1 ==================== Changes in man-pages-3.51 ====================
3 Released: 2013-04-17, Munich
9 The following people contributed patches/fixes or (noted in brackets
10 in the changelog below) reports, notes, and ideas that have been
11 incorporated in changes in this release:
13 Andreas Jaeger <aj@suse.de>
14 Andrew Clayton <andrew@digital-domain.net>
15 Brian M. Carlson <sandals@crustytoothpaste.net>
16 Changhee Han <ch0.han@lge.com>
17 Cyril Hrubis <chrubis@suse.cz>
18 Damien Grassart <damien@grassart.com>
19 David Prévot <taffit@debian.org>
20 Denis Barbier <bouzim@gmail.com>
21 Jeff Moyer <jmoyer@redhat.com>
22 Krzysztof Konopko <krzysztof.konopko@gmail.com>
23 Kyle McMartin <kyle@mcmartin.ca>
24 Mark H Weaver <mhw@netris.org>
25 Michael Kerrisk <mtk.manpages@gmail.com>
26 Mike Frysinger <vapier@gentoo.org>
27 Nicolas Hillegeer <nicolas@hillegeer.com>
28 Pavel Emelyanov <xemul@parallels.com>
29 Peter Schiffer <pschiffe@redhat.com>
30 Radek Pazdera <rpazdera@redhat.com>
31 Ralph Loader <suckfish@ihug.co.nz>
32 Simon Paillard <spaillard@debian.org>
33 The Wanderer <wanderer@fastmail.fm>
35 Apologies if I missed anyone!
38 Newly documented interfaces in existing pages
39 ---------------------------------------------
41 sched_rr_get_interval.2
43 Document /proc/sys/kernel/sched_rr_timeslice_ms
47 Document /proc/[pid]/map_files directory
48 This directory was added in Linux v3.3 and provides info about
49 files being mmap-ed in a way very similar to how /proc/[pid]/fd
52 v2: Added examples of how links look like and noted dependency
53 on kernel config option CONFIG_CHECKPOINT_RESTORE.
55 Document /proc/sys/kernel/shm_rmid_forced
59 Document /proc/sys/kernel/cap_last_cap
67 Global fix: fix placement of word "only"
71 License headers: consistent format
75 Global fix: s/since kernel/since Linux/
77 Various System V IPC pages in Section 2
79 Add "System V" to .TH line and text
80 Make it clear that these pages relate to System V IPC,
84 Changes to individual pages
85 ---------------------------
88 Michael Kerrisk [The Wanderer]
89 Clarify RETURN VALUE for F_OK
90 See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=705293
94 Correct the description of behavior when 'seconds' is 0
97 Michael Kerrisk [Peter Schiffer]
98 Add prototype for syscall to SYNOPSIS
99 And further clarify the distinction between the system call
100 and the wrapper function in the introductory text.
102 Update feature test macro requirements
103 The requirements quietly changed changed in glibc 2.14
105 See also http://www.sourceware.org/bugzilla/show_bug.cgi?id=4749
106 Michael Kerrisk [Mike Frysinger]
107 Clarify differences between clone2() syscall and wrapper function
108 Michael Kerrisk [Mike Frysinger]
109 Note those architectures where the sys_clone argument order differs
110 Michael Kerrisk [Mike Frysinger]
111 Add short subsection noting that blackfin, m68k, and sparc are different
113 Move clone2() text to subsection in description
114 The description of ia64 clone2() should follow the discussion
115 of the raw system call interface.
117 Change subhead for ia64 discussion
121 Recommend that 'tcache' should be specified as NULL nowadays
124 Jeff Moyer, Michael Kerrisk [Cyril Hrubis]
130 The description was rather vague, citing a "list of I/O contexts"
131 and stating that it "can" cancel outstanding requests. This
132 update makes things more concrete so that the reader knows exactly
137 The 'timeout' argument is not updated
138 I looked back through the kernel code, and the timeout was
139 never updated in any case. I've submitted a patch upstream
140 to change the comment above io_getevents.
145 nr_events is technically the number of completion events that can
146 be stored in the completion ring. The wording of the man page:
147 "capable of receiving at least nr_events" seems dubious to me,
148 only because I worry that folks might interpret that to mean
149 'nr_events' total, instead of 'nr_events' concurrently.
151 Further, I've added information on where to find the per-user
152 limit on 'nr_events', /proc/sys/fs/aio-max-nr. Let me know if
153 you think that is not relevant.
