.\" 2015-01, Kees Cook <keescook@chromium.org>
.\" Added PTRACE_O_TRACESECCOMP, PTRACE_EVENT_SECCOMP
.\"
-.\" FIXME The following are undocumented
-.\" PTRACE_GETWMMXREGS ARM
-.\" PTRACE_SETWMMXREGS ARM
-.\" PTRACE_GET_THREAD_AREA Various architectures
-.\" PTRACE_SET_THREAD_AREA Various architectures
-.\" PTRACE_SET_SYSCALL ARM and ARM64
-.\" PTRACE_GETCRUNCHREGS ARM
-.\" PTRACE_SETCRUNCHREGS ARM
-.\" PTRACE_GETVFPREGS ARM and ARM64
-.\" PTRACE_SETVFPREGS ARM and ARM64
-.\" PTRACE_GETHBPREGS ARM and ARM64
-.\" PTRACE_SETHBPREGS ARM and ARM64
-.\" PTRACE_SINGLEBLOCK Various architectures
+.\" FIXME The following are undocumented:
+.\"
+.\" PTRACE_GETWMMXREGS
+.\" PTRACE_SETWMMXREGS
+.\" ARM
+.\" Linux 2.6.12
+.\"
+.\" PTRACE_SET_SYSCALL
+.\" ARM and ARM64
+.\" Linux 2.6.16
+.\" commit 3f471126ee53feb5e9b210ea2f525ed3bb9b7a7f
+.\" Author: Nicolas Pitre <nico@cam.org>
+.\" Date: Sat Jan 14 19:30:04 2006 +0000
+.\"
+.\" PTRACE_GETCRUNCHREGS
+.\" PTRACE_SETCRUNCHREGS
+.\" ARM
+.\" Linux 2.6.18
+.\" commit 3bec6ded282b331552587267d67a06ed7fd95ddd
+.\" Author: Lennert Buytenhek <buytenh@wantstofly.org>
+.\" Date: Tue Jun 27 22:56:18 2006 +0100
+.\"
+.\" PTRACE_GETVFPREGS
+.\" PTRACE_SETVFPREGS
+.\" ARM and ARM64
+.\" Linux 2.6.30
+.\" commit 3d1228ead618b88e8606015cbabc49019981805d
+.\" Author: Catalin Marinas <catalin.marinas@arm.com>
+.\" Date: Wed Feb 11 13:12:56 2009 +0100
+.\"
+.\" PTRACE_GETHBPREGS
+.\" PTRACE_SETHBPREGS
+.\" ARM and ARM64
+.\" Linux 2.6.37
+.\" commit 864232fa1a2f8dfe003438ef0851a56722740f3e
+.\" Author: Will Deacon <will.deacon@arm.com>
+.\" Date: Fri Sep 3 10:42:55 2010 +0100
+.\"
+.\" PTRACE_SINGLEBLOCK
+.\" Since at least Linux 2.4.0 on various architectures
+.\" Since Linux 2.6.25 on x86 (and others?)
+.\" commit 5b88abbf770a0e1975c668743100f42934f385e8
+.\" Author: Roland McGrath <roland@redhat.com>
+.\" Date: Wed Jan 30 13:30:53 2008 +0100
+.\" ptrace: generic PTRACE_SINGLEBLOCK
+.\"
+.\" PTRACE_GETFPXREGS
+.\" PTRACE_SETFPXREGS
+.\" Since at least Linux 2.4.0 on various architectures
+.\"
+.\" PTRACE_GETFDPIC
+.\" PTRACE_GETFDPIC_EXEC
+.\" PTRACE_GETFDPIC_INTERP
+.\" blackfin, c6x, frv, sh
+.\" First appearance in Linux 2.6.11 on frv
+.\"
.\" and others that can be found in the arch/*/include/uapi/asm/ptrace files
.\"
-.TH PTRACE 2 2016-07-17 "Linux" "Linux Programmer's Manual"
+.TH PTRACE 2 2020-02-09 "Linux" "Linux Programmer's Manual"
.SH NAME
ptrace \- process trace
.SH SYNOPSIS
.nf
.B #include <sys/ptrace.h>
-.sp
+.PP
.BI "long ptrace(enum __ptrace_request " request ", pid_t " pid ", "
.BI " void *" addr ", void *" data );
.fi
and examine and change the tracee's memory and registers.
It is primarily used to implement breakpoint debugging and system
call tracing.
-.LP
+.PP
A tracee first needs to be attached to the tracer.
Attachment and subsequent commands are per thread:
in a multithreaded process,
never "a (possibly multithreaded) process".
Ptrace commands are always sent to
a specific tracee using a call of the form
-
+.PP
ptrace(PTRACE_foo, pid, ...)
-
+.PP
where
.I pid
is the thread ID of the corresponding Linux thread.
-.LP
+.PP
(Note that in this page, a "multithreaded process"
means a thread group consisting of threads created using the
.BR clone (2)
.B CLONE_THREAD
flag.)
-.LP
+.PP
A process can initiate a trace by calling
.BR fork (2)
and having the resulting child do a
.B PTRACE_ATTACH
or
.BR PTRACE_SEIZE .
-.LP
+.PP
While being traced, the tracee will stop each time a signal is delivered,
even if the signal is being ignored.
(An exception is
The tracer then causes the tracee to continue,
optionally ignoring the delivered signal
(or even delivering a different signal instead).
-.LP
+.PP
If the
.B PTRACE_O_TRACEEXEC
option is not in effect, all successful calls to
signal,
giving the parent a chance to gain control before the new program
begins execution.
-.LP
+.PP
When the tracer is finished tracing, it can cause the tracee to continue
executing in a normal, untraced mode via
.BR PTRACE_DETACH .
-.LP
+.PP
The value of
.I request
determines the action to be performed:
.BR __SI_FAULT ,
etc.) that are not otherwise exposed to user space.
