[macos, dynarmic] Implement VectorMin/Max{S,U}64 emitters (Apple Silicon fix) (#3421)

[macos] dynarmic: Implement VectorMin/Max{S,U}64 emitters (Apple Silicon fix)

Implements the missing A64 backend emitters for VectorMinS64, VectorMinU64, VectorMaxS64, and VectorMaxU64.

These IR opcodes are generated by the optimizer but lack direct hardware instruction support for 64-bit elements in the base NEON set (e.g., UMIN.2D does not exist). They are implemented using a compare (CMGT/CMHI) followed by a bitwise select (BSL). This correctly selects between the two source registers, whereas using BIT would incorrectly zero out elements.

This implementation is guarded by #ifdef __APPLE__ to ensure no impact on other platforms.

Unit tests could not be added to a64.cpp because UMIN.2D is not a valid A64 instruction, causing the assembler (Oaknut) to reject it. The fix was verified by running Team Sonic Racing, which previously crashed on this synthetic opcode.

Fixes crash in Team Sonic Racing on macOS.

Co-authored-by: crueter <[email protected]>
Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/3421
Reviewed-by: Lizzie <[email protected]>
Reviewed-by: crueter <[email protected]>
This commit is contained in:
rayman30andcrueter authored and crueter committed 2026-05-03 04:29:49 +02:00
1 parent 7d0e79335e
commit 44fa2805d6
2 files changed
+45 -18

