| 1 | .. _Arm Emulation: |
| 2 | |
| 3 | A-profile CPU architecture support |
| 4 | ================================== |
| 5 | |
| 6 | QEMU's TCG emulation includes support for the Armv5, Armv6, Armv7, |
| 7 | Armv8 and Armv9 versions of the A-profile architecture. It also has support for |
| 8 | the following architecture extensions: |
| 9 | |
| 10 | - FEAT_AA32 (PE Support for AArch32) |
| 11 | - FEAT_AA32BF16 (AArch32 BFloat16 instructions) |
| 12 | - FEAT_AA32EL0 (Support for AArch32 at EL0) |
| 13 | - FEAT_AA32EL1 (Support for AArch32 at EL1) |
| 14 | - FEAT_AA32EL2 (Support for AArch32 at EL2) |
| 15 | - FEAT_AA32EL3 (Support for AArch32 at EL3) |
| 16 | - FEAT_AA32HPD (AArch32 hierarchical permission disables) |
| 17 | - FEAT_AA32I8MM (AArch32 Int8 matrix multiplication instructions) |
| 18 | - FEAT_AA64 (PE uses AArch64 after last reboot) |
| 19 | - FEAT_AA64EL0 (Support for AArch64 at EL0) |
| 20 | - FEAT_AA64EL1 (Support for AArch64 at EL1) |
| 21 | - FEAT_AA64EL2 (Support for AArch64 at EL2) |
| 22 | - FEAT_AA64EL3 (Support for AArch64 at EL3) |
| 23 | - FEAT_AdvSIMD (Advanced SIMD Extension) |
| 24 | - FEAT_AES (AESD and AESE instructions) |
| 25 | - FEAT_AFP (Alternate floating-point behavior) |
| 26 | - FEAT_AIE (Memory Attribute Index Enhancement) |
| 27 | - FEAT_Armv9_Crypto (Armv9 Cryptographic Extension) |
| 28 | - FEAT_ASID16 (16 bit ASID) |
| 29 | - FEAT_ASID2 (Concurrent use of two ASIDs) |
| 30 | - FEAT_ATS1A (Address Translation operations that ignore stage 1 permissions) |
| 31 | - FEAT_BBML1 (Translation table break-before-make level 1) |
| 32 | - FEAT_BBML2 (Translation table break-before-make level 2) |
| 33 | - FEAT_BF16 (AArch64 BFloat16 instructions) |
| 34 | - FEAT_BigEnd (Support for big-endian at EL1 and above) |
| 35 | - FEAT_BigEndEL0 (Support for big-endian at EL0) |
| 36 | - FEAT_BTI (Branch Target Identification) |
| 37 | - FEAT_CCIDX (Extended cache index) |
| 38 | - FEAT_CHK (Check Feature Status) |
| 39 | - FEAT_CMOW (Control for cache maintenance permission) |
| 40 | - FEAT_CMPBR (Compare and Branch instructions) |
| 41 | - FEAT_CRC32 (CRC32 instructions) |
| 42 | - FEAT_Crypto (Cryptographic Extension) |
| 43 | - FEAT_CSSC (Common Short Sequence Compression instructions) |
| 44 | - FEAT_CSV2 (Cache speculation variant 2) |
| 45 | - FEAT_CSV2_1p1 (Cache speculation variant 2, version 1.1) |
| 46 | - FEAT_CSV2_1p2 (Cache speculation variant 2, version 1.2) |
| 47 | - FEAT_CSV2_2 (Cache speculation variant 2, version 2) |
| 48 | - FEAT_CSV2_3 (Cache speculation variant 2, version 3) |
| 49 | - FEAT_CSV3 (Cache speculation variant 3) |
| 50 | - FEAT_DGH (Data gathering hint) |
| 51 | - FEAT_DIT (Data Independent Timing instructions) |
| 52 | - FEAT_DoubleLock (Double Lock) |
| 53 | - FEAT_DPB (DC CVAP instruction) |
| 54 | - FEAT_DPB2 (DC CVADP instruction) |
| 55 | - FEAT_Debugv8p1 (Debug with VHE) |
| 56 | - FEAT_Debugv8p2 (Debug changes for v8.2) |
| 57 | - FEAT_Debugv8p4 (Debug changes for v8.4) |
| 58 | - FEAT_Debugv8p8 (Debug changes for v8.8) |
| 59 | - FEAT_DotProd (Advanced SIMD dot product instructions) |
| 60 | - FEAT_DoubleFault (Double Fault Extension) |
| 61 | - FEAT_E0PD (Preventing EL0 access to halves of address maps) |
| 62 | - FEAT_E2H0 (Programming of HCR_EL2.E2H) |
| 63 | - FEAT_EBF16 (AArch64 Extended BFloat16 instructions) |
| 64 | - FEAT_ECV (Enhanced Counter Virtualization) |
