xref: /openbmc/qemu/docs/system/arm/emulation.rst (revision 701bff24)
1A-profile CPU architecture support
2==================================
3
4QEMU's TCG emulation includes support for the Armv5, Armv6, Armv7 and
5Armv8 versions of the A-profile architecture. It also has support for
6the following architecture extensions:
7
8- FEAT_AA32BF16 (AArch32 BFloat16 instructions)
9- FEAT_AA32HPD (AArch32 hierarchical permission disables)
10- FEAT_AA32I8MM (AArch32 Int8 matrix multiplication instructions)
11- FEAT_AES (AESD and AESE instructions)
12- FEAT_BBM at level 2 (Translation table break-before-make levels)
13- FEAT_BF16 (AArch64 BFloat16 instructions)
14- FEAT_BTI (Branch Target Identification)
15- FEAT_CSV2 (Cache speculation variant 2)
16- FEAT_CSV2_1p1 (Cache speculation variant 2, version 1.1)
17- FEAT_CSV2_1p2 (Cache speculation variant 2, version 1.2)
18- FEAT_CSV2_2 (Cache speculation variant 2, version 2)
19- FEAT_CSV3 (Cache speculation variant 3)
20- FEAT_DGH (Data gathering hint)
21- FEAT_DIT (Data Independent Timing instructions)
22- FEAT_DPB (DC CVAP instruction)
23- FEAT_Debugv8p2 (Debug changes for v8.2)
24- FEAT_Debugv8p4 (Debug changes for v8.4)
25- FEAT_DotProd (Advanced SIMD dot product instructions)
26- FEAT_DoubleFault (Double Fault Extension)
27- FEAT_ETS (Enhanced Translation Synchronization)
28- FEAT_FCMA (Floating-point complex number instructions)
29- FEAT_FHM (Floating-point half-precision multiplication instructions)
30- FEAT_FP16 (Half-precision floating-point data processing)
31- FEAT_FRINTTS (Floating-point to integer instructions)
32- FEAT_FlagM (Flag manipulation instructions v2)
33- FEAT_FlagM2 (Enhancements to flag manipulation instructions)
34- FEAT_GTG (Guest translation granule size)
35- FEAT_HCX (Support for the HCRX_EL2 register)
36- FEAT_HPDS (Hierarchical permission disables)
37- FEAT_I8MM (AArch64 Int8 matrix multiplication instructions)
38- FEAT_IDST (ID space trap handling)
39- FEAT_IESB (Implicit error synchronization event)
40- FEAT_JSCVT (JavaScript conversion instructions)
41- FEAT_LOR (Limited ordering regions)
42- FEAT_LPA (Large Physical Address space)
43- FEAT_LPA2 (Large Physical and virtual Address space v2)
44- FEAT_LRCPC (Load-acquire RCpc instructions)
45- FEAT_LRCPC2 (Load-acquire RCpc instructions v2)
46- FEAT_LSE (Large System Extensions)
47- FEAT_LVA (Large Virtual Address space)
48- FEAT_MTE (Memory Tagging Extension)
49- FEAT_MTE2 (Memory Tagging Extension)
50- FEAT_MTE3 (MTE Asymmetric Fault Handling)
51- FEAT_PAN (Privileged access never)
52- FEAT_PAN2 (AT S1E1R and AT S1E1W instruction variants affected by PSTATE.PAN)
53- FEAT_PAuth (Pointer authentication)
54- FEAT_PMULL (PMULL, PMULL2 instructions)
55- FEAT_PMUv3p1 (PMU Extensions v3.1)
56- FEAT_PMUv3p4 (PMU Extensions v3.4)
57- FEAT_PMUv3p5 (PMU Extensions v3.5)
58- FEAT_RAS (Reliability, availability, and serviceability)
59- FEAT_RASv1p1 (RAS Extension v1.1)
60- FEAT_RDM (Advanced SIMD rounding double multiply accumulate instructions)
61- FEAT_RNG (Random number generator)
62- FEAT_S2FWB (Stage 2 forced Write-Back)
63- FEAT_SB (Speculation Barrier)
64- FEAT_SEL2 (Secure EL2)
65- FEAT_SHA1 (SHA1 instructions)
66- FEAT_SHA256 (SHA256 instructions)
67- FEAT_SHA3 (Advanced SIMD SHA3 instructions)
68- FEAT_SHA512 (Advanced SIMD SHA512 instructions)
69- FEAT_SM3 (Advanced SIMD SM3 instructions)
70- FEAT_SM4 (Advanced SIMD SM4 instructions)
71- FEAT_SME (Scalable Matrix Extension)
72- FEAT_SME_FA64 (Full A64 instruction set in Streaming SVE mode)
73- FEAT_SME_F64F64 (Double-precision floating-point outer product instructions)
74- FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions)
75- FEAT_SPECRES (Speculation restriction instructions)
76- FEAT_SSBS (Speculative Store Bypass Safe)
77- FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain)
78- FEAT_TLBIRANGE (TLB invalidate range instructions)
79- FEAT_TTCNP (Translation table Common not private translations)
80- FEAT_TTL (Translation Table Level)
81- FEAT_TTST (Small translation tables)
82- FEAT_UAO (Unprivileged Access Override control)
83- FEAT_VHE (Virtualization Host Extensions)
84- FEAT_VMID16 (16-bit VMID)
85- FEAT_XNX (Translation table stage 2 Unprivileged Execute-never)
86- SVE (The Scalable Vector Extension)
87- SVE2 (The Scalable Vector Extension v2)
88
89For information on the specifics of these extensions, please refer
90to the `Armv8-A Arm Architecture Reference Manual
91<https://developer.arm.com/documentation/ddi0487/latest>`_.
92
93When a specific named CPU is being emulated, only those features which
94are present in hardware for that CPU are emulated. (If a feature is
95not in the list above then it is not supported, even if the real
96hardware should have it.) The ``max`` CPU enables all features.
97
98R-profile CPU architecture support
99==================================
100
101QEMU's TCG emulation support for R-profile CPUs is currently limited.
102We emulate only the Cortex-R5 and Cortex-R5F CPUs.
103
104M-profile CPU architecture support
105==================================
106
107QEMU's TCG emulation includes support for Armv6-M, Armv7-M, Armv8-M, and
108Armv8.1-M versions of the M-profile architucture.  It also has support
109for the following architecture extensions:
110
111- FP (Floating-point Extension)
112- FPCXT (FPCXT access instructions)
113- HP (Half-precision floating-point instructions)
114- LOB (Low Overhead loops and Branch future)
115- M (Main Extension)
116- MPU (Memory Protection Unit Extension)
117- PXN (Privileged Execute Never)
118- RAS (Reliability, Serviceability and Availability): "minimum RAS Extension" only
119- S (Security Extension)
120- ST (System Timer Extension)
121
122For information on the specifics of these extensions, please refer
123to the `Armv8-M Arm Architecture Reference Manual
124<https://developer.arm.com/documentation/ddi0553/latest>`_.
125
126When a specific named CPU is being emulated, only those features which
127are present in hardware for that CPU are emulated. (If a feature is
128not in the list above then it is not supported, even if the real
129hardware should have it.) There is no equivalent of the ``max`` CPU for
130M-profile.
131