.PP
-.in +10n
-.nf
+.in +4n
+.EX
struct ptrace_peeksiginfo_args {
u64 off; /* Ordinal position in queue at which
to start copying signals */
u32 flags; /* PTRACE_PEEKSIGINFO_SHARED or 0 */
s32 nr; /* Number of signals to copy */
};
-.fi
-
+.EE
+.in
+.IP
Currently, there is only one flag,
.BR PTRACE_PEEKSIGINFO_SHARED ,
for dumping signals from the process-wide signal queue.
.TP
.BR PTRACE_O_EXITKILL " (since Linux 3.8)"
.\" commit 992fb6e170639b0849bace8e49bf31bd37c4123
-If a tracer sets this flag, a
+Send a
.B SIGKILL
-signal will be sent to every tracee if the tracer exits.
+signal to the tracee if the tracer exits.
This option is useful for ptrace jailers that
want to ensure that tracees can never escape the tracer's control.
.TP
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_CLONE<<8))
.fi
-
+.IP
The PID of the new process can be retrieved with
.BR PTRACE_GETEVENTMSG .
.IP
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_EXEC<<8))
.fi
-
+.IP
If the execing thread is not a thread group leader,
the thread ID is reset to thread group leader's ID before this stop.
Since Linux 3.0, the former thread ID can be retrieved with
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_EXIT<<8))
.fi
-
+.IP
The tracee's exit status can be retrieved with
.BR PTRACE_GETEVENTMSG .
.IP
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_FORK<<8))
.fi
-
+.IP
The PID of the new process can be retrieved with
.BR PTRACE_GETEVENTMSG .
.TP
.IR "SIGTRAP|0x80" ).
This makes it easy for the tracer to distinguish
normal traps from those caused by a system call.
-.RB ( PTRACE_O_TRACESYSGOOD
-may not work on all architectures.)
.TP
.BR PTRACE_O_TRACEVFORK " (since Linux 2.5.46)"
Stop the tracee at the next
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_VFORK<<8))
.fi
-
+.IP
The PID of the new process can be retrieved with
.BR PTRACE_GETEVENTMSG .
.TP
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_VFORK_DONE<<8))
.fi
-
+.IP
The PID of the new process can (since Linux 2.6.18) be retrieved with
.BR PTRACE_GETEVENTMSG .
.TP
by the tracer will return a
.I status
value such that
-
+.IP
.nf
status>>8 == (SIGTRAP | (PTRACE_EVENT_SECCOMP<<8))
.fi
-
+.IP
While this triggers a
.BR PTRACE_EVENT
-stop, it is similar to a syscall-enter-stop, in that the tracee has
-not yet entered the syscall that seccomp triggered on.
+stop, it is similar to a syscall-enter-stop.
+For details, see the note on
+.B PTRACE_EVENT_SECCOMP
+below.
The seccomp event message data (from the
.BR SECCOMP_RET_DATA
portion of the seccomp filter rule) can be retrieved with
.RI ( addr
is ignored.)
.TP
+.BR PTRACE_SET_SYSCALL " (since Linux 2.6.16)"
+.\" commit 3f471126ee53feb5e9b210ea2f525ed3bb9b7a7f
+When in syscall-enter-stop,
+change the number of the system call that is about to
+be executed to the number specified in the
+.I data
+argument.
+The
+.I addr
+argument is ignored.
+This request is currently
+.\" As of 4.19-rc2
+supported only on arm (and arm64, though only for backwards compatibility),
+.\" commit 27aa55c5e5123fa8b8ad0156559d34d7edff58ca
+but most other architectures have other means of accomplishing this
+(usually by changing the register that the userland code passed the
+system call number in).
+.\" see change_syscall in tools/testing/selftests/seccomp/seccomp_bpf.c
+.\" and also strace's linux/*/set_scno.c files.
+.TP
.BR PTRACE_SYSEMU ", " PTRACE_SYSEMU_SINGLESTEP " (since Linux 2.6.14)"
For
.BR PTRACE_SYSEMU ,
continue and stop on entry to the next system call,
which will not be executed.
+See the documentation on syscall-stops below.
For
.BR PTRACE_SYSEMU_SINGLESTEP ,
do the same but also singlestep if not a system call.
and
.I data
are ignored.)
-
+.IP
Permission to perform a
.BR PTRACE_ATTACH
is governed by a ptrace access mode
must be zero.
.I data
contains a bit mask of ptrace options to activate immediately.
-
+.IP
Permission to perform a
.BR PTRACE_SEIZE
is governed by a ptrace access mode
.B PTRACE_MODE_ATTACH_REALCREDS
check; see below.
+.\"
+.TP
+.BR PTRACE_SECCOMP_GET_FILTER " (since Linux 4.4)"
+.\" commit f8e529ed941ba2bbcbf310b575d968159ce7e895
+This operation allows the tracer to dump the tracee's
+classic BPF filters.
+.IP
+.I addr
+is an integer specifying the index of the filter to be dumped.
+The most recently installed filter has the index 0.
+If
+.I addr
+is greater than the number of installed filters,
+the operation fails with the error
+.BR ENOENT .
+.IP
+.I data
+is either a pointer to a
+.IR "struct sock_filter"
+array that is large enough to store the BPF program,
+or NULL if the program is not to be stored.
+.IP
+Upon success,
+the return value is the number of instructions in the BPF program.
+If
+.I data
+was NULL, then this return value can be used to correctly size the
+.IR "struct sock_filter"
+array passed in a subsequent call.
+.IP
+This operation fails with the error
+.B EACCES
+if the caller does not have the
+.B CAP_SYS_ADMIN
+capability or if the caller is in strict or filter seccomp mode.
+If the filter referred to by
+.I addr
+is not a classic BPF filter, the operation fails with the error
+.BR EMEDIUMTYPE .
