diff --git a/lib/libc/sys/kqueue.2 b/lib/libc/sys/kqueue.2 index d6e949baa24c..e4e4c85d8a77 100644 --- a/lib/libc/sys/kqueue.2 +++ b/lib/libc/sys/kqueue.2 @@ -1,904 +1,910 @@ .\" Copyright (c) 2000 Jonathan Lemon .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" -.Dd March 26, 2023 +.Dd July 22, 2026 .Dt KQUEUE 2 .Os .Sh NAME .Nm kqueue , .Nm kevent .Nd kernel event notification mechanism .Sh LIBRARY .Lb libc .Sh SYNOPSIS .In sys/event.h .Ft int .Fn kqueue "void" .Ft int .Fn kqueuex "u_int flags" .Ft int .Fn kqueue1 "int flags" .Ft int .Fo kevent .Fa "int kq" .Fa "const struct kevent *changelist" .Fa "int nchanges" .Fa "struct kevent *eventlist" .Fa "int nevents" .Fa "const struct timespec *timeout" .Fc .Fn EV_SET "kev" ident filter flags fflags data udata .Sh DESCRIPTION The .Fn kqueue system call provides a generic method of notifying the user when an event happens or a condition holds, based on the results of small pieces of kernel code termed filters. A kevent is identified by the (ident, filter) pair; there may only be one unique kevent per kqueue. .Pp The filter is executed upon the initial registration of a kevent in order to detect whether a preexisting condition is present, and is also executed whenever an event is passed to the filter for evaluation. If the filter determines that the condition should be reported, then the kevent is placed on the kqueue for the user to retrieve. .Pp The filter is also run when the user attempts to retrieve the kevent from the kqueue. If the filter indicates that the condition that triggered the event no longer holds, the kevent is removed from the kqueue and is not returned. .Pp Multiple events which trigger the filter do not result in multiple kevents being placed on the kqueue; instead, the filter will aggregate the events into a single struct kevent. Calling .Fn close on a file descriptor will remove any kevents that reference the descriptor. .Pp The .Fn kqueue system call creates a new kernel event queue and returns a descriptor. The queue is not inherited by a child created with .Xr fork 2 . However, if .Xr rfork 2 is called without the .Dv RFFDG flag, then the descriptor table is shared, which will allow sharing of the kqueue between two processes. .Pp The .Fn kqueuex system call also creates a new kernel event queue, and additionally takes a .Fa flags argument, which is a bitwise-inclusive OR of the following flags: .Bl -tag -width "KQUEUE_CLOEXEC" .It Dv KQUEUE_CLOEXEC The returned file descriptor is automatically closed on .Xr execve 2 .El .Pp The .Fn kqueue system call is equivalent to calling .Fn kqueuex with .Fa flags set to 0. .Pp The .Fn kqueue1 function exists for compatibility with .Nx . The .Fa flags argument accepts zero or more of the following values: .Bl -tag -width O_CLOEXEC .It Dv O_CLOEXEC The returned file descriptor is automatically closed on .Xr execve 2 .El .Pp The .Fn kevent system call is used to register events with the queue, and return any pending events to the user. The .Fa changelist argument is a pointer to an array of .Va kevent structures, as defined in .In sys/event.h . All changes contained in the .Fa changelist are applied before any pending events are read from the queue. The .Fa nchanges argument gives the size of .Fa changelist . The .Fa eventlist argument is a pointer to an array of kevent structures. The .Fa nevents argument determines the size of .Fa eventlist . When .Fa nevents is zero, .Fn kevent will return immediately even if there is a .Fa timeout specified unlike .Xr select 2 . If .Fa timeout is a non-NULL pointer, it specifies a maximum interval to wait for an event, which will be interpreted as a struct timespec. If .Fa timeout is a NULL pointer, .Fn kevent waits indefinitely. To effect a poll, the .Fa timeout argument should be non-NULL, pointing to a zero-valued .Va timespec structure. The same array may be used for the .Fa changelist and .Fa eventlist . .Pp The .Fn EV_SET macro is provided for ease of initializing a kevent structure. .Pp The .Va kevent structure is defined as: .Bd -literal struct kevent { uintptr_t ident; /* identifier for this event */ short filter; /* filter for event */ u_short flags; /* action flags for kqueue */ u_int fflags; /* filter flag value */ int64_t data; /* filter data value */ void *udata; /* opaque user data identifier */ uint64_t ext[4]; /* extensions */ }; .Ed .Pp The fields of .Fa struct kevent are: .Bl -tag -width "Fa filter" .It Fa ident Value used to identify this event. The exact interpretation is determined by the attached filter, but often is a file descriptor. .It Fa filter Identifies the kernel filter used to process this event. The pre-defined system filters are described below. .It Fa flags Actions to perform on the event. .It Fa fflags Filter-specific flags. .It Fa data Filter-specific data value. .It Fa udata Opaque user-defined value passed through the kernel unchanged. .It Fa ext Extended data passed to and from kernel. The meaning of the .Fa ext[0] and .Fa ext[1] members is defined by the filter. If a filter does not use them, these members are passed through the kernel unchanged. The .Fa ext[2] and .Fa ext[3] members are always passed through the kernel unchanged, providing additional user-defined values. .El .Pp The .Va flags field can contain the following values: .Bl -tag -width EV_KEEPUDATA .It Dv EV_ADD Adds the event to the kqueue. Re-adding an existing event will modify the parameters of the original event, and not result in a duplicate entry. Adding an event automatically enables it, unless overridden by the EV_DISABLE flag. .It Dv EV_ENABLE Permit .Fn kevent to return the event if it is triggered. .It Dv EV_DISABLE Disable the event so .Fn kevent will not return it. The filter itself is not disabled. .It Dv EV_DISPATCH Disable the event source immediately after delivery of an event. See .Dv EV_DISABLE above. .It Dv EV_DELETE Removes the event from the kqueue. Events which are attached to file descriptors are automatically deleted on the last close of the descriptor. .It Dv EV_RECEIPT This flag is useful for making bulk changes to a kqueue without draining any pending events. When passed as input, it forces .Dv EV_ERROR to always be returned. When a filter is successfully added the .Va data field will be zero. Note that if this flag is encountered and there is no remaining space in .Fa eventlist to hold the .Dv EV_ERROR event, then subsequent changes will not get processed. .It Dv EV_ONESHOT Return only the first occurrence of the filter being triggered. After the user retrieves the event from the kqueue, it is deleted. .It Dv EV_CLEAR Reset the state of the event after it is retrieved by the user. This is useful for filters which report state transitions instead of the current state. Note that some filters may automatically set this flag internally. .It Dv EV_EOF Filters may set this flag to indicate filter-specific EOF condition. .It Dv EV_ERROR See .Sx RETURN VALUES below. .It Dv EV_KEEPUDATA Preserve the .Fa udata associated with an existing event. This allows other aspects of the event to be modified without requiring the caller to know the .Fa udata value previously registered with the event. This is especially useful with .Dv NOTE_TRIGGER or .Dv EV_ENABLE . This flag may not be used with .Dv EV_ADD . .El .Pp The predefined system filters are listed below. Arguments may be passed to and from the filter via the .Va fflags and .Va data fields in the kevent structure. .Bl -tag -width "Dv EVFILT_PROCDESC" .It Dv EVFILT_READ Takes a descriptor as the identifier, and returns whenever there is data available to read. The behavior of the filter is slightly different depending on the descriptor type. .Bl -tag -width 2n .It Sockets Sockets which have previously been passed to .Xr listen 2 return when there is an incoming connection pending. .Va data contains the size of the listen backlog. .Pp Other socket descriptors return when there is data to be read, subject to the .Dv SO_RCVLOWAT value of the socket buffer. This may be overridden with a per-filter low water mark at the time the filter is added by setting the .Dv NOTE_LOWAT flag in .Va fflags , and specifying the new low water mark in .Va data . On return, .Va data contains the number of bytes of protocol data available to read. .Pp If the read direction of the socket has shutdown, then the filter also sets .Dv EV_EOF in .Va flags , and returns the socket error (if any) in .Va fflags . It is possible for EOF to be returned (indicating the connection is gone) while there is still data pending in the socket buffer. .It Vnodes Returns when the file pointer is not at the end of file. .Va data contains the offset from current position to end of file, and may be negative. .Pp This behavior is different from .Xr poll 2 , where read events are triggered for regular files unconditionally. This event can be triggered unconditionally by setting the .Dv NOTE_FILE_POLL flag in .Va fflags . .It "Fifos, Pipes" Returns when the there is data to read; .Va data contains the number of bytes available. .Pp When the last writer disconnects, the filter will set .Dv EV_EOF in .Va flags . This will be cleared by the filter when a new writer connects, at which point the filter will resume waiting for data to become available before returning. .It "BPF devices" Returns when the BPF buffer is full, the BPF timeout has expired, or when the BPF has .Dq immediate mode enabled and there is any data to read; .Va data contains the number of bytes available. .It Eventfds Returns when the counter is greater than 0; .Va data contains the counter value, which must be cast to .Vt uint64_t . .It Kqueues Returns when pending events are present on the queue; .Va data contains the number of events available. .El .It Dv EVFILT_WRITE Takes a descriptor as the identifier, and returns whenever it is possible to write to the descriptor. For sockets, pipes and fifos, .Va data will contain the amount of space remaining in the write buffer. The filter will set .Dv EV_EOF when the reader disconnects, and for the fifo case, this will be cleared when a new reader connects. Note that this filter is not supported for vnodes. .Pp For sockets, the low water mark and socket error handling is identical to the .Dv EVFILT_READ case. .Pp For eventfds, .Va data will contain the maximum value that can be added to the counter without blocking. .Pp For BPF devices, when the descriptor is attached to an interface the filter always indicates that it is possible to write and .Va data will contain the MTU size of the underlying interface. .It Dv EVFILT_EMPTY Takes a descriptor as the identifier, and returns whenever there is no remaining data in the write buffer. .It Dv EVFILT_AIO Events for this filter are not registered with .Fn kevent directly but are registered via the .Va aio_sigevent member of an asynchronous I/O request when it is scheduled via an asynchronous I/O system call such as .Fn aio_read . The filter returns under the same conditions as .Fn aio_error . For more details on this filter see .Xr sigevent 3 and .Xr aio 4 . .It Dv EVFILT_VNODE Takes a file descriptor as the identifier and the events to watch for in .Va fflags , and returns when one or more of the requested events occurs on the descriptor. The events to monitor are: .Bl -tag -width "Dv NOTE_CLOSE_WRITE" .It Dv NOTE_ATTRIB The file referenced by the descriptor had its attributes changed. .It Dv NOTE_CLOSE A file descriptor referencing the monitored file, was closed. The closed file descriptor did not have write access. .It Dv NOTE_CLOSE_WRITE A file descriptor referencing the monitored file, was closed. The closed file descriptor had write access. .Pp This note, as well as .Dv NOTE_CLOSE , are not activated when files are closed forcibly by .Xr unmount 2 or .Xr revoke 2 . Instead, .Dv NOTE_REVOKE is sent for such events. .It Dv NOTE_DELETE The .Fn unlink system call was called on the file referenced by the descriptor. .It Dv NOTE_EXTEND For regular file, the file referenced by the descriptor was extended. .Pp For directory, reports that a directory entry was added or removed, as the result of rename operation. The .Dv NOTE_EXTEND event is not reported when a name is changed inside the directory. .It Dv NOTE_LINK The link count on the file changed. In particular, the .Dv NOTE_LINK event is reported if a subdirectory was created or deleted inside the directory referenced by the descriptor. .It Dv NOTE_OPEN The file referenced by the descriptor was opened. .It Dv NOTE_READ A read occurred on the file referenced by the descriptor. .It Dv NOTE_RENAME The file referenced by the descriptor was renamed. .It Dv NOTE_REVOKE Access to the file was revoked via .Xr revoke 2 or the underlying file system was unmounted. .It Dv NOTE_WRITE A write occurred on the file referenced by the descriptor. .El .Pp On return, .Va fflags contains the events which triggered the filter. .It Dv EVFILT_PROC Takes the process ID to monitor as the identifier and the events to watch for in .Va fflags , and returns when the process performs one or more of the requested events. If a process can normally see another process, it can attach an event to it. The events to monitor are: .Bl -tag -width "Dv NOTE_TRACKERR" .It Dv NOTE_EXIT The process has exited. The exit status will be stored in .Va data in the same format as the status returned by .Xr wait 2 . +.It Dv NOTE_REAP +The process has been reaped by a call to the +.Xr wait 2 +family of functions. .It Dv NOTE_FORK The process has called -.Fn fork . +.Xr fork 2 +or +.Xr vfork 2 . .It Dv NOTE_EXEC The process has executed a new process via .Xr execve 2 or a similar call. .It Dv NOTE_TRACK Follow a process across .Fn fork calls. The parent process registers a new kevent to monitor the child process using the same .Va fflags as the original event. The child process will signal an event with .Dv NOTE_CHILD set in .Va fflags and the parent PID in .Va data . .Pp If the parent process fails to register a new kevent .Pq usually due to resource limitations , it will signal an event with .Dv NOTE_TRACKERR set in .Va fflags , and the child process will not signal a .Dv NOTE_CHILD event. .El .Pp On return, .Va fflags contains the events which triggered the filter. .It Dv EVFILT_PROCDESC Takes the process descriptor created by .Xr pdfork 2 to monitor as the identifier and the events to watch for in .Va fflags , and returns when the associated process performs one or more of the requested events. The events to monitor are: .Bl -tag -width "Dv NOTE_EXIT" .It Dv NOTE_EXIT The process has exited. The exit status will be stored in .Va data . .El .Pp On return, .Va fflags contains the events which triggered the filter. .It Dv EVFILT_SIGNAL Takes the signal number to monitor as the identifier and returns when the given signal is delivered to the process. This coexists with the .Fn signal and .Fn sigaction facilities, and has a lower precedence. The filter will record all attempts to deliver a signal to a process, even if the signal has been marked as .Dv SIG_IGN , except for the .Dv SIGCHLD signal, which, if ignored, will not be recorded by the filter. Event notification happens after normal signal delivery processing. .Va data returns the number of times the signal has occurred since the last call to .Fn kevent . This filter automatically sets the .Dv EV_CLEAR flag internally. .It Dv EVFILT_TIMER Establishes an arbitrary timer identified by .Va ident . When adding a timer, .Va data specifies the moment to fire the timer (for .Dv NOTE_ABSTIME ) or the timeout period. The timer will be periodic unless .Dv EV_ONESHOT or .Dv NOTE_ABSTIME is specified. On return, .Va data contains the number of times the timeout has expired since the last call to .Fn kevent . For non-monotonic timers, this filter automatically sets the .Dv EV_CLEAR flag internally. .Pp The filter accepts the following flags in the .Va fflags argument: .Bl -tag -width "Dv NOTE_MSECONDS" .It Dv NOTE_SECONDS .Va data is in seconds. .It Dv NOTE_MSECONDS .Va data is in milliseconds. .It Dv NOTE_USECONDS .Va data is in microseconds. .It Dv NOTE_NSECONDS .Va data is in nanoseconds. .It Dv NOTE_ABSTIME The specified expiration time is absolute. .El .Pp If .Va fflags is not set, the default is milliseconds. On return, .Va fflags contains the events which triggered the filter. .Pp Periodic timers with a specified timeout of 0 will be silently adjusted to timeout after 1 of the time units specified by the requested precision in .Va fflags . If an absolute time is specified that has already passed, then it is treated as if the current time were specified and the event will fire as soon as possible. .Pp If an existing timer is re-added, the existing timer will be effectively canceled (throwing away any undelivered record of previous timer expiration) and re-started using the new parameters contained in .Va data and .Va fflags . .Pp There is a system wide limit on the number of timers which is controlled by the .Va kern.kq_calloutmax sysctl. .It Dv EVFILT_USER Establishes a user event identified by .Va ident which is not associated with any kernel mechanism but is triggered by user level code. The lower 24 bits of the .Va fflags may be used for user defined flags and manipulated using the following: .Bl -tag -width "Dv NOTE_FFLAGSMASK" .It Dv NOTE_FFNOP Ignore the input .Va fflags . .It Dv NOTE_FFAND Bitwise AND .Va fflags . .It Dv NOTE_FFOR Bitwise OR .Va fflags . .It Dv NOTE_FFCOPY Copy .Va fflags . .It Dv NOTE_FFCTRLMASK Control mask for .Va fflags . .It Dv NOTE_FFLAGSMASK User defined flag mask for .Va fflags . .El .Pp A user event is triggered for output with the following: .Bl -tag -width "Dv NOTE_FFLAGSMASK" .It Dv NOTE_TRIGGER Cause the event to be triggered. .El .Pp On return, .Va fflags contains the users defined flags in the lower 24 bits. .El .Sh CANCELLATION BEHAVIOUR If .Fa nevents is non-zero, i.e., the function is potentially blocking, the call is a cancellation point. Otherwise, i.e., if .Fa nevents is zero, the call is not cancellable. Cancellation can only occur before any changes are made to the kqueue, or when the call was blocked and no changes to the queue were requested. .Sh RETURN VALUES The .Fn kqueue system call creates a new kernel event queue and returns a file descriptor. If there was an error creating the kernel event queue, a value of -1 is returned and errno set. .Pp The .Fn kevent system call returns the number of events placed in the .Fa eventlist , up to the value given by .Fa nevents . If an error occurs while processing an element of the .Fa changelist and there is enough room in the .Fa eventlist , then the event will be placed in the .Fa eventlist with .Dv EV_ERROR set in .Va flags and the system error in .Va data . Otherwise, .Dv -1 will be returned, and .Dv errno will be set to indicate the error condition. If the time limit expires, then .Fn kevent returns 0. .Sh EXAMPLES .Bd -literal -compact #include #include #include #include #include #include int main(int argc, char **argv) { struct kevent event; /* Event we want to monitor */ struct kevent tevent; /* Event triggered */ int kq, fd, ret; if (argc != 2) err(EXIT_FAILURE, "Usage: %s path\en", argv[0]); fd = open(argv[1], O_RDONLY); if (fd == -1) err(EXIT_FAILURE, "Failed to open '%s'", argv[1]); /* Create kqueue. */ kq = kqueue(); if (kq == -1) err(EXIT_FAILURE, "kqueue() failed"); /* Initialize kevent structure. */ EV_SET(&event, fd, EVFILT_VNODE, EV_ADD | EV_CLEAR, NOTE_WRITE, 0, NULL); /* Attach event to the kqueue. */ ret = kevent(kq, &event, 1, NULL, 0, NULL); if (ret == -1) err(EXIT_FAILURE, "kevent register"); for (;;) { /* Sleep until something happens. */ ret = kevent(kq, NULL, 0, &tevent, 1, NULL); if (ret == -1) { err(EXIT_FAILURE, "kevent wait"); } else if (ret > 0) { if (tevent.flags & EV_ERROR) errx(EXIT_FAILURE, "Event error: %s", strerror(event.data)); else printf("Something was written in '%s'\en", argv[1]); } } /* kqueues are destroyed upon close() */ (void)close(kq); (void)close(fd); } .Ed .Sh ERRORS The .Fn kqueue system call fails if: .Bl -tag -width Er .It Bq Er ENOMEM The kernel failed to allocate enough memory for the kernel queue. .It Bq Er ENOMEM The .Dv RLIMIT_KQUEUES rlimit (see .Xr getrlimit 2 ) for the current user would be exceeded. .It Bq Er EMFILE The per-process descriptor table is full. .It Bq Er ENFILE The system file table is full. .El .Pp The .Fn kevent system call fails if: .Bl -tag -width Er .It Bq Er EACCES The process does not have permission to register a filter. .It Bq Er EFAULT There was an error reading or writing the .Va kevent structure. .It Bq Er EBADF The specified descriptor is invalid. .It Bq Er EINTR A signal was delivered before the timeout expired and before any events were placed on the kqueue for return. .It Bq Er EINTR A cancellation request was delivered to the thread, but not yet handled. .It Bq Er EINVAL The specified time limit or filter is invalid. .It Bq Er EINVAL The specified length of the event or change lists is negative. .It Bq Er ENOENT The event could not be found to be modified or deleted. .It Bq Er ENOMEM No memory was available to register the event or, in the special case of a timer, the maximum number of timers has been exceeded. This maximum is configurable via the .Va kern.kq_calloutmax sysctl. .It Bq Er ESRCH The specified process to attach to does not exist. .El .Pp When .Fn kevent call fails with .Er EINTR error, all changes in the .Fa changelist have been applied. .Sh SEE ALSO .Xr aio_error 2 , .Xr aio_read 2 , .Xr aio_return 2 , .Xr poll 2 , .Xr read 2 , .Xr select 2 , .Xr sigaction 2 , .Xr write 2 , .Xr pthread_setcancelstate 3 , .Xr signal 3 .Rs .