2 .\" $NetBSD: ptrace.2,v 1.2 1995/02/27 12:35:37 cgd Exp $
4 .\" This file is in the public domain.
10 .Nd process tracing and debugging
17 .Fn ptrace "int request" "pid_t pid" "caddr_t addr" "int data"
22 provides tracing and debugging facilities.
31 The tracing process must first attach to the traced process, and then
34 system calls to control the execution of the process, as well as access
35 process memory and register state.
36 For the duration of the tracing session, the traced process will be
38 with its parent process ID (and resulting behavior)
39 changed to the tracing process.
40 It is permissible for a tracing process to attach to more than one
41 other process at a time.
42 When the tracing process has completed its work, it must detach the
43 traced process; if a tracing process exits without first detaching all
44 processes it has attached, those processes will be killed.
46 Most of the time, the traced process runs normally, but when it
51 The tracing process is expected to notice this via
55 signal, examine the state of the stopped process, and cause it to
56 terminate or continue as appropriate.
57 The signal may be a normal process signal, generated as a result of
58 traced process behavior, or use of the
60 system call; alternatively, it may be generated by the tracing facility
61 as a result of attaching, stepping by the tracing
63 or an event in the traced process.
64 The tracing process may choose to intercept the signal, using it to
65 observe process behavior (such as
67 or forward the signal to the process if appropriate.
71 is the mechanism by which all this happens.
73 A traced process may report additional signal stops corresponding to
74 events in the traced process.
75 These additional signal stops are reported as
80 The tracing process can use the
82 request to determine which events are associated with a
87 Note that multiple events may be associated with a single signal.
88 For example, events indicated by the
93 flags are also reported as a system call exit event
95 The signal stop for a new child process enabled via
100 All other additional signal stops use
103 Each traced process has a tracing event mask.
104 An event in the traced process only reports a
105 signal stop if the corresponding flag is set in the tracing event mask.
106 The current set of tracing event flags include:
107 .Bl -tag -width "Dv PTRACE_SYSCALL"
109 Report a stop for a successful invocation of
111 This event is indicated by the
116 .Vt "struct ptrace_lwpinfo" .
118 Report a stop on each system call entry.
119 This event is indicated by the
124 .Vt "struct ptrace_lwpinfo" .
126 Report a stop on each system call exit.
127 This event is indicated by the
132 .Vt "struct ptrace_lwpinfo" .
133 .It Dv PTRACE_SYSCALL
134 Report stops for both system call entry and exit.
136 This event flag controls tracing for new child processes of a traced process.
138 When this event flag is enabled,
139 new child processes will enable tracing and stop before executing their
141 The new child process will include the
146 .Vt "struct ptrace_lwpinfo" .
147 The traced process will report a stop that includes the
150 The process ID of the new child process will also be present in the
153 .Vt "struct ptrace_lwpinfo" .
154 If the new child process was created via
156 the traced process's stop will also include the
159 Note that new child processes will be attached with the default
161 they do not inherit the event mask of the traced process.
163 When this event flag is not enabled,
164 new child processes will execute without tracing enabled.
166 This event flag controls tracing of LWP
168 creation and destruction.
169 When this event is enabled,
170 new LWPs will stop and report an event with
172 set before executing their first instruction,
173 and exiting LWPs will stop and report an event with
175 set before completing their termination.
177 Note that new processes do not report an event for the creation of their
179 and exiting processes do not report an event for the termination of the
182 Report a stop event when a parent process resumes after a
185 When a thread in the traced process creates a new child process via
187 the stop that reports
191 occurs just after the child process is created,
192 but before the thread waits for the child process to stop sharing process
194 If a debugger is not tracing the new child process,
195 it must ensure that no breakpoints are enabled in the shared process
196 memory before detaching from the new child process.
197 This means that no breakpoints are enabled in the parent process either.
201 flag enables a new stop that indicates when the new child process stops
202 sharing the process memory of the parent process.
203 A debugger can reinsert breakpoints in the parent process and resume it
204 in response to this event.
