1# SPDX-License-Identifier: GPL-2.0-only 2# 3# Architectures that offer an FUNCTION_TRACER implementation should 4# select HAVE_FUNCTION_TRACER: 5# 6 7config USER_STACKTRACE_SUPPORT 8 bool 9 10config NOP_TRACER 11 bool 12 13config HAVE_FUNCTION_TRACER 14 bool 15 help 16 See Documentation/trace/ftrace-design.rst 17 18config HAVE_FUNCTION_GRAPH_TRACER 19 bool 20 help 21 See Documentation/trace/ftrace-design.rst 22 23config HAVE_DYNAMIC_FTRACE 24 bool 25 help 26 See Documentation/trace/ftrace-design.rst 27 28config HAVE_DYNAMIC_FTRACE_WITH_REGS 29 bool 30 31config HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS 32 bool 33 34config HAVE_DYNAMIC_FTRACE_WITH_ARGS 35 bool 36 help 37 If this is set, then arguments and stack can be found from 38 the pt_regs passed into the function callback regs parameter 39 by default, even without setting the REGS flag in the ftrace_ops. 40 This allows for use of regs_get_kernel_argument() and 41 kernel_stack_pointer(). 42 43config HAVE_FTRACE_MCOUNT_RECORD 44 bool 45 help 46 See Documentation/trace/ftrace-design.rst 47 48config HAVE_SYSCALL_TRACEPOINTS 49 bool 50 help 51 See Documentation/trace/ftrace-design.rst 52 53config HAVE_FENTRY 54 bool 55 help 56 Arch supports the gcc options -pg with -mfentry 57 58config HAVE_NOP_MCOUNT 59 bool 60 help 61 Arch supports the gcc options -pg with -mrecord-mcount and -nop-mcount 62 63config HAVE_OBJTOOL_MCOUNT 64 bool 65 help 66 Arch supports objtool --mcount 67 68config HAVE_C_RECORDMCOUNT 69 bool 70 help 71 C version of recordmcount available? 72 73config BUILDTIME_MCOUNT_SORT 74 bool 75 default y 76 depends on BUILDTIME_TABLE_SORT && !S390 77 help 78 Sort the mcount_loc section at build time. 79 80config TRACER_MAX_TRACE 81 bool 82 83config TRACE_CLOCK 84 bool 85 86config RING_BUFFER 87 bool 88 select TRACE_CLOCK 89 select IRQ_WORK 90 91config EVENT_TRACING 92 select CONTEXT_SWITCH_TRACER 93 select GLOB 94 bool 95 96config CONTEXT_SWITCH_TRACER 97 bool 98 99config RING_BUFFER_ALLOW_SWAP 100 bool 101 help 102 Allow the use of ring_buffer_swap_cpu. 103 Adds a very slight overhead to tracing when enabled. 104 105config PREEMPTIRQ_TRACEPOINTS 106 bool 107 depends on TRACE_PREEMPT_TOGGLE || TRACE_IRQFLAGS 108 select TRACING 109 default y 110 help 111 Create preempt/irq toggle tracepoints if needed, so that other parts 112 of the kernel can use them to generate or add hooks to them. 113 114# All tracer options should select GENERIC_TRACER. For those options that are 115# enabled by all tracers (context switch and event tracer) they select TRACING. 116# This allows those options to appear when no other tracer is selected. But the 117# options do not appear when something else selects it. We need the two options 118# GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the 119# hiding of the automatic options. 120 121config TRACING 122 bool 123 select RING_BUFFER 124 select STACKTRACE if STACKTRACE_SUPPORT 125 select TRACEPOINTS 126 select NOP_TRACER 127 select BINARY_PRINTF 128 select EVENT_TRACING 129 select TRACE_CLOCK 130 131config GENERIC_TRACER 132 bool 133 select TRACING 134 135# 136# Minimum requirements an architecture has to meet for us to 137# be able to offer generic tracing facilities: 138# 139config TRACING_SUPPORT 140 bool 141 depends on TRACE_IRQFLAGS_SUPPORT 142 depends on STACKTRACE_SUPPORT 143 default y 144 145menuconfig FTRACE 146 bool "Tracers" 147 depends on TRACING_SUPPORT 148 default y if DEBUG_KERNEL 149 help 150 Enable the kernel tracing infrastructure. 151 152if FTRACE 153 154config BOOTTIME_TRACING 155 bool "Boot-time Tracing support" 156 depends on TRACING 157 select BOOT_CONFIG 158 help 159 Enable developer to setup ftrace subsystem via supplemental 160 kernel cmdline at boot time for debugging (tracing) driver 161 initialization and boot process. 162 163config FUNCTION_TRACER 164 bool "Kernel Function Tracer" 165 depends on HAVE_FUNCTION_TRACER 166 select KALLSYMS 167 select GENERIC_TRACER 168 select CONTEXT_SWITCH_TRACER 169 select GLOB 170 select TASKS_RCU if PREEMPTION 171 select TASKS_RUDE_RCU 172 help 173 Enable the kernel to trace every kernel function. This is done 174 by using a compiler feature to insert a small, 5-byte No-Operation 175 instruction at the beginning of every kernel function, which NOP 176 sequence is then dynamically patched into a tracer call when 177 tracing is enabled by the administrator. If it's runtime disabled 178 (the bootup default), then the overhead of the instructions is very 179 small and not measurable even in micro-benchmarks. 