%A Jonathan Lemon .%T "Kqueue: A Generic and Scalable Event Notification Facility" .%I USENIX Association .%B Proceedings of the FREENIX Track: 2001 USENIX Annual Technical Conference .%D June 25-30, 2001 .\".http://www.usenix.org/event/usenix01/freenix01/full_papers/lemon/lemon.pdf .Re .Sh HISTORY The .Fn kqueue and .Fn kevent system calls first appeared in .Fx 4.1 . The .Fn kqueuex system call and .Fn kevent1 function first appeared in .Fx 14.0 . .Sh AUTHORS The .Fn kqueue system and this manual page were written by .An Jonathan Lemon Aq Mt jlemon@FreeBSD.org . .Sh BUGS .Pp In versions older than .Fx 12.0 , .In sys/event.h failed to parse without including .In sys/types.h manually. diff --git a/lib/libsysdecode/flags.c b/lib/libsysdecode/flags.c index 32829d35dbe0..018dfef5f5c2 100644 --- a/lib/libsysdecode/flags.c +++ b/lib/libsysdecode/flags.c @@ -1,1201 +1,1201 @@ /* * Copyright (c) 2006 "David Kirchner" . All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include #define L2CAP_SOCKET_CHECKED #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "support.h" #define X(a) { a, #a }, #define XEND { 0, NULL } #define TABLE_START(n) static struct name_table n[] = { #define TABLE_ENTRY X #define TABLE_END XEND }; #include "tables.h" #undef TABLE_START #undef TABLE_ENTRY #undef TABLE_END const char * sysdecode_atfd(int fd) { if (fd == AT_FDCWD) return ("AT_FDCWD"); return (NULL); } bool sysdecode_atflags(FILE *fp, int flag, int *rem) { return (print_mask_int(fp, atflags, flag, rem)); } static struct name_table semctlops[] = { X(GETNCNT) X(GETPID) X(GETVAL) X(GETALL) X(GETZCNT) X(SETVAL) X(SETALL) X(IPC_RMID) X(IPC_SET) X(IPC_STAT) XEND }; const char * sysdecode_semctl_cmd(int cmd) { return (lookup_value(semctlops, cmd)); } static struct name_table shmctlops[] = { X(IPC_RMID) X(IPC_SET) X(IPC_STAT) XEND }; const char * sysdecode_shmctl_cmd(int cmd) { return (lookup_value(shmctlops, cmd)); } const char * sysdecode_msgctl_cmd(int cmd) { return (sysdecode_shmctl_cmd(cmd)); } static struct name_table semgetflags[] = { X(IPC_CREAT) X(IPC_EXCL) X(SEM_R) X(SEM_A) X((SEM_R>>3)) X((SEM_A>>3)) X((SEM_R>>6)) X((SEM_A>>6)) XEND }; bool sysdecode_semget_flags(FILE *fp, int flag, int *rem) { return (print_mask_int(fp, semgetflags, flag, rem)); } static struct name_table idtypes[] = { X(P_PID) X(P_PPID) X(P_PGID) X(P_SID) X(P_CID) X(P_UID) X(P_GID) X(P_ALL) X(P_LWPID) X(P_TASKID) X(P_PROJID) X(P_POOLID) X(P_JAILID) X(P_CTID) X(P_CPUID) X(P_PSETID) XEND }; /* XXX: idtype is really an idtype_t */ const char * sysdecode_idtype(int idtype) { return (lookup_value(idtypes, idtype)); } /* * [g|s]etsockopt's level argument can either be SOL_SOCKET or a * protocol-specific value. */ const char * sysdecode_sockopt_level(int level) { const char *str; if (level == SOL_SOCKET) return ("SOL_SOCKET"); /* SOL_* constants for Bluetooth sockets. */ str = lookup_value(ngbtsolevel, level); if (str != NULL) return (str); /* * IP and Infiniband sockets use IP protocols as levels. Not all * protocols are valid but it is simpler to just allow all of them. * * XXX: IPPROTO_IP == 0, but UNIX domain sockets use a level of 0 * for private options. */ str = sysdecode_ipproto(level); if (str != NULL) return (str); return (NULL); } bool sysdecode_vmprot(FILE *fp, int type, int *rem) { return (print_mask_int(fp, vmprot, type, rem)); } static struct name_table sockflags[] = { X(SOCK_CLOEXEC) X(SOCK_NONBLOCK) XEND }; bool sysdecode_socket_type(FILE *fp, int type, int *rem) { const char *str; uintmax_t val; bool printed; str = lookup_value(socktype, type & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)); if (str != NULL) { fputs(str, fp); *rem = 0; printed = true; } else { *rem = type & ~(SOCK_CLOEXEC | SOCK_NONBLOCK); printed = false; } val = type & (SOCK_CLOEXEC | SOCK_NONBLOCK); print_mask_part(fp, sockflags, &val, &printed); return (printed); } bool sysdecode_access_mode(FILE *fp, int mode, int *rem) { return (print_mask_int(fp, accessmode, mode, rem)); } /* XXX: 'type' is really an acl_type_t. */ const char * sysdecode_acltype(int type) { return (lookup_value(acltype, type)); } bool sysdecode_cap_fcntlrights(FILE *fp, uint32_t rights, uint32_t *rem) { return (print_mask_int(fp, capfcntl, rights, rem)); } bool sysdecode_close_range_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, closerangeflags, flags, rem)); } const char * sysdecode_extattrnamespace(int namespace) { return (lookup_value(extattrns, namespace)); } const char * sysdecode_fadvice(int advice) { return (lookup_value(fadvisebehav, advice)); } bool sysdecode_open_flags(FILE *fp, int flags, int *rem) { bool printed; int mode; uintmax_t val; mode = flags & O_ACCMODE; flags &= ~O_ACCMODE; switch (mode) { case O_RDONLY: if (flags & O_EXEC) { flags &= ~O_EXEC; fputs("O_EXEC", fp); } else fputs("O_RDONLY", fp); printed = true; mode = 0; break; case O_WRONLY: fputs("O_WRONLY", fp); printed = true; mode = 0; break; case O_RDWR: fputs("O_RDWR", fp); printed = true; mode = 0; break; default: printed = false; } val = (unsigned)flags; print_mask_part(fp, openflags, &val, &printed); if (rem != NULL) *rem = val | mode; return (printed); } bool sysdecode_fcntl_fileflags(FILE *fp, int flags, int *rem) { bool printed; int oflags; /* * The file flags used with F_GETFL/F_SETFL mostly match the * flags passed to open(2). However, a few open-only flag * bits have been repurposed for fcntl-only flags. */ oflags = flags & ~(O_NOFOLLOW | FRDAHEAD); printed = sysdecode_open_flags(fp, oflags, rem); if (flags & O_NOFOLLOW) { fprintf(fp, "%sFPOIXSHM", printed ? "|" : ""); printed = true; } if (flags & FRDAHEAD) { fprintf(fp, "%sFRDAHEAD", printed ? "|" : ""); printed = true; } return (printed); } bool sysdecode_flock_operation(FILE *fp, int operation, int *rem) { return (print_mask_int(fp, flockops, operation, rem)); } static struct name_table getfsstatmode[] = { X(MNT_WAIT) X(MNT_NOWAIT) XEND }; const char * sysdecode_getfsstat_mode(int mode) { return (lookup_value(getfsstatmode, mode)); } const char * sysdecode_getrusage_who(int who) { return (lookup_value(rusage, who)); } static struct name_table kevent_user_ffctrl[] = { X(NOTE_FFNOP) X(NOTE_FFAND) X(NOTE_FFOR) X(NOTE_FFCOPY) XEND }; static struct name_table kevent_rdwr_fflags[] = { X(NOTE_LOWAT) X(NOTE_FILE_POLL) XEND }; static struct name_table kevent_vnode_fflags[] = { X(NOTE_DELETE) X(NOTE_WRITE) X(NOTE_EXTEND) X(NOTE_ATTRIB) X(NOTE_LINK) X(NOTE_RENAME) X(NOTE_REVOKE) X(NOTE_OPEN) X(NOTE_CLOSE) X(NOTE_CLOSE_WRITE) X(NOTE_READ) XEND }; static struct name_table kevent_proc_fflags[] = { - X(NOTE_EXIT) X(NOTE_FORK) X(NOTE_EXEC) X(NOTE_TRACK) X(NOTE_TRACKERR) - X(NOTE_CHILD) XEND + X(NOTE_EXIT) X(NOTE_FORK) X(NOTE_EXEC) X(NOTE_REAP) + X(NOTE_TRACK) X(NOTE_TRACKERR) X(NOTE_CHILD) XEND }; static struct name_table kevent_timer_fflags[] = { X(NOTE_SECONDS) X(NOTE_MSECONDS) X(NOTE_USECONDS) X(NOTE_NSECONDS) X(NOTE_ABSTIME) XEND }; void sysdecode_kevent_fflags(FILE *fp, short filter, int fflags, int base) { int rem; if (fflags == 0) { fputs("0", fp); return; } switch (filter) { case EVFILT_READ: case EVFILT_WRITE: if (!print_mask_int(fp, kevent_rdwr_fflags, fflags, &rem)) fprintf(fp, "%#x", rem); else if (rem != 0) fprintf(fp, "|%#x", rem); break; case EVFILT_VNODE: if (!print_mask_int(fp, kevent_vnode_fflags, fflags, &rem)) fprintf(fp, "%#x", rem); else if (rem != 0) fprintf(fp, "|%#x", rem); break; case EVFILT_PROC: case EVFILT_PROCDESC: if (!print_mask_int(fp, kevent_proc_fflags, fflags, &rem)) fprintf(fp, "%#x", rem); else if (rem != 0) fprintf(fp, "|%#x", rem); break; case EVFILT_TIMER: if (!print_mask_int(fp, kevent_timer_fflags, fflags, &rem)) fprintf(fp, "%#x", rem); else if (rem != 0) fprintf(fp, "|%#x", rem); break; case EVFILT_USER: { unsigned int ctrl, data; ctrl = fflags & NOTE_FFCTRLMASK; data = fflags & NOTE_FFLAGSMASK; if (fflags & NOTE_TRIGGER) { fputs("NOTE_TRIGGER", fp); if (fflags == NOTE_TRIGGER) return; fputc('|', fp); } /* * An event with 'ctrl' == NOTE_FFNOP is either a reported * (output) event for which only 'data' should be output * or a pointless input event. Assume that pointless * input events don't occur in practice. An event with * NOTE_TRIGGER is always an input event. */ if (ctrl != NOTE_FFNOP || fflags & NOTE_TRIGGER) { fprintf(fp, "%s|%#x", lookup_value(kevent_user_ffctrl, ctrl), data); } else { print_integer(fp, data, base); } break; } default: print_integer(fp, fflags, base); break; } } bool sysdecode_kevent_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, keventflags, flags, rem)); } const char * sysdecode_kevent_filter(int filter) { return (lookup_value(keventfilters, filter)); } const char * sysdecode_kldsym_cmd(int cmd) { return (lookup_value(kldsymcmd, cmd)); } const char * sysdecode_kldunload_flags(int flags) { return (lookup_value(kldunloadfflags, flags)); } const char * sysdecode_lio_listio_mode(int mode) { return (lookup_value(lio_listiomodes, mode)); } const char * sysdecode_madvice(int advice) { return (lookup_value(madvisebehav, advice)); } const char * sysdecode_minherit_inherit(int inherit) { return (lookup_value(minheritflags, inherit)); } bool sysdecode_mlockall_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, mlockallflags, flags, rem)); } bool sysdecode_mmap_prot(FILE *fp, int prot, int *rem) { int protm; bool printed; printed = false; protm = PROT_MAX_EXTRACT(prot); prot &= ~PROT_MAX(protm); if (protm != 0) { fputs("PROT_MAX(", fp); printed = print_mask_int(fp, mmapprot, protm, rem); fputs(")|", fp); } return (print_mask_int(fp, mmapprot, prot, rem) || printed); } bool sysdecode_fileflags(FILE *fp, fflags_t flags, fflags_t *rem) { return (print_mask_0(fp, fileflags, flags, rem)); } bool sysdecode_filemode(FILE *fp, int mode, int *rem) { return (print_mask_0(fp, filemode, mode, rem)); } bool sysdecode_mount_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, mountflags, flags, rem)); } bool sysdecode_msync_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, msyncflags, flags, rem)); } const char * sysdecode_nfssvc_flags(int flags) { return (lookup_value(nfssvcflags, flags)); } static struct name_table pipe2flags[] = { X(O_CLOEXEC) X(O_NONBLOCK) XEND }; bool sysdecode_pipe2_flags(FILE *fp, int flags, int *rem) { return (print_mask_0(fp, pipe2flags, flags, rem)); } bool sysdecode_pollfd_events(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, pollfdevents, flags, rem)); } const char * sysdecode_prio_which(int which) { return (lookup_value(prio, which)); } const char * sysdecode_procctl_cmd(int cmd) { return (lookup_value(procctlcmd, cmd)); } const char * sysdecode_ptrace_request(int request) { return (lookup_value(ptraceop, request)); } static struct name_table quotatypes[] = { X(GRPQUOTA) X(USRQUOTA) XEND }; bool sysdecode_quotactl_cmd(FILE *fp, int cmd) { const char *primary, *type; primary = lookup_value(quotactlcmds, cmd >> SUBCMDSHIFT); if (primary == NULL) return (false); fprintf(fp, "QCMD(%s,", primary); type = lookup_value(quotatypes, cmd & SUBCMDMASK); if (type != NULL) fprintf(fp, "%s", type); else fprintf(fp, "%#x", cmd & SUBCMDMASK); fprintf(fp, ")"); return (true); } bool sysdecode_reboot_howto(FILE *fp, int howto, int *rem) { bool printed; /* * RB_AUTOBOOT is special in that its value is zero, but it is * also an implied argument if a different operation is not * requested via RB_HALT, RB_POWERCYCLE, RB_POWEROFF, or * RB_REROOT. */ if (howto != 0 && (howto & (RB_HALT | RB_POWEROFF | RB_REROOT | RB_POWERCYCLE)) == 0) { fputs("RB_AUTOBOOT|", fp); printed = true; } else printed = false; return (print_mask_int(fp, rebootopt, howto, rem) || printed); } bool sysdecode_rfork_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, rforkflags, flags, rem)); } const char * sysdecode_rlimit(int resource) { return (lookup_value(rlimit, resource)); } const char * sysdecode_scheduler_policy(int policy) { return (lookup_value(schedpolicy, policy)); } bool sysdecode_sendfile_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, sendfileflags, flags, rem)); } bool sysdecode_shmat_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, shmatflags, flags, rem)); } const char * sysdecode_shutdown_how(int how) { return (lookup_value(shutdownhow, how)); } const char * sysdecode_sigbus_code(int si_code) { return (lookup_value(sigbuscode, si_code)); } const char * sysdecode_sigchld_code(int si_code) { return (lookup_value(sigchldcode, si_code)); } const char * sysdecode_sigfpe_code(int si_code) { return (lookup_value(sigfpecode, si_code)); } const char * sysdecode_sigill_code(int si_code) { return (lookup_value(sigillcode, si_code)); } const char * sysdecode_sigsegv_code(int si_code) { return (lookup_value(sigsegvcode, si_code)); } const char * sysdecode_sigtrap_code(int si_code) { return (lookup_value(sigtrapcode, si_code)); } const char * sysdecode_sigprocmask_how(int how) { return (lookup_value(sigprocmaskhow, how)); } const char * sysdecode_socketdomain(int domain) { return (lookup_value(sockdomain, domain)); } const char * sysdecode_socket_protocol(int domain, int protocol) { switch (domain) { case PF_INET: case PF_INET6: return (lookup_value(sockipproto, protocol)); default: return (NULL); } } const char * sysdecode_sockaddr_family(int sa_family) { return (lookup_value(sockfamily, sa_family)); } const char * sysdecode_ipproto(int protocol) { return (lookup_value(sockipproto, protocol)); } const char * sysdecode_sockopt_name(int level, int optname) { if (level == SOL_SOCKET) return (lookup_value(sockopt, optname)); if (level == IPPROTO_IP) /* XXX: UNIX domain socket options use a level of 0 also. */ return (lookup_value(sockoptip, optname)); if (level == IPPROTO_IPV6) return (lookup_value(sockoptipv6, optname)); if (level == IPPROTO_SCTP) return (lookup_value(sockoptsctp, optname)); if (level == IPPROTO_TCP) return (lookup_value(sockopttcp, optname)); if (level == IPPROTO_UDP) return (lookup_value(sockoptudp, optname)); if (level == IPPROTO_UDPLITE) return (lookup_value(sockoptudplite, optname)); return (NULL); } bool sysdecode_thr_create_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, thrcreateflags, flags, rem)); } const char * sysdecode_umtx_op(int op) { return (lookup_value(umtxop, op)); } bool sysdecode_umtx_op_flags(FILE *fp, int op, int *rem) { uintmax_t val; bool printed; printed = false; val = (unsigned)op; print_mask_part(fp, umtxopflags, &val, &printed); if (rem != NULL) *rem = val; return (printed); } const char * sysdecode_vmresult(int result) { return (lookup_value(vmresult, result)); } bool sysdecode_wait4_options(FILE *fp, int options, int *rem) { bool printed; int opt6; /* A flags value of 0 is normal. */ if (options == 0) { fputs("0", fp); if (rem != NULL) *rem = 0; return (true); } /* * These flags are implicit and aren't valid flags for wait4() * directly (though they don't fail with EINVAL). */ opt6 = options & (WEXITED | WTRAPPED); options &= ~opt6; printed = print_mask_int(fp, wait6opt, options, rem); if (rem != NULL) *rem |= opt6; return (printed); } bool sysdecode_wait6_options(FILE *fp, int options, int *rem) { return (print_mask_int(fp, wait6opt, options, rem)); } const char * sysdecode_whence(int whence) { return (lookup_value(seekwhence, whence)); } const char * sysdecode_fcntl_cmd(int cmd) { return (lookup_value(fcntlcmd, cmd)); } static struct name_table fcntl_fd_arg[] = { X(FD_CLOEXEC) X(0) XEND }; bool sysdecode_fcntl_arg_p(int cmd) { switch (cmd) { case F_GETFD: case F_GETFL: case F_GETOWN: return (false); default: return (true); } } void sysdecode_fcntl_arg(FILE *fp, int cmd, uintptr_t arg, int base) { int rem; switch (cmd) { case F_SETFD: if (!print_value(fp, fcntl_fd_arg, arg)) print_integer(fp, arg, base); break; case F_SETFL: if (!sysdecode_fcntl_fileflags(fp, arg, &rem)) fprintf(fp, "%#x", rem); else if (rem != 0) fprintf(fp, "|%#x", rem); break; case F_GETLK: case F_SETLK: case F_SETLKW: fprintf(fp, "%p", (void *)arg); break; default: print_integer(fp, arg, base); break; } } bool sysdecode_mmap_flags(FILE *fp, int flags, int *rem) { uintmax_t val; bool printed; int align; /* * MAP_ALIGNED can't be handled directly by print_mask_int(). */ printed = false; align = flags & MAP_ALIGNMENT_MASK; val = (unsigned)flags & ~MAP_ALIGNMENT_MASK; print_mask_part(fp, mmapflags, &val, &printed); if (align != 0) { if (printed) fputc('|', fp); if (align == MAP_ALIGNED_SUPER) fputs("MAP_ALIGNED_SUPER", fp); else fprintf(fp, "MAP_ALIGNED(%d)", align >> MAP_ALIGNMENT_SHIFT); printed = true; } if (rem != NULL) *rem = val; return (printed); } const char * sysdecode_pathconf_name(int name) { return (lookup_value(pathconfname, name)); } const char * sysdecode_rtprio_function(int function) { return (lookup_value(rtpriofuncs, function)); } bool sysdecode_msg_flags(FILE *fp, int flags, int *rem) { return (print_mask_0(fp, msgflags, flags, rem)); } const char * sysdecode_sigcode(int sig, int si_code) { const char *str; str = lookup_value(sigcode, si_code); if (str != NULL) return (str); switch (sig) { case SIGILL: return (sysdecode_sigill_code(si_code)); case SIGBUS: return (sysdecode_sigbus_code(si_code)); case SIGSEGV: return (sysdecode_sigsegv_code(si_code)); case SIGFPE: return (sysdecode_sigfpe_code(si_code)); case SIGTRAP: return (sysdecode_sigtrap_code(si_code)); case SIGCHLD: return (sysdecode_sigchld_code(si_code)); default: return (NULL); } } const char * sysdecode_sysarch_number(int number) { return (lookup_value(sysarchnum, number)); } bool sysdecode_umtx_cvwait_flags(FILE *fp, u_long flags, u_long *rem) { return (print_mask_0ul(fp, umtxcvwaitflags, flags, rem)); } bool sysdecode_umtx_rwlock_flags(FILE *fp, u_long flags, u_long *rem) { return (print_mask_0ul(fp, umtxrwlockflags, flags, rem)); } void sysdecode_cap_rights(FILE *fp, cap_rights_t *rightsp) { cap_rights_t diff, sum, zero; const struct name_table *t; int i; bool comma; for (i = 0; i < CAPARSIZE(rightsp); i++) { if (CAPIDXBIT(rightsp->cr_rights[i]) != 1 << i) { fprintf(fp, "invalid cap_rights_t"); return; } } cap_rights_init(&sum); diff = *rightsp; for (t = caprights, comma = false; t->str != NULL; t++) { if (cap_rights_is_set(rightsp, t->val)) { cap_rights_clear(&diff, t->val); if (cap_rights_is_set(&sum, t->val)) { /* Don't print redundant rights. */ continue; } cap_rights_set(&sum, t->val); fprintf(fp, "%s%s", comma ? "," : "", t->str); comma = true; } } if (!comma) fprintf(fp, "CAP_NONE"); /* * Provide a breadcrumb if some of the provided rights are not included * in the table, likely due to a bug in the mktables script. */ CAP_NONE(&zero); if (!cap_rights_contains(&zero, &diff)) fprintf(fp, ",unknown rights"); } /* * Pre-sort the set of rights, which has a partial ordering defined by the * subset relation. This lets sysdecode_cap_rights() print a list of minimal * length with a single pass over the "caprights" table. */ static void __attribute__((constructor)) sysdecode_cap_rights_init(void) { cap_rights_t tr, qr; struct name_table *t, *q, tmp; bool swapped; do { for (t = caprights, swapped = false; t->str != NULL; t++) { cap_rights_init(&tr, t->val); for (q = t + 1; q->str != NULL; q++) { cap_rights_init(&qr, q->val); if (cap_rights_contains(&qr, &tr)) { tmp = *t; *t = *q; *q = tmp; swapped = true; } } } } while (swapped); } static struct name_table cmsgtypeip[] = { X(IP_RECVDSTADDR) X(IP_RECVTTL) X(IP_RECVOPTS) X(IP_RECVRETOPTS) X(IP_RECVIF) X(IP_RECVTOS) X(IP_FLOWID) X(IP_FLOWTYPE) X(IP_RSSBUCKETID) XEND }; static struct name_table cmsgtypeipv6[] = { #if 0 /* The RFC 2292 defines are kernel space only. */ X(IPV6_2292PKTINFO) X(IPV6_2292HOPLIMIT) X(IPV6_2292HOPOPTS) X(IPV6_2292DSTOPTS) X(IPV6_2292RTHDR) X(IPV6_2292NEXTHOP) #endif X(IPV6_PKTINFO) X(IPV6_HOPLIMIT) X(IPV6_HOPOPTS) X(IPV6_DSTOPTS) X(IPV6_RTHDR) X(IPV6_NEXTHOP) X(IPV6_TCLASS) X(IPV6_FLOWID) X(IPV6_FLOWTYPE) X(IPV6_RSSBUCKETID) X(IPV6_PATHMTU) X(IPV6_RTHDRDSTOPTS) X(IPV6_USE_MIN_MTU) X(IPV6_DONTFRAG) X(IPV6_PREFER_TEMPADDR) XEND }; static struct name_table cmsgtypesctp[] = { X(SCTP_INIT) X(SCTP_SNDRCV) X(SCTP_EXTRCV) X(SCTP_SNDINFO) X(SCTP_RCVINFO) X(SCTP_NXTINFO) X(SCTP_PRINFO) X(SCTP_AUTHINFO) X(SCTP_DSTADDRV4) X(SCTP_DSTADDRV6) XEND }; const char * sysdecode_cmsg_type(int cmsg_level, int cmsg_type) { if (cmsg_level == SOL_SOCKET) return (lookup_value(cmsgtypesocket, cmsg_type)); if (cmsg_level == IPPROTO_IP) return (lookup_value(cmsgtypeip, cmsg_type)); if (cmsg_level == IPPROTO_IPV6) return (lookup_value(cmsgtypeipv6, cmsg_type)); if (cmsg_level == IPPROTO_SCTP) return (lookup_value(cmsgtypesctp, cmsg_type)); return (NULL); } const char * sysdecode_sctp_pr_policy(int policy) { return (lookup_value(sctpprpolicy, policy)); } static struct name_table sctpsndflags[] = { X(SCTP_EOF) X(SCTP_ABORT) X(SCTP_UNORDERED) X(SCTP_ADDR_OVER) X(SCTP_SENDALL) X(SCTP_EOR) X(SCTP_SACK_IMMEDIATELY) XEND }; bool sysdecode_sctp_snd_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, sctpsndflags, flags, rem)); } static struct name_table sctprcvflags[] = { X(SCTP_UNORDERED) XEND }; bool sysdecode_sctp_rcv_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, sctprcvflags, flags, rem)); } static struct name_table sctpnxtflags[] = { X(SCTP_UNORDERED) X(SCTP_COMPLETE) X(SCTP_NOTIFICATION) XEND }; bool sysdecode_sctp_nxt_flags(FILE *fp, int flags, int *rem) { return (print_mask_int(fp, sctpnxtflags, flags, rem)); } static struct name_table sctpsinfoflags[] = { X(SCTP_EOF) X(SCTP_ABORT) X(SCTP_UNORDERED) X(SCTP_ADDR_OVER) X(SCTP_SENDALL) X(SCTP_EOR) X(SCTP_SACK_IMMEDIATELY) XEND }; void sysdecode_sctp_sinfo_flags(FILE *fp, int sinfo_flags) { const char *temp; int rem; bool printed; printed = print_mask_0(fp, sctpsinfoflags, sinfo_flags, &rem); if (rem & ~SCTP_PR_SCTP_ALL) { fprintf(fp, "%s%#x", printed ? "|" : "", rem & ~SCTP_PR_SCTP_ALL); printed = true; rem &= ~SCTP_PR_SCTP_ALL; } if (rem != 0) { temp = sysdecode_sctp_pr_policy(rem); if (temp != NULL) { fprintf(fp, "%s%s", printed ? "|" : "", temp); } else { fprintf(fp, "%s%#x", printed ? "|" : "", rem); } } } bool sysdecode_shmflags(FILE *fp, int flags, int *rem) { return (print_mask_0(fp, shmflags, flags, rem)); } const char * sysdecode_itimer(int which) { return (lookup_value(itimerwhich, which)); } diff --git a/sys/kern/kern_event.c b/sys/kern/kern_event.c index e2f44aea27b4..a735dfaa97df 100644 --- a/sys/kern/kern_event.c +++ b/sys/kern/kern_event.c @@ -1,3122 +1,3125 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 1999,2000,2001 Jonathan Lemon * Copyright 2004 John-Mark Gurney * Copyright (c) 2009 Apple, Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include "opt_ktrace.h" #include "opt_kqueue.h" #ifdef COMPAT_FREEBSD11 #define _WANT_FREEBSD11_KEVENT #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef KTRACE #include #endif #include #ifdef COMPAT_FREEBSD32 #include #include #endif #include static MALLOC_DEFINE(M_KQUEUE, "kqueue", "memory for kqueue system"); /* * This lock is used if multiple kq locks are required. This possibly * should be made into a per proc lock. */ static struct mtx kq_global; MTX_SYSINIT(kq_global, &kq_global, "kqueue order", MTX_DEF); #define KQ_GLOBAL_LOCK(lck, haslck) do { \ if (!haslck) \ mtx_lock(lck); \ haslck = 1; \ } while (0) #define KQ_GLOBAL_UNLOCK(lck, haslck) do { \ if (haslck) \ mtx_unlock(lck); \ haslck = 0; \ } while (0) TASKQUEUE_DEFINE_THREAD(kqueue_ctx); static int kevent_copyout(void *arg, struct kevent *kevp, int count); static int kevent_copyin(void *arg, struct kevent *kevp, int count); static int kqueue_register(struct kqueue *kq, struct kevent *kev, struct thread *td, int mflag); static int kqueue_acquire(struct file *fp, struct kqueue **kqp); static void kqueue_release(struct kqueue *kq, int locked); static void kqueue_destroy(struct kqueue *kq); static void kqueue_drain(struct kqueue *kq, struct thread *td); static int kqueue_expand(struct kqueue *kq, const struct filterops *fops, uintptr_t ident, int mflag); static void kqueue_task(void *arg, int pending); static int kqueue_scan(struct kqueue *kq, int maxevents, struct kevent_copyops *k_ops, const struct timespec *timeout, struct kevent *keva, struct thread *td); static void kqueue_wakeup(struct kqueue *kq); static const struct filterops *kqueue_fo_find(int filt); static void kqueue_fo_release(int filt); struct g_kevent_args; static int kern_kevent_generic(struct thread *td, struct g_kevent_args *uap, struct kevent_copyops *k_ops, const char *struct_name); static fo_ioctl_t kqueue_ioctl; static fo_poll_t kqueue_poll; static fo_kqfilter_t kqueue_kqfilter; static fo_stat_t kqueue_stat; static fo_close_t kqueue_close; static fo_fill_kinfo_t kqueue_fill_kinfo; static const struct fileops kqueueops = { .fo_read = invfo_rdwr, .fo_write = invfo_rdwr, .fo_truncate = invfo_truncate, .fo_ioctl = kqueue_ioctl, .fo_poll = kqueue_poll, .fo_kqfilter = kqueue_kqfilter, .fo_stat = kqueue_stat, .fo_close = kqueue_close, .fo_chmod = invfo_chmod, .fo_chown = invfo_chown, .fo_sendfile = invfo_sendfile, .fo_cmp = file_kcmp_generic, .fo_fill_kinfo = kqueue_fill_kinfo, }; static int knote_attach(struct knote *kn, struct kqueue *kq); static void knote_drop(struct knote *kn, struct thread *td); static void knote_drop_detached(struct knote *kn, struct thread *td); static void knote_enqueue(struct knote *kn); static void knote_dequeue(struct knote *kn); static void knote_init(void *); static struct knote *knote_alloc(int mflag); static void knote_free(struct knote *kn); static void filt_kqdetach(struct knote *kn); static int filt_kqueue(struct knote *kn, long hint); static int filt_procattach(struct knote *kn); static void filt_procdetach(struct knote *kn); static int filt_proc(struct knote *kn, long hint); static int filt_fileattach(struct knote *kn); static void filt_timerexpire(void *knx); static void filt_timerexpire_l(struct knote *kn, bool proc_locked); static int filt_timerattach(struct knote *kn); static void filt_timerdetach(struct knote *kn); static void filt_timerstart(struct knote *kn, sbintime_t to); static void filt_timertouch(struct knote *kn, struct kevent *kev, u_long type); static int filt_timervalidate(struct knote *kn, sbintime_t *to); static int filt_timer(struct knote *kn, long hint); static int filt_userattach(struct knote *kn); static void filt_userdetach(struct knote *kn); static int filt_user(struct knote *kn, long hint); static void filt_usertouch(struct knote *kn, struct kevent *kev, u_long type); static const struct filterops file_filtops = { .f_isfd = 1, .f_attach = filt_fileattach, }; static const struct filterops kqread_filtops = { .f_isfd = 1, .f_detach = filt_kqdetach, .f_event = filt_kqueue, }; /* XXX - move to kern_proc.c? */ static const struct filterops proc_filtops = { .f_isfd = 0, .f_attach = filt_procattach, .f_detach = filt_procdetach, .f_event = filt_proc, }; static const struct filterops timer_filtops = { .f_isfd = 0, .f_attach = filt_timerattach, .f_detach = filt_timerdetach, .f_event = filt_timer, .f_touch = filt_timertouch, }; static const struct filterops user_filtops = { .f_attach = filt_userattach, .f_detach = filt_userdetach, .f_event = filt_user, .f_touch = filt_usertouch, }; static uma_zone_t knote_zone; static unsigned int __exclusive_cache_line kq_ncallouts; static unsigned int kq_calloutmax = 4 * 1024; SYSCTL_UINT(_kern, OID_AUTO, kq_calloutmax, CTLFLAG_RW, &kq_calloutmax, 0, "Maximum number of callouts allocated for kqueue"); /* XXX - ensure not influx ? */ #define KNOTE_ACTIVATE(kn, islock) do { \ if ((islock)) \ mtx_assert(&(kn)->kn_kq->kq_lock, MA_OWNED); \ else \ KQ_LOCK((kn)->kn_kq); \ (kn)->kn_status |= KN_ACTIVE; \ if (((kn)->kn_status & (KN_QUEUED | KN_DISABLED)) == 0) \ knote_enqueue((kn)); \ if (!(islock)) \ KQ_UNLOCK((kn)->kn_kq); \ } while (0) #define KQ_LOCK(kq) do { \ mtx_lock(&(kq)->kq_lock); \ } while (0) #define KQ_FLUX_WAKEUP(kq) do { \ if (((kq)->kq_state & KQ_FLUXWAIT) == KQ_FLUXWAIT) { \ (kq)->kq_state &= ~KQ_FLUXWAIT; \ wakeup((kq)); \ } \ } while (0) #define KQ_UNLOCK_FLUX(kq) do { \ KQ_FLUX_WAKEUP(kq); \ mtx_unlock(&(kq)->kq_lock); \ } while (0) #define KQ_UNLOCK(kq) do { \ mtx_unlock(&(kq)->kq_lock); \ } while (0) #define KQ_OWNED(kq) do { \ mtx_assert(&(kq)->kq_lock, MA_OWNED); \ } while (0) #define KQ_NOTOWNED(kq) do { \ mtx_assert(&(kq)->kq_lock, MA_NOTOWNED); \ } while (0) static struct knlist * kn_list_lock(struct knote *kn) { struct knlist *knl; knl = kn->kn_knlist; if (knl != NULL) knl->kl_lock(knl->kl_lockarg); return (knl); } static void kn_list_unlock(struct knlist *knl) { bool do_free; if (knl == NULL) return; do_free = knl->kl_autodestroy && knlist_empty(knl); knl->kl_unlock(knl->kl_lockarg); if (do_free) { knlist_destroy(knl); free(knl, M_KQUEUE); } } static bool kn_in_flux(struct knote *kn) { return (kn->kn_influx > 0); } static void kn_enter_flux(struct knote *kn) { KQ_OWNED(kn->kn_kq); MPASS(kn->kn_influx < INT_MAX); kn->kn_influx++; } static bool kn_leave_flux(struct knote *kn) { KQ_OWNED(kn->kn_kq); MPASS(kn->kn_influx > 0); kn->kn_influx--; return (kn->kn_influx == 0); } #define KNL_ASSERT_LOCK(knl, islocked) do { \ if (islocked) \ KNL_ASSERT_LOCKED(knl); \ else \ KNL_ASSERT_UNLOCKED(knl); \ } while (0) #ifdef INVARIANTS #define KNL_ASSERT_LOCKED(knl) do { \ knl->kl_assert_lock((knl)->kl_lockarg, LA_LOCKED); \ } while (0) #define KNL_ASSERT_UNLOCKED(knl) do { \ knl->kl_assert_lock((knl)->kl_lockarg, LA_UNLOCKED); \ } while (0) #else /* !INVARIANTS */ #define KNL_ASSERT_LOCKED(knl) do {} while (0) #define KNL_ASSERT_UNLOCKED(knl) do {} while (0) #endif /* INVARIANTS */ #ifndef KN_HASHSIZE #define KN_HASHSIZE 64 /* XXX should be tunable */ #endif #define KN_HASH(val, mask) (((val) ^ (val >> 8)) & (mask)) static int filt_nullattach(struct knote *kn) { return (ENXIO); }; static const struct filterops null_filtops = { .f_isfd = 0, .f_attach = filt_nullattach, }; /* XXX - make SYSINIT to add these, and move into respective modules. */ extern const struct filterops sig_filtops; extern const struct filterops fs_filtops; /* * Table for all system-defined filters. */ static struct mtx filterops_lock; MTX_SYSINIT(kqueue_filterops, &filterops_lock, "protect sysfilt_ops", MTX_DEF); static struct { const struct filterops *for_fop; int for_nolock; int for_refcnt; } sysfilt_ops[EVFILT_SYSCOUNT] = { [~EVFILT_READ] = { &file_filtops, 1 }, [~EVFILT_WRITE] = { &file_filtops, 1 }, [~EVFILT_AIO] = { &null_filtops }, [~EVFILT_VNODE] = { &file_filtops, 1 }, [~EVFILT_PROC] = { &proc_filtops, 1 }, [~EVFILT_SIGNAL] = { &sig_filtops, 1 }, [~EVFILT_TIMER] = { &timer_filtops, 1 }, [~EVFILT_PROCDESC] = { &file_filtops, 1 }, [~EVFILT_FS] = { &fs_filtops, 1 }, [~EVFILT_LIO] = { &null_filtops }, [~EVFILT_USER] = { &user_filtops, 1 }, [~EVFILT_SENDFILE] = { &null_filtops }, [~EVFILT_EMPTY] = { &file_filtops, 1 }, }; /* * Simple redirection for all cdevsw style objects to call their fo_kqfilter * method. */ static int filt_fileattach(struct knote *kn) { return (fo_kqfilter(kn->kn_fp, kn)); } /*ARGSUSED*/ static int kqueue_kqfilter(struct file *fp, struct knote *kn) { struct kqueue *kq = kn->kn_fp->f_data; if (kn->kn_filter != EVFILT_READ) return (EINVAL); kn->kn_status |= KN_KQUEUE; kn->kn_fop = &kqread_filtops; knlist_add(&kq->kq_sel.si_note, kn, 0); return (0); } static void filt_kqdetach(struct knote *kn) { struct kqueue *kq = kn->kn_fp->f_data; knlist_remove(&kq->kq_sel.si_note, kn, 0); } /*ARGSUSED*/ static int filt_kqueue(struct knote *kn, long hint) { struct kqueue *kq = kn->kn_fp->f_data; kn->kn_data = kq->kq_count; return (kn->kn_data > 0); } /* XXX - move to kern_proc.c? */ static int filt_procattach(struct knote *kn) { struct proc *p; int error; bool exiting, immediate; exiting = immediate = false; - if (kn->kn_sfflags & NOTE_EXIT) + if (kn->kn_sfflags & (NOTE_EXIT | NOTE_REAP)) p = pfind_any(kn->kn_id); else p = pfind(kn->kn_id); if (p == NULL) return (ESRCH); if (p->p_flag & P_WEXIT) exiting = true; if ((error = p_cansee(curthread, p))) { PROC_UNLOCK(p); return (error); } kn->kn_ptr.p_proc = p; kn->kn_flags |= EV_CLEAR; /* automatically set */ /* * Internal flag indicating registration done by kernel for the * purposes of getting a NOTE_CHILD notification. */ if (kn->kn_flags & EV_FLAG2) { kn->kn_flags &= ~EV_FLAG2; kn->kn_data = kn->kn_sdata; /* ppid */ kn->kn_fflags = NOTE_CHILD; - kn->kn_sfflags &= ~(NOTE_EXIT | NOTE_EXEC | NOTE_FORK); + kn->kn_sfflags &= ~NOTE_PCTRLMASK; immediate = true; /* Force immediate activation of child note. */ } /* * Internal flag indicating registration done by kernel (for other than * NOTE_CHILD). */ if (kn->kn_flags & EV_FLAG1) { kn->kn_flags &= ~EV_FLAG1; } knlist_add(p->p_klist, kn, 1); /* * Immediately activate any child notes or, in the case of a zombie * target process, exit notes. The latter is necessary to handle the * case where the target process, e.g. a child, dies before the kevent * is registered. */ if (immediate || (exiting && filt_proc(kn, NOTE_EXIT))) KNOTE_ACTIVATE(kn, 0); PROC_UNLOCK(p); return (0); } /* * The knote may be attached to a different process, which may exit, * leaving nothing for the knote to be attached to. So when the process * exits, the knote is marked as DETACHED and also flagged as ONESHOT so * it will be deleted when read out. However, as part of the knote deletion, * this routine is called, so a check is needed to avoid actually performing * a detach, because the original process does not exist any more. */ /* XXX - move to kern_proc.c? */ static void filt_procdetach(struct knote *kn) { knlist_remove(kn->kn_knlist, kn, 0); kn->kn_ptr.p_proc = NULL; } /* XXX - move to kern_proc.c? */ static int filt_proc(struct knote *kn, long hint) { struct proc *p; u_int event; p = kn->kn_ptr.p_proc; if (p == NULL) /* already activated, from attach filter */ return (0); /* Mask off extra data. */ event = (u_int)hint & NOTE_PCTRLMASK; /* If the user is interested in this event, record it. */ if ((kn->kn_sfflags & event) != 0) kn->kn_fflags |= kn->kn_sfflags & event; + /* Report exit status */ + if ((kn->kn_fflags & NOTE_EXIT) != 0) + kn->kn_data = KW_EXITCODE(p->p_xexit, p->p_xsig); + /* Process is gone, so flag the event as finished. */ - if ((event & NOTE_EXIT) != 0) { + if ((event & NOTE_REAP) != 0 || + ((event & NOTE_EXIT) != 0 && (kn->kn_sfflags & NOTE_REAP) == 0)) { kn->kn_flags |= EV_EOF | EV_ONESHOT; kn->kn_ptr.p_proc = NULL; - if (kn->kn_fflags & NOTE_EXIT) - kn->kn_data = KW_EXITCODE(p->p_xexit, p->p_xsig); if (kn->kn_fflags == 0) kn->kn_flags |= EV_DROP; return (1); } return (kn->kn_fflags != 0); } /* * Called when the process forked. It mostly does the same as the * knote(), activating all knotes registered to be activated when the * process forked. Additionally, for each knote attached to the * parent, check whether user wants to track the new process. If so * attach a new knote to it, and immediately report an event with the * child's pid. */ void knote_fork(struct knlist *list, int pid) { struct kqueue *kq; struct knote *kn; struct kevent kev; int error; MPASS(list != NULL); KNL_ASSERT_LOCKED(list); if (SLIST_EMPTY(&list->kl_list)) return; memset(&kev, 0, sizeof(kev)); SLIST_FOREACH(kn, &list->kl_list, kn_selnext) { kq = kn->kn_kq; KQ_LOCK(kq); if (kn_in_flux(kn) && (kn->kn_status & KN_SCAN) == 0) { KQ_UNLOCK(kq); continue; } /* * The same as knote(), activate the event. */ if ((kn->kn_sfflags & NOTE_TRACK) == 0) { if (kn->kn_fop->f_event(kn, NOTE_FORK)) KNOTE_ACTIVATE(kn, 1); KQ_UNLOCK(kq); continue; } /* * The NOTE_TRACK case. In addition to the activation * of the event, we need to register new events to * track the child. Drop the locks in preparation for * the call to kqueue_register(). */ kn_enter_flux(kn); KQ_UNLOCK(kq); list->kl_unlock(list->kl_lockarg); /* * Activate existing knote and register tracking knotes with * new process. * * First register a knote to get just the child notice. This * must be a separate note from a potential NOTE_EXIT * notification since both NOTE_CHILD and NOTE_EXIT are defined * to use the data field (in conflicting ways). */ kev.ident = pid; kev.filter = kn->kn_filter; kev.flags = kn->kn_flags | EV_ADD | EV_ENABLE | EV_ONESHOT | EV_FLAG2; kev.fflags = kn->kn_sfflags; kev.data = kn->kn_id; /* parent */ kev.udata = kn->kn_kevent.udata;/* preserve udata */ error = kqueue_register(kq, &kev, NULL, M_NOWAIT); if (error) kn->kn_fflags |= NOTE_TRACKERR; /* * Then register another knote to track other potential events * from the new process. */ kev.ident = pid; kev.filter = kn->kn_filter; kev.flags = kn->kn_flags | EV_ADD | EV_ENABLE | EV_FLAG1; kev.fflags = kn->kn_sfflags; kev.data = kn->kn_id; /* parent */ kev.udata = kn->kn_kevent.udata;/* preserve udata */ error = kqueue_register(kq, &kev, NULL, M_NOWAIT); /* * Serialize updates to the kn_kevent fields with threads * scanning the queue. */ list->kl_lock(list->kl_lockarg); if (error) kn->kn_fflags |= NOTE_TRACKERR; if (kn->kn_fop->f_event(kn, NOTE_FORK)) { KQ_LOCK(kq); KNOTE_ACTIVATE(kn, 1); } else { KQ_LOCK(kq); } kn_leave_flux(kn); KQ_UNLOCK_FLUX(kq); } } /* * XXX: EVFILT_TIMER should perhaps live in kern_time.c beside the * interval timer support code. */ #define NOTE_TIMER_PRECMASK \ (NOTE_SECONDS | NOTE_MSECONDS | NOTE_USECONDS | NOTE_NSECONDS) static sbintime_t timer2sbintime(int64_t data, int flags) { int64_t secs; /* * Macros for converting to the fractional second portion of an * sbintime_t using 64bit multiplication to improve precision. */ #define NS_TO_SBT(ns) (((ns) * (((uint64_t)1 << 63) / 500000000)) >> 32) #define US_TO_SBT(us) (((us) * (((uint64_t)1 << 63) / 500000)) >> 32) #define MS_TO_SBT(ms) (((ms) * (((uint64_t)1 << 63) / 500)) >> 32) switch (flags & NOTE_TIMER_PRECMASK) { case NOTE_SECONDS: #ifdef __LP64__ if (data > (SBT_MAX / SBT_1S)) return (SBT_MAX); #endif return ((sbintime_t)data << 32); case NOTE_MSECONDS: /* FALLTHROUGH */ case 0: if (data >= 1000) { secs = data / 1000; #ifdef __LP64__ if (secs > (SBT_MAX / SBT_1S)) return (SBT_MAX); #endif return (secs << 32 | MS_TO_SBT(data % 1000)); } return (MS_TO_SBT(data)); case NOTE_USECONDS: if (data >= 1000000) { secs = data / 1000000; #ifdef __LP64__ if (secs > (SBT_MAX / SBT_1S)) return (SBT_MAX); #endif return (secs << 32 | US_TO_SBT(data % 1000000)); } return (US_TO_SBT(data)); case NOTE_NSECONDS: if (data >= 1000000000) { secs = data / 1000000000; #ifdef __LP64__ if (secs > (SBT_MAX / SBT_1S)) return (SBT_MAX); #endif return (secs << 32 | NS_TO_SBT(data % 1000000000)); } return (NS_TO_SBT(data)); default: break; } return (-1); } struct kq_timer_cb_data { struct callout c; struct proc *p; struct knote *kn; int cpuid; int flags; TAILQ_ENTRY(kq_timer_cb_data) link; sbintime_t next; /* next timer event fires at */ sbintime_t to; /* precalculated timer period, 0 for abs */ }; #define KQ_TIMER_CB_ENQUEUED 0x01 static void kqtimer_sched_callout(struct kq_timer_cb_data *kc) { callout_reset_sbt_on(&kc->c, kc->next, 0, filt_timerexpire, kc->kn, kc->cpuid, C_ABSOLUTE); } void kqtimer_proc_continue(struct proc *p) { struct kq_timer_cb_data *kc, *kc1; sbintime_t now; PROC_LOCK_ASSERT(p, MA_OWNED); now = sbinuptime(); TAILQ_FOREACH_SAFE(kc, &p->p_kqtim_stop, link, kc1) { TAILQ_REMOVE(&p->p_kqtim_stop, kc, link); kc->flags &= ~KQ_TIMER_CB_ENQUEUED; if (kc->next <= now) filt_timerexpire_l(kc->kn, true); else kqtimer_sched_callout(kc); } } static void filt_timerexpire_l(struct knote *kn, bool proc_locked) { struct kq_timer_cb_data *kc; struct proc *p; uint64_t delta; sbintime_t now; kc = kn->kn_ptr.p_v; if ((kn->kn_flags & EV_ONESHOT) != 0 || kc->to == 0) { kn->kn_data++; KNOTE_ACTIVATE(kn, 0); return; } now = sbinuptime(); if (now >= kc->next) { delta = (now - kc->next) / kc->to; if (delta == 0) delta = 1; kn->kn_data += delta; kc->next += delta * kc->to; if (now >= kc->next) /* overflow */ kc->next = now + kc->to; KNOTE_ACTIVATE(kn, 0); /* XXX - handle locking */ } /* * Initial check for stopped kc->p is racy. It is fine to * miss the set of the stop flags, at worst we would schedule * one more callout. On the other hand, it is not fine to not * schedule when we we missed clearing of the flags, we * recheck them under the lock and observe consistent state. */ p = kc->p; if (P_SHOULDSTOP(p) || P_KILLED(p)) { if (!proc_locked) PROC_LOCK(p); if (P_SHOULDSTOP(p) || P_KILLED(p)) { if ((kc->flags & KQ_TIMER_CB_ENQUEUED) == 0) { /* * Insert into head so that * kqtimer_proc_continue() does not * iterate into us again. */ kc->flags |= KQ_TIMER_CB_ENQUEUED; TAILQ_INSERT_HEAD(&p->p_kqtim_stop, kc, link); } if (!proc_locked) PROC_UNLOCK(p); return; } if (!proc_locked) PROC_UNLOCK(p); } kqtimer_sched_callout(kc); } static void filt_timerexpire(void *knx) { filt_timerexpire_l(knx, false); } /* * data contains amount of time to sleep */ static int filt_timervalidate(struct knote *kn, sbintime_t *to) { struct bintime bt; sbintime_t sbt; if (kn->kn_sdata < 0) return (EINVAL); if (kn->kn_sdata == 0 && (kn->kn_flags & EV_ONESHOT) == 0) kn->kn_sdata = 1; /* * The only fflags values supported are the timer unit * (precision) and the absolute time indicator. */ if ((kn->kn_sfflags & ~(NOTE_TIMER_PRECMASK | NOTE_ABSTIME)) != 0) return (EINVAL); *to = timer2sbintime(kn->kn_sdata, kn->kn_sfflags); if (*to < 0) return (EINVAL); if ((kn->kn_sfflags & NOTE_ABSTIME) != 0) { getboottimebin(&bt); sbt = bttosbt(bt); *to = MAX(0, *to - sbt); } return (0); } static int filt_timerattach(struct knote *kn) { struct kq_timer_cb_data *kc; sbintime_t to; int error; to = -1; error = filt_timervalidate(kn, &to); if (error != 0) return (error); KASSERT(to > 0 || (kn->kn_flags & EV_ONESHOT) != 0 || (kn->kn_sfflags & NOTE_ABSTIME) != 0, ("%s: periodic timer has a calculated zero timeout", __func__)); KASSERT(to >= 0, ("%s: timer has a calculated negative timeout", __func__)); if (atomic_fetchadd_int(&kq_ncallouts, 1) + 1 > kq_calloutmax) { atomic_subtract_int(&kq_ncallouts, 1); return (ENOMEM); } if ((kn->kn_sfflags & NOTE_ABSTIME) == 0) kn->kn_flags |= EV_CLEAR; /* automatically set */ kn->kn_status &= ~KN_DETACHED; /* knlist_add clears it */ kn->kn_ptr.p_v = kc = malloc(sizeof(*kc), M_KQUEUE, M_WAITOK); kc->kn = kn; kc->p = curproc; kc->cpuid = PCPU_GET(cpuid); kc->flags = 0; callout_init(&kc->c, 1); filt_timerstart(kn, to); return (0); } static void filt_timerstart(struct knote *kn, sbintime_t to) { struct kq_timer_cb_data *kc; kc = kn->kn_ptr.p_v; if ((kn->kn_sfflags & NOTE_ABSTIME) != 0) { kc->next = to; kc->to = 0; } else { kc->next = to + sbinuptime(); kc->to = to; } kqtimer_sched_callout(kc); } static void filt_timerdetach(struct knote *kn) { struct kq_timer_cb_data *kc; unsigned int old __unused; bool pending; kc = kn->kn_ptr.p_v; do { callout_drain(&kc->c); /* * kqtimer_proc_continue() might have rescheduled this callout. * Double-check, using the process mutex as an interlock. */ PROC_LOCK(kc->p); if ((kc->flags & KQ_TIMER_CB_ENQUEUED) != 0) { kc->flags &= ~KQ_TIMER_CB_ENQUEUED; TAILQ_REMOVE(&kc->p->p_kqtim_stop, kc, link); } pending = callout_pending(&kc->c); PROC_UNLOCK(kc->p); } while (pending); free(kc, M_KQUEUE); old = atomic_fetchadd_int(&kq_ncallouts, -1); KASSERT(old > 0, ("Number of callouts cannot become negative")); kn->kn_status |= KN_DETACHED; /* knlist_remove sets it */ } static void filt_timertouch(struct knote *kn, struct kevent *kev, u_long type) { struct kq_timer_cb_data *kc; struct kqueue *kq; sbintime_t to; int error; switch (type) { case EVENT_REGISTER: /* Handle re-added timers that update data/fflags */ if (kev->flags & EV_ADD) { kc = kn->kn_ptr.p_v; /* Drain any existing callout. */ callout_drain(&kc->c); /* Throw away any existing undelivered record * of the timer expiration. This is done under * the presumption that if a process is * re-adding this timer with new parameters, * it is no longer interested in what may have * happened under the old parameters. If it is * interested, it can wait for the expiration, * delete the old timer definition, and then * add the new one. * * This has to be done while the kq is locked: * - if enqueued, dequeue * - make it no longer active * - clear the count of expiration events */ kq = kn->kn_kq; KQ_LOCK(kq); if (kn->kn_status & KN_QUEUED) knote_dequeue(kn); kn->kn_status &= ~KN_ACTIVE; kn->kn_data = 0; KQ_UNLOCK(kq); /* Reschedule timer based on new data/fflags */ kn->kn_sfflags = kev->fflags; kn->kn_sdata = kev->data; error = filt_timervalidate(kn, &to); if (error != 0) { kn->kn_flags |= EV_ERROR; kn->kn_data = error; } else filt_timerstart(kn, to); } break; case EVENT_PROCESS: *kev = kn->kn_kevent; if (kn->kn_flags & EV_CLEAR) { kn->kn_data = 0; kn->kn_fflags = 0; } break; default: panic("filt_timertouch() - invalid type (%ld)", type); break; } } static int filt_timer(struct knote *kn, long hint) { return (kn->kn_data != 0); } static int filt_userattach(struct knote *kn) { /* * EVFILT_USER knotes are not attached to anything in the kernel. */ kn->kn_hook = NULL; if (kn->kn_fflags & NOTE_TRIGGER) kn->kn_hookid = 1; else kn->kn_hookid = 0; return (0); } static void filt_userdetach(__unused struct knote *kn) { /* * EVFILT_USER knotes are not attached to anything in the kernel. */ } static int filt_user(struct knote *kn, __unused long hint) { return (kn->kn_hookid); } static void filt_usertouch(struct knote *kn, struct kevent *kev, u_long type) { u_int ffctrl; switch (type) { case EVENT_REGISTER: if (kev->fflags & NOTE_TRIGGER) kn->kn_hookid = 1; ffctrl = kev->fflags & NOTE_FFCTRLMASK; kev->fflags &= NOTE_FFLAGSMASK; switch (ffctrl) { case NOTE_FFNOP: break; case NOTE_FFAND: kn->kn_sfflags &= kev->fflags; break; case NOTE_FFOR: kn->kn_sfflags |= kev->fflags; break; case NOTE_FFCOPY: kn->kn_sfflags = kev->fflags; break; default: /* XXX Return error? */ break; } kn->kn_sdata = kev->data; if (kev->flags & EV_CLEAR) { kn->kn_hookid = 0; kn->kn_data = 0; kn->kn_fflags = 0; } break; case EVENT_PROCESS: *kev = kn->kn_kevent; kev->fflags = kn->kn_sfflags; kev->data = kn->kn_sdata; if (kn->kn_flags & EV_CLEAR) { kn->kn_hookid = 0; kn->kn_data = 0; kn->kn_fflags = 0; } break; default: panic("filt_usertouch() - invalid type (%ld)", type); break; } } int sys_kqueue(struct thread *td, struct kqueue_args *uap) { return (kern_kqueue(td, 0, NULL)); } int sys_kqueuex(struct thread *td, struct kqueuex_args *uap) { int flags; if ((uap->flags & ~(KQUEUE_CLOEXEC)) != 0) return (EINVAL); flags = 0; if ((uap->flags & KQUEUE_CLOEXEC) != 0) flags |= O_CLOEXEC; return (kern_kqueue(td, flags, NULL)); } static void kqueue_init(struct kqueue *kq) { mtx_init(&kq->kq_lock, "kqueue", NULL, MTX_DEF | MTX_DUPOK); TAILQ_INIT(&kq->kq_head); knlist_init_mtx(&kq->kq_sel.si_note, &kq->kq_lock); TASK_INIT(&kq->kq_task, 0, kqueue_task, kq); } int kern_kqueue(struct thread *td, int flags, struct filecaps *fcaps) { struct filedesc *fdp; struct kqueue *kq; struct file *fp; struct ucred *cred; int fd, error; fdp = td->td_proc->p_fd; cred = td->td_ucred; if (!chgkqcnt(cred->cr_ruidinfo, 1, lim_cur(td, RLIMIT_KQUEUES))) return (ENOMEM); error = falloc_caps(td, &fp, &fd, flags, fcaps); if (error != 0) { chgkqcnt(cred->cr_ruidinfo, -1, 0); return (error); } /* An extra reference on `fp' has been held for us by falloc(). */ kq = malloc(sizeof *kq, M_KQUEUE, M_WAITOK | M_ZERO); kqueue_init(kq); kq->kq_fdp = fdp; kq->kq_cred = crhold(cred); FILEDESC_XLOCK(fdp); TAILQ_INSERT_HEAD(&fdp->fd_kqlist, kq, kq_list); FILEDESC_XUNLOCK(fdp); finit(fp, FREAD | FWRITE, DTYPE_KQUEUE, kq, &kqueueops); fdrop(fp, td); td->td_retval[0] = fd; return (0); } struct g_kevent_args { int fd; const void *changelist; int nchanges; void *eventlist; int nevents; const struct timespec *timeout; }; int sys_kevent(struct thread *td, struct kevent_args *uap) { struct kevent_copyops k_ops = { .arg = uap, .k_copyout = kevent_copyout, .k_copyin = kevent_copyin, .kevent_size = sizeof(struct kevent), }; struct g_kevent_args gk_args = { .fd = uap->fd, .changelist = uap->changelist, .nchanges = uap->nchanges, .eventlist = uap->eventlist, .nevents = uap->nevents, .timeout = uap->timeout, }; return (kern_kevent_generic(td, &gk_args, &k_ops, "kevent")); } static int kern_kevent_generic(struct thread *td, struct g_kevent_args *uap, struct kevent_copyops *k_ops, const char *struct_name) { struct timespec ts, *tsp; #ifdef KTRACE struct kevent *eventlist = uap->eventlist; #endif int error; if (uap->timeout != NULL) { error = copyin(uap->timeout, &ts, sizeof(ts)); if (error) return (error); tsp = &ts; } else tsp = NULL; #ifdef KTRACE if (KTRPOINT(td, KTR_STRUCT_ARRAY)) ktrstructarray(struct_name, UIO_USERSPACE, uap->changelist, uap->nchanges, k_ops->kevent_size); #endif error = kern_kevent(td, uap->fd, uap->nchanges, uap->nevents, k_ops, tsp); #ifdef KTRACE if (error == 0 && KTRPOINT(td, KTR_STRUCT_ARRAY)) ktrstructarray(struct_name, UIO_USERSPACE, eventlist, td->td_retval[0], k_ops->kevent_size); #endif return (error); } /* * Copy 'count' items into the destination list pointed to by uap->eventlist. */ static int kevent_copyout(void *arg, struct kevent *kevp, int count) { struct kevent_args *uap; int error; KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count)); uap = (struct kevent_args *)arg; error = copyout(kevp, uap->eventlist, count * sizeof *kevp); if (error == 0) uap->eventlist += count; return (error); } /* * Copy 'count' items from the list pointed to by uap->changelist. */ static int kevent_copyin(void *arg, struct kevent *kevp, int count) { struct kevent_args *uap; int error; KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count)); uap = (struct kevent_args *)arg; error = copyin(uap->changelist, kevp, count * sizeof *kevp); if (error == 0) uap->changelist += count; return (error); } #ifdef COMPAT_FREEBSD11 static int kevent11_copyout(void *arg, struct kevent *kevp, int count) { struct freebsd11_kevent_args *uap; struct freebsd11_kevent kev11; int error, i; KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count)); uap = (struct freebsd11_kevent_args *)arg; for (i = 0; i < count; i++) { kev11.ident = kevp->ident; kev11.filter = kevp->filter; kev11.flags = kevp->flags; kev11.fflags = kevp->fflags; kev11.data = kevp->data; kev11.udata = kevp->udata; error = copyout(&kev11, uap->eventlist, sizeof(kev11)); if (error != 0) break; uap->eventlist++; kevp++; } return (error); } /* * Copy 'count' items from the list pointed to by uap->changelist. */ static int kevent11_copyin(void *arg, struct kevent *kevp, int count) { struct freebsd11_kevent_args *uap; struct freebsd11_kevent kev11; int error, i; KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count)); uap = (struct freebsd11_kevent_args *)arg; for (i = 0; i < count; i++) { error = copyin(uap->changelist, &kev11, sizeof(kev11)); if (error != 0) break; kevp->ident = kev11.ident; kevp->filter = kev11.filter; kevp->flags = kev11.flags; kevp->fflags = kev11.fflags; kevp->data = (uintptr_t)kev11.data; kevp->udata = kev11.udata; bzero(&kevp->ext, sizeof(kevp->ext)); uap->changelist++; kevp++; } return (error); } int freebsd11_kevent(struct thread *td, struct freebsd11_kevent_args *uap) { struct kevent_copyops k_ops = { .arg = uap, .k_copyout = kevent11_copyout, .k_copyin = kevent11_copyin, .kevent_size = sizeof(struct freebsd11_kevent), }; struct g_kevent_args gk_args = { .fd = uap->fd, .changelist = uap->changelist, .nchanges = uap->nchanges, .eventlist = uap->eventlist, .nevents = uap->nevents, .timeout = uap->timeout, }; return (kern_kevent_generic(td, &gk_args, &k_ops, "freebsd11_kevent")); } #endif int kern_kevent(struct thread *td, int fd, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout) { cap_rights_t rights; struct file *fp; int error; cap_rights_init_zero(&rights); if (nchanges > 0) cap_rights_set_one(&rights, CAP_KQUEUE_CHANGE); if (nevents > 0) cap_rights_set_one(&rights, CAP_KQUEUE_EVENT); error = fget(td, fd, &rights, &fp); if (error != 0) return (error); error = kern_kevent_fp(td, fp, nchanges, nevents, k_ops, timeout); fdrop(fp, td); return (error); } static int kqueue_kevent(struct kqueue *kq, struct thread *td, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout) { struct kevent keva[KQ_NEVENTS]; struct kevent *kevp, *changes; int i, n, nerrors, error; if (nchanges < 0) return (EINVAL); nerrors = 0; while (nchanges > 0) { n = nchanges > KQ_NEVENTS ? KQ_NEVENTS : nchanges; error = k_ops->k_copyin(k_ops->arg, keva, n); if (error) return (error); changes = keva; for (i = 0; i < n; i++) { kevp = &changes[i]; if (!kevp->filter) continue; kevp->flags &= ~EV_SYSFLAGS; error = kqueue_register(kq, kevp, td, M_WAITOK); if (error || (kevp->flags & EV_RECEIPT)) { if (nevents == 0) return (error); kevp->flags = EV_ERROR; kevp->data = error; (void)k_ops->k_copyout(k_ops->arg, kevp, 1); nevents--; nerrors++; } } nchanges -= n; } if (nerrors) { td->td_retval[0] = nerrors; return (0); } return (kqueue_scan(kq, nevents, k_ops, timeout, keva, td)); } int kern_kevent_fp(struct thread *td, struct file *fp, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout) { struct kqueue *kq; int error; error = kqueue_acquire(fp, &kq); if (error != 0) return (error); error = kqueue_kevent(kq, td, nchanges, nevents, k_ops, timeout); kqueue_release(kq, 0); return (error); } /* * Performs a kevent() call on a temporarily created kqueue. This can be * used to perform one-shot polling, similar to poll() and select(). */ int kern_kevent_anonymous(struct thread *td, int nevents, struct kevent_copyops *k_ops) { struct kqueue kq = {}; int error; kqueue_init(&kq); kq.kq_refcnt = 1; error = kqueue_kevent(&kq, td, nevents, nevents, k_ops, NULL); kqueue_drain(&kq, td); kqueue_destroy(&kq); return (error); } int kqueue_add_filteropts(int filt, const struct filterops *filtops) { int error; error = 0; if (filt > 0 || filt + EVFILT_SYSCOUNT < 0) { printf( "trying to add a filterop that is out of range: %d is beyond %d\n", ~filt, EVFILT_SYSCOUNT); return EINVAL; } mtx_lock(&filterops_lock); if (sysfilt_ops[~filt].for_fop != &null_filtops && sysfilt_ops[~filt].for_fop != NULL) error = EEXIST; else { sysfilt_ops[~filt].for_fop = filtops; sysfilt_ops[~filt].for_refcnt = 0; } mtx_unlock(&filterops_lock); return (error); } int kqueue_del_filteropts(int filt) { int error; error = 0; if (filt > 0 || filt + EVFILT_SYSCOUNT < 0) return EINVAL; mtx_lock(&filterops_lock); if (sysfilt_ops[~filt].for_fop == &null_filtops || sysfilt_ops[~filt].for_fop == NULL) error = EINVAL; else if (sysfilt_ops[~filt].for_refcnt != 0) error = EBUSY; else { sysfilt_ops[~filt].for_fop = &null_filtops; sysfilt_ops[~filt].for_refcnt = 0; } mtx_unlock(&filterops_lock); return error; } static const struct filterops * kqueue_fo_find(int filt) { if (filt > 0 || filt + EVFILT_SYSCOUNT < 0) return NULL; if (sysfilt_ops[~filt].for_nolock) return sysfilt_ops[~filt].for_fop; mtx_lock(&filterops_lock); sysfilt_ops[~filt].for_refcnt++; if (sysfilt_ops[~filt].for_fop == NULL) sysfilt_ops[~filt].for_fop = &null_filtops; mtx_unlock(&filterops_lock); return sysfilt_ops[~filt].for_fop; } static void kqueue_fo_release(int filt) { if (filt > 0 || filt + EVFILT_SYSCOUNT < 0) return; if (sysfilt_ops[~filt].for_nolock) return; mtx_lock(&filterops_lock); KASSERT(sysfilt_ops[~filt].for_refcnt > 0, ("filter object refcount not valid on release")); sysfilt_ops[~filt].for_refcnt--; mtx_unlock(&filterops_lock); } /* * A ref to kq (obtained via kqueue_acquire) must be held. */ static int kqueue_register(struct kqueue *kq, struct kevent *kev, struct thread *td, int mflag) { const struct filterops *fops; struct file *fp; struct knote *kn, *tkn; struct knlist *knl; int error, filt, event; int haskqglobal, filedesc_unlock; if ((kev->flags & (EV_ENABLE | EV_DISABLE)) == (EV_ENABLE | EV_DISABLE)) return (EINVAL); fp = NULL; kn = NULL; knl = NULL; error = 0; haskqglobal = 0; filedesc_unlock = 0; filt = kev->filter; fops = kqueue_fo_find(filt); if (fops == NULL) return EINVAL; if (kev->flags & EV_ADD) { /* Reject an invalid flag pair early */ if (kev->flags & EV_KEEPUDATA) { tkn = NULL; error = EINVAL; goto done; } /* * Prevent waiting with locks. Non-sleepable * allocation failures are handled in the loop, only * if the spare knote appears to be actually required. */ tkn = knote_alloc(mflag); } else { tkn = NULL; } findkn: if (fops->f_isfd) { KASSERT(td != NULL, ("td is NULL")); if (kev->ident > INT_MAX) error = EBADF; else error = fget(td, kev->ident, &cap_event_rights, &fp); if (error) goto done; if ((kev->flags & EV_ADD) == EV_ADD && kqueue_expand(kq, fops, kev->ident, M_NOWAIT) != 0) { /* try again */ fdrop(fp, td); fp = NULL; error = kqueue_expand(kq, fops, kev->ident, mflag); if (error) goto done; goto findkn; } if (fp->f_type == DTYPE_KQUEUE) { /* * If we add some intelligence about what we are doing, * we should be able to support events on ourselves. * We need to know when we are doing this to prevent * getting both the knlist lock and the kq lock since * they are the same thing. */ if (fp->f_data == kq) { error = EINVAL; goto done; } /* * Pre-lock the filedesc before the global * lock mutex, see the comment in * kqueue_close(). */ FILEDESC_XLOCK(td->td_proc->p_fd); filedesc_unlock = 1; KQ_GLOBAL_LOCK(&kq_global, haskqglobal); } KQ_LOCK(kq); if (kev->ident < kq->kq_knlistsize) { SLIST_FOREACH(kn, &kq->kq_knlist[kev->ident], kn_link) if (kev->filter == kn->kn_filter) break; } } else { if ((kev->flags & EV_ADD) == EV_ADD) { error = kqueue_expand(kq, fops, kev->ident, mflag); if (error != 0) goto done; } KQ_LOCK(kq); /* * If possible, find an existing knote to use for this kevent. */ if (kev->filter == EVFILT_PROC && (kev->flags & (EV_FLAG1 | EV_FLAG2)) != 0) { /* This is an internal creation of a process tracking * note. Don't attempt to coalesce this with an * existing note. */ ; } else if (kq->kq_knhashmask != 0) { struct klist *list; list = &kq->kq_knhash[ KN_HASH((u_long)kev->ident, kq->kq_knhashmask)]; SLIST_FOREACH(kn, list, kn_link) if (kev->ident == kn->kn_id && kev->filter == kn->kn_filter) break; } } /* knote is in the process of changing, wait for it to stabilize. */ if (kn != NULL && kn_in_flux(kn)) { KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal); if (filedesc_unlock) { FILEDESC_XUNLOCK(td->td_proc->p_fd); filedesc_unlock = 0; } kq->kq_state |= KQ_FLUXWAIT; msleep(kq, &kq->kq_lock, PSOCK | PDROP, "kqflxwt", 0); if (fp != NULL) { fdrop(fp, td); fp = NULL; } goto findkn; } /* * kn now contains the matching knote, or NULL if no match */ if (kn == NULL) { if (kev->flags & EV_ADD) { kn = tkn; tkn = NULL; if (kn == NULL) { KQ_UNLOCK(kq); error = ENOMEM; goto done; } /* * Now that the kqueue is locked, make sure the fd * didn't change out from under us. */ if (fops->f_isfd && fget_noref_unlocked(td->td_proc->p_fd, kev->ident) != fp) { KQ_UNLOCK(kq); tkn = kn; error = EBADF; goto done; } kn->kn_fp = fp; kn->kn_kq = kq; kn->kn_fop = fops; kn->kn_sfflags = kev->fflags; kn->kn_sdata = kev->data; kev->fflags = 0; kev->data = 0; kn->kn_kevent = *kev; kn->kn_kevent.flags &= ~(EV_ADD | EV_DELETE | EV_ENABLE | EV_DISABLE | EV_FORCEONESHOT); kn->kn_status = KN_DETACHED; if ((kev->flags & EV_DISABLE) != 0) kn->kn_status |= KN_DISABLED; kn_enter_flux(kn); error = knote_attach(kn, kq); KQ_UNLOCK(kq); if (error != 0) { tkn = kn; goto done; } /* * We transfer ownership of fops/fp to the knote * structure and avoid releasing them at the end of * this routine, now that all of the remaining exit * paths will knote_drop() to release the reference * counts we held on them above. */ fops = NULL; fp = NULL; if ((error = kn->kn_fop->f_attach(kn)) != 0) { knote_drop_detached(kn, td); goto done; } knl = kn_list_lock(kn); goto done_ev_add; } else { /* No matching knote and the EV_ADD flag is not set. */ KQ_UNLOCK(kq); error = ENOENT; goto done; } } if (kev->flags & EV_DELETE) { kn_enter_flux(kn); KQ_UNLOCK(kq); knote_drop(kn, td); goto done; } if (kev->flags & EV_FORCEONESHOT) { kn->kn_flags |= EV_ONESHOT; KNOTE_ACTIVATE(kn, 1); } if ((kev->flags & EV_ENABLE) != 0) kn->kn_status &= ~KN_DISABLED; else if ((kev->flags & EV_DISABLE) != 0) kn->kn_status |= KN_DISABLED; /* * The user may change some filter values after the initial EV_ADD, * but doing so will not reset any filter which has already been * triggered. */ kn->kn_status |= KN_SCAN; kn_enter_flux(kn); KQ_UNLOCK(kq); knl = kn_list_lock(kn); if ((kev->flags & EV_KEEPUDATA) == 0) kn->kn_kevent.udata = kev->udata; if (!fops->f_isfd && fops->f_touch != NULL) { fops->f_touch(kn, kev, EVENT_REGISTER); } else { kn->kn_sfflags = kev->fflags; kn->kn_sdata = kev->data; } done_ev_add: /* * We can get here with kn->kn_knlist == NULL. This can happen when * the initial attach event decides that the event is "completed" * already, e.g., filt_procattach() is called on a zombie process. It * will call filt_proc() which will remove it from the list, and NULL * kn_knlist. * * KN_DISABLED will be stable while the knote is in flux, so the * unlocked read will not race with an update. */ if ((kn->kn_status & KN_DISABLED) == 0) event = kn->kn_fop->f_event(kn, 0); else event = 0; KQ_LOCK(kq); if (event) kn->kn_status |= KN_ACTIVE; if ((kn->kn_status & (KN_ACTIVE | KN_DISABLED | KN_QUEUED)) == KN_ACTIVE) knote_enqueue(kn); kn->kn_status &= ~KN_SCAN; kn_leave_flux(kn); kn_list_unlock(knl); KQ_UNLOCK_FLUX(kq); done: KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal); if (filedesc_unlock) FILEDESC_XUNLOCK(td->td_proc->p_fd); if (fp != NULL) fdrop(fp, td); knote_free(tkn); if (fops != NULL) kqueue_fo_release(filt); return (error); } static int kqueue_acquire(struct file *fp, struct kqueue **kqp) { int error; struct kqueue *kq; error = 0; kq = fp->f_data; if (fp->f_type != DTYPE_KQUEUE || kq == NULL) return (EINVAL); *kqp = kq; KQ_LOCK(kq); if ((kq->kq_state & KQ_CLOSING) == KQ_CLOSING) { KQ_UNLOCK(kq); return (EBADF); } kq->kq_refcnt++; KQ_UNLOCK(kq); return error; } static void kqueue_release(struct kqueue *kq, int locked) { if (locked) KQ_OWNED(kq); else KQ_LOCK(kq); kq->kq_refcnt--; if (kq->kq_refcnt == 1) wakeup(&kq->kq_refcnt); if (!locked) KQ_UNLOCK(kq); } static void ast_kqueue(struct thread *td, int tda __unused) { taskqueue_quiesce(taskqueue_kqueue_ctx); } static void kqueue_schedtask(struct kqueue *kq) { KQ_OWNED(kq); KASSERT(((kq->kq_state & KQ_TASKDRAIN) != KQ_TASKDRAIN), ("scheduling kqueue task while