-==================== Changes in man-pages-3.64 ====================
+==================== Changes in man-pages-4.07 ====================
-Released: 2014-04-06, Munich
+Released: ????-??-??, Munich
Contributors
in the changelog below) reports, notes, and ideas that have been
incorporated in changes in this release:
-Abhay Sachan <sachan.abhay@gmail.com>
-Alexey Samsonov <samsonov@google.com>
-Andrey Vagin <avagin@openvz.org>
-Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
-Christoph Hellwig <hch@infradead.org>
-David Prévot <taffit@debian.org>
-Eric Dumazet <edumazet@google.com>
-Eric W. Biederman <ebiederm@xmission.com>
-Jan Kara <jack@suse.cz>
-Kir Kolyshkin <kir@openvz.org>
-Michael Kerrisk <mtk.manpages@gmail.com>
-Mike Frysinger <vapier@gentoo.org>
-NeilBrown <neilb@suse.de>
-Peng Haitao <penght@cn.fujitsu.com>
-Peter Hurley <peter@hurleysoftware.com>
-Petr Gajdos <pgajdos@suse.cz>
-Robert P. J. Day <rpjday@crashcourse.ca>
-Vince Weaver <vincent.weaver@maine.edu>
-Yuri Kozlov <yuray@komyakino.ru>
Apologies if I missed anyone!
New and rewritten pages
-----------------------
-open_by_handle_at.2
- Michael Kerrisk [Mike Frysinger, Neil Brown, Aneesh Kumar K.V,
- Christoph Hellwig]
- New page describing name_to_handle_at(2) and open_by_handle_at(2)
-
-inotify.7
- Michael Kerrisk
- Rewrite introductory section
- Reorganize "Limitations and caveats" subsection
- Michael Kerrisk
- Further describe the race when adding a watch to a new subtree
- Michael Kerrisk
- Directory renames may invalidate multiple paths cached by application
- Michael Kerrisk
- Add paragraph on cache consistency checking
- Michael Kerrisk
- Mention cache rebuilding to handle overflow events
- Michael Kerrisk
- Moving an object to another filesystem generates IN_DELETE_SELF
- Michael Kerrisk [Jan Kara]
- Add text on dealing with rename() events
- Michael Kerrisk
- Note rationale and consequences of event coalescing
- Michael Kerrisk [Eric W. Biederman]
- Inotify doesn't work for remote and pseudo filesystems
- Michael Kerrisk
- Add some examples of events generated by various system calls
- Michael Kerrisk
- BUGS: IN_ONESHOT does now cause IN_IGNORED when the watch is dropped
- A silent change as a result of the fanotify work in kernel 2.6.36.
- Michael Kerrisk
- Note that IN_DELETE_SELF will be followed by IN_IGNORED
- Michael Kerrisk
- Note that IN_UNMOUNT will be followed by an IN_IGNORED event
- Michael Kerrisk
- Inotify does not report events for mmap(2) and msync(2)
- Michael Kerrisk
- Add examples of syscalls that trigger IN_ATTRIB
- Michael Kerrisk
- Add some examples of syscalls that trigger IN_MODIFY
- Michael Kerrisk
- execve(2) also generates IN_ACCESS
- Michael Kerrisk
- Add examples of syscalls that trigger IN_CREATE
-
Newly documented interfaces in existing pages
---------------------------------------------
-perf_event_open.2
- Vince Weaver [Michael Kerrisk]
- Document the PERF_FLAG_FD_CLOEXEC flag
- The Linux 3.14 release adds support for the PERF_FLAG_FD_CLOEXEC
- flag.
-
-feature_test_macros.7
- Michael Kerrisk
- Document _LARGEFILE_SOURCE
-
New and changed links
---------------------
-name_to_handle_at.2
- Michael Kerrisk
- New link to new open_by_name_at(2) page
-
Global changes
--------------
-fmemopen.3
-getaddrinfo.3
-mq_notify.3
-offsetof.3
-aio.7
- Michael Kerrisk
- Print size_t/ssize_t values using %z printf() modifier
- There are fewer and fewer systems these days that don't
- support the %z specifier mandated in C99. So replace the
- use of %ld/%lu + (long) cast with %zd/%zu.
-
Changes to individual pages
---------------------------
-bdflush.2
-fsync.2
-sync.2
- Kir Kolyshkin
- SEE ALSO: remove update(8) reference
- Remove reference to update(8) man page, since there is no such
- page. This is an ancient BSD leftover I believe.
-
-chown.2
- Michael Kerrisk
- Note that 'dirfd' can be AT_FDCWD when AT_EMPTY_PATH is used
-
-getxattr.2
- Abhay Sachan
- Fix REURN VALUE description
- A EA can have length zero.
-
-inotify_add_watch.2
- Michael Kerrisk
- ERRORS: add ENAMETOOLONG
-
-inotify_init.2
- Michael Kerrisk
- Add pointer to inotify(7)
-
-link.2
- Michael Kerrisk
- When AT_EMPTY_PATH is specified, 'olddirfd' must not refer to a
- directory
-
-mmap.2
- Andrey Vagin
- The file descriptor for a file mapping must be readable
- There is no difference between MAP_SHARED and MAP_PRIVATE.
