Annotation of wikisrc/tutorials/how_to_enable_and_run_dtrace.mdwn, revision 1.25
1.15 sevan 1: DTrace is a Dynamic Tracing framework developed by Sun and ported to NetBSD. It enables extensive instrumentation of the kernel and user space. See the [DTrace Community Page](http://dtrace.org) for more information.
2: Also see [DTrace Introduction](http://dtrace.org/guide/preface.html), Brendan Gregg's [DTrace one liners](http://www.brendangregg.com/DTrace/dtrace_oneliners.txt) and his notes for [DTrace on FreeBSD](https://wiki.freebsd.org/DTrace/).
1.2 schmonz 3:
1.5 wiki 4: # Current status
1.2 schmonz 5:
1.5 wiki 6: ## Supported platforms
7:
8: DTrace is a work-in-progress effort and it is for x86 systems and some arm boards.
9:
10: * i386 and amd64
1.18 sevan 11: * earm* (evbarm and armv4 based ports)
1.5 wiki 12:
13: ## Supported providers
14:
1.25 ! sevan 15: * DTrace: What to do when a script BEGINs, ENDs, ERRORs
1.5 wiki 16: * FBT: Function Boundary Tracing
1.25 ! sevan 17: * IO: Disk I/O
1.14 sevan 18: * Lockstat: Kernel Lock Statistics
1.25 ! sevan 19: * Proc: Process and thread related events
1.14 sevan 20: * Profile: Time based interrupt event source for Profiling
1.25 ! sevan 21: * SDT: Statically Defined Tracing
1.14 sevan 22: * Syscall: System Calls
23: * Syscall Linux (32bit & 64 bit): System calls via the Linux binary emulation layer
1.25 ! sevan 24: * VFS: Filesystem operations (confined to namecache events at time of writing - 8.99.22)
1.2 schmonz 25:
1.6 riastrad 26: ## TODO for netbsd-7
27:
28: * Measure effect of `options KDTRACE_HOOKS` on system performance.
29: * Determine whether the profile module works and list it here.
1.7 riastrad 30: * Integrate [[riz|users/riz]]'s syscall provider patch.
1.6 riastrad 31:
1.5 wiki 32: # How to use
33:
1.2 schmonz 34: ## Building DTrace
35:
36: You need the following options in your kernel:
37:
1.10 wiz 38: options KDTRACE_HOOKS # kernel DTrace hooks
1.2 schmonz 39: options MODULAR
40:
1.14 sevan 41: Optionally:
1.16 sevan 42:
1.14 sevan 43: options INSECURE # permit modules to loaded from user space once system has gone multiuser and securelevel has been raised.
1.2 schmonz 44:
1.14 sevan 45: A Distribution needs to be built with the options `MKDTRACE=yes` and `MKCTF=yes`, this is taken care of automatically and doesn't need to be specified manually. The list of platforms it is applied to automatically is set in `src/share/mk/bsd.own.mk`
1.2 schmonz 46:
1.14 sevan 47: Set the system to load the solaris and dtrace related modules in `/etc/modules.conf`, for a list of available modules, see `/stand/$MACHINE/$VERSION/modules/`
1.17 sevan 48:
49: For example, add the following to `/etc/modules.conf` (the file may not exist already on a system):
1.2 schmonz 50:
1.22 leot 51: - `solaris`
52: - `dtrace`
53: - `dtrace_sdt`
54: - `dtrace_fbt`
55: - `dtrace_lockstat`
56: - `dtrace_profile`
57: - `dtrace_syscall`
1.2 schmonz 58:
1.14 sevan 59: A `dtrace` device node is created automatically in `/dev/dtrace` when the modules are loaded into place.
1.2 schmonz 60:
61: List the dtrace probes
62:
63: dtrace -l
64:
65: ID PROVIDER MODULE FUNCTION NAME
66: 1 dtrace BEGIN
67: 2 dtrace END
68: 3 dtrace ERROR
69: 4 fbt netbsd AcpiAcquireGlobalLock entry
70: 5 fbt netbsd AcpiAcquireGlobalLock return
71: 6 fbt netbsd AcpiAllocateRootTable entry
72: 7 fbt netbsd AcpiAttachData entry
73: .
74: .
75: 29129 fbt solaris zfs_vop_getattr entry
76: 29130 fbt solaris zfs_vop_getattr return
77: 29131 proc create
78: 29132 proc exec
79: .
80: .
81: 29140 proc lwp_start
82: 29141 proc lwp_exit
83:
84:
1.24 gson 85: ## Running hello world
1.2 schmonz 86:
1.9 wiz 87: Put the following into the file hello.d:
1.2 schmonz 88:
89: BEGIN
90: {
91: trace("Hello world");
92: exit(0);
93: }
94:
95:
96: Run the hello world script:
97:
98: dtrace -s hello.d
99:
100: dtrace: script './hello.d' matched 1 probe
101: CPU ID FUNCTION:NAME
102: 0 1 :BEGIN Hello world
103:
104:
1.23 gson 105: ## A more complex example
106:
107: The following script traces the execution of a sleep operation
1.9 wiz 108: in the kernel. Put it in sleep.d:
1.2 schmonz 109:
110: #pragma D option flowindent
1.23 gson 111:
112: syscall::nanosleep:entry
1.2 schmonz 113: /execname == "sleep" && guard++ == 0/
114: {
115: self->traceme = 1;
116: }
1.23 gson 117:
118: fbt:::
119: /self->traceme/
120: {}
121:
122: syscall::nanosleep:return
1.2 schmonz 123: /self->traceme/
124: {
125: self->traceme = 0;
126: exit(0);
127: }
128:
1.23 gson 129: Start the script running:
130:
131: dtrace -s sleep.d
132:
133: This will take a while as the script instruments every function in the
134: kernel. When it's ready, it will print a message like "dtrace: script
135: 'sleep.d' matched 59268 probes". Then execute a "sleep 2" in another
136: shell.
1.19 sevan 137:
1.24 gson 138: ## Tools included in base
1.19 sevan 139:
140: Starting with NetBSD-8, on builds where `MKDTRACE=yes` is set, scripts from
141: [Brendan Gregg's DTrace toolkit](https://github.com/opendtrace/toolkit/) are installed in base as standard.
142:
1.21 sevan 143: At present, the following scripts are installed in `/usr/sbin`:
1.19 sevan 144:
1.22 leot 145: - `dtruss` - An implementation of the truss utility in DTrace which traces the system calls
146: made by a process
147: - `execsnoop` - snoop on execution of processes as they occur
148: - `opensnoop` - snoop on openning of files as they occur
149: - `procsystime` - print process system call time details.
CVSweb for NetBSD wikisrc <wikimaster@NetBSD.org> software: FreeBSD-CVSweb