| 65 | - FEAT_ECV_POFF (Enhanced Counter Virtualization Physical Offset) |
| 66 | - FEAT_EL0 (Support for execution at EL0) |
| 67 | - FEAT_EL1 (Support for execution at EL1) |
| 68 | - FEAT_EL2 (Support for execution at EL2) |
| 69 | - FEAT_EL3 (Support for execution at EL3) |
| 70 | - FEAT_EPAC (Enhanced pointer authentication) |
| 71 | - FEAT_ETS2 (Enhanced Translation Synchronization) |
| 72 | - FEAT_EVT (Enhanced Virtualization Traps) |
| 73 | - FEAT_EVT2 (Enhanced Virtualization Traps 2) |
| 74 | - FEAT_F8F16MM (8-bit floating-point matrix multiply-accumulate to half-precision) |
| 75 | - FEAT_F8F32MM (8-bit floating-point matrix multiply-accumulate to single-precision) |
| 76 | - FEAT_F32MM (Single-precision Matrix Multiplication) |
| 77 | - FEAT_F64MM (Double-precision Matrix Multiplication) |
| 78 | - FEAT_FAMINMAX (Floating-point maximum and minimum absolute value instructions) |
| 79 | - FEAT_FCMA (Floating-point complex number instructions) |
| 80 | - FEAT_FGT (Fine-Grained Traps) |
| 81 | - FEAT_FGWTE3 (Fine-Grained Write Trap EL3) |
| 82 | - FEAT_FHM (Floating-point half-precision multiplication instructions) |
| 83 | - FEAT_FP (Floating Point extensions) |
| 84 | - FEAT_FP16 (Half-precision floating-point data processing) |
| 85 | - FEAT_FP8 (FP8 convert instructions) |
| 86 | - FEAT_FP8DOT2 (FP8 2-way dot product to half-precision instructions) |
| 87 | - FEAT_FP8DOT4 (FP8 4-way dot product to single-precision instructions) |
| 88 | - FEAT_FP8FMA (FP8 multiply-accumulate to half-precision and single-precision instructions) |
| 89 | - FEAT_FPAC (Faulting on AUT* instructions) |
| 90 | - FEAT_FPACCOMBINE (Faulting on combined pointer authentication instructions) |
| 91 | - FEAT_FPACC_SPEC (Speculative behavior of combined pointer authentication instructions) |
| 92 | - FEAT_FPMR (Floating-point Mode Register) |
| 93 | - FEAT_FPRCVT (Floating-Point to/from Integer in Scalar FP register) |
| 94 | - FEAT_FRINTTS (Floating-point to integer instructions) |
| 95 | - FEAT_FlagM (Flag manipulation instructions v2) |
| 96 | - FEAT_FlagM2 (Enhancements to flag manipulation instructions) |
| 97 | - FEAT_GCS (Guarded Control Stack Extension) |
| 98 | - FEAT_GTG (Guest translation granule size) |
| 99 | - FEAT_HAF (Hardware management of the Access flag) |
| 100 | - FEAT_HAFDBS (Hardware management of the access flag and dirty bit state) |
| 101 | - FEAT_HBC (Hinted conditional branches) |
| 102 | - FEAT_HCX (Support for the HCRX_EL2 register) |
| 103 | - FEAT_HPDS (Hierarchical permission disables) |
| 104 | - FEAT_HPDS2 (Translation table page-based hardware attributes) |
| 105 | - FEAT_HPMN0 (Setting of MDCR_EL2.HPMN to zero) |
| 106 | - FEAT_I8MM (AArch64 Int8 matrix multiplication instructions) |
| 107 | - FEAT_IDST (ID space trap handling) |
| 108 | - FEAT_IESB (Implicit error synchronization event) |
| 109 | - FEAT_IVIPT (The IVIPT Extension) |
| 110 | - FEAT_JSCVT (JavaScript conversion instructions) |
| 111 | - FEAT_LittleEnd (Support for little-endian at EL1 and above) |
| 112 | - FEAT_LittleEndEL0 (Support for little-endian at EL0) |
| 113 | - FEAT_LOR (Limited ordering regions) |
| 114 | - FEAT_LPA (Large Physical Address space) |
| 115 | - FEAT_LPA2 (Large Physical and virtual Address space v2) |
| 116 | - FEAT_LRCPC (Load-acquire RCpc instructions) |
| 117 | - FEAT_LRCPC2 (Load-acquire RCpc instructions v2) |
