1 .. _amu_index: 2 3 ======================================================= 4 Activity Monitors Unit (AMU) extension in AArch64 Linux 5 ======================================================= 6 7 Author: Ionela Voinescu <ionela.voinescu@arm.com> 8 9 Date: 2019-09-10 10 11 This document briefly describes the provision of Activity Monitors Unit 12 support in AArch64 Linux. 13 14 15 Architecture overview 16 --------------------- 17 18 The activity monitors extension is an optional extension introduced by the 19 ARMv8.4 CPU architecture. 20 21 The activity monitors unit, implemented in each CPU, provides performance 22 counters intended for system management use. The AMU extension provides a 23 system register interface to the counter registers and also supports an 24 optional external memory-mapped interface. 25 26 Version 1 of the Activity Monitors architecture implements a counter group 27 of four fixed and architecturally defined 64-bit event counters. 28 29 - CPU cycle counter: increments at the frequency of the CPU. 30 - Constant counter: increments at the fixed frequency of the system 31 clock. 32 - Instructions retired: increments with every architecturally executed 33 instruction. 34 - Memory stall cycles: counts instruction dispatch stall cycles caused by 35 misses in the last level cache within the clock domain. 36 37 When in WFI or WFE these counters do not increment. 38 39 The Activity Monitors architecture provides space for up to 16 architected 40 event counters. Future versions of the architecture may use this space to 41 implement additional architected event counters. 42 43 Additionally, version 1 implements a counter group of up to 16 auxiliary 44 64-bit event counters. 45 46 On cold reset all counters reset to 0. 47 48 49 Basic support 50 ------------- 51 52 The kernel can safely run a mix of CPUs with and without support for the 53 activity monitors extension. Therefore, when CONFIG_ARM64_AMU_EXTN is 54 selected we unconditionally enable the capability to allow any late CPU 55 (secondary or hotplugged) to detect and use the feature. 56 57 When the feature is detected on a CPU, we flag the availability of the 58 feature but this does not guarantee the correct functionality of the 59 counters, only the presence of the extension. 60 61 Firmware (code running at higher exception levels, e.g. arm-tf) support is 62 needed to: 63 64 - Enable access for lower exception levels (EL2 and EL1) to the AMU 65 registers. 66 - Enable the counters. If not enabled these will read as 0. 67 - Save/restore the counters before/after the CPU is being put/brought up 68 from the 'off' power state. 69 70 When using kernels that have this feature enabled but boot with broken 71 firmware the user may experience panics or lockups when accessing the 72 counter registers. Even if these symptoms are not observed, the values 73 returned by the register reads might not correctly reflect reality. Most 74 commonly, the counters will read as 0, indicating that they are not 75 enabled. 76 77 If proper support is not provided in firmware it's best to disable 78 CONFIG_ARM64_AMU_EXTN. To be noted that for security reasons, this does not 79 bypass the setting of AMUSERENR_EL0 to trap accesses from EL0 (userspace) to 80 EL1 (kernel). Therefore, firmware should still ensure accesses to AMU registers 81 are not trapped in EL2/EL3. 82 83 The fixed counters of AMUv1 are accessible though the following system 84 register definitions: 85 86 - SYS_AMEVCNTR0_CORE_EL0 87 - SYS_AMEVCNTR0_CONST_EL0 88 - SYS_AMEVCNTR0_INST_RET_EL0 89 - SYS_AMEVCNTR0_MEM_STALL_EL0 90 91 Auxiliary platform specific counters can be accessed using 92 SYS_AMEVCNTR1_EL0(n), where n is a value between 0 and 15. 93 94 Details can be found in: arch/arm64/include/asm/sysreg.h. 95 96 97 Userspace access 98 ---------------- 99 100 Currently, access from userspace to the AMU registers is disabled due to: 101 102 - Security reasons: they might expose information about code executed in 103 secure mode. 104 - Purpose: AMU counters are intended for system management use. 105 106 Also, the presence of the feature is not visible to userspace. 107 108 109 Virtualization 110 -------------- 111 112 Currently, access from userspace (EL0) and kernelspace (EL1) on the KVM 113 guest side is disabled due to: 114 115 - Security reasons: they might expose information about code executed 116 by other guests or the host. 117 118 Any attempt to access the AMU registers will result in an UNDEFINED 119 exception being injected into the guest.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.