+.IP
+This operation is available if the kernel was configured with both the
+.B CONFIG_SECCOMP_FILTER
+and the
+.B CONFIG_CHECKPOINT_RESTORE
+options.
.TP
.B PTRACE_DETACH
Restart the stopped tracee as for
of which method was used to initiate tracing.
.RI ( addr
is ignored.)
+.\"
+.TP
+.BR PTRACE_GET_THREAD_AREA " (since Linux 2.6.0)"
+This operation performs a similar task to
+.BR get_thread_area (2).
+It reads the TLS entry in the GDT whose index is given in
+.IR addr ,
+placing a copy of the entry into the
+.IR "struct user_desc"
+pointed to by
+.IR data .
+(By contrast with
+.BR get_thread_area (2),
+the
+.I entry_number
+of the
+.IR "struct user_desc"
+is ignored.)
+.TP
+.BR PTRACE_SET_THREAD_AREA " (since Linux 2.6.0)"
+This operation performs a similar task to
+.BR set_thread_area (2).
+It sets the TLS entry in the GDT whose index is given in
+.IR addr ,
+assigning it the data supplied in the
+.IR "struct user_desc"
+pointed to by
+.IR data .
+(By contrast with
+.BR set_thread_area (2),
+the
+.I entry_number
+of the
+.IR "struct user_desc"
+is ignored; in other words,
+this ptrace operation can't be used to allocate a free TLS entry.)
+.TP
+.BR PTRACE_GET_SYSCALL_INFO " (since Linux 5.3)"
+.\" commit 201766a20e30f982ccfe36bebfad9602c3ff574a
+Retrieve information about the system call that caused the stop.
+The information is placed into the buffer pointed by the
+.I data
+argument, which should be a pointer to a buffer of type
+.IR "struct ptrace_syscall_info" .
+The
+.I addr
+argument contains the size of the buffer pointed to
+by the
+.I data
+argument (i.e.,
+.IR "sizeof(struct ptrace_syscall_info)" ).
+The return value contains the number of bytes available
+to be written by the kernel.
+If the size of the data to be written by the kernel exceeds the size
+specified by the
+.I addr
+argument, the output data is truncated.
+.IP
+The
+.I ptrace_syscall_info
+structure contains the following fields:
+.IP
+.in +2n
+.EX
+struct ptrace_syscall_info {
+ __u8 op; /* Type of system call stop */
+ __u32 arch; /* AUDIT_ARCH_* value; see seccomp(2) */
+ __u64 instruction_pointer; /* CPU instruction pointer */
+ __u64 stack_pointer; /* CPU stack pointer */
+ union {
+ struct { /* op == PTRACE_SYSCALL_INFO_ENTRY */
+ __u64 nr; /* System call number */
+ __u64 args[6]; /* System call arguments */
+ } entry;
+ struct { /* op == PTRACE_SYSCALL_INFO_EXIT */
+ __s64 rval; /* System call return value */
+ __u8 is_error; /* System call error flag;
+ Boolean: does rval contain
+ an error value (\-ERRCODE) or
+ a nonerror return value? */
+ } exit;
+ struct { /* op == PTRACE_SYSCALL_INFO_SECCOMP */
+ __u64 nr; /* System call number */
+ __u64 args[6]; /* System call arguments */
+ __u32 ret_data; /* SECCOMP_RET_DATA portion
+ of SECCOMP_RET_TRACE
+ return value */
+ } seccomp;
+ };
+};
+.EE
+.in
+.IP
+The
+.IR op ,
+.IR arch ,
+.IR instruction_pointer ,
+and
+.I stack_pointer
+fields are defined for all kinds of ptrace system call stops.
+The rest of the structure is a union; one should read only those fields
+that are meaningful for the kind of system call stop specified by the
+.IR op
+field.
+.IP
+The
+.I op
+field has one of the following values (defined in
+.IR <linux/ptrace.h>)
+indicating what type of stop occurred and
+which part of the union is filled:
+.RS
+.TP
+.BR PTRACE_SYSCALL_INFO_ENTRY
+The
+.I entry
+component of the union contains information relating to a
+system call entry stop.
+.TP
+.BR PTRACE_SYSCALL_INFO_EXIT
+The
+.I exit
+component of the union contains information relating to a
+system call exit stop.
+.TP
+.BR PTRACE_SYSCALL_INFO_SECCOMP
+The
+.I seccomp
+component of the union contains information relating to a
+.B PTRACE_EVENT_SECCOMP
+stop.
+.TP
+.BR PTRACE_SYSCALL_INFO_NONE
+No component of the union contains relevant information.
+.RE
+.\"
.SS Death under ptrace
When a (possibly multithreaded) process receives a killing signal
(one whose disposition is set to
Tracees report their death to their tracer(s).
Notification of this event is delivered via
.BR waitpid (2).
-.LP
+.PP
Note that the killing signal will first cause signal-delivery-stop
(on one tracee only),
and only after it is injected by the tracer
.I all
tracees within a multithreaded process.
(The term "signal-delivery-stop" is explained below.)
-.LP
+.PP
.B SIGKILL
does not generate signal-delivery-stop and
therefore the tracer can't suppress it.
.B SIGKILL
always kills the process (all its threads),
even if some threads of the process are ptraced.
-.LP
+.PP
When the tracee calls
.BR _exit (2),
it reports its death to its tracer.
Other threads are not affected.
-.LP
+.PP
When any thread executes
.BR exit_group (2),
every tracee in its thread group reports its death to its tracer.
-.LP
+.PP
If the
.B PTRACE_O_TRACEEXIT
option is on,
and when threads are torn down on
.BR execve (2)
in a multithreaded process.
-.LP
+.PP
The tracer cannot assume that the ptrace-stopped tracee exists.
There are many scenarios when the tracee may die while stopped (such as
.BR SIGKILL ).
may return 0 instead.
In other words, the tracee may be "not yet fully dead",
but already refusing ptrace requests.