180 181config FUNCTION_GRAPH_TRACER 182 bool "Kernel Function Graph Tracer" 183 depends on HAVE_FUNCTION_GRAPH_TRACER 184 depends on FUNCTION_TRACER 185 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE 186 default y 187 help 188 Enable the kernel to trace a function at both its return 189 and its entry. 190 Its first purpose is to trace the duration of functions and 191 draw a call graph for each thread with some information like 192 the return value. This is done by setting the current return 193 address on the current task structure into a stack of calls. 194 195config DYNAMIC_FTRACE 196 bool "enable/disable function tracing dynamically" 197 depends on FUNCTION_TRACER 198 depends on HAVE_DYNAMIC_FTRACE 199 default y 200 help 201 This option will modify all the calls to function tracing 202 dynamically (will patch them out of the binary image and 203 replace them with a No-Op instruction) on boot up. During 204 compile time, a table is made of all the locations that ftrace 205 can function trace, and this table is linked into the kernel 206 image. When this is enabled, functions can be individually 207 enabled, and the functions not enabled will not affect 208 performance of the system. 209 210 See the files in /sys/kernel/debug/tracing: 211 available_filter_functions 212 set_ftrace_filter 213 set_ftrace_notrace 214 215 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but 216 otherwise has native performance as long as no tracing is active. 217 218config DYNAMIC_FTRACE_WITH_REGS 219 def_bool y 220 depends on DYNAMIC_FTRACE 221 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS 222 223config DYNAMIC_FTRACE_WITH_DIRECT_CALLS 224 def_bool y 225 depends on DYNAMIC_FTRACE_WITH_REGS 226 depends on HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS 227 228config DYNAMIC_FTRACE_WITH_ARGS 229 def_bool y 230 depends on DYNAMIC_FTRACE 231 depends on HAVE_DYNAMIC_FTRACE_WITH_ARGS 232 233config FUNCTION_PROFILER 234 bool "Kernel function profiler" 235 depends on FUNCTION_TRACER 236 default n 237 help 238 This option enables the kernel function profiler. A file is created 239 in debugfs called function_profile_enabled which defaults to zero. 240 When a 1 is echoed into this file profiling begins, and when a 241 zero is entered, profiling stops. A "functions" file is created in 242 the trace_stat directory; this file shows the list of functions that 243 have been hit and their counters. 244 245 If in doubt, say N. 246 247config STACK_TRACER 248 bool "Trace max stack" 249 depends on HAVE_FUNCTION_TRACER 250 select FUNCTION_TRACER 251 select STACKTRACE 252 select KALLSYMS 253 help 254 This special tracer records the maximum stack footprint of the 255 kernel and displays it in /sys/kernel/debug/tracing/stack_trace. 256 257 This tracer works by hooking into every function call that the 258 kernel executes, and keeping a maximum stack depth value and 259 stack-trace saved. If this is configured with DYNAMIC_FTRACE 260 then it will not have any overhead while the stack tracer 261 is disabled. 262 263 To enable the stack tracer on bootup, pass in 'stacktrace' 264 on the kernel command line. 265 266 The stack tracer can also be enabled or disabled via the 267 sysctl kernel.stack_tracer_enabled 268 269 Say N if unsure. 270 271config TRACE_PREEMPT_TOGGLE 272 bool 273 help 274 Enables hooks which will be called when preemption is first disabled, 275 and last enabled. 276 277config IRQSOFF_TRACER 278 bool "Interrupts-off Latency Tracer" 279 default n 280 depends on TRACE_IRQFLAGS_SUPPORT 281 select TRACE_IRQFLAGS 282 select GENERIC_TRACER 283 select TRACER_MAX_TRACE 284 select RING_BUFFER_ALLOW_SWAP 285 select TRACER_SNAPSHOT 286 select TRACER_SNAPSHOT_PER_CPU_SWAP 287 help 288 This option measures the time spent in irqs-off critical 289 sections, with microsecond accuracy. 290 291 The default measurement method is a maximum search, which is 292 disabled by default and can be runtime (re-)started 293 via: 294 295 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency 296 297 (Note that kernel size and overhead increase with this option 298 enabled. This option and the preempt-off timing option can be 299 used together or separately.) 300 301config PREEMPT_TRACER 302 bool "Preemption-off Latency Tracer" 303 default n 304 depends on PREEMPTION 305 select GENERIC_TRACER 306 select TRACER_MAX_TRACE 307 select RING_BUFFER_ALLOW_SWAP 308 select TRACER_SNAPSHOT 309 select TRACER_SNAPSHOT_PER_CPU_SWAP 310 select TRACE_PREEMPT_TOGGLE 311 help 312 This option measures the time spent in preemption-off critical 313 sections, with microsecond accuracy. 314 315 The default measurement method is a maximum search, which is 316 disabled by default and can be runtime (re-)started 317 via: 318 319 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency 320 321 (Note that kernel size and overhead increase with this option 322 enabled. This option and the irqs-off timing option can be 323 used together or separately.) 