157 Explain use of 'size' argument
160 Michael Kerrisk [Andreas Jaeger]
161 _GNU_SOURCE must be defined to get SEEK_DATE and SEEK_HOLE definitions
162 See http://sourceware.org/bugzilla/show_bug.cgi?id=15312
166 Add pointers to relevant /proc files described in proc(5)
174 Refer to syscall(2) for ABI semantics on certain 32-bit architectures
175 Also: in sync_file_range.2 and posix_fadvise.2 remove description
176 of conventional calling signature as flawed, and in
177 posix_fadvise.2, de-emphasize focus on ARM, and rather phrase
178 as a more general discussion of certain architectures.
182 readdir(2) doesn't exist on x86-64
186 Clarify the discussion of 'semadj'
190 Refer to proc(5) for description of /proc/sys/kernel/shm_rmid_forced
194 Add notes that caution users when passing arguments to syscall()
195 For example, passing 'long long' on ARM-32 requires special
197 Mike Frysinger [Michael Kerrisk]
198 Document the exact calling convention for architecture system calls
199 Mike Frysinger [Kyle McMartin]
200 Add PA-RISC details under calling conventions
201 Michael Kerrisk [Mike Frysinger]
202 Refine discussion of ARM and other ABIs
206 Update kernel version number at start of list
214 Update feature test macro requirements
215 The requirements quietly changed changed in glibc 2.14
217 See also http://www.sourceware.org/bugzilla/show_bug.cgi?id=4749
220 Michael Kerrisk [Ralph Loader]
221 Correct definition of cookie_io_functions_t
225 The thread argument is passed in by value
231 Eliminate the implication that these functions deal with "offsets"
232 The directory position dealt with by the readdir() and
233 friends is not a simple file offset in modern file systems.
234 Typically, it is some kind of cookie value. Add text and
235 make other changes to these pages to eliminate the
236 implication that this is an offset, and warn the reader
237 that directory positions should be treated strictly as
240 In the process, rename the 'offset' argument of seekdir(3)
241 to 'loc', and add some text to readdir(3) to note that
242 the 'd_off' field is the same value returned by telldir(3)
243 at the current directory position.
245 See also https://lwn.net/Articles/544298/
249 Fix prototypes for scalbf() and scalbl()
253 Update feature test macro requirements
254 The requirements quietly changed changed in glibc 2.14
256 See also http://www.sourceware.org/bugzilla/show_bug.cgi?id=4749
259 Michael Kerrisk [Nicolas Hillegeer]
260 Add note on the behavior when 'usecs' is zero
261 POSIX.1-2001 does not specify the behavior in this case
262 and no other system that I checked documented the behavior.
263 Probably, most or all systems do what Linux does in this
264 case: cancel any pending alarm, just as alarm(0) does.
265 Add that info in NOTES.
269 Add byte positions for all EI_xxx fields
270 When describing e_ident, most of the EI_xxx defines mention the
271 exact byte number. This is useful when manually hacking an ELF
272 with a hex editor. However, the last few fields don't do this,
273 which means you have to count things up yourself.
274 Add a single word to each so you don't have to do that.
278 Refer to sched_rr_get_interval(2) for info on sched_rr_timeslice_ms
279 Since Linux 3.9, /proc/sys/kernel/sched_rr_timeslice_ms can
280 be used to change the SCHED_RR quantum.
282 SEE ALSO: Add sysctl(8)
284 Simplify the example of printing out environ
285 The binutils package contains a very handy utility to
286 print out null-byte delimited strings from a file. This
287 can replace a rather complex expression with cat(1)
288 provided as an example for printing out /proc/[pid]/environ.
290 Update /proc/PID/maps example
291 Update to 64-bit example that includes "[heap]", "[stack],
294 Formatting fixes for /proc/PID/maps
296 Document the "pathname" field of /proc/PID/maps
298 Add reference to capabilities(7) for /proc/sys/kernel/cap_last_cap
300 /proc/PID/maps: add a reference to mmap(2)
304 Document IP_MULTICAST_ALL
305 This commit adds documentation for the IP_MULTICAST_ALL socket
308 The option was added to the Linux kernel in 2.6.31:
310 Author Nivedita Singhvi <niv@us.ibm.com>
311 Commit f771bef98004d9d141b085d987a77d06669d4f4f
313 The description is based on a previous one [3] posted by the
314 original author of the code -- Nivedita, but it is slightly
317 I tested it myself and it works as described.
320 [1] http://lxr.free-electrons.com/source/net/ipv4/ip_sockglue.c#L972
321 [2] http://lxr.free-electrons.com/source/net/ipv4/igmp.c#L2267
322 [3] http://patchwork.ozlabs.org/patch/28902/
326 units should use an actual µ
327 The units(7) man page uses an ASCII u in place of the actual Greek
328 letter mu. Since we're in the twenty-first century, with
329 UTF-8-compatible terminals and terminal emulators, we should use
330 the actual letter µ instead of an ASCII approximation.
332 See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=704787