Correct some wrong uses of LLVM intrinsics

crates/core_arch/src/aarch64/neon/generated.rs

@@ -11903,7 +11903,7 @@ pub unsafe fn vluti4q_lane_s8<const LANE: i32>(a: int8x16_t, b: uint8x8_t) -> in
     unsafe extern "unadjusted" {
         #[cfg_attr(
             any(target_arch = "aarch64", target_arch = "arm64ec"),
-            link_name = "llvm.aarch64.neon.vluti4q.lane.v8i8"
+            link_name = "llvm.aarch64.neon.vluti4q.lane.v16i8"
         )]
         fn _vluti4q_lane_s8(a: int8x16_t, b: uint8x8_t, n: i32) -> int8x16_t;
     }

crates/core_arch/src/aarch64/prefetch.rs

@@ -2,7 +2,7 @@
 use stdarch_test::assert_instr;
 
 unsafe extern "unadjusted" {
-    #[link_name = "llvm.prefetch"]
+    #[link_name = "llvm.prefetch.p0"]
     fn prefetch(p: *const i8, rw: i32, loc: i32, ty: i32);
 }
 

crates/core_arch/src/aarch64/sve/mod.rs

@@ -130,7 +130,7 @@ macro_rules! impl_internal_sve_predicate {
             #[target_feature(enable = "sve")]
             unsafe fn sve_into(self) -> svbool_t {
                 #[allow(improper_ctypes)]
-                unsafe extern "C" {
+                unsafe extern "unadjusted" {
                     #[cfg_attr(
                         target_arch = "aarch64",
                         link_name = concat!("llvm.aarch64.sve.convert.to.svbool.nxv", $elt, "i1")
@@ -147,7 +147,7 @@ macro_rules! impl_internal_sve_predicate {
             #[target_feature(enable = "sve")]
             unsafe fn sve_into(self) -> $name {
                 #[allow(improper_ctypes)]
-                unsafe extern "C" {
+                unsafe extern "unadjusted" {
                     #[cfg_attr(
                         target_arch = "aarch64",
                         link_name = concat!("llvm.aarch64.sve.convert.from.svbool.nxv", $elt, "i1")

crates/core_arch/src/mips/msa.rs

@@ -45,7 +45,7 @@ types! {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.mips.add.a.b"]
     fn msa_add_a_b(a: v16i8, b: v16i8) -> v16i8;
     #[link_name = "llvm.mips.add.a.h"]

crates/core_arch/src/nvptx/mod.rs

@@ -19,9 +19,9 @@ mod packed;
 pub use packed::*;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
-    #[link_name = "llvm.nvvm.barrier0"]
-    fn syncthreads() -> ();
+unsafe extern "unadjusted" {
+    #[link_name = "llvm.nvvm.barrier.cta.sync.aligned.all"]
+    fn syncthreads(a: u32) -> ();
     #[link_name = "llvm.nvvm.read.ptx.sreg.ntid.x"]
     fn block_dim_x() -> u32;
     #[link_name = "llvm.nvvm.read.ptx.sreg.ntid.y"]
@@ -54,7 +54,7 @@ unsafe extern "C" {
 #[inline]
 #[unstable(feature = "stdarch_nvptx", issue = "111199")]
 pub unsafe fn _syncthreads() -> () {
-    syncthreads()
+    syncthreads(0)
 }
 
 /// x-th thread-block dimension.

crates/core_arch/src/nvptx/packed.rs

@@ -7,7 +7,7 @@
 use crate::intrinsics::simd::*;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.minimum.v2f16"]
     fn llvm_f16x2_minimum(a: f16x2, b: f16x2) -> f16x2;
     #[link_name = "llvm.maximum.v2f16"]

crates/core_arch/src/powerpc/altivec.rs

@@ -96,7 +96,7 @@ impl From<vector_bool_int> for m32x4 {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.ppc.altivec.lvx"]
     fn lvx(p: *const i8) -> vector_unsigned_int;
 

crates/core_arch/src/powerpc/vsx.rs

@@ -52,7 +52,7 @@ impl From<vector_bool_long> for m64x2 {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.ppc.altivec.vperm"]
     fn vperm(
         a: vector_signed_int,

