xref: /openbmc/qemu/docs/system/arm/emulation.rst (revision 5f69e42d)
1.. _Arm Emulation:
2
3A-profile CPU architecture support
4==================================
5
6QEMU's TCG emulation includes support for the Armv5, Armv6, Armv7 and
7Armv8 versions of the A-profile architecture. It also has support for
8the following architecture extensions:
9
10- FEAT_AA32BF16 (AArch32 BFloat16 instructions)
11- FEAT_AA32EL0 (Support for AArch32 at EL0)
12- FEAT_AA32EL1 (Support for AArch32 at EL1)
13- FEAT_AA32EL2 (Support for AArch32 at EL2)
14- FEAT_AA32EL3 (Support for AArch32 at EL3)
15- FEAT_AA32HPD (AArch32 hierarchical permission disables)
16- FEAT_AA32I8MM (AArch32 Int8 matrix multiplication instructions)
17- FEAT_AA64EL0 (Support for AArch64 at EL0)
18- FEAT_AA64EL1 (Support for AArch64 at EL1)
19- FEAT_AA64EL2 (Support for AArch64 at EL2)
20- FEAT_AA64EL3 (Support for AArch64 at EL3)
21- FEAT_AdvSIMD (Advanced SIMD Extension)
22- FEAT_AES (AESD and AESE instructions)
23- FEAT_Armv9_Crypto (Armv9 Cryptographic Extension)
24- FEAT_ASID16 (16 bit ASID)
25- FEAT_BBM at level 2 (Translation table break-before-make levels)
26- FEAT_BF16 (AArch64 BFloat16 instructions)
27- FEAT_BTI (Branch Target Identification)
28- FEAT_CCIDX (Extended cache index)
29- FEAT_CRC32 (CRC32 instructions)
30- FEAT_Crypto (Cryptographic Extension)
31- FEAT_CSV2 (Cache speculation variant 2)
32- FEAT_CSV2_1p1 (Cache speculation variant 2, version 1.1)
33- FEAT_CSV2_1p2 (Cache speculation variant 2, version 1.2)
34- FEAT_CSV2_2 (Cache speculation variant 2, version 2)
35- FEAT_CSV2_3 (Cache speculation variant 2, version 3)
36- FEAT_CSV3 (Cache speculation variant 3)
37- FEAT_DGH (Data gathering hint)
38- FEAT_DIT (Data Independent Timing instructions)
39- FEAT_DPB (DC CVAP instruction)
40- FEAT_DPB2 (DC CVADP instruction)
41- FEAT_Debugv8p1 (Debug with VHE)
42- FEAT_Debugv8p2 (Debug changes for v8.2)
43- FEAT_Debugv8p4 (Debug changes for v8.4)
44- FEAT_DotProd (Advanced SIMD dot product instructions)
45- FEAT_DoubleFault (Double Fault Extension)
46- FEAT_E0PD (Preventing EL0 access to halves of address maps)
47- FEAT_ECV (Enhanced Counter Virtualization)
48- FEAT_EL0 (Support for execution at EL0)
49- FEAT_EL1 (Support for execution at EL1)
50- FEAT_EL2 (Support for execution at EL2)
51- FEAT_EL3 (Support for execution at EL3)
52- FEAT_EPAC (Enhanced pointer authentication)
53- FEAT_ETS2 (Enhanced Translation Synchronization)
54- FEAT_EVT (Enhanced Virtualization Traps)
55- FEAT_F32MM (Single-precision Matrix Multiplication)
56- FEAT_F64MM (Double-precision Matrix Multiplication)
57- FEAT_FCMA (Floating-point complex number instructions)
58- FEAT_FGT (Fine-Grained Traps)
59- FEAT_FHM (Floating-point half-precision multiplication instructions)
60- FEAT_FP (Floating Point extensions)
61- FEAT_FP16 (Half-precision floating-point data processing)
62- FEAT_FPAC (Faulting on AUT* instructions)
63- FEAT_FPACCOMBINE (Faulting on combined pointer authentication instructions)
64- FEAT_FPACC_SPEC (Speculative behavior of combined pointer authentication instructions)
65- FEAT_FRINTTS (Floating-point to integer instructions)
66- FEAT_FlagM (Flag manipulation instructions v2)
