Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf fixes from Ingo Molnar:

 - Make /sys/devices/cpu/rdpmc based RDPMC enforcement more
   instantaneous

 - decoder: Update the Intel opcode map

 - Various tooling fixes, including a few late optimizations and
   cleanups.

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (23 commits)
  perf script: Fix invalid LBR/binary mismatch error
  perf script: Fix brstackinsn for AUXTRACE
  perf affinity: Add infrastructure to save/restore affinity
  perf pmu: Use file system cache to optimize sysfs access
  perf regs: Make perf_reg_name() return "unknown" instead of NULL
  perf diff: Use llabs() with 64-bit values
  perf diff: Use llabs() with 64-bit values
  perf/x86: Implement immediate enforcement of /sys/devices/cpu/rdpmc value of 0
  perf tools: Allow to link with libbpf dynamicaly
  perf tests: Rename tests/map_groups.c to tests/maps.c
  perf tests: Rename thread-mg-share to thread-maps-share
  perf maps: Rename map_groups.h to maps.h
  perf maps: Rename 'mg' variables to 'maps'
  perf map_symbol: Rename ms->mg to ms->maps
  perf addr_location: Rename al->mg to al->maps
  perf thread: Rename thread->mg to thread->maps
  perf maps: Merge 'struct maps' with 'struct map_groups'
  x86/insn: perf tools: Add some more instructions to the new instructions test
  x86/insn: Add some more Intel instructions to the opcode map
  perf map: Remove unused functions
  ...
diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
index 6e3f0c1..9a89d98 100644
--- a/arch/x86/events/core.c
+++ b/arch/x86/events/core.c
@@ -49,6 +49,7 @@
 	.enabled = 1,
 };
 
+DEFINE_STATIC_KEY_FALSE(rdpmc_never_available_key);
 DEFINE_STATIC_KEY_FALSE(rdpmc_always_available_key);
 
 u64 __read_mostly hw_cache_event_ids
@@ -2181,20 +2182,25 @@
 	if (x86_pmu.attr_rdpmc_broken)
 		return -ENOTSUPP;
 
-	if ((val == 2) != (x86_pmu.attr_rdpmc == 2)) {
+	if (val != x86_pmu.attr_rdpmc) {
 		/*
-		 * Changing into or out of always available, aka
-		 * perf-event-bypassing mode.  This path is extremely slow,
+		 * Changing into or out of never available or always available,
+		 * aka perf-event-bypassing mode. This path is extremely slow,
 		 * but only root can trigger it, so it's okay.
 		 */
+		if (val == 0)
+			static_branch_inc(&rdpmc_never_available_key);
+		else if (x86_pmu.attr_rdpmc == 0)
+			static_branch_dec(&rdpmc_never_available_key);
+
 		if (val == 2)
 			static_branch_inc(&rdpmc_always_available_key);
-		else
+		else if (x86_pmu.attr_rdpmc == 2)
 			static_branch_dec(&rdpmc_always_available_key);
-		on_each_cpu(refresh_pce, NULL, 1);
-	}
 
-	x86_pmu.attr_rdpmc = val;
+		on_each_cpu(refresh_pce, NULL, 1);
+		x86_pmu.attr_rdpmc = val;
+	}
 
 	return count;
 }
diff --git a/arch/x86/include/asm/mmu_context.h b/arch/x86/include/asm/mmu_context.h
index 16ae821..5f33924 100644
--- a/arch/x86/include/asm/mmu_context.h
+++ b/arch/x86/include/asm/mmu_context.h
@@ -26,12 +26,14 @@
 
 #ifdef CONFIG_PERF_EVENTS
 
+DECLARE_STATIC_KEY_FALSE(rdpmc_never_available_key);
 DECLARE_STATIC_KEY_FALSE(rdpmc_always_available_key);
 
 static inline void load_mm_cr4_irqsoff(struct mm_struct *mm)
 {
 	if (static_branch_unlikely(&rdpmc_always_available_key) ||
-	    atomic_read(&mm->context.perf_rdpmc_allowed))
+	    (!static_branch_unlikely(&rdpmc_never_available_key) &&
+	     atomic_read(&mm->context.perf_rdpmc_allowed)))
 		cr4_set_bits_irqsoff(X86_CR4_PCE);
 	else
 		cr4_clear_bits_irqsoff(X86_CR4_PCE);
diff --git a/arch/x86/lib/x86-opcode-map.txt b/arch/x86/lib/x86-opcode-map.txt
index 0a0e911..8908c58 100644
--- a/arch/x86/lib/x86-opcode-map.txt
+++ b/arch/x86/lib/x86-opcode-map.txt
@@ -695,16 +695,28 @@
 4d: vrcp14ss/d Vsd,Hpd,Wsd (66),(ev)
 4e: vrsqrt14ps/d Vpd,Wpd (66),(ev)
 4f: vrsqrt14ss/d Vsd,Hsd,Wsd (66),(ev)
-# Skip 0x50-0x57
+50: vpdpbusd Vx,Hx,Wx (66),(ev)
+51: vpdpbusds Vx,Hx,Wx (66),(ev)
+52: vdpbf16ps Vx,Hx,Wx (F3),(ev) | vpdpwssd Vx,Hx,Wx (66),(ev) | vp4dpwssd Vdqq,Hdqq,Wdq (F2),(ev)
+53: vpdpwssds Vx,Hx,Wx (66),(ev) | vp4dpwssds Vdqq,Hdqq,Wdq (F2),(ev)
+54: vpopcntb/w Vx,Wx (66),(ev)
+55: vpopcntd/q Vx,Wx (66),(ev)
 58: vpbroadcastd Vx,Wx (66),(v)
 59: vpbroadcastq Vx,Wx (66),(v) | vbroadcasti32x2 Vx,Wx (66),(evo)
 5a: vbroadcasti128 Vqq,Mdq (66),(v) | vbroadcasti32x4/64x2 Vx,Wx (66),(evo)
 5b: vbroadcasti32x8/64x4 Vqq,Mdq (66),(ev)
-# Skip 0x5c-0x63
+# Skip 0x5c-0x61
+62: vpexpandb/w Vx,Wx (66),(ev)
+63: vpcompressb/w Wx,Vx (66),(ev)
 64: vpblendmd/q Vx,Hx,Wx (66),(ev)
 65: vblendmps/d Vx,Hx,Wx (66),(ev)
 66: vpblendmb/w Vx,Hx,Wx (66),(ev)
-# Skip 0x67-0x74
+68: vp2intersectd/q Kx,Hx,Wx (F2),(ev)
+# Skip 0x69-0x6f
+70: vpshldvw Vx,Hx,Wx (66),(ev)
+71: vpshldvd/q Vx,Hx,Wx (66),(ev)
+72: vcvtne2ps2bf16 Vx,Hx,Wx (F2),(ev) | vcvtneps2bf16 Vx,Wx (F3),(ev) | vpshrdvw Vx,Hx,Wx (66),(ev)
+73: vpshrdvd/q Vx,Hx,Wx (66),(ev)
 75: vpermi2b/w Vx,Hx,Wx (66),(ev)
 76: vpermi2d/q Vx,Hx,Wx (66),(ev)
 77: vpermi2ps/d Vx,Hx,Wx (66),(ev)
@@ -727,6 +739,7 @@
 8c: vpmaskmovd/q Vx,Hx,Mx (66),(v)
 8d: vpermb/w Vx,Hx,Wx (66),(ev)
 8e: vpmaskmovd/q Mx,Vx,Hx (66),(v)
+8f: vpshufbitqmb Kx,Hx,Wx (66),(ev)
 # 0x0f 0x38 0x90-0xbf (FMA)
 90: vgatherdd/q Vx,Hx,Wx (66),(v) | vpgatherdd/q Vx,Wx (66),(evo)
 91: vgatherqd/q Vx,Hx,Wx (66),(v) | vpgatherqd/q Vx,Wx (66),(evo)
@@ -738,8 +751,8 @@
 97: vfmsubadd132ps/d Vx,Hx,Wx (66),(v)
 98: vfmadd132ps/d Vx,Hx,Wx (66),(v)
 99: vfmadd132ss/d Vx,Hx,Wx (66),(v),(v1)
-9a: vfmsub132ps/d Vx,Hx,Wx (66),(v)
-9b: vfmsub132ss/d Vx,Hx,Wx (66),(v),(v1)
+9a: vfmsub132ps/d Vx,Hx,Wx (66),(v) | v4fmaddps Vdqq,Hdqq,Wdq (F2),(ev)
+9b: vfmsub132ss/d Vx,Hx,Wx (66),(v),(v1) | v4fmaddss Vdq,Hdq,Wdq (F2),(ev)
 9c: vfnmadd132ps/d Vx,Hx,Wx (66),(v)
 9d: vfnmadd132ss/d Vx,Hx,Wx (66),(v),(v1)
 9e: vfnmsub132ps/d Vx,Hx,Wx (66),(v)
@@ -752,8 +765,8 @@
 a7: vfmsubadd213ps/d Vx,Hx,Wx (66),(v)
 a8: vfmadd213ps/d Vx,Hx,Wx (66),(v)
 a9: vfmadd213ss/d Vx,Hx,Wx (66),(v),(v1)
-aa: vfmsub213ps/d Vx,Hx,Wx (66),(v)
-ab: vfmsub213ss/d Vx,Hx,Wx (66),(v),(v1)
+aa: vfmsub213ps/d Vx,Hx,Wx (66),(v) | v4fnmaddps Vdqq,Hdqq,Wdq (F2),(ev)
+ab: vfmsub213ss/d Vx,Hx,Wx (66),(v),(v1) | v4fnmaddss Vdq,Hdq,Wdq (F2),(ev)
 ac: vfnmadd213ps/d Vx,Hx,Wx (66),(v)
 ad: vfnmadd213ss/d Vx,Hx,Wx (66),(v),(v1)
 ae: vfnmsub213ps/d Vx,Hx,Wx (66),(v)
@@ -780,11 +793,12 @@
 cb: sha256rnds2 Vdq,Wdq | vrcp28ss/d Vx,Hx,Wx (66),(ev)
 cc: sha256msg1 Vdq,Wdq | vrsqrt28ps/d Vx,Wx (66),(ev)
 cd: sha256msg2 Vdq,Wdq | vrsqrt28ss/d Vx,Hx,Wx (66),(ev)
+cf: vgf2p8mulb Vx,Wx (66)
 db: VAESIMC Vdq,Wdq (66),(v1)
-dc: VAESENC Vdq,Hdq,Wdq (66),(v1)
-dd: VAESENCLAST Vdq,Hdq,Wdq (66),(v1)
-de: VAESDEC Vdq,Hdq,Wdq (66),(v1)
-df: VAESDECLAST Vdq,Hdq,Wdq (66),(v1)
+dc: vaesenc Vx,Hx,Wx (66)
+dd: vaesenclast Vx,Hx,Wx (66)
+de: vaesdec Vx,Hx,Wx (66)
+df: vaesdeclast Vx,Hx,Wx (66)
 f0: MOVBE Gy,My | MOVBE Gw,Mw (66) | CRC32 Gd,Eb (F2) | CRC32 Gd,Eb (66&F2)
 f1: MOVBE My,Gy | MOVBE Mw,Gw (66) | CRC32 Gd,Ey (F2) | CRC32 Gd,Ew (66&F2)
 f2: ANDN Gy,By,Ey (v)
@@ -848,7 +862,7 @@
 41: vdppd Vdq,Hdq,Wdq,Ib (66),(v1)
 42: vmpsadbw Vx,Hx,Wx,Ib (66),(v1) | vdbpsadbw Vx,Hx,Wx,Ib (66),(evo)
 43: vshufi32x4/64x2 Vx,Hx,Wx,Ib (66),(ev)
-44: vpclmulqdq Vdq,Hdq,Wdq,Ib (66),(v1)
+44: vpclmulqdq Vx,Hx,Wx,Ib (66)
 46: vperm2i128 Vqq,Hqq,Wqq,Ib (66),(v)
 4a: vblendvps Vx,Hx,Wx,Lx (66),(v)
 4b: vblendvpd Vx,Hx,Wx,Lx (66),(v)
@@ -865,7 +879,13 @@
 63: vpcmpistri Vdq,Wdq,Ib (66),(v1)
 66: vfpclassps/d Vk,Wx,Ib (66),(ev)
 67: vfpclassss/d Vk,Wx,Ib (66),(ev)
+70: vpshldw Vx,Hx,Wx,Ib (66),(ev)
+71: vpshldd/q Vx,Hx,Wx,Ib (66),(ev)
+72: vpshrdw Vx,Hx,Wx,Ib (66),(ev)
+73: vpshrdd/q Vx,Hx,Wx,Ib (66),(ev)
 cc: sha1rnds4 Vdq,Wdq,Ib
+ce: vgf2p8affineqb Vx,Wx,Ib (66)
+cf: vgf2p8affineinvqb Vx,Wx,Ib (66)
 df: VAESKEYGEN Vdq,Wdq,Ib (66),(v1)
 f0: RORX Gy,Ey,Ib (F2),(v)
 EndTable
diff --git a/tools/arch/x86/lib/x86-opcode-map.txt b/tools/arch/x86/lib/x86-opcode-map.txt
index 0a0e911..8908c58 100644
--- a/tools/arch/x86/lib/x86-opcode-map.txt
+++ b/tools/arch/x86/lib/x86-opcode-map.txt
@@ -695,16 +695,28 @@
 4d: vrcp14ss/d Vsd,Hpd,Wsd (66),(ev)
 4e: vrsqrt14ps/d Vpd,Wpd (66),(ev)
 4f: vrsqrt14ss/d Vsd,Hsd,Wsd (66),(ev)
-# Skip 0x50-0x57
+50: vpdpbusd Vx,Hx,Wx (66),(ev)
+51: vpdpbusds Vx,Hx,Wx (66),(ev)
+52: vdpbf16ps Vx,Hx,Wx (F3),(ev) | vpdpwssd Vx,Hx,Wx (66),(ev) | vp4dpwssd Vdqq,Hdqq,Wdq (F2),(ev)
+53: vpdpwssds Vx,Hx,Wx (66),(ev) | vp4dpwssds Vdqq,Hdqq,Wdq (F2),(ev)
+54: vpopcntb/w Vx,Wx (66),(ev)
+55: vpopcntd/q Vx,Wx (66),(ev)
 58: vpbroadcastd Vx,Wx (66),(v)
 59: vpbroadcastq Vx,Wx (66),(v) | vbroadcasti32x2 Vx,Wx (66),(evo)
 5a: vbroadcasti128 Vqq,Mdq (66),(v) | vbroadcasti32x4/64x2 Vx,Wx (66),(evo)
 5b: vbroadcasti32x8/64x4 Vqq,Mdq (66),(ev)
-# Skip 0x5c-0x63
+# Skip 0x5c-0x61
+62: vpexpandb/w Vx,Wx (66),(ev)
+63: vpcompressb/w Wx,Vx (66),(ev)
 64: vpblendmd/q Vx,Hx,Wx (66),(ev)
 65: vblendmps/d Vx,Hx,Wx (66),(ev)
 66: vpblendmb/w Vx,Hx,Wx (66),(ev)
-# Skip 0x67-0x74
+68: vp2intersectd/q Kx,Hx,Wx (F2),(ev)
+# Skip 0x69-0x6f
+70: vpshldvw Vx,Hx,Wx (66),(ev)
+71: vpshldvd/q Vx,Hx,Wx (66),(ev)
+72: vcvtne2ps2bf16 Vx,Hx,Wx (F2),(ev) | vcvtneps2bf16 Vx,Wx (F3),(ev) | vpshrdvw Vx,Hx,Wx (66),(ev)
+73: vpshrdvd/q Vx,Hx,Wx (66),(ev)
 75: vpermi2b/w Vx,Hx,Wx (66),(ev)
 76: vpermi2d/q Vx,Hx,Wx (66),(ev)
 77: vpermi2ps/d Vx,Hx,Wx (66),(ev)
@@ -727,6 +739,7 @@
 8c: vpmaskmovd/q Vx,Hx,Mx (66),(v)
 8d: vpermb/w Vx,Hx,Wx (66),(ev)
 8e: vpmaskmovd/q Mx,Vx,Hx (66),(v)
+8f: vpshufbitqmb Kx,Hx,Wx (66),(ev)
 # 0x0f 0x38 0x90-0xbf (FMA)
 90: vgatherdd/q Vx,Hx,Wx (66),(v) | vpgatherdd/q Vx,Wx (66),(evo)
 91: vgatherqd/q Vx,Hx,Wx (66),(v) | vpgatherqd/q Vx,Wx (66),(evo)
@@ -738,8 +751,8 @@
 97: vfmsubadd132ps/d Vx,Hx,Wx (66),(v)
 98: vfmadd132ps/d Vx,Hx,Wx (66),(v)
 99: vfmadd132ss/d Vx,Hx,Wx (66),(v),(v1)
-9a: vfmsub132ps/d Vx,Hx,Wx (66),(v)
-9b: vfmsub132ss/d Vx,Hx,Wx (66),(v),(v1)
+9a: vfmsub132ps/d Vx,Hx,Wx (66),(v) | v4fmaddps Vdqq,Hdqq,Wdq (F2),(ev)
+9b: vfmsub132ss/d Vx,Hx,Wx (66),(v),(v1) | v4fmaddss Vdq,Hdq,Wdq (F2),(ev)
 9c: vfnmadd132ps/d Vx,Hx,Wx (66),(v)
 9d: vfnmadd132ss/d Vx,Hx,Wx (66),(v),(v1)
 9e: vfnmsub132ps/d Vx,Hx,Wx (66),(v)
@@ -752,8 +765,8 @@
 a7: vfmsubadd213ps/d Vx,Hx,Wx (66),(v)
 a8: vfmadd213ps/d Vx,Hx,Wx (66),(v)
 a9: vfmadd213ss/d Vx,Hx,Wx (66),(v),(v1)
-aa: vfmsub213ps/d Vx,Hx,Wx (66),(v)
-ab: vfmsub213ss/d Vx,Hx,Wx (66),(v),(v1)
+aa: vfmsub213ps/d Vx,Hx,Wx (66),(v) | v4fnmaddps Vdqq,Hdqq,Wdq (F2),(ev)
+ab: vfmsub213ss/d Vx,Hx,Wx (66),(v),(v1) | v4fnmaddss Vdq,Hdq,Wdq (F2),(ev)
 ac: vfnmadd213ps/d Vx,Hx,Wx (66),(v)
 ad: vfnmadd213ss/d Vx,Hx,Wx (66),(v),(v1)
 ae: vfnmsub213ps/d Vx,Hx,Wx (66),(v)
@@ -780,11 +793,12 @@
 cb: sha256rnds2 Vdq,Wdq | vrcp28ss/d Vx,Hx,Wx (66),(ev)
 cc: sha256msg1 Vdq,Wdq | vrsqrt28ps/d Vx,Wx (66),(ev)
 cd: sha256msg2 Vdq,Wdq | vrsqrt28ss/d Vx,Hx,Wx (66),(ev)
+cf: vgf2p8mulb Vx,Wx (66)
 db: VAESIMC Vdq,Wdq (66),(v1)
-dc: VAESENC Vdq,Hdq,Wdq (66),(v1)
-dd: VAESENCLAST Vdq,Hdq,Wdq (66),(v1)
-de: VAESDEC Vdq,Hdq,Wdq (66),(v1)
-df: VAESDECLAST Vdq,Hdq,Wdq (66),(v1)
+dc: vaesenc Vx,Hx,Wx (66)
+dd: vaesenclast Vx,Hx,Wx (66)
+de: vaesdec Vx,Hx,Wx (66)
+df: vaesdeclast Vx,Hx,Wx (66)
 f0: MOVBE Gy,My | MOVBE Gw,Mw (66) | CRC32 Gd,Eb (F2) | CRC32 Gd,Eb (66&F2)
 f1: MOVBE My,Gy | MOVBE Mw,Gw (66) | CRC32 Gd,Ey (F2) | CRC32 Gd,Ew (66&F2)
 f2: ANDN Gy,By,Ey (v)
@@ -848,7 +862,7 @@
 41: vdppd Vdq,Hdq,Wdq,Ib (66),(v1)
 42: vmpsadbw Vx,Hx,Wx,Ib (66),(v1) | vdbpsadbw Vx,Hx,Wx,Ib (66),(evo)
 43: vshufi32x4/64x2 Vx,Hx,Wx,Ib (66),(ev)
-44: vpclmulqdq Vdq,Hdq,Wdq,Ib (66),(v1)
+44: vpclmulqdq Vx,Hx,Wx,Ib (66)
 46: vperm2i128 Vqq,Hqq,Wqq,Ib (66),(v)
 4a: vblendvps Vx,Hx,Wx,Lx (66),(v)
 4b: vblendvpd Vx,Hx,Wx,Lx (66),(v)
@@ -865,7 +879,13 @@
 63: vpcmpistri Vdq,Wdq,Ib (66),(v1)
 66: vfpclassps/d Vk,Wx,Ib (66),(ev)
 67: vfpclassss/d Vk,Wx,Ib (66),(ev)
+70: vpshldw Vx,Hx,Wx,Ib (66),(ev)
+71: vpshldd/q Vx,Hx,Wx,Ib (66),(ev)
+72: vpshrdw Vx,Hx,Wx,Ib (66),(ev)
+73: vpshrdd/q Vx,Hx,Wx,Ib (66),(ev)
 cc: sha1rnds4 Vdq,Wdq,Ib
+ce: vgf2p8affineqb Vx,Wx,Ib (66)
+cf: vgf2p8affineinvqb Vx,Wx,Ib (66)
 df: VAESKEYGEN Vdq,Wdq,Ib (66),(v1)
 f0: RORX Gy,Ey,Ib (F2),(v)
 EndTable
diff --git a/tools/build/Makefile.feature b/tools/build/Makefile.feature
index 8a19753..574c2e0 100644
--- a/tools/build/Makefile.feature
+++ b/tools/build/Makefile.feature
@@ -96,7 +96,8 @@
          cxx                            \
          llvm                           \
          llvm-version                   \
-         clang
+         clang                          \
+         libbpf
 
 FEATURE_TESTS ?= $(FEATURE_TESTS_BASIC)
 
diff --git a/tools/build/feature/Makefile b/tools/build/feature/Makefile
index 8499385..f30a890 100644
--- a/tools/build/feature/Makefile
+++ b/tools/build/feature/Makefile
@@ -53,6 +53,7 @@
          test-zlib.bin                          \
          test-lzma.bin                          \
          test-bpf.bin                           \
+         test-libbpf.bin                        \
          test-get_cpuid.bin                     \
          test-sdt.bin                           \
          test-cxx.bin                           \
@@ -270,6 +271,9 @@
 $(OUTPUT)test-bpf.bin:
 	$(BUILD)
 
+$(OUTPUT)test-libbpf.bin:
+	$(BUILD) -lbpf
+
 $(OUTPUT)test-sdt.bin:
 	$(BUILD)
 
diff --git a/tools/build/feature/test-libbpf.c b/tools/build/feature/test-libbpf.c
new file mode 100644
index 0000000..a508756
--- /dev/null
+++ b/tools/build/feature/test-libbpf.c
@@ -0,0 +1,7 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <bpf/libbpf.h>
+
+int main(void)
+{
+	return bpf_object__open("test") ? 0 : -1;
+}
diff --git a/tools/perf/Makefile.config b/tools/perf/Makefile.config
index 1783427..c90f414 100644
--- a/tools/perf/Makefile.config
+++ b/tools/perf/Makefile.config
@@ -483,6 +483,16 @@
     ifeq ($(feature-bpf), 1)
       CFLAGS += -DHAVE_LIBBPF_SUPPORT
       $(call detected,CONFIG_LIBBPF)
+
+      # detecting libbpf without LIBBPF_DYNAMIC, so make VF=1 shows libbpf detection status
+      $(call feature_check,libbpf)
+      ifdef LIBBPF_DYNAMIC
+        ifeq ($(feature-libbpf), 1)
+          EXTLIBS += -lbpf
+        else
+          dummy := $(error Error: No libbpf devel library found, please install libbpf-devel);
+        endif
+      endif
     endif
 
     ifndef NO_DWARF
diff --git a/tools/perf/Makefile.perf b/tools/perf/Makefile.perf
index 1cd2944..eae5d5e 100644
--- a/tools/perf/Makefile.perf
+++ b/tools/perf/Makefile.perf
@@ -116,6 +116,8 @@
 #
 # Define TCMALLOC to enable tcmalloc heap profiling.
 #
+# Define LIBBPF_DYNAMIC to enable libbpf dynamic linking.
+#
 
 # As per kernel Makefile, avoid funny character set dependencies
 unexport LC_ALL
@@ -360,7 +362,9 @@
 
 PERFLIBS = $(LIBAPI) $(LIBTRACEEVENT) $(LIBSUBCMD) $(LIBPERF)
 ifndef NO_LIBBPF
-  PERFLIBS += $(LIBBPF)
+  ifndef LIBBPF_DYNAMIC
+    PERFLIBS += $(LIBBPF)
+  endif
 endif
 
 # We choose to avoid "if .. else if .. else .. endif endif"
diff --git a/tools/perf/arch/arm/tests/dwarf-unwind.c b/tools/perf/arch/arm/tests/dwarf-unwind.c
index 2c35e53..ccfa870 100644
--- a/tools/perf/arch/arm/tests/dwarf-unwind.c
+++ b/tools/perf/arch/arm/tests/dwarf-unwind.c
@@ -3,7 +3,7 @@
 #include "perf_regs.h"
 #include "thread.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "event.h"
 #include "debug.h"
 #include "tests/tests.h"
@@ -26,7 +26,7 @@
 
 	sp = (unsigned long) regs[PERF_REG_ARM_SP];
 
-	map = map_groups__find(thread->mg, (u64)sp);
+	map = maps__find(thread->maps, (u64)sp);
 	if (!map) {
 		pr_debug("failed to get stack map\n");
 		free(buf);
diff --git a/tools/perf/arch/arm64/tests/dwarf-unwind.c b/tools/perf/arch/arm64/tests/dwarf-unwind.c
index a6a407f..46147a4 100644
--- a/tools/perf/arch/arm64/tests/dwarf-unwind.c
+++ b/tools/perf/arch/arm64/tests/dwarf-unwind.c
@@ -3,7 +3,7 @@
 #include "perf_regs.h"
 #include "thread.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "event.h"
 #include "debug.h"
 #include "tests/tests.h"
@@ -26,7 +26,7 @@
 
 	sp = (unsigned long) regs[PERF_REG_ARM64_SP];
 
-	map = map_groups__find(thread->mg, (u64)sp);
+	map = maps__find(thread->maps, (u64)sp);
 	if (!map) {
 		pr_debug("failed to get stack map\n");
 		free(buf);
diff --git a/tools/perf/arch/powerpc/tests/dwarf-unwind.c b/tools/perf/arch/powerpc/tests/dwarf-unwind.c
index 5c178e4..8efd9ed 100644
--- a/tools/perf/arch/powerpc/tests/dwarf-unwind.c
+++ b/tools/perf/arch/powerpc/tests/dwarf-unwind.c
@@ -3,7 +3,7 @@
 #include "perf_regs.h"
 #include "thread.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "event.h"
 #include "debug.h"
 #include "tests/tests.h"
@@ -27,7 +27,7 @@
 
 	sp = (unsigned long) regs[PERF_REG_POWERPC_R1];
 