-.LP
+.PP
The tracer can't assume that the tracee
.I always
ends its life by reporting
it will not respond to signals until
.B SIGCONT
is received.
-.LP
+.PP
There are many kinds of states when the tracee is stopped, and in ptrace
discussions they are often conflated.
Therefore, it is important to use precise terms.
-.LP
+.PP
In this manual page, any stopped state in which the tracee is ready
to accept ptrace commands from the tracer is called
.IR ptrace-stop .
.IR signal-delivery-stop ,
.IR group-stop ,
.IR syscall-stop ,
+.IR "PTRACE_EVENT stops" ,
and so on.
These stopped states are described in detail below.
-.LP
+.PP
When the running tracee enters ptrace-stop, it notifies its tracer using
.BR waitpid (2)
(or one of the other "wait" system calls).
Most of this manual page assumes that the tracer waits with:
-.LP
+.PP
pid = waitpid(pid_or_minus_1, &status, __WALL);
-.LP
+.PP
Ptrace-stopped tracees are reported as returns with
.I pid
greater than 0 and
.\" rules different if user wants to use waitid? Will waitid require
.\" WEXITED?
.\"
-.LP
+.PP
The
.B __WALL
flag does not include the
and
.B WEXITED
flags, but implies their functionality.
-.LP
+.PP
Setting the
.B WCONTINUED
flag when calling
.BR waitpid (2)
is not recommended: the "continued" state is per-process and
consuming it can confuse the real parent of the tracee.
-.LP
+.PP
Use of the
.B WNOHANG
flag may cause
to return 0 ("no wait results available yet")
even if the tracer knows there should be a notification.
Example:
-.nf
-
- errno = 0;
- ptrace(PTRACE_CONT, pid, 0L, 0L);
- if (errno == ESRCH) {
- /* tracee is dead */
- r = waitpid(tracee, &status, __WALL | WNOHANG);
- /* r can still be 0 here! */
- }
-.fi
+.PP
+.in +4n
+.EX
+errno = 0;
+ptrace(PTRACE_CONT, pid, 0L, 0L);
+if (errno == ESRCH) {
+ /* tracee is dead */
+ r = waitpid(tracee, &status, __WALL | WNOHANG);
+ /* r can still be 0 here! */
+}
+.EE
+.in
.\" FIXME .
.\" waitid usage? WNOWAIT?
.\" describe how wait notifications queue (or not queue)
-.LP
+.PP
The following kinds of ptrace-stops exist: signal-delivery-stops,
group-stops,
.B PTRACE_EVENT
with the usual exception that
.B SIGSTOP
can't be blocked.
-.LP
+.PP
Signal-delivery-stop is observed by the tracer as
.BR waitpid (2)
returning with
.SS Signal injection and suppression
After signal-delivery-stop is observed by the tracer,
the tracer should restart the tracee with the call
-.LP
+.PP
ptrace(PTRACE_restart, pid, 0, sig)
-.LP
+.PP
where
.B PTRACE_restart
is one of the restarting ptrace requests.
This operation is called
.I "signal injection"
in this manual page, to distinguish it from signal-delivery-stop.
-.LP
+.PP
The
.I sig
value may be different from the
.I WSTOPSIG(status)
value: the tracer can cause a different signal to be injected.
-.LP
+.PP
Note that a suppressed signal still causes system calls to return
prematurely.
In this case, system calls will be restarted: the tracer will
however, kernel bugs exist which cause some system calls to fail with
.B EINTR
even though no observable signal is injected to the tracee.
-.LP
+.PP
Restarting ptrace commands issued in ptrace-stops other than
signal-delivery-stop are not guaranteed to inject a signal, even if
.I sig
Ptrace users should not try to "create a new signal" this way: use
.BR tgkill (2)
instead.
-.LP
+.PP
The fact that signal injection requests may be ignored
when restarting the tracee after
ptrace stops that are not signal-delivery-stops
is a cause of confusion among ptrace users.
One typical scenario is that the tracer observes group-stop,
mistakes it for signal-delivery-stop, restarts the tracee with
-
+.PP
ptrace(PTRACE_restart, pid, 0, stopsig)
-
+.PP
with the intention of injecting
.IR stopsig ,
but
.I stopsig
gets ignored and the tracee continues to run.
-.LP
+.PP
The
.B SIGCONT
signal has a side effect of waking up (all threads of)
In other words,
.B SIGCONT
may be not the first signal observed by the tracee after it was sent.
-.LP
+.PP
Stopping signals cause (all threads of) a process to enter group-stop.
This side effect happens after signal injection, and therefore can be
suppressed by the tracer.
-.LP
+.PP
In Linux 2.4 and earlier, the
.B SIGSTOP
signal can't be injected.
.\" /* The debugger continued. Ignore SIGSTOP. */
.\" if (signr == SIGSTOP)
.\" continue;
-.LP
+.PP
.B PTRACE_GETSIGINFO
can be used to retrieve a
.I siginfo_t
tracees within the multithreaded process.
As usual, every tracee reports its group-stop separately
to the corresponding tracer.
-.LP
+.PP
Group-stop is observed by the tracer as
.BR waitpid (2)
returning with
.IR WSTOPSIG(status) .
The same result is returned by some other classes of ptrace-stops,
therefore the recommended practice is to perform the call
-.LP
+.PP
ptrace(PTRACE_GETSIGINFO, pid, 0, &siginfo)
-.LP
+.PP
The call can be avoided if the signal is not
.BR SIGSTOP ,
.BR SIGTSTP ,
("no such process") if a
.B SIGKILL
killed the tracee.)
-.LP
+.PP
If tracee was attached using
.BR PTRACE_SEIZE ,
group-stop is indicated by
without requiring an extra
.B PTRACE_GETSIGINFO
call.