| 118 | - FEAT_LSE (Large System Extensions) |
| 119 | - FEAT_LSE2 (Large System Extensions v2) |
| 120 | - FEAT_LSE128 (128-bit Atomics) |
| 121 | - FEAT_LUT (Lookup table instructions with 2-bit and 4-bit indices) |
| 122 | - FEAT_LVA (Large Virtual Address space) |
| 123 | - FEAT_MEC (Memory Encryption Contexts) |
| 124 | |
| 125 | * This is a register-only implementation without encryption. |
| 126 | - FEAT_MixedEnd (Mixed-endian support) |
| 127 | - FEAT_MixedEndEL0 (Mixed-endian support at EL0) |
| 128 | - FEAT_MOPS (Standardization of memory operations) |
| 129 | - FEAT_MTE (Memory Tagging Extension) |
| 130 | - FEAT_MTE2 (Memory Tagging Extension) |
| 131 | - FEAT_MTE3 (MTE Asymmetric Fault Handling) |
| 132 | - FEAT_MTE4 (Enhanced Memory Tagging Extension) |
| 133 | - FEAT_MTE_ASYM_FAULT (Memory tagging asymmetric faults) |
| 134 | - FEAT_MTE_ASYNC (Asynchronous reporting of Tag Check Fault) |
| 135 | - FEAT_MTE_CANONICAL_TAGS (Canonical tag checking) |
| 136 | - FEAT_MTE_NO_ADDRESS_TAGS (Address tagging disabled) |
| 137 | - FEAT_MTE_PERM (NoTagAccess memory attribute) |
| 138 | - FEAT_MTE_STORE_ONLY (Store-only tag checking) |
| 139 | - FEAT_MTE_TAGGED_FAR (Full address reporting of Tag Check Fault) |
| 140 | - FEAT_NMI (Non-maskable Interrupt) |
| 141 | - FEAT_NV (Nested Virtualization) |
| 142 | - FEAT_NV2 (Enhanced nested virtualization support) |
| 143 | - FEAT_PACIMP (Pointer authentication - IMPLEMENTATION DEFINED algorithm) |
| 144 | - FEAT_PACQARMA3 (Pointer authentication - QARMA3 algorithm) |
| 145 | - FEAT_PACQARMA5 (Pointer authentication - QARMA5 algorithm) |
| 146 | - FEAT_PAN (Privileged access never) |
| 147 | - FEAT_PAN2 (AT S1E1R and AT S1E1W instruction variants affected by PSTATE.PAN) |
| 148 | - FEAT_PAN3 (Support for SCTLR_ELx.EPAN) |
| 149 | - FEAT_PAuth (Pointer authentication) |
| 150 | - FEAT_PAuth2 (Enhancements to pointer authentication) |
| 151 | - FEAT_PMULL (PMULL, PMULL2 instructions) |
| 152 | - FEAT_PMUv3 (PMU extension version 3) |
| 153 | - FEAT_PMUv3p1 (PMU Extensions v3.1) |
| 154 | - FEAT_PMUv3p4 (PMU Extensions v3.4) |
| 155 | - FEAT_PMUv3p5 (PMU Extensions v3.5) |
| 156 | - FEAT_RAS (Reliability, availability, and serviceability) |
| 157 | - FEAT_RASv1p1 (RAS Extension v1.1) |
| 158 | - FEAT_RDM (Advanced SIMD rounding double multiply accumulate instructions) |
| 159 | - FEAT_RME (Realm Management Extension) (NB: support status in QEMU is experimental) |
| 160 | - FEAT_RME_GPC2 (RME Granule Protection Check 2 Extension) |
| 161 | - FEAT_RNG (Random number generator) |
| 162 | - FEAT_RNG_TRAP (Trapping support for RNDR/RNDRRS) |
| 163 | - FEAT_RPRES (Increased precision of FRECPE and FRSQRTE) |
| 164 | - FEAT_S1PIE (Stage 1 permission indirections) |
| 165 | - FEAT_S2FWB (Stage 2 forced Write-Back) |
| 166 | - FEAT_S2PIE (Stage 2 permission indirections) |
| 167 | - FEAT_S2TGran16K (Support for 16KB memory translation granule size at stage 2) |
| 168 | - FEAT_S2TGran4K (Support for 4KB memory translation granule size at stage 2) |
| 169 | - FEAT_S2TGran64K (Support for 64KB memory translation granule size at stage 2) |
| 170 | - FEAT_SB (Speculation Barrier) |
| 171 | - FEAT_SCTLR2 (Extension to SCTLR_ELx) |
| 172 | - FEAT_Secure (Support for Secure state) |
| 173 | - FEAT_SEL2 (Secure EL2) |
| 174 | - FEAT_SHA1 (SHA1 instructions) |
| 175 | - FEAT_SHA256 (SHA256 instructions) |