-	map = map_groups__find(thread->mg, (u64)sp);
+	map = maps__find(thread->maps, (u64)sp);
 	if (!map) {
 		pr_debug("failed to get stack map\n");
 		free(buf);
diff --git a/tools/perf/arch/s390/annotate/instructions.c b/tools/perf/arch/s390/annotate/instructions.c
index 2a6662e4..0e13663 100644
--- a/tools/perf/arch/s390/annotate/instructions.c
+++ b/tools/perf/arch/s390/annotate/instructions.c
@@ -38,7 +38,7 @@
 		return -1;
 	target.addr = map__objdump_2mem(map, ops->target.addr);
 
-	if (map_groups__find_ams(ms->mg, &target) == 0 &&
+	if (maps__find_ams(ms->maps, &target) == 0 &&
 	    map__rip_2objdump(target.ms.map, map->map_ip(target.ms.map, target.addr)) == ops->target.addr)
 		ops->target.sym = target.ms.sym;
 
diff --git a/tools/perf/arch/x86/tests/dwarf-unwind.c b/tools/perf/arch/x86/tests/dwarf-unwind.c
index 6ad0a1c..ef43be9 100644
--- a/tools/perf/arch/x86/tests/dwarf-unwind.c
+++ b/tools/perf/arch/x86/tests/dwarf-unwind.c
@@ -3,7 +3,7 @@
 #include "perf_regs.h"
 #include "thread.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "event.h"
 #include "debug.h"
 #include "tests/tests.h"
@@ -27,7 +27,7 @@
 
 	sp = (unsigned long) regs[PERF_REG_X86_SP];
 