324 325config SCHED_TRACER 326 bool "Scheduling Latency Tracer" 327 select GENERIC_TRACER 328 select CONTEXT_SWITCH_TRACER 329 select TRACER_MAX_TRACE 330 select TRACER_SNAPSHOT 331 help 332 This tracer tracks the latency of the highest priority task 333 to be scheduled in, starting from the point it has woken up. 334 335config HWLAT_TRACER 336 bool "Tracer to detect hardware latencies (like SMIs)" 337 select GENERIC_TRACER 338 help 339 This tracer, when enabled will create one or more kernel threads, 340 depending on what the cpumask file is set to, which each thread 341 spinning in a loop looking for interruptions caused by 342 something other than the kernel. For example, if a 343 System Management Interrupt (SMI) takes a noticeable amount of 344 time, this tracer will detect it. This is useful for testing 345 if a system is reliable for Real Time tasks. 346 347 Some files are created in the tracing directory when this 348 is enabled: 349 350 hwlat_detector/width - time in usecs for how long to spin for 351 hwlat_detector/window - time in usecs between the start of each 352 iteration 353 354 A kernel thread is created that will spin with interrupts disabled 355 for "width" microseconds in every "window" cycle. It will not spin 356 for "window - width" microseconds, where the system can 357 continue to operate. 358 359 The output will appear in the trace and trace_pipe files. 360 361 When the tracer is not running, it has no affect on the system, 362 but when it is running, it can cause the system to be 363 periodically non responsive. Do not run this tracer on a 364 production system. 365 366 To enable this tracer, echo in "hwlat" into the current_tracer 367 file. Every time a latency is greater than tracing_thresh, it will 368 be recorded into the ring buffer. 369 370config OSNOISE_TRACER 371 bool "OS Noise tracer" 372 select GENERIC_TRACER 373 help 374 In the context of high-performance computing (HPC), the Operating 375 System Noise (osnoise) refers to the interference experienced by an 376 application due to activities inside the operating system. In the 377 context of Linux, NMIs, IRQs, SoftIRQs, and any other system thread 378 can cause noise to the system. Moreover, hardware-related jobs can 379 also cause noise, for example, via SMIs. 380 381 The osnoise tracer leverages the hwlat_detector by running a similar 382 loop with preemption, SoftIRQs and IRQs enabled, thus allowing all 383 the sources of osnoise during its execution. The osnoise tracer takes 384 note of the entry and exit point of any source of interferences, 385 increasing a per-cpu interference counter. It saves an interference 386 counter for each source of interference. The interference counter for 387 NMI, IRQs, SoftIRQs, and threads is increased anytime the tool 388 observes these interferences' entry events. When a noise happens 389 without any interference from the operating system level, the 390 hardware noise counter increases, pointing to a hardware-related 391 noise. In this way, osnoise can account for any source of 392 interference. At the end of the period, the osnoise tracer prints 393 the sum of all noise, the max single noise, the percentage of CPU 394 available for the thread, and the counters for the noise sources. 395 396 In addition to the tracer, a set of tracepoints were added to 397 facilitate the identification of the osnoise source. 398 399 The output will appear in the trace and trace_pipe files. 400 401 To enable this tracer, echo in "osnoise" into the current_tracer 402 file. 403 404config TIMERLAT_TRACER 405 bool "Timerlat tracer" 406 select OSNOISE_TRACER 407 select GENERIC_TRACER 408 help 409 The timerlat tracer aims to help the preemptive kernel developers 410 to find sources of wakeup latencies of real-time threads. 411 412 The tracer creates a per-cpu kernel thread with real-time priority. 413 The tracer thread sets a periodic timer to wakeup itself, and goes 414 to sleep waiting for the timer to fire. At the wakeup, the thread 415 then computes a wakeup latency value as the difference between 416 the current time and the absolute time that the timer was set 417 to expire. 418 419 The tracer prints two lines at every activation. The first is the 420 timer latency observed at the hardirq context before the 421 activation of the thread. The second is the timer latency observed 422 by the thread, which is the same level that cyclictest reports. The 423 ACTIVATION ID field serves to relate the irq execution to its 424 respective thread execution. 