6 perf-c2c - Shared Data C2C/HITM Analyzer.
11 'perf c2c record' [<options>] <command>
12 'perf c2c record' [<options>] \-- [<record command options>] <command>
13 'perf c2c report' [<options>]
17 C2C stands for Cache To Cache.
19 The perf c2c tool provides means for Shared Data C2C/HITM analysis. It allows
20 you to track down the cacheline contentions.
22 On Intel, the tool is based on load latency and precise store facility events
23 provided by Intel CPUs. On PowerPC, the tool uses random instruction sampling
24 with thresholding feature. On AMD, the tool uses IBS op pmu (due to hardware
25 limitations, perf c2c is not supported on Zen3 cpus). On Arm64 it uses SPE to
26 sample load and store operations, therefore hardware and kernel support is
27 required. See linkperf:perf-arm-spe[1] for a setup guide. Due to the
28 statistical nature of Arm SPE sampling, not every memory operation will be
32 - memory address of the access
33 - type of the access (load and store details)
34 - latency (in cycles) of the load access
36 The c2c tool provide means to record this data and report back access details
37 for cachelines with highest contention - highest number of HITM accesses.
39 The basic workflow with this tool follows the standard record/report phase.
40 User uses the record command to record events data and report command to
48 Select the PMU event. Use 'perf c2c record -e list'
49 to list available events.
53 Be more verbose (show counter open errors, etc).
57 Configure mem-loads latency. Supported on Intel and Arm64 processors
58 only. Ignored on other archs.
62 Configure all used events to run in kernel space.
66 Configure all used events to run in user space.
76 Be more verbose (show counter open errors, etc).
80 Specify the input file to process.
84 Show extra node info in report (see NODE INFO section)
88 Specify sorting fields for single cacheline display.
89 Following fields are available: tid,pid,iaddr,dso
94 Setup callchains parameters.
95 Please refer to perf-report man page for details.
98 Force the stdio output (see STDIO OUTPUT)
101 Display only statistic tables and force stdio mode.
104 Display full length of symbols.
107 Do not display Source:Line column.
110 Show all captured HITM lines, with no regard to HITM % 0.0005 limit.
114 Don't do ownership validation.
118 Switch to HITM type (rmt, lcl) or peer snooping type (peer) to display
119 and sort on. Total HITMs (tot) as default, except Arm64 uses peer mode
123 Show callgraph with stitched LBRs, which may have more complete
124 callgraph. The perf.data file must have been obtained using
125 perf c2c record --call-graph lbr.
126 Disabled by default. In common cases with call stack overflows,
127 it can recreate better call stacks than the default lbr call stack
128 output. But this approach is not foolproof. There can be cases
129 where it creates incorrect call stacks from incorrect matches.
130 The known limitations include exception handing such as
131 setjmp/longjmp will have calls/returns not match.
134 Group the detection of shared cacheline events into double cacheline
135 granularity. Some architectures have an Adjacent Cacheline Prefetch
136 feature, which causes cacheline sharing to behave like the cacheline
141 The perf c2c record command setup options related to HITM cacheline analysis
142 and calls standard perf record command.
144 Following perf record options are configured by default:
145 (check perf record man page for details)
147 -W,-d,--phys-data,--sample-cpu
149 Unless specified otherwise with '-e' option, following events are monitored by
152 cpu/mem-loads,ldlat=30/P
159 and following on PowerPC:
164 User can pass any 'perf record' option behind '--' mark, like (to enable
165 callchains and system wide monitoring):
167 $ perf c2c record -- -g -a
169 Please check RECORD OPTIONS section for specific c2c record options.
173 The perf c2c report command displays shared data analysis. It comes in two
174 display modes: stdio and tui (default).