67- FEAT_FlagM2 (Enhancements to flag manipulation instructions)
68- FEAT_GTG (Guest translation granule size)
69- FEAT_HAFDBS (Hardware management of the access flag and dirty bit state)
70- FEAT_HBC (Hinted conditional branches)
71- FEAT_HCX (Support for the HCRX_EL2 register)
72- FEAT_HPDS (Hierarchical permission disables)
73- FEAT_HPDS2 (Translation table page-based hardware attributes)
74- FEAT_HPMN0 (Setting of MDCR_EL2.HPMN to zero)
75- FEAT_I8MM (AArch64 Int8 matrix multiplication instructions)
76- FEAT_IDST (ID space trap handling)
77- FEAT_IESB (Implicit error synchronization event)
78- FEAT_JSCVT (JavaScript conversion instructions)
79- FEAT_LOR (Limited ordering regions)
80- FEAT_LPA (Large Physical Address space)
81- FEAT_LPA2 (Large Physical and virtual Address space v2)
82- FEAT_LRCPC (Load-acquire RCpc instructions)
83- FEAT_LRCPC2 (Load-acquire RCpc instructions v2)
84- FEAT_LSE (Large System Extensions)
85- FEAT_LSE2 (Large System Extensions v2)
86- FEAT_LVA (Large Virtual Address space)
87- FEAT_MixedEnd (Mixed-endian support)
88- FEAT_MixdEndEL0 (Mixed-endian support at EL0)
89- FEAT_MOPS (Standardization of memory operations)
90- FEAT_MTE (Memory Tagging Extension)
91- FEAT_MTE2 (Memory Tagging Extension)
92- FEAT_MTE3 (MTE Asymmetric Fault Handling)
93- FEAT_MTE_ASYM_FAULT (Memory tagging asymmetric faults)
94- FEAT_NMI (Non-maskable Interrupt)
95- FEAT_NV (Nested Virtualization)
96- FEAT_NV2 (Enhanced nested virtualization support)
97- FEAT_PACIMP (Pointer authentication - IMPLEMENTATION DEFINED algorithm)
98- FEAT_PACQARMA3 (Pointer authentication - QARMA3 algorithm)
99- FEAT_PACQARMA5 (Pointer authentication - QARMA5 algorithm)
100- FEAT_PAN (Privileged access never)
101- FEAT_PAN2 (AT S1E1R and AT S1E1W instruction variants affected by PSTATE.PAN)
102- FEAT_PAN3 (Support for SCTLR_ELx.EPAN)
103- FEAT_PAuth (Pointer authentication)
104- FEAT_PAuth2 (Enhancements to pointer authentication)
105- FEAT_PMULL (PMULL, PMULL2 instructions)
106- FEAT_PMUv3 (PMU extension version 3)
107- FEAT_PMUv3p1 (PMU Extensions v3.1)
108- FEAT_PMUv3p4 (PMU Extensions v3.4)
109- FEAT_PMUv3p5 (PMU Extensions v3.5)
110- FEAT_RAS (Reliability, availability, and serviceability)
111- FEAT_RASv1p1 (RAS Extension v1.1)
112- FEAT_RDM (Advanced SIMD rounding double multiply accumulate instructions)
113- FEAT_RME (Realm Management Extension) (NB: support status in QEMU is experimental)
114- FEAT_RNG (Random number generator)
115- FEAT_S2FWB (Stage 2 forced Write-Back)
116- FEAT_SB (Speculation Barrier)
117- FEAT_SEL2 (Secure EL2)
118- FEAT_SHA1 (SHA1 instructions)
119- FEAT_SHA256 (SHA256 instructions)
120- FEAT_SHA3 (Advanced SIMD SHA3 instructions)
121- FEAT_SHA512 (Advanced SIMD SHA512 instructions)
122- FEAT_SM3 (Advanced SIMD SM3 instructions)
123- FEAT_SM4 (Advanced SIMD SM4 instructions)
124- FEAT_SME (Scalable Matrix Extension)
125- FEAT_SME_FA64 (Full A64 instruction set in Streaming SVE mode)
126- FEAT_SME_F64F64 (Double-precision floating-point outer product instructions)
127- FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions)
128- FEAT_SVE (Scalable Vector Extension)
