| 1 |
NAME |
| 2 |
Linux::AIO - linux-specific aio implemented using clone |
| 3 |
|
| 4 |
SYNOPSIS |
| 5 |
use Linux::AIO; |
| 6 |
|
| 7 |
DESCRIPTION |
| 8 |
This module implements asynchronous I/O using the means available to |
| 9 |
Linux - clone. It does not hook into the POSIX aio_* functions because |
| 10 |
Linux does not yet support these in the kernel (even as of 2.6.12, only |
| 11 |
O_DIRECT files are supported) and even if, it would only allow aio_read |
| 12 |
and write, not open, stat and so on. |
| 13 |
|
| 14 |
Instead, in this module a number of (non-posix) threads are started that |
| 15 |
execute your read/writes and signal their completion. You don't need |
| 16 |
thread support in your libc or perl, and the threads created by this |
| 17 |
module will not be visible to the pthreads library. |
| 18 |
|
| 19 |
NOTICE: the threads created by this module will automatically be killed |
| 20 |
when the thread calling min_parallel exits. Make sure you only ever call |
| 21 |
min_parallel from the same thread that loaded this module. |
| 22 |
|
| 23 |
Although the module will work with in the presence of other threads, it |
| 24 |
is not reentrant, so use appropriate locking yourself. |
| 25 |
|
| 26 |
API NOTES |
| 27 |
All the "aio_*" calls are more or less thin wrappers around the syscall |
| 28 |
with the same name (sans "aio_"). The arguments are similar or |
| 29 |
identical, and they all accept an additional $callback argument which |
| 30 |
must be a code reference. This code reference will get called with the |
| 31 |
syscall return code (e.g. most syscalls return -1 on error, unlike perl, |
| 32 |
which usually delivers "false") as it's sole argument when the given |
| 33 |
syscall has been executed asynchronously. |
| 34 |
|
| 35 |
All functions that expect a filehandle will also accept a file |
| 36 |
descriptor. |
| 37 |
|
| 38 |
Linux::AIO::min_parallel $nthreads |
| 39 |
Set the minimum number of AIO threads to $nthreads. The default is |
| 40 |
1, which means a single asynchronous operation can be done at one |
| 41 |
time (the number of outstanding operations, however, is unlimited). |
| 42 |
|
| 43 |
It is recommended to keep the number of threads low, as some linux |
| 44 |
kernel versions will scale negatively with the number of threads |
| 45 |
(higher parallelity => MUCH higher latency). |
| 46 |
|
| 47 |
Under normal circumstances you don't need to call this function, as |
| 48 |
this module automatically starts a single async thread. |
| 49 |
|
| 50 |
Linux::AIO::max_parallel $nthreads |
| 51 |
Sets the maximum number of AIO threads to $nthreads. If more than |
| 52 |
the specified number of threads are currently running, kill them. |
| 53 |
This function blocks until the limit is reached. |
| 54 |
|
| 55 |
This module automatically runs "max_parallel 0" at program end, to |
| 56 |
ensure that all threads are killed and that there are no outstanding |
| 57 |
requests. |
| 58 |
|
| 59 |
Under normal circumstances you don't need to call this function. |
| 60 |
|
| 61 |
$fileno = Linux::AIO::poll_fileno |
| 62 |
Return the *request result pipe filehandle*. This filehandle must be |
| 63 |
polled for reading by some mechanism outside this module (e.g. Event |
| 64 |
or select, see below). If the pipe becomes readable you have to call |
| 65 |
"poll_cb" to check the results. |
| 66 |
|
| 67 |
See "poll_cb" for an example. |
| 68 |
|
| 69 |
Linux::AIO::poll_cb |
| 70 |
Process all outstanding events on the result pipe. You have to call |
| 71 |
this regularly. Returns the number of events processed. Returns |
| 72 |
immediately when no events are outstanding. |
| 73 |
|
| 74 |
You can use Event to multiplex, e.g.: |
| 75 |
|
| 76 |
Event->io (fd => Linux::AIO::poll_fileno, |
| 77 |
poll => 'r', async => 1, |
| 78 |
cb => \&Linux::AIO::poll_cb); |
| 79 |
|
| 80 |
Linux::AIO::poll_wait |
| 81 |
Wait till the result filehandle becomes ready for reading (simply |
| 82 |
does a select on the filehandle. This is useful if you want to |
| 83 |
synchronously wait for some requests to finish). |
| 84 |
|
| 85 |
See "nreqs" for an example. |
| 86 |
|
| 87 |
Linux::AIO::nreqs |
| 88 |
Returns the number of requests currently outstanding. |