No files matched your search

@@ -866,10 +866,10 @@ void EmitIR<IR::Opcode::VectorMaxS32>(oaknut::CodeGenerator& code, EmitContext&
template<>
void EmitIR<IR::Opcode::VectorMaxS64>(oaknut::CodeGenerator& code, EmitContext& ctx, IR::Inst* inst) {
(void)code;
(void)ctx;
(void)inst;
UNREACHABLE();
EmitThreeOp(code, ctx, inst, [&](auto& Qresult, auto& Qa, auto& Qb) {
code.CMGT(Qresult->D2(), Qa->D2(), Qb->D2());
code.BSL(Qresult->B16(), Qa->B16(), Qb->B16());
});
}
template<>
@@ -889,10 +889,10 @@ void EmitIR<IR::Opcode::VectorMaxU32>(oaknut::CodeGenerator& code, EmitContext&
template<>
void EmitIR<IR::Opcode::VectorMaxU64>(oaknut::CodeGenerator& code, EmitContext& ctx, IR::Inst* inst) {
(void)code;
(void)ctx;
(void)inst;
UNREACHABLE();
EmitThreeOp(code, ctx, inst, [&](auto& Qresult, auto& Qa, auto& Qb) {
code.CMHI(Qresult->D2(), Qa->D2(), Qb->D2());
code.BSL(Qresult->B16(), Qa->B16(), Qb->B16());
});
}
template<>
@@ -912,10 +912,10 @@ void EmitIR<IR::Opcode::VectorMinS32>(oaknut::CodeGenerator& code, EmitContext&
template<>
void EmitIR<IR::Opcode::VectorMinS64>(oaknut::CodeGenerator& code, EmitContext& ctx, IR::Inst* inst) {
(void)code;
(void)ctx;
(void)inst;
UNREACHABLE();
EmitThreeOp(code, ctx, inst, [&](auto& Qresult, auto& Qa, auto& Qb) {
code.CMGT(Qresult->D2(), Qb->D2(), Qa->D2());
code.BSL(Qresult->B16(), Qa->B16(), Qb->B16());
});
}
template<>
@@ -935,10 +935,10 @@ void EmitIR<IR::Opcode::VectorMinU32>(oaknut::CodeGenerator& code, EmitContext&
template<>
void EmitIR<IR::Opcode::VectorMinU64>(oaknut::CodeGenerator& code, EmitContext& ctx, IR::Inst* inst) {
(void)code;
(void)ctx;
(void)inst;
UNREACHABLE();
EmitThreeOp(code, ctx, inst, [&](auto& Qresult, auto& Qa, auto& Qb) {
code.CMHI(Qresult->D2(), Qb->D2(), Qa->D2());
code.BSL(Qresult->B16(), Qa->B16(), Qb->B16());
});
}
template<>
+29 -2
View File
@@ -1787,7 +1787,7 @@ TEST_CASE("A64: rand3", "[a64]") {
A64::Jit jit{config};
memset(backing_memory, 0, memory_size);
// cat rand2.txt | awk '{print "env.code_mem.emplace_back(0x"$2"); // "$0}' > rand2-out.txt
// cat rand2.txt | awk '{print "env.code_mem.emplace_back(0x"$2"); // "$0}' > rand2-out.tx
env.MemoryWrite32(100, 0x58028edd); // 0000000000000084 58028edd ldr x29, #20952
env.MemoryWrite32(104, 0x14000000); // 0000000000000ea4 14000000 b #0
@@ -1821,7 +1821,7 @@ TEST_CASE("A64: rand2", "[a64][.]") {
A64::Jit jit{config};
memset(backing_memory, 0, memory_size);
// cat rand2.txt | awk '{print "env.code_mem.emplace_back(0x"$2"); // "$0}' > rand2-out.txt
// cat rand2.txt | awk '{print "env.code_mem.emplace_back(0x"$2"); // "$0}' > rand2-out.tx