-	map = map_groups__find(thread->mg, (u64)sp);
+	map = maps__find(thread->maps, (u64)sp);
 	if (!map) {
 		pr_debug("failed to get stack map\n");
 		free(buf);
diff --git a/tools/perf/arch/x86/tests/insn-x86-dat-32.c b/tools/perf/arch/x86/tests/insn-x86-dat-32.c
index 58f8f2a0..e6461ab 100644
--- a/tools/perf/arch/x86/tests/insn-x86-dat-32.c
+++ b/tools/perf/arch/x86/tests/insn-x86-dat-32.c
@@ -667,6 +667,86 @@
 "62 f2 55 0f 4f f4    \tvrsqrt14ss %xmm4,%xmm5,%xmm6{%k7}",},
 {{0x62, 0xf2, 0xd5, 0x0f, 0x4f, 0xf4, }, 6, 0, "", "",
 "62 f2 d5 0f 4f f4    \tvrsqrt14sd %xmm4,%xmm5,%xmm6{%k7}",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x50, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 50 d9    \tvpdpbusd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x50, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 50 d9    \tvpdpbusd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x50, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 50 d9    \tvpdpbusd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x50, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 50 9c c8 78 56 34 12 \tvpdpbusd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x51, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 51 d9    \tvpdpbusds %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x51, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 51 d9    \tvpdpbusds %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x51, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 51 d9    \tvpdpbusds %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x51, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 51 9c c8 78 56 34 12 \tvpdpbusds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6e, 0x08, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6e 08 52 d9    \tvdpbf16ps %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6e, 0x28, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6e 28 52 d9    \tvdpbf16ps %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6e, 0x48, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6e 48 52 d9    \tvdpbf16ps %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6e, 0x48, 0x52, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6e 48 52 9c c8 78 56 34 12 \tvdpbf16ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 52 d9    \tvpdpwssd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 52 d9    \tvpdpwssd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 52 d9    \tvpdpwssd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x52, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 52 9c c8 78 56 34 12 \tvpdpwssd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x52, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 52 20    \tvp4dpwssd (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x52, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 52 a4 c8 78 56 34 12 \tvp4dpwssd 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x53, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 53 d9    \tvpdpwssds %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x53, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 53 d9    \tvpdpwssds %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x53, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 53 d9    \tvpdpwssds %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x53, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 53 9c c8 78 56 34 12 \tvpdpwssds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x53, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 53 20    \tvp4dpwssds (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x53, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 53 a4 c8 78 56 34 12 \tvp4dpwssds 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 08 54 d1    \tvpopcntb %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 28 54 d1    \tvpopcntb %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 48 54 d1    \tvpopcntb %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x54, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 54 94 c8 78 56 34 12 \tvpopcntb 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 08 54 d1    \tvpopcntw %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 28 54 d1    \tvpopcntw %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 48 54 d1    \tvpopcntw %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x54, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 54 94 c8 78 56 34 12 \tvpopcntw 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 08 55 d1    \tvpopcntd %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 28 55 d1    \tvpopcntd %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 48 55 d1    \tvpopcntd %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x55, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 55 94 c8 78 56 34 12 \tvpopcntd 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 08 55 d1    \tvpopcntq %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 28 55 d1    \tvpopcntq %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 48 55 d1    \tvpopcntq %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x55, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 55 94 c8 78 56 34 12 \tvpopcntq 0x12345678(%eax,%ecx,8),%zmm2",},
 {{0xc4, 0xe2, 0x79, 0x59, 0xf4, }, 5, 0, "", "",
 "c4 e2 79 59 f4       \tvpbroadcastq %xmm4,%xmm6",},
 {{0x62, 0xf2, 0x7d, 0x48, 0x59, 0xf7, }, 6, 0, "", "",
@@ -681,6 +761,38 @@
 "62 f2 7d 48 5b 31    \tvbroadcasti32x8 (%ecx),%zmm6",},
 {{0x62, 0xf2, 0xfd, 0x48, 0x5b, 0x31, }, 6, 0, "", "",
 "62 f2 fd 48 5b 31    \tvbroadcasti64x4 (%ecx),%zmm6",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 08 62 d1    \tvpexpandb %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 28 62 d1    \tvpexpandb %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 48 62 d1    \tvpexpandb %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x62, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 62 94 c8 78 56 34 12 \tvpexpandb 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 08 62 d1    \tvpexpandw %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 28 62 d1    \tvpexpandw %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 48 62 d1    \tvpexpandw %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x62, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 62 94 c8 78 56 34 12 \tvpexpandw 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 7d 08 63 ca    \tvpcompressb %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 7d 28 63 ca    \tvpcompressb %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 7d 48 63 ca    \tvpcompressb %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x63, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 63 94 c8 78 56 34 12 \tvpcompressb %zmm2,0x12345678(%eax,%ecx,8)",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 fd 08 63 ca    \tvpcompressw %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 fd 28 63 ca    \tvpcompressw %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 fd 48 63 ca    \tvpcompressw %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x63, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 63 94 c8 78 56 34 12 \tvpcompressw %zmm2,0x12345678(%eax,%ecx,8)",},
 {{0x62, 0xf2, 0x55, 0x48, 0x64, 0xf4, }, 6, 0, "", "",
 "62 f2 55 48 64 f4    \tvpblendmd %zmm4,%zmm5,%zmm6",},
 {{0x62, 0xf2, 0xd5, 0x48, 0x64, 0xf4, }, 6, 0, "", "",
@@ -693,6 +805,86 @@
 "62 f2 55 48 66 f4    \tvpblendmb %zmm4,%zmm5,%zmm6",},
 {{0x62, 0xf2, 0xd5, 0x48, 0x66, 0xf4, }, 6, 0, "", "",
 "62 f2 d5 48 66 f4    \tvpblendmw %zmm4,%zmm5,%zmm6",},
+{{0x62, 0xf2, 0x6f, 0x08, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 08 68 d9    \tvp2intersectd %xmm1,%xmm2,%k3",},
+{{0x62, 0xf2, 0x6f, 0x28, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 28 68 d9    \tvp2intersectd %ymm1,%ymm2,%k3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 48 68 d9    \tvp2intersectd %zmm1,%zmm2,%k3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x68, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6f 48 68 9c c8 78 56 34 12 \tvp2intersectd 0x12345678(%eax,%ecx,8),%zmm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x08, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 ef 08 68 d9    \tvp2intersectq %xmm1,%xmm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x28, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 ef 28 68 d9    \tvp2intersectq %ymm1,%ymm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x48, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 ef 48 68 d9    \tvp2intersectq %zmm1,%zmm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x48, 0x68, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ef 48 68 9c c8 78 56 34 12 \tvp2intersectq 0x12345678(%eax,%ecx,8),%zmm2,%k3",},
+{{0x62, 0xf2, 0xed, 0x08, 0x70, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 70 d9    \tvpshldvw %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x70, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 70 d9    \tvpshldvw %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x70, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 70 d9    \tvpshldvw %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x70, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 70 9c c8 78 56 34 12 \tvpshldvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 71 d9    \tvpshldvd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 71 d9    \tvpshldvd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 71 d9    \tvpshldvd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x71, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 71 9c c8 78 56 34 12 \tvpshldvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x08, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 71 d9    \tvpshldvq %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 71 d9    \tvpshldvq %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 71 d9    \tvpshldvq %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x71, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 71 9c c8 78 56 34 12 \tvpshldvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6f, 0x08, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 08 72 d9    \tvcvtne2ps2bf16 %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6f, 0x28, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 28 72 d9    \tvcvtne2ps2bf16 %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 48 72 d9    \tvcvtne2ps2bf16 %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x72, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6f 48 72 9c c8 78 56 34 12 \tvcvtne2ps2bf16 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7e, 0x08, 0x72, 0xd1, }, 6, 0, "", "",
+"62 f2 7e 08 72 d1    \tvcvtneps2bf16 %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7e, 0x28, 0x72, 0xd1, }, 6, 0, "", "",
+"62 f2 7e 28 72 d1    \tvcvtneps2bf16 %ymm1,%xmm2",},
+{{0x62, 0xf2, 0x7e, 0x48, 0x72, 0xd1, }, 6, 0, "", "",
+"62 f2 7e 48 72 d1    \tvcvtneps2bf16 %zmm1,%ymm2",},
+{{0x62, 0xf2, 0x7e, 0x48, 0x72, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7e 48 72 94 c8 78 56 34 12 \tvcvtneps2bf16 0x12345678(%eax,%ecx,8),%ymm2",},
+{{0x62, 0xf2, 0xed, 0x08, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 72 d9    \tvpshrdvw %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 72 d9    \tvpshrdvw %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 72 d9    \tvpshrdvw %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x72, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 72 9c c8 78 56 34 12 \tvpshrdvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 73 d9    \tvpshrdvd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 73 d9    \tvpshrdvd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 73 d9    \tvpshrdvd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x73, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 73 9c c8 78 56 34 12 \tvpshrdvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x08, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 73 d9    \tvpshrdvq %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 73 d9    \tvpshrdvq %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 73 d9    \tvpshrdvq %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x73, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 73 9c c8 78 56 34 12 \tvpshrdvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
 {{0x62, 0xf2, 0x55, 0x48, 0x75, 0xf4, }, 6, 0, "", "",
 "62 f2 55 48 75 f4    \tvpermi2b %zmm4,%zmm5,%zmm6",},
 {{0x62, 0xf2, 0xd5, 0x48, 0x75, 0xf4, }, 6, 0, "", "",
@@ -745,6 +937,14 @@
 "62 f2 55 48 8d f4    \tvpermb %zmm4,%zmm5,%zmm6",},
 {{0x62, 0xf2, 0xd5, 0x48, 0x8d, 0xf4, }, 6, 0, "", "",
 "62 f2 d5 48 8d f4    \tvpermw %zmm4,%zmm5,%zmm6",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x8f, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 8f d9    \tvpshufbitqmb %xmm1,%xmm2,%k3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x8f, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 8f d9    \tvpshufbitqmb %ymm1,%ymm2,%k3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x8f, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 8f d9    \tvpshufbitqmb %zmm1,%zmm2,%k3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x8f, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 8f 9c c8 78 56 34 12 \tvpshufbitqmb 0x12345678(%eax,%ecx,8),%zmm2,%k3",},
 {{0xc4, 0xe2, 0x69, 0x90, 0x4c, 0x7d, 0x02, }, 7, 0, "", "",
 "c4 e2 69 90 4c 7d 02 \tvpgatherdd %xmm2,0x2(%ebp,%xmm7,2),%xmm1",},
 {{0xc4, 0xe2, 0xe9, 0x90, 0x4c, 0x7d, 0x04, }, 7, 0, "", "",
@@ -761,6 +961,38 @@
 "62 f2 7d 49 91 b4 fd 7b 00 00 00 \tvpgatherqd 0x7b(%ebp,%zmm7,8),%ymm6{%k1}",},
 {{0x62, 0xf2, 0xfd, 0x49, 0x91, 0xb4, 0xfd, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
 "62 f2 fd 49 91 b4 fd 7b 00 00 00 \tvpgatherqq 0x7b(%ebp,%zmm7,8),%zmm6{%k1}",},
+{{0xc4, 0xe2, 0x69, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 9a d9       \tvfmsub132ps %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d 9a d9       \tvfmsub132ps %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x9a, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 9a d9    \tvfmsub132ps %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x9a, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 9a 9c c8 78 56 34 12 \tvfmsub132ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0xe9, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 9a d9       \tvfmsub132pd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xed, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 ed 9a d9       \tvfmsub132pd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x9a, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 9a d9    \tvfmsub132pd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x9a, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 9a 9c c8 78 56 34 12 \tvfmsub132pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x9a, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 9a 20    \tv4fmaddps (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x9a, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 9a a4 c8 78 56 34 12 \tv4fmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0xc4, 0xe2, 0x69, 0x9b, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 9b d9       \tvfmsub132ss %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x69, 0x9b, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 69 9b 9c c8 78 56 34 12 \tvfmsub132ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0x9b, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 9b d9       \tvfmsub132sd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0x9b, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 e9 9b 9c c8 78 56 34 12 \tvfmsub132sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x7f, 0x08, 0x9b, 0x20, }, 6, 0, "", "",
+"62 f2 7f 08 9b 20    \tv4fmaddss (%eax),%xmm0,%xmm4",},
+{{0x62, 0xf2, 0x7f, 0x08, 0x9b, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 08 9b a4 c8 78 56 34 12 \tv4fmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4",},
 {{0x62, 0xf2, 0x7d, 0x49, 0xa0, 0xb4, 0xfd, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
 "62 f2 7d 49 a0 b4 fd 7b 00 00 00 \tvpscatterdd %zmm6,0x7b(%ebp,%zmm7,8){%k1}",},
 {{0x62, 0xf2, 0xfd, 0x49, 0xa0, 0xb4, 0xfd, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
@@ -777,6 +1009,38 @@
 "62 f2 7d 49 a3 b4 fd 7b 00 00 00 \tvscatterqps %ymm6,0x7b(%ebp,%zmm7,8){%k1}",},
 {{0x62, 0xf2, 0xfd, 0x49, 0xa3, 0xb4, 0xfd, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
 "62 f2 fd 49 a3 b4 fd 7b 00 00 00 \tvscatterqpd %zmm6,0x7b(%ebp,%zmm7,8){%k1}",},
+{{0xc4, 0xe2, 0x69, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 aa d9       \tvfmsub213ps %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d aa d9       \tvfmsub213ps %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xaa, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 aa d9    \tvfmsub213ps %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xaa, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 aa 9c c8 78 56 34 12 \tvfmsub213ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0xe9, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 aa d9       \tvfmsub213pd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xed, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 ed aa d9       \tvfmsub213pd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0xaa, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 aa d9    \tvfmsub213pd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0xaa, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 aa 9c c8 78 56 34 12 \tvfmsub213pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0xaa, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 aa 20    \tv4fnmaddps (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0xaa, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 aa a4 c8 78 56 34 12 \tv4fnmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0xc4, 0xe2, 0x69, 0xab, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 ab d9       \tvfmsub213ss %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x69, 0xab, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 69 ab 9c c8 78 56 34 12 \tvfmsub213ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0xab, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 ab d9       \tvfmsub213sd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0xab, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 e9 ab 9c c8 78 56 34 12 \tvfmsub213sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x7f, 0x08, 0xab, 0x20, }, 6, 0, "", "",
+"62 f2 7f 08 ab 20    \tv4fnmaddss (%eax),%xmm0,%xmm4",},
+{{0x62, 0xf2, 0x7f, 0x08, 0xab, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 08 ab a4 c8 78 56 34 12 \tv4fnmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4",},
 {{0x62, 0xf2, 0xd5, 0x48, 0xb4, 0xf4, }, 6, 0, "", "",
 "62 f2 d5 48 b4 f4    \tvpmadd52luq %zmm4,%zmm5,%zmm6",},
 {{0x62, 0xf2, 0xd5, 0x48, 0xb5, 0xf4, }, 6, 0, "", "",
@@ -805,6 +1069,50 @@
 "62 f2 4d 0f cd fd    \tvrsqrt28ss %xmm5,%xmm6,%xmm7{%k7}",},
 {{0x62, 0xf2, 0xcd, 0x0f, 0xcd, 0xfd, }, 6, 0, "", "",
 "62 f2 cd 0f cd fd    \tvrsqrt28sd %xmm5,%xmm6,%xmm7{%k7}",},
+{{0x66, 0x0f, 0x38, 0xcf, 0xd9, }, 5, 0, "", "",
+"66 0f 38 cf d9       \tgf2p8mulb %xmm1,%xmm3",},
+{{0x66, 0x0f, 0x38, 0xcf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"66 0f 38 cf 9c c8 78 56 34 12 \tgf2p8mulb 0x12345678(%eax,%ecx,8),%xmm3",},
+{{0xc4, 0xe2, 0x69, 0xcf, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 cf d9       \tvgf2p8mulb %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xcf, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d cf d9       \tvgf2p8mulb %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xcf, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 cf d9    \tvgf2p8mulb %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xcf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 cf 9c c8 78 56 34 12 \tvgf2p8mulb 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xdc, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 dc d9       \tvaesenc %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xdc, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d dc d9       \tvaesenc %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdc, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 dc d9    \tvaesenc %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdc, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 dc 9c c8 78 56 34 12 \tvaesenc 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xdd, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 dd d9       \tvaesenclast %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xdd, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d dd d9       \tvaesenclast %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdd, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 dd d9    \tvaesenclast %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdd, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 dd 9c c8 78 56 34 12 \tvaesenclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xde, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 de d9       \tvaesdec %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xde, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d de d9       \tvaesdec %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xde, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 de d9    \tvaesdec %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xde, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 de 9c c8 78 56 34 12 \tvaesdec 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xdf, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 df d9       \tvaesdeclast %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xdf, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d df d9       \tvaesdeclast %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdf, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 df d9    \tvaesdeclast %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 df 9c c8 78 56 34 12 \tvaesdeclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
 {{0x62, 0xf3, 0x4d, 0x48, 0x03, 0xfd, 0x12, }, 7, 0, "", "",
 "62 f3 4d 48 03 fd 12 \tvalignd $0x12,%zmm5,%zmm6,%zmm7",},
 {{0x62, 0xf3, 0xcd, 0x48, 0x03, 0xfd, 0x12, }, 7, 0, "", "",
@@ -905,6 +1213,12 @@
 "62 f3 4d 48 43 fd 12 \tvshufi32x4 $0x12,%zmm5,%zmm6,%zmm7",},
 {{0x62, 0xf3, 0xcd, 0x48, 0x43, 0xfd, 0x12, }, 7, 0, "", "",
 "62 f3 cd 48 43 fd 12 \tvshufi64x2 $0x12,%zmm5,%zmm6,%zmm7",},
+{{0xc4, 0xe3, 0x69, 0x44, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 69 44 d9 12    \tvpclmulqdq $0x12,%xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe3, 0x6d, 0x44, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 6d 44 d9 12    \tvpclmulqdq $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0x6d, 0x48, 0x44, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 48 44 d9 12 \tvpclmulqdq $0x12,%zmm1,%zmm2,%zmm3",},
 {{0x62, 0xf3, 0x4d, 0x48, 0x50, 0xfd, 0x12, }, 7, 0, "", "",
 "62 f3 4d 48 50 fd 12 \tvrangeps $0x12,%zmm5,%zmm6,%zmm7",},
 {{0x62, 0xf3, 0xcd, 0x48, 0x50, 0xfd, 0x12, }, 7, 0, "", "",
@@ -937,6 +1251,58 @@
 "62 f3 7d 08 67 ef 12 \tvfpclassss $0x12,%xmm7,%k5",},
 {{0x62, 0xf3, 0xfd, 0x08, 0x67, 0xef, 0x12, }, 7, 0, "", "",
 "62 f3 fd 08 67 ef 12 \tvfpclasssd $0x12,%xmm7,%k5",},
+{{0x62, 0xf3, 0xed, 0x08, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 70 d9 12 \tvpshldw $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 70 d9 12 \tvpshldw $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 70 d9 12 \tvpshldw $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf3, 0x6d, 0x08, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 08 71 d9 12 \tvpshldd $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0x6d, 0x28, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 28 71 d9 12 \tvpshldd $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0x6d, 0x48, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 48 71 d9 12 \tvpshldd $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf3, 0xed, 0x08, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 71 d9 12 \tvpshldq $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 71 d9 12 \tvpshldq $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 71 d9 12 \tvpshldq $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf3, 0xed, 0x08, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 72 d9 12 \tvpshrdw $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 72 d9 12 \tvpshrdw $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 72 d9 12 \tvpshrdw $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf3, 0x6d, 0x08, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 08 73 d9 12 \tvpshrdd $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0x6d, 0x28, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 28 73 d9 12 \tvpshrdd $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0x6d, 0x48, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 48 73 d9 12 \tvpshrdd $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf3, 0xed, 0x08, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 73 d9 12 \tvpshrdq $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 73 d9 12 \tvpshrdq $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 73 d9 12 \tvpshrdq $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x66, 0x0f, 0x3a, 0xce, 0xd9, 0x12, }, 6, 0, "", "",
+"66 0f 3a ce d9 12    \tgf2p8affineqb $0x12,%xmm1,%xmm3",},
+{{0xc4, 0xe3, 0xe9, 0xce, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 e9 ce d9 12    \tvgf2p8affineqb $0x12,%xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe3, 0xed, 0xce, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 ed ce d9 12    \tvgf2p8affineqb $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0xce, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 ce d9 12 \tvgf2p8affineqb $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x66, 0x0f, 0x3a, 0xcf, 0xd9, 0x12, }, 6, 0, "", "",
+"66 0f 3a cf d9 12    \tgf2p8affineinvqb $0x12,%xmm1,%xmm3",},
+{{0xc4, 0xe3, 0xe9, 0xcf, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 e9 cf d9 12    \tvgf2p8affineinvqb $0x12,%xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe3, 0xed, 0xcf, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 ed cf d9 12    \tvgf2p8affineinvqb $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0xcf, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 cf d9 12 \tvgf2p8affineinvqb $0x12,%zmm1,%zmm2,%zmm3",},
 {{0x62, 0xf1, 0x4d, 0x48, 0x72, 0xc5, 0x12, }, 7, 0, "", "",
 "62 f1 4d 48 72 c5 12 \tvprord $0x12,%zmm5,%zmm6",},
 {{0x62, 0xf1, 0xcd, 0x48, 0x72, 0xc5, 0x12, }, 7, 0, "", "",
diff --git a/tools/perf/arch/x86/tests/insn-x86-dat-64.c b/tools/perf/arch/x86/tests/insn-x86-dat-64.c
index 656f8ae..567eccc 100644
--- a/tools/perf/arch/x86/tests/insn-x86-dat-64.c
+++ b/tools/perf/arch/x86/tests/insn-x86-dat-64.c
@@ -587,6 +587,112 @@
 "62 02 35 07 4f d0    \tvrsqrt14ss %xmm24,%xmm25,%xmm26{%k7}",},
 {{0x62, 0x02, 0xb5, 0x07, 0x4f, 0xd0, }, 6, 0, "", "",
 "62 02 b5 07 4f d0    \tvrsqrt14sd %xmm24,%xmm25,%xmm26{%k7}",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x50, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 50 d9    \tvpdpbusd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x50, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 50 d9    \tvpdpbusd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x50, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 50 d9    \tvpdpbusd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x50, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 50 9c c8 78 56 34 12 \tvpdpbusd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x50, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 50 9c c8 78 56 34 12 \tvpdpbusd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x51, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 51 d9    \tvpdpbusds %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x51, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 51 d9    \tvpdpbusds %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x51, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 51 d9    \tvpdpbusds %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x51, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 51 9c c8 78 56 34 12 \tvpdpbusds 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x51, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 51 9c c8 78 56 34 12 \tvpdpbusds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6e, 0x08, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6e 08 52 d9    \tvdpbf16ps %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6e, 0x28, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6e 28 52 d9    \tvdpbf16ps %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6e, 0x48, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6e 48 52 d9    \tvdpbf16ps %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6e, 0x48, 0x52, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6e 48 52 9c c8 78 56 34 12 \tvdpbf16ps 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6e, 0x48, 0x52, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6e 48 52 9c c8 78 56 34 12 \tvdpbf16ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 52 d9    \tvpdpwssd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 52 d9    \tvpdpwssd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x52, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 52 d9    \tvpdpwssd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x52, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 52 9c c8 78 56 34 12 \tvpdpwssd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x52, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 52 9c c8 78 56 34 12 \tvpdpwssd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x52, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 52 20    \tvp4dpwssd (%rax),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0x52, 0x20, }, 7, 0, "", "",
+"67 62 f2 7f 48 52 20 \tvp4dpwssd (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x52, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 52 a4 c8 78 56 34 12 \tvp4dpwssd 0x12345678(%rax,%rcx,8),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0x52, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7f 48 52 a4 c8 78 56 34 12 \tvp4dpwssd 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x53, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 53 d9    \tvpdpwssds %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x53, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 53 d9    \tvpdpwssds %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x53, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 53 d9    \tvpdpwssds %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x53, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 53 9c c8 78 56 34 12 \tvpdpwssds 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x53, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 53 9c c8 78 56 34 12 \tvpdpwssds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x53, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 53 20    \tvp4dpwssds (%rax),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0x53, 0x20, }, 7, 0, "", "",
+"67 62 f2 7f 48 53 20 \tvp4dpwssds (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x53, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 53 a4 c8 78 56 34 12 \tvp4dpwssds 0x12345678(%rax,%rcx,8),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0x53, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7f 48 53 a4 c8 78 56 34 12 \tvp4dpwssds 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 08 54 d1    \tvpopcntb %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 28 54 d1    \tvpopcntb %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 48 54 d1    \tvpopcntb %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x54, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 54 94 c8 78 56 34 12 \tvpopcntb 0x12345678(%rax,%rcx,8),%zmm2",},
+{{0x67, 0x62, 0xf2, 0x7d, 0x48, 0x54, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7d 48 54 94 c8 78 56 34 12 \tvpopcntb 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 08 54 d1    \tvpopcntw %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 28 54 d1    \tvpopcntw %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x54, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 48 54 d1    \tvpopcntw %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x54, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 54 94 c8 78 56 34 12 \tvpopcntw 0x12345678(%rax,%rcx,8),%zmm2",},
+{{0x67, 0x62, 0xf2, 0xfd, 0x48, 0x54, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 fd 48 54 94 c8 78 56 34 12 \tvpopcntw 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 08 55 d1    \tvpopcntd %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 28 55 d1    \tvpopcntd %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 48 55 d1    \tvpopcntd %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x55, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 55 94 c8 78 56 34 12 \tvpopcntd 0x12345678(%rax,%rcx,8),%zmm2",},
+{{0x67, 0x62, 0xf2, 0x7d, 0x48, 0x55, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7d 48 55 94 c8 78 56 34 12 \tvpopcntd 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 08 55 d1    \tvpopcntq %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 28 55 d1    \tvpopcntq %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x55, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 48 55 d1    \tvpopcntq %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x55, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 55 94 c8 78 56 34 12 \tvpopcntq 0x12345678(%rax,%rcx,8),%zmm2",},
+{{0x67, 0x62, 0xf2, 0xfd, 0x48, 0x55, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 fd 48 55 94 c8 78 56 34 12 \tvpopcntq 0x12345678(%eax,%ecx,8),%zmm2",},
 {{0xc4, 0xe2, 0x79, 0x59, 0xf4, }, 5, 0, "", "",
 "c4 e2 79 59 f4       \tvpbroadcastq %xmm4,%xmm6",},
 {{0x62, 0x02, 0x7d, 0x48, 0x59, 0xd3, }, 6, 0, "", "",
@@ -601,6 +707,46 @@
 "62 62 7d 48 5b 21    \tvbroadcasti32x8 (%rcx),%zmm28",},
 {{0x62, 0x62, 0xfd, 0x48, 0x5b, 0x11, }, 6, 0, "", "",
 "62 62 fd 48 5b 11    \tvbroadcasti64x4 (%rcx),%zmm26",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 08 62 d1    \tvpexpandb %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 28 62 d1    \tvpexpandb %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 7d 48 62 d1    \tvpexpandb %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x62, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 62 94 c8 78 56 34 12 \tvpexpandb 0x12345678(%rax,%rcx,8),%zmm2",},
+{{0x67, 0x62, 0xf2, 0x7d, 0x48, 0x62, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7d 48 62 94 c8 78 56 34 12 \tvpexpandb 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 08 62 d1    \tvpexpandw %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 28 62 d1    \tvpexpandw %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x62, 0xd1, }, 6, 0, "", "",
+"62 f2 fd 48 62 d1    \tvpexpandw %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x62, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 62 94 c8 78 56 34 12 \tvpexpandw 0x12345678(%rax,%rcx,8),%zmm2",},
+{{0x67, 0x62, 0xf2, 0xfd, 0x48, 0x62, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 fd 48 62 94 c8 78 56 34 12 \tvpexpandw 0x12345678(%eax,%ecx,8),%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x08, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 7d 08 63 ca    \tvpcompressb %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7d, 0x28, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 7d 28 63 ca    \tvpcompressb %ymm1,%ymm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 7d 48 63 ca    \tvpcompressb %zmm1,%zmm2",},
+{{0x62, 0xf2, 0x7d, 0x48, 0x63, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7d 48 63 94 c8 78 56 34 12 \tvpcompressb %zmm2,0x12345678(%rax,%rcx,8)",},
+{{0x67, 0x62, 0xf2, 0x7d, 0x48, 0x63, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7d 48 63 94 c8 78 56 34 12 \tvpcompressb %zmm2,0x12345678(%eax,%ecx,8)",},
+{{0x62, 0xf2, 0xfd, 0x08, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 fd 08 63 ca    \tvpcompressw %xmm1,%xmm2",},
+{{0x62, 0xf2, 0xfd, 0x28, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 fd 28 63 ca    \tvpcompressw %ymm1,%ymm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x63, 0xca, }, 6, 0, "", "",
+"62 f2 fd 48 63 ca    \tvpcompressw %zmm1,%zmm2",},
+{{0x62, 0xf2, 0xfd, 0x48, 0x63, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 fd 48 63 94 c8 78 56 34 12 \tvpcompressw %zmm2,0x12345678(%rax,%rcx,8)",},
+{{0x67, 0x62, 0xf2, 0xfd, 0x48, 0x63, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 fd 48 63 94 c8 78 56 34 12 \tvpcompressw %zmm2,0x12345678(%eax,%ecx,8)",},
 {{0x62, 0x02, 0x25, 0x40, 0x64, 0xe2, }, 6, 0, "", "",
 "62 02 25 40 64 e2    \tvpblendmd %zmm26,%zmm27,%zmm28",},
 {{0x62, 0x02, 0xa5, 0x40, 0x64, 0xe2, }, 6, 0, "", "",
@@ -613,6 +759,106 @@
 "62 02 25 40 66 e2    \tvpblendmb %zmm26,%zmm27,%zmm28",},
 {{0x62, 0x02, 0xa5, 0x40, 0x66, 0xe2, }, 6, 0, "", "",
 "62 02 a5 40 66 e2    \tvpblendmw %zmm26,%zmm27,%zmm28",},
+{{0x62, 0xf2, 0x6f, 0x08, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 08 68 d9    \tvp2intersectd %xmm1,%xmm2,%k3",},
+{{0x62, 0xf2, 0x6f, 0x28, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 28 68 d9    \tvp2intersectd %ymm1,%ymm2,%k3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 48 68 d9    \tvp2intersectd %zmm1,%zmm2,%k3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x68, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6f 48 68 9c c8 78 56 34 12 \tvp2intersectd 0x12345678(%rax,%rcx,8),%zmm2,%k3",},
+{{0x67, 0x62, 0xf2, 0x6f, 0x48, 0x68, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6f 48 68 9c c8 78 56 34 12 \tvp2intersectd 0x12345678(%eax,%ecx,8),%zmm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x08, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 ef 08 68 d9    \tvp2intersectq %xmm1,%xmm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x28, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 ef 28 68 d9    \tvp2intersectq %ymm1,%ymm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x48, 0x68, 0xd9, }, 6, 0, "", "",
+"62 f2 ef 48 68 d9    \tvp2intersectq %zmm1,%zmm2,%k3",},
+{{0x62, 0xf2, 0xef, 0x48, 0x68, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ef 48 68 9c c8 78 56 34 12 \tvp2intersectq 0x12345678(%rax,%rcx,8),%zmm2,%k3",},
+{{0x67, 0x62, 0xf2, 0xef, 0x48, 0x68, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ef 48 68 9c c8 78 56 34 12 \tvp2intersectq 0x12345678(%eax,%ecx,8),%zmm2,%k3",},
+{{0x62, 0xf2, 0xed, 0x08, 0x70, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 70 d9    \tvpshldvw %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x70, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 70 d9    \tvpshldvw %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x70, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 70 d9    \tvpshldvw %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x70, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 70 9c c8 78 56 34 12 \tvpshldvw 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0xed, 0x48, 0x70, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ed 48 70 9c c8 78 56 34 12 \tvpshldvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 71 d9    \tvpshldvd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 71 d9    \tvpshldvd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 71 d9    \tvpshldvd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x71, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 71 9c c8 78 56 34 12 \tvpshldvd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x71, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 71 9c c8 78 56 34 12 \tvpshldvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x08, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 71 d9    \tvpshldvq %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 71 d9    \tvpshldvq %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x71, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 71 d9    \tvpshldvq %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x71, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 71 9c c8 78 56 34 12 \tvpshldvq 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0xed, 0x48, 0x71, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ed 48 71 9c c8 78 56 34 12 \tvpshldvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6f, 0x08, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 08 72 d9    \tvcvtne2ps2bf16 %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6f, 0x28, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 28 72 d9    \tvcvtne2ps2bf16 %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 6f 48 72 d9    \tvcvtne2ps2bf16 %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6f, 0x48, 0x72, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6f 48 72 9c c8 78 56 34 12 \tvcvtne2ps2bf16 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6f, 0x48, 0x72, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6f 48 72 9c c8 78 56 34 12 \tvcvtne2ps2bf16 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7e, 0x08, 0x72, 0xd1, }, 6, 0, "", "",
+"62 f2 7e 08 72 d1    \tvcvtneps2bf16 %xmm1,%xmm2",},
+{{0x62, 0xf2, 0x7e, 0x28, 0x72, 0xd1, }, 6, 0, "", "",
+"62 f2 7e 28 72 d1    \tvcvtneps2bf16 %ymm1,%xmm2",},
+{{0x62, 0xf2, 0x7e, 0x48, 0x72, 0xd1, }, 6, 0, "", "",
+"62 f2 7e 48 72 d1    \tvcvtneps2bf16 %zmm1,%ymm2",},
+{{0x62, 0xf2, 0x7e, 0x48, 0x72, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7e 48 72 94 c8 78 56 34 12 \tvcvtneps2bf16 0x12345678(%rax,%rcx,8),%ymm2",},
+{{0x67, 0x62, 0xf2, 0x7e, 0x48, 0x72, 0x94, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7e 48 72 94 c8 78 56 34 12 \tvcvtneps2bf16 0x12345678(%eax,%ecx,8),%ymm2",},
+{{0x62, 0xf2, 0xed, 0x08, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 72 d9    \tvpshrdvw %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 72 d9    \tvpshrdvw %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x72, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 72 d9    \tvpshrdvw %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x72, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 72 9c c8 78 56 34 12 \tvpshrdvw 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0xed, 0x48, 0x72, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ed 48 72 9c c8 78 56 34 12 \tvpshrdvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 73 d9    \tvpshrdvd %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 73 d9    \tvpshrdvd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 73 d9    \tvpshrdvd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x73, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 73 9c c8 78 56 34 12 \tvpshrdvd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x73, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 73 9c c8 78 56 34 12 \tvpshrdvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x08, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 08 73 d9    \tvpshrdvq %xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf2, 0xed, 0x28, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 28 73 d9    \tvpshrdvq %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x73, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 73 d9    \tvpshrdvq %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x73, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 73 9c c8 78 56 34 12 \tvpshrdvq 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0xed, 0x48, 0x73, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ed 48 73 9c c8 78 56 34 12 \tvpshrdvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
 {{0x62, 0x02, 0x35, 0x40, 0x75, 0xd0, }, 6, 0, "", "",
 "62 02 35 40 75 d0    \tvpermi2b %zmm24,%zmm25,%zmm26",},
 {{0x62, 0x02, 0xa5, 0x40, 0x75, 0xe2, }, 6, 0, "", "",
@@ -667,6 +913,16 @@
 "62 02 25 40 8d e2    \tvpermb %zmm26,%zmm27,%zmm28",},
 {{0x62, 0x02, 0xa5, 0x40, 0x8d, 0xe2, }, 6, 0, "", "",
 "62 02 a5 40 8d e2    \tvpermw %zmm26,%zmm27,%zmm28",},
+{{0x62, 0xf2, 0x6d, 0x08, 0x8f, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 08 8f d9    \tvpshufbitqmb %xmm1,%xmm2,%k3",},
+{{0x62, 0xf2, 0x6d, 0x28, 0x8f, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 28 8f d9    \tvpshufbitqmb %ymm1,%ymm2,%k3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x8f, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 8f d9    \tvpshufbitqmb %zmm1,%zmm2,%k3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x8f, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 8f 9c c8 78 56 34 12 \tvpshufbitqmb 0x12345678(%rax,%rcx,8),%zmm2,%k3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x8f, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 8f 9c c8 78 56 34 12 \tvpshufbitqmb 0x12345678(%eax,%ecx,8),%zmm2,%k3",},
 {{0xc4, 0xe2, 0x69, 0x90, 0x4c, 0x7d, 0x02, }, 7, 0, "", "",
 "c4 e2 69 90 4c 7d 02 \tvpgatherdd %xmm2,0x2(%rbp,%xmm7,2),%xmm1",},
 {{0xc4, 0xe2, 0xe9, 0x90, 0x4c, 0x7d, 0x04, }, 7, 0, "", "",
@@ -683,6 +939,54 @@
 "62 22 7d 41 91 94 dd 7b 00 00 00 \tvpgatherqd 0x7b(%rbp,%zmm27,8),%ymm26{%k1}",},
 {{0x62, 0x22, 0xfd, 0x41, 0x91, 0x94, 0xdd, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
 "62 22 fd 41 91 94 dd 7b 00 00 00 \tvpgatherqq 0x7b(%rbp,%zmm27,8),%zmm26{%k1}",},
+{{0xc4, 0xe2, 0x69, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 9a d9       \tvfmsub132ps %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d 9a d9       \tvfmsub132ps %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x9a, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 9a d9    \tvfmsub132ps %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0x9a, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 9a 9c c8 78 56 34 12 \tvfmsub132ps 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0x9a, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 9a 9c c8 78 56 34 12 \tvfmsub132ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0xe9, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 9a d9       \tvfmsub132pd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xed, 0x9a, 0xd9, }, 5, 0, "", "",
+"c4 e2 ed 9a d9       \tvfmsub132pd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x9a, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 9a d9    \tvfmsub132pd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0x9a, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 9a 9c c8 78 56 34 12 \tvfmsub132pd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0xed, 0x48, 0x9a, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ed 48 9a 9c c8 78 56 34 12 \tvfmsub132pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x9a, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 9a 20    \tv4fmaddps (%rax),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0x9a, 0x20, }, 7, 0, "", "",
+"67 62 f2 7f 48 9a 20 \tv4fmaddps (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0x9a, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 9a a4 c8 78 56 34 12 \tv4fmaddps 0x12345678(%rax,%rcx,8),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0x9a, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7f 48 9a a4 c8 78 56 34 12 \tv4fmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0xc4, 0xe2, 0x69, 0x9b, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 9b d9       \tvfmsub132ss %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x69, 0x9b, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 69 9b 9c c8 78 56 34 12 \tvfmsub132ss 0x12345678(%rax,%rcx,8),%xmm2,%xmm3",},
+{{0x67, 0xc4, 0xe2, 0x69, 0x9b, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"67 c4 e2 69 9b 9c c8 78 56 34 12 \tvfmsub132ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0x9b, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 9b d9       \tvfmsub132sd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0x9b, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 e9 9b 9c c8 78 56 34 12 \tvfmsub132sd 0x12345678(%rax,%rcx,8),%xmm2,%xmm3",},
+{{0x67, 0xc4, 0xe2, 0xe9, 0x9b, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"67 c4 e2 e9 9b 9c c8 78 56 34 12 \tvfmsub132sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x7f, 0x08, 0x9b, 0x20, }, 6, 0, "", "",
+"62 f2 7f 08 9b 20    \tv4fmaddss (%rax),%xmm0,%xmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x08, 0x9b, 0x20, }, 7, 0, "", "",
+"67 62 f2 7f 08 9b 20 \tv4fmaddss (%eax),%xmm0,%xmm4",},
+{{0x62, 0xf2, 0x7f, 0x08, 0x9b, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 08 9b a4 c8 78 56 34 12 \tv4fmaddss 0x12345678(%rax,%rcx,8),%xmm0,%xmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x08, 0x9b, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7f 08 9b a4 c8 78 56 34 12 \tv4fmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4",},
 {{0x62, 0x22, 0x7d, 0x41, 0xa0, 0xa4, 0xed, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
 "62 22 7d 41 a0 a4 ed 7b 00 00 00 \tvpscatterdd %zmm28,0x7b(%rbp,%zmm29,8){%k1}",},
 {{0x62, 0x22, 0xfd, 0x41, 0xa0, 0x94, 0xdd, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
@@ -699,6 +1003,54 @@
 "62 b2 7d 41 a3 b4 ed 7b 00 00 00 \tvscatterqps %ymm6,0x7b(%rbp,%zmm29,8){%k1}",},
 {{0x62, 0x22, 0xfd, 0x41, 0xa3, 0xa4, 0xed, 0x7b, 0x00, 0x00, 0x00, }, 11, 0, "", "",
 "62 22 fd 41 a3 a4 ed 7b 00 00 00 \tvscatterqpd %zmm28,0x7b(%rbp,%zmm29,8){%k1}",},
+{{0xc4, 0xe2, 0x69, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 aa d9       \tvfmsub213ps %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d aa d9       \tvfmsub213ps %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xaa, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 aa d9    \tvfmsub213ps %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xaa, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 aa 9c c8 78 56 34 12 \tvfmsub213ps 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0xaa, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 aa 9c c8 78 56 34 12 \tvfmsub213ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0xe9, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 aa d9       \tvfmsub213pd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xed, 0xaa, 0xd9, }, 5, 0, "", "",
+"c4 e2 ed aa d9       \tvfmsub213pd %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0xaa, 0xd9, }, 6, 0, "", "",
+"62 f2 ed 48 aa d9    \tvfmsub213pd %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0xed, 0x48, 0xaa, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 ed 48 aa 9c c8 78 56 34 12 \tvfmsub213pd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0xed, 0x48, 0xaa, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 ed 48 aa 9c c8 78 56 34 12 \tvfmsub213pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x7f, 0x48, 0xaa, 0x20, }, 6, 0, "", "",
+"62 f2 7f 48 aa 20    \tv4fnmaddps (%rax),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0xaa, 0x20, }, 7, 0, "", "",
+"67 62 f2 7f 48 aa 20 \tv4fnmaddps (%eax),%zmm0,%zmm4",},
+{{0x62, 0xf2, 0x7f, 0x48, 0xaa, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 48 aa a4 c8 78 56 34 12 \tv4fnmaddps 0x12345678(%rax,%rcx,8),%zmm0,%zmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x48, 0xaa, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7f 48 aa a4 c8 78 56 34 12 \tv4fnmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4",},
+{{0xc4, 0xe2, 0x69, 0xab, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 ab d9       \tvfmsub213ss %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x69, 0xab, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 69 ab 9c c8 78 56 34 12 \tvfmsub213ss 0x12345678(%rax,%rcx,8),%xmm2,%xmm3",},
+{{0x67, 0xc4, 0xe2, 0x69, 0xab, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"67 c4 e2 69 ab 9c c8 78 56 34 12 \tvfmsub213ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0xab, 0xd9, }, 5, 0, "", "",
+"c4 e2 e9 ab d9       \tvfmsub213sd %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0xe9, 0xab, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"c4 e2 e9 ab 9c c8 78 56 34 12 \tvfmsub213sd 0x12345678(%rax,%rcx,8),%xmm2,%xmm3",},
+{{0x67, 0xc4, 0xe2, 0xe9, 0xab, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"67 c4 e2 e9 ab 9c c8 78 56 34 12 \tvfmsub213sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3",},
+{{0x62, 0xf2, 0x7f, 0x08, 0xab, 0x20, }, 6, 0, "", "",
+"62 f2 7f 08 ab 20    \tv4fnmaddss (%rax),%xmm0,%xmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x08, 0xab, 0x20, }, 7, 0, "", "",
+"67 62 f2 7f 08 ab 20 \tv4fnmaddss (%eax),%xmm0,%xmm4",},
+{{0x62, 0xf2, 0x7f, 0x08, 0xab, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 7f 08 ab a4 c8 78 56 34 12 \tv4fnmaddss 0x12345678(%rax,%rcx,8),%xmm0,%xmm4",},
+{{0x67, 0x62, 0xf2, 0x7f, 0x08, 0xab, 0xa4, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 7f 08 ab a4 c8 78 56 34 12 \tv4fnmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4",},
 {{0x62, 0x02, 0xa5, 0x40, 0xb4, 0xe2, }, 6, 0, "", "",
 "62 02 a5 40 b4 e2    \tvpmadd52luq %zmm26,%zmm27,%zmm28",},
 {{0x62, 0x02, 0xa5, 0x40, 0xb5, 0xe2, }, 6, 0, "", "",
@@ -727,6 +1079,62 @@
 "62 02 15 07 cd f4    \tvrsqrt28ss %xmm28,%xmm29,%xmm30{%k7}",},
 {{0x62, 0x02, 0xad, 0x07, 0xcd, 0xd9, }, 6, 0, "", "",
 "62 02 ad 07 cd d9    \tvrsqrt28sd %xmm25,%xmm26,%xmm27{%k7}",},
+{{0x66, 0x0f, 0x38, 0xcf, 0xd9, }, 5, 0, "", "",
+"66 0f 38 cf d9       \tgf2p8mulb %xmm1,%xmm3",},
+{{0x66, 0x0f, 0x38, 0xcf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 10, 0, "", "",
+"66 0f 38 cf 9c c8 78 56 34 12 \tgf2p8mulb 0x12345678(%rax,%rcx,8),%xmm3",},
+{{0x67, 0x66, 0x0f, 0x38, 0xcf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"67 66 0f 38 cf 9c c8 78 56 34 12 \tgf2p8mulb 0x12345678(%eax,%ecx,8),%xmm3",},
+{{0xc4, 0xe2, 0x69, 0xcf, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 cf d9       \tvgf2p8mulb %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xcf, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d cf d9       \tvgf2p8mulb %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xcf, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 cf d9    \tvgf2p8mulb %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xcf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 cf 9c c8 78 56 34 12 \tvgf2p8mulb 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0xcf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 cf 9c c8 78 56 34 12 \tvgf2p8mulb 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xdc, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 dc d9       \tvaesenc %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xdc, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d dc d9       \tvaesenc %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdc, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 dc d9    \tvaesenc %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdc, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 dc 9c c8 78 56 34 12 \tvaesenc 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0xdc, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 dc 9c c8 78 56 34 12 \tvaesenc 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xdd, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 dd d9       \tvaesenclast %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xdd, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d dd d9       \tvaesenclast %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdd, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 dd d9    \tvaesenclast %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdd, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 dd 9c c8 78 56 34 12 \tvaesenclast 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0xdd, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 dd 9c c8 78 56 34 12 \tvaesenclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xde, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 de d9       \tvaesdec %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xde, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d de d9       \tvaesdec %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xde, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 de d9    \tvaesdec %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xde, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 de 9c c8 78 56 34 12 \tvaesdec 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0xde, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 de 9c c8 78 56 34 12 \tvaesdec 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
+{{0xc4, 0xe2, 0x69, 0xdf, 0xd9, }, 5, 0, "", "",
+"c4 e2 69 df d9       \tvaesdeclast %xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe2, 0x6d, 0xdf, 0xd9, }, 5, 0, "", "",
+"c4 e2 6d df d9       \tvaesdeclast %ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdf, 0xd9, }, 6, 0, "", "",
+"62 f2 6d 48 df d9    \tvaesdeclast %zmm1,%zmm2,%zmm3",},
+{{0x62, 0xf2, 0x6d, 0x48, 0xdf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 11, 0, "", "",
+"62 f2 6d 48 df 9c c8 78 56 34 12 \tvaesdeclast 0x12345678(%rax,%rcx,8),%zmm2,%zmm3",},
+{{0x67, 0x62, 0xf2, 0x6d, 0x48, 0xdf, 0x9c, 0xc8, 0x78, 0x56, 0x34, 0x12, }, 12, 0, "", "",
+"67 62 f2 6d 48 df 9c c8 78 56 34 12 \tvaesdeclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3",},
 {{0x62, 0x03, 0x15, 0x40, 0x03, 0xf4, 0x12, }, 7, 0, "", "",
 "62 03 15 40 03 f4 12 \tvalignd $0x12,%zmm28,%zmm29,%zmm30",},
 {{0x62, 0x03, 0xad, 0x40, 0x03, 0xd9, 0x12, }, 7, 0, "", "",
@@ -827,6 +1235,14 @@
 "62 03 2d 40 43 d9 12 \tvshufi32x4 $0x12,%zmm25,%zmm26,%zmm27",},
 {{0x62, 0x03, 0x95, 0x40, 0x43, 0xf4, 0x12, }, 7, 0, "", "",
 "62 03 95 40 43 f4 12 \tvshufi64x2 $0x12,%zmm28,%zmm29,%zmm30",},
+{{0xc4, 0xe3, 0x69, 0x44, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 69 44 d9 12    \tvpclmulqdq $0x12,%xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe3, 0x6d, 0x44, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 6d 44 d9 12    \tvpclmulqdq $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0x6d, 0x48, 0x44, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 48 44 d9 12 \tvpclmulqdq $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0x2d, 0x40, 0x44, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 2d 40 44 d9 12 \tvpclmulqdq $0x12,%zmm25,%zmm26,%zmm27",},
 {{0x62, 0x03, 0x2d, 0x40, 0x50, 0xd9, 0x12, }, 7, 0, "", "",
 "62 03 2d 40 50 d9 12 \tvrangeps $0x12,%zmm25,%zmm26,%zmm27",},
 {{0x62, 0x03, 0x95, 0x40, 0x50, 0xf4, 0x12, }, 7, 0, "", "",
@@ -859,6 +1275,74 @@
 "62 93 7d 08 67 eb 12 \tvfpclassss $0x12,%xmm27,%k5",},
 {{0x62, 0x93, 0xfd, 0x08, 0x67, 0xee, 0x12, }, 7, 0, "", "",
 "62 93 fd 08 67 ee 12 \tvfpclasssd $0x12,%xmm30,%k5",},
+{{0x62, 0xf3, 0xed, 0x08, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 70 d9 12 \tvpshldw $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 70 d9 12 \tvpshldw $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 70 d9 12 \tvpshldw $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0xad, 0x40, 0x70, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 ad 40 70 d9 12 \tvpshldw $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x62, 0xf3, 0x6d, 0x08, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 08 71 d9 12 \tvpshldd $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0x6d, 0x28, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 28 71 d9 12 \tvpshldd $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0x6d, 0x48, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 48 71 d9 12 \tvpshldd $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0x2d, 0x40, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 2d 40 71 d9 12 \tvpshldd $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x62, 0xf3, 0xed, 0x08, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 71 d9 12 \tvpshldq $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 71 d9 12 \tvpshldq $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 71 d9 12 \tvpshldq $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0xad, 0x40, 0x71, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 ad 40 71 d9 12 \tvpshldq $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x62, 0xf3, 0xed, 0x08, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 72 d9 12 \tvpshrdw $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 72 d9 12 \tvpshrdw $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 72 d9 12 \tvpshrdw $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0xad, 0x40, 0x72, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 ad 40 72 d9 12 \tvpshrdw $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x62, 0xf3, 0x6d, 0x08, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 08 73 d9 12 \tvpshrdd $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0x6d, 0x28, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 28 73 d9 12 \tvpshrdd $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0x6d, 0x48, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 6d 48 73 d9 12 \tvpshrdd $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0x2d, 0x40, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 2d 40 73 d9 12 \tvpshrdd $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x62, 0xf3, 0xed, 0x08, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 08 73 d9 12 \tvpshrdq $0x12,%xmm1,%xmm2,%xmm3",},
+{{0x62, 0xf3, 0xed, 0x28, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 28 73 d9 12 \tvpshrdq $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 73 d9 12 \tvpshrdq $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0xad, 0x40, 0x73, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 ad 40 73 d9 12 \tvpshrdq $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x66, 0x0f, 0x3a, 0xce, 0xd9, 0x12, }, 6, 0, "", "",
+"66 0f 3a ce d9 12    \tgf2p8affineqb $0x12,%xmm1,%xmm3",},
+{{0xc4, 0xe3, 0xe9, 0xce, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 e9 ce d9 12    \tvgf2p8affineqb $0x12,%xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe3, 0xed, 0xce, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 ed ce d9 12    \tvgf2p8affineqb $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0xce, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 ce d9 12 \tvgf2p8affineqb $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0xad, 0x40, 0xce, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 ad 40 ce d9 12 \tvgf2p8affineqb $0x12,%zmm25,%zmm26,%zmm27",},
+{{0x66, 0x0f, 0x3a, 0xcf, 0xd9, 0x12, }, 6, 0, "", "",
+"66 0f 3a cf d9 12    \tgf2p8affineinvqb $0x12,%xmm1,%xmm3",},
+{{0xc4, 0xe3, 0xe9, 0xcf, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 e9 cf d9 12    \tvgf2p8affineinvqb $0x12,%xmm1,%xmm2,%xmm3",},
+{{0xc4, 0xe3, 0xed, 0xcf, 0xd9, 0x12, }, 6, 0, "", "",
+"c4 e3 ed cf d9 12    \tvgf2p8affineinvqb $0x12,%ymm1,%ymm2,%ymm3",},
+{{0x62, 0xf3, 0xed, 0x48, 0xcf, 0xd9, 0x12, }, 7, 0, "", "",
+"62 f3 ed 48 cf d9 12 \tvgf2p8affineinvqb $0x12,%zmm1,%zmm2,%zmm3",},
+{{0x62, 0x03, 0xad, 0x40, 0xcf, 0xd9, 0x12, }, 7, 0, "", "",
+"62 03 ad 40 cf d9 12 \tvgf2p8affineinvqb $0x12,%zmm25,%zmm26,%zmm27",},
 {{0x62, 0x91, 0x2d, 0x40, 0x72, 0xc1, 0x12, }, 7, 0, "", "",
 "62 91 2d 40 72 c1 12 \tvprord $0x12,%zmm25,%zmm26",},
 {{0x62, 0x91, 0xad, 0x40, 0x72, 0xc1, 0x12, }, 7, 0, "", "",
diff --git a/tools/perf/arch/x86/tests/insn-x86-dat-src.c b/tools/perf/arch/x86/tests/insn-x86-dat-src.c
index dd85a3a..ddbf07c5 100644
--- a/tools/perf/arch/x86/tests/insn-x86-dat-src.c
+++ b/tools/perf/arch/x86/tests/insn-x86-dat-src.c
@@ -510,6 +510,82 @@
 	asm volatile("vrsqrt14ss %xmm24,%xmm25,%xmm26{%k7}");
 	asm volatile("vrsqrt14sd %xmm24,%xmm25,%xmm26{%k7}");
 