-.LP
+.PP
As of Linux 2.6.38,
after the tracer sees the tracee ptrace-stop and until it
restarts or kills it, the tracee will not run,
death) to the tracer, even if the tracer enters into another
.BR waitpid (2)
call.
-.LP
+.PP
The kernel behavior described in the previous paragraph
causes a problem with transparent handling of stopping signals.
If the tracer restarts the tracee after group-stop,
this would cause the
.B SIGCONT
signals to have no effect on the tracee.
-.LP
+.PP
Since Linux 3.4, there is a method to overcome this problem: instead of
.BR PTRACE_CONT ,
a
options, the tracee will enter ptrace-stops called
.B PTRACE_EVENT
stops.
-.LP
+.PP
.B PTRACE_EVENT
stops are observed by the tracer as
.BR waitpid (2)
and
.I WSTOPSIG(status)
returns
-.BR SIGTRAP .
+.BR SIGTRAP
+(or for
+.BR PTRACE_EVENT_STOP ,
+returns the stopping signal if tracee is in a group-stop).
An additional bit is set in the higher byte of the status word:
the value
.I status>>8
will be
-
- (SIGTRAP | PTRACE_EVENT_foo << 8).
-
+.PP
+ ((PTRACE_EVENT_foo<<8) | SIGTRAP).
+.PP
The following events exist:
.TP
.B PTRACE_EVENT_VFORK
.B CLONE_VFORK
flag,
but after the child unblocked this tracee by exiting or execing.
-.LP
+.PP
For all four stops described above,
the stop occurs in the parent (i.e., the tracee),
not in the newly created thread.
.BR SECCOMP_RET_DATA
portion of the seccomp filter rule) can be retrieved with
.BR PTRACE_GETEVENTMSG .
-.LP
+The semantics of this stop are described in
+detail in a separate section below.
+.PP
.B PTRACE_GETSIGINFO
on
.B PTRACE_EVENT
.IR "(event<<8)\ |\ SIGTRAP" .
.SS Syscall-stops
If the tracee was restarted by
-.BR PTRACE_SYSCALL ,
+.BR PTRACE_SYSCALL
+or
+.BR PTRACE_SYSEMU ,
the tracee enters
-syscall-enter-stop just prior to entering any system call.
-If the tracer restarts the tracee with
+syscall-enter-stop just prior to entering any system call (which
+will not be executed if the restart was using
+.BR PTRACE_SYSEMU ,
+regardless of any change made to registers at this point or how the
+tracee is restarted after this stop).
+No matter which method caused the syscall-entry-stop,
+if the tracer restarts the tracee with
.BR PTRACE_SYSCALL ,
the tracee enters syscall-exit-stop when the system call is finished,
or if it is interrupted by a signal.
(That is, signal-delivery-stop never happens between syscall-enter-stop
and syscall-exit-stop; it happens
.I after
-syscall-exit-stop.)
-.LP
+syscall-exit-stop.).
+If the tracee is continued using any other method (including
+.BR PTRACE_SYSEMU ),
+no syscall-exit-stop occurs.
+Note that all mentions
+.BR PTRACE_SYSEMU
+apply equally to
+.BR PTRACE_SYSEMU_SINGLESTEP.
+.PP
+However, even if the tracee was continued using
+.BR PTRACE_SYSCALL ,
+it is not guaranteed that the next stop will be a syscall-exit-stop.
Other possibilities are that the tracee may stop in a
.B PTRACE_EVENT
-stop, exit (if it entered
+stop (including seccomp stops), exit (if it entered
.BR _exit (2)
or
.BR exit_group (2)),
happened in another thread,
and that thread is not traced by the same tracer;
this situation is discussed later).
-.LP
+.PP
Syscall-enter-stop and syscall-exit-stop are observed by the tracer as
.BR waitpid (2)
returning with
.I WSTOPSIG(status)
will give the value
.IR "(SIGTRAP\ |\ 0x80)" .
-.LP
+.PP
Syscall-stops can be distinguished from signal-delivery-stop with
.B SIGTRAP
by querying
.TP
.IR si_code " == SIGTRAP or " si_code " == (SIGTRAP|0x80)"
This is a syscall-stop.
-.LP
+.PP
However, syscall-stops happen very often (twice per system call),
and performing
.B PTRACE_GETSIGINFO
for every syscall-stop may be somewhat expensive.
-.LP
+.PP
Some architectures allow the cases to be distinguished
by examining registers.
For example, on x86,
in other words, it looks like a
"stray syscall-exit-stop" and can be detected this way.
But such detection is fragile and is best avoided.
-.LP
+.PP
Using the
.B PTRACE_O_TRACESYSGOOD
option is the recommended method to distinguish syscall-stops
from other kinds of ptrace-stops,
since it is reliable and does not incur a performance penalty.
-.LP
+.PP
Syscall-enter-stop and syscall-exit-stop are
indistinguishable from each other by the tracer.
The tracer needs to keep track of the sequence of
ptrace-stops in order to not misinterpret syscall-enter-stop as
syscall-exit-stop or vice versa.
-The rule is that syscall-enter-stop is
+In general, a syscall-enter-stop is
always followed by syscall-exit-stop,
.B PTRACE_EVENT
-stop or the tracee's death;
+stop, or the tracee's death;
no other kinds of ptrace-stop can occur in between.
-.LP
+However, note that seccomp stops (see below) can cause syscall-exit-stops,
+without preceding syscall-entry-stops.
+If seccomp is in use, care needs
+to be taken not to misinterpret such stops as syscall-entry-stops.
+.PP
If after syscall-enter-stop,
the tracer uses a restarting command other than
.BR PTRACE_SYSCALL ,
syscall-exit-stop is not generated.
-.LP
+.PP
.B PTRACE_GETSIGINFO
on syscall-stops returns
.B SIGTRAP
.B SIGTRAP
or
.IR (SIGTRAP|0x80) .
