diff options
| author | Niels Möller <nisse@lysator.liu.se> | 2020-11-30 20:23:01 +0100 |
|---|---|---|
| committer | Niels Möller <nisse@lysator.liu.se> | 2020-11-30 20:23:01 +0100 |
| commit | efdd88ec59d2653bda53070448c8eb1889ca3544 (patch) | |
| tree | f6f7a29b911fe6c75fefcea84a2d55d1242647ba | |
| parent | 42443755a94ccb0c99ce5dde6432ad4311c1ca5d (diff) | |
| download | nettle-efdd88ec59d2653bda53070448c8eb1889ca3544.tar.gz | |
ppc: Add byte-swapping to chacha_4core, for big-endian builds.
| -rw-r--r-- | powerpc64/p7/chacha-4core.asm | 13 |
1 files changed, 13 insertions, 0 deletions
diff --git a/powerpc64/p7/chacha-4core.asm b/powerpc64/p7/chacha-4core.asm index 49a801be..29d3f3c1 100644 --- a/powerpc64/p7/chacha-4core.asm +++ b/powerpc64/p7/chacha-4core.asm @@ -184,6 +184,19 @@ C Load state and splat vadduwm v10, v10, T2 vadduwm v11, v11, T2 +IF_BE(` + C Output always stored in little-endian byte order. + C Can reuse T0 and T1 to construct permutation mask. + li r9, 0 + lvsl T0, r9, r9 C 00 01 02 03 ... 0c 0d 0e 0f + vspltisb T1, 0x03 C 03 03 03 03 ... 03 03 03 03 + vxor T1, T1, T0 C 03 02 01 00 ... 0f 0e 0d 0c + + forloop(i, 0, 15, ` + vperm `v'i, `v'i, `v'i, T1 + ') +') + stxvw4x VSR(v0), 0, DST stxvw4x VSR(v4), r6, DST stxvw4x VSR(v8), r7, DST |