGBA RGBA RGBA */
- s0 = s & n0xFF; /* R B R B R B R B */
- s1 = s >> n8; /* G A G A G A G A */
+ memcpy( &s, base, 16 ); /* RGBA RGBA RGBA RGBA */
+ s0 = s & n0xFF; /* R B R B R B R B */
+ s1 = s >> n8; /* G A G A G A G A */
- a = vector_shuffle( s1, ma ); /* A A A A A A A A */
- s1 |= o1; /* G 1 G 1 G 1 G 1 */
- s0 = vector_shuffle( s0, m0 ); /* B R B R B R B R */
+ a = vector_shuffle( s1, ma ); /* A A A A A A A A */
+ s1 |= o1; /* G 1 G 1 G 1 G 1 */
+ s0 = vector_shuffle( s0, m0 ); /* B R B R B R B R */
- s0 *= a;
- s1 *= a;
- s0 += n0x80;
- s1 += n0x80;
- s0 = ( s0 + ( s0 >> n8 ) ) >> n8;
- s1 = ( s1 + ( s1 >> n8 ) ) >> n8;
+ s0 *= a;
+ s1 *= a;
+ s0 += n0x80;
+ s1 += n0x80;
+ s0 = ( s0 + ( s0 >> n8 ) ) >> n8;
+ s1 = ( s1 + ( s1 >> n8 ) ) >> n8;
- s = s0 | ( s1 << n8 );
- memcpy( base, &s, 16 );
+ s = s0 | ( s1 << n8 );
+ memcpy( base, &s, 16 );
+ }
}
#endif /* use `vector_size' */