205 This event is indicated by setting the
206 .Dv PL_FLAG_VFORK_DONE
210 The default tracing event mask when attaching to a process via
218 All other event flags are disabled.
222 argument specifies what operation is being performed; the meaning of
223 the rest of the arguments depends on the operation, but except for one
224 special case noted below, all
226 calls are made by the tracing process, and the
228 argument specifies the process ID of the traced process
229 or a corresponding thread ID.
234 .Bl -tag -width "Dv PT_GET_EVENT_MASK"
236 This request is the only one used by the traced process; it declares
237 that the process expects to be traced by its parent.
238 All the other arguments are ignored.
239 (If the parent process does not expect to trace the child, it will
240 probably be rather confused by the results; once the traced process
241 stops, it cannot be made to continue except via
243 When a process has used this request and calls
245 or any of the routines built on it
248 it will stop before executing the first instruction of the new image.
249 Also, any setuid or setgid bits on the executable being executed will
251 If the child was created by
257 flag specified, the debugging events are reported to the parent
261 .It Dv PT_READ_I , Dv PT_READ_D
262 These requests read a single
264 of data from the traced process's address space.
267 has allowed for machines with distinct address spaces for instruction
268 and data, which is why there are two requests: conceptually,
270 reads from the instruction space and
272 reads from the data space.
275 implementation, these two requests are completely identical.
278 argument specifies the address
279 (in the traced process's virtual address space)
280 at which the read is to be done.
281 This address does not have to meet any alignment constraints.
282 The value read is returned as the return value from
284 .It Dv PT_WRITE_I , Dv PT_WRITE_D
285 These requests parallel
289 except that they write rather than read.
292 argument supplies the value to be written.
294 This request allows reading and writing arbitrary amounts of data in
295 the traced process's address space.
298 argument specifies a pointer to a
299 .Vt "struct ptrace_io_desc" ,
300 which is defined as follows:
302 struct ptrace_io_desc {
303 int piod_op; /* I/O operation */
304 void *piod_offs; /* child offset */
305 void *piod_addr; /* parent offset */
306 size_t piod_len; /* request length */
310 * Operations in piod_op.
312 #define PIOD_READ_D 1 /* Read from D space */
313 #define PIOD_WRITE_D 2 /* Write to D space */
314 #define PIOD_READ_I 3 /* Read from I space */
315 #define PIOD_WRITE_I 4 /* Write to I space */
321 The actual number of bytes read or written is stored in
325 The traced process continues execution.
329 is an address specifying the place where execution is to be resumed
330 (a new value for the program counter),
332 .Po Vt caddr_t Pc Ns 1
333 to indicate that execution is to pick up where it left off.
337 provides a signal number to be delivered to the traced process as it
338 resumes execution, or 0 if no signal is to be sent.
340 The traced process is single stepped one instruction.
345 .Po Vt caddr_t Pc Ns 1 .
349 provides a signal number to be delivered to the traced process as it
350 resumes execution, or 0 if no signal is to be sent.
352 The traced process terminates, as if
356 given as the signal to be delivered.
358 This request allows a process to gain control of an otherwise
359 unrelated process and begin tracing it.
360 It does not need any cooperation from the to-be-traced process.
364 specifies the process ID of the to-be-traced process, and the other
365 two arguments are ignored.
366 This request requires that the target process must have the same real
367 UID as the tracing process, and that it must not be executing a setuid
368 or setgid executable.
369 (If the tracing process is running as root, these restrictions do not
371 The tracing process will see the newly-traced process stop and may
372 then control it as if it had been traced all along.
374 This request is like PT_CONTINUE, except that it does not allow
375 specifying an alternate place to continue execution, and after it
376 succeeds, the traced process is no longer traced and continues
379 This request reads the traced process's machine registers into the
388 This request is the converse of
390 it loads the traced process's machine registers from the
399 This request reads the traced process's floating-point registers into
409 This request is the converse of
411 it loads the traced process's floating-point registers from the
420 This request reads the traced process's debug registers into
430 This request is the converse of
432 it loads the traced process's debug registers from the
441 This request can be used to obtain information about the kernel thread,
442 also known as light-weight process, that caused the traced process to stop.
