1perf-list(1) 2============ 3 4NAME 5---- 6perf-list - List all symbolic event types 7 8SYNOPSIS 9-------- 10[verse] 11'perf list' [hw|sw|cache|tracepoint|pmu|event_glob] 12 13DESCRIPTION 14----------- 15This command displays the symbolic event types which can be selected in the 16various perf commands with the -e option. 17 18[[EVENT_MODIFIERS]] 19EVENT MODIFIERS 20--------------- 21 22Events can optionally have a modifier by appending a colon and one or 23more modifiers. Modifiers allow the user to restrict the events to be 24counted. The following modifiers exist: 25 26 u - user-space counting 27 k - kernel counting 28 h - hypervisor counting 29 I - non idle counting 30 G - guest counting (in KVM guests) 31 H - host counting (not in KVM guests) 32 p - precise level 33 P - use maximum detected precise level 34 S - read sample value (PERF_SAMPLE_READ) 35 D - pin the event to the PMU 36 37The 'p' modifier can be used for specifying how precise the instruction 38address should be. The 'p' modifier can be specified multiple times: 39 40 0 - SAMPLE_IP can have arbitrary skid 41 1 - SAMPLE_IP must have constant skid 42 2 - SAMPLE_IP requested to have 0 skid 43 3 - SAMPLE_IP must have 0 skid 44 45For Intel systems precise event sampling is implemented with PEBS 46which supports up to precise-level 2. 47 48On AMD systems it is implemented using IBS (up to precise-level 2). 49The precise modifier works with event types 0x76 (cpu-cycles, CPU 50clocks not halted) and 0xC1 (micro-ops retired). Both events map to 51IBS execution sampling (IBS op) with the IBS Op Counter Control bit 52(IbsOpCntCtl) set respectively (see AMD64 Architecture Programmer’s 53Manual Volume 2: System Programming, 13.3 Instruction-Based 54Sampling). Examples to use IBS: 55 56 perf record -a -e cpu-cycles:p ... # use ibs op counting cycles 57 perf record -a -e r076:p ... # same as -e cpu-cycles:p 58 perf record -a -e r0C1:p ... # use ibs op counting micro-ops 59 60RAW HARDWARE EVENT DESCRIPTOR 61----------------------------- 62Even when an event is not available in a symbolic form within perf right now, 63it can be encoded in a per processor specific way. 64 65For instance For x86 CPUs NNN represents the raw register encoding with the 66layout of IA32_PERFEVTSELx MSRs (see [Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide] Figure 30-1 Layout 67of IA32_PERFEVTSELx MSRs) or AMD's PerfEvtSeln (see [AMD64 Architecture Programmer’s Manual Volume 2: System Programming], Page 344, 68Figure 13-7 Performance Event-Select Register (PerfEvtSeln)). 69 70Note: Only the following bit fields can be set in x86 counter 71registers: event, umask, edge, inv, cmask. Esp. guest/host only and 72OS/user mode flags must be setup using <<EVENT_MODIFIERS, EVENT 73MODIFIERS>>. 74 75Example: 76 77If the Intel docs for a QM720 Core i7 describe an event as: 78 79 Event Umask Event Mask 80 Num. Value Mnemonic Description Comment 81 82 A8H 01H LSD.UOPS Counts the number of micro-ops Use cmask=1 and 83 delivered by loop stream detector invert to count 84 cycles 85 86raw encoding of 0x1A8 can be used: 87 88 perf stat -e r1a8 -a sleep 1 89 perf record -e r1a8 ... 90 91You should refer to the processor specific documentation for getting these 92details. Some of them are referenced in the SEE ALSO section below. 93 94PARAMETERIZED EVENTS 95-------------------- 96 97Some pmu events listed by 'perf-list' will be displayed with '?' in them. For 98example: 99 100 hv_gpci/dtbp_ptitc,phys_processor_idx=?/ 101 102This means that when provided as an event, a value for '?' must 103also be supplied. For example: 104 105 perf stat -C 0 -e 'hv_gpci/dtbp_ptitc,phys_processor_idx=0x2/' ... 106 107OPTIONS 108------- 109 110Without options all known events will be listed. 111 112To limit the list use: 113 114. 'hw' or 'hardware' to list hardware events such as cache-misses, etc. 115 116. 'sw' or 'software' to list software events such as context switches, etc. 117 118. 'cache' or 'hwcache' to list hardware cache events such as L1-dcache-loads, etc. 119 120. 'tracepoint' to list all tracepoint events, alternatively use 121 'subsys_glob:event_glob' to filter by tracepoint subsystems such as sched, 122 block, etc. 123 124. 'pmu' to print the kernel supplied PMU events. 125 126. If none of the above is matched, it will apply the supplied glob to all 127 events, printing the ones that match. 128 129. As a last resort, it will do a substring search in all event names. 130 131One or more types can be used at the same time, listing the events for the 132types specified. 133 134Support raw format: 135 136. '--raw-dump', shows the raw-dump of all the events. 137. '--raw-dump [hw|sw|cache|tracepoint|pmu|event_glob]', shows the raw-dump of 138 a certain kind of events. 139 140SEE ALSO 141-------- 142linkperf:perf-stat[1], linkperf:perf-top[1], 143linkperf:perf-record[1], 144http://www.intel.com/Assets/PDF/manual/253669.pdf[Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide], 145http://support.amd.com/us/Processor_TechDocs/24593_APM_v2.pdf[AMD64 Architecture Programmer’s Manual Volume 2: System Programming] 146