@cpp-linter/clang-tidy-linux-arm64 18.1.5 → 20.1.0

package/lib/clang/20/include/amxtransposeintrin.h

@@ -0,0 +1,248 @@
+/* ===--- amxtransposeintrin.h - AMX_TRANSPOSE intrinsics -*- C++ -*---------===
+ *
+ * Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+ * See https://llvm.org/LICENSE.txt for license information.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ *
+ * ===-----------------------------------------------------------------------===
+ */
+
+#ifndef __IMMINTRIN_H
+#error "Never use <amxtransposeintrin.h> directly; use <immintrin.h> instead."
+#endif /* __IMMINTRIN_H */
+
+#ifndef __AMX_TRANSPOSEINTRIN_H
+#define __AMX_TRANSPOSEINTRIN_H
+#ifdef __x86_64__
+
+#define __DEFAULT_FN_ATTRS_TRANSPOSE                                           \
+  __attribute__((__always_inline__, __nodebug__, __target__("amx-transpose")))
+
+#define _tile_2rpntlvwz0(tdst, base, stride)                                   \
+  __builtin_ia32_t2rpntlvwz0(tdst, base, stride)
+#define _tile_2rpntlvwz0t1(tdst, base, stride)                                 \
+  __builtin_ia32_t2rpntlvwz0t1(tdst, base, stride)
+#define _tile_2rpntlvwz1(tdst, base, stride)                                   \
+  __builtin_ia32_t2rpntlvwz1(tdst, base, stride)
+#define _tile_2rpntlvwz1t1(tdst, base, stride)                                 \
+  __builtin_ia32_t2rpntlvwz1t1(tdst, base, stride)
+
+/// Transpose 32-bit elements from \a src and write the result to \a dst.
+///
+/// \headerfile <immintrin.h>
+///
+/// \code
+/// void _tile_transposed(__tile dst, __tile src);
+/// \endcode
+///
+/// This intrinsic corresponds to the <c> TTRANSPOSED </c> instruction.
+///
+/// \param dst
+/// 	The destination tile. Max size is 1024 Bytes.
+/// \param src
+/// 	The source tile. Max size is 1024 Bytes.
+///
+/// \code{.operation}
+///
+/// FOR i := 0 TO (dst.rows-1)
+/// 	tmp[511:0] := 0
+/// 	FOR j := 0 TO (dst.colsb/4-1)
+/// 		tmp.dword[j] := src.row[j].dword[i]
+/// 	ENDFOR
+/// 	dst.row[i] := tmp
+/// ENDFOR
+///
+/// zero_upper_rows(dst, dst.rows)
+/// zero_tileconfig_start()
+/// \endcode
+#define _tile_transposed(dst, src) __builtin_ia32_ttransposed(dst, src)
+
+static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz0_internal(
+    unsigned short row, unsigned short col0, unsigned short col1,
+    _tile1024i *dst0, _tile1024i *dst1, const void *base,
+    __SIZE_TYPE__ stride) {
+  // Use __tile1024i_1024a* to escape the alignment check in
+  // clang/test/Headers/x86-intrinsics-headers-clean.cpp
+  __builtin_ia32_t2rpntlvwz0_internal(row, col0, col1, (_tile1024i_1024a *)dst0,
+                                      (_tile1024i_1024a *)dst1, base,
+                                      (__SIZE_TYPE__)(stride));
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz0t1_internal(
+    unsigned short row, unsigned short col0, unsigned short col1,
+    _tile1024i *dst0, _tile1024i *dst1, const void *base,
+    __SIZE_TYPE__ stride) {
+  __builtin_ia32_t2rpntlvwz0t1_internal(
+      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
+      (__SIZE_TYPE__)(stride));
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz1_internal(
+    unsigned short row, unsigned short col0, unsigned short col1,
+    _tile1024i *dst0, _tile1024i *dst1, const void *base,
+    __SIZE_TYPE__ stride) {
+  __builtin_ia32_t2rpntlvwz1_internal(row, col0, col1, (_tile1024i_1024a *)dst0,
+                                      (_tile1024i_1024a *)dst1, base,
+                                      (__SIZE_TYPE__)(stride));
+}
+
+static __inline__ void __DEFAULT_FN_ATTRS_TRANSPOSE _tile_2rpntlvwz1t1_internal(
+    unsigned short row, unsigned short col0, unsigned short col1,
