Changes

   1 ==================== Changes in man-pages-4.16 ====================
   2
   3 Released: ????-??-??, Munich
   4
   5
   6 Contributors
   7 ------------
   8
   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:
  12
  13 Adam Borowski <kilobyte@angband.pl>
  14 Andy Owen <andrew.owen@dolby.com>
  15 Carlos O'Donell <carlos@redhat.com>
  16 Carsten Grohmann <carstengrohmann@gmx.de>
  17 Elvira Khabirova <lineprinter@altlinux.org>
  18 Enrique Garcia <cquike@arcor.de>
  19 Frederic Brault <fbrault@xyalis.com>
  20 Heinrich Schuchardt <xypron.glpk@gmx.de>
  21 Howard Johnson <hwj@BridgeportContractor.com>
  22 Jakub Wilk <jwilk@jwilk.net>
  23 Jan Kara <jack@suse.cz>
  24 Jann Horn <jannh@google.com>
  25 John Hubbard <jhubbard@nvidia.com>
  26 Jürg Billeter <j@bitron.ch>
  27 Konstantin Grinemayer <cdlscpmv@gmail.com>
  28 Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
  29 Martin Mares <mj@ucw.cz>
  30 Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
  31 Mattias Andrée <maandree@kth.se>
  32 Michael Kerrisk <mtk.manpages@gmail.com>
  33 Michal Hocko <mhocko@suse.com>
  34 Mike Frysinger <vapier@gentoo.org>
  35 Nikos Mavrogiannopoulos <nmavrogi@redhat.com>
  36 Robin Kuzmin <kuzmin.robin@gmail.com>
  37 Ross Zwisler <ross.zwisler@linux.intel.com>
  38 Rusty Russell <rusty@rustcorp.com.au>
  39 Serge E. Hallyn <serge@hallyn.com>
  40 Song Liu <songliubraving@fb.com>
  41 Tomi Salminen <tsalminen@forcepoint.com>
  42
  43 Apologies if I missed anyone!
  44
  45
  46 Newly documented interfaces in existing pages
  47 ---------------------------------------------
  48
  49 membarrier.2
  50     Mathieu Desnoyers  [Michael Kerrisk]
  51         Document new membarrier commands introduced in Linux 4.16
  52             Document the following membarrier commands introduced in
  53             Linux 4.16:
  54
  55                 MEMBARRIER_CMD_GLOBAL_EXPEDITED
  56                     (the old enum label MEMBARRIER_CMD_SHARED is now an
  57                     alias to preserve header backward compatibility)
  58                 MEMBARRIER_CMD_REGISTER_GLOBAL_EXPEDITED
  59                 MEMBARRIER_CMD_PRIVATE_EXPEDITED_SYNC_CORE
  60                 MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_SYNC_CORE
  61
  62 mmap.2
  63     Jan Kara  [Ross Zwisler, Michael Kerrisk]
  64         Add description of MAP_SHARED_VALIDATE and MAP_SYNC
  65     Michal Hocko  [John Hubbard, Michael Kerrisk, Jann Horn]
  66         Document new MAP_FIXED_NOREPLACE flag
  67             4.17+ kernels offer a new MAP_FIXED_NOREPLACE flag which allows
  68             the caller to atomically probe for a given address range.
  69
  70 readv.2
  71 io_submit.2
  72     Jürg Billeter
  73         Document RWF_APPEND added in Linux 4.16
  74
  75 capabilities.7
  76     Michael Kerrisk
  77         Describe file capability versioning
  78     Michael Kerrisk  [Serge E. Hallyn]
  79         Document namespaced-file capabilities
  80         [There's still more work to be done on this new text]
  81
  82
  83 Changes to individual pages
  84 ---------------------------
  85
  86 bpf.2
  87     Michael Kerrisk
  88         Update list of architectures that support JITed eBPF
  89             And note kernel version numbers where support is added.
  90     Michael Kerrisk
  91         Kernel 4.15 added CONFIG_BPF_JIT_ALWAYS_ON
  92             This causes the JIT compiler to be always on and
  93             forces bpf_jit_enable to 1.