+	/* AVX-512: Op code 0f 38 50 */
+
+	asm volatile("vpdpbusd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpbusd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpbusd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpbusd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpdpbusd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 51 */
+
+	asm volatile("vpdpbusds %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpbusds %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpbusds %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpbusds 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpdpbusds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 52 */
+
+	asm volatile("vdpbf16ps %xmm1, %xmm2, %xmm3");
+	asm volatile("vdpbf16ps %ymm1, %ymm2, %ymm3");
+	asm volatile("vdpbf16ps %zmm1, %zmm2, %zmm3");
+	asm volatile("vdpbf16ps 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vdpbf16ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vpdpwssd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpwssd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpwssd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpwssd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpdpwssd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vp4dpwssd (%rax), %zmm0, %zmm4");
+	asm volatile("vp4dpwssd (%eax), %zmm0, %zmm4");
+	asm volatile("vp4dpwssd 0x12345678(%rax,%rcx,8),%zmm0,%zmm4");
+	asm volatile("vp4dpwssd 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 53 */
+
+	asm volatile("vpdpwssds %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpwssds %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpwssds %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpwssds 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpdpwssds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vp4dpwssds (%rax), %zmm0, %zmm4");
+	asm volatile("vp4dpwssds (%eax), %zmm0, %zmm4");
+	asm volatile("vp4dpwssds 0x12345678(%rax,%rcx,8),%zmm0,%zmm4");
+	asm volatile("vp4dpwssds 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 54 */
+
+	asm volatile("vpopcntb %xmm1, %xmm2");
+	asm volatile("vpopcntb %ymm1, %ymm2");
+	asm volatile("vpopcntb %zmm1, %zmm2");
+	asm volatile("vpopcntb 0x12345678(%rax,%rcx,8),%zmm2");
+	asm volatile("vpopcntb 0x12345678(%eax,%ecx,8),%zmm2");
+
+	asm volatile("vpopcntw %xmm1, %xmm2");
+	asm volatile("vpopcntw %ymm1, %ymm2");
+	asm volatile("vpopcntw %zmm1, %zmm2");
+	asm volatile("vpopcntw 0x12345678(%rax,%rcx,8),%zmm2");
+	asm volatile("vpopcntw 0x12345678(%eax,%ecx,8),%zmm2");
+
+	/* AVX-512: Op code 0f 38 55 */
+
+	asm volatile("vpopcntd %xmm1, %xmm2");
+	asm volatile("vpopcntd %ymm1, %ymm2");
+	asm volatile("vpopcntd %zmm1, %zmm2");
+	asm volatile("vpopcntd 0x12345678(%rax,%rcx,8),%zmm2");
+	asm volatile("vpopcntd 0x12345678(%eax,%ecx,8),%zmm2");
+
+	asm volatile("vpopcntq %xmm1, %xmm2");
+	asm volatile("vpopcntq %ymm1, %ymm2");
+	asm volatile("vpopcntq %zmm1, %zmm2");
+	asm volatile("vpopcntq 0x12345678(%rax,%rcx,8),%zmm2");
+	asm volatile("vpopcntq 0x12345678(%eax,%ecx,8),%zmm2");
+
 	/* AVX-512: Op code 0f 38 59 */
 
 	asm volatile("vpbroadcastq %xmm4,%xmm6");
@@ -526,6 +602,34 @@
 	asm volatile("vbroadcasti32x8 (%rcx),%zmm28");
 	asm volatile("vbroadcasti64x4 (%rcx),%zmm26");
 
+	/* AVX-512: Op code 0f 38 62 */
+
+	asm volatile("vpexpandb %xmm1, %xmm2");
+	asm volatile("vpexpandb %ymm1, %ymm2");
+	asm volatile("vpexpandb %zmm1, %zmm2");
+	asm volatile("vpexpandb 0x12345678(%rax,%rcx,8),%zmm2");
+	asm volatile("vpexpandb 0x12345678(%eax,%ecx,8),%zmm2");
+
+	asm volatile("vpexpandw %xmm1, %xmm2");
+	asm volatile("vpexpandw %ymm1, %ymm2");
+	asm volatile("vpexpandw %zmm1, %zmm2");
+	asm volatile("vpexpandw 0x12345678(%rax,%rcx,8),%zmm2");
+	asm volatile("vpexpandw 0x12345678(%eax,%ecx,8),%zmm2");
+
+	/* AVX-512: Op code 0f 38 63 */
+
+	asm volatile("vpcompressb %xmm1, %xmm2");
+	asm volatile("vpcompressb %ymm1, %ymm2");
+	asm volatile("vpcompressb %zmm1, %zmm2");
+	asm volatile("vpcompressb %zmm2,0x12345678(%rax,%rcx,8)");
+	asm volatile("vpcompressb %zmm2,0x12345678(%eax,%ecx,8)");
+
+	asm volatile("vpcompressw %xmm1, %xmm2");
+	asm volatile("vpcompressw %ymm1, %ymm2");
+	asm volatile("vpcompressw %zmm1, %zmm2");
+	asm volatile("vpcompressw %zmm2,0x12345678(%rax,%rcx,8)");
+	asm volatile("vpcompressw %zmm2,0x12345678(%eax,%ecx,8)");
+
 	/* AVX-512: Op code 0f 38 64 */
 
 	asm volatile("vpblendmd %zmm26,%zmm27,%zmm28");
@@ -541,6 +645,76 @@
 	asm volatile("vpblendmb %zmm26,%zmm27,%zmm28");
 	asm volatile("vpblendmw %zmm26,%zmm27,%zmm28");
 
+	/* AVX-512: Op code 0f 38 68 */
+
+	asm volatile("vp2intersectd %xmm1, %xmm2, %k3");
+	asm volatile("vp2intersectd %ymm1, %ymm2, %k3");
+	asm volatile("vp2intersectd %zmm1, %zmm2, %k3");
+	asm volatile("vp2intersectd 0x12345678(%rax,%rcx,8),%zmm2,%k3");
+	asm volatile("vp2intersectd 0x12345678(%eax,%ecx,8),%zmm2,%k3");
+
+	asm volatile("vp2intersectq %xmm1, %xmm2, %k3");
+	asm volatile("vp2intersectq %ymm1, %ymm2, %k3");
+	asm volatile("vp2intersectq %zmm1, %zmm2, %k3");
+	asm volatile("vp2intersectq 0x12345678(%rax,%rcx,8),%zmm2,%k3");
+	asm volatile("vp2intersectq 0x12345678(%eax,%ecx,8),%zmm2,%k3");
+
+	/* AVX-512: Op code 0f 38 70 */
+
+	asm volatile("vpshldvw %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshldvw %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshldvw %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshldvw 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpshldvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 71 */
+
+	asm volatile("vpshldvd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshldvd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshldvd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshldvd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpshldvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vpshldvq %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshldvq %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshldvq %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshldvq 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpshldvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 72 */
+
+	asm volatile("vcvtne2ps2bf16 %xmm1, %xmm2, %xmm3");
+	asm volatile("vcvtne2ps2bf16 %ymm1, %ymm2, %ymm3");
+	asm volatile("vcvtne2ps2bf16 %zmm1, %zmm2, %zmm3");
+	asm volatile("vcvtne2ps2bf16 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vcvtne2ps2bf16 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vcvtneps2bf16 %xmm1, %xmm2");
+	asm volatile("vcvtneps2bf16 %ymm1, %xmm2");
+	asm volatile("vcvtneps2bf16 %zmm1, %ymm2");
+	asm volatile("vcvtneps2bf16 0x12345678(%rax,%rcx,8),%ymm2");
+	asm volatile("vcvtneps2bf16 0x12345678(%eax,%ecx,8),%ymm2");
+
+	asm volatile("vpshrdvw %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshrdvw %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshrdvw %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshrdvw 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpshrdvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 73 */
+
+	asm volatile("vpshrdvd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshrdvd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshrdvd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshrdvd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpshrdvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vpshrdvq %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshrdvq %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshrdvq %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshrdvq 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vpshrdvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
 	/* AVX-512: Op code 0f 38 75 */
 
 	asm volatile("vpermi2b %zmm24,%zmm25,%zmm26");
@@ -613,6 +787,14 @@
 	asm volatile("vpermb %zmm26,%zmm27,%zmm28");
 	asm volatile("vpermw %zmm26,%zmm27,%zmm28");
 
+	/* AVX-512: Op code 0f 38 8f */
+
+	asm volatile("vpshufbitqmb %xmm1, %xmm2, %k3");
+	asm volatile("vpshufbitqmb %ymm1, %ymm2, %k3");
+	asm volatile("vpshufbitqmb %zmm1, %zmm2, %k3");
+	asm volatile("vpshufbitqmb 0x12345678(%rax,%rcx,8),%zmm2,%k3");
+	asm volatile("vpshufbitqmb 0x12345678(%eax,%ecx,8),%zmm2,%k3");
+
 	/* AVX-512: Op code 0f 38 90 */
 
 	asm volatile("vpgatherdd %xmm2,0x02(%rbp,%xmm7,2),%xmm1");
@@ -627,6 +809,40 @@
 	asm volatile("vpgatherqd 0x7b(%rbp,%zmm27,8),%ymm26{%k1}");
 	asm volatile("vpgatherqq 0x7b(%rbp,%zmm27,8),%zmm26{%k1}");
 
+	/* AVX-512: Op code 0f 38 9a */
+
+	asm volatile("vfmsub132ps %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132ps %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub132ps %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub132ps 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vfmsub132ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vfmsub132pd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132pd %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub132pd %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub132pd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vfmsub132pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("v4fmaddps (%rax), %zmm0, %zmm4");
+	asm volatile("v4fmaddps (%eax), %zmm0, %zmm4");
+	asm volatile("v4fmaddps 0x12345678(%rax,%rcx,8),%zmm0,%zmm4");
+	asm volatile("v4fmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 9b */
+
+	asm volatile("vfmsub132ss %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132ss 0x12345678(%rax,%rcx,8),%xmm2,%xmm3");
+	asm volatile("vfmsub132ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("vfmsub132sd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132sd 0x12345678(%rax,%rcx,8),%xmm2,%xmm3");
+	asm volatile("vfmsub132sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("v4fmaddss (%rax), %xmm0, %xmm4");
+	asm volatile("v4fmaddss (%eax), %xmm0, %xmm4");
+	asm volatile("v4fmaddss 0x12345678(%rax,%rcx,8),%xmm0,%xmm4");
+	asm volatile("v4fmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4");
+
 	/* AVX-512: Op code 0f 38 a0 */
 
 	asm volatile("vpscatterdd %zmm28,0x7b(%rbp,%zmm29,8){%k1}");
@@ -647,6 +863,40 @@
 	asm volatile("vscatterqps %ymm6,0x7b(%rbp,%zmm29,8){%k1}");
 	asm volatile("vscatterqpd %zmm28,0x7b(%rbp,%zmm29,8){%k1}");
 
+	/* AVX-512: Op code 0f 38 aa */
+
+	asm volatile("vfmsub213ps %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213ps %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub213ps %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub213ps 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vfmsub213ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vfmsub213pd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213pd %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub213pd %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub213pd 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vfmsub213pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("v4fnmaddps (%rax), %zmm0, %zmm4");
+	asm volatile("v4fnmaddps (%eax), %zmm0, %zmm4");
+	asm volatile("v4fnmaddps 0x12345678(%rax,%rcx,8),%zmm0,%zmm4");
+	asm volatile("v4fnmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 ab */
+
+	asm volatile("vfmsub213ss %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213ss 0x12345678(%rax,%rcx,8),%xmm2,%xmm3");
+	asm volatile("vfmsub213ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("vfmsub213sd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213sd 0x12345678(%rax,%rcx,8),%xmm2,%xmm3");
+	asm volatile("vfmsub213sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("v4fnmaddss (%rax), %xmm0, %xmm4");
+	asm volatile("v4fnmaddss (%eax), %xmm0, %xmm4");
+	asm volatile("v4fnmaddss 0x12345678(%rax,%rcx,8),%xmm0,%xmm4");
+	asm volatile("v4fnmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4");
+
 	/* AVX-512: Op code 0f 38 b4 */
 
 	asm volatile("vpmadd52luq %zmm26,%zmm27,%zmm28");
@@ -685,6 +935,50 @@
 	asm volatile("vrsqrt28ss %xmm28,%xmm29,%xmm30{%k7}");
 	asm volatile("vrsqrt28sd %xmm25,%xmm26,%xmm27{%k7}");
 
+	/* AVX-512: Op code 0f 38 cf */
+
+	asm volatile("gf2p8mulb %xmm1, %xmm3");
+	asm volatile("gf2p8mulb 0x12345678(%rax,%rcx,8),%xmm3");
+	asm volatile("gf2p8mulb 0x12345678(%eax,%ecx,8),%xmm3");
+
+	asm volatile("vgf2p8mulb %xmm1, %xmm2, %xmm3");
+	asm volatile("vgf2p8mulb %ymm1, %ymm2, %ymm3");
+	asm volatile("vgf2p8mulb %zmm1, %zmm2, %zmm3");
+	asm volatile("vgf2p8mulb 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vgf2p8mulb 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 dc */
+
+	asm volatile("vaesenc %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesenc %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesenc %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesenc 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vaesenc 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 dd */
+
+	asm volatile("vaesenclast %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesenclast %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesenclast %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesenclast 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vaesenclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 de */
+
+	asm volatile("vaesdec %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesdec %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesdec %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesdec 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vaesdec 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 df */
+
+	asm volatile("vaesdeclast %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesdeclast %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesdeclast %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesdeclast 0x12345678(%rax,%rcx,8),%zmm2,%zmm3");
+	asm volatile("vaesdeclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
 	/* AVX-512: Op code 0f 3a 03 */
 
 	asm volatile("valignd $0x12,%zmm28,%zmm29,%zmm30");
@@ -804,6 +1098,13 @@
 	asm volatile("vshufi32x4 $0x12,%zmm25,%zmm26,%zmm27");
 	asm volatile("vshufi64x2 $0x12,%zmm28,%zmm29,%zmm30");
 
+	/* AVX-512: Op code 0f 3a 44 */
+
+	asm volatile("vpclmulqdq $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpclmulqdq $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpclmulqdq $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpclmulqdq $0x12,%zmm25,%zmm26,%zmm27");
+
 	/* AVX-512: Op code 0f 3a 50 */
 
 	asm volatile("vrangeps $0x12,%zmm25,%zmm26,%zmm27");
@@ -844,6 +1145,62 @@
 	asm volatile("vfpclassss $0x12,%xmm27,%k5");
 	asm volatile("vfpclasssd $0x12,%xmm30,%k5");
 
+	/* AVX-512: Op code 0f 3a 70 */
+
+	asm volatile("vpshldw $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshldw $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshldw $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpshldw $0x12,%zmm25,%zmm26,%zmm27");
+
+	/* AVX-512: Op code 0f 3a 71 */
+
+	asm volatile("vpshldd $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshldd $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshldd $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpshldd $0x12,%zmm25,%zmm26,%zmm27");
+
+	asm volatile("vpshldq $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshldq $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshldq $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpshldq $0x12,%zmm25,%zmm26,%zmm27");
+
+	/* AVX-512: Op code 0f 3a 72 */
+
+	asm volatile("vpshrdw $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshrdw $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshrdw $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpshrdw $0x12,%zmm25,%zmm26,%zmm27");
+
+	/* AVX-512: Op code 0f 3a 73 */
+
+	asm volatile("vpshrdd $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshrdd $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshrdd $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpshrdd $0x12,%zmm25,%zmm26,%zmm27");
+
+	asm volatile("vpshrdq $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshrdq $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshrdq $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vpshrdq $0x12,%zmm25,%zmm26,%zmm27");
+
+	/* AVX-512: Op code 0f 3a ce */
+
+	asm volatile("gf2p8affineqb $0x12,%xmm1,%xmm3");
+
+	asm volatile("vgf2p8affineqb $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vgf2p8affineqb $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vgf2p8affineqb $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vgf2p8affineqb $0x12,%zmm25,%zmm26,%zmm27");
+
+	/* AVX-512: Op code 0f 3a cf */
+
+	asm volatile("gf2p8affineinvqb $0x12,%xmm1,%xmm3");
+
+	asm volatile("vgf2p8affineinvqb $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vgf2p8affineinvqb $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vgf2p8affineinvqb $0x12,%zmm1,%zmm2,%zmm3");
+	asm volatile("vgf2p8affineinvqb $0x12,%zmm25,%zmm26,%zmm27");
+
 	/* AVX-512: Op code 0f 72 (Grp13) */
 
 	asm volatile("vprord $0x12,%zmm25,%zmm26");
@@ -1946,6 +2303,69 @@
 	asm volatile("vrsqrt14ss %xmm4,%xmm5,%xmm6{%k7}");
 	asm volatile("vrsqrt14sd %xmm4,%xmm5,%xmm6{%k7}");
 
+	/* AVX-512: Op code 0f 38 50 */
+
+	asm volatile("vpdpbusd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpbusd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpbusd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpbusd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 51 */
+
+	asm volatile("vpdpbusds %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpbusds %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpbusds %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpbusds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 52 */
+
+	asm volatile("vdpbf16ps %xmm1, %xmm2, %xmm3");
+	asm volatile("vdpbf16ps %ymm1, %ymm2, %ymm3");
+	asm volatile("vdpbf16ps %zmm1, %zmm2, %zmm3");
+	asm volatile("vdpbf16ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vpdpwssd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpwssd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpwssd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpwssd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vp4dpwssd (%eax), %zmm0, %zmm4");
+	asm volatile("vp4dpwssd 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 53 */
+
+	asm volatile("vpdpwssds %xmm1, %xmm2, %xmm3");
+	asm volatile("vpdpwssds %ymm1, %ymm2, %ymm3");
+	asm volatile("vpdpwssds %zmm1, %zmm2, %zmm3");
+	asm volatile("vpdpwssds 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vp4dpwssds (%eax), %zmm0, %zmm4");
+	asm volatile("vp4dpwssds 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 54 */
+
+	asm volatile("vpopcntb %xmm1, %xmm2");
+	asm volatile("vpopcntb %ymm1, %ymm2");
+	asm volatile("vpopcntb %zmm1, %zmm2");
+	asm volatile("vpopcntb 0x12345678(%eax,%ecx,8),%zmm2");
+
+	asm volatile("vpopcntw %xmm1, %xmm2");
+	asm volatile("vpopcntw %ymm1, %ymm2");
+	asm volatile("vpopcntw %zmm1, %zmm2");
+	asm volatile("vpopcntw 0x12345678(%eax,%ecx,8),%zmm2");
+
+	/* AVX-512: Op code 0f 38 55 */
+
+	asm volatile("vpopcntd %xmm1, %xmm2");
+	asm volatile("vpopcntd %ymm1, %ymm2");
+	asm volatile("vpopcntd %zmm1, %zmm2");
+	asm volatile("vpopcntd 0x12345678(%eax,%ecx,8),%zmm2");
+
+	asm volatile("vpopcntq %xmm1, %xmm2");
+	asm volatile("vpopcntq %ymm1, %ymm2");
+	asm volatile("vpopcntq %zmm1, %zmm2");
+	asm volatile("vpopcntq 0x12345678(%eax,%ecx,8),%zmm2");
+
 	/* AVX-512: Op code 0f 38 59 */
 
 	asm volatile("vpbroadcastq %xmm4,%xmm6");
@@ -1962,6 +2382,30 @@
 	asm volatile("vbroadcasti32x8 (%ecx),%zmm6");
 	asm volatile("vbroadcasti64x4 (%ecx),%zmm6");
 
+	/* AVX-512: Op code 0f 38 62 */
+
+	asm volatile("vpexpandb %xmm1, %xmm2");
+	asm volatile("vpexpandb %ymm1, %ymm2");
+	asm volatile("vpexpandb %zmm1, %zmm2");
+	asm volatile("vpexpandb 0x12345678(%eax,%ecx,8),%zmm2");
+
+	asm volatile("vpexpandw %xmm1, %xmm2");
+	asm volatile("vpexpandw %ymm1, %ymm2");
+	asm volatile("vpexpandw %zmm1, %zmm2");
+	asm volatile("vpexpandw 0x12345678(%eax,%ecx,8),%zmm2");
+
+	/* AVX-512: Op code 0f 38 63 */
+
+	asm volatile("vpcompressb %xmm1, %xmm2");
+	asm volatile("vpcompressb %ymm1, %ymm2");
+	asm volatile("vpcompressb %zmm1, %zmm2");
+	asm volatile("vpcompressb %zmm2,0x12345678(%eax,%ecx,8)");
+
+	asm volatile("vpcompressw %xmm1, %xmm2");
+	asm volatile("vpcompressw %ymm1, %ymm2");
+	asm volatile("vpcompressw %zmm1, %zmm2");
+	asm volatile("vpcompressw %zmm2,0x12345678(%eax,%ecx,8)");
+
 	/* AVX-512: Op code 0f 38 64 */
 
 	asm volatile("vpblendmd %zmm4,%zmm5,%zmm6");
@@ -1977,6 +2421,66 @@
 	asm volatile("vpblendmb %zmm4,%zmm5,%zmm6");
 	asm volatile("vpblendmw %zmm4,%zmm5,%zmm6");
 
+	/* AVX-512: Op code 0f 38 68 */
+
+	asm volatile("vp2intersectd %xmm1, %xmm2, %k3");
+	asm volatile("vp2intersectd %ymm1, %ymm2, %k3");
+	asm volatile("vp2intersectd %zmm1, %zmm2, %k3");
+	asm volatile("vp2intersectd 0x12345678(%eax,%ecx,8),%zmm2,%k3");
+
+	asm volatile("vp2intersectq %xmm1, %xmm2, %k3");
+	asm volatile("vp2intersectq %ymm1, %ymm2, %k3");
+	asm volatile("vp2intersectq %zmm1, %zmm2, %k3");
+	asm volatile("vp2intersectq 0x12345678(%eax,%ecx,8),%zmm2,%k3");
+
+	/* AVX-512: Op code 0f 38 70 */
+
+	asm volatile("vpshldvw %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshldvw %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshldvw %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshldvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 71 */
+
+	asm volatile("vpshldvd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshldvd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshldvd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshldvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vpshldvq %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshldvq %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshldvq %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshldvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 72 */
+
+	asm volatile("vcvtne2ps2bf16 %xmm1, %xmm2, %xmm3");
+	asm volatile("vcvtne2ps2bf16 %ymm1, %ymm2, %ymm3");
+	asm volatile("vcvtne2ps2bf16 %zmm1, %zmm2, %zmm3");
+	asm volatile("vcvtne2ps2bf16 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vcvtneps2bf16 %xmm1, %xmm2");
+	asm volatile("vcvtneps2bf16 %ymm1, %xmm2");
+	asm volatile("vcvtneps2bf16 %zmm1, %ymm2");
+	asm volatile("vcvtneps2bf16 0x12345678(%eax,%ecx,8),%ymm2");
+
+	asm volatile("vpshrdvw %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshrdvw %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshrdvw %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshrdvw 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 73 */
+
+	asm volatile("vpshrdvd %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshrdvd %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshrdvd %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshrdvd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vpshrdvq %xmm1, %xmm2, %xmm3");
+	asm volatile("vpshrdvq %ymm1, %ymm2, %ymm3");
+	asm volatile("vpshrdvq %zmm1, %zmm2, %zmm3");
+	asm volatile("vpshrdvq 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
 	/* AVX-512: Op code 0f 38 75 */
 
 	asm volatile("vpermi2b %zmm4,%zmm5,%zmm6");
@@ -2048,6 +2552,13 @@
 	asm volatile("vpermb %zmm4,%zmm5,%zmm6");
 	asm volatile("vpermw %zmm4,%zmm5,%zmm6");
 
+	/* AVX-512: Op code 0f 38 8f */
+
+	asm volatile("vpshufbitqmb %xmm1, %xmm2, %k3");
+	asm volatile("vpshufbitqmb %ymm1, %ymm2, %k3");
+	asm volatile("vpshufbitqmb %zmm1, %zmm2, %k3");
+	asm volatile("vpshufbitqmb 0x12345678(%eax,%ecx,8),%zmm2,%k3");
+
 	/* AVX-512: Op code 0f 38 90 */
 
 	asm volatile("vpgatherdd %xmm2,0x02(%ebp,%xmm7,2),%xmm1");
@@ -2062,6 +2573,32 @@
 	asm volatile("vpgatherqd 0x7b(%ebp,%zmm7,8),%ymm6{%k1}");
 	asm volatile("vpgatherqq 0x7b(%ebp,%zmm7,8),%zmm6{%k1}");
 