+    _tile1024i *dst0, _tile1024i *dst1, const void *base,
+    __SIZE_TYPE__ stride) {
+  __builtin_ia32_t2rpntlvwz1t1_internal(
+      row, col0, col1, (_tile1024i_1024a *)dst0, (_tile1024i_1024a *)dst1, base,
+      (__SIZE_TYPE__)(stride));
+}
+
+// This is internal intrinsic. C/C++ user should avoid calling it directly.
+static __inline__ _tile1024i __DEFAULT_FN_ATTRS_TRANSPOSE
+_tile_transposed_internal(unsigned short m, unsigned short n, _tile1024i src) {
+  return __builtin_ia32_ttransposed_internal(m, n, src);
+}
+
+/// Converts a pair of tiles from memory into VNNI format, and places the
+/// results in a pair of destinations specified by dst. The pair of tiles
+/// in memory is specified via a tsib; the second tile is after the first
+/// one, separated by the same stride that separates each row.
+/// The tile configuration for the destination tiles indicates the amount
+/// of data to read from memory. The instruction will load a number of rows
+/// that is equal to twice the number of rows in tmm1. The size of each row
+/// is equal to the average width of the destination tiles. If the second
+/// tile is configured with zero rows and columns, only the first tile will
+/// be written.
+/// Provides a hint to the implementation that the data will likely not be
+/// reused in the near future and the data caching can be optimized.
+///
+/// \headerfile <immintrin.h>
+///
+/// This intrinsic corresponds to the <c> T2RPNTLVWZ0 </c> instruction.
+///
+/// \param dst0
+///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param dst1
+///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param base
+///    A pointer to base address.
+/// \param stride
+///    The stride between the rows' data to be loaded in memory.
+__DEFAULT_FN_ATTRS_TRANSPOSE
+static void __tile_2rpntlvwz0(__tile1024i *dst0, __tile1024i *dst1,
+                              const void *base, __SIZE_TYPE__ stride) {
+  _tile_2rpntlvwz0_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
+                            &dst1->tile, base, stride);
+}
+
+/// Converts a pair of tiles from memory into VNNI format, and places the
+/// results in a pair of destinations specified by dst. The pair of tiles
+/// in memory is specified via a tsib; the second tile is after the first
+/// one, separated by the same stride that separates each row.
+/// The tile configuration for the destination tiles indicates the amount
+/// of data to read from memory. The instruction will load a number of rows
+/// that is equal to twice the number of rows in tmm1. The size of each row
+/// is equal to the average width of the destination tiles. If the second
+/// tile is configured with zero rows and columns, only the first tile will
+/// be written.
+///
+/// \headerfile <immintrin.h>
+///
+/// This intrinsic corresponds to the <c> T2RPNTLVWZ0T1 </c> instruction.
+///
+/// \param dst0
+///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param dst1
+///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param base
+///    A pointer to base address.
+/// \param stride
+///    The stride between the rows' data to be loaded in memory.
+__DEFAULT_FN_ATTRS_TRANSPOSE
+static void __tile_2rpntlvwz0t1(__tile1024i *dst0, __tile1024i *dst1,
+                                const void *base, __SIZE_TYPE__ stride) {
+  _tile_2rpntlvwz0t1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
+                              &dst1->tile, base, stride);
+}
+
+/// Converts a pair of tiles from memory into VNNI format, and places the
+/// results in a pair of destinations specified by dst. The pair of tiles