129- FEAT_SVE_AES (Scalable Vector AES instructions)
130- FEAT_SVE_BitPerm (Scalable Vector Bit Permutes instructions)
131- FEAT_SVE_PMULL128 (Scalable Vector PMULL instructions)
132- FEAT_SVE_SHA3 (Scalable Vector SHA3 instructions)
133- FEAT_SVE_SM4 (Scalable Vector SM4 instructions)
134- FEAT_SVE2 (Scalable Vector Extension version 2)
135- FEAT_SPECRES (Speculation restriction instructions)
136- FEAT_SSBS (Speculative Store Bypass Safe)
137- FEAT_TGran16K (Support for 16KB memory translation granule size at stage 1)
138- FEAT_TGran4K (Support for 4KB memory translation granule size at stage 1)
139- FEAT_TGran64K (Support for 64KB memory translation granule size at stage 1)
140- FEAT_TIDCP1 (EL0 use of IMPLEMENTATION DEFINED functionality)
141- FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain)
142- FEAT_TLBIRANGE (TLB invalidate range instructions)
143- FEAT_TTCNP (Translation table Common not private translations)
144- FEAT_TTL (Translation Table Level)
145- FEAT_TTST (Small translation tables)
146- FEAT_UAO (Unprivileged Access Override control)
147- FEAT_VHE (Virtualization Host Extensions)
148- FEAT_VMID16 (16-bit VMID)
149- FEAT_WFxT (WFE and WFI instructions with timeout)
150- FEAT_XNX (Translation table stage 2 Unprivileged Execute-never)
151
152For information on the specifics of these extensions, please refer
153to the `Armv8-A Arm Architecture Reference Manual
154<https://developer.arm.com/documentation/ddi0487/latest>`_.
155
156When a specific named CPU is being emulated, only those features which
157are present in hardware for that CPU are emulated. (If a feature is
158not in the list above then it is not supported, even if the real
159hardware should have it.) The ``max`` CPU enables all features.
160
161R-profile CPU architecture support
162==================================
163
164QEMU's TCG emulation support for R-profile CPUs is currently limited.
165We emulate only the Cortex-R5 and Cortex-R5F CPUs.
166
167M-profile CPU architecture support
168==================================
169
170QEMU's TCG emulation includes support for Armv6-M, Armv7-M, Armv8-M, and
171Armv8.1-M versions of the M-profile architucture.  It also has support
172for the following architecture extensions:
173
174- FP (Floating-point Extension)
175- FPCXT (FPCXT access instructions)
176- HP (Half-precision floating-point instructions)
177- LOB (Low Overhead loops and Branch future)
178- M (Main Extension)
179- MPU (Memory Protection Unit Extension)
180- PXN (Privileged Execute Never)
181- RAS (Reliability, Serviceability and Availability): "minimum RAS Extension" only
182- S (Security Extension)
183- ST (System Timer Extension)
184
185For information on the specifics of these extensions, please refer
186to the `Armv8-M Arm Architecture Reference Manual
187<https://developer.arm.com/documentation/ddi0553/latest>`_.
188
189When a specific named CPU is being emulated, only those features which
190are present in hardware for that CPU are emulated. (If a feature is
191not in the list above then it is not supported, even if the real
192hardware should have it.) There is no equivalent of the ``max`` CPU for
193M-profile.
194