-
-open.2
- Michael Kerrisk
- SEE ALSO: add open_by_name_at(2)
-
- Various clean-ups and bug fixes after review by Mike Frysinger.
-
-perf_event_open.2
- Vince Weaver
- Document PERF_EVENT_IOC_PERIOD behavior change
- Linux 3.14 (in commit bad7192b842c83e580747ca57104dd51fe08c223)
- changes the perf_event PERF_EVENT_IOC_PERIOD ioctl() behavior
- on all architectures to update immediately, to match the behavior
- found on ARM.
-
-stat.2
- Michael Kerrisk
- Note that 'dirfd' can be AT_FDCWD when AT_EMPTY_PATH is used
-
-syscalls.2
- Michael Kerrisk
- Add sched_getattr() and sched_setattr()
- And update kernel version to 3.14
-
-abort.3
- Peng Haitao
- ATTRIBUTES: Note function that is thread-safe
- The function abort() is thread safe.
-
-confstr.3
- Michael Kerrisk
- SEE ALSO: add getconf(1), fpathconf(3), sysconf(3), pathconf(3)
-
-exit.3
- Peng Haitao
- ATTRIBUTES: Note function that is not thread-safe
- The function exit() is not thread safe.
-
-fenv.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions feclearexcept(), fegetexceptflag(), feraiseexcept(),
- fesetexceptflag(), fetestexcept(), fegetround(), fesetround(),
- fegetenv(), feholdexcept(), fesetenv(), feupdateenv(),
- feenableexcept(), fedisableexcept() and fegetexcept() are thread
- safe.
-
-fpathconf.3
- Michael Kerrisk
- SEE ALSO: add confstr(3)
-
-fseek.3
- Michael Kerrisk [Petr Gajdos]
- Document EINVAL error for negative file offset
-
-fseeko.3
- Michael Kerrisk
- Add feature test macro requirements
-
-fts.3
- Michael Kerrisk [Mike Frysinger]
- Remove mention of "32-bit systems" in BUGS
-
-fwide.3
-wprintf.3
- Michael Kerrisk [Robert P. J. Day]
- Remove mention of bogus _ISOC95_SOURCE feature test macro
-
-getline.3
- Alexey Samsonov
- Caller should free the allocated buffer even if getline() failed
- Relevant discussion in glibc bugzilla:
- https://sourceware.org/bugzilla/show_bug.cgi?id=5666
-
-getloadavg.3
- Peng Haitao
- ATTRIBUTES: Note function that is thread-safe
- The function getloadavg() is thread safe.
-
-getpt.3
- Peng Haitao
- ATTRIBUTES: Note function that is thread-safe
- The function getpt() is thread safe.
-
-if_nametoindex.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions if_nametoindex() and if_indextoname() are thread safe.
-
-index.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions index() and rindex() are thread safe.
-
-mkfifo.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions mkfifo() and mkfifoat() are thread safe.
-
-netlink.3
- Michael Kerrisk
- SEE ALSO: make the reference for libnetlink the libnetlink(3) man page
-
-random.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions random(), srandom(), initstate() and setstate()
- are thread safe.
-
-random_r.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions random_r(), srandom_r(), initstate_r() and
- setstate_r() are thread safe.
-
-sigvec.3
- Peng Haitao
- ATTRIBUTES: Note functions that are thread-safe
- The functions sigvec(), sigblock(), sigsetmask() and
- siggetmask() are thread safe.
-
- The macro sigmask() is thread safe.
-
-sysconf.3
- Michael Kerrisk
- SEE ALSO: add confstr(3)
-
-termios.3
- Michael Kerrisk [Peter Hurley]
- Fix error in discussion of MIN > 0, TIME == 0 noncanonical mode
- As reported by Peter Hurley, for the MIN > 0, TIME == 0 case:
-
- read() may unblock when MIN bytes are available but return
- up to the 'count' parameter if more input arrives in between
- waking and copying into the user buffer.
- ...
- read() may also _not_ return until MIN bytes have been
- received, even if 'count' bytes have been received.
- Michael Kerrisk
- Add a note on interaction of O_NONBLOCK with noncanonical MIN/TIME
- POSIX leaves the behavior open.
- Michael Kerrisk
- Clarify termination conditions for MIN > 0, TIME > 0
- Michael Kerrisk
- Clarify behavior if data is available before noncanonical read()
- Michael Kerrisk
- Add descriptive titles to noncanonical read() cases
-
-symlink.7
- Michael Kerrisk
- Add subsection on opening a symbolic link with O_PATH
- Michael Kerrisk
- name_to_handle_at(2) and open_by_handle_at(2) optionally follow symlinks
- Michael Kerrisk
- Mention use of readlink(2) to read contents of a symlink
-
-tcp.7
- Michael Kerrisk [Eric Dumazet]
- Document /proc/sys/net/ipv4/tcp_autocorking
- Text heavily based on Documentation/networking/ip-sysctl.txt