  94
  95 execve.2
  96     Michael Kerrisk
  97         Note that describing execve as "executing a new process" is misleading
  98             This misdescription is so common that it's worth calling it out
  99             explicitly.
 100     Michael Kerrisk
 101         More explicitly describe effect of execve() in the opening paragraph
 102
 103 fallocate.2
 104     Michael Kerrisk
 105         Since Linux 4.16, Btrfs supports FALLOC_FL_ZERO_RANGE
 106
 107 getrlimit.2
 108     Michael Kerrisk
 109         CAP_SYS_RESOURCE capability is required in *initial user namespace*
 110
 111 io_submit.2
 112     Michael Kerrisk
 113         Add kernel version numbers for various 'aio_rw_flags' flags
 114     Michael Kerrisk
 115         Place 'aio_rw_flags' in alphabetical order
 116
 117 mmap.2
 118     Jann Horn
 119         MAP_FIXED is okay if the address range has been reserved
 120             Clarify that MAP_FIXED is appropriate if the specified address
 121             range has been reserved using an existing mapping, but shouldn't
 122             be used otherwise.
 123     Michael Kerrisk
 124         Move the text on MAP_FIXED to NOTES
 125             This text has become rather long, making it it somewhat
 126             unwieldy in the discussion of the mmap() flags. Therefore,
 127             move it to NOTES, with a pointer in DESCRIPTION referring
 128             the reader to NOTES.
 129     Michael Kerrisk  [Heinrich Schuchardt]
 130         Clarify that when addr==NULL, address chosen by kernel is page-aligned
 131     Michael Kerrisk
 132         Add a little historical detail on the obsolete MAP_DENYWRITE
 133
 134 mount.2
 135     Michael Kerrisk
 136         ERRORS: add EBUSY for the case of trying to stack same mount twice
 137     Michael Kerrisk
 138         Remove a couple of obsolete EBUSY errors
 139             As far as I can tell, these EBUSY errors disappeared
 140             with the addition of stackable mounts in Linux 2.4.
 141
 142 msgget.2
 143 semget.2
 144 shmget.2
 145     Michael Kerrisk
 146         The purpose of "flags" == 0 is to obtain ID of an existing IPC object
 147             This was implied in these pages, but the meaning of "flags" == 0
 148             could be more explicit, as indicated by questions such as
 149             https://stackoverflow.com/questions/49833569/flag-value-of-semget-function
 150
 151 open.2
 152     Jann Horn
 153         Document more -ETXTBSY conditions
 154     Jakub Wilk
 155         Add missing argument for snprintf() in example code
 156
 157 perf_event_open.2
 158     Song Liu
 159         Add type kprobe and uprobe
 160             Two new types kprobe and uprobe are being added to
 161             perf_event_open(), which allow creating kprobe or
 162             uprobe with perf_event_open. This patch adds
 163             information about these types.
 164
 165 ptrace.2
 166     Jann Horn
 167         Copy retval info for SECCOMP_GET_FILTER to right section
 168             The "RETURN VALUE" section made a claim that was incorrect for
 169             PTRACE_SECCOMP_GET_FILTER. Explicitly describe the behavior of
 170             PTRACE_SECCOMP_GET_FILTER in the "RETURN VALUE" section (as
 171             usual), but leave the now duplicate description in the section
 172             describing PTRACE_SECCOMP_GET_FILTER, since the
 173             PTRACE_SECCOMP_GET_FILTER section would otherwise probably become
 174             harder to understand.