445 argument specifies a pointer to a
446 .Vt "struct ptrace_lwpinfo" ,
447 which is defined as follows:
449 struct ptrace_lwpinfo {
455 siginfo_t pl_siginfo;
456 char pl_tdname[MAXCOMLEN + 1];
458 u_int pl_syscall_code;
459 u_int pl_syscall_narg;
465 argument is to be set to the size of the structure known to the caller.
466 This allows the structure to grow without affecting older programs.
469 .Vt "struct ptrace_lwpinfo"
470 have the following meaning:
471 .Bl -tag -width indent -compact
475 Event that caused the stop.
476 Currently defined events are:
477 .Bl -tag -width "Dv PL_EVENT_SIGNAL" -compact
480 .It Dv PL_EVENT_SIGNAL
481 Thread stopped due to the pending signal
484 Flags that specify additional details about observed stop.
485 Currently defined flags are:
486 .Bl -tag -width indent -compact
488 The thread stopped due to system call entry, right after the kernel is entered.
489 The debugger may examine syscall arguments that are stored in memory and
490 registers according to the ABI of the current process, and modify them,
493 The thread is stopped immediately before syscall is returning to the usermode.
494 The debugger may examine system call return values in the ABI-defined registers
499 is set, this flag may be additionally specified to inform that the
500 program being executed by debuggee process has been changed by successful
501 execution of a system call from the
508 .Vt "struct ptrace_lwpinfo"
509 contains valid information.
510 .It Dv PL_FLAG_FORKED
511 Indicates that the process is returning from a call to
513 that created a new child process.
514 The process identifier of the new process is available in the
517 .Vt "struct ptrace_lwpinfo" .
519 The flag is set for first event reported from a new child which is
520 automatically attached when
524 This flag is set for the first event reported from a new LWP when
527 It is reported along with
529 .It Dv PL_FLAG_EXITED
530 This flag is set for the last event reported by an exiting LWP when
533 Note that this event is not reported when the last LWP in a process exits.
534 The termination of the last thread is reported via a normal process exit
536 .It Dv PL_FLAG_VFORKED
537 Indicates that the thread is returning from a call to
539 that created a new child process.
540 This flag is set in addition to
542 .It Dv PL_FLAG_VFORK_DONE
543 Indicates that the thread has resumed after a child process created via
545 has stopped sharing its address space with the traced process.
548 The current signal mask of the LWP
550 The current pending set of signals for the LWP.
551 Note that signals that are delivered to the process would not appear
552 on an LWP siglist until the thread is selected for delivery.
554 The siginfo that accompanies the signal pending.
562 The name of the thread.
564 The process identifier of the new child process.
571 .It Va pl_syscall_code
572 The ABI-specific identifier of the current system call.
573 Note that for indirect system calls this field reports the indirected
581 .It Va pl_syscall_narg
582 The number of arguments passed to the current system call not counting
583 the system call identifier.
584 Note that for indirect system calls this field reports the arguments
585 passed to the indirected system call.
594 This request returns the number of kernel threads associated with the
597 This request can be used to get the current thread list.
598 A pointer to an array of type
602 with the array size specified by
604 The return value from
606 is the count of array entries filled in.
608 This request will turn on single stepping of the specified process.
609 Stepping is automatically disabled when a single step trap is caught.
611 This request will turn off single stepping of the specified process.
613 This request will suspend the specified thread.
615 This request will resume the specified thread.
617 This request will set the
619 event flag to trace all future system call entries and continue the process.
624 arguments are used the same as for
627 This request will set the
629 event flag to trace all future system call exits and continue the process.
634 arguments are used the same as for
637 This request will set the
639 event flag to trace all future system call entries and exits and continue
645 arguments are used the same as for
647 .It Dv PT_GET_SC_ARGS
648 For the thread which is stopped in either
652 state, that is, on entry or exit to a syscall,
653 this request fetches the syscall arguments.
