Source file src/runtime/extern.go
1 // Copyright 2009 The Go Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style 3 // license that can be found in the LICENSE file. 4 5 /* 6 Package runtime contains operations that interact with Go's runtime system, 7 such as functions to control goroutines. It also includes the low-level type information 8 used by the reflect package; see [reflect]'s documentation for the programmable 9 interface to the run-time type system. 10 11 # Environment Variables 12 13 The following environment variables ($name or %name%, depending on the host 14 operating system) control the run-time behavior of Go programs. The meanings 15 and use may change from release to release. 16 17 The GOGC variable sets the initial garbage collection target percentage. 18 A collection is triggered when the ratio of freshly allocated data to live data 19 remaining after the previous collection reaches this percentage. The default 20 is GOGC=100. Setting GOGC=off disables the garbage collector entirely. 21 [runtime/debug.SetGCPercent] allows changing this percentage at run time. 22 23 The GOMEMLIMIT variable sets a soft memory limit for the runtime. This memory limit 24 includes the Go heap and all other memory managed by the runtime, and excludes 25 external memory sources such as mappings of the binary itself, memory managed in 26 other languages, and memory held by the operating system on behalf of the Go 27 program. GOMEMLIMIT is a numeric value in bytes with an optional unit suffix. 28 The supported suffixes include B, KiB, MiB, GiB, and TiB. These suffixes 29 represent quantities of bytes as defined by the IEC 80000-13 standard. That is, 30 they are based on powers of two: KiB means 2^10 bytes, MiB means 2^20 bytes, 31 and so on. The default setting is [math.MaxInt64], which effectively disables the 32 memory limit. [runtime/debug.SetMemoryLimit] allows changing this limit at run 33 time. 34 35 The GODEBUG variable controls debugging variables within the runtime. 36 It is a comma-separated list of name=val pairs setting these named variables: 37 38 clobberfree: setting clobberfree=1 causes the garbage collector to 39 clobber the memory content of an object with bad content when it frees 40 the object. 41 42 cpu.*: cpu.all=off disables the use of all optional instruction set extensions. 43 cpu.extension=off disables use of instructions from the specified instruction set extension. 44 extension is the lower case name for the instruction set extension such as sse41 or avx 45 as listed in internal/cpu package. As an example cpu.avx=off disables runtime detection 46 and thereby use of AVX instructions. 47 48 cgocheck: setting cgocheck=0 disables all checks for packages 49 using cgo to incorrectly pass Go pointers to non-Go code. 50 Setting cgocheck=1 (the default) enables relatively cheap 51 checks that may miss some errors. A more complete, but slow, 52 cgocheck mode can be enabled using GOEXPERIMENT (which 53 requires a rebuild), see https://pkg.go.dev/internal/goexperiment for details. 54 55 disablethp: setting disablethp=1 on Linux disables transparent huge pages for the heap. 56 It has no effect on other platforms. disablethp is meant for compatibility with versions 57 of Go before 1.21, which stopped working around a Linux kernel default that can result 58 in significant memory overuse. See https://go.dev/issue/64332. This setting will be 59 removed in a future release, so operators should tweak their Linux configuration to suit 60 their needs before then. See https://go.dev/doc/gc-guide#Linux_transparent_huge_pages. 61 62 dontfreezetheworld: by default, the start of a fatal panic or throw 63 "freezes the world", preempting all threads to stop all running 64 goroutines, which makes it possible to traceback all goroutines, and 65 keeps their state close to the point of panic. Setting 66 dontfreezetheworld=1 disables this preemption, allowing goroutines to 67 continue executing during panic processing. Note that goroutines that 68 naturally enter the scheduler will still stop. This can be useful when 69 debugging the runtime scheduler, as freezetheworld perturbs scheduler 70 state and thus may hide problems. 71 72 efence: setting efence=1 causes the allocator to run in a mode 73 where each object is allocated on a unique page and addresses are 74 never recycled. 75 76 gccheckmark: setting gccheckmark=1 enables verification of the 77 garbage collector's concurrent mark phase by performing a 78 second mark pass while the world is stopped. If the second 79 pass finds a reachable object that was not found by concurrent 80 mark, the garbage collector will panic. 81 82 gcpacertrace: setting gcpacertrace=1 causes the garbage collector to 83 print information about the internal state of the concurrent pacer. 84 85 gcshrinkstackoff: setting gcshrinkstackoff=1 disables moving goroutines 86 onto smaller stacks. In this mode, a goroutine's stack can only grow. 87 88 gcstoptheworld: setting gcstoptheworld=1 disables concurrent garbage collection, 89 making every garbage collection a stop-the-world event. Setting gcstoptheworld=2 90 also disables concurrent sweeping after the garbage collection finishes. 91 92 gctrace: setting gctrace=1 causes the garbage collector to emit a single line to standard 93 error at each collection, summarizing the amount of memory collected and the 94 length of the pause. The format of this line is subject to change. Included in 95 the explanation below is also the relevant runtime/metrics metric for each field. 96 Currently, it is: 97 gc # @#s #%: #+#+# ms clock, #+#/#/#+# ms cpu, #->#-># MB, # MB goal, # MB stacks, #MB globals, # P 98 where the fields are as follows: 99 gc # the GC number, incremented at each GC 100 @#s time in seconds since program start 101 #% percentage of time spent in GC since program start 102 #+...+# wall-clock/CPU times for the phases of the GC 103 #->#-># MB heap size at GC start, at GC end, and live heap, or /gc/scan/heap:bytes 104 # MB goal goal heap size, or /gc/heap/goal:bytes 105 # MB stacks estimated scannable stack size, or /gc/scan/stack:bytes 106 # MB globals scannable global size, or /gc/scan/globals:bytes 107 # P number of processors used, or /sched/gomaxprocs:threads 108 The phases are stop-the-world (STW) sweep termination, concurrent 109 mark and scan, and STW mark termination. The CPU times 110 for mark/scan are broken down in to assist time (GC performed in 111 line with allocation), background GC time, and idle GC time. 112 If the line ends with "(forced)", this GC was forced by a 113 runtime.GC() call. 114 115 harddecommit: setting harddecommit=1 causes memory that is returned to the OS to 116 also have protections removed on it. This is the only mode of operation on Windows, 117 but is helpful in debugging scavenger-related issues on other platforms. Currently, 118 only supported on Linux. 119 120 inittrace: setting inittrace=1 causes the runtime to emit a single line to standard 121 error for each package with init work, summarizing the execution time and memory 122 allocation. No information is printed for inits executed as part of plugin loading 123 and for packages without both user defined and compiler generated init work. 124 The format of this line is subject to change. Currently, it is: 125 init # @#ms, # ms clock, # bytes, # allocs 126 where the fields are as follows: 127 init # the package name 128 @# ms time in milliseconds when the init started since program start 129 # clock wall-clock time for package initialization work 130 # bytes memory allocated on the heap 131 # allocs number of heap allocations 132 133 madvdontneed: setting madvdontneed=0 will use MADV_FREE 134 instead of MADV_DONTNEED on Linux when returning memory to the 135 kernel. This is more efficient, but means RSS numbers will 136 drop only when the OS is under memory pressure. On the BSDs and 137 Illumos/Solaris, setting madvdontneed=1 will use MADV_DONTNEED instead 138 of MADV_FREE. This is less efficient, but causes RSS numbers to drop 139 more quickly. 140 141 memprofilerate: setting memprofilerate=X will update the value of runtime.MemProfileRate. 142 When set to 0 memory profiling is disabled. Refer to the description of 143 MemProfileRate for the default value. 144 145 profstackdepth: profstackdepth=128 (the default) will set the maximum stack 146 depth used by all pprof profilers except for the CPU profiler to 128 frames. 147 Stack traces that exceed this limit will be truncated to the limit starting 148 from the leaf frame. Setting profstackdepth to any value above 1024 will 149 silently default to 1024. Future versions of Go may remove this limitation 150 and extend profstackdepth to apply to the CPU profiler and execution tracer. 151 152 pagetrace: setting pagetrace=/path/to/file will write out a trace of page events 153 that can be viewed, analyzed, and visualized using the x/debug/cmd/pagetrace tool. 154 Build your program with GOEXPERIMENT=pagetrace to enable this functionality. Do not 155 enable this functionality if your program is a setuid binary as it introduces a security 156 risk in that scenario. Currently not supported on Windows, plan9 or js/wasm. Setting this 157 option for some applications can produce large traces, so use with care. 158 159 panicnil: setting panicnil=1 disables the runtime error when calling panic with nil 160 interface value or an untyped nil. 161 162 runtimecontentionstacks: setting runtimecontentionstacks=1 enables inclusion of call stacks 163 related to contention on runtime-internal locks in the "mutex" profile, subject to the 164 MutexProfileFraction setting. When runtimecontentionstacks=0, contention on 165 runtime-internal locks will report as "runtime._LostContendedRuntimeLock". When 166 runtimecontentionstacks=1, the call stacks will correspond to the unlock call that released 167 the lock. But instead of the value corresponding to the amount of contention that call 168 stack caused, it corresponds to the amount of time the caller of unlock had to wait in its 169 original call to lock. A future release is expected to align those and remove this setting. 