+	/* AVX-512: Op code 0f 38 9a */
+
+	asm volatile("vfmsub132ps %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132ps %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub132ps %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub132ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vfmsub132pd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132pd %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub132pd %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub132pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("v4fmaddps (%eax), %zmm0, %zmm4");
+	asm volatile("v4fmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 9b */
+
+	asm volatile("vfmsub132ss %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("vfmsub132sd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub132sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("v4fmaddss (%eax), %xmm0, %xmm4");
+	asm volatile("v4fmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4");
+
 	/* AVX-512: Op code 0f 38 a0 */
 
 	asm volatile("vpscatterdd %zmm6,0x7b(%ebp,%zmm7,8){%k1}");
@@ -2082,6 +2619,32 @@
 	asm volatile("vscatterqps %ymm6,0x7b(%ebp,%zmm7,8){%k1}");
 	asm volatile("vscatterqpd %zmm6,0x7b(%ebp,%zmm7,8){%k1}");
 
+	/* AVX-512: Op code 0f 38 aa */
+
+	asm volatile("vfmsub213ps %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213ps %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub213ps %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub213ps 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("vfmsub213pd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213pd %ymm1, %ymm2, %ymm3");
+	asm volatile("vfmsub213pd %zmm1, %zmm2, %zmm3");
+	asm volatile("vfmsub213pd 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	asm volatile("v4fnmaddps (%eax), %zmm0, %zmm4");
+	asm volatile("v4fnmaddps 0x12345678(%eax,%ecx,8),%zmm0,%zmm4");
+
+	/* AVX-512: Op code 0f 38 ab */
+
+	asm volatile("vfmsub213ss %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213ss 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("vfmsub213sd %xmm1, %xmm2, %xmm3");
+	asm volatile("vfmsub213sd 0x12345678(%eax,%ecx,8),%xmm2,%xmm3");
+
+	asm volatile("v4fnmaddss (%eax), %xmm0, %xmm4");
+	asm volatile("v4fnmaddss 0x12345678(%eax,%ecx,8),%xmm0,%xmm4");
+
 	/* AVX-512: Op code 0f 38 b4 */
 
 	asm volatile("vpmadd52luq %zmm4,%zmm5,%zmm6");
@@ -2120,6 +2683,44 @@
 	asm volatile("vrsqrt28ss %xmm5,%xmm6,%xmm7{%k7}");
 	asm volatile("vrsqrt28sd %xmm5,%xmm6,%xmm7{%k7}");
 
+	/* AVX-512: Op code 0f 38 cf */
+
+	asm volatile("gf2p8mulb %xmm1, %xmm3");
+	asm volatile("gf2p8mulb 0x12345678(%eax,%ecx,8),%xmm3");
+
+	asm volatile("vgf2p8mulb %xmm1, %xmm2, %xmm3");
+	asm volatile("vgf2p8mulb %ymm1, %ymm2, %ymm3");
+	asm volatile("vgf2p8mulb %zmm1, %zmm2, %zmm3");
+	asm volatile("vgf2p8mulb 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 dc */
+
+	asm volatile("vaesenc %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesenc %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesenc %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesenc 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 dd */
+
+	asm volatile("vaesenclast %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesenclast %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesenclast %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesenclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 de */
+
+	asm volatile("vaesdec %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesdec %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesdec %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesdec 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 38 df */
+
+	asm volatile("vaesdeclast %xmm1, %xmm2, %xmm3");
+	asm volatile("vaesdeclast %ymm1, %ymm2, %ymm3");
+	asm volatile("vaesdeclast %zmm1, %zmm2, %zmm3");
+	asm volatile("vaesdeclast 0x12345678(%eax,%ecx,8),%zmm2,%zmm3");
+
 	/* AVX-512: Op code 0f 3a 03 */
 
 	asm volatile("valignd $0x12,%zmm5,%zmm6,%zmm7");
@@ -2239,6 +2840,12 @@
 	asm volatile("vshufi32x4 $0x12,%zmm5,%zmm6,%zmm7");
 	asm volatile("vshufi64x2 $0x12,%zmm5,%zmm6,%zmm7");
 
+	/* AVX-512: Op code 0f 3a 44 */
+
+	asm volatile("vpclmulqdq $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpclmulqdq $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpclmulqdq $0x12,%zmm1,%zmm2,%zmm3");
+
 	/* AVX-512: Op code 0f 3a 50 */
 
 	asm volatile("vrangeps $0x12,%zmm5,%zmm6,%zmm7");
@@ -2279,6 +2886,54 @@
 	asm volatile("vfpclassss $0x12,%xmm7,%k5");
 	asm volatile("vfpclasssd $0x12,%xmm7,%k5");
 
+	/* AVX-512: Op code 0f 3a 70 */
+
+	asm volatile("vpshldw $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshldw $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshldw $0x12,%zmm1,%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 3a 71 */
+
+	asm volatile("vpshldd $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshldd $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshldd $0x12,%zmm1,%zmm2,%zmm3");
+
+	asm volatile("vpshldq $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshldq $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshldq $0x12,%zmm1,%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 3a 72 */
+
+	asm volatile("vpshrdw $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshrdw $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshrdw $0x12,%zmm1,%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 3a 73 */
+
+	asm volatile("vpshrdd $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshrdd $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshrdd $0x12,%zmm1,%zmm2,%zmm3");
+
+	asm volatile("vpshrdq $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vpshrdq $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vpshrdq $0x12,%zmm1,%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 3a ce */
+
+	asm volatile("gf2p8affineqb $0x12,%xmm1,%xmm3");
+
+	asm volatile("vgf2p8affineqb $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vgf2p8affineqb $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vgf2p8affineqb $0x12,%zmm1,%zmm2,%zmm3");
+
+	/* AVX-512: Op code 0f 3a cf */
+
+	asm volatile("gf2p8affineinvqb $0x12,%xmm1,%xmm3");
+
+	asm volatile("vgf2p8affineinvqb $0x12,%xmm1,%xmm2,%xmm3");
+	asm volatile("vgf2p8affineinvqb $0x12,%ymm1,%ymm2,%ymm3");
+	asm volatile("vgf2p8affineinvqb $0x12,%zmm1,%zmm2,%zmm3");
+
 	/* AVX-512: Op code 0f 72 (Grp13) */
 
 	asm volatile("vprord $0x12,%zmm5,%zmm6");
diff --git a/tools/perf/arch/x86/util/event.c b/tools/perf/arch/x86/util/event.c
index d1044df..ac45015 100644
--- a/tools/perf/arch/x86/util/event.c
+++ b/tools/perf/arch/x86/util/event.c
@@ -18,8 +18,7 @@
 {
 	int rc = 0;
 	struct map *pos;
-	struct map_groups *kmaps = &machine->kmaps;
-	struct maps *maps = &kmaps->maps;
+	struct maps *kmaps = &machine->kmaps;
 	union perf_event *event = zalloc(sizeof(event->mmap) +
 					 machine->id_hdr_size);
 
@@ -29,7 +28,7 @@
 		return -1;
 	}
 
-	maps__for_each_entry(maps, pos) {
+	maps__for_each_entry(kmaps, pos) {
 		struct kmap *kmap;
 		size_t size;
 
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index 376dbf1..f8b6ae5 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -547,8 +547,8 @@
 	if (!pairs_left && !pairs_right)
 		return 0;
 
-	l = labs(left->diff.cycles);
-	r = labs(right->diff.cycles);
+	l = llabs(left->diff.cycles);
+	r = llabs(right->diff.cycles);
 	return r - l;
 }
 
@@ -646,7 +646,7 @@
 			if (i >= he->block_info->num || i >= NUM_SPARKS)
 				break;
 
-			val = labs(pair->block_info->cycles_spark[i] -
+			val = llabs(pair->block_info->cycles_spark[i] -
 				     he->block_info->cycles_spark[i]);
 
 			update_spark_value(pair->diff.svals, NUM_SPARKS,
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index ab0f6e5..830d563 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -780,11 +780,6 @@
 	return printed;
 }
 
-static int map_groups__fprintf_task(struct map_groups *mg, int indent, FILE *fp)
-{
-	return maps__fprintf_task(&mg->maps, indent, fp);
-}
-
 static void task__print_level(struct task *task, FILE *fp, int level)
 {
 	struct thread *thread = task->thread;
@@ -795,7 +790,7 @@
 
 	fprintf(fp, "%s\n", thread__comm_str(thread));
 
-	map_groups__fprintf_task(thread->mg, comm_indent, fp);
+	maps__fprintf_task(thread->maps, comm_indent, fp);
 
 	if (!list_empty(&task->children)) {
 		list_for_each_entry(child, &task->children, list)
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index f86c5cc..e2406b2 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -448,7 +448,7 @@
 		       "selected. Hence, no address to lookup the source line number.\n");
 		return -EINVAL;
 	}
-	if (PRINT_FIELD(BRSTACKINSN) &&
+	if (PRINT_FIELD(BRSTACKINSN) && !allow_user_set &&
 	    !(perf_evlist__combined_branch_type(session->evlist) &
 	      PERF_SAMPLE_BRANCH_ANY)) {
 		pr_err("Display of branch stack assembler requested, but non all-branch filter set\n"
@@ -932,6 +932,48 @@
 	return len;
 }
 
+static int map__fprintf_srccode(struct map *map, u64 addr, FILE *fp, struct srccode_state *state)
+{
+	char *srcfile;
+	int ret = 0;
+	unsigned line;
+	int len;
+	char *srccode;
+
+	if (!map || !map->dso)
+		return 0;
+	srcfile = get_srcline_split(map->dso,
+				    map__rip_2objdump(map, addr),
+				    &line);
+	if (!srcfile)
+		return 0;
+
+	/* Avoid redundant printing */
+	if (state &&
+	    state->srcfile &&
+	    !strcmp(state->srcfile, srcfile) &&
+	    state->line == line) {
+		free(srcfile);
+		return 0;
+	}
+
+	srccode = find_sourceline(srcfile, line, &len);
+	if (!srccode)
+		goto out_free_line;
+
+	ret = fprintf(fp, "|%-8d %.*s", line, len, srccode);
+
+	if (state) {
+		state->srcfile = srcfile;
+		state->line = line;
+	}
+	return ret;
+
+out_free_line:
+	free(srcfile);
+	return ret;
+}
+
 static int print_srccode(struct thread *thread, u8 cpumode, uint64_t addr)
 {
 	struct addr_location al;
@@ -1084,7 +1126,7 @@
 				insn++;
 			}
 		}
-		if (off != (unsigned)len)
+		if (off != end - start)
 			printed += fprintf(fp, "\tmismatch of LBR data and executable\n");
 	}
 
diff --git a/tools/perf/tests/Build b/tools/perf/tests/Build
index e72acce..a3c595f 100644
--- a/tools/perf/tests/Build
+++ b/tools/perf/tests/Build
@@ -27,7 +27,7 @@
 perf-y += task-exit.o
 perf-y += sw-clock.o
 perf-y += mmap-thread-lookup.o
-perf-y += thread-mg-share.o
+perf-y += thread-maps-share.o
 perf-y += switch-tracking.o
 perf-y += keep-tracking.o
 perf-y += code-reading.o
@@ -52,7 +52,7 @@
 perf-y += clang.o
 perf-y += unit_number__scnprintf.o
 perf-y += mem2node.o
-perf-y += map_groups.o
+perf-y += maps.o
 perf-y += time-utils-test.o
 
 $(OUTPUT)tests/llvm-src-base.c: tests/bpf-script-example.c tests/Build
diff --git a/tools/perf/tests/builtin-test.c b/tools/perf/tests/builtin-test.c
index 8b286e9..7115aa3 100644
--- a/tools/perf/tests/builtin-test.c
+++ b/tools/perf/tests/builtin-test.c
@@ -166,8 +166,8 @@
 		.func = test__mmap_thread_lookup,
 	},
 	{
-		.desc = "Share thread mg",
-		.func = test__thread_mg_share,
+		.desc = "Share thread maps",
+		.func = test__thread_maps_share,
 	},
 	{
 		.desc = "Sort output of hist entries",
@@ -297,8 +297,8 @@
 		.func = test__time_utils,
 	},
 	{
-		.desc = "map_groups__merge_in",
-		.func = test__map_groups__merge_in,
+		.desc = "maps__merge_in",
+		.func = test__maps__merge_in,
 	},
 	{
 		.func = NULL,
diff --git a/tools/perf/tests/code-reading.c b/tools/perf/tests/code-reading.c
index 1f017e1..6fe221d 100644
--- a/tools/perf/tests/code-reading.c
+++ b/tools/perf/tests/code-reading.c
@@ -276,7 +276,7 @@
 		len = al.map->end - addr;
 
 	/* Read the object code using perf */
-	ret_len = dso__data_read_offset(al.map->dso, thread->mg->machine,
+	ret_len = dso__data_read_offset(al.map->dso, thread->maps->machine,
 					al.addr, buf1, len);
 	if (ret_len != len) {
 		pr_debug("dso__data_read_offset failed\n");
diff --git a/tools/perf/tests/map_groups.c b/tools/perf/tests/maps.c
similarity index 82%
rename from tools/perf/tests/map_groups.c
rename to tools/perf/tests/maps.c
index 6b9f1cd..edcbc70 100644
--- a/tools/perf/tests/map_groups.c
+++ b/tools/perf/tests/maps.c
@@ -3,7 +3,7 @@
 #include <linux/kernel.h>
 #include "tests.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "dso.h"
 #include "debug.h"
 
@@ -13,12 +13,12 @@
 	u64 end;
 };
 
-static int check_maps(struct map_def *merged, unsigned int size, struct map_groups *mg)
+static int check_maps(struct map_def *merged, unsigned int size, struct maps *maps)
 {
 	struct map *map;
 	unsigned int i = 0;
 
-	map_groups__for_each_entry(mg, map) {
+	maps__for_each_entry(maps, map) {
 		if (i > 0)
 			TEST_ASSERT_VAL("less maps expected", (map && i < size) || (!map && i == size));
 
@@ -33,9 +33,9 @@
 	return TEST_OK;
 }
 
-int test__map_groups__merge_in(struct test *t __maybe_unused, int subtest __maybe_unused)
+int test__maps__merge_in(struct test *t __maybe_unused, int subtest __maybe_unused)
 {
-	struct map_groups mg;
+	struct maps maps;
 	unsigned int i;
 	struct map_def bpf_progs[] = {
 		{ "bpf_prog_1", 200, 300 },
@@ -64,7 +64,7 @@
 	struct map *map_kcore1, *map_kcore2, *map_kcore3;
 	int ret;
 
-	map_groups__init(&mg, NULL);
+	maps__init(&maps, NULL);
 
 	for (i = 0; i < ARRAY_SIZE(bpf_progs); i++) {
 		struct map *map;
@@ -74,7 +74,7 @@
 
 		map->start = bpf_progs[i].start;
 		map->end   = bpf_progs[i].end;
-		map_groups__insert(&mg, map);
+		maps__insert(&maps, map);
 		map__put(map);
 	}
 
@@ -99,22 +99,22 @@
 	map_kcore3->start = 880;
 	map_kcore3->end   = 1100;
 
-	ret = map_groups__merge_in(&mg, map_kcore1);
+	ret = maps__merge_in(&maps, map_kcore1);
 	TEST_ASSERT_VAL("failed to merge map", !ret);
 
-	ret = check_maps(merged12, ARRAY_SIZE(merged12), &mg);
+	ret = check_maps(merged12, ARRAY_SIZE(merged12), &maps);
 	TEST_ASSERT_VAL("merge check failed", !ret);
 
-	ret = map_groups__merge_in(&mg, map_kcore2);
+	ret = maps__merge_in(&maps, map_kcore2);
 	TEST_ASSERT_VAL("failed to merge map", !ret);
 
-	ret = check_maps(merged12, ARRAY_SIZE(merged12), &mg);
+	ret = check_maps(merged12, ARRAY_SIZE(merged12), &maps);
 	TEST_ASSERT_VAL("merge check failed", !ret);
 
-	ret = map_groups__merge_in(&mg, map_kcore3);
+	ret = maps__merge_in(&maps, map_kcore3);
 	TEST_ASSERT_VAL("failed to merge map", !ret);
 
-	ret = check_maps(merged3, ARRAY_SIZE(merged3), &mg);
+	ret = check_maps(merged3, ARRAY_SIZE(merged3), &maps);
 	TEST_ASSERT_VAL("merge check failed", !ret);
 	return TEST_OK;
 }
diff --git a/tools/perf/tests/tests.h b/tools/perf/tests/tests.h
index 9837b6e..25aea38 100644
--- a/tools/perf/tests/tests.h
+++ b/tools/perf/tests/tests.h
@@ -73,7 +73,7 @@
 int test__expr(struct test *test, int subtest);
 int test__hists_filter(struct test *test, int subtest);
 int test__mmap_thread_lookup(struct test *test, int subtest);
-int test__thread_mg_share(struct test *test, int subtest);
+int test__thread_maps_share(struct test *test, int subtest);
 int test__hists_output(struct test *test, int subtest);
 int test__hists_cumulate(struct test *test, int subtest);
 int test__switch_tracking(struct test *test, int subtest);
@@ -107,7 +107,7 @@
 int test__clang_subtest_get_nr(void);
 int test__unit_number__scnprint(struct test *test, int subtest);
 int test__mem2node(struct test *t, int subtest);
-int test__map_groups__merge_in(struct test *t, int subtest);
+int test__maps__merge_in(struct test *t, int subtest);
 int test__time_utils(struct test *t, int subtest);
 
 bool test__bp_signal_is_supported(void);
diff --git a/tools/perf/tests/thread-mg-share.c b/tools/perf/tests/thread-maps-share.c
similarity index 65%
rename from tools/perf/tests/thread-mg-share.c
rename to tools/perf/tests/thread-maps-share.c
index cbac717..9371484 100644
--- a/tools/perf/tests/thread-mg-share.c
+++ b/tools/perf/tests/thread-maps-share.c
@@ -4,7 +4,7 @@
 #include "thread.h"
 #include "debug.h"
 
-int test__thread_mg_share(struct test *test __maybe_unused, int subtest __maybe_unused)
+int test__thread_maps_share(struct test *test __maybe_unused, int subtest __maybe_unused)
 {
 	struct machines machines;
 	struct machine *machine;
@@ -12,16 +12,16 @@
 	/* thread group */
 	struct thread *leader;
 	struct thread *t1, *t2, *t3;
-	struct map_groups *mg;
+	struct maps *maps;
 
 	/* other process */
 	struct thread *other, *other_leader;
-	struct map_groups *other_mg;
+	struct maps *other_maps;
 
 	/*
 	 * This test create 2 processes abstractions (struct thread)
 	 * with several threads and checks they properly share and
-	 * maintain map groups info (struct map_groups).
+	 * maintain maps info (struct maps).
 	 *
 	 * thread group (pid: 0, tids: 0, 1, 2, 3)
 	 * other  group (pid: 4, tids: 4, 5)
@@ -42,17 +42,17 @@
 	TEST_ASSERT_VAL("failed to create threads",
 			leader && t1 && t2 && t3 && other);
 
-	mg = leader->mg;
-	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&mg->refcnt), 4);
+	maps = leader->maps;
+	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&maps->refcnt), 4);
 
-	/* test the map groups pointer is shared */
-	TEST_ASSERT_VAL("map groups don't match", mg == t1->mg);
-	TEST_ASSERT_VAL("map groups don't match", mg == t2->mg);
-	TEST_ASSERT_VAL("map groups don't match", mg == t3->mg);
+	/* test the maps pointer is shared */
+	TEST_ASSERT_VAL("maps don't match", maps == t1->maps);
+	TEST_ASSERT_VAL("maps don't match", maps == t2->maps);
+	TEST_ASSERT_VAL("maps don't match", maps == t3->maps);
 
 	/*
 	 * Verify the other leader was created by previous call.
-	 * It should have shared map groups with no change in
+	 * It should have shared maps with no change in
 	 * refcnt.
 	 */
 	other_leader = machine__find_thread(machine, 4, 4);
@@ -70,26 +70,26 @@
 	machine__remove_thread(machine, other);
 	machine__remove_thread(machine, other_leader);
 
-	other_mg = other->mg;
-	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&other_mg->refcnt), 2);
+	other_maps = other->maps;
+	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&other_maps->refcnt), 2);
 
-	TEST_ASSERT_VAL("map groups don't match", other_mg == other_leader->mg);
+	TEST_ASSERT_VAL("maps don't match", other_maps == other_leader->maps);
 
 	/* release thread group */
 	thread__put(leader);
-	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&mg->refcnt), 3);
+	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&maps->refcnt), 3);
 
 	thread__put(t1);
-	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&mg->refcnt), 2);
+	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&maps->refcnt), 2);
 
 	thread__put(t2);
-	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&mg->refcnt), 1);
+	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&maps->refcnt), 1);
 
 	thread__put(t3);
 
 	/* release other group  */
 	thread__put(other_leader);
-	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&other_mg->refcnt), 1);
+	TEST_ASSERT_EQUAL("wrong refcnt", refcount_read(&other_maps->refcnt), 1);
 
 	thread__put(other);
 
diff --git a/tools/perf/tests/vmlinux-kallsyms.c b/tools/perf/tests/vmlinux-kallsyms.c
index ff64907..193b7c9 100644
--- a/tools/perf/tests/vmlinux-kallsyms.c
+++ b/tools/perf/tests/vmlinux-kallsyms.c
@@ -190,10 +190,9 @@
 		 * so use the short name, less descriptive but the same ("[kernel]" in
 		 * both cases.
 		 */
-		pair = map_groups__find_by_name(&kallsyms.kmaps,
-						(map->dso->kernel ?
-							map->dso->short_name :
-							map->dso->name));
+		pair = maps__find_by_name(&kallsyms.kmaps, (map->dso->kernel ?
+								map->dso->short_name :
+								map->dso->name));
 		if (pair) {
 			pair->priv = 1;
 		} else {
@@ -213,7 +212,7 @@
 		mem_start = vmlinux_map->unmap_ip(vmlinux_map, map->start);
 		mem_end = vmlinux_map->unmap_ip(vmlinux_map, map->end);
 
-		pair = map_groups__find(&kallsyms.kmaps, mem_start);
+		pair = maps__find(&kallsyms.kmaps, mem_start);
 		if (pair == NULL || pair->priv)
 			continue;
 
diff --git a/tools/perf/ui/browsers/annotate.c b/tools/perf/ui/browsers/annotate.c
index 992705c..badbddb 100644
--- a/tools/perf/ui/browsers/annotate.c
+++ b/tools/perf/ui/browsers/annotate.c
@@ -430,7 +430,7 @@
 		return true;
 	}
 
-	target_ms.mg  = ms->mg;
+	target_ms.maps = ms->maps;
 	target_ms.map = ms->map;
 	target_ms.sym = dl->ops.target.sym;
 	pthread_mutex_unlock(&notes->lock);
diff --git a/tools/perf/ui/stdio/hist.c b/tools/perf/ui/stdio/hist.c
index 132056c..2ab2af4 100644
--- a/tools/perf/ui/stdio/hist.c
+++ b/tools/perf/ui/stdio/hist.c
@@ -8,7 +8,7 @@
 #include "../../util/event.h"
 #include "../../util/hist.h"
 #include "../../util/map.h"
-#include "../../util/map_groups.h"
+#include "../../util/maps.h"
 #include "../../util/symbol.h"
 #include "../../util/sort.h"
 #include "../../util/evsel.h"
@@ -885,7 +885,7 @@
 		}
 
 		if (h->ms.map == NULL && verbose > 1) {
-			map_groups__fprintf(h->thread->mg, fp);
+			maps__fprintf(h->thread->maps, fp);
 			fprintf(fp, "%.10s end\n", graph_dotted_line);
 		}
 	}
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index b8e05a1..07da6c7 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -49,6 +49,7 @@
 perf-y += callchain.o
 perf-y += values.o
 perf-y += debug.o
+perf-y += fncache.o
 perf-y += machine.o
 perf-y += map.o
 perf-y += pstack.o
@@ -76,6 +77,7 @@
 perf-y += hist.o
 perf-y += util.o
 perf-y += cpumap.o
+perf-y += affinity.o
 perf-y += cputopo.o
 perf-y += cgroup.o
 perf-y += target.o
diff --git a/tools/perf/util/affinity.c b/tools/perf/util/affinity.c
new file mode 100644
index 0000000..a5e31f8
--- /dev/null
+++ b/tools/perf/util/affinity.c
@@ -0,0 +1,73 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Manage affinity to optimize IPIs inside the kernel perf API. */
+#define _GNU_SOURCE 1
+#include <sched.h>
+#include <stdlib.h>
+#include <linux/bitmap.h>
+#include <linux/zalloc.h>
+#include "perf.h"
+#include "cpumap.h"
+#include "affinity.h"
+
+static int get_cpu_set_size(void)
+{
+	int sz = cpu__max_cpu() + 8 - 1;
+	/*
+	 * sched_getaffinity doesn't like masks smaller than the kernel.
+	 * Hopefully that's big enough.
+	 */
+	if (sz < 4096)
+		sz = 4096;
+	return sz / 8;
+}
+
+int affinity__setup(struct affinity *a)
+{
+	int cpu_set_size = get_cpu_set_size();
+
+	a->orig_cpus = bitmap_alloc(cpu_set_size * 8);
+	if (!a->orig_cpus)
+		return -1;
+	sched_getaffinity(0, cpu_set_size, (cpu_set_t *)a->orig_cpus);
+	a->sched_cpus = bitmap_alloc(cpu_set_size * 8);
+	if (!a->sched_cpus) {
+		zfree(&a->orig_cpus);
+		return -1;
+	}
+	bitmap_zero((unsigned long *)a->sched_cpus, cpu_set_size);
+	a->changed = false;
+	return 0;
+}
+
+/*
+ * perf_event_open does an IPI internally to the target CPU.
+ * It is more efficient to change perf's affinity to the target
+ * CPU and then set up all events on that CPU, so we amortize
+ * CPU communication.
+ */
+void affinity__set(struct affinity *a, int cpu)
+{
+	int cpu_set_size = get_cpu_set_size();
+
+	if (cpu == -1)
+		return;
+	a->changed = true;
+	set_bit(cpu, a->sched_cpus);
+	/*
+	 * We ignore errors because affinity is just an optimization.
+	 * This could happen for example with isolated CPUs or cpusets.
+	 * In this case the IPIs inside the kernel's perf API still work.
+	 */
+	sched_setaffinity(0, cpu_set_size, (cpu_set_t *)a->sched_cpus);
+	clear_bit(cpu, a->sched_cpus);
+}
+
+void affinity__cleanup(struct affinity *a)
+{
+	int cpu_set_size = get_cpu_set_size();
+
+	if (a->changed)
+		sched_setaffinity(0, cpu_set_size, (cpu_set_t *)a->orig_cpus);
+	zfree(&a->sched_cpus);
+	zfree(&a->orig_cpus);
+}
diff --git a/tools/perf/util/affinity.h b/tools/perf/util/affinity.h
new file mode 100644
index 0000000..0ad6a18
--- /dev/null
+++ b/tools/perf/util/affinity.h
@@ -0,0 +1,17 @@
+// SPDX-License-Identifier: GPL-2.0
+#ifndef PERF_AFFINITY_H
+#define PERF_AFFINITY_H 1
+
+#include <stdbool.h>
+
+struct affinity {
+	unsigned long *orig_cpus;
+	unsigned long *sched_cpus;
+	bool changed;
+};
+
+void affinity__cleanup(struct affinity *a);
+void affinity__set(struct affinity *a, int cpu);
+int affinity__setup(struct affinity *a);
+
+#endif // PERF_AFFINITY_H
diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c
index 5ea9a45..f5e77ed 100644
--- a/tools/perf/util/annotate.c
+++ b/tools/perf/util/annotate.c
@@ -23,7 +23,7 @@
 #include "dso.h"
 #include "env.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "symbol.h"
 #include "srcline.h"
 #include "units.h"
@@ -271,7 +271,7 @@
 find_target:
 	target.addr = map__objdump_2mem(map, ops->target.addr);
 
