1 ==================== Changes in man-pages-5.04 ====================
3 Released: 2019-08-02, 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 Andrew Price <anprice@redhat.com>
14 Christian Brauner <christian.brauner@ubuntu.com>
15 Florian Weimer <fweimer@redhat.com>
16 Jakub Wilk <jwilk@jwilk.net>
17 Jan Kara <jack@suse.cz>
18 Jann Horn <jannh@google.com>
19 Kenigbolo Meya Stephen <kenigbol@ut.ee>
20 Marko Myllynen <myllynen@redhat.com>
21 Michael Kerrisk <mtk.manpages@gmail.com>
22 Mikael Magnusson <mikachu@gmail.com>
23 Robert Edmonds <edmonds@debian.org>
24 Silviu Popescu <silviupopescu1990@gmail.com>
25 Torin Carey <torin@tcarey.uk>
26 Witold Baryluk <witold.baryluk@gmail.com>
27 Yang Xu <xuyang2018.jy@cn.fujitsu.com>
29 Apologies if I missed anyone!
32 Newly documented interfaces in existing pages
33 ---------------------------------------------
36 Michael Kerrisk [Christian Brauner, Jakub Wilk]
41 Add P_PIDFD for waiting on a child referred to by a PID file descriptor
45 Refresh against kernel v5.4-rc7
56 Changes to individual pages
57 ---------------------------
61 Rename arguments for consistency with clone3()
62 Make the names of the clone() arguments the same as the fields
63 in the clone3() 'args' struct:
70 Consistently order paragraphs for CLONE_NEW* flags
71 Sometimes the descriptions of these flags mentioned the
72 corresponding section 7 namespace manual page and then the
73 required capabilities, and sometimes the order was the was
74 the reverse. Make it consistent.
75 Michael Kerrisk [Christian Brauner, Jann Horn]
76 EXAMPLE: Allocate child's stack using mmap(2) rather than malloc(3)
77 Christian Brauner suggested mmap(MAP_STACK), rather than
78 malloc(), as the canonical way of allocating a stack for the
79 child of clone(), and Jann Horn noted some reasons why
80 (MAP_STACK exists elsewhere, and mmap() returns a page-aligned
81 block of memory, which is useful if we want to set up a guard
82 page at the end of the stack).
83 Michael Kerrisk [Christian Brauner]
84 Tidy up the description of CLONE_DETACHED
85 The obsolete CLONE_DETACHED flag has never been properly
86 documented, but now the discussion CLONE_PIDFD also requires
87 mention of CLONE_DETACHED. So, properly document CLONE_DETACHED,
88 and mention its interactions with CLONE_PIDFD.
89 Michael Kerrisk [Christian Brauner]
90 Give the introductory paragraph a new coat of paint
91 Change the text in the introductory paragraph (which was written
92 20 years ago) to reflect the fact that clone*() does more things
95 Remove wording that suggests CLONE_NEW* flags are for containers
96 These flags are used for implementing many other interesting
99 Remove various details that are already covered in namespaces pages
100 Remove details of UTS, IPC, and network namespaces that are
101 already covered in the corresponding namespaces pages in section 7.
106 Adjust references to namespaces(7)
107 Adjust references to namespaces(7) to be references to pages
108 describing specific namespace types.
112 Add gfs2 to the list of punch hole-capable filesystems
115 Michael Kerrisk [Robert Edmonds]
116 Emphasize that FS_IOC_GETFLAGS and FS_IOC_SETFLAGS argument is 'int *'
120 Add reference to ioctl(2) SEE ALSO section
121 The referenced section lists various pages that document ioctls.
125 Note that MAP_STACK exists on some other systems
127 Some rewording of the description of MAP_STACK
128 Reword a little to allow for the fact that there are now
129 *two* reasons to consider using this flag.
133 Note the waitid() use case for PID file descriptors
135 Add a subsection header "Use cases for PID file descriptors"
137 Make it a little more explicit the CLONE_PIDFD returns a PID FD
141 EXAMPLE: allocate stack using mmap() MAP_STACK rather than malloc()
145 Add some details about Q_QUOTAON
150 Switch to "considerate language"
154 POLLIN_SET/POLLOUT_SET/POLLEX_SET are now defined in terms of EPOLL*
155 Since kernel commit a9a08845e9acbd224e4ee466f5c1275ed50054e8, the
156 equivalence between select() and poll()/epoll is defined in terms
157 of the EPOLL* constants, rather than the POLL* constants.
161 waitid() can be used to wait on children in same process group as caller
162 Since Linux 5.4, idtype == P_PGID && id == 0 can be used to wait
163 on children in same process group as caller.
165 Clarify semantics of waitpid(0, ...)
166 As noted in kernel commit 821cc7b0b205c0df64cce59aacc330af251fa8f7,
167 threads create an ambiguity: what if the calling process's PGID
168 is changed by another thread while waitpid(0, ...) is blocked?
169 So, clarify that waitpid(0, ...) means wait for children whose
170 PGID matches the caller's PGID at the time of the call to
174 Michael Kerrisk [Witold Baryluk]
175 Clarify that AT_BASE_PLATFORM and AT_EXECFN return pointers to strings
176 See https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=942207
180 Attempt clarify domain/search interaction
181 The domain directive is historic at this point; it should not
185 Michael Kerrisk [Silviu Popescu]
186 Small wording fix in description of SIOCGIFCONF
187 SIOCGIFCONF returns "network layer" addresses (not "transport
192 Add a little more detail on scope of UTS namespaces