176 The report command workflow is following:
177 - sort all the data based on the cacheline address
178 - store access details for each cacheline
179 - sort all cachelines based on user settings
182 In general perf report output consist of 2 basic views:
183 1) most expensive cachelines list
184 2) offsets details for each cacheline
186 For each cacheline in the 1) list we display following data:
187 (Both stdio and TUI modes follow the same fields output)
190 - zero based index to identify the cacheline
193 - cacheline address (hex number)
195 Rmt/Lcl Hitm (Display with HITM types)
196 - cacheline percentage of all Remote/Local HITM accesses
198 Peer Snoop (Display with peer type)
199 - cacheline percentage of all peer accesses
201 LLC Load Hitm - Total, LclHitm, RmtHitm (For display with HITM types)
202 - count of Total/Local/Remote load HITMs
204 Load Peer - Total, Local, Remote (For display with peer type)
205 - count of Total/Local/Remote load from peer cache or DRAM
208 - sum of all cachelines accesses
211 - sum of all load accesses
214 - sum of all store accesses
216 Store Reference - L1Hit, L1Miss, N/A
217 L1Hit - store accesses that hit L1
218 L1Miss - store accesses that missed L1
219 N/A - store accesses with memory level is not available
221 Core Load Hit - FB, L1, L2
222 - count of load hits in FB (Fill Buffer), L1 and L2 cache
224 LLC Load Hit - LlcHit, LclHitm
225 - count of LLC load accesses, includes LLC hits and LLC HITMs
227 RMT Load Hit - RmtHit, RmtHitm
228 - count of remote load accesses, includes remote hits and remote HITMs;
229 on Arm neoverse cores, RmtHit is used to account remote accesses,
230 includes remote DRAM or any upward cache level in remote node
233 - count of local and remote DRAM accesses
235 For each offset in the 2) list we display following data:
237 HITM - Rmt, Lcl (Display with HITM types)
238 - % of Remote/Local HITM accesses for given offset within cacheline
240 Peer Snoop - Rmt, Lcl (Display with peer type)
241 - % of Remote/Local peer accesses for given offset within cacheline
243 Store Refs - L1 Hit, L1 Miss, N/A
244 - % of store accesses that hit L1, missed L1 and N/A (no available) memory
245 level for given offset within cacheline
247 Data address - Offset
251 - pid of the process responsible for the accesses
254 - tid of the process responsible for the accesses
257 - code address responsible for the accesses
259 cycles - rmt hitm, lcl hitm, load (Display with HITM types)
260 - sum of cycles for given accesses - Remote/Local HITM and generic load
262 cycles - rmt peer, lcl peer, load (Display with peer type)
263 - sum of cycles for given accesses - Remote/Local peer load and generic load
266 - number of cpus that participated on the access
269 - code symbol related to the 'Code address' value
272 - shared object name related to the 'Code address' value
275 - source information related to the 'Code address' value
278 - nodes participating on the access (see NODE INFO section)
282 The 'Node' field displays nodes that accesses given cacheline
283 offset. Its output comes in 3 flavors:
284 - node IDs separated by ','
285 - node IDs with stats for each ID, in following format:
286 Node{cpus %hitms %stores} (Display with HITM types)
287 Node{cpus %peers %stores} (Display with peer type)
288 - node IDs with list of affected CPUs in following format:
291 User can switch between above flavors with -N option or
292 use 'n' key to interactively switch in TUI mode.
296 User can specify how to sort offsets for cacheline.
298 Following fields are available and governs the final
299 output fields set for cacheline offsets output:
301 tid - coalesced by process TIDs
302 pid - coalesced by process PIDs
303 iaddr - coalesced by code address, following fields are displayed:
304 Code address, Code symbol, Shared Object, Source line
305 dso - coalesced by shared object
307 By default the coalescing is setup with 'pid,iaddr'.
311 The stdio output displays data on standard output.
313 Following tables are displayed:
314 Trace Event Information
315 - overall statistics of memory accesses
317 Global Shared Cache Line Event Information
318 - overall statistics on shared cachelines
320 Shared Data Cache Line Table
321 - list of most expensive cachelines
323 Shared Cache Line Distribution Pareto
324 - list of all accessed offsets for each cacheline
328 The TUI output provides interactive interface to navigate
329 through cachelines list and to display offset details.
331 For details please refer to the help window by pressing '?' key.
335 Although Don Zickus, Dick Fowles and Joe Mario worked together
336 to get this implemented, we got lots of early help from Arnaldo
337 Carvalho de Melo, Stephane Eranian, Jiri Olsa and Andi Kleen.
341 Check Joe's blog on c2c tool for detailed use case explanation:
342 https://joemario.github.io/blog/2016/09/01/c2c-blog/
346 linkperf:perf-record[1], linkperf:perf-mem[1], linkperf:perf-arm-spe[1]