655 The arguments are copied out into the buffer pointed to by the
657 pointer, sequentially.
658 Each syscall argument is stored as the machine word.
659 Kernel copies out as many arguments as the syscall accepts,
663 .Vt struct ptrace_lwpinfo ,
664 but not more than the
666 bytes in total are copied.
668 Fetch the system call return values on exit from a syscall.
669 This request is only valid for threads stopped in a syscall
675 argument specifies a pointer to a
676 .Vt "struct ptrace_sc_ret" ,
677 which is defined as follows:
679 struct ptrace_sc_ret {
680 register_t sr_retval[2];
687 argument is set to the size of the structure.
689 If the system call completed successfully,
691 is set to zero and the return values of the system call are saved in
693 If the system call failed to execute,
695 field is set to a positive
698 If the system call completed in an unusual fashion,
700 is set to a negative value:
701 .Bl -tag -width Dv EJUSTRETURN -compact
703 System call will be restarted.
705 System call completed sucessfully but did not set a return value
712 .It Dv PT_FOLLOW_FORK
713 This request controls tracing for new child processes of a traced process.
718 is set in the traced process's event tracing mask.
723 is cleared from the traced process's event tracing mask.
725 This request controls tracing of LWP creation and destruction.
730 is set in the traced process's event tracing mask.
735 is cleared from the traced process's event tracing mask.
736 .It Dv PT_GET_EVENT_MASK
737 This request reads the traced process's event tracing mask into the
738 integer pointed to by
740 The size of the integer must be passed in
742 .It Dv PT_SET_EVENT_MASK
743 This request sets the traced process's event tracing mask from the
744 integer pointed to by
746 The size of the integer must be passed in
748 .It Dv PT_VM_TIMESTAMP
749 This request returns the generation number or timestamp of the memory map of
750 the traced process as the return value from
752 This provides a low-cost way for the tracing process to determine if the
753 VM map changed since the last time this request was made.
755 This request is used to iterate over the entries of the VM map of the traced
759 argument specifies a pointer to a
760 .Vt "struct ptrace_vm_entry" ,
761 which is defined as follows:
763 struct ptrace_vm_entry {
777 The first entry is returned by setting
780 Subsequent entries are returned by leaving
782 unmodified from the value returned by previous requests.
785 field can be used to detect changes to the VM map while iterating over the
787 The tracing process can then take appropriate action, such as restarting.
790 to a non-zero value on entry, the pathname of the backing object is returned
791 in the buffer pointed to by
793 provided the entry is backed by a vnode.
796 field is updated with the actual length of the pathname (including the
797 terminating null character).
800 field is the offset within the backing object at which the range starts.
801 The range is located in the VM space at
811 .Sh ARM MACHINE-SPECIFIC REQUESTS
812 .Bl -tag -width "Dv PT_SETVFPREGS"
816 machine state in the buffer pointed to by
825 machine state from the buffer pointed to by
832 .Sh x86 MACHINE-SPECIFIC REQUESTS
833 .Bl -tag -width "Dv PT_GETXSTATE_INFO"
835 Copy the XMM FPU state into the buffer pointed to by the
838 The buffer has the same layout as the 32-bit save buffer for the
842 This request is only valid for i386 programs, both on native 32-bit
843 systems and on amd64 kernels.
844 For 64-bit amd64 programs, the XMM state is reported as part of
845 the FPU state returned by the
853 Load the XMM FPU state for the thread from the buffer pointed to
856 The buffer has the same layout as the 32-bit load buffer for the
862 this request is only valid for i386 programs.
867 .It Dv PT_GETXSTATE_INFO
868 Report which XSAVE FPU extensions are supported by the CPU
869 and allowed in userspace programs.
872 argument must point to a variable of type
873 .Vt struct ptrace_xstate_info ,
874 which contains the information on the request return.
875 .Vt struct ptrace_xstate_info
876 is defined as follows:
878 struct ptrace_xstate_info {
885 field is a bitmask of the currently enabled extensions.
886 The meaning of the bits is defined in the Intel and AMD
887 processor documentation.