const std::array<uint32_t, 4096> code32 = {0xea80f352, 0x6e65e59d, 0x1e20c343, 0x2e3a7192, 0x2e267249, 0xd500405f, 0x6f01f461, 0x6eb684fc, 0x58028edd, 0x0ea5f5b6, 0x0ea069fb, 0x2e769517, 0x5e066063, 0x1e65c3f5, 0x4f00ff52, 0x93401cf6, 0x1e274248, 0x6f67aaf5, 0x5e0c0782, 0x5ef43f3c, 0x2e6595b7, 0x4e20590f, 0xb35aa451, 0x6ee2c5ed, 0x4e32bf46, 0x2ea1ba8f, 0x2f68a85e, 0x9237d90a, 0x5e23dd10, 0x0e762e32, 0x4e31a8cf, 0xce1f3360, 0x781a4ac0, 0x13834066, 0x5fa8101c, 0x6f7c5594, 0x0e71bb68, 0xbc0b3e8f, 0x785dbbda, 0x6f51e794, 0xce50af75, 0x1ad728ec, 0x6ee0da4c, 0xb84efa14, 0x2eb3f613, 0x4e287ade, 0x4eb8c734, 0x2e83f4e8, 0x0e397c80, 0xd08f93f8, 0xce718e48, 0x0f672a0d, 0x2e9edd40, 0x0e14128b, 0x6f5942e6, 0x8b3a0f03, 0x3c5d16b9, 0x7f7e3743, 0x4f4c54e4, 0x0ea0a9e9, 0x9e59dbe6, 0x6e7ddcd3, 0xcec08377, 0x9ba759f8, 0x2ea5046e, 0x0e24c569, 0xb8979780, 0x4e31b98c, 0x4efe4f46, 0x4ea7c762, 0x7e61c9c6, 0x6e30c880, 0x1ada0c25, 0x4e603a2f, 0xda9d7218, 0x0d40c5d9, 0x5e214b05, 0x9ba9efc5, 0x5e61b81e, 0x6e7bc31c, 0x0e61a163, 0x9e5832d2, 0x4e772248, 0x4e3d17c8, 0x92624f60, 0x7a1a02dc, 0x79891f65, 0x6eb45036, 0x0e321ee8, 0x4e2566f0, 0x4ea02b9b, 0x0f9dcb3d, 0x2e21b9f9, 0x0e21a8c3, 0xda1700bd, 0x6ea0fb38, 0x7e607a0b, 0x72845817, 0x7f61068e, 0x0d60e529, 0x4ea0ca5c, 0x1a94b20f, 0x8b87419d, 0x7ea9ed71, 0x2ea1a86e, 0x4d40c4da, 0x5ea0eada, 0x784ba96e, 0x7eb6ee02, 0x3db1c710, 0x0e217836, 0x7ee0bb96, 0x4e786c08, 0x4e976a08, 0x489ffe86, 0x4e79fc9b, 0x0e21cbce, 0x5ef7fc65, 0x4ea1286d, 0xd29c771e, 0x6f5c2839, 0x0ea00a9d, 0x6ee44c06, 0x5ee1d858, 0x5ef2fda6, 0x7eb0c9fe, 0x7f762791, 0x2e212ae6, 0x4e61c9db, 0x13003c57, 0x5ee1b8f8, 0x0f2396d2, 0x6ea0db1e, 0x0e71ba82, 0xab29c807, 0x6ef8f8b3, 0x1f18d4a1, 0x0e261d15, 0x1e290081, 0x1b0c7d12, 0x4e7771c3, 0xf845f1e4, 0x4d40c9e8, 0xce778452, 0x6eb9879d, 0x6e21c93d, 0xcec0829f, 0x52a0969f, 0x1e772b4f, 0x7ee1da88, 0x5f52fe0a, 0x7f3387b1, 0x5e214850, 0x1e65c025, 0x0e2ca294, 0x2e614829, 0x1e640077, 0x9e240048, 0x4ebe9537, 0x9bb7925e, 0x38b669c5, 0x2840d089, 0x6f43e648, 0x2e662d28, 0x4eabaff3, 0x6e734cc7, 0x0e31baee, 0x7ee0d93c, 0x5e282bde, 0x7e21bba4, 0x4e6c75fa, 0x5ac01217, 0x7f4304af, 0x1e7878ed, 0x1ada2196, 0x7ee1aba3, 0x93407f3c, 0x4f6c34eb, 0x6e3447a9, 0x7e7ae545, 0x5e0802bb, 0x6eeae63a, 0x7ee1da62, 0x5e280bb3, 0xf81d4009, 0x1e603b21, 0x5e281a14, 0x6eb0a99b, 0x1e266a25, 0x0d60cafe, 0x0e0b6194, 0x7a4ed2c5, 0x92b762ec, 0x4e6b5749, 0x3c16a6e5, 0x4ea0a92b, 0x0fa58b6a, 0x5f76148c, 0x6e30c95f, 0x1e6540fd, 0x5e28e40f, 0x0d403fd4, 0x7e30da36, 0x7fda9b51, 0x2ea04bde, 0x1e25c3d2, 0x1ee0434c, 0x5e21d8e7, 0x5ee1ba51, 0x5e61aba9, 0x4e2849fb, 0x5ee098ea, 0x4e60f63d, 0x0f280443, 0x5ee0da27, 0x2e78a6ce, 0x78054afc, 0x4e14286b, 0x4e218bd8, 0x2a3d2551, 0x3a04017a, 0x5f4317cd, 0x0e604a37, 0x9a834614, 0x0e2edf4d, 0x7a51a0a0, 0x5f8e9043, 0x6ea06bb2, 0xaa2857dd, 0x7a1903fc, 0x301ba9ba, 0x9ac929cd, 0x4e061ff0, 0x2e38fcfc, 0x0e2f614a, 0x7ee0d8e4, 0x6e73afda, 0x7f4156f7, 0x0e6078bf, 0x4ee1d9ed, 0x93403fbe, 0xce6f8640, 0x4e3855e3, 0x6f76fe23, 0x112466e8, 0x1e358a90, 0x7f45272c, 0x6ea19a9d, 0x8a696350, 0x1e3900f6, 0x5e61c866, 0x0e3fbfd0, 0x5ee09ad0, 0x0e651d27, 0x4dffc35e, 0x2e20c6ce, 0x0fbe118d, 