170 171 invalidptr: invalidptr=1 (the default) causes the garbage collector and stack 172 copier to crash the program if an invalid pointer value (for example, 1) 173 is found in a pointer-typed location. Setting invalidptr=0 disables this check. 174 This should only be used as a temporary workaround to diagnose buggy code. 175 The real fix is to not store integers in pointer-typed locations. 176 177 sbrk: setting sbrk=1 replaces the memory allocator and garbage collector 178 with a trivial allocator that obtains memory from the operating system and 179 never reclaims any memory. 180 181 scavtrace: setting scavtrace=1 causes the runtime to emit a single line to standard 182 error, roughly once per GC cycle, summarizing the amount of work done by the 183 scavenger as well as the total amount of memory returned to the operating system 184 and an estimate of physical memory utilization. The format of this line is subject 185 to change, but currently it is: 186 scav # KiB work (bg), # KiB work (eager), # KiB total, #% util 187 where the fields are as follows: 188 # KiB work (bg) the amount of memory returned to the OS in the background since 189 the last line 190 # KiB work (eager) the amount of memory returned to the OS eagerly since the last line 191 # KiB now the amount of address space currently returned to the OS 192 #% util the fraction of all unscavenged heap memory which is in-use 193 If the line ends with "(forced)", then scavenging was forced by a 194 debug.FreeOSMemory() call. 195 196 scheddetail: setting schedtrace=X and scheddetail=1 causes the scheduler to emit 197 detailed multiline info every X milliseconds, describing state of the scheduler, 198 processors, threads and goroutines. 199 200 schedtrace: setting schedtrace=X causes the scheduler to emit a single line to standard 201 error every X milliseconds, summarizing the scheduler state. 202 203 tracebackancestors: setting tracebackancestors=N extends tracebacks with the stacks at 204 which goroutines were created, where N limits the number of ancestor goroutines to 205 report. This also extends the information returned by runtime.Stack. 206 Setting N to 0 will report no ancestry information. 207 208 tracefpunwindoff: setting tracefpunwindoff=1 forces the execution tracer to 209 use the runtime's default stack unwinder instead of frame pointer unwinding. 210 This increases tracer overhead, but could be helpful as a workaround or for 211 debugging unexpected regressions caused by frame pointer unwinding. 212 213 traceadvanceperiod: the approximate period in nanoseconds between trace generations. Only 214 applies if a program is built with GOEXPERIMENT=exectracer2. Used primarily for testing 215 and debugging the execution tracer. 216 217 tracecheckstackownership: setting tracecheckstackownership=1 enables a debug check in the 218 execution tracer to double-check stack ownership before taking a stack trace. 219 220 asyncpreemptoff: asyncpreemptoff=1 disables signal-based 221 asynchronous goroutine preemption. This makes some loops 222 non-preemptible for long periods, which may delay GC and 223 goroutine scheduling. This is useful for debugging GC issues 224 because it also disables the conservative stack scanning used 225 for asynchronously preempted goroutines. 226 227 The [net] and [net/http] packages also refer to debugging variables in GODEBUG. 228 See the documentation for those packages for details. 229 230 The GOMAXPROCS variable limits the number of operating system threads that 231 can execute user-level Go code simultaneously. There is no limit to the number of threads 232 that can be blocked in system calls on behalf of Go code; those do not count against 233 the GOMAXPROCS limit. This package's [GOMAXPROCS] function queries and changes 234 the limit. 235 236 The GORACE variable configures the race detector, for programs built using -race. 237 See the [Race Detector article] for details. 238 239 The GOTRACEBACK variable controls the amount of output generated when a Go 240 program fails due to an unrecovered panic or an unexpected runtime condition. 241 By default, a failure prints a stack trace for the current goroutine, 242 eliding functions internal to the run-time system, and then exits with exit code 2. 243 The failure prints stack traces for all goroutines if there is no current goroutine 244 or the failure is internal to the run-time. 245 GOTRACEBACK=none omits the goroutine stack traces entirely. 246 GOTRACEBACK=single (the default) behaves as described above. 247 GOTRACEBACK=all adds stack traces for all user-created goroutines. 248 GOTRACEBACK=system is like “all” but adds stack frames for run-time functions 249 and shows goroutines created internally by the run-time. 250 GOTRACEBACK=crash is like “system” but crashes in an operating system-specific 251 manner instead of exiting. For example, on Unix systems, the crash raises 252 SIGABRT to trigger a core dump. 253 GOTRACEBACK=wer is like “crash” but doesn't disable Windows Error Reporting (WER). 254 For historical reasons, the GOTRACEBACK settings 0, 1, and 2 are synonyms for 255 none, all, and system, respectively. 