 175
 176 readv.2
 177     Michael Kerrisk
 178         Remove redundant sentence
 179
 180 seccomp.2
 181     Michael Kerrisk
 182         Note that execve() may change syscall numbers during life of process
 183             On a multiarch/multi-ABI platform such as modern x86, each
 184             architecture/ABI (x86-64, x32, i386)has its own syscall numbers,
 185             which means a seccomp() filter may see different syscall numbers
 186             over the life of the process if that process uses execve() to
 187             execute programs that has a different architectures/ABIs.
 188     Michael Kerrisk
 189         Note which architectures support seccomp BPF
 190     Michael Kerrisk
 191         In EXAMPLE, clearly note that x32 syscalls are >= X32_SYSCALL_BIT
 192
 193 shutdown.2
 194     Carsten Grohmann
 195         SEE ALSO: add close(2)
 196
 197 syscall.2
 198     Adam Borowski
 199         Add riscv
 200
 201 wait.2
 202     Michael Kerrisk  [Robin Kuzmin]
 203         wait() and waitpid() block the calling thread (not process)
 204
 205 wait4.2
 206     Michael Kerrisk  [Martin Mares]
 207         Soften the warning against the use of wait3()/wait4()
 208             These functions are nonstandard, but there is no replacement.
 209
 210             See https://bugzilla.kernel.org/show_bug.cgi?id=199215
 211
 212 crypt.3
 213 encrypt.3
 214     Carlos O'Donell  [Michael Kerrisk]
 215         Add notes about _XOPEN_CRYPT
 216             The distribution may choose not to support _XOPEN_CRYPT in the
 217             case that the distribution has transitioned from glibc crypt to
 218             libxcrypt.
 219
 220 fseek.3
 221     Michael Kerrisk  [Andy Owen]
 222         ERRORS: EBADF should be ESPIPE
 223     Michael Kerrisk
 224         Improve EPIPE error text
 225
 226 getcwd.3
 227     Carlos O'Donell
 228         Mention that "(unreachable)" is no longer returned for glibc >= 2.27.
 229
 230 makedev.3
 231     Michael Kerrisk
 232         Since glibc 2.28, <sys/types.h> no longer defines these macros
 233
 234 pthread_create.3
 235     Frederic Brault
 236         Note default thread stack size for several architectures
 237
 238 tsearch.3
 239     Jann Horn
 240         Clarify items vs nodes
 241             The manpage claimed that tsearch() returns a pointer to a data
 242             item. This is incorrect; tsearch() returns a pointer to the
 243             corresponding tree node, which can also be interpreted as a
 244             pointer to a pointer to the data item.
 245
 246             Since this API is quite unintuitive, also add a clarifying
 247             sentence.
 248     Jann Horn
 249         tdelete() can return dangling pointers
 250             POSIX says that deleting the root node must cause tdelete() to
 251             return some unspecified non-NULL pointer. Glibc implements it by
 252             returning a dangling pointer to the (freed) root node.
 253             Therefore, explicitly note that tdelete() may return bad pointers
 254             that must not be accessed.