425 426 The tracer is build on top of osnoise tracer, and the osnoise: 427 events can be used to trace the source of interference from NMI, 428 IRQs and other threads. It also enables the capture of the 429 stacktrace at the IRQ context, which helps to identify the code 430 path that can cause thread delay. 431 432config MMIOTRACE 433 bool "Memory mapped IO tracing" 434 depends on HAVE_MMIOTRACE_SUPPORT && PCI 435 select GENERIC_TRACER 436 help 437 Mmiotrace traces Memory Mapped I/O access and is meant for 438 debugging and reverse engineering. It is called from the ioremap 439 implementation and works via page faults. Tracing is disabled by 440 default and can be enabled at run-time. 441 442 See Documentation/trace/mmiotrace.rst. 443 If you are not helping to develop drivers, say N. 444 445config ENABLE_DEFAULT_TRACERS 446 bool "Trace process context switches and events" 447 depends on !GENERIC_TRACER 448 select TRACING 449 help 450 This tracer hooks to various trace points in the kernel, 451 allowing the user to pick and choose which trace point they 452 want to trace. It also includes the sched_switch tracer plugin. 453 454config FTRACE_SYSCALLS 455 bool "Trace syscalls" 456 depends on HAVE_SYSCALL_TRACEPOINTS 457 select GENERIC_TRACER 458 select KALLSYMS 459 help 460 Basic tracer to catch the syscall entry and exit events. 461 462config TRACER_SNAPSHOT 463 bool "Create a snapshot trace buffer" 464 select TRACER_MAX_TRACE 465 help 466 Allow tracing users to take snapshot of the current buffer using the 467 ftrace interface, e.g.: 468 469 echo 1 > /sys/kernel/debug/tracing/snapshot 470 cat snapshot 471 472config TRACER_SNAPSHOT_PER_CPU_SWAP 473 bool "Allow snapshot to swap per CPU" 474 depends on TRACER_SNAPSHOT 475 select RING_BUFFER_ALLOW_SWAP 476 help 477 Allow doing a snapshot of a single CPU buffer instead of a 478 full swap (all buffers). If this is set, then the following is 479 allowed: 480 481 echo 1 > /sys/kernel/debug/tracing/per_cpu/cpu2/snapshot 482 483 After which, only the tracing buffer for CPU 2 was swapped with 484 the main tracing buffer, and the other CPU buffers remain the same. 485 486 When this is enabled, this adds a little more overhead to the 487 trace recording, as it needs to add some checks to synchronize 488 recording with swaps. But this does not affect the performance 489 of the overall system. This is enabled by default when the preempt 490 or irq latency tracers are enabled, as those need to swap as well 491 and already adds the overhead (plus a lot more). 492 493config TRACE_BRANCH_PROFILING 494 bool 495 select GENERIC_TRACER 496 497choice 498 prompt "Branch Profiling" 499 default BRANCH_PROFILE_NONE 500 help 501 The branch profiling is a software profiler. It will add hooks 502 into the C conditionals to test which path a branch takes. 503 504 The likely/unlikely profiler only looks at the conditions that 505 are annotated with a likely or unlikely macro. 506 507 The "all branch" profiler will profile every if-statement in the 508 kernel. This profiler will also enable the likely/unlikely 509 profiler. 510 511 Either of the above profilers adds a bit of overhead to the system. 512 If unsure, choose "No branch profiling". 513 514config BRANCH_PROFILE_NONE 515 bool "No branch profiling" 516 help 517 No branch profiling. Branch profiling adds a bit of overhead. 518 Only enable it if you want to analyse the branching behavior. 519 Otherwise keep it disabled. 520 521config PROFILE_ANNOTATED_BRANCHES 522 bool "Trace likely/unlikely profiler" 523 select TRACE_BRANCH_PROFILING 524 help 525 This tracer profiles all likely and unlikely macros 526 in the kernel. It will display the results in: 527 528 /sys/kernel/debug/tracing/trace_stat/branch_annotated 529 530 Note: this will add a significant overhead; only turn this 531 on if you need to profile the system's use of these macros. 532 533config PROFILE_ALL_BRANCHES 534 bool "Profile all if conditionals" if !FORTIFY_SOURCE 535 select TRACE_BRANCH_PROFILING 536 help 537 This tracer profiles all branch conditions. Every if () 538 taken in the kernel is recorded whether it hit or miss. 539 The results will be displayed in: 540 541 /sys/kernel/debug/tracing/trace_stat/branch_all 542 543 This option also enables the likely/unlikely profiler. 544 545 This configuration, when enabled, will impose a great overhead 546 on the system. This should only be enabled when the system 547 is to be analyzed in much detail. 548endchoice 549 550config TRACING_BRANCHES 551 bool 552 help 553 Selected by tracers that will trace the likely and unlikely 554 conditions. This prevents the tracers themselves from being 555 profiled. Profiling the tracing infrastructure can only happen 556 when the likelys and unlikelys are not being traced. 