{
    "mode": "perldoc",
    "parameter": "fork",
    "section": "",
    "url": "https://www.chedong.com/phpMan.php/perldoc/fork/json",
    "generated": "2026-06-03T04:24:24Z",
    "synopsis": "NOTE:  As of the 5.8.0 release, fork() emulation has considerably\nmatured.  However, there are still a few known bugs and differences\nfrom real fork() that might affect you.  See the \"BUGS\" and\n\"CAVEATS AND LIMITATIONS\" sections below.\nPerl provides a fork() keyword that corresponds to the Unix system call of the same name. On\nmost Unix-like platforms where the fork() system call is available, Perl's fork() simply calls\nit.\nOn some platforms such as Windows where the fork() system call is not available, Perl can be\nbuilt to emulate fork() at the interpreter level. While the emulation is designed to be as\ncompatible as possible with the real fork() at the level of the Perl program, there are certain\nimportant differences that stem from the fact that all the pseudo child \"processes\" created this\nway live in the same real process as far as the operating system is concerned.\nThis document provides a general overview of the capabilities and limitations of the fork()\nemulation. Note that the issues discussed here are not applicable to platforms where a real",
    "sections": {
        "NAME": {
            "content": "perlfork - Perl's fork() emulation\n",
            "subsections": []
        },
        "SYNOPSIS": {
            "content": "NOTE:  As of the 5.8.0 release, fork() emulation has considerably\nmatured.  However, there are still a few known bugs and differences\nfrom real fork() that might affect you.  See the \"BUGS\" and\n\"CAVEATS AND LIMITATIONS\" sections below.\n\nPerl provides a fork() keyword that corresponds to the Unix system call of the same name. On\nmost Unix-like platforms where the fork() system call is available, Perl's fork() simply calls\nit.\n\nOn some platforms such as Windows where the fork() system call is not available, Perl can be\nbuilt to emulate fork() at the interpreter level. While the emulation is designed to be as\ncompatible as possible with the real fork() at the level of the Perl program, there are certain\nimportant differences that stem from the fact that all the pseudo child \"processes\" created this\nway live in the same real process as far as the operating system is concerned.\n\nThis document provides a general overview of the capabilities and limitations of the fork()\nemulation. Note that the issues discussed here are not applicable to platforms where a real",
            "subsections": [
                {
                    "name": "fork",
                    "content": ""
                }
            ]
        },
        "DESCRIPTION": {
            "content": "The fork() emulation is implemented at the level of the Perl interpreter. What this means in\ngeneral is that running fork() will actually clone the running interpreter and all its state,\nand run the cloned interpreter in a separate thread, beginning execution in the new thread just\nafter the point where the fork() was called in the parent. We will refer to the thread that\nimplements this child \"process\" as the pseudo-process.\n\nTo the Perl program that called fork(), all this is designed to be transparent. The parent\nreturns from the fork() with a pseudo-process ID that can be subsequently used in any\nprocess-manipulation functions; the child returns from the fork() with a value of 0 to signify\nthat it is the child pseudo-process.\n",
            "subsections": [
                {
                    "name": "Behavior of other Perl features in forked pseudo-processes",
                    "content": "Most Perl features behave in a natural way within pseudo-processes.\n\n$$ or $PROCESSID\nThis special variable is correctly set to the pseudo-process ID. It can be used to\nidentify pseudo-processes within a particular session. Note that this value is subject\nto recycling if any pseudo-processes are launched after others have been wait()-ed on.\n\n%ENV    Each pseudo-process maintains its own virtual environment. Modifications to %ENV affect\nthe virtual environment, and are only visible within that pseudo-process, and in any\nprocesses (or pseudo-processes) launched from it.\n"
                },
                {
                    "name": "chdir",
                    "content": "Each pseudo-process maintains its own virtual idea of the current directory.\nModifications to the current directory using chdir() are only visible within that\npseudo-process, and in any processes (or pseudo-processes) launched from it. All file\nand directory accesses from the pseudo-process will correctly map the virtual working\ndirectory to the real working directory appropriately.\n"
                },
                {
                    "name": "wait",
                    "content": "wait() and waitpid() can be passed a pseudo-process ID returned by fork(). These calls\nwill properly wait for the termination of the pseudo-process and return its status.\n"
                },
                {
                    "name": "kill",
