xref: /linux/tools/perf/Documentation/perf-c2c.txt (revision b61104e7a6349bd2c2b3e2fb3260d87f15eda8f4)
1perf-c2c(1)
2===========
3
4NAME
5----
6perf-c2c - Shared Data C2C/HITM Analyzer.
7
8SYNOPSIS
9--------
10[verse]
11'perf c2c record' [<options>] <command>
12'perf c2c record' [<options>] \-- [<record command options>] <command>
13'perf c2c report' [<options>]
14
15DESCRIPTION
16-----------
17C2C stands for Cache To Cache.
18
19The perf c2c tool provides means for Shared Data C2C/HITM analysis. It allows
20you to track down the cacheline contentions.
21
22On Intel, the tool is based on load latency and precise store facility events
23provided by Intel CPUs. On PowerPC, the tool uses random instruction sampling
24with thresholding feature. On AMD, the tool uses IBS op pmu (due to hardware
25limitations, perf c2c is not supported on Zen3 cpus). On Arm64 it uses SPE to
26sample load and store operations, therefore hardware and kernel support is
27required. See linkperf:perf-arm-spe[1] for a setup guide. Due to the
28statistical nature of Arm SPE sampling, not every memory operation will be
29sampled.
30
31These events provide:
32  - memory address of the access
33  - type of the access (load and store details)
34  - latency (in cycles) of the load access
35
36The c2c tool provide means to record this data and report back access details
37for cachelines with highest contention - highest number of HITM accesses.
38
39The basic workflow with this tool follows the standard record/report phase.
40User uses the record command to record events data and report command to
41display it.
42
43
44RECORD OPTIONS
45--------------
46-e::
47--event=::
48	Select the PMU event. Use 'perf c2c record -e list'
49	to list available events.
50
51-v::
52--verbose::
53	Be more verbose (show counter open errors, etc).
54
55-l::
56--ldlat::
57	Configure mem-loads latency. Supported on Intel, Arm64 and some AMD
58	processors. Ignored on other archs.
59
60	On supported AMD processors:
61	- /sys/bus/event_source/devices/ibs_op/caps/ldlat file contains '1'.
62	- Supported latency values are 128 to 2048 (both inclusive).
63	- Latency value which is a multiple of 128 incurs a little less profiling
64	  overhead compared to other values.
65	- Load latency filtering is disabled by default.
66
67-k::
68--all-kernel::
69	Configure all used events to run in kernel space.
70
71-u::
72--all-user::
73	Configure all used events to run in user space.
74
75REPORT OPTIONS
76--------------
77-k::
78--vmlinux=<file>::
79	vmlinux pathname
80
81-v::
82--verbose::
83	Be more verbose (show counter open errors, etc).
84
85-i::
86--input::
87	Specify the input file to process.
88
89-N::
90--node-info::
91	Show extra node info in report (see NODE INFO section)
92
93-c::
94--coalesce::
95	Specify sorting fields for single cacheline display.
96	Following fields are available: tid,pid,iaddr,dso
97	(see COALESCE)
98
99-g::
100--call-graph::
101	Setup callchains parameters.
102	Please refer to perf-report man page for details.
103
104--stdio::
105	Force the stdio output (see STDIO OUTPUT)
106
107--stats::
108	Display only statistic tables and force stdio mode.
109
110--full-symbols::
111	Display full length of symbols.
112
113--no-source::
114	Do not display Source:Line column.
115
116--show-all::
117	Show all captured HITM lines, with no regard to HITM % 0.0005 limit.
118
119-f::
120--force::
121	Don't do ownership validation.
122
123-d::
124--display::
125	Switch to HITM type (rmt, lcl) or peer snooping type (peer) to display
126	and sort on. Total HITMs (tot) as default, except Arm64 uses peer mode
127	as default.
128
129--stitch-lbr::
130	Show callgraph with stitched LBRs, which may have more complete
131	callgraph. The perf.data file must have been obtained using
132	perf c2c record --call-graph lbr.
133	Disabled by default. In common cases with call stack overflows,
134	it can recreate better call stacks than the default lbr call stack
135	output. But this approach is not foolproof. There can be cases
136	where it creates incorrect call stacks from incorrect matches.
137	The known limitations include exception handing such as
138	setjmp/longjmp will have calls/returns not match.
139
140--double-cl::
141	Group the detection of shared cacheline events into double cacheline
142	granularity. Some architectures have an Adjacent Cacheline Prefetch
143	feature, which causes cacheline sharing to behave like the cacheline
144	size is doubled.
145
146-M::
147--disassembler-style=::
148	Set disassembler style for objdump.
149
150--objdump=<path>::
151        Path to objdump binary.
152
153C2C RECORD
154----------
155The perf c2c record command setup options related to HITM cacheline analysis
156and calls standard perf record command.
157
158Following perf record options are configured by default:
159(check perf record man page for details)
160
161  -W,-d,--phys-data,--sample-cpu
162
163Unless specified otherwise with '-e' option, following events are monitored by
164default on Intel:
165
166  cpu/mem-loads,ldlat=30/P
167  cpu/mem-stores/P
168
169following on AMD:
170
171  ibs_op//
172
173and following on PowerPC:
174
175  cpu/mem-loads/
176  cpu/mem-stores/
177
178User can pass any 'perf record' option behind '--' mark, like (to enable
179callchains and system wide monitoring):
180
181  $ perf c2c record -- -g -a
182
183Please check RECORD OPTIONS section for specific c2c record options.
184
185C2C REPORT
186----------
187The perf c2c report command displays shared data analysis.  It comes in two
188display modes: stdio and tui (default).