-.SS PTRACE_SINGLESTEP, PTRACE_SYSEMU, PTRACE_SYSEMU_SINGLESTEP stops
+.\"
+.SS PTRACE_EVENT_SECCOMP stops (Linux 3.5 to 4.7)
+The behavior of
+.BR PTRACE_EVENT_SECCOMP
+stops and their interaction with other kinds
+of ptrace stops has changed between kernel versions.
+This documents the behavior
+from their introduction until Linux 4.7 (inclusive).
+The behavior in later kernel versions is documented in the next section.
+.PP
+A
+.BR PTRACE_EVENT_SECCOMP
+stop occurs whenever a
+.BR SECCOMP_RET_TRACE
+rule is triggered.
+This is independent of which methods was used to restart the system call.
+Notably, seccomp still runs even if the tracee was restarted using
+.BR PTRACE_SYSEMU
+and this system call is unconditionally skipped.
+.PP
+Restarts from this stop will behave as if the stop had occurred right
+before the system call in question.
+In particular, both
+.BR PTRACE_SYSCALL
+and
+.BR PTRACE_SYSEMU
+will normally cause a subsequent syscall-entry-stop.
+However, if after the
+.BR PTRACE_EVENT_SECCOMP
+the system call number is negative,
+both the syscall-entry-stop and the system call itself will be skipped.
+This means that if the system call number is negative after a
+.BR PTRACE_EVENT_SECCOMP
+and the tracee is restarted using
+.BR PTRACE_SYSCALL ,
+the next observed stop will be a syscall-exit-stop,
+rather than the syscall-entry-stop that might have been expected.
+.\"
+.SS PTRACE_EVENT_SECCOMP stops (since Linux 4.8)
+Starting with Linux 4.8,
+.\" commit 93e35efb8de45393cf61ed07f7b407629bf698ea
+the
+.BR PTRACE_EVENT_SECCOMP
+stop was reordered to occur between syscall-entry-stop and
+syscall-exit-stop.
+Note that seccomp no longer runs (and no
+.B PTRACE_EVENT_SECCOMP
+will be reported) if the system call is skipped due to
+.BR PTRACE_SYSEMU .
+.PP
+Functionally, a
+.B PTRACE_EVENT_SECCOMP
+stop functions comparably
+to a syscall-entry-stop (i.e., continuations using
+.BR PTRACE_SYSCALL
+will cause syscall-exit-stops,
+the system call number may be changed and any other modified registers
+are visible to the to-be-executed system call as well).
+Note that there may be,
+but need not have been a preceding syscall-entry-stop.
+.PP
+After a
+.BR PTRACE_EVENT_SECCOMP
+stop, seccomp will be rerun, with a
+.BR SECCOMP_RET_TRACE
+rule now functioning the same as a
+.BR SECCOMP_RET_ALLOW .
+Specifically, this means that if registers are not modified during the
+.BR PTRACE_EVENT_SECCOMP
+stop, the system call will then be allowed.
+.\"
+.SS PTRACE_SINGLESTEP stops
[Details of these kinds of stops are yet to be documented.]
.\"
.\" FIXME .
-.\" document stops occurring with PTRACE_SINGLESTEP, PTRACE_SYSEMU,
-.\" PTRACE_SYSEMU_SINGLESTEP
+.\" document stops occurring with PTRACE_SINGLESTEP
+.\"
.SS Informational and restarting ptrace commands
Most ptrace commands (all except
.BR PTRACE_ATTACH ,
.BR PTRACE_KILL )
require the tracee to be in a ptrace-stop, otherwise they fail with
.BR ESRCH .
-.LP
+.PP
When the tracee is in ptrace-stop,
the tracer can read and write data to
the tracee using informational commands.
These commands leave the tracee in ptrace-stopped state:
-.LP
-.nf
- ptrace(PTRACE_PEEKTEXT/PEEKDATA/PEEKUSER, pid, addr, 0);
- ptrace(PTRACE_POKETEXT/POKEDATA/POKEUSER, pid, addr, long_val);
- ptrace(PTRACE_GETREGS/GETFPREGS, pid, 0, &struct);
- ptrace(PTRACE_SETREGS/SETFPREGS, pid, 0, &struct);
- ptrace(PTRACE_GETREGSET, pid, NT_foo, &iov);
- ptrace(PTRACE_SETREGSET, pid, NT_foo, &iov);
- ptrace(PTRACE_GETSIGINFO, pid, 0, &siginfo);
- ptrace(PTRACE_SETSIGINFO, pid, 0, &siginfo);
- ptrace(PTRACE_GETEVENTMSG, pid, 0, &long_var);
- ptrace(PTRACE_SETOPTIONS, pid, 0, PTRACE_O_flags);
-.fi
-.LP
+.PP
+.in +4n
+.EX
+ptrace(PTRACE_PEEKTEXT/PEEKDATA/PEEKUSER, pid, addr, 0);
+ptrace(PTRACE_POKETEXT/POKEDATA/POKEUSER, pid, addr, long_val);
+ptrace(PTRACE_GETREGS/GETFPREGS, pid, 0, &struct);
+ptrace(PTRACE_SETREGS/SETFPREGS, pid, 0, &struct);
+ptrace(PTRACE_GETREGSET, pid, NT_foo, &iov);
+ptrace(PTRACE_SETREGSET, pid, NT_foo, &iov);
+ptrace(PTRACE_GETSIGINFO, pid, 0, &siginfo);
+ptrace(PTRACE_SETSIGINFO, pid, 0, &siginfo);
+ptrace(PTRACE_GETEVENTMSG, pid, 0, &long_var);
+ptrace(PTRACE_SETOPTIONS, pid, 0, PTRACE_O_flags);
+.EE
+.in
+.PP
Note that some errors are not reported.
For example, setting signal information
.RI ( siginfo )
.B PTRACE_GETEVENTMSG
may succeed and return some random value if current ptrace-stop
is not documented as returning a meaningful event message.