crates/core_arch/src/powerpc64/vsx.rs

@@ -17,7 +17,7 @@ use stdarch_test::assert_instr;
 use crate::mem::transmute;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.ppc.vsx.lxvl"]
     fn lxvl(a: *const u8, l: usize) -> vector_signed_int;
 

crates/core_arch/src/wasm32/memory.rs

@@ -2,9 +2,11 @@
 use stdarch_test::assert_instr;
 
 unsafe extern "unadjusted" {
-    #[link_name = "llvm.wasm.memory.grow"]
+    #[cfg_attr(target_pointer_width = "32", link_name = "llvm.wasm.memory.grow.i32")]
+    #[cfg_attr(target_pointer_width = "64", link_name = "llvm.wasm.memory.grow.i64")]
     fn llvm_memory_grow(mem: u32, pages: usize) -> usize;
-    #[link_name = "llvm.wasm.memory.size"]
+    #[cfg_attr(target_pointer_width = "32", link_name = "llvm.wasm.memory.size.i32")]
+    #[cfg_attr(target_pointer_width = "64", link_name = "llvm.wasm.memory.size.i64")]
     fn llvm_memory_size(mem: u32) -> usize;
 }
 

crates/core_arch/src/x86/aes.rs

@@ -13,7 +13,7 @@ use crate::core_arch::x86::__m128i;
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.aesni.aesdec"]
     fn aesdec(a: __m128i, round_key: __m128i) -> __m128i;
     #[link_name = "llvm.x86.aesni.aesdeclast"]

crates/core_arch/src/x86/avx.rs

@@ -3292,7 +3292,7 @@ pub const fn _mm256_cvtss_f32(a: __m256) -> f32 {
 
 // LLVM intrinsics used in the above functions
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx.round.pd.256"]
     fn roundpd256(a: __m256d, b: i32) -> __m256d;
     #[link_name = "llvm.x86.avx.round.ps.256"]

crates/core_arch/src/x86/avx2.rs

@@ -3906,7 +3906,7 @@ pub const fn _mm256_extract_epi16<const INDEX: i32>(a: __m256i) -> i32 {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx2.pmadd.wd"]
     fn pmaddwd(a: i16x16, b: i16x16) -> i32x8;
     #[link_name = "llvm.x86.avx2.pmadd.ub.sw"]

crates/core_arch/src/x86/avx512bf16.rs

@@ -9,7 +9,7 @@ use crate::intrinsics::simd::*;
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512bf16.cvtne2ps2bf16.128"]
     fn cvtne2ps2bf16(a: f32x4, b: f32x4) -> i16x8;
     #[link_name = "llvm.x86.avx512bf16.cvtne2ps2bf16.256"]

crates/core_arch/src/x86/avx512bitalg.rs

@@ -27,7 +27,7 @@ use crate::mem::transmute;
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.vpshufbitqmb.512"]
     fn bitshuffle_512(data: i8x64, indices: i8x64) -> __mmask64;
     #[link_name = "llvm.x86.avx512.vpshufbitqmb.256"]

crates/core_arch/src/x86/avx512bw.rs

@@ -12764,7 +12764,7 @@ pub unsafe fn _mm_mask_cvtusepi16_storeu_epi8(mem_addr: *mut i8, k: __mmask8, a:
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.pmul.hr.sw.512"]
     fn vpmulhrsw(a: i16x32, b: i16x32) -> i16x32;
 

crates/core_arch/src/x86/avx512cd.rs

@@ -563,7 +563,7 @@ pub const fn _mm_maskz_lzcnt_epi64(k: __mmask8, a: __m128i) -> __m128i {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.conflict.d.512"]
     fn vpconflictd(a: i32x16) -> i32x16;
     #[link_name = "llvm.x86.avx512.conflict.d.256"]

crates/core_arch/src/x86/avx512dq.rs

@@ -7235,7 +7235,7 @@ pub fn _mm_mask_fpclass_ss_mask<const IMM8: i32>(k1: __mmask8, a: __m128) -> __m
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.sitofp.round.v2f64.v2i64"]
     fn vcvtqq2pd_128(a: i64x2, rounding: i32) -> f64x2;
     #[link_name = "llvm.x86.avx512.sitofp.round.v4f64.v4i64"]