draining")); if ((kq->kq_state & KQ_TASKSCHED) != KQ_TASKSCHED) { taskqueue_enqueue(taskqueue_kqueue_ctx, &kq->kq_task); kq->kq_state |= KQ_TASKSCHED; ast_sched(curthread, TDA_KQUEUE); } } /* * Expand the kq to make sure we have storage for fops/ident pair. * * Return 0 on success (or no work necessary), return errno on failure. */ static int kqueue_expand(struct kqueue *kq, const struct filterops *fops, uintptr_t ident, int mflag) { struct klist *list, *tmp_knhash, *to_free; u_long tmp_knhashmask; int error, fd, size; KQ_NOTOWNED(kq); error = 0; to_free = NULL; if (fops->f_isfd) { fd = ident; size = atomic_load_int(&kq->kq_knlistsize); if (size <= fd) { do { size += KQEXTENT; } while (size <= fd); list = malloc(size * sizeof(*list), M_KQUEUE, mflag); if (list == NULL) return ENOMEM; KQ_LOCK(kq); if ((kq->kq_state & KQ_CLOSING) != 0) { to_free = list; error = EBADF; } else if (kq->kq_knlistsize >= size) { to_free = list; } else { if (kq->kq_knlist != NULL) { bcopy(kq->kq_knlist, list, kq->kq_knlistsize * sizeof(*list)); to_free = kq->kq_knlist; kq->kq_knlist = NULL; } bzero((caddr_t)list + kq->kq_knlistsize * sizeof(*list), (size - kq->kq_knlistsize) * sizeof(*list)); kq->kq_knlistsize = size; kq->kq_knlist = list; } MPASS(error != 0 || kq->kq_knlistsize > fd); KQ_UNLOCK(kq); } } else { if (kq->kq_knhashmask == 0) { tmp_knhash = hashinit_flags(KN_HASHSIZE, M_KQUEUE, &tmp_knhashmask, (mflag & M_WAITOK) != 0 ? HASH_WAITOK : HASH_NOWAIT); if (tmp_knhash == NULL) return (ENOMEM); KQ_LOCK(kq); if ((kq->kq_state & KQ_CLOSING) != 0) { to_free = tmp_knhash; error = EBADF; } else if (kq->kq_knhashmask == 0) { kq->kq_knhash = tmp_knhash; kq->kq_knhashmask = tmp_knhashmask; } else { to_free = tmp_knhash; } KQ_UNLOCK(kq); } } free(to_free, M_KQUEUE); KQ_NOTOWNED(kq); return (error); } static void kqueue_task(void *arg, int pending) { struct kqueue *kq; int haskqglobal; haskqglobal = 0; kq = arg; KQ_GLOBAL_LOCK(&kq_global, haskqglobal); KQ_LOCK(kq); KNOTE_LOCKED(&kq->kq_sel.si_note, 0); kq->kq_state &= ~KQ_TASKSCHED; if ((kq->kq_state & KQ_TASKDRAIN) == KQ_TASKDRAIN) { wakeup(&kq->kq_state); } KQ_UNLOCK(kq); KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal); } /* * Scan, update kn_data (if not ONESHOT), and copyout triggered events. * We treat KN_MARKER knotes as if they are in flux. */ static int kqueue_scan(struct kqueue *kq, int maxevents, struct kevent_copyops *k_ops, const struct timespec *tsp, struct kevent *keva, struct thread *td) { struct kevent *kevp; struct knote *kn, *marker; struct knlist *knl; sbintime_t asbt, rsbt; int count, error, haskqglobal, influx, nkev, touch; count = maxevents; nkev = 0; error = 0; haskqglobal = 0; if (maxevents == 0) goto done_nl; if (maxevents < 0) { error = EINVAL; goto done_nl; } rsbt = 0; if (tsp != NULL) { if (!timespecvalid_interval(tsp)) { error = EINVAL; goto done_nl; } if (timespecisset(tsp)) { if (tsp->tv_sec <= INT32_MAX) { rsbt = tstosbt(*tsp); if (TIMESEL(&asbt, rsbt)) asbt += tc_tick_sbt; if (asbt <= SBT_MAX - rsbt) asbt += rsbt; else asbt = 0; rsbt >>= tc_precexp; } else asbt = 0; } else asbt = -1; } else asbt = 0; marker = knote_alloc(M_WAITOK); marker->kn_status = KN_MARKER; KQ_LOCK(kq); retry: kevp = keva; if (kq->kq_count == 0) { if (asbt == -1) { error = EWOULDBLOCK; } else { kq->kq_state |= KQ_SLEEP; error = msleep_sbt(kq, &kq->kq_lock, PSOCK | PCATCH, "kqread", asbt, rsbt, C_ABSOLUTE); } if (error == 0) goto retry; /* don't restart after signals... */ if (error == ERESTART) error = EINTR; else if (error == EWOULDBLOCK) error = 0; goto done; } TAILQ_INSERT_TAIL(&kq->kq_head, marker, kn_tqe); influx = 0; while (count) { KQ_OWNED(kq); kn = TAILQ_FIRST(&kq->kq_head); if ((kn->kn_status == KN_MARKER && kn != marker) || kn_in_flux(kn)) { if (influx) { influx = 0; KQ_FLUX_WAKEUP(kq); } kq->kq_state |= KQ_FLUXWAIT; error = msleep(kq, &kq->kq_lock, PSOCK, "kqflxwt", 0); continue; } TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe); if ((kn->kn_status & KN_DISABLED) == KN_DISABLED) { kn->kn_status &= ~KN_QUEUED; kq->kq_count--; continue; } if (kn == marker) { KQ_FLUX_WAKEUP(kq); if (count == maxevents) goto retry; goto done; } KASSERT(!kn_in_flux(kn), ("knote %p is unexpectedly in flux", kn)); if ((kn->kn_flags & EV_DROP) == EV_DROP) { kn->kn_status &= ~KN_QUEUED; kn_enter_flux(kn); kq->kq_count--; KQ_UNLOCK(kq); /* * We don't need to lock the list since we've * marked it as in flux. */ knote_drop(kn, td); KQ_LOCK(kq); continue; } else if ((kn->kn_flags & EV_ONESHOT) == EV_ONESHOT) { kn->kn_status &= ~KN_QUEUED; kn_enter_flux(kn); kq->kq_count--; KQ_UNLOCK(kq); /* * We don't need to lock the list since we've * marked the knote as being in flux. */ *kevp = kn->kn_kevent; knote_drop(kn, td); KQ_LOCK(kq); kn = NULL; } else { kn->kn_status |= KN_SCAN; kn_enter_flux(kn); KQ_UNLOCK(kq); if ((kn->kn_status & KN_KQUEUE) == KN_KQUEUE) KQ_GLOBAL_LOCK(&kq_global, haskqglobal); knl = kn_list_lock(kn); if (kn->kn_fop->f_event(kn, 0) == 0) { KQ_LOCK(kq); KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal); kn->kn_status &= ~(KN_QUEUED | KN_ACTIVE | KN_SCAN); kn_leave_flux(kn); kq->kq_count--; kn_list_unlock(knl); influx = 1; continue; } touch = (!kn->kn_fop->f_isfd && kn->kn_fop->f_touch != NULL); if (touch) kn->kn_fop->f_touch(kn, kevp, EVENT_PROCESS); else *kevp = kn->kn_kevent; KQ_LOCK(kq); KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal); if (kn->kn_flags & (EV_CLEAR | EV_DISPATCH)) { /* * Manually clear knotes who weren't * 'touch'ed. */ if (touch == 0 && kn->kn_flags & EV_CLEAR) { kn->kn_data = 0; kn->kn_fflags = 0; } if (kn->kn_flags & EV_DISPATCH) kn->kn_status |= KN_DISABLED; kn->kn_status &= ~(KN_QUEUED | KN_ACTIVE); kq->kq_count--; } else TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe); kn->kn_status &= ~KN_SCAN; kn_leave_flux(kn); kn_list_unlock(knl); influx = 1; } /* we are returning a copy to the user */ kevp++; nkev++; count--; if (nkev == KQ_NEVENTS) { influx = 0; KQ_UNLOCK_FLUX(kq); error = k_ops->k_copyout(k_ops->arg, keva, nkev); nkev = 0; kevp = keva; KQ_LOCK(kq); if (error) break; } } TAILQ_REMOVE(&kq->kq_head, marker, kn_tqe); done: KQ_OWNED(kq); KQ_UNLOCK_FLUX(kq); knote_free(marker); done_nl: KQ_NOTOWNED(kq); if (nkev != 0) error = k_ops->k_copyout(k_ops->arg, keva, nkev); td->td_retval[0] = maxevents - count; return (error); } /*ARGSUSED*/ static int kqueue_ioctl(struct file *fp, u_long cmd, void *data, struct ucred *active_cred, struct thread *td) { /* * Enabling sigio causes two major problems: * 1) infinite recursion: * Synopsys: kevent is being used to track signals and have FIOASYNC * set. On receipt of a signal this will cause a kqueue to recurse * into itself over and over. Sending the sigio causes the kqueue * to become ready, which in turn posts sigio again, forever. * Solution: this can be solved by setting a flag in the kqueue that * we have a SIGIO in progress. * 2) locking problems: * Synopsys: Kqueue is a leaf subsystem, but adding signalling puts * us above the proc and pgrp locks. * Solution: Post a signal using an async mechanism, being sure to * record a generation count in the delivery so that we do not deliver * a signal to the wrong process. * * Note, these two mechanisms are somewhat mutually exclusive! */ #if 0 struct kqueue *kq; kq = fp->f_data; switch (cmd) { case FIOASYNC: if (*(int *)data) { kq->kq_state |= KQ_ASYNC; } else { kq->kq_state &= ~KQ_ASYNC; } return (0); case FIOSETOWN: return (fsetown(*(int *)data, &kq->kq_sigio)); case FIOGETOWN: *(int *)data = fgetown(&kq->kq_sigio); return (0); } #endif return (ENOTTY); } /*ARGSUSED*/ static int kqueue_poll(struct file *fp, int events, struct ucred *active_cred, struct thread *td) { struct kqueue *kq; int revents = 0; int error; if ((error = kqueue_acquire(fp, &kq))) return POLLERR; KQ_LOCK(kq); if (events & (POLLIN | POLLRDNORM)) { if (kq->kq_count) { revents |= events & (POLLIN | POLLRDNORM); } else { selrecord(td, &kq->kq_sel); if (SEL_WAITING(&kq->kq_sel)) kq->kq_state |= KQ_SEL; } } kqueue_release(kq, 1); KQ_UNLOCK(kq); return (revents); } /*ARGSUSED*/ static int kqueue_stat(struct file *fp, struct stat *st, struct ucred *active_cred) { bzero((void *)st, sizeof *st); /* * We no longer return kq_count because the unlocked value is useless. * If you spent all this time getting the count, why not spend your * syscall better by calling kevent? * * XXX - This is needed for libc_r. */ st->st_mode = S_IFIFO; return (0); } static void kqueue_drain(struct kqueue *kq, struct thread *td) { struct knote *kn; int i; KQ_LOCK(kq); KASSERT((kq->kq_state & KQ_CLOSING) != KQ_CLOSING, ("kqueue already closing")); kq->kq_state |= KQ_CLOSING; if (kq->kq_refcnt > 1) msleep(&kq->kq_refcnt, &kq->kq_lock, PSOCK, "kqclose", 0); KASSERT(kq->kq_refcnt == 1, ("other refs are out there!")); KASSERT(knlist_empty(&kq->kq_sel.si_note), ("kqueue's knlist not empty")); for (i = 0; i < kq->kq_knlistsize; i++) { while ((kn = SLIST_FIRST(&kq->kq_knlist[i])) != NULL) { if (kn_in_flux(kn)) { kq->kq_state |= KQ_FLUXWAIT; msleep(kq, &kq->kq_lock, PSOCK, "kqclo1", 0); continue; } kn_enter_flux(kn); KQ_UNLOCK(kq); knote_drop(kn, td); KQ_LOCK(kq); } } if (kq->kq_knhashmask != 0) { for (i = 0; i <= kq->kq_knhashmask; i++) { while ((kn = SLIST_FIRST(&kq->kq_knhash[i])) != NULL) { if (kn_in_flux(kn)) { kq->kq_state |= KQ_FLUXWAIT; msleep(kq, &kq->kq_lock, PSOCK, "kqclo2", 0); continue; } kn_enter_flux(kn); KQ_UNLOCK(kq); knote_drop(kn, td); KQ_LOCK(kq); } } } if ((kq->kq_state & KQ_TASKSCHED) == KQ_TASKSCHED) { kq->kq_state |= KQ_TASKDRAIN; msleep(&kq->kq_state, &kq->kq_lock, PSOCK, "kqtqdr", 0); } if ((kq->kq_state & KQ_SEL) == KQ_SEL) { selwakeuppri(&kq->kq_sel, PSOCK); if (!SEL_WAITING(&kq->kq_sel)) kq->kq_state &= ~KQ_SEL; } KQ_UNLOCK(kq); } static void kqueue_destroy(struct kqueue *kq) { KASSERT(kq->kq_fdp == NULL, ("kqueue still attached to a file descriptor")); seldrain(&kq->kq_sel); knlist_destroy(&kq->kq_sel.si_note); mtx_destroy(&kq->kq_lock); if (kq->kq_knhash != NULL) free(kq->kq_knhash, M_KQUEUE); if (kq->kq_knlist != NULL) free(kq->kq_knlist, M_KQUEUE); funsetown(&kq->kq_sigio); } /*ARGSUSED*/ static int kqueue_close(struct file *fp, struct thread *td) { struct kqueue *kq = fp->f_data; struct filedesc *fdp; int error; int filedesc_unlock; if ((error = kqueue_acquire(fp, &kq))) return error; kqueue_drain(kq, td); /* * We could be called due to the knote_drop() doing fdrop(), * called from kqueue_register(). In this case the global * lock is owned, and filedesc sx is locked before, to not * take the sleepable lock after non-sleepable. */ fdp = kq->kq_fdp; kq->kq_fdp = NULL; if (!sx_xlocked(FILEDESC_LOCK(fdp))) { FILEDESC_XLOCK(fdp); filedesc_unlock = 1; } else filedesc_unlock = 0; TAILQ_REMOVE(&fdp->fd_kqlist, kq, kq_list); if (filedesc_unlock) FILEDESC_XUNLOCK(fdp); kqueue_destroy(kq); chgkqcnt(kq->kq_cred->cr_ruidinfo, -1, 0); crfree(kq->kq_cred); free(kq, M_KQUEUE); fp->f_data = NULL; return (0); } static int kqueue_fill_kinfo(struct file *fp, struct kinfo_file *kif, struct filedesc *fdp) { struct kqueue *kq = fp->f_data; kif->kf_type = KF_TYPE_KQUEUE; kif->kf_un.kf_kqueue.kf_kqueue_addr = (uintptr_t)kq; kif->kf_un.kf_kqueue.kf_kqueue_count = kq->kq_count; kif->kf_un.kf_kqueue.kf_kqueue_state = kq->kq_state; return (0); } static void kqueue_wakeup(struct kqueue *kq) { KQ_OWNED(kq); if ((kq->kq_state & KQ_SLEEP) == KQ_SLEEP) { kq->kq_state &= ~KQ_SLEEP; wakeup(kq); } if ((kq->kq_state & KQ_SEL) == KQ_SEL) { selwakeuppri(&kq->kq_sel, PSOCK); if (!SEL_WAITING(&kq->kq_sel)) kq->kq_state &= ~KQ_SEL; } if (!knlist_empty(&kq->kq_sel.si_note)) kqueue_schedtask(kq); if ((kq->kq_state & KQ_ASYNC) == KQ_ASYNC) { pgsigio(&kq->kq_sigio, SIGIO, 0); } } /* * Walk down a list of knotes, activating them if their event has triggered. * * There is a possibility to optimize in the case of one kq watching another. * Instead of scheduling a task to wake it up, you could pass enough state * down the chain to make up the parent kqueue. Make this code functional * first. */ void knote(struct knlist *list, long hint, int lockflags) { struct kqueue *kq; struct knote *kn, *tkn; int error; if (list == NULL) return; KNL_ASSERT_LOCK(list, lockflags & KNF_LISTLOCKED); if ((lockflags & KNF_LISTLOCKED) == 0) list->kl_lock(list->kl_lockarg); /* * If we unlock the list lock (and enter influx), we can * eliminate the kqueue scheduling, but this will introduce * four lock/unlock's for each knote to test. Also, marker * would be needed to keep iteration position, since filters * or other threads could remove events. */ SLIST_FOREACH_SAFE(kn, &list->kl_list, kn_selnext, tkn) { kq = kn->kn_kq; KQ_LOCK(kq); if (kn_in_flux(kn) && (kn->kn_status & KN_SCAN) == 0) { /* * Do not process the influx notes, except for * the influx coming from the kq unlock in the * kqueue_scan(). In the later case, we do * not interfere with the scan, since the code * fragment in kqueue_scan() locks the knlist, * and cannot proceed until we finished. */ KQ_UNLOCK(kq); } else if ((lockflags & KNF_NOKQLOCK) != 0) { kn_enter_flux(kn); KQ_UNLOCK(kq); error = kn->kn_fop->f_event(kn, hint); KQ_LOCK(kq); kn_leave_flux(kn); if (error) KNOTE_ACTIVATE(kn, 1); KQ_UNLOCK_FLUX(kq); } else { if (kn->kn_fop->f_event(kn, hint)) KNOTE_ACTIVATE(kn, 1); KQ_UNLOCK(kq); } } if ((lockflags & KNF_LISTLOCKED) == 0) list->kl_unlock(list->kl_lockarg); } /* * add a knote to a knlist */ void knlist_add(struct knlist *knl, struct knote *kn, int islocked) { KNL_ASSERT_LOCK(knl, islocked); KQ_NOTOWNED(kn->kn_kq); KASSERT(kn_in_flux(kn), ("knote %p not in flux", kn)); KASSERT((kn->kn_status & KN_DETACHED) != 0, ("knote %p was not detached", kn)); KASSERT(kn->kn_knlist == NULL, ("knote %p was already on knlist %p", kn, kn->kn_knlist)); if (!islocked) knl->kl_lock(knl->kl_lockarg); SLIST_INSERT_HEAD(&knl->kl_list, kn, kn_selnext); if (!islocked) knl->kl_unlock(knl->kl_lockarg); KQ_LOCK(kn->kn_kq); kn->kn_knlist = knl; kn->kn_status &= ~KN_DETACHED; KQ_UNLOCK(kn->kn_kq); } static void knlist_remove_kq(struct knlist *knl, struct knote *kn, int knlislocked, int kqislocked) { KASSERT(!kqislocked || knlislocked, ("kq locked w/o knl locked")); KNL_ASSERT_LOCK(knl, knlislocked); mtx_assert(&kn->kn_kq->kq_lock, kqislocked ? MA_OWNED : MA_NOTOWNED); KASSERT(kqislocked || kn_in_flux(kn), ("knote %p not in flux", kn)); KASSERT((kn->kn_status & KN_DETACHED) == 0, ("knote %p was already detached", kn)); KASSERT(kn->kn_knlist == knl, ("knote %p was not on knlist %p", kn, knl)); if (!knlislocked) knl->kl_lock(knl->kl_lockarg); SLIST_REMOVE(&knl->kl_list, kn, knote, kn_selnext); kn->kn_knlist = NULL; if (!knlislocked) kn_list_unlock(knl); if (!kqislocked) KQ_LOCK(kn->kn_kq); kn->kn_status |= KN_DETACHED; if (!kqislocked) KQ_UNLOCK(kn->kn_kq); } /* * remove knote from the specified knlist */ void knlist_remove(struct knlist *knl, struct knote *kn, int islocked) { knlist_remove_kq(knl, kn, islocked, 0); } int knlist_empty(struct knlist *knl) { KNL_ASSERT_LOCKED(knl); return (SLIST_EMPTY(&knl->kl_list)); } static struct mtx knlist_lock; MTX_SYSINIT(knlist_lock, &knlist_lock, "knlist lock for lockless objects", MTX_DEF); static void knlist_mtx_lock(void *arg); static void knlist_mtx_unlock(void *arg); static void knlist_mtx_lock(void *arg) { mtx_lock((struct mtx *)arg); } static void knlist_mtx_unlock(void *arg) { mtx_unlock((struct mtx *)arg); } static void knlist_mtx_assert_lock(void *arg, int what) { if (what == LA_LOCKED) mtx_assert((struct mtx *)arg, MA_OWNED); else mtx_assert((struct mtx *)arg, MA_NOTOWNED); } void knlist_init(struct knlist *knl, void *lock, void (*kl_lock)(void *), void (*kl_unlock)(void *), void (*kl_assert_lock)(void *, int)) { if (lock == NULL) knl->kl_lockarg = &knlist_lock; else knl->kl_lockarg = lock; if (kl_lock == NULL) knl->kl_lock = knlist_mtx_lock; else knl->kl_lock = kl_lock; if (kl_unlock == NULL) knl->kl_unlock = knlist_mtx_unlock; else knl->kl_unlock = kl_unlock; if (kl_assert_lock == NULL) knl->kl_assert_lock = knlist_mtx_assert_lock; else knl->kl_assert_lock = kl_assert_lock; knl->kl_autodestroy = 0; SLIST_INIT(&knl->kl_list); } void knlist_init_mtx(struct knlist *knl, struct mtx *lock) { knlist_init(knl, lock, NULL, NULL, NULL); } struct knlist * knlist_alloc(struct mtx *lock) { struct knlist *knl; knl = malloc(sizeof(struct knlist), M_KQUEUE, M_WAITOK); knlist_init_mtx(knl, lock); return (knl); } void knlist_destroy(struct knlist *knl) { KASSERT(KNLIST_EMPTY(knl), ("destroying knlist %p with knotes on it", knl)); } void knlist_detach(struct knlist *knl) { KNL_ASSERT_LOCKED(knl); knl->kl_autodestroy = 1; if (knlist_empty(knl)) { knlist_destroy(knl); free(knl, M_KQUEUE); } } /* * Even if we are locked, we may need to drop the lock to allow any influx * knotes time to "settle". */ void knlist_cleardel(struct knlist *knl, struct thread *td, int islocked, int killkn) { struct knote *kn, *kn2; struct kqueue *kq; KASSERT(!knl->kl_autodestroy, ("cleardel for autodestroy %p", knl)); if (islocked) KNL_ASSERT_LOCKED(knl); else { KNL_ASSERT_UNLOCKED(knl); knl->kl_lock(knl->kl_lockarg); } for (;;) { /* * Each pass removes as many knotes as we can before dropping * into FLUXWAIT. Active knotes are simply detached and either * freed or converted to one-shot, as the attached subject is * essentially disappearing. */ SLIST_FOREACH_SAFE(kn, &knl->kl_list, kn_selnext, kn2) { kq = kn->kn_kq; KQ_LOCK(kq); if (kn_in_flux(kn)) { KQ_UNLOCK(kq); continue; } knlist_remove_kq(knl, kn, 1, 1); if (killkn) { kn_enter_flux(kn); KQ_UNLOCK(kq); knote_drop_detached(kn, td); } else { /* Make sure cleared knotes disappear soon */ kn->kn_flags |= EV_EOF | EV_ONESHOT; KQ_UNLOCK(kq); } kq = NULL; } if (SLIST_EMPTY(&knl->kl_list)) break; /* there are still in flux knotes remaining */ kn = SLIST_FIRST(&knl->kl_list); kq = kn->kn_kq; KQ_LOCK(kq); KASSERT(kn_in_flux(kn), ("knote removed w/o list lock")); knl->kl_unlock(knl->kl_lockarg); kq->kq_state |= KQ_FLUXWAIT; msleep(kq, &kq->kq_lock, PSOCK | PDROP, "kqkclr", 0); kq = NULL; knl->kl_lock(knl->kl_lockarg); } if (islocked) KNL_ASSERT_LOCKED(knl); else { knl->kl_unlock(knl->kl_lockarg); KNL_ASSERT_UNLOCKED(knl); } } /* * Remove all knotes referencing a specified fd must be called with FILEDESC * lock. This prevents a race where a new fd comes along and occupies the * entry and we attach a knote to the fd. */ void knote_fdclose(struct thread *td, int fd) { struct filedesc *fdp = td->td_proc->p_fd; struct kqueue *kq; struct knote *kn; FILEDESC_XLOCK_ASSERT(fdp); /* * We shouldn't have to worry about new kevents appearing on fd * since filedesc is locked. */ TAILQ_FOREACH(kq, &fdp->fd_kqlist, kq_list) { KQ_LOCK(kq); if (kq->kq_knlistsize <= fd || SLIST_EMPTY(&kq->kq_knlist[fd])) { KQ_UNLOCK(kq); continue; } while ((kn = SLIST_FIRST(&kq->kq_knlist[fd])) != NULL) { if (kn_in_flux(kn)) { /* * Wait for this knote to stabilize, it could be * the case that it's in the process of being * dropped anyways. */ kq->kq_state |= KQ_FLUXWAIT; msleep(kq, &kq->kq_lock, PSOCK, "kqflxwt", 0); continue; } kn_enter_flux(kn); KQ_UNLOCK(kq); knote_drop(kn, td); KQ_LOCK(kq); } KQ_UNLOCK_FLUX(kq); } } static int knote_attach(struct knote *kn, struct kqueue *kq) { struct klist *list; KASSERT(kn_in_flux(kn), ("knote %p not marked influx", kn)); KQ_OWNED(kq); if ((kq->kq_state & KQ_CLOSING) != 0) return (EBADF); if (kn->kn_fop->f_isfd) { if (kn->kn_id >= kq->kq_knlistsize) return (ENOMEM); list = &kq->kq_knlist[kn->kn_id]; } else { if (kq->kq_knhash == NULL) return (ENOMEM); list = &kq->kq_knhash[KN_HASH(kn->kn_id, kq->kq_knhashmask)]; } SLIST_INSERT_HEAD(list, kn, kn_link); return (0); } static void knote_drop(struct knote *kn, struct thread *td) { if ((kn->kn_status & KN_DETACHED) == 0) kn->kn_fop->f_detach(kn); knote_drop_detached(kn, td); } static void knote_drop_detached(struct knote *kn, struct thread *td) { struct kqueue *kq; struct klist *list; kq = kn->kn_kq; KASSERT((kn->kn_status & KN_DETACHED) != 0, ("knote %p still attached", kn)); KQ_NOTOWNED(kq); KQ_LOCK(kq); for (;;) { KASSERT(kn->kn_influx >= 1, ("knote_drop called on %p with influx %d", kn, kn->kn_influx)); if (kn->kn_influx == 1) break; kq->kq_state |= KQ_FLUXWAIT; msleep(kq, &kq->kq_lock, PSOCK, "kqflxwt", 0); } if (kn->kn_fop->f_isfd) list = &kq->kq_knlist[kn->kn_id]; else list = &kq->kq_knhash[KN_HASH(kn->kn_id, kq->kq_knhashmask)]; if (!SLIST_EMPTY(list)) SLIST_REMOVE(list, kn, knote, kn_link); if (kn->kn_status & KN_QUEUED) knote_dequeue(kn); KQ_UNLOCK_FLUX(kq); if (kn->kn_fop->f_isfd) { fdrop(kn->kn_fp, td); kn->kn_fp = NULL; } kqueue_fo_release(kn->kn_kevent.filter); kn->kn_fop = NULL; knote_free(kn); } static void knote_enqueue(struct knote *kn) { struct kqueue *kq = kn->kn_kq; KQ_OWNED(kn->kn_kq); KASSERT((kn->kn_status & KN_QUEUED) == 0, ("knote already queued")); TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe); kn->kn_status |= KN_QUEUED; kq->kq_count++; kqueue_wakeup(kq); } static void knote_dequeue(struct knote *kn) { struct kqueue *kq = kn->kn_kq; KQ_OWNED(kn->kn_kq); KASSERT(kn->kn_status & KN_QUEUED, ("knote not queued")); TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe); kn->kn_status &= ~KN_QUEUED; kq->kq_count--; } static void knote_init(void *dummy __unused) { knote_zone = uma_zcreate("KNOTE", sizeof(struct knote), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); ast_register(TDA_KQUEUE, ASTR_ASTF_REQUIRED, 0, ast_kqueue); } SYSINIT(knote, SI_SUB_PSEUDO, SI_ORDER_ANY, knote_init, NULL); static struct knote * knote_alloc(int mflag) { return (uma_zalloc(knote_zone, mflag | M_ZERO)); } static void knote_free(struct knote *kn) { uma_zfree(knote_zone, kn); } /* * Register the kev w/ the kq specified by fd. */ int kqfd_register(int fd, struct kevent *kev, struct thread *td, int mflag) { struct kqueue *kq; struct file *fp; cap_rights_t rights; int error; error = fget(td, fd, cap_rights_init_one(&rights, CAP_KQUEUE_CHANGE), &fp); if (error != 0) return (error); if ((error = kqueue_acquire(fp, &kq)) != 0) goto noacquire; error = kqueue_register(kq, kev, td, mflag); kqueue_release(kq, 0); noacquire: fdrop(fp, td); return (error); } struct knote_status_export_bit { int kn_status_bit; int knt_status_bit; }; #define ST(name) \ { .kn_status_bit = KN_##name, .knt_status_bit = KNOTE_STATUS_##name } static const struct knote_status_export_bit knote_status_export_bits[] = { ST(ACTIVE), ST(QUEUED), ST(DISABLED), ST(DETACHED), ST(KQUEUE), }; #undef ST static int knote_status_export(int kn_status) { const struct knote_status_export_bit *b; unsigned i; int res; res = 0; for (i = 0; i < nitems(knote_status_export_bits); i++) { b = &knote_status_export_bits[i]; if ((kn_status & b->kn_status_bit) != 0) res |= b->knt_status_bit; } return (res); } static int kern_proc_kqueue_report_one(struct sbuf *s, struct proc *p, int kq_fd, struct kqueue *kq, struct knote *kn, bool compat32 __unused) { struct kinfo_knote kin; #ifdef COMPAT_FREEBSD32 struct kinfo_knote32 kin32; #endif int error; if (kn->kn_status == KN_MARKER) return (0); memset(&kin, 0, sizeof(kin)); kin.knt_kq_fd = kq_fd; memcpy(&kin.knt_event, &kn->kn_kevent, sizeof(struct kevent)); kin.knt_status = knote_status_export(kn->kn_status); kn_enter_flux(kn); KQ_UNLOCK_FLUX(kq); if (kn->kn_fop->f_userdump != NULL) (void)kn->kn_fop->f_userdump(p, kn, &kin); #ifdef COMPAT_FREEBSD32 if (compat32) { freebsd32_kinfo_knote_to_32(&kin, &kin32); error = sbuf_bcat(s, &kin32, sizeof(kin32)); } else #endif error = sbuf_bcat(s, &kin, sizeof(kin)); KQ_LOCK(kq); kn_leave_flux(kn); return (error); } static int kern_proc_kqueue_report(struct sbuf *s, struct proc *p, int kq_fd, struct kqueue *kq, bool compat32) { struct knote *kn; int error, i; error = 0; KQ_LOCK(kq); for (i = 0; i < kq->kq_knlistsize; i++) { SLIST_FOREACH(kn, &kq->kq_knlist[i], kn_link) { error = kern_proc_kqueue_report_one(s, p, kq_fd, kq, kn, compat32); if (error != 0) goto out; } } if (kq->kq_knhashmask == 0) goto out; for (i = 0; i <= kq->kq_knhashmask; i++) { SLIST_FOREACH(kn, &kq->kq_knhash[i], kn_link) { error = kern_proc_kqueue_report_one(s, p, kq_fd, kq, kn, compat32); if (error != 0) goto out; } } out: KQ_UNLOCK_FLUX(kq); return (error); } struct kern_proc_kqueues_out1_cb_args { struct sbuf *s; bool compat32; }; static int kern_proc_kqueues_out1_cb(struct proc *p, int fd, struct file *fp, void *arg) { struct kqueue *kq; struct kern_proc_kqueues_out1_cb_args *a; if (fp->f_type != DTYPE_KQUEUE) return (0); a = arg; kq = fp->f_data; return (kern_proc_kqueue_report(a->s, p, fd, kq, a->compat32)); } static int kern_proc_kqueues_out1(struct thread *td, struct proc *p, struct sbuf *s, bool compat32) { struct kern_proc_kqueues_out1_cb_args a; a.s = s; a.compat32 = compat32; return (fget_remote_foreach(td, p, kern_proc_kqueues_out1_cb, &a)); } int kern_proc_kqueues_out(struct proc *p, struct sbuf *sb, size_t maxlen, bool compat32) { struct sbuf *s, sm; size_t sb_len; int error; if (maxlen == -1 || maxlen == 0) sb_len = 128; else sb_len = maxlen; s = sbuf_new(&sm, NULL, sb_len, maxlen == -1 ? SBUF_AUTOEXTEND : SBUF_FIXEDLEN); error = kern_proc_kqueues_out1(curthread, p, s, compat32); sbuf_finish(s); if (error == 0) { sbuf_bcat(sb, sbuf_data(s), MIN(sbuf_len(s), maxlen == -1 ? SIZE_T_MAX : maxlen)); } sbuf_delete(s); return (error); } static int sysctl_kern_proc_kqueue_one(struct thread *td, struct sbuf *s, struct proc *p, int kq_fd, bool compat32) { struct file *fp; struct kqueue *kq; int error; error = fget_remote(td, p, kq_fd, &fp); if (error == 0) { if (fp->f_type != DTYPE_KQUEUE) { error = EINVAL; } else { kq = fp->f_data; error = kern_proc_kqueue_report(s, p, kq_fd, kq, compat32); } fdrop(fp, td); } return (error); } static int sysctl_kern_proc_kqueue(SYSCTL_HANDLER_ARGS) { struct thread *td; struct proc *p; struct sbuf *s, sm; int error, error1, *name; bool compat32; name = (int *)arg1; if ((u_int)arg2 > 2 || (u_int)arg2 == 0) return (EINVAL); td = curthread; #ifdef COMPAT_FREEBSD32 compat32 = SV_CURPROC_FLAG(SV_ILP32); #else compat32 = false; #endif error = pget((pid_t)name[0], PGET_NOTWEXIT, &p); if (error != 0) return (error); _PHOLD(p); execve_block_wait(td, p); error = p_candebug(td, p); if (error != 0) goto out1; PROC_UNLOCK(p); s = sbuf_new_for_sysctl(&sm, NULL, 0, req); if (s == NULL) { error = ENOMEM; goto out; } sbuf_clear_flags(s, SBUF_INCLUDENUL); if ((u_int)arg2 == 1) { error = kern_proc_kqueues_out1(td, p, s, compat32); } else { error = sysctl_kern_proc_kqueue_one(td, s, p, name[1] /* kq_fd */, compat32); } error1 = sbuf_finish(s); if (error == 0) error = error1; sbuf_delete(s); out: PROC_LOCK(p); out1: execve_unblock(td, p); _PRELE(p); PROC_UNLOCK(p); return (error); } static SYSCTL_NODE(_kern_proc, KERN_PROC_KQUEUE, kq, CTLFLAG_RD | CTLFLAG_MPSAFE, sysctl_kern_proc_kqueue, "KQueue events"); diff --git a/sys/kern/kern_exit.c b/sys/kern/kern_exit.c index baa5cad2bdc8..3cfb55fe3711 100644 --- a/sys/kern/kern_exit.c +++ b/sys/kern/kern_exit.c @@ -1,1469 +1,1470 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1982, 1986, 1989, 1991, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)kern_exit.c 8.7 (Berkeley) 2/12/94 */ #include #include "opt_ddb.h" #include "opt_ktrace.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* for acct_process() function prototype */ #include #include #include #include #include #include #include #ifdef KTRACE #include #endif #include #include #include #include #include #include #include #include #include #ifdef KDTRACE_HOOKS #include dtrace_execexit_func_t dtrace_fasttrap_exit; #endif SDT_PROVIDER_DECLARE(proc); SDT_PROBE_DEFINE1(proc, , , exit, "int"); static int kern_kill_on_dbg_exit = 1; SYSCTL_INT(_kern, OID_AUTO, kill_on_debugger_exit, CTLFLAG_RWTUN, &kern_kill_on_dbg_exit, 0, "Kill ptraced processes when debugger exits"); static bool kern_wait_dequeue_sigchld = 1; SYSCTL_BOOL(_kern, OID_AUTO, wait_dequeue_sigchld, CTLFLAG_RWTUN, &kern_wait_dequeue_sigchld, 0, "Dequeue SIGCHLD on wait(2) for live process"); struct proc * proc_realparent(struct proc *child) { struct proc *p, *parent; sx_assert(&proctree_lock, SX_LOCKED); if ((child->p_treeflag & P_TREE_ORPHANED) == 0) return (child->p_pptr->p_pid == child->p_oppid ? child->p_pptr : child->p_reaper); for (p = child; (p->p_treeflag & P_TREE_FIRST_ORPHAN) == 0;) { /* Cannot use LIST_PREV(), since the list head is not known. */ p = __containerof(p->p_orphan.le_prev, struct proc, p_orphan.le_next); KASSERT((p->p_treeflag & P_TREE_ORPHANED) != 0, ("missing P_ORPHAN %p", p)); } parent = __containerof(p->p_orphan.le_prev, struct proc, p_orphans.lh_first); return (parent); } static void reaper_clear(struct proc *p, struct proc *rp) { struct proc *p1; bool clear; sx_assert(&proctree_lock, SX_XLOCKED); LIST_REMOVE(p, p_reapsibling); if (p->p_reapsubtree == 1) return; clear = true; LIST_FOREACH(p1, &rp->p_reaplist, p_reapsibling) { if (p1->p_reapsubtree == p->p_reapsubtree) { clear = false; break; } } if (clear) proc_id_clear(PROC_ID_REAP, p->p_reapsubtree); } void reaper_abandon_children(struct proc *p, bool exiting) { struct proc *p1, *p2, *ptmp; sx_assert(&proctree_lock, SX_XLOCKED); KASSERT(p != initproc, ("reaper_abandon_children for initproc")); if ((p->p_treeflag & P_TREE_REAPER) == 0) return; p1 = p->p_reaper; LIST_FOREACH_SAFE(p2, &p->p_reaplist, p_reapsibling, ptmp) { reaper_clear(p2, p); p2->p_reaper = p1; p2->p_reapsubtree = p->p_reapsubtree; LIST_INSERT_HEAD(&p1->p_reaplist, p2, p_reapsibling); if (exiting && p2->p_pptr == p) { PROC_LOCK(p2); proc_reparent(p2, p1, true); PROC_UNLOCK(p2); } } KASSERT(LIST_EMPTY(&p->p_reaplist), ("p_reaplist not empty")); p->p_treeflag &= ~P_TREE_REAPER; } void proc_clear_orphan(struct proc *p) { struct proc *p1; sx_assert(&proctree_lock, SA_XLOCKED); if ((p->p_treeflag & P_TREE_ORPHANED) == 0) return; if ((p->p_treeflag & P_TREE_FIRST_ORPHAN) != 0) { p1 = LIST_NEXT(p, p_orphan); if (p1 != NULL) p1->p_treeflag |= P_TREE_FIRST_ORPHAN; p->p_treeflag &= ~P_TREE_FIRST_ORPHAN; } LIST_REMOVE(p, p_orphan); p->p_treeflag &= ~P_TREE_ORPHANED; } void exit_onexit(struct proc *p) { MPASS(p->p_numthreads == 1); umtx_thread_exit(FIRST_THREAD_IN_PROC(p)); } /* * exit -- death of process. */ int sys_exit(struct thread *td, struct exit_args *uap) { exit1(td, uap->rval, 0); __unreachable(); } void proc_set_p2_wexit(struct proc *p) { PROC_LOCK_ASSERT(p, MA_OWNED); p->p_flag2 |= P2_WEXIT; } /* * Exit: deallocate address space and other resources, change proc state to * zombie, and unlink proc from allproc and parent's lists. Save exit status * and rusage for wait(). Check for child processes and orphan them. */ void exit1(struct thread *td, int rval, int signo) { struct proc *p, *nq, *q, *t; struct thread *tdt; ksiginfo_t *ksi, *ksi1; int signal_parent; mtx_assert(&Giant, MA_NOTOWNED); KASSERT(rval == 0 || signo == 0, ("exit1 rv %d sig %d", rval, signo)); TSPROCEXIT(td->td_proc->p_pid); p = td->td_proc; /* * In case we're rebooting we just let init die in order to * work around an issues where pid 1 might get a fatal signal. * For instance, if network interface serving NFS root is * going down due to reboot, page-in requests for text are * failing. */ if (p == initproc && rebooting == 0) { printf("init died (signal %d, exit %d)\n", signo, rval); panic("Going nowhere without my init!"); } /* * Process deferred operations, designated with ASTF_KCLEAR. * For instance, we need to deref SU mp, since the thread does * not return to userspace, and wait for geom to stabilize. */ ast_kclear(td); /* * MUST abort all other threads before proceeding past here. */ PROC_LOCK(p); proc_set_p2_wexit(p); /* * First check if some other thread or external request got * here before us. If so, act appropriately: exit or suspend. * We must ensure that stop requests are handled before we set * P_WEXIT. */ thread_suspend_check(0); while (p->p_flag & P_HADTHREADS) { /* * Kill off the other threads. This requires * some co-operation from other parts of the kernel * so it may not be instantaneous. With this state set * any thread attempting to interruptibly * sleep will return immediately with EINTR or EWOULDBLOCK * which will hopefully force them to back out to userland * freeing resources as they go. Any thread attempting * to return to userland will thread_exit() from ast(). * thread_exit() will unsuspend us when the last of the * other threads exits. * If there is already a thread singler after resumption, * calling thread_single() will fail; in that case, we just * re-check all suspension request, the thread should * either be suspended there or exit. */ if (!thread_single(p, SINGLE_EXIT)) /* * All other activity in this process is now * stopped. Threading support has been turned * off. */ break; /* * Recheck for new stop or suspend requests which * might appear while process lock was dropped in * thread_single(). */ thread_suspend_check(0); } KASSERT(p->p_numthreads == 1, ("exit1: proc %p exiting with %d threads", p, p->p_numthreads)); racct_sub(p, RACCT_NTHR, 1); /* Let event handler change exit status */ p->p_xexit = rval; p->p_xsig = signo; /* * Ignore any pending request to stop due to a stop signal. * Once P_WEXIT is set, future requests will be ignored as * well. */ p->p_flag &= ~P_STOPPED_SIG; KASSERT(!P_SHOULDSTOP(p), ("exiting process is stopped")); /* Note that we are exiting. */ p->p_flag |= P_WEXIT; /* * Wait for any processes that have a hold on our vmspace to * release their reference. */ while (p->p_lock > 0) msleep(&p->p_lock, &p->p_mtx, PWAIT, "exithold", 0); MPASS(p->p_execblock == 0); PROC_UNLOCK(p); /* Drain the limit callout while we don't have the proc locked */ callout_drain(&p->p_limco); #ifdef AUDIT /* * The Sun BSM exit token contains two components: an exit status as * passed to exit(), and a return value to indicate what sort of exit * it was. The exit status is WEXITSTATUS(rv), but it's not clear * what the return value is. */ AUDIT_ARG_EXIT(rval, 0); AUDIT_SYSCALL_EXIT(0, td); #endif /* Are we a task leader with peers? */ if (p->p_peers != NULL && p == p->p_leader) { mtx_lock(&ppeers_lock); q = p->p_peers; while (q != NULL) { PROC_LOCK(q); kern_psignal(q, SIGKILL); PROC_UNLOCK(q); q = q->p_peers; } while (p->p_peers != NULL) msleep(p, &ppeers_lock, PWAIT, "exit1", 0); mtx_unlock(&ppeers_lock); } itimers_exit(p); /* * Check if any loadable modules need anything done at process exit. * E.g. SYSV IPC stuff. * Event handler could change exit status. * XXX what if one of these generates an error? */ EVENTHANDLER_DIRECT_INVOKE(process_exit, p); /* * If parent is waiting for us to exit or exec, * P_PPWAIT is set; we will wakeup the parent below. */ PROC_LOCK(p); stopprofclock(p); p->p_ptevents = 0; /* * Stop the real interval timer. If the handler is currently * executing, prevent it from rearming itself and let it finish. */ p->p_flag2 &= ~P2_ITSTOPPED; if (timevalisset(&p->p_realtimer.it_value) && callout_stop(&p->p_itcallout) == 0) { timevalclear(&p->p_realtimer.it_interval); PROC_UNLOCK(p); callout_drain(&p->p_itcallout); } else { PROC_UNLOCK(p); } if (p->p_sysent->sv_onexit != NULL) p->p_sysent->sv_onexit(p); seltdfini(td); /* * Reset any sigio structures pointing to us as a result of * F_SETOWN with our pid. The P_WEXIT flag interlocks with fsetown(). */ funsetownlst(&p->p_sigiolst); /* * Close open files and release open-file table. * This may block! */ pdescfree(td); fdescfree(td); /* * Remove ourself from our leader's peer list and wake our leader. */ if (p->p_leader->p_peers != NULL) { mtx_lock(&ppeers_lock); if (p->p_leader->p_peers != NULL) { q = p->p_leader; while (q->p_peers != p) q = q->p_peers; q->p_peers = p->p_peers; wakeup(p->p_leader); } mtx_unlock(&ppeers_lock); } exec_free_abi_mappings(p); vmspace_exit(td); (void)acct_process(td); #ifdef KTRACE ktrprocexit(td); #endif /* * Release reference to text vnode etc */ if (p->p_textvp != NULL) { vrele(p->p_textvp); p->p_textvp = NULL; } if (p->p_textdvp != NULL) { vrele(p->p_textdvp); p->p_textdvp = NULL; } if (p->p_binname != NULL) { free(p->p_binname, M_PARGS); p->p_binname = NULL; } /* * Release our limits structure. */ lim_free(p->p_limit); p->p_limit = NULL; tidhash_remove(td); /* * Call machine-dependent code to release any * machine-dependent resources other than the address space. * The address space is released by "vmspace_exitfree(p)" in * vm_waitproc(). */ cpu_exit(td); WITNESS_WARN(WARN_PANIC, NULL, "process (pid %d) exiting", p->p_pid); /* * Remove from allproc. It still sits in the hash. */ sx_xlock(&allproc_lock); LIST_REMOVE(p, p_list); #ifdef DDB /* * Used by ddb's 'ps' command to find this process via the * pidhash. */ p->p_list.le_prev = NULL; #endif prison_proc_unlink(p->p_ucred->cr_prison, p); sx_xunlock(&allproc_lock); sx_xlock(&proctree_lock); if ((p->p_flag & (P_TRACED | P_PPWAIT | P_PPTRACE)) != 0) { PROC_LOCK(p); p->p_flag &= ~(P_TRACED | P_PPWAIT | P_PPTRACE); PROC_UNLOCK(p); } /* * killjobc() might drop and re-acquire proctree_lock to * revoke control tty if exiting process was a session leader. */ killjobc(); /* * Reparent all children processes: * - traced ones to the original parent (or init if we are that parent) * - the rest to init */ q = LIST_FIRST(&p->p_children); if (q != NULL) /* only need this if any child is S_ZOMB */ wakeup(q->p_reaper); for (; q != NULL; q = nq) { nq = LIST_NEXT(q, p_sibling); ksi = ksiginfo_alloc(M_WAITOK); PROC_LOCK(q); q->p_sigparent = SIGCHLD; if ((q->p_flag & P_TRACED) == 0) { proc_reparent(q, q->p_reaper, true); if (q->p_state == PRS_ZOMBIE) { /* * Inform reaper about the reparented * zombie, since wait(2) has something * new to report. Guarantee queueing * of the SIGCHLD signal, similar to * the _exit() behaviour, by providing * our ksiginfo. Ksi is freed by the * signal delivery. */ if (q->p_ksi == NULL) { ksi1 = NULL; } else { ksiginfo_copy(q->p_ksi, ksi); ksi->ksi_flags |= KSI_INS; ksi1 = ksi; ksi = NULL; } PROC_LOCK(q->p_reaper); pksignal(q->p_reaper, SIGCHLD, ksi1); PROC_UNLOCK(q->p_reaper); } else if (q->p_pdeathsig > 0) { /* * The child asked to received a signal * when we exit. */ kern_psignal(q, q->p_pdeathsig); } } else { /* * Traced processes are killed by default * since their existence means someone is * screwing up. */ t = proc_realparent(q); if (t == p) { proc_reparent(q, q->p_reaper, true); } else { PROC_LOCK(t); proc_reparent(q, t, true); PROC_UNLOCK(t); } /* * Since q was found on our children list, the * proc_reparent() call moved q to the orphan * list due to present P_TRACED flag. Clear * orphan link for q now while q is locked. */ proc_clear_orphan(q); q->p_flag &= ~P_TRACED; q->p_flag2 &= ~P2_PTRACE_FSTP; q->p_ptevents = 0; p->p_xthread = NULL; FOREACH_THREAD_IN_PROC(q, tdt) { tdt->td_dbgflags &= ~(TDB_SUSPEND | TDB_XSIG | TDB_FSTP); tdt->td_xsig = 0; } if (kern_kill_on_dbg_exit) { q->p_flag &= ~P_STOPPED_TRACE; kern_psignal(q, SIGKILL); } else if ((q->p_flag & (P_STOPPED_TRACE | P_STOPPED_SIG)) != 0) { sigqueue_delete_proc(q, SIGTRAP); ptrace_unsuspend(q); } } PROC_UNLOCK(q); if (ksi != NULL) ksiginfo_free(ksi); } /* * Also get rid of our orphans. */ while ((q = LIST_FIRST(&p->p_orphans)) != NULL) { PROC_LOCK(q); KASSERT(q->p_oppid == p->p_pid, ("orphan %p of %p has unexpected oppid %d", q, p, q->p_oppid)); q->p_oppid = q->p_reaper->p_pid; /* * If we are the real parent of this process * but it has been reparented to a debugger, then * check if it asked for a signal when we exit. */ if (q->p_pdeathsig > 0) kern_psignal(q, q->p_pdeathsig); CTR2(KTR_PTRACE, "exit: pid %d, clearing orphan %d", p->p_pid, q->p_pid); proc_clear_orphan(q); PROC_UNLOCK(q); } #ifdef KDTRACE_HOOKS if (SDT_PROBES_ENABLED()) { int reason = CLD_EXITED; if (WCOREDUMP(signo)) reason = CLD_DUMPED; else if (WIFSIGNALED(signo)) reason = CLD_KILLED; SDT_PROBE1(proc, , , exit, reason); } #endif /* Save exit status. */ PROC_LOCK(p); p->p_xthread = td; if (p->p_sysent->sv_ontdexit != NULL) p->p_sysent->sv_ontdexit(td); #ifdef KDTRACE_HOOKS /* * Tell the DTrace fasttrap provider about the exit if it * has declared an interest. */ if (dtrace_fasttrap_exit) dtrace_fasttrap_exit(p); #endif /* * Notify interested parties of our demise. */ KNOTE_LOCKED(p->p_klist, NOTE_EXIT); /* * If this is a process with a descriptor, we may not need to deliver * a signal to the parent. proctree_lock is held over * procdesc_exit() to serialize concurrent calls to close() and * exit(). */ signal_parent = 0; if (p->p_procdesc == NULL || procdesc_exit(p)) { /* * Notify parent that we're gone. If parent has the * PS_NOCLDWAIT flag set, or if the handler is set to SIG_IGN, * notify process 1 instead (and hope it will handle this * situation). */ PROC_LOCK(p->p_pptr); mtx_lock(&p->p_pptr->p_sigacts->ps_mtx); if (p->p_pptr->p_sigacts->ps_flag & (PS_NOCLDWAIT | PS_CLDSIGIGN)) { struct proc *pp; mtx_unlock(&p->p_pptr->p_sigacts->ps_mtx); pp = p->p_pptr; PROC_UNLOCK(pp); proc_reparent(p, p->p_reaper, true); p->p_sigparent = SIGCHLD; PROC_LOCK(p->p_pptr); /* * Notify parent, so in case he was wait(2)ing or * executing waitpid(2) with our pid, he will * continue. */ wakeup(pp); } else mtx_unlock(&p->p_pptr->p_sigacts->ps_mtx); if (p->p_pptr == p->p_reaper || p->p_pptr == initproc) { signal_parent = 1; } else if (p->p_sigparent != 0) { if (p->p_sigparent == SIGCHLD) { signal_parent = 1; } else { /* LINUX thread */ signal_parent = 2; } } } else PROC_LOCK(p->p_pptr); sx_xunlock(&proctree_lock); if (signal_parent == 1) { childproc_exited(p); } else if (signal_parent == 2) { kern_psignal(p->p_pptr, p->p_sigparent); } /* Tell the prison that we are gone. */ prison_proc_free(p->p_ucred->cr_prison); /* * The state PRS_ZOMBIE prevents other processes from sending * signal to the process, to avoid memory leak, we free memory * for signal queue at the time when the state is set. */ sigqueue_flush(&p->p_sigqueue); sigqueue_flush(&td->td_sigqueue); /* * We have to wait until after acquiring all locks before * changing p_state. We need to avoid all possible context * switches (including ones from blocking on a mutex) while * marked as a zombie. We also have to set the zombie state * before we release the parent process' proc lock to avoid * a lost wakeup. So, we first call wakeup, then we grab the * sched lock, update the state, and release the parent process' * proc lock. */ wakeup(p->p_pptr); cv_broadcast(&p->p_pwait); sched_exit(p->p_pptr, td); PROC_SLOCK(p); p->p_state = PRS_ZOMBIE; PROC_UNLOCK(p->p_pptr); /* * Save our children's rusage information in our exit rusage. */ PROC_STATLOCK(p); ruadd(&p->p_ru, &p->p_rux, &p->p_stats->p_cru, &p->p_crux); PROC_STATUNLOCK(p); /* * Make sure the scheduler takes this thread out of its tables etc. * This will also release this thread's reference to the ucred. * Other thread parts to release include pcb bits and such. */ thread_exit(); } #ifndef _SYS_SYSPROTO_H_ struct abort2_args { char *why; int nargs; void **args; }; #endif int sys_abort2(struct thread *td, struct abort2_args *uap) { void *uargs[16]; void **uargsp; int error, nargs; nargs = uap->nargs; if (nargs < 0 || nargs > nitems(uargs)) nargs = -1; uargsp = NULL; if (nargs > 0) { if (uap->args != NULL) { error = copyin(uap->args, uargs, nargs * sizeof(void *)); if (error != 0) nargs = -1; else uargsp = uargs; } else nargs = -1; } return (kern_abort2(td, uap->why, nargs, uargsp)); } /* * kern_abort2() * Arguments: * why - user pointer to why * nargs - number of arguments copied or -1 if an error occurred in copying * args - pointer to an array of pointers in kernel format */ int kern_abort2(struct thread *td, const char *why, int nargs, void **uargs) { struct proc *p = td->td_proc; struct sbuf *sb; int error, i, sig; /* * Do it right now so we can log either proper call of abort2(), or * note, that invalid argument was passed. 