-.LP
+.PP
The call
-
+.PP
ptrace(PTRACE_SETOPTIONS, pid, 0, PTRACE_O_flags);
-
+.PP
affects one tracee.
The tracee's current flags are replaced.
Flags are inherited by new tracees created and "auto-attached" via active
or
.BR PTRACE_O_TRACECLONE
options.
-.LP
+.PP
Another group of commands makes the ptrace-stopped tracee run.
They have the form:
-.LP
+.PP
ptrace(cmd, pid, 0, sig);
-.LP
+.PP
where
.I cmd
is
.IR sig .)
.SS Attaching and detaching
A thread can be attached to the tracer using the call
-
+.PP
ptrace(PTRACE_ATTACH, pid, 0, 0);
-
+.PP
or
-
+.PP
ptrace(PTRACE_SEIZE, pid, 0, PTRACE_O_flags);
-
+.PP
.B PTRACE_ATTACH
sends
.B SIGSTOP
.B SIGSTOP
may be lost.
.\"
-.\" FIXME . Describe how to attach to a thread which is already group-stopped.
-.LP
+.\" FIXME Describe how to attach to a thread which is already group-stopped.
+.PP
Since attaching sends
.B SIGSTOP
and the tracer usually suppresses it, this may cause a stray
.B EINTR
return from the currently executing system call in the tracee,
as described in the "Signal injection and suppression" section.
-.LP
+.PP
Since Linux 3.4,
.B PTRACE_SEIZE
can be used instead of
use
.B PTRACE_INTERRUPT
command.
-.LP
+.PP
The request
-
+.PP
ptrace(PTRACE_TRACEME, 0, 0, 0);
-
+.PP
turns the calling thread into a tracee.
The thread continues to run (doesn't enter ptrace-stop).
A common practice is to follow the
.B PTRACE_TRACEME
with
-
+.PP
raise(SIGSTOP);
-
+.PP
and allow the parent (which is our tracer now) to observe our
signal-delivery-stop.
-.LP
+.PP
If the
.BR PTRACE_O_TRACEFORK ,
.BR PTRACE_O_TRACEVFORK ,
.B SIGSTOP
is delivered to the children, causing them to enter
signal-delivery-stop after they exit the system call which created them.
-.LP
+.PP
Detaching of the tracee is performed by:
-
+.PP
ptrace(PTRACE_DETACH, pid, 0, sig);
-
+.PP
.B PTRACE_DETACH
is a restarting operation;
therefore it requires the tracee to be in ptrace-stop.
Otherwise, the
.I sig
parameter may be silently ignored.
-.LP
+.PP
If the tracee is running when the tracer wants to detach it,
the usual solution is to send
.B SIGSTOP
the tracee is not already ptrace-stopped,
because no signal delivery happens while it is\(emnot even
.BR SIGSTOP .
-.\" FIXME . Describe how to detach from a group-stopped tracee so that it
-.\" doesn't run, but continues to wait for SIGCONT.
-.LP
+.\" FIXME Describe how to detach from a group-stopped tracee so that it
+.\" doesn't run, but continues to wait for SIGCONT.
+.PP
If the tracer dies, all tracees are automatically detached and restarted,
unless they were in group-stop.
Handling of restart from group-stop is currently buggy,
.BR execve (2)
it will appear as if it has become a tracee of
the tracer of the execing tracee.
-.LP
+.PP
All of the above effects are the artifacts of
the thread ID change in the tracee.
-.LP
+.PP
The
.B PTRACE_O_TRACEEXEC
option is the recommended tool for dealing with this situation.
.B SIGTRAP
generation on
.BR execve (2).
-.LP
+.PP
When the tracer receives
.B PTRACE_EVENT_EXEC
stop notification,
it is guaranteed that except this tracee and the thread group leader,
no other threads from the process are alive.
-.LP
+.PP
On receiving the
.B PTRACE_EVENT_EXEC
stop notification,
data structures describing the threads of this process,
and retain only one data structure\(emone which
describes the single still running tracee, with
-
+.PP
thread ID == thread group ID == process ID.
-.LP
+.PP
Example: two threads call
.BR execve (2)
at the same time:
-.LP
+.PP
.nf
*** we get syscall-enter-stop in thread 1: **
PID1 execve("/bin/foo", "foo" <unfinished ...>
*** we get syscall-exit-stop for PID0: **
PID0 <... execve resumed> ) = 0
.fi
-.LP
+.PP
If the
.B PTRACE_O_TRACEEXEC
option is
.RI ( SI_USER ).
This signal may be blocked by signal mask,
and thus may be delivered (much) later.
-.LP
+.PP
Usually, the tracer (for example,
.BR strace (1))
would not want to show this extra post-execve
several kinds of
.BR waitpid (2)
notifications when the child process is traced by some other process.
-.LP
+.PP
Many of these bugs have been fixed, but as of Linux 2.6.38 several still
exist; see BUGS below.
-.LP
+.PP
As of Linux 2.6.38, the following is believed to work correctly:
.IP * 3
exit/death by signal is reported first to the tracer, then,
On success, the
.B PTRACE_PEEK*
requests return the requested data (but see NOTES),
-while other requests return zero.
-.LP
+the
+.B PTRACE_SECCOMP_GET_FILTER
+request returns the number of instructions in the BPF program, and
+other requests return zero.
+.PP
On error, all requests return \-1, and
.I errno
is set appropriately.
.I 0L
or
.IR "(void\ *)\ 0".
-.LP
+.PP
In Linux kernels before 2.6.26,
.\" See commit 00cd5c37afd5f431ac186dd131705048c0a11fdb
.BR init (1),
the process with PID 1, may not be traced.
-.LP
+.PP
A tracees parent continues to be the tracer even if that tracer calls
.BR execve (2).
-.LP
+.PP
The layout of the contents of memory and the USER area are
quite operating-system- and architecture-specific.