557 558config BRANCH_TRACER 559 bool "Trace likely/unlikely instances" 560 depends on TRACE_BRANCH_PROFILING 561 select TRACING_BRANCHES 562 help 563 This traces the events of likely and unlikely condition 564 calls in the kernel. The difference between this and the 565 "Trace likely/unlikely profiler" is that this is not a 566 histogram of the callers, but actually places the calling 567 events into a running trace buffer to see when and where the 568 events happened, as well as their results. 569 570 Say N if unsure. 571 572config BLK_DEV_IO_TRACE 573 bool "Support for tracing block IO actions" 574 depends on SYSFS 575 depends on BLOCK 576 select RELAY 577 select DEBUG_FS 578 select TRACEPOINTS 579 select GENERIC_TRACER 580 select STACKTRACE 581 help 582 Say Y here if you want to be able to trace the block layer actions 583 on a given queue. Tracing allows you to see any traffic happening 584 on a block device queue. For more information (and the userspace 585 support tools needed), fetch the blktrace tools from: 586 587 git://git.kernel.dk/blktrace.git 588 589 Tracing also is possible using the ftrace interface, e.g.: 590 591 echo 1 > /sys/block/sda/sda1/trace/enable 592 echo blk > /sys/kernel/debug/tracing/current_tracer 593 cat /sys/kernel/debug/tracing/trace_pipe 594 595 If unsure, say N. 596 597config KPROBE_EVENTS 598 depends on KPROBES 599 depends on HAVE_REGS_AND_STACK_ACCESS_API 600 bool "Enable kprobes-based dynamic events" 601 select TRACING 602 select PROBE_EVENTS 603 select DYNAMIC_EVENTS 604 default y 605 help 606 This allows the user to add tracing events (similar to tracepoints) 607 on the fly via the ftrace interface. See 608 Documentation/trace/kprobetrace.rst for more details. 609 610 Those events can be inserted wherever kprobes can probe, and record 611 various register and memory values. 612 613 This option is also required by perf-probe subcommand of perf tools. 614 If you want to use perf tools, this option is strongly recommended. 615 616config KPROBE_EVENTS_ON_NOTRACE 617 bool "Do NOT protect notrace function from kprobe events" 618 depends on KPROBE_EVENTS 619 depends on DYNAMIC_FTRACE 620 default n 621 help 622 This is only for the developers who want to debug ftrace itself 623 using kprobe events. 624 625 If kprobes can use ftrace instead of breakpoint, ftrace related 626 functions are protected from kprobe-events to prevent an infinite 627 recursion or any unexpected execution path which leads to a kernel 628 crash. 629 630 This option disables such protection and allows you to put kprobe 631 events on ftrace functions for debugging ftrace by itself. 632 Note that this might let you shoot yourself in the foot. 633 634 If unsure, say N. 635 636config UPROBE_EVENTS 637 bool "Enable uprobes-based dynamic events" 638 depends on ARCH_SUPPORTS_UPROBES 639 depends on MMU 640 depends on PERF_EVENTS 641 select UPROBES 642 select PROBE_EVENTS 643 select DYNAMIC_EVENTS 644 select TRACING 645 default y 646 help 647 This allows the user to add tracing events on top of userspace 648 dynamic events (similar to tracepoints) on the fly via the trace 649 events interface. Those events can be inserted wherever uprobes 650 can probe, and record various registers. 651 This option is required if you plan to use perf-probe subcommand 652 of perf tools on user space applications. 653 654config BPF_EVENTS 655 depends on BPF_SYSCALL 656 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS 657 bool 658 default y 659 help 660 This allows the user to attach BPF programs to kprobe, uprobe, and 661 tracepoint events. 662 663config DYNAMIC_EVENTS 664 def_bool n 665 666config PROBE_EVENTS 667 def_bool n 668 669config BPF_KPROBE_OVERRIDE 670 bool "Enable BPF programs to override a kprobed function" 671 depends on BPF_EVENTS 672 depends on FUNCTION_ERROR_INJECTION 673 default n 674 help 675 Allows BPF to override the execution of a probed function and 676 set a different return value. This is used for error injection. 