512 is big enough to * handle 16 arguments' descriptions with additional comments. */ sb = sbuf_new(NULL, NULL, 512, SBUF_FIXEDLEN); sbuf_clear(sb); sbuf_printf(sb, "%s(pid %d uid %d) aborted: ", p->p_comm, p->p_pid, td->td_ucred->cr_uid); /* * Since we can't return from abort2(), send SIGKILL in cases, where * abort2() was called improperly */ sig = SIGKILL; /* Prevent from DoSes from user-space. */ if (nargs == -1) goto out; KASSERT(nargs >= 0 && nargs <= 16, ("called with too many args (%d)", nargs)); /* * Limit size of 'reason' string to 128. Will fit even when * maximal number of arguments was chosen to be logged. */ if (why != NULL) { error = sbuf_copyin(sb, why, 128); if (error < 0) goto out; } else { sbuf_cat(sb, "(null)"); } if (nargs > 0) { sbuf_putc(sb, '('); for (i = 0; i < nargs; i++) sbuf_printf(sb, "%s%p", i == 0 ? "" : ", ", uargs[i]); sbuf_putc(sb, ')'); } /* * Final stage: arguments were proper, string has been * successfully copied from userspace, and copying pointers * from user-space succeed. */ sig = SIGABRT; out: if (sig == SIGKILL) { sbuf_trim(sb); sbuf_cat(sb, " (Reason text inaccessible)"); } sbuf_cat(sb, "\n"); sbuf_finish(sb); log(LOG_INFO, "%s", sbuf_data(sb)); sbuf_delete(sb); PROC_LOCK(p); sigexit(td, sig); /* NOTREACHED */ } #ifdef COMPAT_43 /* * The dirty work is handled by kern_wait(). */ int owait(struct thread *td, struct owait_args *uap __unused) { int error, status; error = kern_wait(td, WAIT_ANY, &status, 0, NULL); if (error == 0) td->td_retval[1] = status; return (error); } #endif /* COMPAT_43 */ /* * The dirty work is handled by kern_wait(). */ int sys_wait4(struct thread *td, struct wait4_args *uap) { struct rusage ru, *rup; int error, status; if (uap->rusage != NULL) rup = &ru; else rup = NULL; error = kern_wait(td, uap->pid, &status, uap->options, rup); if (uap->status != NULL && error == 0 && td->td_retval[0] != 0) error = copyout(&status, uap->status, sizeof(status)); if (uap->rusage != NULL && error == 0 && td->td_retval[0] != 0) error = copyout(&ru, uap->rusage, sizeof(struct rusage)); return (error); } int sys_wait6(struct thread *td, struct wait6_args *uap) { struct __wrusage wru, *wrup; siginfo_t si, *sip; idtype_t idtype; id_t id; int error, status; idtype = uap->idtype; id = uap->id; if (uap->wrusage != NULL) wrup = &wru; else wrup = NULL; if (uap->info != NULL) { sip = &si; bzero(sip, sizeof(*sip)); } else sip = NULL; /* * We expect all callers of wait6() to know about WEXITED and * WTRAPPED. */ error = kern_wait6(td, idtype, id, &status, uap->options, wrup, sip); if (uap->status != NULL && error == 0 && td->td_retval[0] != 0) error = copyout(&status, uap->status, sizeof(status)); if (uap->wrusage != NULL && error == 0 && td->td_retval[0] != 0) error = copyout(&wru, uap->wrusage, sizeof(wru)); if (uap->info != NULL && error == 0) error = copyout(&si, uap->info, sizeof(si)); return (error); } /* * Reap the remains of a zombie process and optionally return status and * rusage. Asserts and will release both the proctree_lock and the process * lock as part of its work. */ void proc_reap(struct thread *td, struct proc *p, int *status, int options) { struct proc *q, *t; sx_assert(&proctree_lock, SA_XLOCKED); PROC_LOCK_ASSERT(p, MA_OWNED); KASSERT(p->p_state == PRS_ZOMBIE, ("proc_reap: !PRS_ZOMBIE")); mtx_spin_wait_unlocked(&p->p_slock); q = td->td_proc; if (status) *status = KW_EXITCODE(p->p_xexit, p->p_xsig); if (options & WNOWAIT) { /* * Only poll, returning the status. Caller does not wish to * release the proc struct just yet. */ PROC_UNLOCK(p); sx_xunlock(&proctree_lock); return; } PROC_LOCK(q); sigqueue_take(p->p_ksi); PROC_UNLOCK(q); /* * If we got the child via a ptrace 'attach', we need to give it back * to the old parent. */ if (p->p_oppid != p->p_pptr->p_pid) { PROC_UNLOCK(p); t = proc_realparent(p); PROC_LOCK(t); PROC_LOCK(p); CTR2(KTR_PTRACE, "wait: traced child %d moved back to parent %d", p->p_pid, t->p_pid); proc_reparent(p, t, false); PROC_UNLOCK(p); pksignal(t, SIGCHLD, p->p_ksi); wakeup(t); cv_broadcast(&p->p_pwait); PROC_UNLOCK(t); sx_xunlock(&proctree_lock); return; } PROC_UNLOCK(p); /* * Remove other references to this process to ensure we have an * exclusive reference. */ sx_xlock(PIDHASHLOCK(p->p_pid)); LIST_REMOVE(p, p_hash); sx_xunlock(PIDHASHLOCK(p->p_pid)); LIST_REMOVE(p, p_sibling); reaper_abandon_children(p, true); reaper_clear(p, p->p_reaper); PROC_LOCK(p); proc_clear_orphan(p); PROC_UNLOCK(p); leavepgrp(p); if (p->p_procdesc != NULL) procdesc_reap(p); sx_xunlock(&proctree_lock); proc_id_clear(PROC_ID_PID, p->p_pid); PROC_LOCK(p); + KNOTE_LOCKED(p->p_klist, NOTE_REAP); knlist_detach(p->p_klist); p->p_klist = NULL; PROC_UNLOCK(p); /* * Removal from allproc list and process group list paired with * PROC_LOCK which was executed during that time should guarantee * nothing can reach this process anymore. As such further locking * is unnecessary. */ p->p_xexit = p->p_xsig = 0; /* XXX: why? */ PROC_LOCK(q); ruadd(&q->p_stats->p_cru, &q->p_crux, &p->p_ru, &p->p_rux); PROC_UNLOCK(q); /* * Destroy resource accounting information associated with the process. */ #ifdef RACCT if (racct_enable) { PROC_LOCK(p); racct_sub(p, RACCT_NPROC, 1); PROC_UNLOCK(p); } #endif racct_proc_exit(p); /* * Free credentials, arguments, and sigacts, and decrement the count of * processes running with this uid. */ proc_unset_cred(p, true); pargs_drop(p->p_args); p->p_args = NULL; sigacts_free(p->p_sigacts); p->p_sigacts = NULL; /* * Do any thread-system specific cleanups. */ thread_wait(p); /* * Give vm and machine-dependent layer a chance to free anything that * cpu_exit couldn't release while still running in process context. */ vm_waitproc(p); #ifdef MAC mac_proc_destroy(p); #endif KASSERT(FIRST_THREAD_IN_PROC(p), ("proc_reap: no residual thread!")); uma_zfree(proc_zone, p); atomic_add_int(&nprocs, -1); } static int proc_to_reap(struct thread *td, struct proc *p, idtype_t idtype, id_t id, int *status, int options, struct __wrusage *wrusage, siginfo_t *siginfo, int check_only) { struct rusage *rup; sx_assert(&proctree_lock, SA_XLOCKED); PROC_LOCK(p); switch (idtype) { case P_ALL: if (p->p_procdesc == NULL || (p->p_pptr == td->td_proc && (p->p_flag & P_TRACED) != 0)) { break; } PROC_UNLOCK(p); return (0); case P_PID: if (p->p_pid != (pid_t)id) { PROC_UNLOCK(p); return (0); } break; case P_PGID: if (p->p_pgid != (pid_t)id) { PROC_UNLOCK(p); return (0); } break; case P_SID: if (p->p_session->s_sid != (pid_t)id) { PROC_UNLOCK(p); return (0); } break; case P_UID: if (p->p_ucred->cr_uid != (uid_t)id) { PROC_UNLOCK(p); return (0); } break; case P_GID: if (p->p_ucred->cr_gid != (gid_t)id) { PROC_UNLOCK(p); return (0); } break; case P_JAILID: if (p->p_ucred->cr_prison->pr_id != (int)id) { PROC_UNLOCK(p); return (0); } break; /* * It seems that the thread structures get zeroed out * at process exit. This makes it impossible to * support P_SETID, P_CID or P_CPUID. */ default: PROC_UNLOCK(p); return (0); } if (p_canwait(td, p)) { PROC_UNLOCK(p); return (0); } if (((options & WEXITED) == 0) && (p->p_state == PRS_ZOMBIE)) { PROC_UNLOCK(p); return (0); } /* * This special case handles a kthread spawned by linux_clone * (see linux_misc.c). The linux_wait4 and linux_waitpid * functions need to be able to distinguish between waiting * on a process and waiting on a thread. It is a thread if * p_sigparent is not SIGCHLD, and the WLINUXCLONE option * signifies we want to wait for threads and not processes. */ if ((p->p_sigparent != SIGCHLD) ^ ((options & WLINUXCLONE) != 0)) { PROC_UNLOCK(p); return (0); } if (siginfo != NULL) { bzero(siginfo, sizeof(*siginfo)); siginfo->si_errno = 0; /* * SUSv4 requires that the si_signo value is always * SIGCHLD. Obey it despite the rfork(2) interface * allows to request other signal for child exit * notification. */ siginfo->si_signo = SIGCHLD; /* * This is still a rough estimate. We will fix the * cases TRAPPED, STOPPED, and CONTINUED later. */ if (WCOREDUMP(p->p_xsig)) { siginfo->si_code = CLD_DUMPED; siginfo->si_status = WTERMSIG(p->p_xsig); } else if (WIFSIGNALED(p->p_xsig)) { siginfo->si_code = CLD_KILLED; siginfo->si_status = WTERMSIG(p->p_xsig); } else { siginfo->si_code = CLD_EXITED; siginfo->si_status = p->p_xexit; } siginfo->si_pid = p->p_pid; siginfo->si_uid = p->p_ucred->cr_uid; /* * The si_addr field would be useful additional * detail, but apparently the PC value may be lost * when we reach this point. bzero() above sets * siginfo->si_addr to NULL. */ } /* * There should be no reason to limit resources usage info to * exited processes only. A snapshot about any resources used * by a stopped process may be exactly what is needed. */ if (wrusage != NULL) { rup = &wrusage->wru_self; *rup = p->p_ru; PROC_STATLOCK(p); calcru(p, &rup->ru_utime, &rup->ru_stime); PROC_STATUNLOCK(p); rup = &wrusage->wru_children; *rup = p->p_stats->p_cru; calccru(p, &rup->ru_utime, &rup->ru_stime); } if (p->p_state == PRS_ZOMBIE && !check_only) { proc_reap(td, p, status, options); return (-1); } return (1); } int kern_wait(struct thread *td, pid_t pid, int *status, int options, struct rusage *rusage) { struct __wrusage wru, *wrup; idtype_t idtype; id_t id; int ret; /* * Translate the special pid values into the (idtype, pid) * pair for kern_wait6. The WAIT_MYPGRP case is handled by * kern_wait6() on its own. */ if (pid == WAIT_ANY) { idtype = P_ALL; id = 0; } else if (pid < 0) { idtype = P_PGID; id = (id_t)-pid; } else { idtype = P_PID; id = (id_t)pid; } if (rusage != NULL) wrup = &wru; else wrup = NULL; /* * For backward compatibility we implicitly add flags WEXITED * and WTRAPPED here. */ options |= WEXITED | WTRAPPED; ret = kern_wait6(td, idtype, id, status, options, wrup, NULL); if (rusage != NULL) *rusage = wru.wru_self; return (ret); } static void report_alive_proc(struct thread *td, struct proc *p, siginfo_t *siginfo, int *status, int options, int si_code) { bool cont; PROC_LOCK_ASSERT(p, MA_OWNED); sx_assert(&proctree_lock, SA_XLOCKED); MPASS(si_code == CLD_TRAPPED || si_code == CLD_STOPPED || si_code == CLD_CONTINUED); cont = si_code == CLD_CONTINUED; if ((options & WNOWAIT) == 0) { if (cont) p->p_flag &= ~P_CONTINUED; else p->p_flag |= P_WAITED; if (kern_wait_dequeue_sigchld && (td->td_proc->p_sysent->sv_flags & SV_SIG_WAITNDQ) == 0) { PROC_LOCK(td->td_proc); sigqueue_take(p->p_ksi); PROC_UNLOCK(td->td_proc); } } sx_xunlock(&proctree_lock); if (siginfo != NULL) { siginfo->si_code = si_code; siginfo->si_status = cont ? SIGCONT : p->p_xsig; } if (status != NULL) *status = cont ? SIGCONT : W_STOPCODE(p->p_xsig); td->td_retval[0] = p->p_pid; PROC_UNLOCK(p); } int kern_wait6(struct thread *td, idtype_t idtype, id_t id, int *status, int options, struct __wrusage *wrusage, siginfo_t *siginfo) { struct proc *p, *q; pid_t pid; int error, nfound, ret; bool report; AUDIT_ARG_VALUE((int)idtype); /* XXX - This is likely wrong! */ AUDIT_ARG_PID((pid_t)id); /* XXX - This may be wrong! */ AUDIT_ARG_VALUE(options); q = td->td_proc; if ((pid_t)id == WAIT_MYPGRP && (idtype == P_PID || idtype == P_PGID)) { PROC_LOCK(q); id = (id_t)q->p_pgid; PROC_UNLOCK(q); idtype = P_PGID; } /* If we don't know the option, just return. */ if ((options & ~(WUNTRACED | WNOHANG | WCONTINUED | WNOWAIT | WEXITED | WTRAPPED | WLINUXCLONE)) != 0) return (EINVAL); if ((options & (WEXITED | WUNTRACED | WCONTINUED | WTRAPPED)) == 0) { /* * We will be unable to find any matching processes, * because there are no known events to look for. * Prefer to return error instead of blocking * indefinitely. */ return (EINVAL); } loop: if (q->p_flag & P_STATCHILD) { PROC_LOCK(q); q->p_flag &= ~P_STATCHILD; PROC_UNLOCK(q); } sx_xlock(&proctree_lock); loop_locked: nfound = 0; LIST_FOREACH(p, &q->p_children, p_sibling) { pid = p->p_pid; ret = proc_to_reap(td, p, idtype, id, status, options, wrusage, siginfo, 0); if (ret == 0) continue; else if (ret != 1) { td->td_retval[0] = pid; return (0); } nfound++; PROC_LOCK_ASSERT(p, MA_OWNED); if ((options & WTRAPPED) != 0 && (p->p_flag & P_TRACED) != 0) { PROC_SLOCK(p); report = ((p->p_flag & (P_STOPPED_TRACE | P_STOPPED_SIG)) && p->p_suspcount == p->p_numthreads && (p->p_flag & P_WAITED) == 0); PROC_SUNLOCK(p); if (report) { CTR4(KTR_PTRACE, "wait: returning trapped pid %d status %#x " "(xstat %d) xthread %d", p->p_pid, W_STOPCODE(p->p_xsig), p->p_xsig, p->p_xthread != NULL ? p->p_xthread->td_tid : -1); report_alive_proc(td, p, siginfo, status, options, CLD_TRAPPED); return (0); } } if ((options & WUNTRACED) != 0 && (p->p_flag & P_STOPPED_SIG) != 0) { PROC_SLOCK(p); report = (p->p_suspcount == p->p_numthreads && ((p->p_flag & P_WAITED) == 0)); PROC_SUNLOCK(p); if (report) { report_alive_proc(td, p, siginfo, status, options, CLD_STOPPED); return (0); } } if ((options & WCONTINUED) != 0 && (p->p_flag & P_CONTINUED) != 0) { report_alive_proc(td, p, siginfo, status, options, CLD_CONTINUED); return (0); } PROC_UNLOCK(p); } /* * Look in the orphans list too, to allow the parent to * collect it's child exit status even if child is being * debugged. * * Debugger detaches from the parent upon successful * switch-over from parent to child. At this point due to * re-parenting the parent loses the child to debugger and a * wait4(2) call would report that it has no children to wait * for. By maintaining a list of orphans we allow the parent * to successfully wait until the child becomes a zombie. */ if (nfound == 0) { LIST_FOREACH(p, &q->p_orphans, p_orphan) { ret = proc_to_reap(td, p, idtype, id, NULL, options, NULL, NULL, 1); if (ret != 0) { KASSERT(ret != -1, ("reaped an orphan (pid %d)", (int)td->td_retval[0])); PROC_UNLOCK(p); nfound++; break; } } } if (nfound == 0) { sx_xunlock(&proctree_lock); return (ECHILD); } if (options & WNOHANG) { sx_xunlock(&proctree_lock); td->td_retval[0] = 0; return (0); } PROC_LOCK(q); if (q->p_flag & P_STATCHILD) { q->p_flag &= ~P_STATCHILD; PROC_UNLOCK(q); goto loop_locked; } sx_xunlock(&proctree_lock); error = msleep(q, &q->p_mtx, PWAIT | PCATCH | PDROP, "wait", 0); if (error) return (error); goto loop; } void proc_add_orphan(struct proc *child, struct proc *parent) { sx_assert(&proctree_lock, SX_XLOCKED); KASSERT((child->p_flag & P_TRACED) != 0, ("proc_add_orphan: not traced")); if (LIST_EMPTY(&parent->p_orphans)) { child->p_treeflag |= P_TREE_FIRST_ORPHAN; LIST_INSERT_HEAD(&parent->p_orphans, child, p_orphan); } else { LIST_INSERT_AFTER(LIST_FIRST(&parent->p_orphans), child, p_orphan); } child->p_treeflag |= P_TREE_ORPHANED; } /* * Make process 'parent' the new parent of process 'child'. * Must be called with an exclusive hold of proctree lock. */ void proc_reparent(struct proc *child, struct proc *parent, bool set_oppid) { sx_assert(&proctree_lock, SX_XLOCKED); PROC_LOCK_ASSERT(child, MA_OWNED); if (child->p_pptr == parent) return; PROC_LOCK(child->p_pptr); sigqueue_take(child->p_ksi); PROC_UNLOCK(child->p_pptr); LIST_REMOVE(child, p_sibling); LIST_INSERT_HEAD(&parent->p_children, child, p_sibling); proc_clear_orphan(child); if ((child->p_flag & P_TRACED) != 0) { proc_add_orphan(child, child->p_pptr); } child->p_pptr = parent; if (set_oppid) child->p_oppid = parent->p_pid; } diff --git a/sys/kern/sys_procdesc.c b/sys/kern/sys_procdesc.c index f1aeb4066fa4..c6781aac2523 100644 --- a/sys/kern/sys_procdesc.c +++ b/sys/kern/sys_procdesc.c @@ -1,568 +1,571 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2009, 2016 Robert N. M. Watson * All rights reserved. * * This software was developed at the University of Cambridge Computer * Laboratory with support from a grant from Google, Inc. * * Portions of this software were developed by BAE Systems, the University of * Cambridge Computer Laboratory, and Memorial University under DARPA/AFRL * contract FA8650-15-C-7558 ("CADETS"), as part of the DARPA Transparent * Computing (TC) research program. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /*- * FreeBSD process descriptor facility. * * Some processes are represented by a file descriptor, which will be used in * preference to signaling and pids for the purposes of process management, * and is, in effect, a form of capability. When a process descriptor is * used with a process, it ceases to be visible to certain traditional UNIX * process facilities, such as waitpid(2). * * Some semantics: * * - At most one process descriptor will exist for any process, although * references to that descriptor may be held from many processes (or even * be in flight between processes over a local domain socket). * - Last close on the process descriptor will terminate the process using * SIGKILL and reparent it to init so that there's a process to reap it * when it's done exiting. * - If the process exits before the descriptor is closed, it will not * generate SIGCHLD on termination, or be picked up by waitpid(). * - The pdkill(2) system call may be used to deliver a signal to the process * using its process descriptor. * - The pdwait4(2) system call may be used to block (or not) on a process * descriptor to collect termination information. * * Open questions: * * - Will we want to add a pidtoprocdesc(2) system call to allow process * descriptors to be created for processes without pdfork(2)? */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include FEATURE(process_descriptors, "Process Descriptors"); MALLOC_DEFINE(M_PROCDESC, "procdesc", "process descriptors"); static fo_poll_t procdesc_poll; static fo_kqfilter_t procdesc_kqfilter; static fo_stat_t procdesc_stat; static fo_close_t procdesc_close; static fo_fill_kinfo_t procdesc_fill_kinfo; static fo_cmp_t procdesc_cmp; static const struct fileops procdesc_ops = { .fo_read = invfo_rdwr, .fo_write = invfo_rdwr, .fo_truncate = invfo_truncate, .fo_ioctl = invfo_ioctl, .fo_poll = procdesc_poll, .fo_kqfilter = procdesc_kqfilter, .fo_stat = procdesc_stat, .fo_close = procdesc_close, .fo_chmod = invfo_chmod, .fo_chown = invfo_chown, .fo_sendfile = invfo_sendfile, .fo_fill_kinfo = procdesc_fill_kinfo, .fo_cmp = procdesc_cmp, .fo_flags = DFLAG_PASSABLE, }; /* * Return a locked process given a process descriptor, or ESRCH if it has * died. */ int procdesc_find(struct thread *td, int fd, const cap_rights_t *rightsp, struct proc **p) { struct procdesc *pd; struct file *fp; int error; error = fget(td, fd, rightsp, &fp); if (error) return (error); if (fp->f_type != DTYPE_PROCDESC) { error = EINVAL; goto out; } pd = fp->f_data; sx_slock(&proctree_lock); if (pd->pd_proc != NULL) { *p = pd->pd_proc; PROC_LOCK(*p); } else error = ESRCH; sx_sunlock(&proctree_lock); out: fdrop(fp, td); return (error); } /* * Function to be used by procstat(1) sysctls when returning procdesc * information. */ pid_t procdesc_pid(struct file *fp_procdesc) { struct procdesc *pd; KASSERT(fp_procdesc->f_type == DTYPE_PROCDESC, ("procdesc_pid: !procdesc")); pd = fp_procdesc->f_data; return (pd->pd_pid); } /* * Retrieve the PID associated with a process descriptor. */ int kern_pdgetpid(struct thread *td, int fd, const cap_rights_t *rightsp, pid_t *pidp) { struct file *fp; int error; error = fget(td, fd, rightsp, &fp); if (error) return (error); if (fp->f_type != DTYPE_PROCDESC) { error = EBADF; goto out; } *pidp = procdesc_pid(fp); out: fdrop(fp, td); return (error); } /* * System call to return the pid of a process given its process descriptor. */ int sys_pdgetpid(struct thread *td, struct pdgetpid_args *uap) { pid_t pid; int error; AUDIT_ARG_FD(uap->fd); error = kern_pdgetpid(td, uap->fd, &cap_pdgetpid_rights, &pid); if (error == 0) error = copyout(&pid, uap->pidp, sizeof(pid)); return (error); } /* * When a new process is forked by pdfork(), a file descriptor is allocated * by the fork code first, then the process is forked, and then we get a * chance to set up the process descriptor. Failure is not permitted at this * point, so procdesc_new() must succeed. */ void procdesc_new(struct proc *p, int flags) { struct procdesc *pd; pd = malloc(sizeof(*pd), M_PROCDESC, M_WAITOK | M_ZERO); pd->pd_proc = p; pd->pd_pid = p->p_pid; p->p_procdesc = pd; pd->pd_flags = 0; if (flags & PD_DAEMON) pd->pd_flags |= PDF_DAEMON; PROCDESC_LOCK_INIT(pd); knlist_init_mtx(&pd->pd_selinfo.si_note, &pd->pd_lock); /* * Process descriptors start out with two references: one from their * struct file, and the other from their struct proc. */ refcount_init(&pd->pd_refcount, 2); } /* * Create a new process decriptor for the process that refers to it. */ int procdesc_falloc(struct thread *td, struct file **resultfp, int *resultfd, int flags, struct filecaps *fcaps) { int fflags; fflags = 0; if (flags & PD_CLOEXEC) fflags = O_CLOEXEC; return (falloc_caps(td, resultfp, resultfd, fflags, fcaps)); } /* * Initialize a file with a process descriptor. */ void procdesc_finit(struct procdesc *pdp, struct file *fp) { finit(fp, FREAD | FWRITE, DTYPE_PROCDESC, pdp, &procdesc_ops); } static void procdesc_free(struct procdesc *pd) { /* * When the last reference is released, we assert that the descriptor * has been closed, but not that the process has exited, as we will * detach the descriptor before the process dies if the descript is * closed, as we can't wait synchronously. */ if (refcount_release(&pd->pd_refcount)) { KASSERT(pd->pd_proc == NULL, ("procdesc_free: pd_proc != NULL")); KASSERT((pd->pd_flags & PDF_CLOSED), ("procdesc_free: !PDF_CLOSED")); seldrain(&pd->pd_selinfo); knlist_destroy(&pd->pd_selinfo.si_note); PROCDESC_LOCK_DESTROY(pd); free(pd, M_PROCDESC); } } /* * procdesc_exit() - notify a process descriptor that its process is exiting. * We use the proctree_lock to ensure that process exit either happens * strictly before or strictly after a concurrent call to procdesc_close(). */ int procdesc_exit(struct proc *p) { struct procdesc *pd; sx_assert(&proctree_lock, SA_XLOCKED); PROC_LOCK_ASSERT(p, MA_OWNED); KASSERT(p->p_procdesc != NULL, ("procdesc_exit: p_procdesc NULL")); pd = p->p_procdesc; PROCDESC_LOCK(pd); KASSERT((pd->pd_flags & PDF_CLOSED) == 0 || p->p_pptr == p->p_reaper, ("procdesc_exit: closed && parent not reaper")); pd->pd_flags |= PDF_EXITED; pd->pd_xstat = KW_EXITCODE(p->p_xexit, p->p_xsig); /* * If the process descriptor has been closed, then we have nothing * to do; return 1 so that init will get SIGCHLD and do the reaping. * Clean up the procdesc now rather than letting it happen during * that reap. */ if (pd->pd_flags & PDF_CLOSED) { PROCDESC_UNLOCK(pd); pd->pd_proc = NULL; p->p_procdesc = NULL; procdesc_free(pd); return (1); } selwakeup(&pd->pd_selinfo); KNOTE_LOCKED(&pd->pd_selinfo.si_note, NOTE_EXIT); PROCDESC_UNLOCK(pd); return (0); } /* * When a process descriptor is reaped, perhaps as a result of close() or * pdwait4(), release the process's reference on the process descriptor. */ void procdesc_reap(struct proc *p) { struct procdesc *pd; sx_assert(&proctree_lock, SA_XLOCKED); KASSERT(p->p_procdesc != NULL, ("procdesc_reap: p_procdesc == NULL")); pd = p->p_procdesc; pd->pd_proc = NULL; p->p_procdesc = NULL; procdesc_free(pd); } /* * procdesc_close() - last close on a process descriptor. If the process is * still running, terminate with SIGKILL (unless PDF_DAEMON is set) and let * its reaper clean up the mess; if not, we have to clean up the zombie * ourselves. */ static int procdesc_close(struct file *fp, struct thread *td) { struct procdesc *pd; struct proc *p; KASSERT(fp->f_type == DTYPE_PROCDESC, ("procdesc_close: !procdesc")); pd = fp->f_data; fp->f_ops = &badfileops; fp->f_data = NULL; sx_xlock(&proctree_lock); PROCDESC_LOCK(pd); pd->pd_flags |= PDF_CLOSED; PROCDESC_UNLOCK(pd); p = pd->pd_proc; if (p == NULL) { /* * This is the case where process' exit status was already * collected and procdesc_reap() was already called. */ sx_xunlock(&proctree_lock); } else { PROC_LOCK(p); AUDIT_ARG_PROCESS(p); if (p->p_state == PRS_ZOMBIE) { /* * If the process is already dead and just awaiting * reaping, do that now. This will release the * process's reference to the process descriptor when it * calls back into procdesc_reap(). */ proc_reap(curthread, p, NULL, 0); } else { /* * If the process is not yet dead, we need to kill it, * but we can't wait around synchronously for it to go * away, as that path leads to madness (and deadlocks). * First, detach the process from its descriptor so that * its exit status will be reported normally. */ pd->pd_proc = NULL; p->p_procdesc = NULL; procdesc_free(pd); /* * Next, reparent it to its reaper (usually init(8)) so * that there's someone to pick up the pieces; finally, * terminate with prejudice. */ p->p_sigparent = SIGCHLD; if ((p->p_flag & P_TRACED) == 0) { proc_reparent(p, p->p_reaper, true); } else { proc_clear_orphan(p); p->p_oppid = p->p_reaper->p_pid; proc_add_orphan(p, p->p_reaper); } if ((pd->pd_flags & PDF_DAEMON) == 0) kern_psignal(p, SIGKILL); PROC_UNLOCK(p); sx_xunlock(&proctree_lock); } } /* * Release the file descriptor's reference on the process descriptor. */ procdesc_free(pd); return (0); } static int procdesc_poll(struct file *fp, int events, struct ucred *active_cred, struct thread *td) { struct procdesc *pd; int revents; revents = 0; pd = fp->f_data; PROCDESC_LOCK(pd); if (pd->pd_flags & PDF_EXITED) revents |= POLLHUP; else selrecord(td, &pd->pd_selinfo); PROCDESC_UNLOCK(pd); return (revents); } static void procdesc_kqops_detach(struct knote *kn) { struct procdesc *pd; pd = kn->kn_fp->f_data; knlist_remove(&pd->pd_selinfo.si_note, kn, 0); } static int procdesc_kqops_event(struct knote *kn, long hint) { struct procdesc *pd; u_int event; pd = kn->kn_fp->f_data; if (hint == 0) { /* * Initial test after registration. Generate a NOTE_EXIT in * case the process already terminated before registration. */ event = pd->pd_flags & PDF_EXITED ? NOTE_EXIT : 0; } else { /* Mask off extra data. */ event = (u_int)hint & NOTE_PCTRLMASK; } /* If the user is interested in this event, record it. */ if ((kn->kn_sfflags & event) != 0) kn->kn_fflags |= kn->kn_sfflags & event; + /* Report exit status */ + if ((kn->kn_fflags & NOTE_EXIT) != 0) + kn->kn_data = pd->pd_xstat; + /* Process is gone, so flag the event as finished. */ - if (event == NOTE_EXIT) { + if ((event & NOTE_REAP) != 0 || + ((event & NOTE_EXIT) != 0 && (kn->kn_sfflags & NOTE_REAP) == 0)) { kn->kn_flags |= EV_EOF | EV_ONESHOT; - if (kn->kn_fflags & NOTE_EXIT) - kn->kn_data = pd->pd_xstat; if (kn->kn_fflags == 0) kn->kn_flags |= EV_DROP; return (1); } return (kn->kn_fflags != 0); } static const struct filterops procdesc_kqops = { .f_isfd = 1, .f_detach = procdesc_kqops_detach, .f_event = procdesc_kqops_event, }; static int procdesc_kqfilter(struct file *fp, struct knote *kn) { struct procdesc *pd; pd = fp->f_data; switch (kn->kn_filter) { case EVFILT_PROCDESC: kn->kn_fop = &procdesc_kqops; kn->kn_flags |= EV_CLEAR; knlist_add(&pd->pd_selinfo.si_note, kn, 0); return (0); default: return (EINVAL); } } static int procdesc_stat(struct file *fp, struct stat *sb, struct ucred *active_cred) { struct procdesc *pd; struct timeval pstart, boottime; /* * XXXRW: Perhaps we should cache some more information from the * process so that we can return it reliably here even after it has * died. For example, caching its credential data. */ bzero(sb, sizeof(*sb)); pd = fp->f_data; sx_slock(&proctree_lock); if (pd->pd_proc != NULL) { PROC_LOCK(pd->pd_proc); AUDIT_ARG_PROCESS(pd->pd_proc); /* Set birth and [acm] times to process start time. */ pstart = pd->pd_proc->p_stats->p_start; getboottime(&boottime); timevaladd(&pstart, &boottime); TIMEVAL_TO_TIMESPEC(&pstart, &sb->st_birthtim); sb->st_atim = sb->st_birthtim; sb->st_ctim = sb->st_birthtim; sb->st_mtim = sb->st_birthtim; if (pd->pd_proc->p_state != PRS_ZOMBIE) sb->st_mode = S_IFREG | S_IRWXU; else sb->st_mode = S_IFREG; sb->st_uid = pd->pd_proc->p_ucred->cr_ruid; sb->st_gid = pd->pd_proc->p_ucred->cr_rgid; PROC_UNLOCK(pd->pd_proc); } else sb->st_mode = S_IFREG; sx_sunlock(&proctree_lock); return (0); } static int procdesc_fill_kinfo(struct file *fp, struct kinfo_file *kif, struct filedesc *fdp) { struct procdesc *pdp; kif->kf_type = KF_TYPE_PROCDESC; pdp = fp->f_data; kif->kf_un.kf_proc.kf_pid = pdp->pd_pid; return (0); } static int procdesc_cmp(struct file *fp1, struct file *fp2, struct thread *td) { struct procdesc *pdp1, *pdp2; if (fp2->f_type != DTYPE_PROCDESC) return (3); pdp1 = fp1->f_data; pdp2 = fp2->f_data; return (kcmp_cmp((uintptr_t)pdp1->pd_pid, (uintptr_t)pdp2->pd_pid)); } diff --git a/sys/sys/event.h b/sys/sys/event.h index a1cd224a960e..246ebdedd004 100644 --- a/sys/sys/event.h +++ b/sys/sys/event.h @@ -1,378 +1,375 @@ /*- * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 1999,2000,2001 Jonathan Lemon * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef _SYS_EVENT_H_ #define _SYS_EVENT_H_ #include #include #define EVFILT_READ (-1) #define EVFILT_WRITE (-2) #define EVFILT_AIO (-3) /* attached to aio requests */ #define EVFILT_VNODE (-4) /* attached to vnodes */ #define EVFILT_PROC (-5) /* attached to struct proc */ #define EVFILT_SIGNAL (-6) /* attached to struct proc */ #define EVFILT_TIMER (-7) /* timers */ #define EVFILT_PROCDESC (-8) /* attached to process descriptors */ #define EVFILT_FS (-9) /* filesystem events */ #define EVFILT_LIO (-10) /* attached to lio requests */ #define EVFILT_USER (-11) /* User events */ #define EVFILT_SENDFILE (-12) /* attached to sendfile requests */ #define EVFILT_EMPTY (-13) /* empty send socket buf */ #define EVFILT_SYSCOUNT 13 #if defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L #define EV_SET(kevp_, a, b, c, d, e, f) do { \ *(kevp_) = (struct kevent){ \ .ident = (a), \ .filter = (b), \ .flags = (c), \ .fflags = (d), \ .data = (e), \ .udata = (f), \ .ext = {0}, \ }; \ } while (0) #else /* Pre-C99 or not STDC (e.g., C++) */ /* * The definition of the local variable kevp could possibly conflict * with a user-defined value passed in parameters a-f. */ #define EV_SET(kevp_, a, b, c, d, e, f) do { \ struct kevent *kevp = (kevp_); \ (kevp)->ident = (a); \ (kevp)->filter = (b); \ (kevp)->flags = (c); \ (kevp)->fflags = (d); \ (kevp)->data = (e); \ (kevp)->udata = (f); \ (kevp)->ext[0] = 0; \ (kevp)->ext[1] = 0; \ (kevp)->ext[2] = 0; \ (kevp)->ext[3] = 0; \ } while (0) #endif struct kevent { __uintptr_t ident; /* identifier for this event */ short filter; /* filter for event */ unsigned short flags; /* action flags for kqueue */ unsigned int fflags; /* filter flag value */ __int64_t data; /* filter data value */ void *udata; /* opaque user data identifier */ __uint64_t ext[4]; /* extensions */ }; #if defined(_WANT_FREEBSD11_KEVENT) /* Older structure used in FreeBSD 11.x and older. */ struct freebsd11_kevent { __uintptr_t ident; /* identifier for this event */ short filter; /* filter for event */ unsigned short flags; unsigned int fflags; __intptr_t data; void *udata; /* opaque user data identifier */ }; #endif #if defined(_WANT_KEVENT32) || defined(_KERNEL) #include struct kevent32 { __uint32_t ident; /* identifier for this event */ short filter; /* filter for event */ unsigned short flags; unsigned int fflags; #ifndef __amd64__ __uint32_t pad0; #endif freebsd32_uint64_t data; __uint32_t udata; /* opaque user data identifier */ #ifndef __amd64__ __uint32_t pad1; #endif freebsd32_uint64_t ext[4]; }; #ifdef _WANT_FREEBSD11_KEVENT struct freebsd11_kevent32 { __uint32_t ident; /* identifier for this event */ short filter; /* filter for event */ unsigned short flags; unsigned int fflags; __int32_t data; __uint32_t udata; /* opaque user data identifier */ }; #endif #endif /* actions */ #define EV_ADD 0x0001 /* add event to kq (implies enable) */ #define EV_DELETE 0x0002 /* delete event from kq */ #define EV_ENABLE 0x0004 /* enable event */ #define EV_DISABLE 0x0008 /* disable event (not reported) */ #define EV_FORCEONESHOT 0x0100 /* enable _ONESHOT and force trigger */ #define EV_KEEPUDATA 0x0200 /* do not update the udata field */ /* flags */ #define EV_ONESHOT 0x0010 /* only report one occurrence */ #define EV_CLEAR 0x0020 /* clear event state after reporting */ #define EV_RECEIPT 0x0040 /* force EV_ERROR on success, data=0 */ #define EV_DISPATCH 0x0080 /* disable event after reporting */ #define EV_SYSFLAGS 0xF000 /* reserved by system */ #define EV_DROP 0x1000 /* note should be dropped */ #define EV_FLAG1 0x2000 /* filter-specific flag */ #define EV_FLAG2 0x4000 /* filter-specific flag */ /* returned values */ #define EV_EOF 0x8000 /* EOF detected */ #define EV_ERROR 0x4000 /* error, data contains errno */ /* * data/hint flags/masks for EVFILT_USER, shared with userspace * * On input, the top two bits of fflags specifies how the lower twenty four * bits should be applied to the stored value of fflags. * * On output, the top two bits will always be set to NOTE_FFNOP and the * remaining twenty four bits will contain the stored fflags value. */ #define NOTE_FFNOP 0x00000000 /* ignore input fflags */ #define NOTE_FFAND 0x40000000 /* AND fflags */ #define NOTE_FFOR 0x80000000 /* OR fflags */ #define NOTE_FFCOPY 0xc0000000 /* copy fflags */ #define NOTE_FFCTRLMASK 0xc0000000 /* masks for operations */ #define NOTE_FFLAGSMASK 0x00ffffff #define NOTE_TRIGGER 0x01000000 /* Cause the event to be triggered for output. */ /* * data/hint flags for EVFILT_{READ|WRITE}, shared with userspace */ #define NOTE_LOWAT 0x0001 /* low water mark */ #define NOTE_FILE_POLL 0x0002 /* behave like poll() */ /* * data/hint flags for EVFILT_VNODE, shared with userspace */ #define NOTE_DELETE 0x0001 /* vnode was removed */ #define NOTE_WRITE 0x0002 /* data contents changed */ #define NOTE_EXTEND 0x0004 /* size increased */ #define NOTE_ATTRIB 0x0008 /* attributes changed */ #define NOTE_LINK 0x0010 /* link count changed */ #define NOTE_RENAME 0x0020 /* vnode was renamed */ #define NOTE_REVOKE 0x0040 /* vnode access was revoked */ #define NOTE_OPEN 0x0080 /* vnode was opened */ #define NOTE_CLOSE 0x0100 /* file closed, fd did not allowed write */ #define NOTE_CLOSE_WRITE 0x0200 /* file closed, fd did allowed write */ #define NOTE_READ 0x0400 /* file was read */ /* * data/hint flags for EVFILT_PROC and EVFILT_PROCDESC, shared with userspace */ #define NOTE_EXIT 0x80000000 /* process exited */ #define NOTE_FORK 0x40000000 /* process forked */ #define NOTE_EXEC 0x20000000 /* process exec'd */ -#define NOTE_PCTRLMASK 0xf0000000 /* mask for hint bits */ +#define NOTE_SIGNAL 0x08000000 /* process received a signal, + shared with EVFIL_SIGNAL */ +#define NOTE_REAP 0x04000000 /* process reaped */ +#define NOTE_PCTRLMASK 0xec000000 /* mask for hint bits */ #define NOTE_PDATAMASK 0x000fffff /* mask for pid */ /* additional flags for EVFILT_PROC */ #define NOTE_TRACK 0x00000001 /* follow across forks */ #define NOTE_TRACKERR 0x00000002 /* could not track child */ #define NOTE_CHILD 0x00000004 /* am a child process */ /* additional flags for EVFILT_TIMER */ #define NOTE_SECONDS 0x00000001 /* data is seconds */ #define NOTE_MSECONDS 0x00000002 /* data is milliseconds */ #define NOTE_USECONDS 0x00000004 /* data is microseconds */ #define NOTE_NSECONDS 0x00000008 /* data is nanoseconds */ #define NOTE_ABSTIME 0x00000010 /* timeout is absolute */ /* Flags for kqueuex(2) */ #define KQUEUE_CLOEXEC 0x00000001 /* close on exec */ struct knote; SLIST_HEAD(klist, knote); struct kqueue; TAILQ_HEAD(kqlist, kqueue); struct knlist { struct klist kl_list; void (*kl_lock)(void *); /* lock function */ void (*kl_unlock)(void *); void (*kl_assert_lock)(void *, int); void *kl_lockarg; /* argument passed to lock functions */ int kl_autodestroy; }; #ifdef _KERNEL /* * Flags for knote call */ #define KNF_LISTLOCKED 0x0001 /* knlist is locked */ #define KNF_NOKQLOCK 0x0002 /* do not keep KQ_LOCK */ #define KNOTE(list, hint, flags) knote(list, hint, flags) #define KNOTE_LOCKED(list, hint) knote(list, hint, KNF_LISTLOCKED) #define KNOTE_UNLOCKED(list, hint) knote(list, hint, 0) #define KNLIST_EMPTY(list) SLIST_EMPTY(&(list)->kl_list) -/* - * Flag indicating hint is a signal. Used by EVFILT_SIGNAL, and also - * shared by EVFILT_PROC (all knotes attached to p->p_klist) - */ -#define NOTE_SIGNAL 0x08000000 - /* * Hint values for the optional f_touch event filter. If f_touch is not set * to NULL and f_isfd is zero the f_touch filter will be called with the type * argument set to EVENT_REGISTER during a kevent() system call. It is also * called under the same conditions with the type argument set to EVENT_PROCESS * when the event has been triggered. */ #define EVENT_REGISTER 1 #define EVENT_PROCESS 2 struct kinfo_knote; struct proc; struct filterops { int f_isfd; /* true if ident == filedescriptor */ int (*f_attach)(struct knote *kn); void (*f_detach)(struct knote *kn); int (*f_event)(struct knote *kn, long hint); void (*f_touch)(struct knote *kn, struct kevent *kev, u_long type); int (*f_userdump)(struct proc *p, struct knote *kn, struct kinfo_knote *kin); }; /* * An in-flux knote cannot be dropped from its kq while the kq is * unlocked. If the KN_SCAN flag is not set, a thread can only set * kn_influx when it is exclusive owner of the knote state, and can * modify kn_status as if it had the KQ lock. KN_SCAN must not be set * on a knote which is already in flux. * * kn_sfflags, kn_sdata, and kn_kevent are protected by the knlist lock. */ struct knote { SLIST_ENTRY(knote) kn_link; /* for kq */ SLIST_ENTRY(knote) kn_selnext; /* for struct selinfo */ struct knlist *kn_knlist; /* f_attach populated */ TAILQ_ENTRY(knote) kn_tqe; struct kqueue *kn_kq; /* which queue we are on */ struct kevent kn_kevent; void *kn_hook; int kn_hookid; int kn_status; /* protected by kq lock */ #define KN_ACTIVE 0x01 /* event has been triggered */ #define KN_QUEUED 0x02 /* event is on queue */ #define KN_DISABLED 0x04 /* event is disabled */ #define KN_DETACHED 0x08 /* knote is detached */ #define KN_MARKER 0x20 /* ignore this knote */ #define KN_KQUEUE 0x40 /* this knote belongs to a kq */ #define KN_SCAN 0x100 /* flux set in kqueue_scan() */ int kn_influx; int kn_sfflags; /* saved filter flags */ int64_t kn_sdata; /* saved data field */ union { struct file *p_fp; /* file data pointer */ struct proc *p_proc; /* proc pointer */ struct kaiocb *p_aio; /* AIO job pointer */ struct aioliojob *p_lio; /* LIO job pointer */ void *p_v; /* generic other pointer */ } kn_ptr; const struct filterops *kn_fop; #define kn_id kn_kevent.ident #define kn_filter kn_kevent.filter #define kn_flags kn_kevent.flags #define kn_fflags kn_kevent.fflags #define kn_data kn_kevent.data #define kn_fp kn_ptr.p_fp }; struct kevent_copyops { void *arg; int (*k_copyout)(void *arg, struct kevent *kevp, int count); int (*k_copyin)(void *arg, struct kevent *kevp, int count); size_t kevent_size; }; struct thread; struct proc; struct knlist; struct mtx; struct rwlock; void knote(struct knlist *list, long hint, int lockflags); void knote_fork(struct knlist *list, int pid); struct knlist *knlist_alloc(struct mtx *lock); void knlist_detach(struct knlist *knl); void knlist_add(struct knlist *knl, struct knote *kn, int islocked); void knlist_remove(struct knlist *knl, struct knote *kn, int islocked); int knlist_empty(struct knlist *knl); void knlist_init(struct knlist *knl, void *lock, void (*kl_lock)(void *), void (*kl_unlock)(void *), void (*kl_assert_lock)(void *, int)); void knlist_init_mtx(struct knlist *knl, struct mtx *lock); void knlist_destroy(struct knlist *knl); void knlist_cleardel(struct knlist *knl, struct thread *td, int islocked, int killkn); #define knlist_clear(knl, islocked) \ knlist_cleardel((knl), NULL, (islocked), 0) #define knlist_delete(knl, td, islocked) \ knlist_cleardel((knl), (td), (islocked), 1) void knote_fdclose(struct thread *p, int fd); int kqfd_register(int fd, struct kevent *kev, struct thread *p, int mflag); int kqueue_add_filteropts(int filt, const struct filterops *filtops); int kqueue_del_filteropts(int filt); void kqueue_drain_schedtask(void); #else /* !_KERNEL */ #include struct timespec; __BEGIN_DECLS int kqueue(void); int kqueuex(unsigned flags); int kqueue1(int flags); int kevent(int kq, const struct kevent *changelist, int nchanges, struct kevent *eventlist, int nevents, const struct timespec *timeout); __END_DECLS #endif /* !_KERNEL */ #endif /* !_SYS_EVENT_H_ */