1# 2# Architectures that offer an FUNCTION_TRACER implementation should 3# select HAVE_FUNCTION_TRACER: 4# 5 6config USER_STACKTRACE_SUPPORT 7 bool 8 9config NOP_TRACER 10 bool 11 12config HAVE_FTRACE_NMI_ENTER 13 bool 14 help 15 See Documentation/trace/ftrace-design.txt 16 17config HAVE_FUNCTION_TRACER 18 bool 19 help 20 See Documentation/trace/ftrace-design.txt 21 22config HAVE_FUNCTION_GRAPH_TRACER 23 bool 24 help 25 See Documentation/trace/ftrace-design.txt 26 27config HAVE_DYNAMIC_FTRACE 28 bool 29 help 30 See Documentation/trace/ftrace-design.txt 31 32config HAVE_DYNAMIC_FTRACE_WITH_REGS 33 bool 34 35config HAVE_FTRACE_MCOUNT_RECORD 36 bool 37 help 38 See Documentation/trace/ftrace-design.txt 39 40config HAVE_SYSCALL_TRACEPOINTS 41 bool 42 help 43 See Documentation/trace/ftrace-design.txt 44 45config HAVE_FENTRY 46 bool 47 help 48 Arch supports the gcc options -pg with -mfentry 49 50config HAVE_C_RECORDMCOUNT 51 bool 52 help 53 C version of recordmcount available? 54 55config TRACER_MAX_TRACE 56 bool 57 58config TRACE_CLOCK 59 bool 60 61config RING_BUFFER 62 bool 63 select TRACE_CLOCK 64 select IRQ_WORK 65 66config FTRACE_NMI_ENTER 67 bool 68 depends on HAVE_FTRACE_NMI_ENTER 69 default y 70 71config EVENT_TRACING 72 select CONTEXT_SWITCH_TRACER 73 select GLOB 74 bool 75 76config CONTEXT_SWITCH_TRACER 77 bool 78 79config RING_BUFFER_ALLOW_SWAP 80 bool 81 help 82 Allow the use of ring_buffer_swap_cpu. 83 Adds a very slight overhead to tracing when enabled. 84 85# All tracer options should select GENERIC_TRACER. For those options that are 86# enabled by all tracers (context switch and event tracer) they select TRACING. 87# This allows those options to appear when no other tracer is selected. But the 88# options do not appear when something else selects it. We need the two options 89# GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the 90# hiding of the automatic options. 91 92config TRACING 93 bool 94 select DEBUG_FS 95 select RING_BUFFER 96 select STACKTRACE if STACKTRACE_SUPPORT 97 select TRACEPOINTS 98 select NOP_TRACER 99 select BINARY_PRINTF 100 select EVENT_TRACING 101 select TRACE_CLOCK 102 103config GENERIC_TRACER 104 bool 105 select TRACING 106 107# 108# Minimum requirements an architecture has to meet for us to 109# be able to offer generic tracing facilities: 110# 111config TRACING_SUPPORT 112 bool 113 # PPC32 has no irqflags tracing support, but it can use most of the 114 # tracers anyway, they were tested to build and work. Note that new 115 # exceptions to this list aren't welcomed, better implement the 116 # irqflags tracing for your architecture. 117 depends on TRACE_IRQFLAGS_SUPPORT || PPC32 118 depends on STACKTRACE_SUPPORT 119 default y 120 121if TRACING_SUPPORT 122 123menuconfig FTRACE 124 bool "Tracers" 125 default y if DEBUG_KERNEL 126 help 127 Enable the kernel tracing infrastructure. 128 129if FTRACE 130 131config FUNCTION_TRACER 132 bool "Kernel Function Tracer" 133 depends on HAVE_FUNCTION_TRACER 134 select KALLSYMS 135 select GENERIC_TRACER 136 select CONTEXT_SWITCH_TRACER 137 select GLOB 138 select TASKS_RCU if PREEMPT 139 help 140 Enable the kernel to trace every kernel function. This is done 141 by using a compiler feature to insert a small, 5-byte No-Operation 142 instruction at the beginning of every kernel function, which NOP 143 sequence is then dynamically patched into a tracer call when 144 tracing is enabled by the administrator. If it's runtime disabled 145 (the bootup default), then the overhead of the instructions is very 146 small and not measurable even in micro-benchmarks. 147 148config FUNCTION_GRAPH_TRACER 149 bool "Kernel Function Graph Tracer" 150 depends on HAVE_FUNCTION_GRAPH_TRACER 151 depends on FUNCTION_TRACER 152 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE 153 default y 154 help 155 Enable the kernel to trace a function at both its return 156 and its entry. 157 Its first purpose is to trace the duration of functions and 158 draw a call graph for each thread with some information like 159 the return value. This is done by setting the current return 160 address on the current task structure into a stack of calls. 161 162 163config IRQSOFF_TRACER 164 bool "Interrupts-off Latency Tracer" 165 default n 166 depends on TRACE_IRQFLAGS_SUPPORT 167 depends on !ARCH_USES_GETTIMEOFFSET 168 select TRACE_IRQFLAGS 169 select GENERIC_TRACER 170 select TRACER_MAX_TRACE 171 select RING_BUFFER_ALLOW_SWAP 172 select TRACER_SNAPSHOT 173 select TRACER_SNAPSHOT_PER_CPU_SWAP 174 help 175 This option measures the time spent in irqs-off critical 176 sections, with microsecond accuracy. 177 178 The default measurement method is a maximum search, which is 179 disabled by default and can be runtime (re-)started 180 via: 181 182 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency 183 184 (Note that kernel size and overhead increase with this option 185 enabled. This option and the preempt-off timing option can be 186 used together or separately.) 187 188config PREEMPT_TRACER 189 bool "Preemption-off Latency Tracer" 190 default n 191 depends on !ARCH_USES_GETTIMEOFFSET 192 depends on PREEMPT 193 select GENERIC_TRACER 194 select TRACER_MAX_TRACE 195 select RING_BUFFER_ALLOW_SWAP 196 select TRACER_SNAPSHOT 197 select TRACER_SNAPSHOT_PER_CPU_SWAP 198 help 199 This option measures the time spent in preemption-off critical 200 sections, with microsecond accuracy. 201 202 The default measurement method is a maximum search, which is 203 disabled by default and can be runtime (re-)started 204 via: 205 206 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency 207 208 (Note that kernel size and overhead increase with this option 209 enabled. This option and the irqs-off timing option can be 210 used together or separately.) 211 212config SCHED_TRACER 213 bool "Scheduling Latency Tracer" 214 select GENERIC_TRACER 215 select CONTEXT_SWITCH_TRACER 216 select TRACER_MAX_TRACE 217 select TRACER_SNAPSHOT 218 help 219 This tracer tracks the latency of the highest priority task 220 to be scheduled in, starting from the point it has woken up. 221 222config HWLAT_TRACER 223 bool "Tracer to detect hardware latencies (like SMIs)" 224 select GENERIC_TRACER 225 help 226 This tracer, when enabled will create one or more kernel threads, 227 depening on what the cpumask file is set to, which each thread 228 spinning in a loop looking for interruptions caused by 229 something other than the kernel. For example, if a 230 System Management Interrupt (SMI) takes a noticeable amount of 231 time, this tracer will detect it. This is useful for testing 232 if a system is reliable for Real Time tasks. 233 234 Some files are created in the tracing directory when this 235 is enabled: 236 237 hwlat_detector/width - time in usecs for how long to spin for 238 hwlat_detector/window - time in usecs between the start of each 239 iteration 240 241 A kernel thread is created that will spin with interrupts disabled 242 for "width" microseconds in every "widow" cycle. It will not spin 243 for "window - width" microseconds, where the system can 244 continue to operate. 245 246 The output will appear in the trace and trace_pipe files. 247 248 When the tracer is not running, it has no affect on the system, 249 but when it is running, it can cause the system to be 250 periodically non responsive. Do not run this tracer on a 251 production system. 252 253 To enable this tracer, echo in "hwlat" into the current_tracer 254 file. Every time a latency is greater than tracing_thresh, it will 255 be recorded into the ring buffer. 256 257config ENABLE_DEFAULT_TRACERS 258 bool "Trace process context switches and events" 259 depends on !GENERIC_TRACER 260 select TRACING 261 help 262 This tracer hooks to various trace points in the kernel, 263 allowing the user to pick and choose which trace point they 264 want to trace. It also includes the sched_switch tracer plugin. 265 266config FTRACE_SYSCALLS 267 bool "Trace syscalls" 268 depends on HAVE_SYSCALL_TRACEPOINTS 269 select GENERIC_TRACER 270 select KALLSYMS 271 help 272 Basic tracer to catch the syscall entry and exit events. 273 274config TRACER_SNAPSHOT 275 bool "Create a snapshot trace buffer" 276 select TRACER_MAX_TRACE 277 help 278 Allow tracing users to take snapshot of the current buffer using the 279 ftrace interface, e.g.: 280 281 echo 1 > /sys/kernel/debug/tracing/snapshot 282 cat snapshot 283 284config TRACER_SNAPSHOT_PER_CPU_SWAP 285 bool "Allow snapshot to swap per CPU" 286 depends on TRACER_SNAPSHOT 287 select RING_BUFFER_ALLOW_SWAP 288 help 289 Allow doing a snapshot of a single CPU buffer instead of a 290 full swap (all buffers). If this is set, then the following is 291 allowed: 292 293 echo 1 > /sys/kernel/debug/tracing/per_cpu/cpu2/snapshot 294 295 After which, only the tracing buffer for CPU 2 was swapped with 296 the main tracing buffer, and the other CPU buffers remain the same. 297 298 When this is enabled, this adds a little more overhead to the 299 trace recording, as it needs to add some checks to synchronize 300 recording with swaps. But this does not affect the performance 301 of the overall system. This is enabled by default when the preempt 302 or irq latency tracers are enabled, as those need to swap as well 303 and already adds the overhead (plus a lot more). 304 305config TRACE_BRANCH_PROFILING 306 bool 307 select GENERIC_TRACER 308 309choice 310 prompt "Branch Profiling" 311 default BRANCH_PROFILE_NONE 312 help 313 The branch profiling is a software profiler. It will add hooks 314 into the C conditionals to test which path a branch takes. 315 316 The likely/unlikely profiler only looks at the conditions that 317 are annotated with a likely or unlikely macro. 318 319 The "all branch" profiler will profile every if-statement in the 320 kernel. This profiler will also enable the likely/unlikely 321 profiler. 322 323 Either of the above profilers adds a bit of overhead to the system. 324 If unsure, choose "No branch profiling". 325 326config BRANCH_PROFILE_NONE 327 bool "No branch profiling" 328 help 329 No branch profiling. Branch profiling adds a bit of overhead. 330 Only enable it if you want to analyse the branching behavior. 331 Otherwise keep it disabled. 332 333config PROFILE_ANNOTATED_BRANCHES 334 bool "Trace likely/unlikely profiler" 335 select TRACE_BRANCH_PROFILING 336 help 337 This tracer profiles all likely and unlikely macros 338 in the kernel. It will display the results in: 339 340 /sys/kernel/debug/tracing/trace_stat/branch_annotated 341 342 Note: this will add a significant overhead; only turn this 343 on if you need to profile the system's use of these macros. 344 345config PROFILE_ALL_BRANCHES 346 bool "Profile all if conditionals" 347 select TRACE_BRANCH_PROFILING 348 help 349 This tracer profiles all branch conditions. Every if () 350 taken in the kernel is recorded whether it hit or miss. 351 The results will be displayed in: 352 353 /sys/kernel/debug/tracing/trace_stat/branch_all 354 355 This option also enables the likely/unlikely profiler. 356 357 This configuration, when enabled, will impose a great overhead 358 on the system. This should only be enabled when the system 359 is to be analyzed in much detail. 360endchoice 361 362config TRACING_BRANCHES 363 bool 364 help 365 Selected by tracers that will trace the likely and unlikely 366 conditions. This prevents the tracers themselves from being 367 profiled. Profiling the tracing infrastructure can only happen 368 when the likelys and unlikelys are not being traced. 369 370config BRANCH_TRACER 371 bool "Trace likely/unlikely instances" 372 depends on TRACE_BRANCH_PROFILING 373 select TRACING_BRANCHES 374 help 375 This traces the events of likely and unlikely condition 376 calls in the kernel. The difference between this and the 377 "Trace likely/unlikely profiler" is that this is not a 378 histogram of the callers, but actually places the calling 379 events into a running trace buffer to see when and where the 380 events happened, as well as their results. 381 382 Say N if unsure. 