Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

aarch64-macos-none baseline #194

Closed
alexrp opened this issue Sep 23, 2024 · 2 comments · Fixed by #198
Closed

aarch64-macos-none baseline #194

alexrp opened this issue Sep 23, 2024 · 2 comments · Fixed by #198

Comments

@alexrp
Copy link
Member

alexrp commented Sep 23, 2024

Using: ./build aarch64-macos-none baseline

/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1028:79: error: always_inline function 'vdupq_n_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint16x8_t equalMask = vreinterpretq_u16_u8(vceqq_u8(chunk, vdupq_n_u8(tag)));
                                                                              ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1028:63: error: always_inline function 'vceqq_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint16x8_t equalMask = vreinterpretq_u16_u8(vceqq_u8(chunk, vdupq_n_u8(tag)));
                                                              ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1028:42: error: always_inline function 'vreinterpretq_u16_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint16x8_t equalMask = vreinterpretq_u16_u8(vceqq_u8(chunk, vdupq_n_u8(tag)));
                                         ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1030:35: error: always_inline function 'vreinterpretq_u32_u16' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint32x4_t t1 = vreinterpretq_u32_u16(vsriq_n_u16(t0, t0, 14));
                                  ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1031:35: error: always_inline function 'vreinterpretq_u64_u32' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint64x2_t t2 = vreinterpretq_u64_u32(vshrq_n_u32(t1, 14));
                                  ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1032:35: error: always_inline function 'vreinterpretq_u8_u64' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t t3 = vreinterpretq_u8_u64(vsraq_n_u64(t2, t2, 28));
                                  ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1038:39: error: always_inline function 'vreinterpretq_u8_u16' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t chunk0 = vreinterpretq_u8_u16(chunk.val[0]);
                                      ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1039:39: error: always_inline function 'vreinterpretq_u8_u16' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t chunk1 = vreinterpretq_u8_u16(chunk.val[1]);
                                      ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1040:60: error: always_inline function 'vdupq_n_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t equalMask0 = vceqq_u8(chunk0, vdupq_n_u8(tag));
                                                           ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1040:43: error: always_inline function 'vceqq_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t equalMask0 = vceqq_u8(chunk0, vdupq_n_u8(tag));
                                          ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1041:60: error: always_inline function 'vdupq_n_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t equalMask1 = vceqq_u8(chunk1, vdupq_n_u8(tag));
                                                           ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1041:43: error: always_inline function 'vceqq_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x16_t equalMask1 = vceqq_u8(chunk1, vdupq_n_u8(tag));
                                          ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1042:47: error: always_inline function 'vreinterpretq_s16_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const int8x8_t pack0 = vqmovn_s16(vreinterpretq_s16_u8(equalMask0));
                                              ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1042:36: error: always_inline function 'vqmovn_s16' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const int8x8_t pack0 = vqmovn_s16(vreinterpretq_s16_u8(equalMask0));
                                   ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1043:47: error: always_inline function 'vreinterpretq_s16_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const int8x8_t pack1 = vqmovn_s16(vreinterpretq_s16_u8(equalMask1));
                                              ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1043:36: error: always_inline function 'vqmovn_s16' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const int8x8_t pack1 = vqmovn_s16(vreinterpretq_s16_u8(equalMask1));
                                   ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1044:34: error: always_inline function 'vreinterpret_u8_s8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x8_t t0 = vreinterpret_u8_s8(pack0);
                                 ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1045:34: error: always_inline function 'vreinterpret_u8_s8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x8_t t1 = vreinterpret_u8_s8(pack1);
                                 ^
/home/alexrp/Source/zig-bootstrap/zstd/lib/compress/zstd_lazy.c:1047:36: error: always_inline function 'vuzp_u8' requires target feature 'aes', but would be inlined into function 'ZSTD_row_getMatchMask' that is compiled without support for 'aes'
            const uint8x8x2_t t3 = vuzp_u8(t2, t0);
                                   ^
:1:1: error: too many errors emitted, stopping now

It works fine when using apple_a14 instead of baseline.

@nektro
Copy link

nektro commented Sep 23, 2024

It works fine when using apple_a14 instead of baseline.

since this is a proprietary operating system and apple_a14 is the oldest chip it has been released with, im not sure what the action item is here

@alexrp
Copy link
Member Author

alexrp commented Sep 23, 2024

baseline is the default CPU targeted by the compiler. It should either just work or something should be changed in the compiler to pick apple_a14 as the actual baseline when targeting aarch64-macos-*. I don't particularly mind which.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants