x86-64: Use ZMM16-ZMM31 in AVX512 memmove family functions

Update ifunc-memmove.h to select the function optimized with AVX512
instructions using ZMM16-ZMM31 registers to avoid RTM abort with usable
AVX512VL since VZEROUPPER isn't needed at function exit.

(cherry picked from commit e4fda4631017e49d4ee5a2755db34289b6860fa4)

3 files changed, 35 insertions, 19 deletions

diff --git a/sysdeps/x86_64/multiarch/ifunc-impl-list.c b/sysdeps/x86_64/multiarch/ifunc-impl-list.c
index 80b4db8d05..ed25b4497c 100644
--- a/sysdeps/x86_64/multiarch/ifunc-impl-list.c
+++ b/sysdeps/x86_64/multiarch/ifunc-impl-list.c
@@ -83,10 +83,10 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      HAS_ARCH_FEATURE (AVX512F_Usable),
 			      __memmove_chk_avx512_no_vzeroupper)
 	      IFUNC_IMPL_ADD (array, i, __memmove_chk,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memmove_chk_avx512_unaligned)
 	      IFUNC_IMPL_ADD (array, i, __memmove_chk,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memmove_chk_avx512_unaligned_erms)
 	      IFUNC_IMPL_ADD (array, i, __memmove_chk,
 			      HAS_ARCH_FEATURE (AVX_Usable),
@@ -148,10 +148,10 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      HAS_ARCH_FEATURE (AVX512F_Usable),
 			      __memmove_avx512_no_vzeroupper)
 	      IFUNC_IMPL_ADD (array, i, memmove,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memmove_avx512_unaligned)
 	      IFUNC_IMPL_ADD (array, i, memmove,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memmove_avx512_unaligned_erms)
 	      IFUNC_IMPL_ADD (array, i, memmove, HAS_CPU_FEATURE (SSSE3),
 			      __memmove_ssse3_back)
@@ -697,10 +697,10 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      HAS_ARCH_FEATURE (AVX512F_Usable),
 			      __memcpy_chk_avx512_no_vzeroupper)
 	      IFUNC_IMPL_ADD (array, i, __memcpy_chk,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memcpy_chk_avx512_unaligned)
 	      IFUNC_IMPL_ADD (array, i, __memcpy_chk,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memcpy_chk_avx512_unaligned_erms)
 	      IFUNC_IMPL_ADD (array, i, __memcpy_chk,
 			      HAS_ARCH_FEATURE (AVX_Usable),
@@ -766,10 +766,10 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      HAS_ARCH_FEATURE (AVX512F_Usable),
 			      __memcpy_avx512_no_vzeroupper)
 	      IFUNC_IMPL_ADD (array, i, memcpy,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memcpy_avx512_unaligned)
 	      IFUNC_IMPL_ADD (array, i, memcpy,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __memcpy_avx512_unaligned_erms)
 	      IFUNC_IMPL_ADD (array, i, memcpy, 1, __memcpy_sse2_unaligned)
 	      IFUNC_IMPL_ADD (array, i, memcpy, 1,
@@ -783,10 +783,10 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      HAS_ARCH_FEATURE (AVX512F_Usable),
 			      __mempcpy_chk_avx512_no_vzeroupper)
 	      IFUNC_IMPL_ADD (array, i, __mempcpy_chk,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __mempcpy_chk_avx512_unaligned)
 	      IFUNC_IMPL_ADD (array, i, __mempcpy_chk,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __mempcpy_chk_avx512_unaligned_erms)
 	      IFUNC_IMPL_ADD (array, i, __mempcpy_chk,
 			      HAS_ARCH_FEATURE (AVX_Usable),
@@ -828,10 +828,10 @@ __libc_ifunc_impl_list (const char *name, struct libc_ifunc_impl *array,
 			      HAS_ARCH_FEATURE (AVX512F_Usable),
 			      __mempcpy_avx512_no_vzeroupper)
 	      IFUNC_IMPL_ADD (array, i, mempcpy,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __mempcpy_avx512_unaligned)
 	      IFUNC_IMPL_ADD (array, i, mempcpy,
-			      HAS_ARCH_FEATURE (AVX512F_Usable),
+			      HAS_ARCH_FEATURE (AVX512VL_Usable),
 			      __mempcpy_avx512_unaligned_erms)
 	      IFUNC_IMPL_ADD (array, i, mempcpy,
 			      HAS_ARCH_FEATURE (AVX_Usable),
diff --git a/sysdeps/x86_64/multiarch/ifunc-memmove.h b/sysdeps/x86_64/multiarch/ifunc-memmove.h
index fe003b28e1..554a3f8be8 100644
--- a/sysdeps/x86_64/multiarch/ifunc-memmove.h
+++ b/sysdeps/x86_64/multiarch/ifunc-memmove.h
@@ -56,13 +56,15 @@ IFUNC_SELECTOR (void)
   if (CPU_FEATURES_ARCH_P (cpu_features, AVX512F_Usable)
       && !CPU_FEATURES_ARCH_P (cpu_features, Prefer_No_AVX512))
     {
-      if (CPU_FEATURES_ARCH_P (cpu_features, Prefer_No_VZEROUPPER))
-	return OPTIMIZE (avx512_no_vzeroupper);
+      if (CPU_FEATURES_ARCH_P (cpu_features, AVX512VL_Usable))
+	{
+	if (CPU_FEATURES_CPU_P (cpu_features, ERMS))
+	    return OPTIMIZE (avx512_unaligned_erms);
 
-      if (CPU_FEATURES_CPU_P (cpu_features, ERMS))
-	return OPTIMIZE (avx512_unaligned_erms);
+	  return OPTIMIZE (avx512_unaligned);
+	}
 
-      return OPTIMIZE (avx512_unaligned);
+      return OPTIMIZE (avx512_no_vzeroupper);
     }
 
   if (CPU_FEATURES_ARCH_P (cpu_features, AVX_Fast_Unaligned_Load))
diff --git a/sysdeps/x86_64/multiarch/memmove-avx512-unaligned-erms.S b/sysdeps/x86_64/multiarch/memmove-avx512-unaligned-erms.S
index aac1515cf6..7dad1ad74c 100644
--- a/sysdeps/x86_64/multiarch/memmove-avx512-unaligned-erms.S
+++ b/sysdeps/x86_64/multiarch/memmove-avx512-unaligned-erms.S
@@ -1,11 +1,25 @@
 #if IS_IN (libc)
 # define VEC_SIZE	64
-# define VEC(i)		zmm##i
+# define XMM0		xmm16
+# define XMM1		xmm17
+# define YMM0		ymm16
+# define YMM1		ymm17
+# define VEC0		zmm16
+# define VEC1		zmm17
+# define VEC2		zmm18
+# define VEC3		zmm19
+# define VEC4		zmm20
+# define VEC5		zmm21
+# define VEC6		zmm22
+# define VEC7		zmm23
+# define VEC8		zmm24
+# define VEC(i)		VEC##i
 # define VMOVNT		vmovntdq
 # define VMOVU		vmovdqu64
 # define VMOVA		vmovdqa64
+# define VZEROUPPER
 
-# define SECTION(p)		p##.avx512
+# define SECTION(p)		p##.evex512
 # define MEMMOVE_SYMBOL(p,s)	p##_avx512_##s
 
 # include "memmove-vec-unaligned-erms.S"