383 384config STACK_TRACER 385 bool "Trace max stack" 386 depends on HAVE_FUNCTION_TRACER 387 select FUNCTION_TRACER 388 select STACKTRACE 389 select KALLSYMS 390 help 391 This special tracer records the maximum stack footprint of the 392 kernel and displays it in /sys/kernel/debug/tracing/stack_trace. 393 394 This tracer works by hooking into every function call that the 395 kernel executes, and keeping a maximum stack depth value and 396 stack-trace saved. If this is configured with DYNAMIC_FTRACE 397 then it will not have any overhead while the stack tracer 398 is disabled. 399 400 To enable the stack tracer on bootup, pass in 'stacktrace' 401 on the kernel command line. 402 403 The stack tracer can also be enabled or disabled via the 404 sysctl kernel.stack_tracer_enabled 405 406 Say N if unsure. 407 408config BLK_DEV_IO_TRACE 409 bool "Support for tracing block IO actions" 410 depends on SYSFS 411 depends on BLOCK 412 select RELAY 413 select DEBUG_FS 414 select TRACEPOINTS 415 select GENERIC_TRACER 416 select STACKTRACE 417 help 418 Say Y here if you want to be able to trace the block layer actions 419 on a given queue. Tracing allows you to see any traffic happening 420 on a block device queue. For more information (and the userspace 421 support tools needed), fetch the blktrace tools from: 422 423 git://git.kernel.dk/blktrace.git 424 425 Tracing also is possible using the ftrace interface, e.g.: 426 427 echo 1 > /sys/block/sda/sda1/trace/enable 428 echo blk > /sys/kernel/debug/tracing/current_tracer 429 cat /sys/kernel/debug/tracing/trace_pipe 430 431 If unsure, say N. 432 433config KPROBE_EVENTS 434 depends on KPROBES 435 depends on HAVE_REGS_AND_STACK_ACCESS_API 436 bool "Enable kprobes-based dynamic events" 437 select TRACING 438 select PROBE_EVENTS 439 default y 440 help 441 This allows the user to add tracing events (similar to tracepoints) 442 on the fly via the ftrace interface. See 443 Documentation/trace/kprobetrace.txt for more details. 444 445 Those events can be inserted wherever kprobes can probe, and record 446 various register and memory values. 447 448 This option is also required by perf-probe subcommand of perf tools. 449 If you want to use perf tools, this option is strongly recommended. 450 451config UPROBE_EVENTS 452 bool "Enable uprobes-based dynamic events" 453 depends on ARCH_SUPPORTS_UPROBES 454 depends on MMU 455 depends on PERF_EVENTS 456 select UPROBES 457 select PROBE_EVENTS 458 select TRACING 459 default y 460 help 461 This allows the user to add tracing events on top of userspace 462 dynamic events (similar to tracepoints) on the fly via the trace 463 events interface. Those events can be inserted wherever uprobes 464 can probe, and record various registers. 465 This option is required if you plan to use perf-probe subcommand 466 of perf tools on user space applications. 467 468config BPF_EVENTS 469 depends on BPF_SYSCALL 470 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS 471 bool 472 default y 473 help 474 This allows the user to attach BPF programs to kprobe events. 475 476config PROBE_EVENTS 477 def_bool n 478 479config DYNAMIC_FTRACE 480 bool "enable/disable function tracing dynamically" 481 depends on FUNCTION_TRACER 482 depends on HAVE_DYNAMIC_FTRACE 483 default y 484 help 485 This option will modify all the calls to function tracing 486 dynamically (will patch them out of the binary image and 487 replace them with a No-Op instruction) on boot up. During 488 compile time, a table is made of all the locations that ftrace 489 can function trace, and this table is linked into the kernel 490 image. When this is enabled, functions can be individually 491 enabled, and the functions not enabled will not affect 492 performance of the system. 493 494 See the files in /sys/kernel/debug/tracing: 495 available_filter_functions 496 set_ftrace_filter 497 set_ftrace_notrace 498 499 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but 500 otherwise has native performance as long as no tracing is active. 