The offset supplied, and the data returned,
might not entirely match with the definition of
.IR "struct user" .
.\" See http://lkml.org/lkml/2008/5/8/375
-.LP
+.PP
The size of a "word" is determined by the operating-system variant
(e.g., for 32-bit Linux it is 32 bits).
-.LP
+.PP
This page documents the way the
.BR ptrace ()
call works currently in Linux.
and the results of checks performed by any enabled Linux Security Module
(LSM)\(emfor example, SELinux, Yama, or Smack\(emand by the commoncap LSM
(which is always invoked).
-
+.PP
Prior to Linux 2.6.27, all access checks were of a single type.
Since Linux 2.6.27,
.\" commit 006ebb40d3d65338bd74abb03b945f8d60e362bd
.B PTRACE_MODE_ATTACH_REALCREDS
Defined as
.BR "PTRACE_MODE_ATTACH | PTRACE_MODE_REALCREDS" .
-.fi
.PP
One further modifier can be ORed with the access mode:
.TP
.IR /proc ).
These names are used in other manual pages to provide a simple
shorthand for labeling the different kernel checks.
-
+.PP
The algorithm employed for ptrace access mode checking determines whether
the calling process is allowed to perform the corresponding action
on the target process.
files, the "calling process" is the one opening the file,
and the process with the corresponding PID is the "target process".)
The algorithm is as follows:
-.IP 1. 4
+.IP 1. 3
If the calling thread and the target thread are in the same
thread group, access is always allowed.
.IP 2.
.BR credentials (7),
the filesystem UID and GID almost always have the same values
as the corresponding effective IDs.)
-
+.IP
Otherwise, the access mode specifies
.BR PTRACE_MODE_REALCREDS ,
so use the caller's real UID and GID for the checks in the next step.
.RS
.IP \(bu 2
The caller and the target process are in the same user namespace,
-and the caller's capabilities are a proper superset of the target process's
+and the caller's capabilities are a superset of the target process's
.I permitted
capabilities.
.IP \(bu
(e.g., a GPG agent or an SSH session) owned by the user in order
to gain additional credentials that may exist in memory
and thus expand the scope of the attack.
-
+.PP
More precisely, the Yama LSM limits two types of operations:
.IP * 3
Any operation that performs a ptrace access mode
.BR ptrace ()
.BR PTRACE_ATTACH .
(See the "Ptrace access mode checking" discussion above.)
-
+.IP
.IP *
.BR ptrace ()
.BR PTRACE_TRACEME .
No additional restrictions on operations that perform
.BR PTRACE_MODE_ATTACH
checks (beyond those imposed by the commoncap and other LSMs).
-
+.IP
The use of
.BR PTRACE_TRACEME
is unchanged.
By default,
the predefined relationship is that the target process
must be a descendant of the caller.
-
+.IP
A target process can employ the
.BR prctl (2)
.B PR_SET_PTRACER
.BR PTRACE_MODE_ATTACH
operations on the target.
See the kernel source file
+.IR Documentation/admin\-guide/LSM/Yama.rst
+.\" commit 90bb766440f2147486a2acc3e793d7b8348b0c22
+(or
.IR Documentation/security/Yama.txt
+before Linux 4.13)
for further details.
-
+.IP
The use of
.BR PTRACE_TRACEME
is unchanged.
.BR PTRACE_MODE_ATTACH
operations or trace children that employ
.BR PTRACE_TRACEME .
-
+.IP
Once this value has been written to the file, it cannot be changed.
.PP
With respect to values 1 and 2,
to
.BR PTRACE_OLDSETOPTIONS ,
if that is defined.
-.LP
+.PP
Group-stop notifications are sent to the tracer, but not to real parent.
Last confirmed on 2.6.38.6.
-.LP
+.PP
If a thread group leader is traced and exits by calling
.BR _exit (2),
.\" Note from Denys Vlasenko:
the thread group leader instead of restarting it in this case.
Last confirmed on 2.6.38.6.
.\" FIXME . need to test/verify this scenario
-.LP
+.PP
A
.B SIGKILL
signal may still cause a
.B SIGKILL
is meant to always immediately kill tasks even under ptrace.
Last confirmed on Linux 3.13.
-.LP
+.PP
Some system calls return with
.B EINTR
if a signal was sent to a tracee, but delivery was suppressed by the tracer.
file descriptor.
The usual symptom of this bug is that when you attach to
a quiescent process with the command
-
- strace \-p <process-ID>
-
+.PP
+.in +4n
+.EX
+strace \-p <process-ID>
+.EE
+.in
+.PP
then, instead of the usual
and expected one-line output such as
-.nf
-
- restart_syscall(<... resuming interrupted call ...>_
-
-.fi
+.PP
+.in +4n
+.EX
+restart_syscall(<... resuming interrupted call ...>_
+.EE
+.in
+.PP
or
-.nf
-
- select(6, [5], NULL, [5], NULL_
-
-.fi
+.PP
+.in +4n
+.EX
+select(6, [5], NULL, [5], NULL_
+.EE
+.in
+.PP
('_' denotes the cursor position), you observe more than one line.
For example:
-.nf
-
+.PP
+.in +4n
+.EX
clock_gettime(CLOCK_MONOTONIC, {15370, 690928118}) = 0
epoll_wait(4,_
-
-.fi
+.EE
+.in
+.PP
What is not visible here is that the process was blocked in
.BR epoll_wait (2)
before
errors may behave in an unintended way upon an
.BR strace (1)
attach.)
+.PP
+Contrary to the normal rules, the glibc wrapper for
+.BR ptrace ()
+can set
+.I errno
+to zero.
.SH SEE ALSO
.BR gdb (1),
+.BR ltrace (1),
.BR strace (1),
.BR clone (2),
.BR execve (2),
projects / thirdparty / man-pages.git / blobdiff