crates/core_arch/src/x86/avx512f.rs

@@ -44215,7 +44215,7 @@ pub const _MM_PERM_DDDC: _MM_PERM_ENUM = 0xFE;
 pub const _MM_PERM_DDDD: _MM_PERM_ENUM = 0xFF;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.sqrt.ps.512"]
     fn vsqrtps(a: f32x16, rounding: i32) -> f32x16;
     #[link_name = "llvm.x86.avx512.sqrt.pd.512"]

crates/core_arch/src/x86/avx512ifma.rs

@@ -347,7 +347,7 @@ pub fn _mm_maskz_madd52lo_epu64(k: __mmask8, a: __m128i, b: __m128i, c: __m128i)
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.vpmadd52l.uq.128"]
     fn vpmadd52luq_128(z: __m128i, x: __m128i, y: __m128i) -> __m128i;
     #[link_name = "llvm.x86.avx512.vpmadd52h.uq.128"]

crates/core_arch/src/x86/avx512vbmi.rs

@@ -453,7 +453,7 @@ pub fn _mm_maskz_multishift_epi64_epi8(k: __mmask16, a: __m128i, b: __m128i) ->
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.vpermi2var.qi.512"]
     fn vpermi2b(a: i8x64, idx: i8x64, b: i8x64) -> i8x64;
     #[link_name = "llvm.x86.avx512.vpermi2var.qi.256"]

crates/core_arch/src/x86/avx512vbmi2.rs

@@ -2383,7 +2383,7 @@ pub const fn _mm_maskz_shrdi_epi16<const IMM8: i32>(
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.mask.compress.store.w.512"]
     fn vcompressstorew(mem: *mut i8, data: i16x32, mask: u32);
     #[link_name = "llvm.x86.avx512.mask.compress.store.w.256"]

crates/core_arch/src/x86/avx512vnni.rs

@@ -873,7 +873,7 @@ pub fn _mm256_dpwuuds_epi32(src: __m256i, a: __m256i, b: __m256i) -> __m256i {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.vpdpwssd.512"]
     fn vpdpwssd(src: i32x16, a: i16x32, b: i16x32) -> i32x16;
     #[link_name = "llvm.x86.avx512.vpdpwssd.256"]

crates/core_arch/src/x86/avx512vp2intersect.rs

@@ -110,7 +110,7 @@ pub unsafe fn _mm512_2intersect_epi64(
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.avx512.vp2intersect.d.128"]
     fn vp2intersectd_128(a: i32x4, b: i32x4) -> (u8, u8);
     #[link_name = "llvm.x86.avx512.vp2intersect.q.128"]

crates/core_arch/src/x86/avxneconvert.rs

@@ -176,7 +176,7 @@ pub fn _mm256_cvtneps_avx_pbh(a: __m256) -> __m128bh {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.vbcstnebf162ps128"]
     fn bcstnebf162ps_128(a: *const bf16) -> __m128;
     #[link_name = "llvm.x86.vbcstnebf162ps256"]

crates/core_arch/src/x86/bmi1.rs

@@ -131,7 +131,7 @@ pub const fn _mm_tzcnt_32(x: u32) -> i32 {
     x.trailing_zeros() as i32
 }
 
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.bmi.bextr.32"]
     fn x86_bmi_bextr_32(x: u32, y: u32) -> u32;
 }

crates/core_arch/src/x86/bmi2.rs

@@ -67,7 +67,7 @@ pub fn _pext_u32(a: u32, mask: u32) -> u32 {
     unsafe { x86_bmi2_pext_32(a, mask) }
 }
 
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.bmi.bzhi.32"]
     fn x86_bmi2_bzhi_32(x: u32, y: u32) -> u32;
     #[link_name = "llvm.x86.bmi.pdep.32"]

crates/core_arch/src/x86/fxsr.rs

@@ -4,7 +4,7 @@
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.fxsave"]
     fn fxsave(p: *mut u8);
     #[link_name = "llvm.x86.fxrstor"]

crates/core_arch/src/x86/gfni.rs

@@ -23,7 +23,7 @@ use crate::mem::transmute;
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.vgf2p8affineinvqb.512"]
     fn vgf2p8affineinvqb_512(x: i8x64, a: i8x64, imm8: u8) -> i8x64;
     #[link_name = "llvm.x86.vgf2p8affineinvqb.256"]

crates/core_arch/src/x86/pclmulqdq.rs

@@ -11,7 +11,7 @@ use crate::core_arch::x86::__m128i;
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.pclmulqdq"]
     fn pclmulqdq(a: __m128i, round_key: __m128i, imm8: u8) -> __m128i;
 }

crates/core_arch/src/x86/rtm.rs

@@ -16,7 +16,7 @@
 #[cfg(test)]
 use stdarch_test::assert_instr;
 