+/// in memory is specified via a tsib; the second tile is after the first
+/// one, separated by the same stride that separates each row.
+/// The tile configuration for the destination tiles indicates the amount
+/// of data to read from memory. The instruction will load a number of rows
+/// that is equal to twice the number of rows in tmm1. The size of each row
+/// is equal to the average width of the destination tiles. If the second
+/// tile is configured with zero rows and columns, only the first tile will
+/// be written. The last row will be not be read from memory but instead
+/// filled with zeros.
+/// Provides a hint to the implementation that the data will likely not be
+/// reused in the near future and the data caching can be optimized.
+///
+/// \headerfile <immintrin.h>
+///
+/// This intrinsic corresponds to the <c> T2RPNTLVWZ1 </c> instruction.
+///
+/// \param dst0
+///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param dst1
+///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param base
+///    A pointer to base address.
+/// \param stride
+///    The stride between the rows' data to be loaded in memory.
+__DEFAULT_FN_ATTRS_TRANSPOSE
+static void __tile_2rpntlvwz1(__tile1024i *dst0, __tile1024i *dst1,
+                              const void *base, __SIZE_TYPE__ stride) {
+  _tile_2rpntlvwz1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
+                            &dst1->tile, base, stride);
+}
+
+/// Converts a pair of tiles from memory into VNNI format, and places the
+/// results in a pair of destinations specified by dst. The pair of tiles
+/// in memory is specified via a tsib; the second tile is after the first
+/// one, separated by the same stride that separates each row.
+/// The tile configuration for the destination tiles indicates the amount
+/// of data to read from memory. The instruction will load a number of rows
+/// that is equal to twice the number of rows in tmm1. The size of each row
+/// is equal to the average width of the destination tiles. If the second
+/// tile is configured with zero rows and columns, only the first tile will
+/// be written. The last row will be not be read from memory but instead
+/// filled with zeros.
+/// Provides a hint to the implementation that the data will likely not be
+/// reused in the near future and the data caching can be optimized.
+///
+/// \headerfile <immintrin.h>
+///
+/// This intrinsic corresponds to the <c> T2RPNTLVWZ1T1 </c> instruction.
+///
+/// \param dst0
+///    First tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param dst1
+///    Second tile of destination tile pair. Max size is 1024i*2 Bytes.
+/// \param base
+///    A pointer to base address.
+/// \param stride
+///    The stride between the rows' data to be loaded in memory.
+__DEFAULT_FN_ATTRS_TRANSPOSE
+static void __tile_2rpntlvwz1t1(__tile1024i *dst0, __tile1024i *dst1,
+                                const void *base, __SIZE_TYPE__ stride) {
+  _tile_2rpntlvwz1t1_internal(dst0->row, dst0->col, dst1->col, &dst0->tile,
+                              &dst1->tile, base, stride);
+}
+
+/// Transpose 32-bit elements from src and write the result to dst.
+///
+/// \headerfile <immintrin.h>
+///
+/// This intrinsic corresponds to the <c> TTRANSPOSED </c> instruction.
+///
+/// \param dst
+///    The destination tile. Max size is 1024 Bytes.
+/// \param src
+///    The source tile. Max size is 1024 Bytes.
+__DEFAULT_FN_ATTRS_TRANSPOSE
+static void __tile_transposed(__tile1024i *dst, __tile1024i src) {
+  dst->tile = _tile_transposed_internal(dst->row, dst->col, src.tile);
+}
+
+#endif /* __x86_64__ */
+#endif /* __AMX_TRANSPOSEINTRIN_H */