890 field reports the length of the XSAVE area for storing the hardware
891 state for currently enabled extensions in the format defined by the x86
897 argument value must be equal to the size of the
898 .Vt struct ptrace_xstate_info .
900 Return the content of the XSAVE area for the thread.
903 argument points to the buffer where the content is copied, and the
905 argument specifies the size of the buffer.
906 The kernel copies out as much content as allowed by the buffer size.
907 The buffer layout is specified by the layout of the save area for the
911 Load the XSAVE state for the thread from the buffer specified by the
914 The buffer size is passed in the
917 The buffer must be at least as large as the
921 to allow the complete x87 FPU and XMM state load.
922 It must not be larger than the XSAVE state length, as reported by the
925 .Vt struct ptrace_xstate_info
927 .Dv PT_GETXSTATE_INFO
929 Layout of the buffer is identical to the layout of the load area for the
933 Return the value of the base used when doing segmented
934 memory addressing using the %fs segment register.
937 argument points to an
939 variable where the base value is stored.
947 request, but returns the base for the %gs segment register.
949 Set the base for the %fs segment register to the value pointed to
956 variable containing the new base.
964 request, but sets the base for the %gs segment register.
966 .Sh PowerPC MACHINE-SPECIFIC REQUESTS
967 .Bl -tag -width "Dv PT_SETVRREGS"
971 machine state in the buffer pointed to by
980 machine state from the buffer pointed to by
987 Return doubleword 1 of the thread's
989 registers VSR0-VSR31 in the buffer pointed to by
996 Set doubleword 1 of the thread's
998 registers VSR0-VSR31 from the buffer pointed to by
1003 argument is ignored.
1006 Additionally, other machine-specific requests can exist.
1008 Most requests return 0 on success and \-1 on error.
1009 Some requests can cause
1013 as a non-error value, among them are
1017 which return the value read from the process memory on success.
1020 can be set to 0 before the call and checked afterwards.
1024 implementation always sets
1026 to 0 before calling into the kernel, both for historic reasons and for
1027 consistency with other operating systems.
1028 It is recommended to assign zero to
1030 explicitly for forward compatibility.
1034 system call may fail if:
1037 .Bl -bullet -compact
1039 No process having the specified process ID exists.
1042 .Bl -bullet -compact
1044 A process attempted to use
1051 was not one of the legal requests.
1058 was neither 0 nor a legal signal number.
1067 was attempted on a process with no valid register set.
1068 (This is normally true only of system processes.)
1071 was given an invalid value for
1073 This can also be caused by changes to the VM map of the process.
1079 was less than or equal to zero, or larger than the
1081 structure known to the kernel.
1085 provided to the x86-specific
1086 .Dv PT_GETXSTATE_INFO
1087 request was not equal to the size of the
1088 .Vt struct ptrace_xstate_info .
1092 provided to the x86-specific
1094 request was less than the size of the x87 plus the XMM save area.
1098 provided to the x86-specific
1100 request was larger than returned in the
1103 .Vt struct ptrace_xstate_info
1105 .Dv PT_GETXSTATE_INFO
1108 The base value, provided to the amd64-specific requests
1112 pointed outside of the valid user address space.
1113 This error will not occur in 32-bit programs.
1116 .Bl -bullet -compact
1119 was attempted on a process that was already being traced.
1121 A request attempted to manipulate a process that was being traced by
1122 some process other than the one making the request.
1127 specified a process that was not stopped.
1130 .Bl -bullet -compact
1135 attempted to manipulate a process that was not being traced at all.
1137 An attempt was made to use
1139 on a process in violation of the requirements listed under
1144 .Bl -bullet -compact
1147 previously returned the last entry of the memory map.
1148 No more entries exist.
1150 .It Bq Er ENAMETOOLONG
1151 .Bl -bullet -compact
1154 cannot return the pathname of the backing object because the buffer is not big
1157 holds the minimum buffer size required on return.
1165 .Xr i386_clr_watch 3 ,
1166 .Xr i386_set_watch 3
1170 function appeared in