                    "content": "returned by fork(). The outcome of kill on a pseudo-process is unpredictable and it\nshould not be used except under dire circumstances, because the operating system may not\nguarantee integrity of the process resources when a running thread is terminated. The\nprocess which implements the pseudo-processes can be blocked and the Perl interpreter\nhangs. Note that using \"kill('KILL', ...)\" on a pseudo-process() may typically cause\nmemory leaks, because the thread that implements the pseudo-process does not get a\nchance to clean up its resources.\n\n\"kill('TERM', ...)\" can also be used on pseudo-processes, but the signal will not be\ndelivered while the pseudo-process is blocked by a system call, e.g. waiting for a\nsocket to connect, or trying to read from a socket with no data available. Starting in\nPerl 5.14 the parent process will not wait for children to exit once they have been\nsignalled with \"kill('TERM', ...)\" to avoid deadlock during process exit. You will have\nto explicitly call waitpid() to make sure the child has time to clean-up itself, but you\nare then also responsible that the child is not blocking on I/O either.\n"
                },
                {
                    "name": "exec",
                    "content": "separate process and waits for it to complete before exiting with the same exit status\nas that process. This means that the process ID reported within the running executable\nwill be different from what the earlier Perl fork() might have returned. Similarly, any\nprocess manipulation functions applied to the ID returned by fork() will affect the\nwaiting pseudo-process that called exec(), not the real process it is waiting for after\nthe exec().\n\nWhen exec() is called inside a pseudo-process then DESTROY methods and END blocks will\nstill be called after the external process returns.\n"
                },
                {
                    "name": "exit",
                    "content": "for any outstanding child pseudo-processes. Note that this means that the process as a\nwhole will not exit unless all running pseudo-processes have exited. See below for some\nlimitations with open filehandles.\n\nOpen handles to files, directories and network sockets\nAll open handles are dup()-ed in pseudo-processes, so that closing any handles in one\nprocess does not affect the others. See below for some limitations.\n"
                },
                {
                    "name": "Resource limits",
                    "content": "In the eyes of the operating system, pseudo-processes created via the fork() emulation are\nsimply threads in the same process. This means that any process-level limits imposed by the\noperating system apply to all pseudo-processes taken together. This includes any limits imposed\nby the operating system on the number of open file, directory and socket handles, limits on disk\nspace usage, limits on memory size, limits on CPU utilization etc.\n"
                },
                {
                    "name": "Killing the parent process",
                    "content": "If the parent process is killed (either using Perl's kill() builtin, or using some external\nmeans) all the pseudo-processes are killed as well, and the whole process exits.\n"
                },
                {
                    "name": "Lifetime of the parent process and pseudo-processes",
                    "content": "During the normal course of events, the parent process and every pseudo-process started by it\nwill wait for their respective pseudo-children to complete before they exit. This means that the\nparent and every pseudo-child created by it that is also a pseudo-parent will only exit after\ntheir pseudo-children have exited.\n\nStarting with Perl 5.14 a parent will not wait() automatically for any child that has been\nsignalled with \"kill('TERM', ...)\" to avoid a deadlock in case the child is blocking on I/O and\nnever receives the signal.\n"
                }
            ]
        },
        "CAVEATS AND LIMITATIONS": {
            "content": "BEGIN blocks\nThe fork() emulation will not work entirely correctly when called from within a BEGIN\nblock. The forked copy will run the contents of the BEGIN block, but will not continue\nparsing the source stream after the BEGIN block. For example, consider the following\ncode:\n\nBEGIN {\nfork and exit;          # fork child and exit the parent\nprint \"inner\\n\";\n}\nprint \"outer\\n\";\n\nThis will print:\n\ninner\n\nrather than the expected:\n\ninner\nouter\n\nThis limitation arises from fundamental technical difficulties in cloning and restarting\nthe stacks used by the Perl parser in the middle of a parse.\n\nOpen filehandles\nAny filehandles open at the time of the fork() will be dup()-ed. Thus, the files can be\nclosed independently in the parent and child, but beware that the dup()-ed handles will\nstill share the same seek pointer. Changing the seek position in the parent will change\nit in the child and vice-versa. One can avoid this by opening files that need distinct\nseek pointers separately in the child.