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.xbegin"]
     fn x86_xbegin() -> i32;
     #[link_name = "llvm.x86.xend"]

crates/core_arch/src/x86/sha.rs

@@ -1,7 +1,7 @@
 use crate::core_arch::{simd::*, x86::*};
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.sha1msg1"]
     fn sha1msg1(a: i32x4, b: i32x4) -> i32x4;
     #[link_name = "llvm.x86.sha1msg2"]

crates/core_arch/src/x86/sse.rs

@@ -1987,7 +1987,7 @@ pub const fn _MM_TRANSPOSE4_PS(
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.sse.rcp.ss"]
     fn rcpss(a: __m128) -> __m128;
     #[link_name = "llvm.x86.sse.rcp.ps"]
@@ -2040,7 +2040,7 @@ unsafe extern "C" {
     fn stmxcsr(p: *mut i8);
     #[link_name = "llvm.x86.sse.ldmxcsr"]
     fn ldmxcsr(p: *const i8);
-    #[link_name = "llvm.prefetch"]
+    #[link_name = "llvm.prefetch.p0"]
     fn prefetch(p: *const i8, rw: i32, loc: i32, ty: i32);
     #[link_name = "llvm.x86.sse.cmp.ss"]
     fn cmpss(a: __m128, b: __m128, imm8: i8) -> __m128;

crates/core_arch/src/x86/sse2.rs

@@ -3236,7 +3236,7 @@ pub const fn _mm_unpacklo_pd(a: __m128d, b: __m128d) -> __m128d {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.sse2.pause"]
     fn pause();
     #[link_name = "llvm.x86.sse2.clflush"]

crates/core_arch/src/x86/sse3.rs

@@ -178,7 +178,7 @@ pub const fn _mm_moveldup_ps(a: __m128) -> __m128 {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.sse3.ldu.dq"]
     fn lddqu(mem_addr: *const i8) -> i8x16;
 }

crates/core_arch/src/x86/sse41.rs

@@ -1181,7 +1181,7 @@ pub unsafe fn _mm_stream_load_si128(mem_addr: *const __m128i) -> __m128i {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.sse41.insertps"]
     fn insertps(a: __m128, b: __m128, imm8: u8) -> __m128;
     #[link_name = "llvm.x86.sse41.dppd"]

crates/core_arch/src/x86/sse42.rs

@@ -569,7 +569,7 @@ pub const fn _mm_cmpgt_epi64(a: __m128i, b: __m128i) -> __m128i {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     // SSE 4.2 string and text comparison ops
     #[link_name = "llvm.x86.sse42.pcmpestrm128"]
     fn pcmpestrm128(a: i8x16, la: i32, b: i8x16, lb: i32, imm8: i8) -> u8x16;

crates/core_arch/src/x86/sse4a.rs

@@ -6,7 +6,7 @@ use crate::core_arch::{simd::*, x86::*};
 use stdarch_test::assert_instr;
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.sse4a.extrq"]
     fn extrq(x: i64x2, y: i8x16) -> i64x2;
     #[link_name = "llvm.x86.sse4a.extrqi"]

crates/core_arch/src/x86/ssse3.rs

@@ -345,7 +345,7 @@ pub fn _mm_sign_epi32(a: __m128i, b: __m128i) -> __m128i {
 }
 
 #[allow(improper_ctypes)]
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.ssse3.pshuf.b.128"]
     fn pshufb128(a: u8x16, b: u8x16) -> u8x16;
 

crates/core_arch/src/x86/tbm.rs

@@ -13,7 +13,7 @@
 #[cfg(test)]
 use stdarch_test::assert_instr;
 
-unsafe extern "C" {
+unsafe extern "unadjusted" {
     #[link_name = "llvm.x86.tbm.bextri.u32"]
     fn bextri_u32(a: u32, control: u32) -> u32;
 }