haraldwolff/qemu-patch-raspberry4
1
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

target-arm: Don't overflow when calculating value for signed VABAL

Browse source 
In the VABAL instruction we take the absolute difference of two
values of size x and store it in a result of size 2x. This means
we have to be careful to calculate the absolute difference using
a wide enough type that we don't accidentally overflow.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
This commit is contained in:
Peter Maydell 2011-04-11 16:32:08 +01:00 committed by Aurelien Jarno
parent cc9453f457
commit 4d9ad7f793
1 changed files with 21 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

38
target-arm/neon_helper.c
View file

@ -1514,9 +1514,13 @@ uint64_t HELPER(neon_addl_saturate_s64)(uint64_t a, uint64_t b)
return result; return result;
} }
#define DO_ABD(dest, x, y, type) do { \ /* We have to do the arithmetic in a larger type than
type tmp_x = x; \ * the input type, because for example with a signed 32 bit
type tmp_y = y; \ * op the absolute difference can overflow a signed 32 bit value.
*/
#define DO_ABD(dest, x, y, intype, arithtype) do { \
arithtype tmp_x = (intype)(x); \
arithtype tmp_y = (intype)(y); \
dest = ((tmp_x > tmp_y) ? tmp_x - tmp_y : tmp_y - tmp_x); \ dest = ((tmp_x > tmp_y) ? tmp_x - tmp_y : tmp_y - tmp_x); \
} while(0) } while(0)
@ -1524,12 +1528,12 @@ uint64_t HELPER(neon_abdl_u16)(uint32_t a, uint32_t b)
{ {
uint64_t tmp; uint64_t tmp;
uint64_t result; uint64_t result;
DO_ABD(result, a, b, uint8_t); DO_ABD(result, a, b, uint8_t, uint32_t);
DO_ABD(tmp, a >> 8, b >> 8, uint8_t); DO_ABD(tmp, a >> 8, b >> 8, uint8_t, uint32_t);
result |= tmp << 16; result |= tmp << 16;
DO_ABD(tmp, a >> 16, b >> 16, uint8_t); DO_ABD(tmp, a >> 16, b >> 16, uint8_t, uint32_t);
result |= tmp << 32; result |= tmp << 32;
DO_ABD(tmp, a >> 24, b >> 24, uint8_t); DO_ABD(tmp, a >> 24, b >> 24, uint8_t, uint32_t);
result |= tmp << 48; result |= tmp << 48;
return result; return result;
} }
@ -1538,12 +1542,12 @@ uint64_t HELPER(neon_abdl_s16)(uint32_t a, uint32_t b)
{ {
uint64_t tmp; uint64_t tmp;
uint64_t result; uint64_t result;
DO_ABD(result, a, b, int8_t); DO_ABD(result, a, b, int8_t, int32_t);
DO_ABD(tmp, a >> 8, b >> 8, int8_t); DO_ABD(tmp, a >> 8, b >> 8, int8_t, int32_t);
result |= tmp << 16; result |= tmp << 16;
DO_ABD(tmp, a >> 16, b >> 16, int8_t); DO_ABD(tmp, a >> 16, b >> 16, int8_t, int32_t);
result |= tmp << 32; result |= tmp << 32;
DO_ABD(tmp, a >> 24, b >> 24, int8_t); DO_ABD(tmp, a >> 24, b >> 24, int8_t, int32_t);
result |= tmp << 48; result |= tmp << 48;
return result; return result;
} }
@ -1552,8 +1556,8 @@ uint64_t HELPER(neon_abdl_u32)(uint32_t a, uint32_t b)
{ {
uint64_t tmp; uint64_t tmp;
uint64_t result; uint64_t result;
DO_ABD(result, a, b, uint16_t); DO_ABD(result, a, b, uint16_t, uint32_t);
DO_ABD(tmp, a >> 16, b >> 16, uint16_t); DO_ABD(tmp, a >> 16, b >> 16, uint16_t, uint32_t);
return result | (tmp << 32); return result | (tmp << 32);
} }
@ -1561,22 +1565,22 @@ uint64_t HELPER(neon_abdl_s32)(uint32_t a, uint32_t b)
{ {
uint64_t tmp; uint64_t tmp;
uint64_t result; uint64_t result;
DO_ABD(result, a, b, int16_t); DO_ABD(result, a, b, int16_t, int32_t);
DO_ABD(tmp, a >> 16, b >> 16, int16_t); DO_ABD(tmp, a >> 16, b >> 16, int16_t, int32_t);
return result | (tmp << 32); return result | (tmp << 32);
} }
uint64_t HELPER(neon_abdl_u64)(uint32_t a, uint32_t b) uint64_t HELPER(neon_abdl_u64)(uint32_t a, uint32_t b)
{ {
uint64_t result; uint64_t result;
DO_ABD(result, a, b, uint32_t); DO_ABD(result, a, b, uint32_t, uint64_t);
return result; return result;
} }
uint64_t HELPER(neon_abdl_s64)(uint32_t a, uint32_t b) uint64_t HELPER(neon_abdl_s64)(uint32_t a, uint32_t b)
{ {
uint64_t result; uint64_t result;
DO_ABD(result, a, b, int32_t); DO_ABD(result, a, b, int32_t, int64_t);
return result; return result;
} }
#undef DO_ABD #undef DO_ABD