256 The [runtime/debug.SetTraceback] function allows increasing the 257 amount of output at run time, but it cannot reduce the amount below that 258 specified by the environment variable. 259 260 The GOARCH, GOOS, GOPATH, and GOROOT environment variables complete 261 the set of Go environment variables. They influence the building of Go programs 262 (see [cmd/go] and [go/build]). 263 GOARCH, GOOS, and GOROOT are recorded at compile time and made available by 264 constants or functions in this package, but they do not influence the execution 265 of the run-time system. 266 267 # Security 268 269 On Unix platforms, Go's runtime system behaves slightly differently when a 270 binary is setuid/setgid or executed with setuid/setgid-like properties, in order 271 to prevent dangerous behaviors. On Linux this is determined by checking for the 272 AT_SECURE flag in the auxiliary vector, on the BSDs and Solaris/Illumos it is 273 determined by checking the issetugid syscall, and on AIX it is determined by 274 checking if the uid/gid match the effective uid/gid. 275 276 When the runtime determines the binary is setuid/setgid-like, it does three main 277 things: 278 - The standard input/output file descriptors (0, 1, 2) are checked to be open. 279 If any of them are closed, they are opened pointing at /dev/null. 280 - The value of the GOTRACEBACK environment variable is set to 'none'. 281 - When a signal is received that terminates the program, or the program 282 encounters an unrecoverable panic that would otherwise override the value 283 of GOTRACEBACK, the goroutine stack, registers, and other memory related 284 information are omitted. 285 286 [Race Detector article]: https://go.dev/doc/articles/race_detector 287 */ 288 package runtime 289 290 import ( 291 "internal/goarch" 292 "internal/goos" 293 ) 294 295 // Caller reports file and line number information about function invocations on 296 // the calling goroutine's stack. The argument skip is the number of stack frames 297 // to ascend, with 0 identifying the caller of Caller. (For historical reasons the 298 // meaning of skip differs between Caller and [Callers].) The return values report the 299 // program counter, file name, and line number within the file of the corresponding 300 // call. The boolean ok is false if it was not possible to recover the information. 301 func Caller(skip int) (pc uintptr, file string, line int, ok bool) { 302 rpc := make([]uintptr, 1) 303 n := callers(skip+1, rpc) 304 if n < 1 { 305 return 306 } 307 frame, _ := CallersFrames(rpc).Next() 308 return frame.PC, frame.File, frame.Line, frame.PC != 0 309 } 310 311 // Callers fills the slice pc with the return program counters of function invocations 312 // on the calling goroutine's stack. The argument skip is the number of stack frames 313 // to skip before recording in pc, with 0 identifying the frame for Callers itself and 314 // 1 identifying the caller of Callers. 315 // It returns the number of entries written to pc. 316 // 317 // To translate these PCs into symbolic information such as function 318 // names and line numbers, use [CallersFrames]. CallersFrames accounts 319 // for inlined functions and adjusts the return program counters into 320 // call program counters. Iterating over the returned slice of PCs 321 // directly is discouraged, as is using [FuncForPC] on any of the 322 // returned PCs, since these cannot account for inlining or return 323 // program counter adjustment. 324 func Callers(skip int, pc []uintptr) int { 325 // runtime.callers uses pc.array==nil as a signal 326 // to print a stack trace. Pick off 0-length pc here 327 // so that we don't let a nil pc slice get to it. 328 if len(pc) == 0 { 329 return 0 330 } 331 return callers(skip, pc) 332 } 333 334 var defaultGOROOT string // set by cmd/link 335 336 // GOROOT returns the root of the Go tree. It uses the 337 // GOROOT environment variable, if set at process start, 338 // or else the root used during the Go build. 339 func GOROOT() string { 340 s := gogetenv("GOROOT") 341 if s != "" { 342 return s 343 } 344 return defaultGOROOT 345 } 346 347 // buildVersion is the Go tree's version string at build time. 348 // 349 // If any GOEXPERIMENTs are set to non-default values, it will include 350 // "X:<GOEXPERIMENT>". 351 // 352 // This is set by the linker. 353 // 354 // This is accessed by "go version <binary>". 355 var buildVersion string 356 357 // Version returns the Go tree's version string. 358 // It is either the commit hash and date at the time of the build or, 359 // when possible, a release tag like "go1.3". 360 func Version() string { 361 return buildVersion 362 } 363 364 // GOOS is the running program's operating system target: 365 // one of darwin, freebsd, linux, and so on. 366 // To view possible combinations of GOOS and GOARCH, run "go tool dist list". 367 const GOOS string = goos.GOOS 368 369 // GOARCH is the running program's architecture target: 370 // one of 386, amd64, arm, s390x, and so on. 371 const GOARCH string = goarch.GOARCH 372