target-arm: Correct ABD's handling of negative zeroes
authorPeter Maydell <peter.maydell@linaro.org>
Fri, 11 Mar 2011 08:12:24 +0000 (08:12 +0000)
committerAurelien Jarno <aurelien@aurel32.net>
Sun, 3 Apr 2011 15:19:38 +0000 (17:19 +0200)
commit9b234d39f0c0a52bbd05f6322bd3378bc44c4526
tree945bc26f5ce9c2bcdccd0019435732d3b2af3093
parent8160728f3fa8e4d03b2fb254ea4b434e7a91b6c9
target-arm: Correct ABD's handling of negative zeroes

Implement ABD by taking the absolute value of the difference
of the operands (as the ARM ARM specifies) rather than by
flipping the order of the operands to the subtract based
on the results of a comparison. The latter approch gives
the wrong answers for some edge cases like negative zero.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
target-arm/neon_helper.c