-	if (map_groups__find_ams(ms->mg, &target) == 0 &&
+	if (maps__find_ams(ms->maps, &target) == 0 &&
 	    map__rip_2objdump(target.ms.map, map->map_ip(target.ms.map, target.addr)) == ops->target.addr)
 		ops->target.sym = target.ms.sym;
 
@@ -391,7 +391,7 @@
 	 * Actual navigation will come next, with further understanding of how
 	 * the symbol searching and disassembly should be done.
 	 */
-	if (map_groups__find_ams(ms->mg, &target) == 0 &&
+	if (maps__find_ams(ms->maps, &target) == 0 &&
 	    map__rip_2objdump(target.ms.map, map->map_ip(target.ms.map, target.addr)) == ops->target.addr)
 		ops->target.sym = target.ms.sym;
 
@@ -1545,7 +1545,7 @@
 			.ms = { .map = map, },
 		};
 
-		if (!map_groups__find_ams(args->ms.mg, &target) &&
+		if (!maps__find_ams(args->ms.maps, &target) &&
 		    target.ms.sym->start == target.al_addr)
 			dl->ops.target.sym = target.ms.sym;
 	}
diff --git a/tools/perf/util/bpf-event.c b/tools/perf/util/bpf-event.c
index f7ed5d1..a3207d9 100644
--- a/tools/perf/util/bpf-event.c
+++ b/tools/perf/util/bpf-event.c
@@ -52,9 +52,7 @@
 	for (i = 0; i < info_linear->info.nr_jited_ksyms; i++) {
 		u64 *addrs = (u64 *)(uintptr_t)(info_linear->info.jited_ksyms);
 		u64 addr = addrs[i];
-		struct map *map;
-
-		map = map_groups__find(&machine->kmaps, addr);
+		struct map *map = maps__find(&machine->kmaps, addr);
 
 		if (map) {
 			map->dso->binary_type = DSO_BINARY_TYPE__BPF_PROG_INFO;
diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c
index 5cefce3..818aa4e 100644
--- a/tools/perf/util/callchain.c
+++ b/tools/perf/util/callchain.c
@@ -1106,7 +1106,7 @@
 int fill_callchain_info(struct addr_location *al, struct callchain_cursor_node *node,
 			bool hide_unresolved)
 {
-	al->mg	= node->ms.mg;
+	al->maps = node->ms.maps;
 	al->map = node->ms.map;
 	al->sym = node->ms.sym;
 	al->srcline = node->srcline;
@@ -1119,8 +1119,8 @@
 			goto out;
 	}
 
-	if (al->mg == &al->mg->machine->kmaps) {
-		if (machine__is_host(al->mg->machine)) {
+	if (al->maps == &al->maps->machine->kmaps) {
+		if (machine__is_host(al->maps->machine)) {
 			al->cpumode = PERF_RECORD_MISC_KERNEL;
 			al->level = 'k';
 		} else {
@@ -1128,7 +1128,7 @@
 			al->level = 'g';
 		}
 	} else {
-		if (machine__is_host(al->mg->machine)) {
+		if (machine__is_host(al->maps->machine)) {
 			al->cpumode = PERF_RECORD_MISC_USER;
 			al->level = '.';
 		} else if (perf_guest) {
diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c
index f5f855f..5471045 100644
--- a/tools/perf/util/cs-etm.c
+++ b/tools/perf/util/cs-etm.c
@@ -2569,7 +2569,7 @@
 	if (err)
 		goto err_delete_thread;
 
-	if (thread__init_map_groups(etm->unknown_thread, etm->machine)) {
+	if (thread__init_maps(etm->unknown_thread, etm->machine)) {
 		err = -ENOMEM;
 		goto err_delete_thread;
 	}
diff --git a/tools/perf/util/db-export.c b/tools/perf/util/db-export.c
index d029faf..db74471 100644
--- a/tools/perf/util/db-export.c
+++ b/tools/perf/util/db-export.c
@@ -181,7 +181,7 @@
 	if (al->map) {
 		struct dso *dso = al->map->dso;
 
-		err = db_export__dso(dbe, dso, al->mg->machine);
+		err = db_export__dso(dbe, dso, al->maps->machine);
 		if (err)
 			return err;
 		*dso_db_id = dso->db_id;
@@ -251,7 +251,7 @@
 		 */
 		al.sym = node->ms.sym;
 		al.map = node->ms.map;
-		al.mg  = thread->mg;
+		al.maps = thread->maps;
 		al.addr = node->ip;
 
 		if (al.map && !al.sym)
@@ -360,13 +360,13 @@
 	if (err)
 		return err;
 
-	err = db_export__machine(dbe, al->mg->machine);
+	err = db_export__machine(dbe, al->maps->machine);
 	if (err)
 		return err;
 
-	main_thread = thread__main_thread(al->mg->machine, thread);
+	main_thread = thread__main_thread(al->maps->machine, thread);
 
-	err = db_export__threads(dbe, thread, main_thread, al->mg->machine, &comm);
+	err = db_export__threads(dbe, thread, main_thread, al->maps->machine, &comm);
 	if (err)
 		goto out_put;
 
@@ -380,7 +380,7 @@
 		goto out_put;
 
 	if (dbe->cpr) {
-		struct call_path *cp = call_path_from_sample(dbe, al->mg->machine,
+		struct call_path *cp = call_path_from_sample(dbe, al->maps->machine,
 							     thread, sample,
 							     evsel);
 		if (cp) {
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 0141b26..c5447ff 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -457,11 +457,11 @@
 struct map *thread__find_map(struct thread *thread, u8 cpumode, u64 addr,
 			     struct addr_location *al)
 {
-	struct map_groups *mg = thread->mg;
-	struct machine *machine = mg->machine;
+	struct maps *maps = thread->maps;
+	struct machine *machine = maps->machine;
 	bool load_map = false;
 
-	al->mg = mg;
+	al->maps = maps;
 	al->thread = thread;
 	al->addr = addr;
 	al->cpumode = cpumode;
@@ -474,13 +474,13 @@
 
 	if (cpumode == PERF_RECORD_MISC_KERNEL && perf_host) {
 		al->level = 'k';
-		al->mg = mg = &machine->kmaps;
+		al->maps = maps = &machine->kmaps;
 		load_map = true;
 	} else if (cpumode == PERF_RECORD_MISC_USER && perf_host) {
 		al->level = '.';
 	} else if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL && perf_guest) {
 		al->level = 'g';
-		al->mg = mg = &machine->kmaps;
+		al->maps = maps = &machine->kmaps;
 		load_map = true;
 	} else if (cpumode == PERF_RECORD_MISC_GUEST_USER && perf_guest) {
 		al->level = 'u';
@@ -500,7 +500,7 @@
 		return NULL;
 	}
 
-	al->map = map_groups__find(mg, al->addr);
+	al->map = maps__find(maps, al->addr);
 	if (al->map != NULL) {
 		/*
 		 * Kernel maps might be changed when loading symbols so loading
@@ -523,7 +523,7 @@
 				struct addr_location *al)
 {
 	struct map *map = thread__find_map(thread, cpumode, addr, al);
-	struct machine *machine = thread->mg->machine;
+	struct machine *machine = thread->maps->machine;
 	u8 addr_cpumode = machine__addr_cpumode(machine, cpumode, addr);
 
 	if (map || addr_cpumode == cpumode)
diff --git a/tools/perf/util/fncache.c b/tools/perf/util/fncache.c
new file mode 100644
index 0000000..6225cbc
--- /dev/null
+++ b/tools/perf/util/fncache.c
@@ -0,0 +1,63 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Manage a cache of file names' existence */
+#include <stdlib.h>
+#include <unistd.h>
+#include <string.h>
+#include <linux/list.h>
+#include "fncache.h"
+
+struct fncache {
+	struct hlist_node nd;
+	bool res;
+	char name[];
+};
+
+#define FNHSIZE 61
+
+static struct hlist_head fncache_hash[FNHSIZE];
+
+unsigned shash(const unsigned char *s)
+{
+	unsigned h = 0;
+	while (*s)
+		h = 65599 * h + *s++;
+	return h ^ (h >> 16);
+}
+
+static bool lookup_fncache(const char *name, bool *res)
+{
+	int h = shash((const unsigned char *)name) % FNHSIZE;
+	struct fncache *n;
+
+	hlist_for_each_entry(n, &fncache_hash[h], nd) {
+		if (!strcmp(n->name, name)) {
+			*res = n->res;
+			return true;
+		}
+	}
+	return false;
+}
+
+static void update_fncache(const char *name, bool res)
+{
+	struct fncache *n = malloc(sizeof(struct fncache) + strlen(name) + 1);
+	int h = shash((const unsigned char *)name) % FNHSIZE;
+
+	if (!n)
+		return;
+	strcpy(n->name, name);
+	n->res = res;
+	hlist_add_head(&n->nd, &fncache_hash[h]);
+}
+
+/* No LRU, only use when bounded in some other way. */
+bool file_available(const char *name)
+{
+	bool res;
+
+	if (lookup_fncache(name, &res))
+		return res;
+	res = access(name, R_OK) == 0;
+	update_fncache(name, res);
+	return res;
+}
diff --git a/tools/perf/util/fncache.h b/tools/perf/util/fncache.h
new file mode 100644
index 0000000..fe020be
--- /dev/null
+++ b/tools/perf/util/fncache.h
@@ -0,0 +1,7 @@
+#ifndef _FCACHE_H
+#define _FCACHE_H 1
+
+unsigned shash(const unsigned char *s);
+bool file_available(const char *name);
+
+#endif
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index 0a8d72a..ca5a8f4 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -692,7 +692,7 @@
 			.ino = ns ? ns->link_info[CGROUP_NS_INDEX].ino : 0,
 		},
 		.ms = {
-			.mg	= al->mg,
+			.maps	= al->maps,
 			.map	= al->map,
 			.sym	= al->sym,
 		},
@@ -760,7 +760,7 @@
 		.block_info = block_info,
 		.hists = hists,
 		.ms = {
-			.mg  = al->mg,
+			.maps = al->maps,
 			.map = al->map,
 			.sym = al->sym,
 		},
@@ -895,7 +895,7 @@
 	if (iter->curr >= iter->total)
 		return 0;
 
-	al->mg  = bi[i].to.ms.mg;
+	al->maps = bi[i].to.ms.maps;
 	al->map = bi[i].to.ms.map;
 	al->sym = bi[i].to.ms.sym;
 	al->addr = bi[i].to.addr;
@@ -1072,7 +1072,7 @@
 		.comm = thread__comm(al->thread),
 		.ip = al->addr,
 		.ms = {
-			.mg  = al->mg,
+			.maps = al->maps,
 			.map = al->map,
 			.sym = al->sym,
 		},
diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c
index 409afc6..33cf892 100644
--- a/tools/perf/util/intel-pt.c
+++ b/tools/perf/util/intel-pt.c
@@ -3296,7 +3296,7 @@
 	err = thread__set_comm(pt->unknown_thread, "unknown", 0);
 	if (err)
 		goto err_delete_thread;
-	if (thread__init_map_groups(pt->unknown_thread, pt->machine)) {
+	if (thread__init_maps(pt->unknown_thread, pt->machine)) {
 		err = -ENOMEM;
 		goto err_delete_thread;
 	}
diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
index e2a312c..416d174 100644
--- a/tools/perf/util/machine.c
+++ b/tools/perf/util/machine.c
@@ -86,7 +86,7 @@
 	int err = -ENOMEM;
 
 	memset(machine, 0, sizeof(*machine));
-	map_groups__init(&machine->kmaps, machine);
+	maps__init(&machine->kmaps, machine);
 	RB_CLEAR_NODE(&machine->rb_node);
 	dsos__init(&machine->dsos);
 
@@ -217,7 +217,7 @@
 		return;
 
 	machine__destroy_kernel_maps(machine);
-	map_groups__exit(&machine->kmaps);
+	maps__exit(&machine->kmaps);
 	dsos__exit(&machine->dsos);
 	machine__exit_vdso(machine);
 	zfree(&machine->root_dir);
@@ -412,28 +412,28 @@
 	if (!leader)
 		goto out_err;
 
-	if (!leader->mg)
-		leader->mg = map_groups__new(machine);
+	if (!leader->maps)
+		leader->maps = maps__new(machine);
 
-	if (!leader->mg)
+	if (!leader->maps)
 		goto out_err;
 
-	if (th->mg == leader->mg)
+	if (th->maps == leader->maps)
 		return;
 
-	if (th->mg) {
+	if (th->maps) {
 		/*
 		 * Maps are created from MMAP events which provide the pid and
 		 * tid.  Consequently there never should be any maps on a thread
 		 * with an unknown pid.  Just print an error if there are.
 		 */
-		if (!map_groups__empty(th->mg))
+		if (!maps__empty(th->maps))
 			pr_err("Discarding thread maps for %d:%d\n",
 			       th->pid_, th->tid);
-		map_groups__put(th->mg);
+		maps__put(th->maps);
 	}
 
-	th->mg = map_groups__get(leader->mg);
+	th->maps = maps__get(leader->maps);
 out_put:
 	thread__put(leader);
 	return;
@@ -536,14 +536,13 @@
 		rb_insert_color_cached(&th->rb_node, &threads->entries, leftmost);
 
 		/*
-		 * We have to initialize map_groups separately
-		 * after rb tree is updated.
+		 * We have to initialize maps separately after rb tree is updated.
 		 *
 		 * The reason is that we call machine__findnew_thread
-		 * within thread__init_map_groups to find the thread
+		 * within thread__init_maps to find the thread
 		 * leader and that would screwed the rb tree.
 		 */
-		if (thread__init_map_groups(th, machine)) {
+		if (thread__init_maps(th, machine)) {
 			rb_erase_cached(&th->rb_node, &threads->entries);
 			RB_CLEAR_NODE(&th->rb_node);
 			thread__put(th);
@@ -724,9 +723,8 @@
 					     struct perf_sample *sample __maybe_unused)
 {
 	struct symbol *sym;
-	struct map *map;
+	struct map *map = maps__find(&machine->kmaps, event->ksymbol.addr);
 
-	map = map_groups__find(&machine->kmaps, event->ksymbol.addr);
 	if (!map) {
 		map = dso__new_map(event->ksymbol.name);
 		if (!map)
@@ -734,7 +732,7 @@
 
 		map->start = event->ksymbol.addr;
 		map->end = map->start + event->ksymbol.len;
-		map_groups__insert(&machine->kmaps, map);
+		maps__insert(&machine->kmaps, map);
 	}
 
 	sym = symbol__new(map->map_ip(map, map->start),
@@ -752,9 +750,9 @@
 {
 	struct map *map;
 
-	map = map_groups__find(&machine->kmaps, event->ksymbol.addr);
+	map = maps__find(&machine->kmaps, event->ksymbol.addr);
 	if (map)
-		map_groups__remove(&machine->kmaps, map);
+		maps__remove(&machine->kmaps, map);
 
 	return 0;
 }
@@ -790,9 +788,9 @@
 	if (map == NULL)
 		goto out;
 
-	map_groups__insert(&machine->kmaps, map);
+	maps__insert(&machine->kmaps, map);
 
-	/* Put the map here because map_groups__insert alread got it */
+	/* Put the map here because maps__insert alread got it */
 	map__put(map);
 out:
 	/* put the dso here, corresponding to  machine__findnew_module_dso */
@@ -977,7 +975,7 @@
 	kmap->kmaps = &machine->kmaps;
 	strlcpy(kmap->name, xm->name, KMAP_NAME_LEN);
 
-	map_groups__insert(&machine->kmaps, map);
+	maps__insert(&machine->kmaps, map);
 
 	pr_debug2("Added extra kernel map %s %" PRIx64 "-%" PRIx64 "\n",
 		  kmap->name, map->start, map->end);
@@ -1022,8 +1020,7 @@
 int machine__map_x86_64_entry_trampolines(struct machine *machine,
 					  struct dso *kernel)
 {
-	struct map_groups *kmaps = &machine->kmaps;
-	struct maps *maps = &kmaps->maps;
+	struct maps *kmaps = &machine->kmaps;
 	int nr_cpus_avail, cpu;
 	bool found = false;
 	struct map *map;
@@ -1033,14 +1030,14 @@
 	 * In the vmlinux case, pgoff is a virtual address which must now be
 	 * mapped to a vmlinux offset.
 	 */
-	maps__for_each_entry(maps, map) {
+	maps__for_each_entry(kmaps, map) {
 		struct kmap *kmap = __map__kmap(map);
 		struct map *dest_map;
 
 		if (!kmap || !is_entry_trampoline(kmap->name))
 			continue;
 
-		dest_map = map_groups__find(kmaps, map->pgoff);
+		dest_map = maps__find(kmaps, map->pgoff);
 		if (dest_map != map)
 			map->pgoff = dest_map->map_ip(dest_map, map->pgoff);
 		found = true;
@@ -1102,7 +1099,7 @@
 		return -1;
 
 	kmap->kmaps = &machine->kmaps;
-	map_groups__insert(&machine->kmaps, map);
+	maps__insert(&machine->kmaps, map);
 
 	return 0;
 }
@@ -1116,7 +1113,7 @@
 		return;
 
 	kmap = map__kmap(map);
-	map_groups__remove(&machine->kmaps, map);
+	maps__remove(&machine->kmaps, map);
 	if (kmap && kmap->ref_reloc_sym) {
 		zfree((char **)&kmap->ref_reloc_sym->name);
 		zfree(&kmap->ref_reloc_sym);
@@ -1211,7 +1208,7 @@
 		 * kernel, with modules between them, fixup the end of all
 		 * sections.
 		 */
-		map_groups__fixup_end(&machine->kmaps);
+		maps__fixup_end(&machine->kmaps);
 	}
 
 	return ret;
@@ -1262,11 +1259,10 @@
 	       dso->symtab_type == DSO_BINARY_TYPE__GUEST_KMODULE;
 }
 
-static int map_groups__set_module_path(struct map_groups *mg, const char *path,
-				       struct kmod_path *m)
+static int maps__set_module_path(struct maps *maps, const char *path, struct kmod_path *m)
 {
 	char *long_name;
-	struct map *map = map_groups__find_by_name(mg, m->name);
+	struct map *map = maps__find_by_name(maps, m->name);
 
 	if (map == NULL)
 		return 0;
@@ -1290,8 +1286,7 @@
 	return 0;
 }
 
-static int map_groups__set_modules_path_dir(struct map_groups *mg,
-				const char *dir_name, int depth)
+static int maps__set_modules_path_dir(struct maps *maps, const char *dir_name, int depth)
 {
 	struct dirent *dent;
 	DIR *dir = opendir(dir_name);
@@ -1323,8 +1318,7 @@
 					continue;
 			}
 
-			ret = map_groups__set_modules_path_dir(mg, path,
-							       depth + 1);
+			ret = maps__set_modules_path_dir(maps, path, depth + 1);
 			if (ret < 0)
 				goto out;
 		} else {
@@ -1335,7 +1329,7 @@
 				goto out;
 
 			if (m.kmod)
-				ret = map_groups__set_module_path(mg, path, &m);
+				ret = maps__set_module_path(maps, path, &m);
 
 			zfree(&m.name);
 
@@ -1362,7 +1356,7 @@
 		 machine->root_dir, version);
 	free(version);
 
-	return map_groups__set_modules_path_dir(&machine->kmaps, modules_path, 0);
+	return maps__set_modules_path_dir(&machine->kmaps, modules_path, 0);
 }
 int __weak arch__fix_module_text_start(u64 *start __maybe_unused,
 				u64 *size __maybe_unused,
@@ -1435,11 +1429,11 @@
 	struct map *map = machine__kernel_map(machine);
 
 	map__get(map);
-	map_groups__remove(&machine->kmaps, map);
+	maps__remove(&machine->kmaps, map);
 
 	machine__set_kernel_mmap(machine, start, end);
 
-	map_groups__insert(&machine->kmaps, map);
+	maps__insert(&machine->kmaps, map);
 	map__put(map);
 }
 
@@ -1940,7 +1934,7 @@
 
 	ams->addr = ip;
 	ams->al_addr = al.addr;
-	ams->ms.mg  = al.mg;
+	ams->ms.maps = al.maps;
 	ams->ms.sym = al.sym;
 	ams->ms.map = al.map;
 	ams->phys_addr = 0;
@@ -1958,7 +1952,7 @@
 
 	ams->addr = addr;
 	ams->al_addr = al.addr;
-	ams->ms.mg  = al.mg;
+	ams->ms.maps = al.maps;
 	ams->ms.sym = al.sym;
 	ams->ms.map = al.map;
 	ams->phys_addr = phys_addr;
@@ -2075,7 +2069,7 @@
 		iter_cycles = iter->cycles;
 	}
 
-	ms.mg  = al.mg;
+	ms.maps = al.maps;
 	ms.map = al.map;
 	ms.sym = al.sym;
 	srcline = callchain_srcline(&ms, al.addr);
diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h
index 499be20..be0a930 100644
--- a/tools/perf/util/machine.h
+++ b/tools/perf/util/machine.h
@@ -4,7 +4,7 @@
 
 #include <sys/types.h>
 #include <linux/rbtree.h>
-#include "map_groups.h"
+#include "maps.h"
 #include "dsos.h"
 #include "rwsem.h"
 
@@ -51,7 +51,7 @@
 	struct vdso_info  *vdso_info;
 	struct perf_env   *env;
 	struct dsos	  dsos;
-	struct map_groups kmaps;
+	struct maps	  kmaps;
 	struct map	  *vmlinux_map;
 	u64		  kernel_start;
 	pid_t		  *current_tid;
@@ -83,7 +83,7 @@
 static inline
 struct maps *machine__kernel_maps(struct machine *machine)
 {
-	return &machine->kmaps.maps;
+	return &machine->kmaps;
 }
 
 int machine__get_kernel_start(struct machine *machine);
@@ -212,7 +212,7 @@
 struct symbol *machine__find_kernel_symbol(struct machine *machine, u64 addr,
 					   struct map **mapp)
 {
-	return map_groups__find_symbol(&machine->kmaps, addr, mapp);
+	return maps__find_symbol(&machine->kmaps, addr, mapp);
 }
 
 static inline
@@ -220,7 +220,7 @@
 						   const char *name,
 						   struct map **mapp)
 {
-	return map_groups__find_symbol_by_name(&machine->kmaps, name, mapp);
+	return maps__find_symbol_by_name(&machine->kmaps, name, mapp);
 }
 
 int arch__fix_module_text_start(u64 *start, u64 *size, const char *name);
diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c
index 744bfba..fdd5bdd 100644
--- a/tools/perf/util/map.c
+++ b/tools/perf/util/map.c
@@ -433,51 +433,6 @@
 	return ret;
 }
 
-int map__fprintf_srccode(struct map *map, u64 addr,
-			 FILE *fp,
-			 struct srccode_state *state)
-{
-	char *srcfile;
-	int ret = 0;
-	unsigned line;
-	int len;
-	char *srccode;
-
-	if (!map || !map->dso)
-		return 0;
-	srcfile = get_srcline_split(map->dso,
-				    map__rip_2objdump(map, addr),
-				    &line);
-	if (!srcfile)
-		return 0;
-
-	/* Avoid redundant printing */
-	if (state &&
-	    state->srcfile &&
-	    !strcmp(state->srcfile, srcfile) &&
-	    state->line == line) {
-		free(srcfile);
-		return 0;
-	}
-
-	srccode = find_sourceline(srcfile, line, &len);
-	if (!srccode)
-		goto out_free_line;
-
-	ret = fprintf(fp, "|%-8d %.*s", line, len, srccode);
-
-	if (state) {
-		state->srcfile = srcfile;
-		state->line = line;
-	}
-	return ret;
-
-out_free_line:
-	free(srcfile);
-	return ret;
-}
-
-
 void srccode_state_free(struct srccode_state *state)
 {
 	zfree(&state->srcfile);
@@ -557,73 +512,71 @@
 	return ip + map->reloc;
 }
 
-static void maps__init(struct maps *maps)
+void maps__init(struct maps *maps, struct machine *machine)
 {
 	maps->entries = RB_ROOT;
 	init_rwsem(&maps->lock);
+	maps->machine = machine;
+	maps->last_search_by_name = NULL;
+	maps->nr_maps = 0;
+	maps->maps_by_name = NULL;
+	refcount_set(&maps->refcnt, 1);
 }
 
-void map_groups__init(struct map_groups *mg, struct machine *machine)
-{
-	maps__init(&mg->maps);
-	mg->machine = machine;
-	mg->last_search_by_name = NULL;
-	mg->nr_maps = 0;
-	mg->maps_by_name = NULL;
-	refcount_set(&mg->refcnt, 1);
-}
-
-static void __map_groups__free_maps_by_name(struct map_groups *mg)
+static void __maps__free_maps_by_name(struct maps *maps)
 {
 	/*
 	 * Free everything to try to do it from the rbtree in the next search
 	 */
-	zfree(&mg->maps_by_name);
-	mg->nr_maps_allocated = 0;
+	zfree(&maps->maps_by_name);
+	maps->nr_maps_allocated = 0;
 }
 
-void map_groups__insert(struct map_groups *mg, struct map *map)
+void maps__insert(struct maps *maps, struct map *map)
 {
-	struct maps *maps = &mg->maps;
-
 	down_write(&maps->lock);
 	__maps__insert(maps, map);
-	++mg->nr_maps;
+	++maps->nr_maps;
 
 	/*
 	 * If we already performed some search by name, then we need to add the just
 	 * inserted map and resort.
 	 */
-	if (mg->maps_by_name) {
-		if (mg->nr_maps > mg->nr_maps_allocated) {
-			int nr_allocate = mg->nr_maps * 2;
-			struct map **maps_by_name = realloc(mg->maps_by_name, nr_allocate * sizeof(map));
+	if (maps->maps_by_name) {
+		if (maps->nr_maps > maps->nr_maps_allocated) {
+			int nr_allocate = maps->nr_maps * 2;
+			struct map **maps_by_name = realloc(maps->maps_by_name, nr_allocate * sizeof(map));
 
 			if (maps_by_name == NULL) {
-				__map_groups__free_maps_by_name(mg);
+				__maps__free_maps_by_name(maps);
 				return;
 			}
 
-			mg->maps_by_name = maps_by_name;
-			mg->nr_maps_allocated = nr_allocate;
+			maps->maps_by_name = maps_by_name;
+			maps->nr_maps_allocated = nr_allocate;
 		}
-		mg->maps_by_name[mg->nr_maps - 1] = map;
-		__map_groups__sort_by_name(mg);
+		maps->maps_by_name[maps->nr_maps - 1] = map;
+		__maps__sort_by_name(maps);
 	}
 	up_write(&maps->lock);
 }
 