 255
 256 elf.5
 257     Michael Kerrisk
 258         SEE ALSO: add patchelf(1)
 259
 260 filesystems.5
 261     Michael Kerrisk
 262         Add an entry for tmpfs(5)
 263
 264 group.5
 265     Michael Kerrisk
 266         SEE ALSO: add vigr(8)
 267
 268 passwd.5
 269     Michael Kerrisk
 270         SEE ALSO: add vipw(8)
 271
 272 sysfs.5
 273     Michael Kerrisk
 274         Add brief note on /sys/fs/smackfs
 275
 276 tmpfs.5
 277     Mike Frysinger
 278         Document current mount options
 279             Some of this content is moved from the mount(8) man page.
 280             Style was based on proc(5) sections.
 281     Michael Kerrisk
 282         Remove reference to mount(8) for discussion of mount options
 283             The mount options are now described in this page.
 284     Michael Kerrisk
 285         SEE ALSO: add Documentation/vm/transhuge.txt
 286     Michael Kerrisk
 287         Reformat 'huge' and 'mpol' mount option values as lists
 288     Michael Kerrisk
 289         Describe 'mpol' mount options
 290             Based on text from Documentation/filesystems/tmpfs.txt.
 291     Michael Kerrisk
 292         Document 'huge' mount options
 293             Based on text from Documentation/vm/transhuge.txt.
 294     Michael Kerrisk
 295         SEE ALSO: add set_mempolicy(2)
 296     Michael Kerrisk
 297         Document mpol=local mount option
 298
 299 capabilities.7
 300     Michael Kerrisk
 301         Remove redundant mention of PTRACE_SECCOMP_GET_FILTER
 302
 303 cgroups.7
 304     Michael Kerrisk
 305         cgroup.events transitions generate POLLERR as well as POLLPRI
 306
 307 mount_namespaces.7
 308     Michael Kerrisk
 309         Note another case where shared "peer groups" are formed
 310
 311 namespaces.7
 312     Michael Kerrisk  [Konstantin Khlebnikov]
 313         Mention that device ID should also be checked when comparing NS symlinks
 314             When comparing two namespaces symlinks to see if they refer to
 315             the same namespace, both the inode number and the device ID
 316             should be compared. This point was already made clear in
 317             ioctl_ns(2), but was missing from this page.
 318     Michael Kerrisk
 319         Note an idiosyncrasy of /proc/[pid]/ns/pid_for_children
 320             /proc/[pid]/ns/pid_for_children has a value only after first
 321             child is created in PID namespace. Verified by experiment.
 322
 323 network_namespaces.7
 324     Michael Kerrisk
 325         Network namespaces isolate the UNIX domain abstract socket namespace
 326     Michael Kerrisk
 327         Add cross reference to unix(7)
 328             For further information on UNIX domain abstract sockets.
 329
 330 posixoptions.7
 331     Carlos O'Donell
 332         Expand XSI Options groups
 333             We define in detail the X/Open System Interfaces i.e. _XOPEN_UNIX
 334             and all of the X/Open System Interfaces (XSI) Options Groups.
 335
 336             The XSI options groups include encryption, realtime, advanced
 337             realtime, realtime threads, advanced realtime threads, tracing,
 338             streams, and legacy interfaces.
 339     Michael Kerrisk
 340         Use a more consistent, less cluttered layout for option lists
 341     Michael Kerrisk
 342         Make function lists more consistent and less cluttered
 343             Use more consistent layout for lists of functions, and
 344             remove punctuation from the lists to make them less cluttered.
 345
 346 socket.7
 347     Michael Kerrisk  [Tomi Salminen]
 348         Fix error in SO_INCOMING_CPU code snippet
 349             The last argument is passed by value, not reference.
 350
 351 time.7
 352     Michael Kerrisk  [Enrique Garcia]
 353         Mention clock_gettime()/clock_settime() rather than [gs]ettimeofday()
 354             gettimeofday() is declared obsolete by POSIX. Mention instead
 355             the modern APIs for working with the realtime clock.
 356
 357             See https://bugzilla.kernel.org/show_bug.cgi?id=199049
 358
 359 unix.7
 360     Michael Kerrisk  [Rusty Russell]
 361         ERRORS: add EBADF for sending closed file descriptor with SCM_RIGHTS
 362
 363 vdso.7
 364     Michael Kerrisk
 365         VDSO symbols (system calls) are not visible to seccomp(2) filters
 366
 367 xattr.7
 368     Michael Kerrisk
 369         SEE ALSO: add selinux(8)
 370
 371 ld.so.8
 372     Mike Frysinger
 373         Make lack of separator escaping explicit
 374             Make it clear that the delimiters in LD_PRELOAD, LD_LIBRARY_PATH,
 375             and LD_AUDIT cannot be escaped so people don't try various methods
 376             (such as \:) to workaround it.
 377     Michael Kerrisk
 378         Remove unneeded mention of PATH in discussion of LD_LIBRARY_PATH
 379             This brief sentence doesn't add value to the text.