501 502config DYNAMIC_FTRACE_WITH_REGS 503 def_bool y 504 depends on DYNAMIC_FTRACE 505 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS 506 507config FUNCTION_PROFILER 508 bool "Kernel function profiler" 509 depends on FUNCTION_TRACER 510 default n 511 help 512 This option enables the kernel function profiler. A file is created 513 in debugfs called function_profile_enabled which defaults to zero. 514 When a 1 is echoed into this file profiling begins, and when a 515 zero is entered, profiling stops. A "functions" file is created in 516 the trace_stats directory; this file shows the list of functions that 517 have been hit and their counters. 518 519 If in doubt, say N. 520 521config FTRACE_MCOUNT_RECORD 522 def_bool y 523 depends on DYNAMIC_FTRACE 524 depends on HAVE_FTRACE_MCOUNT_RECORD 525 526config FTRACE_SELFTEST 527 bool 528 529config FTRACE_STARTUP_TEST 530 bool "Perform a startup test on ftrace" 531 depends on GENERIC_TRACER 532 select FTRACE_SELFTEST 533 help 534 This option performs a series of startup tests on ftrace. On bootup 535 a series of tests are made to verify that the tracer is 536 functioning properly. It will do tests on all the configured 537 tracers of ftrace. 538 539config EVENT_TRACE_TEST_SYSCALLS 540 bool "Run selftest on syscall events" 541 depends on FTRACE_STARTUP_TEST 542 help 543 This option will also enable testing every syscall event. 544 It only enables the event and disables it and runs various loads 545 with the event enabled. This adds a bit more time for kernel boot 546 up since it runs this on every system call defined. 547 548 TBD - enable a way to actually call the syscalls as we test their 549 events 550 551config MMIOTRACE 552 bool "Memory mapped IO tracing" 553 depends on HAVE_MMIOTRACE_SUPPORT && PCI 554 select GENERIC_TRACER 555 help 556 Mmiotrace traces Memory Mapped I/O access and is meant for 557 debugging and reverse engineering. It is called from the ioremap 558 implementation and works via page faults. Tracing is disabled by 559 default and can be enabled at run-time. 560 561 See Documentation/trace/mmiotrace.txt. 562 If you are not helping to develop drivers, say N. 563 564config TRACING_MAP 565 bool 566 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG 567 help 568 tracing_map is a special-purpose lock-free map for tracing, 569 separated out as a stand-alone facility in order to allow it 570 to be shared between multiple tracers. It isn't meant to be 571 generally used outside of that context, and is normally 572 selected by tracers that use it. 573 574config HIST_TRIGGERS 575 bool "Histogram triggers" 576 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG 577 select TRACING_MAP 578 select TRACING 579 default n 580 help 581 Hist triggers allow one or more arbitrary trace event fields 582 to be aggregated into hash tables and dumped to stdout by 583 reading a debugfs/tracefs file. They're useful for 584 gathering quick and dirty (though precise) summaries of 585 event activity as an initial guide for further investigation 586 using more advanced tools. 587 588 See Documentation/trace/events.txt. 589 If in doubt, say N. 590 591config MMIOTRACE_TEST 592 tristate "Test module for mmiotrace" 593 depends on MMIOTRACE && m 594 help 595 This is a dumb module for testing mmiotrace. It is very dangerous 596 as it will write garbage to IO memory starting at a given address. 597 However, it should be safe to use on e.g. unused portion of VRAM. 598 599 Say N, unless you absolutely know what you are doing. 600 601config TRACEPOINT_BENCHMARK 602 bool "Add tracepoint that benchmarks tracepoints" 603 help 604 This option creates the tracepoint "benchmark:benchmark_event". 605 When the tracepoint is enabled, it kicks off a kernel thread that 606 goes into an infinite loop (calling cond_sched() to let other tasks 607 run), and calls the tracepoint. Each iteration will record the time 608 it took to write to the tracepoint and the next iteration that 609 data will be passed to the tracepoint itself. That is, the tracepoint 610 will report the time it took to do the previous tracepoint. 