package/lib/clang/{18 → 20}/include/arm_acle.h

@@ -75,6 +75,14 @@ static __inline__ void __attribute__((__always_inline__, __nodebug__)) __yield(v
 #define __dbg(t) __builtin_arm_dbg(t)
 #endif
 
+#if defined(__ARM_64BIT_STATE) && __ARM_64BIT_STATE
+#define _CHKFEAT_GCS 1
+static __inline__ uint64_t __attribute__((__always_inline__, __nodebug__))
+__chkfeat(uint64_t __features) {
+  return __builtin_arm_chkfeat(__features) ^ __features;
+}
+#endif
+
 /* 7.5 Swap */
 static __inline__ uint32_t __attribute__((__always_inline__, __nodebug__))
 __swp(uint32_t __x, volatile uint32_t *__p) {
@@ -109,7 +117,7 @@ __swp(uint32_t __x, volatile uint32_t *__p) {
 #endif
 
 /* 7.7 NOP */
-#if !defined(_MSC_VER) || !defined(__aarch64__)
+#if !defined(_MSC_VER) || (!defined(__aarch64__) && !defined(__arm64ec__))
 static __inline__ void __attribute__((__always_inline__, __nodebug__)) __nop(void) {
   __builtin_arm_nop();
 }
@@ -256,28 +264,28 @@ __rbitl(unsigned long __t) {
 }
 
 /* 8.3 16-bit multiplications */
-#if defined(__ARM_FEATURE_DSP) && __ARM_FEATURE_DSP
-static __inline__ int32_t __attribute__((__always_inline__,__nodebug__))
+#if defined(__ARM_32BIT_STATE) && __ARM_32BIT_STATE
+static __inline__ int32_t __attribute__((__always_inline__,__nodebug__, target("dsp")))
 __smulbb(int32_t __a, int32_t __b) {
   return __builtin_arm_smulbb(__a, __b);
 }
-static __inline__ int32_t __attribute__((__always_inline__,__nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__,__nodebug__, target("dsp")))
 __smulbt(int32_t __a, int32_t __b) {
   return __builtin_arm_smulbt(__a, __b);
 }
-static __inline__ int32_t __attribute__((__always_inline__,__nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__,__nodebug__, target("dsp")))
 __smultb(int32_t __a, int32_t __b) {
   return __builtin_arm_smultb(__a, __b);
 }
-static __inline__ int32_t __attribute__((__always_inline__,__nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__,__nodebug__, target("dsp")))
 __smultt(int32_t __a, int32_t __b) {
   return __builtin_arm_smultt(__a, __b);
 }
-static __inline__ int32_t __attribute__((__always_inline__,__nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__,__nodebug__, target("dsp")))
 __smulwb(int32_t __a, int32_t __b) {
   return __builtin_arm_smulwb(__a, __b);
 }
-static __inline__ int32_t __attribute__((__always_inline__,__nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__,__nodebug__, target("dsp")))
 __smulwt(int32_t __a, int32_t __b) {
   return __builtin_arm_smulwt(__a, __b);
 }
@@ -296,46 +304,46 @@ __smulwt(int32_t __a, int32_t __b) {
 #endif
 
 /* 8.4.2 Saturating addition and subtraction intrinsics */
-#if defined(__ARM_FEATURE_DSP) && __ARM_FEATURE_DSP
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+#if defined(__ARM_32BIT_STATE) && __ARM_32BIT_STATE
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __qadd(int32_t __t, int32_t __v) {
   return __builtin_arm_qadd(__t, __v);
 }
 
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __qsub(int32_t __t, int32_t __v) {
   return __builtin_arm_qsub(__t, __v);
 }
 
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __qdbl(int32_t __t) {
   return __builtin_arm_qadd(__t, __t);
 }
 #endif
 
-/* 8.4.3 Accumultating multiplications */
-#if defined(__ARM_FEATURE_DSP) && __ARM_FEATURE_DSP
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+/* 8.4.3 Accumulating multiplications */
+#if defined(__ARM_32BIT_STATE) && __ARM_32BIT_STATE
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __smlabb(int32_t __a, int32_t __b, int32_t __c) {
   return __builtin_arm_smlabb(__a, __b, __c);
 }
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __smlabt(int32_t __a, int32_t __b, int32_t __c) {
   return __builtin_arm_smlabt(__a, __b, __c);
 }
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __smlatb(int32_t __a, int32_t __b, int32_t __c) {
   return __builtin_arm_smlatb(__a, __b, __c);
 }
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __smlatt(int32_t __a, int32_t __b, int32_t __c) {
   return __builtin_arm_smlatt(__a, __b, __c);
 }
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __smlawb(int32_t __a, int32_t __b, int32_t __c) {
   return __builtin_arm_smlawb(__a, __b, __c);
 }
-static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
+static __inline__ int32_t __attribute__((__always_inline__, __nodebug__, target("dsp")))
 __smlawt(int32_t __a, int32_t __b, int32_t __c) {
   return __builtin_arm_smlawt(__a, __b, __c);
 }
@@ -545,7 +553,7 @@ __usub16(uint16x2_t __a, uint16x2_t __b) {
 }
 #endif
 