189
190The report command workflow is following:
191  - sort all the data based on the cacheline address
192  - store access details for each cacheline
193  - sort all cachelines based on user settings
194  - display data
195
196In general perf report output consist of 2 basic views:
197  1) most expensive cachelines list
198  2) offsets details for each cacheline
199
200For each cacheline in the 1) list we display following data:
201(Both stdio and TUI modes follow the same fields output)
202
203  Index
204  - zero based index to identify the cacheline
205
206  Cacheline
207  - cacheline address (hex number)
208
209  Rmt/Lcl Hitm (Display with HITM types)
210  - cacheline percentage of all Remote/Local HITM accesses
211
212  Peer Snoop (Display with peer type)
213  - cacheline percentage of all peer accesses
214
215  LLC Load Hitm - Total, LclHitm, RmtHitm (For display with HITM types)
216  - count of Total/Local/Remote load HITMs
217
218  Load Peer - Total, Local, Remote (For display with peer type)
219  - count of Total/Local/Remote load from peer cache or DRAM
220
221  Total records
222  - sum of all cachelines accesses
223
224  Total loads
225  - sum of all load accesses
226
227  Total stores
228  - sum of all store accesses
229
230  Store Reference - L1Hit, L1Miss, N/A
231    L1Hit - store accesses that hit L1
232    L1Miss - store accesses that missed L1
233    N/A - store accesses with memory level is not available
234
235  Core Load Hit - FB, L1, L2
236  - count of load hits in FB (Fill Buffer), L1 and L2 cache
237
238  LLC Load Hit - LlcHit, LclHitm
239  - count of LLC load accesses, includes LLC hits and LLC HITMs
240
241  RMT Load Hit - RmtHit, RmtHitm
242  - count of remote load accesses, includes remote hits and remote HITMs;
243    on Arm neoverse cores, RmtHit is used to account remote accesses,
244    includes remote DRAM or any upward cache level in remote node
245
246  Load Dram - Lcl, Rmt
247  - count of local and remote DRAM accesses
248
249For each offset in the 2) list we display following data:
250
251  HITM - Rmt, Lcl (Display with HITM types)
252  - % of Remote/Local HITM accesses for given offset within cacheline
253
254  Peer Snoop - Rmt, Lcl (Display with peer type)
255  - % of Remote/Local peer accesses for given offset within cacheline
256
257  Store Refs - L1 Hit, L1 Miss, N/A
258  - % of store accesses that hit L1, missed L1 and N/A (no available) memory
259    level for given offset within cacheline
260
261  Data address - Offset
262  - offset address
263
264  Pid
265  - pid of the process responsible for the accesses
266
267  Tid
268  - tid of the process responsible for the accesses
269
270  Code address
271  - code address responsible for the accesses
272
273  cycles - rmt hitm, lcl hitm, load (Display with HITM types)
274    - sum of cycles for given accesses - Remote/Local HITM and generic load
275
276  cycles - rmt peer, lcl peer, load (Display with peer type)
277    - sum of cycles for given accesses - Remote/Local peer load and generic load
278
279  cpu cnt
280    - number of cpus that participated on the access
281
282  Symbol
283    - code symbol related to the 'Code address' value
284
285  Shared Object
286    - shared object name related to the 'Code address' value
287
288  Source:Line
289    - source information related to the 'Code address' value
290
291  Node
292    - nodes participating on the access (see NODE INFO section)
293
294NODE INFO
295---------
296The 'Node' field displays nodes that accesses given cacheline
297offset. Its output comes in 3 flavors:
298  - node IDs separated by ','
299  - node IDs with stats for each ID, in following format:
300      Node{cpus %hitms %stores} (Display with HITM types)
301      Node{cpus %peers %stores} (Display with peer type)
302  - node IDs with list of affected CPUs in following format:
303      Node{cpu list}
304
305User can switch between above flavors with -N option or
306use 'n' key to interactively switch in TUI mode.
307
308COALESCE
309--------
310User can specify how to sort offsets for cacheline.
311
312Following fields are available and governs the final
313output fields set for cacheline offsets output:
314
315  tid   - coalesced by process TIDs
316  pid   - coalesced by process PIDs
317  iaddr - coalesced by code address, following fields are displayed:
318             Code address, Code symbol, Shared Object, Source line
319  dso   - coalesced by shared object
320
321By default the coalescing is setup with 'pid,iaddr'.
322
323STDIO OUTPUT
324------------
325The stdio output displays data on standard output.
326
327Following tables are displayed:
328  Trace Event Information
329  - overall statistics of memory accesses
330
331  Global Shared Cache Line Event Information
332  - overall statistics on shared cachelines
333
334  Shared Data Cache Line Table
335  - list of most expensive cachelines
336
337  Shared Cache Line Distribution Pareto
338  - list of all accessed offsets for each cacheline
339
340TUI OUTPUT
341----------
342The TUI output provides interactive interface to navigate
343through cachelines list and to display offset details.
344
345For details please refer to the help window by pressing '?' key.
346
347CREDITS
348-------
349Although Don Zickus, Dick Fowles and Joe Mario worked together
350to get this implemented, we got lots of early help from Arnaldo
351Carvalho de Melo, Stephane Eranian, Jiri Olsa and Andi Kleen.
352
353C2C BLOG
354--------
355Check Joe's blog on c2c tool for detailed use case explanation:
356  https://joemario.github.io/blog/2016/09/01/c2c-blog/
357
358SEE ALSO
359--------
360linkperf:perf-record[1], linkperf:perf-mem[1], linkperf:perf-arm-spe[1]
361