611 The string written to the tracepoint is a static string of 128 bytes 612 to keep the time the same. The initial string is simply a write of 613 "START". The second string records the cold cache time of the first 614 write which is not added to the rest of the calculations. 615 616 As it is a tight loop, it benchmarks as hot cache. That's fine because 617 we care most about hot paths that are probably in cache already. 618 619 An example of the output: 620 621 START 622 first=3672 [COLD CACHED] 623 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 624 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 625 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 626 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 627 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 628 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 629 630 631config RING_BUFFER_BENCHMARK 632 tristate "Ring buffer benchmark stress tester" 633 depends on RING_BUFFER 634 help 635 This option creates a test to stress the ring buffer and benchmark it. 636 It creates its own ring buffer such that it will not interfere with 637 any other users of the ring buffer (such as ftrace). It then creates 638 a producer and consumer that will run for 10 seconds and sleep for 639 10 seconds. Each interval it will print out the number of events 640 it recorded and give a rough estimate of how long each iteration took. 641 642 It does not disable interrupts or raise its priority, so it may be 643 affected by processes that are running. 644 645 If unsure, say N. 646 647config RING_BUFFER_STARTUP_TEST 648 bool "Ring buffer startup self test" 649 depends on RING_BUFFER 650 help 651 Run a simple self test on the ring buffer on boot up. Late in the 652 kernel boot sequence, the test will start that kicks off 653 a thread per cpu. Each thread will write various size events 654 into the ring buffer. Another thread is created to send IPIs 655 to each of the threads, where the IPI handler will also write 656 to the ring buffer, to test/stress the nesting ability. 657 If any anomalies are discovered, a warning will be displayed 658 and all ring buffers will be disabled. 659 660 The test runs for 10 seconds. This will slow your boot time 661 by at least 10 more seconds. 662 663 At the end of the test, statics and more checks are done. 664 It will output the stats of each per cpu buffer. What 665 was written, the sizes, what was read, what was lost, and 666 other similar details. 667 668 If unsure, say N 669 670config TRACE_ENUM_MAP_FILE 671 bool "Show enum mappings for trace events" 672 depends on TRACING 673 help 674 The "print fmt" of the trace events will show the enum names instead 675 of their values. This can cause problems for user space tools that 676 use this string to parse the raw data as user space does not know 677 how to convert the string to its value. 678 679 To fix this, there's a special macro in the kernel that can be used 680 to convert the enum into its value. If this macro is used, then the 681 print fmt strings will have the enums converted to their values. 682 683 If something does not get converted properly, this option can be 684 used to show what enums the kernel tried to convert. 685 686 This option is for debugging the enum conversions. A file is created 687 in the tracing directory called "enum_map" that will show the enum 688 names matched with their values and what trace event system they 689 belong too. 690 691 Normally, the mapping of the strings to values will be freed after 692 boot up or module load. With this option, they will not be freed, as 693 they are needed for the "enum_map" file. Enabling this option will 694 increase the memory footprint of the running kernel. 695 696 If unsure, say N 697 698config TRACING_EVENTS_GPIO 699 bool "Trace gpio events" 700 depends on GPIOLIB 701 default y 702 help 703 Enable tracing events for gpio subsystem 704 705endif # FTRACE 706 707endif # TRACING_SUPPORT 708 709