677 678config FTRACE_MCOUNT_RECORD 679 def_bool y 680 depends on DYNAMIC_FTRACE 681 depends on HAVE_FTRACE_MCOUNT_RECORD 682 683config FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY 684 bool 685 depends on FTRACE_MCOUNT_RECORD 686 687config FTRACE_MCOUNT_USE_CC 688 def_bool y 689 depends on $(cc-option,-mrecord-mcount) 690 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY 691 depends on FTRACE_MCOUNT_RECORD 692 693config FTRACE_MCOUNT_USE_OBJTOOL 694 def_bool y 695 depends on HAVE_OBJTOOL_MCOUNT 696 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY 697 depends on !FTRACE_MCOUNT_USE_CC 698 depends on FTRACE_MCOUNT_RECORD 699 700config FTRACE_MCOUNT_USE_RECORDMCOUNT 701 def_bool y 702 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY 703 depends on !FTRACE_MCOUNT_USE_CC 704 depends on !FTRACE_MCOUNT_USE_OBJTOOL 705 depends on FTRACE_MCOUNT_RECORD 706 707config TRACING_MAP 708 bool 709 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG 710 help 711 tracing_map is a special-purpose lock-free map for tracing, 712 separated out as a stand-alone facility in order to allow it 713 to be shared between multiple tracers. It isn't meant to be 714 generally used outside of that context, and is normally 715 selected by tracers that use it. 716 717config SYNTH_EVENTS 718 bool "Synthetic trace events" 719 select TRACING 720 select DYNAMIC_EVENTS 721 default n 722 help 723 Synthetic events are user-defined trace events that can be 724 used to combine data from other trace events or in fact any 725 data source. Synthetic events can be generated indirectly 726 via the trace() action of histogram triggers or directly 727 by way of an in-kernel API. 728 729 See Documentation/trace/events.rst or 730 Documentation/trace/histogram.rst for details and examples. 731 732 If in doubt, say N. 733 734config HIST_TRIGGERS 735 bool "Histogram triggers" 736 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG 737 select TRACING_MAP 738 select TRACING 739 select DYNAMIC_EVENTS 740 select SYNTH_EVENTS 741 default n 742 help 743 Hist triggers allow one or more arbitrary trace event fields 744 to be aggregated into hash tables and dumped to stdout by 745 reading a debugfs/tracefs file. They're useful for 746 gathering quick and dirty (though precise) summaries of 747 event activity as an initial guide for further investigation 748 using more advanced tools. 749 750 Inter-event tracing of quantities such as latencies is also 751 supported using hist triggers under this option. 752 753 See Documentation/trace/histogram.rst. 754 If in doubt, say N. 755 756config TRACE_EVENT_INJECT 757 bool "Trace event injection" 758 depends on TRACING 759 help 760 Allow user-space to inject a specific trace event into the ring 761 buffer. This is mainly used for testing purpose. 762 763 If unsure, say N. 764 765config TRACEPOINT_BENCHMARK 766 bool "Add tracepoint that benchmarks tracepoints" 767 help 768 This option creates the tracepoint "benchmark:benchmark_event". 769 When the tracepoint is enabled, it kicks off a kernel thread that 770 goes into an infinite loop (calling cond_resched() to let other tasks 771 run), and calls the tracepoint. Each iteration will record the time 772 it took to write to the tracepoint and the next iteration that 773 data will be passed to the tracepoint itself. That is, the tracepoint 774 will report the time it took to do the previous tracepoint. 775 The string written to the tracepoint is a static string of 128 bytes 776 to keep the time the same. The initial string is simply a write of 777 "START". The second string records the cold cache time of the first 778 write which is not added to the rest of the calculations. 779 780 As it is a tight loop, it benchmarks as hot cache. That's fine because 781 we care most about hot paths that are probably in cache already. 782 783 An example of the output: 784 785 START 786 first=3672 [COLD CACHED] 787 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712 788 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337 789 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064 790 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411 791 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389 792 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666 793 794 795config RING_BUFFER_BENCHMARK 796 tristate "Ring buffer benchmark stress tester" 797 depends on RING_BUFFER 798 help 799 This option creates a test to stress the ring buffer and benchmark it. 800 It creates its own ring buffer such that it will not interfere with 801 any other users of the ring buffer (such as ftrace). It then creates 802 a producer and consumer that will run for 10 seconds and sleep for 803 10 seconds. Each interval it will print out the number of events 804 it recorded and give a rough estimate of how long each iteration took. 805 806 It does not disable interrupts or raise its priority, so it may be 807 affected by processes that are running. 808 809 If unsure, say N. 810 811config TRACE_EVAL_MAP_FILE 812 bool "Show eval mappings for trace events" 813 depends on TRACING 814 help 815 The "print fmt" of the trace events will show the enum/sizeof names 816 instead of their values. This can cause problems for user space tools 817 that use this string to parse the raw data as user space does not know 818 how to convert the string to its value. 