… | |
… | |
436 | |
436 | |
437 | Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts |
437 | Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts |
438 | reading at byte offset C<$in_offset>, and starts writing at the current |
438 | reading at byte offset C<$in_offset>, and starts writing at the current |
439 | file offset of C<$out_fh>. Because of that, it is not safe to issue more |
439 | file offset of C<$out_fh>. Because of that, it is not safe to issue more |
440 | than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each |
440 | than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each |
441 | other. |
441 | other. The same C<$in_fh> works fine though, as this function does not |
|
|
442 | move or use the file offset of C<$in_fh>. |
442 | |
443 | |
443 | Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than |
444 | Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than |
444 | are written, and there is no way to find out how many bytes have been read |
445 | are written, and there is no way to find out how many more bytes have been |
445 | from C<aio_sendfile> alone, as C<aio_sendfile> only provides the number of |
446 | read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the |
446 | bytes written to C<$out_fh>. Only if the result value equals C<$length> |
447 | number of bytes written to C<$out_fh>. Only if the result value equals |
447 | one can assume that C<$length> bytes have been read. |
448 | C<$length> one can assume that C<$length> bytes have been read. |
448 | |
449 | |
449 | Unlike with other C<aio_> functions, it makes a lot of sense to use |
450 | Unlike with other C<aio_> functions, it makes a lot of sense to use |
450 | C<aio_sendfile> on non-blocking sockets, as long as one end (typically |
451 | C<aio_sendfile> on non-blocking sockets, as long as one end (typically |
451 | the C<$in_fh>) is a file - the file I/O will then be asynchronous, while |
452 | the C<$in_fh>) is a file - the file I/O will then be asynchronous, while |
452 | the socket I/O will be non-blocking. Note, however, that you can run into |
453 | the socket I/O will be non-blocking. Note, however, that you can run |
453 | a trap where C<aio_sendfile> reads some data with readahead, then fails |
454 | into a trap where C<aio_sendfile> reads some data with readahead, then |
454 | to write all data, and when the socket is ready the next time, the data |
455 | fails to write all data, and when the socket is ready the next time, the |
455 | in the cache is already lost, forcing C<aio_sendfile> to again hit the |
456 | data in the cache is already lost, forcing C<aio_sendfile> to again hit |
456 | disk. Explicit C<aio_read> + C<aio_write> let's you control resource usage |
457 | the disk. Explicit C<aio_read> + C<aio_write> let's you better control |
457 | much better. |
458 | resource usage. |
458 | |
459 | |
459 | This call tries to make use of a native C<sendfile> syscall to provide |
460 | This call tries to make use of a native C<sendfile>-like syscall to |
460 | zero-copy operation. For this to work, C<$out_fh> should refer to a |
461 | provide zero-copy operation. For this to work, C<$out_fh> should refer to |
461 | socket, and C<$in_fh> should refer to an mmap'able file. |
462 | a socket, and C<$in_fh> should refer to an mmap'able file. |
462 | |
463 | |
463 | If a native sendfile cannot be found or it fails with C<ENOSYS>, |
464 | If a native sendfile cannot be found or it fails with C<ENOSYS>, |
464 | C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or C<ENOTSOCK>, |
465 | C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or |
465 | it will be emulated, so you can call C<aio_sendfile> on any type of |
466 | C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any |
466 | filehandle regardless of the limitations of the operating system. |
467 | type of filehandle regardless of the limitations of the operating system. |
|
|
468 | |
|
|
469 | As native sendfile syscalls (as practically any non-POSIX interface hacked |
|
|
470 | together in a hurry to improve benchmark numbers) tend to be rather buggy |
|
|
471 | on many systems, this implementation tries to work around some known bugs |
|
|
472 | in Linux and FreeBSD kernels (probably others, too), but that might fail, |
|
|
473 | so you really really should check the return value of C<aio_sendfile> - |
|
|
474 | fewre bytes than expected might have been transferred. |
467 | |
475 | |
468 | |
476 | |
469 | =item aio_readahead $fh,$offset,$length, $callback->($retval) |
477 | =item aio_readahead $fh,$offset,$length, $callback->($retval) |
470 | |
478 | |
471 | C<aio_readahead> populates the page cache with data from a file so that |
479 | C<aio_readahead> populates the page cache with data from a file so that |
… | |
… | |
862 | if ($_[0] && $! == EXDEV) { |
870 | if ($_[0] && $! == EXDEV) { |
863 | aioreq_pri $pri; |
871 | aioreq_pri $pri; |
864 | add $grp aio_copy $src, $dst, sub { |
872 | add $grp aio_copy $src, $dst, sub { |
865 | $grp->result ($_[0]); |
873 | $grp->result ($_[0]); |
866 | |
874 | |
867 | if (!$_[0]) { |
875 | unless ($_[0]) { |
868 | aioreq_pri $pri; |
876 | aioreq_pri $pri; |
869 | add $grp aio_unlink $src; |
877 | add $grp aio_unlink $src; |
870 | } |
878 | } |
871 | }; |
879 | }; |
872 | } else { |
880 | } else { |