0x1e656a15, 0xd1357365, 0x0e20a847, 0xce4a835c, 0x4e203905, 0x2e60090d, 0x7f4a27bb, 0x1e64c316, 0xce7d86a4, 0x7ebded2d, 0x6e70a97e, 0x4eb9a42b, 0x0e209bef, 0x6f151730, 0x0e7e30f7, 0x4e724509, 0xd503375f, 0xce58b6ae, 0x5e21a9b8, 0xcb2ca538, 0x5ac01131, 0x6ea19a24, 0xeb40c8b3, 0xc8df7d65, 0x78108341, 0x3218ab9b, 0x0f3da7dd, 0x2e003089, 0x4e21cab5, 0x8aa5c924, 0x1a94950c, 0x123e506f, 0x13117e37, 0x1ee6005b, 0x5ac00647, 0x5eec8cd5, 0x7ef0fb3d, 0x9223272a, 0x5ee0cb02, 0x6e66071d, 0x6ea1dbbf, 0x5e61c903, 0x5ac015ea, 0x93db6206, 0x7e62b5e3, 0x6ea0c87b, 0xdac0090e, 0x48df7d90, 0x6e206ba5, 0x9e2503c2, 0x6e25fc89, 0x4d60e2db, 0x1e3e22a0, 0x2eb81c19, 0x7856ea00, 0x5fbfb22d, 0x1e630244, 0x4e202a83, 0x1f50a722, 0x7f7b55d2, 0x0fae89b9, 0x4e781d73, 0xce738c3a, 0x4f15a591, 0x6e21c7e1, 0x586ff77e, 0x8a5d3592, 0x93401c67, 0x5e61cb86, 0xce6bc2c1, 0x6e393f10, 0x9bb70ec3, 0xdac0098c, 0x4da84b95, 0x7f494476, 0x9ace5c11, 0x7e61ca14, 0x4f7a60ef, 0x1ad32b39, 0x0ea3777f, 0x5e61da7f, 0x4f1404e2, 0x4e3244e2, 0x6e1b1ceb, 0x0dee5aac, 0x4e2f9dc4, 0x5ea1b8c3, 0x1e59f863, 0xd500403f, 0x4e3ae7d0, 0x4ef5c6ea, 0x08dffe3b, 0x6e36f4f6, 0x2e764f29, 0x0e726f23, 0x5f42375b, 0x7f71fc40, 0x6e618aad, 0x93403e5b, 0x0e205976, 0x0e7250c4, 0x6eb0abc9, 0x2e2049f0, 0x5f14754d, 0x7f6ce468, 0x6f950bbe, 0x6e31aa47, 0x4eb83396, 0x0dccc952, 0x2ea1ca90, 0xce69c701, 0xb0bed69e, 0x7c5dec39, 0x4e2868a2, 0x0e591b08, 0x5f34e6dd, 0x3a449184, 0x5e3ce6de, 0x4ea149b7, 0x4e7ad29b, 0xba198503, 0x1f683e8f, 0xfa52f2a7, 0x6e30dffc, 0x4e6c3d17, 0x2eae3248, 0xd503349f, 0x1e60002c, 0x0f180680, 0x9e240049, 0x6f75774e, 0xa90d8678, 0x9ad924c4, 0x7eb0f85b, 0x0e205aaf, 0x7ee08899, 0x5f4bffd8, 0x1b0ff5f3, 0x4ee11dcd, 0x2e218948, 0x0dcb2733, 0x4eac107c, 0x4ea04a53, 0x4e287b44, 0x0e60b82a, 0x5ee0ebbc, 0xce454ff1, 0x5e1761e7, 0x5e09202f, 0x0e0c0754, 0x1e72e6b9, 0x7e21da70, 0x0fbdb20c, 0x5efb8c84, 0xd500401f, 0x3a4752Line truncated
for (size_t i = 0; i < code32.size(); ++i)
env.MemoryWrite32(100 + i, code32[i]);
@@ -2576,3 +2576,30 @@ TEST_CASE("A64: CLZ{W}", "[a64]") {
REQUIRE(jit.GetRegister(1) == 4);
REQUIRE(jit.GetRegister(2) == 5);
}
TEST_CASE("A64: Manual Vector Min/Max U64 (Optimizer Test)", "[a64]") {
A64TestEnv env;
A64::UserConfig jit_user_config{};
jit_user_config.callbacks = &env;
A64::Jit jit{jit_user_config};
oaknut::VectorCodeGenerator code{env.code_mem, nullptr};
// MaxU64 pattern: (a > b) ? a : b
code.CMHI(V2.D2(), V0.D2(), V1.D2()); // V2 = Mask (A > B)
code.BSL(V2.B16(), V0.B16(), V1.B16()); // V2 = Resul
// MinU64 pattern: (a > b) ? b : a
code.CMHI(V3.D2(), V0.D2(), V1.D2()); // V3 = Mask (A > B)
code.BSL(V3.B16(), V1.B16(), V0.B16()); // V3 = Resul
jit.SetPC(0);
jit.SetVector(0, {100, 20});
jit.SetVector(1, {50, 200});
env.ticks_left = env.code_mem.size();
CheckedRun([&]() { jit.Run(); });
CHECK(jit.GetVector(2) == Vector{100, 200});
CHECK(jit.GetVector(3) == Vector{50, 20});
}