-void map_groups__remove(struct map_groups *mg, struct map *map)
+static void __maps__remove(struct maps *maps, struct map *map)
 {
-	struct maps *maps = &mg->maps;
+	rb_erase_init(&map->rb_node, &maps->entries);
+	map__put(map);
+}
+
+void maps__remove(struct maps *maps, struct map *map)
+{
 	down_write(&maps->lock);
-	if (mg->last_search_by_name == map)
-		mg->last_search_by_name = NULL;
+	if (maps->last_search_by_name == map)
+		maps->last_search_by_name = NULL;
 
 	__maps__remove(maps, map);
-	--mg->nr_maps;
-	if (mg->maps_by_name)
-		__map_groups__free_maps_by_name(mg);
+	--maps->nr_maps;
+	if (maps->maps_by_name)
+		__maps__free_maps_by_name(maps);
 	up_write(&maps->lock);
 }
 
@@ -637,50 +590,44 @@
 	}
 }
 
-static void maps__exit(struct maps *maps)
+void maps__exit(struct maps *maps)
 {
 	down_write(&maps->lock);
 	__maps__purge(maps);
 	up_write(&maps->lock);
 }
 
-void map_groups__exit(struct map_groups *mg)
+bool maps__empty(struct maps *maps)
 {
-	maps__exit(&mg->maps);
+	return !maps__first(maps);
 }
 
-bool map_groups__empty(struct map_groups *mg)
+struct maps *maps__new(struct machine *machine)
 {
-	return !maps__first(&mg->maps);
+	struct maps *maps = zalloc(sizeof(*maps));
+
+	if (maps != NULL)
+		maps__init(maps, machine);
+
+	return maps;
 }
 
-struct map_groups *map_groups__new(struct machine *machine)
+void maps__delete(struct maps *maps)
 {
-	struct map_groups *mg = zalloc(sizeof(*mg));
-
-	if (mg != NULL)
-		map_groups__init(mg, machine);
-
-	return mg;
+	maps__exit(maps);
+	unwind__finish_access(maps);
+	free(maps);
 }
 
-void map_groups__delete(struct map_groups *mg)
+void maps__put(struct maps *maps)
 {
-	map_groups__exit(mg);
-	unwind__finish_access(mg);
-	free(mg);
+	if (maps && refcount_dec_and_test(&maps->refcnt))
+		maps__delete(maps);
 }
 
-void map_groups__put(struct map_groups *mg)
+struct symbol *maps__find_symbol(struct maps *maps, u64 addr, struct map **mapp)
 {
-	if (mg && refcount_dec_and_test(&mg->refcnt))
-		map_groups__delete(mg);
-}
-
-struct symbol *map_groups__find_symbol(struct map_groups *mg,
-				       u64 addr, struct map **mapp)
-{
-	struct map *map = map_groups__find(mg, addr);
+	struct map *map = maps__find(maps, addr);
 
 	/* Ensure map is loaded before using map->map_ip */
 	if (map != NULL && map__load(map) >= 0) {
@@ -699,8 +646,7 @@
 	return ip >= map->start && ip < map->end;
 }
 
-struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name,
-					 struct map **mapp)
+struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name, struct map **mapp)
 {
 	struct symbol *sym;
 	struct map *pos;
@@ -727,19 +673,12 @@
 	return sym;
 }
 
-struct symbol *map_groups__find_symbol_by_name(struct map_groups *mg,
-					       const char *name,
-					       struct map **mapp)
-{
-	return maps__find_symbol_by_name(&mg->maps, name, mapp);
-}
-
-int map_groups__find_ams(struct map_groups *mg, struct addr_map_symbol *ams)
+int maps__find_ams(struct maps *maps, struct addr_map_symbol *ams)
 {
 	if (ams->addr < ams->ms.map->start || ams->addr >= ams->ms.map->end) {
-		if (mg == NULL)
+		if (maps == NULL)
 			return -1;
-		ams->ms.map = map_groups__find(mg, ams->addr);
+		ams->ms.map = maps__find(maps, ams->addr);
 		if (ams->ms.map == NULL)
 			return -1;
 	}
@@ -750,7 +689,7 @@
 	return ams->ms.sym ? 0 : -1;
 }
 
-static size_t maps__fprintf(struct maps *maps, FILE *fp)
+size_t maps__fprintf(struct maps *maps, FILE *fp)
 {
 	size_t printed = 0;
 	struct map *pos;
@@ -771,19 +710,8 @@
 	return printed;
 }
 
-size_t map_groups__fprintf(struct map_groups *mg, FILE *fp)
+int maps__fixup_overlappings(struct maps *maps, struct map *map, FILE *fp)
 {
-	return maps__fprintf(&mg->maps, fp);
-}
-
-static void __map_groups__insert(struct map_groups *mg, struct map *map)
-{
-	__maps__insert(&mg->maps, map);
-}
-
-int map_groups__fixup_overlappings(struct map_groups *mg, struct map *map, FILE *fp)
-{
-	struct maps *maps = &mg->maps;
 	struct rb_root *root;
 	struct rb_node *next, *first;
 	int err = 0;
@@ -848,7 +776,7 @@
 			}
 
 			before->end = map->start;
-			__map_groups__insert(mg, before);
+			__maps__insert(maps, before);
 			if (verbose >= 2 && !use_browser)
 				map__fprintf(before, fp);
 			map__put(before);
@@ -865,7 +793,7 @@
 			after->start = map->end;
 			after->pgoff += map->end - pos->start;
 			assert(pos->map_ip(pos, map->end) == after->map_ip(after, map->end));
-			__map_groups__insert(mg, after);
+			__maps__insert(maps, after);
 			if (verbose >= 2 && !use_browser)
 				map__fprintf(after, fp);
 			map__put(after);
@@ -886,31 +814,30 @@
 /*
  * XXX This should not really _copy_ te maps, but refcount them.
  */
-int map_groups__clone(struct thread *thread, struct map_groups *parent)
+int maps__clone(struct thread *thread, struct maps *parent)
 {
-	struct map_groups *mg = thread->mg;
+	struct maps *maps = thread->maps;
 	int err = -ENOMEM;
 	struct map *map;
-	struct maps *maps = &parent->maps;
 
-	down_read(&maps->lock);
+	down_read(&parent->lock);
 
-	maps__for_each_entry(maps, map) {
+	maps__for_each_entry(parent, map) {
 		struct map *new = map__clone(map);
 		if (new == NULL)
 			goto out_unlock;
 
-		err = unwind__prepare_access(mg, new, NULL);
+		err = unwind__prepare_access(maps, new, NULL);
 		if (err)
 			goto out_unlock;
 
-		map_groups__insert(mg, new);
+		maps__insert(maps, new);
 		map__put(new);
 	}
 
 	err = 0;
 out_unlock:
-	up_read(&maps->lock);
+	up_read(&parent->lock);
 	return err;
 }
 
@@ -935,26 +862,6 @@
 	map__get(map);
 }
 
-void maps__insert(struct maps *maps, struct map *map)
-{
-	down_write(&maps->lock);
-	__maps__insert(maps, map);
-	up_write(&maps->lock);
-}
-
-void __maps__remove(struct maps *maps, struct map *map)
-{
-	rb_erase_init(&map->rb_node, &maps->entries);
-	map__put(map);
-}
-
-void maps__remove(struct maps *maps, struct map *map)
-{
-	down_write(&maps->lock);
-	__maps__remove(maps, map);
-	up_write(&maps->lock);
-}
-
 struct map *maps__find(struct maps *maps, u64 ip)
 {
 	struct rb_node *p;
@@ -1018,7 +925,7 @@
 	return kmap;
 }
 
-struct map_groups *map__kmaps(struct map *map)
+struct maps *map__kmaps(struct map *map)
 {
 	struct kmap *kmap = map__kmap(map);
 
diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h
index 5e88998..067036e 100644
--- a/tools/perf/util/map.h
+++ b/tools/perf/util/map.h
@@ -12,11 +12,8 @@
 #include <linux/types.h>
 
 struct dso;
-struct ip_callchain;
-struct ref_reloc_sym;
-struct map_groups;
+struct maps;
 struct machine;
-struct evsel;
 
 struct map {
 	union {
@@ -45,7 +42,7 @@
 
 struct kmap *__map__kmap(struct map *map);
 struct kmap *map__kmap(struct map *map);
-struct map_groups *map__kmaps(struct map *map);
+struct maps *map__kmaps(struct map *map);
 
 static inline u64 map__map_ip(struct map *map, u64 ip)
 {
@@ -138,19 +135,12 @@
 int map__fprintf_srcline(struct map *map, u64 addr, const char *prefix,
 			 FILE *fp);
 
-struct srccode_state;
-
-int map__fprintf_srccode(struct map *map, u64 addr,
-			 FILE *fp, struct srccode_state *state);
-
 int map__load(struct map *map);
 struct symbol *map__find_symbol(struct map *map, u64 addr);
 struct symbol *map__find_symbol_by_name(struct map *map, const char *name);
 void map__fixup_start(struct map *map);
 void map__fixup_end(struct map *map);
 
-void map__reloc_vmlinux(struct map *map);
-
 int map__set_kallsyms_ref_reloc_sym(struct map *map, const char *symbol_name,
 				    u64 addr);
 
diff --git a/tools/perf/util/map_groups.h b/tools/perf/util/map_groups.h
deleted file mode 100644
index 63ed211..0000000
--- a/tools/perf/util/map_groups.h
+++ /dev/null
@@ -1,106 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-#ifndef __PERF_MAP_GROUPS_H
-#define __PERF_MAP_GROUPS_H
-
-#include <linux/refcount.h>
-#include <linux/rbtree.h>
-#include <stdio.h>
-#include <stdbool.h>
-#include <linux/types.h>
-#include "rwsem.h"
-
-struct ref_reloc_sym;
-struct machine;
-struct map;
-struct thread;
-
-struct maps {
-	struct rb_root      entries;
-	struct rw_semaphore lock;
-};
-
-void maps__insert(struct maps *maps, struct map *map);
-void maps__remove(struct maps *maps, struct map *map);
-void __maps__remove(struct maps *maps, struct map *map);
-struct map *maps__find(struct maps *maps, u64 addr);
-struct map *maps__first(struct maps *maps);
-struct map *map__next(struct map *map);
-
-#define maps__for_each_entry(maps, map) \
-	for (map = maps__first(maps); map; map = map__next(map))
-
-#define maps__for_each_entry_safe(maps, map, next) \
-	for (map = maps__first(maps), next = map__next(map); map; map = next, next = map__next(map))
-
-struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name, struct map **mapp);
-
-struct map_groups {
-	struct maps	 maps;
-	struct machine	 *machine;
-	struct map	 *last_search_by_name;
-	struct map	 **maps_by_name;
-	refcount_t	 refcnt;
-	unsigned int	 nr_maps;
-	unsigned int	 nr_maps_allocated;
-#ifdef HAVE_LIBUNWIND_SUPPORT
-	void				*addr_space;
-	struct unwind_libunwind_ops	*unwind_libunwind_ops;
-#endif
-};
-
-#define KMAP_NAME_LEN 256
-
-struct kmap {
-	struct ref_reloc_sym *ref_reloc_sym;
-	struct map_groups    *kmaps;
-	char		     name[KMAP_NAME_LEN];
-};
-
-struct map_groups *map_groups__new(struct machine *machine);
-void map_groups__delete(struct map_groups *mg);
-bool map_groups__empty(struct map_groups *mg);
-
-static inline struct map_groups *map_groups__get(struct map_groups *mg)
-{
-	if (mg)
-		refcount_inc(&mg->refcnt);
-	return mg;
-}
-
-void map_groups__put(struct map_groups *mg);
-void map_groups__init(struct map_groups *mg, struct machine *machine);
-void map_groups__exit(struct map_groups *mg);
-int map_groups__clone(struct thread *thread, struct map_groups *parent);
-size_t map_groups__fprintf(struct map_groups *mg, FILE *fp);
-
-void map_groups__insert(struct map_groups *mg, struct map *map);
-
-void map_groups__remove(struct map_groups *mg, struct map *map);
-
-static inline struct map *map_groups__find(struct map_groups *mg, u64 addr)
-{
-	return maps__find(&mg->maps, addr);
-}
-
-#define map_groups__for_each_entry(mg, map) \
-	for (map = maps__first(&mg->maps); map; map = map__next(map))
-
-#define map_groups__for_each_entry_safe(mg, map, next) \
-	for (map = maps__first(&mg->maps), next = map__next(map); map; map = next, next = map__next(map))
-
-struct symbol *map_groups__find_symbol(struct map_groups *mg, u64 addr, struct map **mapp);
-struct symbol *map_groups__find_symbol_by_name(struct map_groups *mg, const char *name, struct map **mapp);
-
-struct addr_map_symbol;
-
-int map_groups__find_ams(struct map_groups *mg, struct addr_map_symbol *ams);
-
-int map_groups__fixup_overlappings(struct map_groups *mg, struct map *map, FILE *fp);
-
-struct map *map_groups__find_by_name(struct map_groups *mg, const char *name);
-
-int map_groups__merge_in(struct map_groups *kmaps, struct map *new_map);
-
-void __map_groups__sort_by_name(struct map_groups *mg);
-
-#endif // __PERF_MAP_GROUPS_H
diff --git a/tools/perf/util/map_symbol.h b/tools/perf/util/map_symbol.h
index 2964d97..5b8ca93 100644
--- a/tools/perf/util/map_symbol.h
+++ b/tools/perf/util/map_symbol.h
@@ -4,12 +4,12 @@
 
 #include <linux/types.h>
 
-struct map_groups;
+struct maps;
 struct map;
 struct symbol;
 
 struct map_symbol {
-	struct map_groups *mg;
+	struct maps   *maps;
 	struct map    *map;
 	struct symbol *sym;
 };
diff --git a/tools/perf/util/maps.h b/tools/perf/util/maps.h
new file mode 100644
index 0000000..3dd000d
--- /dev/null
+++ b/tools/perf/util/maps.h
@@ -0,0 +1,87 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __PERF_MAPS_H
+#define __PERF_MAPS_H
+
+#include <linux/refcount.h>
+#include <linux/rbtree.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <linux/types.h>
+#include "rwsem.h"
+
+struct ref_reloc_sym;
+struct machine;
+struct map;
+struct maps;
+struct thread;
+
+struct map *maps__find(struct maps *maps, u64 addr);
+struct map *maps__first(struct maps *maps);
+struct map *map__next(struct map *map);
+
+#define maps__for_each_entry(maps, map) \
+	for (map = maps__first(maps); map; map = map__next(map))
+
+#define maps__for_each_entry_safe(maps, map, next) \
+	for (map = maps__first(maps), next = map__next(map); map; map = next, next = map__next(map))
+
+struct maps {
+	struct rb_root      entries;
+	struct rw_semaphore lock;
+	struct machine	 *machine;
+	struct map	 *last_search_by_name;
+	struct map	 **maps_by_name;
+	refcount_t	 refcnt;
+	unsigned int	 nr_maps;
+	unsigned int	 nr_maps_allocated;
+#ifdef HAVE_LIBUNWIND_SUPPORT
+	void				*addr_space;
+	struct unwind_libunwind_ops	*unwind_libunwind_ops;
+#endif
+};
+
+#define KMAP_NAME_LEN 256
+
+struct kmap {
+	struct ref_reloc_sym *ref_reloc_sym;
+	struct maps	     *kmaps;
+	char		     name[KMAP_NAME_LEN];
+};
+
+struct maps *maps__new(struct machine *machine);
+void maps__delete(struct maps *maps);
+bool maps__empty(struct maps *maps);
+
+static inline struct maps *maps__get(struct maps *maps)
+{
+	if (maps)
+		refcount_inc(&maps->refcnt);
+	return maps;
+}
+
+void maps__put(struct maps *maps);
+void maps__init(struct maps *maps, struct machine *machine);
+void maps__exit(struct maps *maps);
+int maps__clone(struct thread *thread, struct maps *parent);
+size_t maps__fprintf(struct maps *maps, FILE *fp);
+
+void maps__insert(struct maps *maps, struct map *map);
+
+void maps__remove(struct maps *maps, struct map *map);
+
+struct symbol *maps__find_symbol(struct maps *maps, u64 addr, struct map **mapp);
+struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name, struct map **mapp);
+
+struct addr_map_symbol;
+
+int maps__find_ams(struct maps *maps, struct addr_map_symbol *ams);
+
+int maps__fixup_overlappings(struct maps *maps, struct map *map, FILE *fp);
+
+struct map *maps__find_by_name(struct maps *maps, const char *name);
+
+int maps__merge_in(struct maps *kmaps, struct map *new_map);
+
+void __maps__sort_by_name(struct maps *maps);
+
+#endif // __PERF_MAPS_H
diff --git a/tools/perf/util/perf_regs.h b/tools/perf/util/perf_regs.h
index e014c2c..a454991 100644
--- a/tools/perf/util/perf_regs.h
+++ b/tools/perf/util/perf_regs.h
@@ -41,7 +41,7 @@
 
 static inline const char *perf_reg_name(int id __maybe_unused)
 {
-	return NULL;
+	return "unknown";
 }
 
 static inline int perf_reg_value(u64 *valp __maybe_unused,
diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
index e8d3489..8b99fd3 100644
--- a/tools/perf/util/pmu.c
+++ b/tools/perf/util/pmu.c
@@ -24,6 +24,7 @@
 #include "pmu-events/pmu-events.h"
 #include "string2.h"
 #include "strbuf.h"
+#include "fncache.h"
 
 struct perf_pmu_format {
 	char *name;
@@ -82,7 +83,6 @@
  */
 static int pmu_format(const char *name, struct list_head *format)
 {
-	struct stat st;
 	char path[PATH_MAX];
 	const char *sysfs = sysfs__mountpoint();
 
@@ -92,8 +92,8 @@
 	snprintf(path, PATH_MAX,
 		 "%s" EVENT_SOURCE_DEVICE_PATH "%s/format", sysfs, name);
 
-	if (stat(path, &st) < 0)
-		return 0;	/* no error if format does not exist */
+	if (!file_available(path))
+		return 0;
 
 	if (perf_pmu__format_parse(path, format))
 		return -1;
@@ -475,7 +475,6 @@
  */
 static int pmu_aliases(const char *name, struct list_head *head)
 {
-	struct stat st;
 	char path[PATH_MAX];
 	const char *sysfs = sysfs__mountpoint();
 
@@ -485,8 +484,8 @@
 	snprintf(path, PATH_MAX,
 		 "%s/bus/event_source/devices/%s/events", sysfs, name);
 
-	if (stat(path, &st) < 0)
-		return 0;	 /* no error if 'events' does not exist */
+	if (!file_available(path))
+		return 0;
 
 	if (pmu_aliases_parse(path, head))
 		return -1;
@@ -525,7 +524,6 @@
  */
 static int pmu_type(const char *name, __u32 *type)
 {
-	struct stat st;
 	char path[PATH_MAX];
 	FILE *file;
 	int ret = 0;
@@ -537,7 +535,7 @@
 	snprintf(path, PATH_MAX,
 		 "%s" EVENT_SOURCE_DEVICE_PATH "%s/type", sysfs, name);
 
-	if (stat(path, &st) < 0)
+	if (access(path, R_OK) < 0)
 		return -1;
 
 	file = fopen(path, "r");
@@ -628,14 +626,11 @@
 static bool pmu_is_uncore(const char *name)
 {
 	char path[PATH_MAX];
-	struct perf_cpu_map *cpus;
-	const char *sysfs = sysfs__mountpoint();
+	const char *sysfs;
 
+	sysfs = sysfs__mountpoint();
 	snprintf(path, PATH_MAX, CPUS_TEMPLATE_UNCORE, sysfs, name);
-	cpus = __pmu_cpumask(path);
-	perf_cpu_map__put(cpus);
-
-	return !!cpus;
+	return file_available(path);
 }
 
 /*
@@ -645,7 +640,6 @@
  */
 static int is_arm_pmu_core(const char *name)
 {
-	struct stat st;
 	char path[PATH_MAX];
 	const char *sysfs = sysfs__mountpoint();
 
@@ -655,10 +649,7 @@
 	/* Look for cpu sysfs (specific to arm) */
 	scnprintf(path, PATH_MAX, "%s/bus/event_source/devices/%s/cpus",
 				sysfs, name);
-	if (stat(path, &st) == 0)
-		return 1;
-
-	return 0;
+	return file_available(path);
 }
 
 static char *perf_pmu__getcpuid(struct perf_pmu *pmu)
@@ -1544,7 +1535,6 @@
 
 static FILE *perf_pmu__open_file(struct perf_pmu *pmu, const char *name)
 {
-	struct stat st;
 	char path[PATH_MAX];
 	const char *sysfs;
 
@@ -1554,10 +1544,8 @@
 
 	snprintf(path, PATH_MAX,
 		 "%s" EVENT_SOURCE_DEVICE_PATH "%s/%s", sysfs, pmu->name, name);
-
-	if (stat(path, &st) < 0)
+	if (!file_available(path))
 		return NULL;
-
 	return fopen(path, "r");
 }
 
diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index 52b2d16..eea132f 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -28,7 +28,7 @@
 #include "dso.h"
 #include "color.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "symbol.h"
 #include <api/fs/fs.h>
 #include "trace-event.h"	/* For __maybe_unused */
@@ -321,7 +321,7 @@
 		char module_name[128];
 
 		snprintf(module_name, sizeof(module_name), "[%s]", module);
-		map = map_groups__find_by_name(&host_machine->kmaps, module_name);
+		map = maps__find_by_name(&host_machine->kmaps, module_name);
 		if (map) {
 			dso = map->dso;
 			goto found;
diff --git a/tools/perf/util/python-ext-sources b/tools/perf/util/python-ext-sources
index 9af1838..e7279ea 100644
--- a/tools/perf/util/python-ext-sources
+++ b/tools/perf/util/python-ext-sources
@@ -33,3 +33,4 @@
 util/string.c
 util/symbol_fprintf.c
 util/units.c
+util/affinity.c
diff --git a/tools/perf/util/scripting-engines/trace-event-python.c b/tools/perf/util/scripting-engines/trace-event-python.c
index 9581a90..80ca5d0 100644
--- a/tools/perf/util/scripting-engines/trace-event-python.c
+++ b/tools/perf/util/scripting-engines/trace-event-python.c
@@ -1127,7 +1127,7 @@
 
 	tuple_set_u64(t, 0, es->db_id);
 	tuple_set_u64(t, 1, es->evsel->db_id);
-	tuple_set_u64(t, 2, es->al->mg->machine->db_id);
+	tuple_set_u64(t, 2, es->al->maps->machine->db_id);
 	tuple_set_u64(t, 3, es->al->thread->db_id);
 	tuple_set_u64(t, 4, es->comm_db_id);
 	tuple_set_u64(t, 5, es->dso_db_id);
diff --git a/tools/perf/util/srccode.c b/tools/perf/util/srccode.c
index d84ed8b..c29edaa 100644
--- a/tools/perf/util/srccode.c
+++ b/tools/perf/util/srccode.c
@@ -16,6 +16,7 @@
 #include "srccode.h"
 #include "debug.h"
 #include <internal/lib.h> // page_size
+#include "fncache.h"
 
 #define MAXSRCCACHE (32*1024*1024)
 #define MAXSRCFILES     64
@@ -36,14 +37,6 @@
 static long map_total_sz;
 static int num_srcfiles;
 
-static unsigned shash(unsigned char *s)
-{
-	unsigned h = 0;
-	while (*s)
-		h = 65599 * h + *s++;
-	return h ^ (h >> 16);
-}
-
 static int countlines(char *map, int maplen)
 {
 	int numl;
diff --git a/tools/perf/util/symbol-elf.c b/tools/perf/util/symbol-elf.c
index 16776d5..6658fbf 100644
--- a/tools/perf/util/symbol-elf.c
+++ b/tools/perf/util/symbol-elf.c
@@ -9,7 +9,7 @@
 
 #include "dso.h"
 #include "map.h"
-#include "map_groups.h"
+#include "maps.h"
 #include "symbol.h"
 #include "symsrc.h"
 #include "demangle-java.h"
@@ -844,7 +844,7 @@
 
 static int dso__process_kernel_symbol(struct dso *dso, struct map *map,
 				      GElf_Sym *sym, GElf_Shdr *shdr,
-				      struct map_groups *kmaps, struct kmap *kmap,
+				      struct maps *kmaps, struct kmap *kmap,
 				      struct dso **curr_dsop, struct map **curr_mapp,
 				      const char *section_name,
 				      bool adjust_kernel_syms, bool kmodule, bool *remap_kernel)
@@ -876,8 +876,8 @@
 			/* Ensure maps are correctly ordered */
 			if (kmaps) {
 				map__get(map);
-				map_groups__remove(kmaps, map);
-				map_groups__insert(kmaps, map);
+				maps__remove(kmaps, map);
+				maps__insert(kmaps, map);
 				map__put(map);
 			}
 		}
@@ -902,7 +902,7 @@
 
 	snprintf(dso_name, sizeof(dso_name), "%s%s", dso->short_name, section_name);
 
-	curr_map = map_groups__find_by_name(kmaps, dso_name);
+	curr_map = maps__find_by_name(kmaps, dso_name);
 	if (curr_map == NULL) {
 		u64 start = sym->st_value;
 
@@ -928,7 +928,7 @@
 			curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
 		}
 		curr_dso->symtab_type = dso->symtab_type;
-		map_groups__insert(kmaps, curr_map);
+		maps__insert(kmaps, curr_map);
 		/*
 		 * Add it before we drop the referece to curr_map, i.e. while
 		 * we still are sure to have a reference to this DSO via
@@ -950,7 +950,7 @@
 		  struct symsrc *runtime_ss, int kmodule)
 {
 	struct kmap *kmap = dso->kernel ? map__kmap(map) : NULL;
-	struct map_groups *kmaps = kmap ? map__kmaps(map) : NULL;
+	struct maps *kmaps = kmap ? map__kmaps(map) : NULL;
 	struct map *curr_map = map;
 	struct dso *curr_dso = dso;
 	Elf_Data *symstrs, *secstrs;
@@ -1162,7 +1162,7 @@
 			 * We need to fixup this here too because we create new
 			 * maps here, for things like vsyscall sections.
 			 */
-			map_groups__fixup_end(kmaps);
+			maps__fixup_end(kmaps);
 		}
 	}
 	err = nr;
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index db9667a..3b379b1 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -239,9 +239,8 @@
 		curr->end = roundup(curr->start, 4096) + 4096;
 }
 
-void map_groups__fixup_end(struct map_groups *mg)
+void maps__fixup_end(struct maps *maps)
 {
-	struct maps *maps = &mg->maps;
 	struct map *prev = NULL, *curr;
 
 	down_write(&maps->lock);
@@ -698,7 +697,7 @@
 	return kallsyms__parse(filename, dso, map__process_kallsym_symbol);
 }
 
-static int map_groups__split_kallsyms_for_kcore(struct map_groups *kmaps, struct dso *dso)
+static int maps__split_kallsyms_for_kcore(struct maps *kmaps, struct dso *dso)
 {
 	struct map *curr_map;
 	struct symbol *pos;
@@ -724,7 +723,7 @@
 		if (module)
 			*module = '\0';
 