819 820 To fix this, there's a special macro in the kernel that can be used 821 to convert an enum/sizeof into its value. If this macro is used, then 822 the print fmt strings will be converted to their values. 823 824 If something does not get converted properly, this option can be 825 used to show what enums/sizeof the kernel tried to convert. 826 827 This option is for debugging the conversions. A file is created 828 in the tracing directory called "eval_map" that will show the 829 names matched with their values and what trace event system they 830 belong too. 831 832 Normally, the mapping of the strings to values will be freed after 833 boot up or module load. With this option, they will not be freed, as 834 they are needed for the "eval_map" file. Enabling this option will 835 increase the memory footprint of the running kernel. 836 837 If unsure, say N. 838 839config FTRACE_RECORD_RECURSION 840 bool "Record functions that recurse in function tracing" 841 depends on FUNCTION_TRACER 842 help 843 All callbacks that attach to the function tracing have some sort 844 of protection against recursion. Even though the protection exists, 845 it adds overhead. This option will create a file in the tracefs 846 file system called "recursed_functions" that will list the functions 847 that triggered a recursion. 848 849 This will add more overhead to cases that have recursion. 850 851 If unsure, say N 852 853config FTRACE_RECORD_RECURSION_SIZE 854 int "Max number of recursed functions to record" 855 default 128 856 depends on FTRACE_RECORD_RECURSION 857 help 858 This defines the limit of number of functions that can be 859 listed in the "recursed_functions" file, that lists all 860 the functions that caused a recursion to happen. 861 This file can be reset, but the limit can not change in 862 size at runtime. 863 864config RING_BUFFER_RECORD_RECURSION 865 bool "Record functions that recurse in the ring buffer" 866 depends on FTRACE_RECORD_RECURSION 867 # default y, because it is coupled with FTRACE_RECORD_RECURSION 868 default y 869 help 870 The ring buffer has its own internal recursion. Although when 871 recursion happens it wont cause harm because of the protection, 872 but it does cause an unwanted overhead. Enabling this option will 873 place where recursion was detected into the ftrace "recursed_functions" 874 file. 875 876 This will add more overhead to cases that have recursion. 877 878config GCOV_PROFILE_FTRACE 879 bool "Enable GCOV profiling on ftrace subsystem" 880 depends on GCOV_KERNEL 881 help 882 Enable GCOV profiling on ftrace subsystem for checking 883 which functions/lines are tested. 884 885 If unsure, say N. 886 887 Note that on a kernel compiled with this config, ftrace will 888 run significantly slower. 889 890config FTRACE_SELFTEST 891 bool 892 893config FTRACE_STARTUP_TEST 894 bool "Perform a startup test on ftrace" 895 depends on GENERIC_TRACER 896 select FTRACE_SELFTEST 897 help 898 This option performs a series of startup tests on ftrace. On bootup 899 a series of tests are made to verify that the tracer is 900 functioning properly. It will do tests on all the configured 901 tracers of ftrace. 902 903config EVENT_TRACE_STARTUP_TEST 904 bool "Run selftest on trace events" 905 depends on FTRACE_STARTUP_TEST 906 default y 907 help 908 This option performs a test on all trace events in the system. 909 It basically just enables each event and runs some code that 910 will trigger events (not necessarily the event it enables) 911 This may take some time run as there are a lot of events. 912 913config EVENT_TRACE_TEST_SYSCALLS 914 bool "Run selftest on syscall events" 915 depends on EVENT_TRACE_STARTUP_TEST 916 help 917 This option will also enable testing every syscall event. 918 It only enables the event and disables it and runs various loads 919 with the event enabled. This adds a bit more time for kernel boot 920 up since it runs this on every system call defined. 921 922 TBD - enable a way to actually call the syscalls as we test their 923 events 924 925config FTRACE_SORT_STARTUP_TEST 926 bool "Verify compile time sorting of ftrace functions" 927 depends on DYNAMIC_FTRACE 928 depends on BUILDTIME_MCOUNT_SORT 929 help 930 Sorting of the mcount_loc sections that is used to find the 931 where the ftrace knows where to patch functions for tracing 932 and other callbacks is done at compile time. But if the sort 933 is not done correctly, it will cause non-deterministic failures. 