| 89 |
|
| 90 |
Example: wait till there are no outstanding requests anymore: |
| 91 |
|
| 92 |
Linux::AIO::poll_wait while Linux::AIO::nreqs; |
| 93 |
|
| 94 |
aio_open $pathname, $flags, $mode, $callback |
| 95 |
Asynchronously open or create a file and call the callback with the |
| 96 |
filedescriptor (NOT a perl filehandle, sorry for that, but watch |
| 97 |
out, this might change in the future). |
| 98 |
|
| 99 |
The $mode argument is a bitmask. See the "Fcntl" module for a list. |
| 100 |
They are the same as used in "sysopen". |
| 101 |
|
| 102 |
Example: |
| 103 |
|
| 104 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
| 105 |
if ($_[0] >= 0) { |
| 106 |
open my $fh, "<&$_[0]"; # create a copy for perl |
| 107 |
aio_close $_[0], sub { }; # close the aio handle |
| 108 |
print "open successful, fh is $fh\n"; |
| 109 |
... |
| 110 |
} else { |
| 111 |
die "open failed: $!\n"; |
| 112 |
} |
| 113 |
}; |
| 114 |
|
| 115 |
aio_close $fh, $callback |
| 116 |
Asynchronously close a file and call the callback with the result |
| 117 |
code. |
| 118 |
|
| 119 |
aio_read $fh,$offset,$length, $data,$dataoffset,$callback |
| 120 |
aio_write $fh,$offset,$length, $data,$dataoffset,$callback |
| 121 |
Reads or writes "length" bytes from the specified "fh" and "offset" |
| 122 |
into the scalar given by "data" and offset "dataoffset" and calls |
| 123 |
the callback without the actual number of bytes read (or -1 on |
| 124 |
error, just like the syscall). |
| 125 |
|
| 126 |
Example: Read 15 bytes at offset 7 into scalar $buffer, strating at |
| 127 |
offset 0 within the scalar: |
| 128 |
|
| 129 |
aio_read $fh, 7, 15, $buffer, 0, sub { |
| 130 |
$_[0] >= 0 or die "read error: $!"; |
| 131 |
print "read <$buffer>\n"; |
| 132 |
}; |
| 133 |
|
| 134 |
aio_readahead $fh,$offset,$length, $callback |
| 135 |
Asynchronously reads the specified byte range into the page cache, |
| 136 |
using the "readahead" syscall. |
| 137 |
|
| 138 |
readahead() populates the page cache with data from a file so that |
| 139 |
subsequent reads from that file will not block on disk I/O. The |
| 140 |
$offset argument specifies the starting point from which data is to |
| 141 |
be read and $length specifies the number of bytes to be read. I/O is |
| 142 |
performed in whole pages, so that offset is effectively rounded down |
| 143 |
to a page boundary and bytes are read up to the next page boundary |
| 144 |
greater than or equal to (off-set+length). aio_readahead() does not |
| 145 |
read beyond the end of the file. The current file offset of the file |
| 146 |
is left unchanged. |
| 147 |
|
| 148 |
aio_stat $fh_or_path, $callback |
| 149 |
aio_lstat $fh, $callback |
| 150 |
Works like perl's "stat" or "lstat" in void context. The callback |
| 151 |
will be called after the stat and the results will be available |
| 152 |
using "stat _" or "-s _" etc... |
| 153 |
|
| 154 |
Currently, the stats are always 64-bit-stats, i.e. instead of |
| 155 |
returning an error when stat'ing a large file, the results will be |
| 156 |
silently truncated unless perl itself is compiled with large file |
| 157 |
support. |
| 158 |
|
| 159 |
Example: Print the length of /etc/passwd: |
| 160 |
|
| 161 |
aio_stat "/etc/passwd", sub { |
| 162 |
$_[0] and die "stat failed: $!"; |
| 163 |
print "size is ", -s _, "\n"; |
| 164 |
}; |
| 165 |
|
| 166 |
aio_unlink $pathname, $callback |
| 167 |
Asynchronously unlink (delete) a file and call the callback with the |
| 168 |
result code. |
| 169 |
|
| 170 |
aio_fsync $fh, $callback |
| 171 |
Asynchronously call fsync on the given filehandle and call the |
| 172 |
callback with the fsync result code. |
| 173 |
|
| 174 |
aio_fdatasync $fh, $callback |
| 175 |
Asynchronously call fdatasync on the given filehandle and call the |
| 176 |
callback with the fdatasync result code. |
| 177 |
|
| 178 |
BUGS |
| 179 |
This module has been extensively tested in a large and very busy |
| 180 |
webserver for many years now. |
| 181 |
|
| 182 |
- aio_open gives a fd, but all other functions expect a perl filehandle. |
| 183 |
|
| 184 |
SEE ALSO |
| 185 |
Coro. |
| 186 |
|
| 187 |
AUTHOR |
| 188 |
Marc Lehmann <schmorp@schmorp.de> |
| 189 |
http://home.schmorp.de/ |
| 190 |
|