-		curr_map = map_groups__find(kmaps, pos->start);
+		curr_map = maps__find(kmaps, pos->start);
 
 		if (!curr_map) {
 			symbol__delete(pos);
@@ -751,8 +750,8 @@
  * kernel range is broken in several maps, named [kernel].N, as we don't have
  * the original ELF section names vmlinux have.
  */
-static int map_groups__split_kallsyms(struct map_groups *kmaps, struct dso *dso, u64 delta,
-				      struct map *initial_map)
+static int maps__split_kallsyms(struct maps *kmaps, struct dso *dso, u64 delta,
+				struct map *initial_map)
 {
 	struct machine *machine;
 	struct map *curr_map = initial_map;
@@ -797,7 +796,7 @@
 					dso__set_loaded(curr_map->dso);
 				}
 
-				curr_map = map_groups__find_by_name(kmaps, module);
+				curr_map = maps__find_by_name(kmaps, module);
 				if (curr_map == NULL) {
 					pr_debug("%s/proc/{kallsyms,modules} "
 					         "inconsistency while looking "
@@ -864,7 +863,7 @@
 			}
 
 			curr_map->map_ip = curr_map->unmap_ip = identity__map_ip;
-			map_groups__insert(kmaps, curr_map);
+			maps__insert(kmaps, curr_map);
 			++kernel_range;
 		} else if (delta) {
 			/* Kernel was relocated at boot time */
@@ -1049,8 +1048,7 @@
 	return ret;
 }
 
-static int do_validate_kcore_modules(const char *filename,
-				  struct map_groups *kmaps)
+static int do_validate_kcore_modules(const char *filename, struct maps *kmaps)
 {
 	struct rb_root modules = RB_ROOT;
 	struct map *old_map;
@@ -1060,7 +1058,7 @@
 	if (err)
 		return err;
 
-	map_groups__for_each_entry(kmaps, old_map) {
+	maps__for_each_entry(kmaps, old_map) {
 		struct module_info *mi;
 
 		if (!__map__is_kmodule(old_map)) {
@@ -1107,7 +1105,7 @@
 static int validate_kcore_modules(const char *kallsyms_filename,
 				  struct map *map)
 {
-	struct map_groups *kmaps = map__kmaps(map);
+	struct maps *kmaps = map__kmaps(map);
 	char modules_filename[PATH_MAX];
 
 	if (!kmaps)
@@ -1167,15 +1165,15 @@
 }
 
 /*
- * Merges map into map_groups by splitting the new map
- * within the existing map regions.
+ * Merges map into maps by splitting the new map within the existing map
+ * regions.
  */
-int map_groups__merge_in(struct map_groups *kmaps, struct map *new_map)
+int maps__merge_in(struct maps *kmaps, struct map *new_map)
 {
 	struct map *old_map;
 	LIST_HEAD(merged);
 
-	map_groups__for_each_entry(kmaps, old_map) {
+	maps__for_each_entry(kmaps, old_map) {
 		/* no overload with this one */
 		if (new_map->end < old_map->start ||
 		    new_map->start >= old_map->end)
@@ -1232,12 +1230,12 @@
 	while (!list_empty(&merged)) {
 		old_map = list_entry(merged.next, struct map, node);
 		list_del_init(&old_map->node);
-		map_groups__insert(kmaps, old_map);
+		maps__insert(kmaps, old_map);
 		map__put(old_map);
 	}
 
 	if (new_map) {
-		map_groups__insert(kmaps, new_map);
+		maps__insert(kmaps, new_map);
 		map__put(new_map);
 	}
 	return 0;
@@ -1246,7 +1244,7 @@
 static int dso__load_kcore(struct dso *dso, struct map *map,
 			   const char *kallsyms_filename)
 {
-	struct map_groups *kmaps = map__kmaps(map);
+	struct maps *kmaps = map__kmaps(map);
 	struct kcore_mapfn_data md;
 	struct map *old_map, *new_map, *replacement_map = NULL, *next;
 	struct machine *machine;
@@ -1295,14 +1293,14 @@
 	}
 
 	/* Remove old maps */
-	map_groups__for_each_entry_safe(kmaps, old_map, next) {
+	maps__for_each_entry_safe(kmaps, old_map, next) {
 		/*
 		 * We need to preserve eBPF maps even if they are
 		 * covered by kcore, because we need to access
 		 * eBPF dso for source data.
 		 */
 		if (old_map != map && !__map__is_bpf_prog(old_map))
-			map_groups__remove(kmaps, old_map);
+			maps__remove(kmaps, old_map);
 	}
 	machine->trampolines_mapped = false;
 
@@ -1331,8 +1329,8 @@
 			map->unmap_ip	= new_map->unmap_ip;
 			/* Ensure maps are correctly ordered */
 			map__get(map);
-			map_groups__remove(kmaps, map);
-			map_groups__insert(kmaps, map);
+			maps__remove(kmaps, map);
+			maps__insert(kmaps, map);
 			map__put(map);
 			map__put(new_map);
 		} else {
@@ -1341,7 +1339,7 @@
 			 * and ensure that current maps (eBPF)
 			 * stay intact.
 			 */
-			if (map_groups__merge_in(kmaps, new_map))
+			if (maps__merge_in(kmaps, new_map))
 				goto out_err;
 		}
 	}
@@ -1433,9 +1431,9 @@
 		dso->symtab_type = DSO_BINARY_TYPE__KALLSYMS;
 
 	if (!no_kcore && !dso__load_kcore(dso, map, filename))
-		return map_groups__split_kallsyms_for_kcore(kmap->kmaps, dso);
+		return maps__split_kallsyms_for_kcore(kmap->kmaps, dso);
 	else
-		return map_groups__split_kallsyms(kmap->kmaps, dso, delta, map);
+		return maps__split_kallsyms(kmap->kmaps, dso, delta, map);
 }
 
 int dso__load_kallsyms(struct dso *dso, const char *filename,
@@ -1772,68 +1770,67 @@
 	return strcmp(name, map->dso->short_name);
 }
 
-void __map_groups__sort_by_name(struct map_groups *mg)
+void __maps__sort_by_name(struct maps *maps)
 {
-	qsort(mg->maps_by_name, mg->nr_maps, sizeof(struct map *), map__strcmp);
+	qsort(maps->maps_by_name, maps->nr_maps, sizeof(struct map *), map__strcmp);
 }
 
-static int map__groups__sort_by_name_from_rbtree(struct map_groups *mg)
+static int map__groups__sort_by_name_from_rbtree(struct maps *maps)
 {
 	struct map *map;
-	struct map **maps_by_name = realloc(mg->maps_by_name, mg->nr_maps * sizeof(map));
+	struct map **maps_by_name = realloc(maps->maps_by_name, maps->nr_maps * sizeof(map));
 	int i = 0;
 
 	if (maps_by_name == NULL)
 		return -1;
 
-	mg->maps_by_name = maps_by_name;
-	mg->nr_maps_allocated = mg->nr_maps;
+	maps->maps_by_name = maps_by_name;
+	maps->nr_maps_allocated = maps->nr_maps;
 
-	maps__for_each_entry(&mg->maps, map)
+	maps__for_each_entry(maps, map)
 		maps_by_name[i++] = map;
 
-	__map_groups__sort_by_name(mg);
+	__maps__sort_by_name(maps);
 	return 0;
 }
 
-static struct map *__map_groups__find_by_name(struct map_groups *mg, const char *name)
+static struct map *__maps__find_by_name(struct maps *maps, const char *name)
 {
 	struct map **mapp;
 
-	if (mg->maps_by_name == NULL &&
-	    map__groups__sort_by_name_from_rbtree(mg))
+	if (maps->maps_by_name == NULL &&
+	    map__groups__sort_by_name_from_rbtree(maps))
 		return NULL;
 
-	mapp = bsearch(name, mg->maps_by_name, mg->nr_maps, sizeof(*mapp), map__strcmp_name);
+	mapp = bsearch(name, maps->maps_by_name, maps->nr_maps, sizeof(*mapp), map__strcmp_name);
 	if (mapp)
 		return *mapp;
 	return NULL;
 }
 
-struct map *map_groups__find_by_name(struct map_groups *mg, const char *name)
+struct map *maps__find_by_name(struct maps *maps, const char *name)
 {
-	struct maps *maps = &mg->maps;
 	struct map *map;
 
 	down_read(&maps->lock);
 
-	if (mg->last_search_by_name && strcmp(mg->last_search_by_name->dso->short_name, name) == 0) {
-		map = mg->last_search_by_name;
+	if (maps->last_search_by_name && strcmp(maps->last_search_by_name->dso->short_name, name) == 0) {
+		map = maps->last_search_by_name;
 		goto out_unlock;
 	}
 	/*
-	 * If we have mg->maps_by_name, then the name isn't in the rbtree,
-	 * as mg->maps_by_name mirrors the rbtree when lookups by name are
+	 * If we have maps->maps_by_name, then the name isn't in the rbtree,
+	 * as maps->maps_by_name mirrors the rbtree when lookups by name are
 	 * made.
 	 */
-	map = __map_groups__find_by_name(mg, name);
-	if (map || mg->maps_by_name != NULL)
+	map = __maps__find_by_name(maps, name);
+	if (map || maps->maps_by_name != NULL)
 		goto out_unlock;
 
 	/* Fallback to traversing the rbtree... */
 	maps__for_each_entry(maps, map)
 		if (strcmp(map->dso->short_name, name) == 0) {
-			mg->last_search_by_name = map;
+			maps->last_search_by_name = map;
 			goto out_unlock;
 		}
 
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index 0b718cc..93fc43d 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -21,7 +21,7 @@
 
 struct dso;
 struct map;
-struct map_groups;
+struct maps;
 struct option;
 
 /*
@@ -108,7 +108,7 @@
 
 struct addr_location {
 	struct thread *thread;
-	struct map_groups *mg;
+	struct maps   *maps;
 	struct map    *map;
 	struct symbol *sym;
 	const char    *srcline;
@@ -186,7 +186,7 @@
 void symbols__insert(struct rb_root_cached *symbols, struct symbol *sym);
 void symbols__fixup_duplicate(struct rb_root_cached *symbols);
 void symbols__fixup_end(struct rb_root_cached *symbols);
-void map_groups__fixup_end(struct map_groups *mg);
+void maps__fixup_end(struct maps *maps);
 
 typedef int (*mapfn_t)(u64 start, u64 len, u64 pgoff, void *data);
 int file__read_maps(int fd, bool exe, mapfn_t mapfn, void *data,
diff --git a/tools/perf/util/synthetic-events.c b/tools/perf/util/synthetic-events.c
index 48c3f8b..c423298 100644
--- a/tools/perf/util/synthetic-events.c
+++ b/tools/perf/util/synthetic-events.c
@@ -493,7 +493,7 @@
 
 		/*
 		 * send mmap only for thread group leader
-		 * see thread__init_map_groups
+		 * see thread__init_maps()
 		 */
 		if (pid == tgid &&
 		    perf_event__synthesize_mmap_events(tool, mmap_event, pid, tgid,
diff --git a/tools/perf/util/thread-stack.c b/tools/perf/util/thread-stack.c
index cd8a948..0885967 100644
--- a/tools/perf/util/thread-stack.c
+++ b/tools/perf/util/thread-stack.c
@@ -134,8 +134,8 @@
 	if (err)
 		return err;
 
-	if (thread->mg && thread->mg->machine) {
-		struct machine *machine = thread->mg->machine;
+	if (thread->maps && thread->maps->machine) {
+		struct machine *machine = thread->maps->machine;
 		const char *arch = perf_env__arch(machine->env);
 
 		ts->kernel_start = machine__kernel_start(machine);
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c
index 0a277a9..28b7193 100644
--- a/tools/perf/util/thread.c
+++ b/tools/perf/util/thread.c
@@ -19,21 +19,21 @@
 
 #include <api/fs/fs.h>
 
-int thread__init_map_groups(struct thread *thread, struct machine *machine)
+int thread__init_maps(struct thread *thread, struct machine *machine)
 {
 	pid_t pid = thread->pid_;
 
 	if (pid == thread->tid || pid == -1) {
-		thread->mg = map_groups__new(machine);
+		thread->maps = maps__new(machine);
 	} else {
 		struct thread *leader = __machine__findnew_thread(machine, pid, pid);
 		if (leader) {
-			thread->mg = map_groups__get(leader->mg);
+			thread->maps = maps__get(leader->maps);
 			thread__put(leader);
 		}
 	}
 
-	return thread->mg ? 0 : -1;
+	return thread->maps ? 0 : -1;
 }
 
 struct thread *thread__new(pid_t pid, pid_t tid)
@@ -86,9 +86,9 @@
 
 	thread_stack__free(thread);
 
-	if (thread->mg) {
-		map_groups__put(thread->mg);
-		thread->mg = NULL;
+	if (thread->maps) {
+		maps__put(thread->maps);
+		thread->maps = NULL;
 	}
 	down_write(&thread->namespaces_lock);
 	list_for_each_entry_safe(namespaces, tmp_namespaces,
@@ -251,7 +251,7 @@
 		list_add(&new->list, &thread->comm_list);
 
 		if (exec)
-			unwind__flush_access(thread->mg);
+			unwind__flush_access(thread->maps);
 	}
 
 	thread->comm_set = true;
@@ -324,19 +324,19 @@
 size_t thread__fprintf(struct thread *thread, FILE *fp)
 {
 	return fprintf(fp, "Thread %d %s\n", thread->tid, thread__comm_str(thread)) +
-	       map_groups__fprintf(thread->mg, fp);
+	       maps__fprintf(thread->maps, fp);
 }
 
 int thread__insert_map(struct thread *thread, struct map *map)
 {
 	int ret;
 
-	ret = unwind__prepare_access(thread->mg, map, NULL);
+	ret = unwind__prepare_access(thread->maps, map, NULL);
 	if (ret)
 		return ret;
 
-	map_groups__fixup_overlappings(thread->mg, map, stderr);
-	map_groups__insert(thread->mg, map);
+	maps__fixup_overlappings(thread->maps, map, stderr);
+	maps__insert(thread->maps, map);
 
 	return 0;
 }
@@ -345,13 +345,13 @@
 {
 	bool initialized = false;
 	int err = 0;
-	struct maps *maps = &thread->mg->maps;
+	struct maps *maps = thread->maps;
 	struct map *map;
 
 	down_read(&maps->lock);
 
 	maps__for_each_entry(maps, map) {
-		err = unwind__prepare_access(thread->mg, map, &initialized);
+		err = unwind__prepare_access(thread->maps, map, &initialized);
 		if (err || initialized)
 			break;
 	}
@@ -371,21 +371,19 @@
 	return err;
 }
 
-static int thread__clone_map_groups(struct thread *thread,
-				    struct thread *parent,
-				    bool do_maps_clone)
+static int thread__clone_maps(struct thread *thread, struct thread *parent, bool do_maps_clone)
 {
 	/* This is new thread, we share map groups for process. */
 	if (thread->pid_ == parent->pid_)
 		return thread__prepare_access(thread);
 
-	if (thread->mg == parent->mg) {
+	if (thread->maps == parent->maps) {
 		pr_debug("broken map groups on thread %d/%d parent %d/%d\n",
 			 thread->pid_, thread->tid, parent->pid_, parent->tid);
 		return 0;
 	}
 	/* But this one is new process, copy maps. */
-	return do_maps_clone ? map_groups__clone(thread, parent->mg) : 0;
+	return do_maps_clone ? maps__clone(thread, parent->maps) : 0;
 }
 
 int thread__fork(struct thread *thread, struct thread *parent, u64 timestamp, bool do_maps_clone)
@@ -401,7 +399,7 @@
 	}
 
 	thread->ppid = parent->tid;
-	return thread__clone_map_groups(thread, parent, do_maps_clone);
+	return thread__clone_maps(thread, parent, do_maps_clone);
 }
 
 void thread__find_cpumode_addr_location(struct thread *thread, u64 addr,
diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h
index 51bdb9a..20b96b5 100644
--- a/tools/perf/util/thread.h
+++ b/tools/perf/util/thread.h
@@ -25,7 +25,7 @@
 		struct rb_node	 rb_node;
 		struct list_head node;
 	};
-	struct map_groups	*mg;
+	struct maps		*maps;
 	pid_t			pid_; /* Not all tools update this */
 	pid_t			tid;
 	pid_t			ppid;
@@ -53,7 +53,7 @@
 struct comm;
 
 struct thread *thread__new(pid_t pid, pid_t tid);
-int thread__init_map_groups(struct thread *thread, struct machine *machine);
+int thread__init_maps(struct thread *thread, struct machine *machine);
 void thread__delete(struct thread *thread);
 
 struct thread *thread__get(struct thread *thread);
diff --git a/tools/perf/util/unwind-libdw.c b/tools/perf/util/unwind-libdw.c
index d2a8df0..7a3dbc2 100644
--- a/tools/perf/util/unwind-libdw.c
+++ b/tools/perf/util/unwind-libdw.c
@@ -81,7 +81,7 @@
 		return -1;
 
 	e->ip	  = ip;
-	e->ms.mg  = al.mg;
+	e->ms.maps = al.maps;
 	e->ms.map = al.map;
 	e->ms.sym = al.sym;
 
@@ -200,7 +200,7 @@
 	struct unwind_info *ui, ui_buf = {
 		.sample		= data,
 		.thread		= thread,
-		.machine	= thread->mg->machine,
+		.machine	= thread->maps->machine,
 		.cb		= cb,
 		.arg		= arg,
 		.max_stack	= max_stack,
diff --git a/tools/perf/util/unwind-libunwind-local.c b/tools/perf/util/unwind-libunwind-local.c
index 6d53347..b4649f5 100644
--- a/tools/perf/util/unwind-libunwind-local.c
+++ b/tools/perf/util/unwind-libunwind-local.c
@@ -578,7 +578,7 @@
 	e.ms.sym = thread__find_symbol(thread, PERF_RECORD_MISC_USER, ip, &al);
 	e.ip     = ip;
 	e.ms.map = al.map;
-	e.ms.mg  = al.mg;
+	e.ms.maps = al.maps;
 
 	pr_debug("unwind: %s:ip = 0x%" PRIx64 " (0x%" PRIx64 ")\n",
 		 al.sym ? al.sym->name : "''",
@@ -616,26 +616,26 @@
 	.get_proc_name		= get_proc_name,
 };
 
-static int _unwind__prepare_access(struct map_groups *mg)
+static int _unwind__prepare_access(struct maps *maps)
 {
-	mg->addr_space = unw_create_addr_space(&accessors, 0);
-	if (!mg->addr_space) {
+	maps->addr_space = unw_create_addr_space(&accessors, 0);
+	if (!maps->addr_space) {
 		pr_err("unwind: Can't create unwind address space.\n");
 		return -ENOMEM;
 	}
 
-	unw_set_caching_policy(mg->addr_space, UNW_CACHE_GLOBAL);
+	unw_set_caching_policy(maps->addr_space, UNW_CACHE_GLOBAL);
 	return 0;
 }
 
-static void _unwind__flush_access(struct map_groups *mg)
+static void _unwind__flush_access(struct maps *maps)
 {
-	unw_flush_cache(mg->addr_space, 0, 0);
+	unw_flush_cache(maps->addr_space, 0, 0);
 }
 
-static void _unwind__finish_access(struct map_groups *mg)
+static void _unwind__finish_access(struct maps *maps)
 {
-	unw_destroy_addr_space(mg->addr_space);
+	unw_destroy_addr_space(maps->addr_space);
 }
 
 static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,
@@ -660,7 +660,7 @@
 	 */
 	if (max_stack - 1 > 0) {
 		WARN_ONCE(!ui->thread, "WARNING: ui->thread is NULL");
-		addr_space = ui->thread->mg->addr_space;
+		addr_space = ui->thread->maps->addr_space;
 
 		if (addr_space == NULL)
 			return -1;
@@ -709,7 +709,7 @@
 	struct unwind_info ui = {
 		.sample       = data,
 		.thread       = thread,
-		.machine      = thread->mg->machine,
+		.machine      = thread->maps->machine,
 	};
 
 	if (!data->user_regs.regs)
diff --git a/tools/perf/util/unwind-libunwind.c b/tools/perf/util/unwind-libunwind.c
index a24fb57..e89a547 100644
--- a/tools/perf/util/unwind-libunwind.c
+++ b/tools/perf/util/unwind-libunwind.c
@@ -12,14 +12,12 @@
 struct unwind_libunwind_ops __weak *x86_32_unwind_libunwind_ops;
 struct unwind_libunwind_ops __weak *arm64_unwind_libunwind_ops;
 
-static void unwind__register_ops(struct map_groups *mg,
-			  struct unwind_libunwind_ops *ops)
+static void unwind__register_ops(struct maps *maps, struct unwind_libunwind_ops *ops)
 {
-	mg->unwind_libunwind_ops = ops;
+	maps->unwind_libunwind_ops = ops;
 }
 
-int unwind__prepare_access(struct map_groups *mg, struct map *map,
-			   bool *initialized)
+int unwind__prepare_access(struct maps *maps, struct map *map, bool *initialized)
 {
 	const char *arch;
 	enum dso_type dso_type;
@@ -29,7 +27,7 @@
 	if (!dwarf_callchain_users)
 		return 0;
 
-	if (mg->addr_space) {
+	if (maps->addr_space) {
 		pr_debug("unwind: thread map already set, dso=%s\n",
 			 map->dso->name);
 		if (initialized)
@@ -38,14 +36,14 @@
 	}
 
 	/* env->arch is NULL for live-mode (i.e. perf top) */
-	if (!mg->machine->env || !mg->machine->env->arch)
+	if (!maps->machine->env || !maps->machine->env->arch)
 		goto out_register;
 
-	dso_type = dso__type(map->dso, mg->machine);
+	dso_type = dso__type(map->dso, maps->machine);
 	if (dso_type == DSO__TYPE_UNKNOWN)
 		return 0;
 
-	arch = perf_env__arch(mg->machine->env);
+	arch = perf_env__arch(maps->machine->env);
 
 	if (!strcmp(arch, "x86")) {
 		if (dso_type != DSO__TYPE_64BIT)
@@ -60,31 +58,31 @@
 		return 0;
 	}
 out_register:
-	unwind__register_ops(mg, ops);
+	unwind__register_ops(maps, ops);
 
-	err = mg->unwind_libunwind_ops->prepare_access(mg);
+	err = maps->unwind_libunwind_ops->prepare_access(maps);
 	if (initialized)
 		*initialized = err ? false : true;
 	return err;
 }
 
-void unwind__flush_access(struct map_groups *mg)
+void unwind__flush_access(struct maps *maps)
 {
-	if (mg->unwind_libunwind_ops)
-		mg->unwind_libunwind_ops->flush_access(mg);
+	if (maps->unwind_libunwind_ops)
+		maps->unwind_libunwind_ops->flush_access(maps);
 }
 
-void unwind__finish_access(struct map_groups *mg)
+void unwind__finish_access(struct maps *maps)
 {
-	if (mg->unwind_libunwind_ops)
-		mg->unwind_libunwind_ops->finish_access(mg);
+	if (maps->unwind_libunwind_ops)
+		maps->unwind_libunwind_ops->finish_access(maps);
 }
 
 int unwind__get_entries(unwind_entry_cb_t cb, void *arg,
 			 struct thread *thread,
 			 struct perf_sample *data, int max_stack)
 {
-	if (thread->mg->unwind_libunwind_ops)
-		return thread->mg->unwind_libunwind_ops->get_entries(cb, arg, thread, data, max_stack);
+	if (thread->maps->unwind_libunwind_ops)
+		return thread->maps->unwind_libunwind_ops->get_entries(cb, arg, thread, data, max_stack);
 	return 0;
 }
diff --git a/tools/perf/util/unwind.h b/tools/perf/util/unwind.h
index 50337c9..ab8ad46 100644
--- a/tools/perf/util/unwind.h
+++ b/tools/perf/util/unwind.h
@@ -6,7 +6,7 @@
 #include <linux/types.h>
 #include "util/map_symbol.h"
 
-struct map_groups;
+struct maps;
 struct perf_sample;
 struct thread;
 
@@ -18,9 +18,9 @@
 typedef int (*unwind_entry_cb_t)(struct unwind_entry *entry, void *arg);
 
 struct unwind_libunwind_ops {
-	int (*prepare_access)(struct map_groups *mg);
-	void (*flush_access)(struct map_groups *mg);
-	void (*finish_access)(struct map_groups *mg);
+	int (*prepare_access)(struct maps *maps);
+	void (*flush_access)(struct maps *maps);
+	void (*finish_access)(struct maps *maps);
 	int (*get_entries)(unwind_entry_cb_t cb, void *arg,
 			   struct thread *thread,
 			   struct perf_sample *data, int max_stack);
@@ -45,20 +45,19 @@
 #endif
 
 int LIBUNWIND__ARCH_REG_ID(int regnum);
-int unwind__prepare_access(struct map_groups *mg, struct map *map,
-			   bool *initialized);
-void unwind__flush_access(struct map_groups *mg);
-void unwind__finish_access(struct map_groups *mg);
+int unwind__prepare_access(struct maps *maps, struct map *map, bool *initialized);
+void unwind__flush_access(struct maps *maps);
+void unwind__finish_access(struct maps *maps);
 #else
-static inline int unwind__prepare_access(struct map_groups *mg __maybe_unused,
+static inline int unwind__prepare_access(struct maps *maps __maybe_unused,
 					 struct map *map __maybe_unused,
 					 bool *initialized __maybe_unused)
 {
 	return 0;
 }
 
-static inline void unwind__flush_access(struct map_groups *mg __maybe_unused) {}
-static inline void unwind__finish_access(struct map_groups *mg __maybe_unused) {}
+static inline void unwind__flush_access(struct maps *maps __maybe_unused) {}
+static inline void unwind__finish_access(struct maps *maps __maybe_unused) {}
 #endif
 #else
 static inline int
@@ -71,14 +70,14 @@
 	return 0;
 }
 
-static inline int unwind__prepare_access(struct map_groups *mg __maybe_unused,
+static inline int unwind__prepare_access(struct maps *maps __maybe_unused,
 					 struct map *map __maybe_unused,
 					 bool *initialized __maybe_unused)
 {
 	return 0;
 }
 
-static inline void unwind__flush_access(struct map_groups *mg __maybe_unused) {}
-static inline void unwind__finish_access(struct map_groups *mg __maybe_unused) {}
+static inline void unwind__flush_access(struct maps *maps __maybe_unused) {}
+static inline void unwind__finish_access(struct maps *maps __maybe_unused) {}
 #endif /* HAVE_DWARF_UNWIND_SUPPORT */
 #endif /* __UNWIND_H */
diff --git a/tools/perf/util/vdso.c b/tools/perf/util/vdso.c
index 6e00793..3cc91ad 100644
--- a/tools/perf/util/vdso.c
+++ b/tools/perf/util/vdso.c
@@ -144,7 +144,7 @@
 	enum dso_type dso_type = DSO__TYPE_UNKNOWN;
 	struct map *map;
 
-	map_groups__for_each_entry(thread->mg, map) {
+	maps__for_each_entry(thread->maps, map) {
 		struct dso *dso = map->dso;
 		if (!dso || dso->long_name[0] != '/')
 			continue;