934 When this is set, the sorted sections will be verified that they 935 are in deed sorted and will warn if they are not. 936 937 If unsure, say N 938 939config RING_BUFFER_STARTUP_TEST 940 bool "Ring buffer startup self test" 941 depends on RING_BUFFER 942 help 943 Run a simple self test on the ring buffer on boot up. Late in the 944 kernel boot sequence, the test will start that kicks off 945 a thread per cpu. Each thread will write various size events 946 into the ring buffer. Another thread is created to send IPIs 947 to each of the threads, where the IPI handler will also write 948 to the ring buffer, to test/stress the nesting ability. 949 If any anomalies are discovered, a warning will be displayed 950 and all ring buffers will be disabled. 951 952 The test runs for 10 seconds. This will slow your boot time 953 by at least 10 more seconds. 954 955 At the end of the test, statics and more checks are done. 956 It will output the stats of each per cpu buffer. What 957 was written, the sizes, what was read, what was lost, and 958 other similar details. 959 960 If unsure, say N 961 962config RING_BUFFER_VALIDATE_TIME_DELTAS 963 bool "Verify ring buffer time stamp deltas" 964 depends on RING_BUFFER 965 help 966 This will audit the time stamps on the ring buffer sub 967 buffer to make sure that all the time deltas for the 968 events on a sub buffer matches the current time stamp. 969 This audit is performed for every event that is not 970 interrupted, or interrupting another event. A check 971 is also made when traversing sub buffers to make sure 972 that all the deltas on the previous sub buffer do not 973 add up to be greater than the current time stamp. 974 975 NOTE: This adds significant overhead to recording of events, 976 and should only be used to test the logic of the ring buffer. 977 Do not use it on production systems. 978 979 Only say Y if you understand what this does, and you 980 still want it enabled. Otherwise say N 981 982config MMIOTRACE_TEST 983 tristate "Test module for mmiotrace" 984 depends on MMIOTRACE && m 985 help 986 This is a dumb module for testing mmiotrace. It is very dangerous 987 as it will write garbage to IO memory starting at a given address. 988 However, it should be safe to use on e.g. unused portion of VRAM. 989 990 Say N, unless you absolutely know what you are doing. 991 992config PREEMPTIRQ_DELAY_TEST 993 tristate "Test module to create a preempt / IRQ disable delay thread to test latency tracers" 994 depends on m 995 help 996 Select this option to build a test module that can help test latency 997 tracers by executing a preempt or irq disable section with a user 998 configurable delay. The module busy waits for the duration of the 999 critical section. 1000 1001 For example, the following invocation generates a burst of three 1002 irq-disabled critical sections for 500us: 1003 modprobe preemptirq_delay_test test_mode=irq delay=500 burst_size=3 1004 1005 What's more, if you want to attach the test on the cpu which the latency 1006 tracer is running on, specify cpu_affinity=cpu_num at the end of the 1007 command. 1008 1009 If unsure, say N 1010 1011config SYNTH_EVENT_GEN_TEST 1012 tristate "Test module for in-kernel synthetic event generation" 1013 depends on SYNTH_EVENTS 1014 help 1015 This option creates a test module to check the base 1016 functionality of in-kernel synthetic event definition and 1017 generation. 1018 1019 To test, insert the module, and then check the trace buffer 1020 for the generated sample events. 1021 1022 If unsure, say N. 1023 1024config KPROBE_EVENT_GEN_TEST 1025 tristate "Test module for in-kernel kprobe event generation" 1026 depends on KPROBE_EVENTS 1027 help 1028 This option creates a test module to check the base 1029 functionality of in-kernel kprobe event definition. 1030 1031 To test, insert the module, and then check the trace buffer 1032 for the generated kprobe events. 1033 1034 If unsure, say N. 1035 1036config HIST_TRIGGERS_DEBUG 1037 bool "Hist trigger debug support" 1038 depends on HIST_TRIGGERS 1039 help 1040 Add "hist_debug" file for each event, which when read will 1041 dump out a bunch of internal details about the hist triggers 1042 defined on that event. 1043 1044 The hist_debug file serves a couple of purposes: 1045 1046 - Helps developers verify that nothing is broken. 1047 1048 - Provides educational information to support the details 1049 of the hist trigger internals as described by 1050 Documentation/trace/histogram-design.rst. 1051 1052 The hist_debug output only covers the data structures 1053 related to the histogram definitions themselves and doesn't 1054 display the internals of map buckets or variable values of 1055 running histograms. 1056 1057 If unsure, say N. 1058 1059endif # FTRACE 1060