diff options
| author | Peter Maydell | 2020-11-02 17:52:15 +0100 |
|---|---|---|
| committer | Peter Maydell | 2020-11-02 17:52:15 +0100 |
| commit | 552714c0812a10e5cff239bd29928e5fcb8d8b3b (patch) | |
| tree | 8a172f5921689dcdb7cc542f851aa70bc4c4d3ca /target/arm/vec_helper.c | |
| parent | target/arm: Improve do_prewiden_3d (diff) | |
| download | qemu-552714c0812a10e5cff239bd29928e5fcb8d8b3b.tar.gz qemu-552714c0812a10e5cff239bd29928e5fcb8d8b3b.tar.xz qemu-552714c0812a10e5cff239bd29928e5fcb8d8b3b.zip | |
target/arm: Fix float16 pairwise Neon ops on big-endian hosts
In the neon_padd/pmax/pmin helpers for float16, a cut-and-paste error
meant we were using the H4() address swizzler macro rather than the
H2() which is required for 2-byte data. This had no effect on
little-endian hosts but meant we put the result data into the
destination Dreg in the wrong order on big-endian hosts.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20201028191712.4910-2-peter.maydell@linaro.org
Diffstat (limited to 'target/arm/vec_helper.c')
| -rw-r--r-- | target/arm/vec_helper.c | 8 |
1 files changed, 4 insertions, 4 deletions
diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c index a973454e4f..30d76d05be 100644 --- a/target/arm/vec_helper.c +++ b/target/arm/vec_helper.c @@ -1858,10 +1858,10 @@ DO_ABA(gvec_uaba_d, uint64_t) r2 = float16_##OP(m[H2(0)], m[H2(1)], fpst); \ r3 = float16_##OP(m[H2(2)], m[H2(3)], fpst); \ \ - d[H4(0)] = r0; \ - d[H4(1)] = r1; \ - d[H4(2)] = r2; \ - d[H4(3)] = r3; \ + d[H2(0)] = r0; \ + d[H2(1)] = r1; \ + d[H2(2)] = r2; \ + d[H2(3)] = r3; \ } DO_NEON_PAIRWISE(neon_padd, add) |