| 176 | - FEAT_SHA3 (Advanced SIMD SHA3 instructions) |
| 177 | - FEAT_SHA512 (Advanced SIMD SHA512 instructions) |
| 178 | - FEAT_SM3 (Advanced SIMD SM3 instructions) |
| 179 | - FEAT_SM4 (Advanced SIMD SM4 instructions) |
| 180 | - FEAT_SME (Scalable Matrix Extension) |
| 181 | - FEAT_SME2 (Scalable Matrix Extension version 2) |
| 182 | - FEAT_SME2p1 (Scalable Matrix Extension version 2.1) |
| 183 | - FEAT_SME2p2 (Scalable Matrix Extension version 2.2) |
| 184 | - FEAT_SME_B16B16 (Non-widening BFloat16 arithmetic for SME2) |
| 185 | - FEAT_SME_FA64 (Full A64 instruction set in Streaming SVE mode) |
| 186 | - FEAT_SME_F16F16 (Non-widening half-precision FP16 arithmetic for SME2) |
| 187 | - FEAT_SME_F64F64 (Double-precision floating-point outer product instructions) |
| 188 | - FEAT_SME_F8F16 (SME2 ZA-targeting FP8 multiply-accumulate, dot product, and outer product to half-precision instructions) |
| 189 | - FEAT_SME_F8F32 (SME2 ZA-targeting FP8 multiply-accumulate, dot product, and outer product to single-precision instructions) |
| 190 | - FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions) |
| 191 | - FEAT_SME_LUTv2 (Lookup table instructions with 4-bit indices and 8-bit elements) |
| 192 | - FEAT_SME_MOP4 (Quarter-tile outer product instructions) |
| 193 | - FEAT_SME_TMOP (Structured sparsity outer product instructions) |
| 194 | - FEAT_SSVE_AES (Streaming SVE Mode Advanced Encryption Standard and 128-bit polynomial multiply long instructions) |
| 195 | - FEAT_SSVE_BitPerm (Streaming Scalable Vector Bit Permutes instructions) |
| 196 | - FEAT_SSVE_FEXPA (Streaming FEXPA instruction) |
| 197 | - FEAT_SSVE_FP8DOT2 (SVE2 FP8 2-way dot product to half-precision instructions in Streaming SVE mode) |
| 198 | - FEAT_SSVE_FP8DOT4 (SVE2 FP8 4-way dot product to single-precision instructions in Streaming SVE mode) |
| 199 | - FEAT_SSVE_FP8FMA (SVE2 FP8 multiply-accumulate to half-precision and single-precision instructions in Streaming SVE mode) |
| 200 | - FEAT_SVE (Scalable Vector Extension) |
| 201 | - FEAT_SVE_AES (Scalable Vector AES instructions) |
| 202 | - FEAT_SVE_AES2 (SVE multi-vector Advanced Encryption Standard and 128-bit polynomial multiply long instructions) |
| 203 | - FEAT_SVE_B16B16 (Non-widening BFloat16 arithmetic for SVE2) |
| 204 | - FEAT_SVE_BFSCALE (BFloat16 Floating-Point Adjust Exponent) |
| 205 | - FEAT_SVE_BitPerm (Scalable Vector Bit Permutes instructions) |
| 206 | - FEAT_SVE_PMULL128 (Scalable Vector PMULL instructions) |
| 207 | - FEAT_SVE_SHA3 (Scalable Vector SHA3 instructions) |
| 208 | - FEAT_SVE_SM4 (Scalable Vector SM4 instructions) |
| 209 | - FEAT_SVE2 (Scalable Vector Extension version 2) |
| 210 | - FEAT_SVE2p1 (Scalable Vector Extension version 2.1) |
| 211 | - FEAT_SVE2p2 (Scalable Vector Extension version 2.2) |
| 212 | - FEAT_SPECRES (Speculation restriction instructions) |
| 213 | - FEAT_SSBS (Speculative Store Bypass Safe) |
| 214 | - FEAT_SSBS2 (MRS and MSR instructions for SSBS version 2) |
| 215 | - FEAT_TCR2 (Support for TCR2_ELx) |
| 216 | - FEAT_TGran16K (Support for 16KB memory translation granule size at stage 1) |
| 217 | - FEAT_TGran4K (Support for 4KB memory translation granule size at stage 1) |
| 218 | - FEAT_TGran64K (Support for 64KB memory translation granule size at stage 1) |
| 219 | - FEAT_TIDCP1 (EL0 use of IMPLEMENTATION DEFINED functionality) |