-/* 8.5.10 Parallel 16-bit multiplications */
+/* 8.5.10 Parallel 16-bit multiplication */
 #if defined(__ARM_FEATURE_SIMD32) && __ARM_FEATURE_SIMD32
 static __inline__ int32_t __attribute__((__always_inline__, __nodebug__))
 __smlad(int16x2_t __a, int16x2_t __b, int32_t __c) {
@@ -613,8 +621,6 @@ __rintnf(float __a) {
 #endif
 
 /* 8.8 CRC32 intrinsics */
-#if (defined(__ARM_FEATURE_CRC32) && __ARM_FEATURE_CRC32) ||                   \
-    (defined(__ARM_64BIT_STATE) && __ARM_64BIT_STATE)
 static __inline__ uint32_t __attribute__((__always_inline__, __nodebug__, target("crc")))
 __crc32b(uint32_t __a, uint8_t __b) {
   return __builtin_arm_crc32b(__a, __b);
@@ -654,7 +660,6 @@ static __inline__ uint32_t __attribute__((__always_inline__, __nodebug__, target
 __crc32cd(uint32_t __a, uint64_t __b) {
   return __builtin_arm_crc32cd(__a, __b);
 }
-#endif
 
 /* 8.6 Floating-point data-processing intrinsics */
 /* Armv8.3-A Javascript conversion intrinsic */
@@ -748,7 +753,7 @@ __arm_st64bv0(void *__addr, data512_t __value) {
 #define __arm_wsrf(sysreg, v) __arm_wsr(sysreg, __builtin_bit_cast(uint32_t, v))
 #define __arm_wsrf64(sysreg, v) __arm_wsr64(sysreg, __builtin_bit_cast(uint64_t, v))
 
-/* 10.3 Memory Tagging Extensions (MTE) Intrinsics */
+/* 10.3 MTE intrinsics */
 #if defined(__ARM_64BIT_STATE) && __ARM_64BIT_STATE
 #define __arm_mte_create_random_tag(__ptr, __mask)  __builtin_arm_irg(__ptr, __mask)
 #define __arm_mte_increment_tag(__ptr, __tag_offset)  __builtin_arm_addg(__ptr, __tag_offset)
@@ -757,7 +762,7 @@ __arm_st64bv0(void *__addr, data512_t __value) {
 #define __arm_mte_set_tag(__ptr) __builtin_arm_stg(__ptr)
 #define __arm_mte_ptrdiff(__ptra, __ptrb) __builtin_arm_subp(__ptra, __ptrb)
 
-/* 18 Memory Operations Intrinsics */
+/* 18 memcpy family of operations intrinsics - MOPS */
 #define __arm_mops_memset_tag(__tagged_address, __value, __size)    \
   __builtin_arm_mops_memset_tag(__tagged_address, __value, __size)
 #endif
@@ -855,6 +860,24 @@ __rndrrs(uint64_t *__p) {
 }
 #endif
 
+/* 11.2 Guarded Control Stack intrinsics */
+#if defined(__ARM_64BIT_STATE) && __ARM_64BIT_STATE
+static __inline__ void * __attribute__((__always_inline__, __nodebug__))
+__gcspr() {
+  return (void *)__builtin_arm_rsr64("gcspr_el0");
+}
+
+static __inline__ uint64_t __attribute__((__always_inline__, __nodebug__, target("gcs")))
+__gcspopm() {
+  return __builtin_arm_gcspopm(0);
+}
+
+static __inline__ const void * __attribute__((__always_inline__, __nodebug__, target("gcs")))
+__gcsss(const void *__stack) {
+  return __builtin_arm_gcsss(__stack);
+}
+#endif
+
 #if defined(__cplusplus)
 }
 #endif

package/lib/clang/{18 → 20}/include/arm_fp16.h

@@ -29,7 +29,7 @@
 typedef __fp16 float16_t;
 #define __ai static __inline__ __attribute__((__always_inline__, __nodebug__))
 
-#if defined(__aarch64__)
+#if defined(__aarch64__) || defined(__arm64ec__)
 #define vabdh_f16(__p0, __p1) __extension__ ({ \
   float16_t __ret; \
   float16_t __s0 = __p0; \