\n\nOn some operating systems, notably Solaris and Unixware, calling \"exit()\" from a child\nprocess will flush and close open filehandles in the parent, thereby corrupting the\nfilehandles. On these systems, calling \"exit()\" is suggested instead. \"exit()\" is\navailable in Perl through the \"POSIX\" module. Please consult your system's manpages for\nmore information on this.\n\nOpen directory handles\nPerl will completely read from all open directory handles until they reach the end of\nthe stream. It will then seekdir() back to the original location and all future\nreaddir() requests will be fulfilled from the cache buffer. That means that neither the\ndirectory handle held by the parent process nor the one held by the child process will\nsee any changes made to the directory after the fork() call.\n\nNote that rewinddir() has a similar limitation on Windows and will not force readdir()\nto read the directory again either. Only a newly opened directory handle will reflect\nchanges to the directory.\n\nForking pipe open() not yet implemented\nThe \"open(FOO, \"|-\")\" and \"open(BAR, \"-|\")\" constructs are not yet implemented. This\nlimitation can be easily worked around in new code by creating a pipe explicitly. The\nfollowing example shows how to write to a forked child:\n\n# simulate open(FOO, \"|-\")\nsub pipetofork ($) {\nmy $parent = shift;\npipe my $child, $parent or die;\nmy $pid = fork();\ndie \"fork() failed: $!\" unless defined $pid;\nif ($pid) {\nclose $child;\n}\nelse {\nclose $parent;\nopen(STDIN, \"<&=\" . fileno($child)) or die;\n}\n$pid;\n}\n\nif (pipetofork('FOO')) {\n# parent\nprint FOO \"pipetofork\\n\";\nclose FOO;\n}\nelse {\n# child\nwhile (<STDIN>) { print; }\nexit(0);\n}\n\nAnd this one reads from the child:\n\n# simulate open(FOO, \"-|\")\nsub pipefromfork ($) {\nmy $parent = shift;\npipe $parent, my $child or die;\nmy $pid = fork();\ndie \"fork() failed: $!\" unless defined $pid;\nif ($pid) {\nclose $child;\n}\nelse {\nclose $parent;\nopen(STDOUT, \">&=\" . fileno($child)) or die;\n}\n$pid;\n}\n\nif (pipefromfork('BAR')) {\n# parent\nwhile (<BAR>) { print; }\nclose BAR;\n}\nelse {\n# child\nprint \"pipefromfork\\n\";\nexit(0);\n}\n\nForking pipe open() constructs will be supported in future.\n\nGlobal state maintained by XSUBs\nExternal subroutines (XSUBs) that maintain their own global state may not work\ncorrectly. Such XSUBs will either need to maintain locks to protect simultaneous access\nto global data from different pseudo-processes, or maintain all their state on the Perl\nsymbol table, which is copied naturally when fork() is called. A callback mechanism that\nprovides extensions an opportunity to clone their state will be provided in the near\nfuture.\n\nInterpreter embedded in larger application\nThe fork() emulation may not behave as expected when it is executed in an application\nwhich embeds a Perl interpreter and calls Perl APIs that can evaluate bits of Perl code.\nThis stems from the fact that the emulation only has knowledge about the Perl\ninterpreter's own data structures and knows nothing about the containing application's\nstate. For example, any state carried on the application's own call stack is out of\nreach.\n\nThread-safety of extensions\nSince the fork() emulation runs code in multiple threads, extensions calling into\nnon-thread-safe libraries may not work reliably when calling fork(). As Perl's threading\nsupport gradually becomes more widely adopted even on platforms with a native fork(),\nsuch extensions are expected to be fixed for thread-safety.\n",
            "subsections": []
        },
        "PORTABILITY CAVEATS": {
            "content": "In portable Perl code, \"kill(9, $child)\" must not be used on forked processes. Killing a forked\nprocess is unsafe and has unpredictable results. See \"kill()\", above.\n",
            "subsections": []
        },
        "BUGS": {
            "content": "*       Having pseudo-process IDs be negative integers breaks down for the integer -1 because\nthe wait() and waitpid() functions treat this number as being special. The tacit\nassumption in the current implementation is that the system never allocates a thread ID\nof 1 for user threads. A better representation for pseudo-process IDs will be\nimplemented in future.\n\n*       In certain cases, the OS-level handles created by the pipe(), socket(), and accept()\noperators are apparently not duplicated accurately in pseudo-processes. This only\nhappens in some situations, but where it does happen, it may result in deadlocks between\nthe read and write ends of pipe handles, or inability to send or receive data across\nsocket handles.\n\n*       This document may be incomplete in some respects.\n",
            "subsections": []
        },
        "AUTHOR": {
            "content": "Support for concurrent interpreters and the fork() emulation was implemented by ActiveState,\nwith funding from Microsoft Corporation.\n\nThis document is authored and maintained by Gurusamy Sarathy <gsar@activestate.com>.\n",
            "subsections": []
        },
        "SEE ALSO": {
            "content": "\"fork\" in perlfunc, perlipc\n",
            "subsections": []
        }
    },
    "summary": "perlfork - Perl's fork() emulation",
    "flags": [],
    "examples": [],
    "see_also": []
}