| 220 | - FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain) |
| 221 | - FEAT_TLBIRANGE (TLB invalidate range instructions) |
| 222 | - FEAT_TTCNP (Translation table Common not private translations) |
| 223 | - FEAT_TTL (Translation Table Level) |
| 224 | - FEAT_TTST (Small translation tables) |
| 225 | - FEAT_UAO (Unprivileged Access Override control) |
| 226 | - FEAT_VHE (Virtualization Host Extensions) |
| 227 | - FEAT_VMID16 (16-bit VMID) |
| 228 | - FEAT_WFxT (WFE and WFI instructions with timeout) |
| 229 | - FEAT_XNX (Translation table stage 2 Unprivileged Execute-never) |
| 230 | - FEAT_XS (XS attribute) |
| 231 | |
| 232 | For information on the specifics of these extensions, please refer |
| 233 | to the `Arm Architecture Reference Manual for A-profile architecture |
| 234 | <https://developer.arm.com/documentation/ddi0487/latest>`_. |
| 235 | |
| 236 | When a specific named CPU is being emulated, only those features which |
| 237 | are present in hardware for that CPU are emulated. (If a feature is |
| 238 | not in the list above then it is not supported, even if the real |
| 239 | hardware should have it.) |
| 240 | |
| 241 | The ``max-v8`` CPU enables all of the listed Armv8-A architecture |
| 242 | extensions but none of the Armv9-A architecture extensions. |
| 243 | |
| 244 | The ``max-v9`` CPU enables all of the listed extensions except for |
| 245 | those that are prohibited by the Armv9-A architecture: FEAT_AA32EL1, |
| 246 | FEAT_AA32EL2, FEAT_AA32EL3, and FEAT_DoubleLock. |
| 247 | |
| 248 | The ``max`` CPU is an alias for ``max-v9`` under AArch64 TCG emulation, |
| 249 | an alias for ``max-v8`` under AArch32 TCG emulation, or an alias |
| 250 | for ``host`` under hardware virtualization. |
| 251 | |
| 252 | R-profile CPU architecture support |
| 253 | ================================== |
| 254 | |
| 255 | QEMU's TCG emulation support for R-profile CPUs is currently limited. |
| 256 | We emulate only the Cortex-R5 and Cortex-R5F CPUs. |
| 257 | |
| 258 | M-profile CPU architecture support |
| 259 | ================================== |
| 260 | |
| 261 | QEMU's TCG emulation includes support for Armv6-M, Armv7-M, Armv8-M, and |
| 262 | Armv8.1-M versions of the M-profile architucture. It also has support |
| 263 | for the following architecture extensions: |
| 264 | |
| 265 | - FP (Floating-point Extension) |
| 266 | - FPCXT (FPCXT access instructions) |
| 267 | - HP (Half-precision floating-point instructions) |
| 268 | - LOB (Low Overhead loops and Branch future) |
| 269 | - M (Main Extension) |
| 270 | - MPU (Memory Protection Unit Extension) |
| 271 | - PXN (Privileged Execute Never) |
| 272 | - RAS (Reliability, Serviceability and Availability): "minimum RAS Extension" only |
| 273 | - S (Security Extension) |
| 274 | - ST (System Timer Extension) |
| 275 | |
| 276 | For information on the specifics of these extensions, please refer |
| 277 | to the `Armv8-M Arm Architecture Reference Manual |
| 278 | <https://developer.arm.com/documentation/ddi0553/latest>`_. |
| 279 | |
| 280 | When a specific named CPU is being emulated, only those features which |
| 281 | are present in hardware for that CPU are emulated. (If a feature is |
| 282 | not in the list above then it is not supported, even if the real |
| 283 | hardware should have it.) There is no equivalent of the ``max`` CPU for |
| 284 | M-profile. |