==================== Changes in man-pages-4.15 ==================== Released: 2018-02-02, Palo Alto Contributors ------------ The following people contributed patches/fixes or (noted in brackets in the changelog below) reports, notes, and ideas that have been incorporated in changes in this release: Adam Liddell Andrea Parri Andries E. Brouwer Elie Roudninski Eric Benton Florian Weimer G. Branden Robinson Jakub Wilk Joel Williamson John Hubbard Jorgen Hansen Keno Fischer Michael Kerrisk Michal Hocko NeilBrown Nikola Forró Nikolay Borisov Pradeep Kumar QingFeng Hao Ricardo Biehl Pasquali roblabla Roman Gushchin Shawn Landden Stefan Hajnoczi Stefan Raspl Tejun Heo Apologies if I missed anyone! New and rewritten pages ----------------------- s390_sthyi.2 QingFeng Hao [Michael Kerrisk] New page for s390-specific s390_sthyi(2) network_namespaces.7 Michael Kerrisk New page describing network namespaces Based on content moved from namespaces(7) Michael Kerrisk Reorganize text No content changes... Michael Kerrisk When a NW namespace is freed, veth devices are destroyed vsock.7 Stefan Hajnoczi [Jorgen Hansen, Michael Kerrisk] Document the VSOCK socket address family Newly documented interfaces in existing pages --------------------------------------------- cgroups.7 Michael Kerrisk [Tejun Heo] Document cgroups v2 "thread mode" Michael Kerrisk [Tejun Heo] Document cgroup v2 delegation via the 'nsdelegate' mount option Michael Kerrisk Document the cgroup.max.depth and cgroup.max.descendants files Michael Kerrisk Document 'release_agent' mount option Michael Kerrisk [Roman Gushchin] Document /sys/kernel/cgroup/delegate Michael Kerrisk [Roman Gushchin] Document /sys/kernel/cgroup/features Michael Kerrisk [Roman Gushchin] Document cgroups v2 cgroup.stat file Global changes -------------- Various pages G. Branden Robinson Standardize on "nonzero" Also add this term to the style guide in man-pages(7). Changes to individual pages --------------------------- bpf.2 Nikolay Borisov Sync list of supported map types with 4.14 kernel copy_file_range.2 Michael Kerrisk Library support was added in glibc 2.27 Shawn Landden glibc provides a user-space emulation where the system call is absent Florian Weimer EFBIG errors are possible, similar to write(2) Michael Kerrisk ERRORS: add EISDIR Michael Kerrisk Order ERRORS alphabetically Michael Kerrisk Add comment to code example explaining use of syscall(2) fcntl.2 read.2 write.2 NeilBrown Document "Lost locks" as cause for EIO. If an advisory lock is lost, then read/write requests on any affected file descriptor can return EIO - for NFSv4 at least. memfd_create.2 Michael Kerrisk glibc support for memfd_create() was added in version 2.27 mlock.2 Michael Kerrisk Make details for MLOCK_ONFAULT a little more explicit Michael Kerrisk glibc support for mlock2() is added in version 2.27 mmap.2 John Hubbard [Michael Hocko] MAP_FIXED is no longer discouraged MAP_FIXED has been widely used for a very long time, yet the man page still claims that "the use of this option is discouraged". John Hubbard MAP_FIXED updated documentation -- Expand the documentation to discuss the hazards in enough detail to allow avoiding them. -- Mention the upcoming MAP_FIXED_SAFE flag. -- Enhance the alignment requirement slightly. mount.2 Keno Fischer [Michael Kerrisk] Add EINVAL error condition when MS_BINDing MNT_LOCKED submounts mprotect.2 pkey_alloc.2 Michael Kerrisk Glibc support for memory protection keys was added in version 2.27 perf_event_open.2 Michael Kerrisk SEE ALSO: add perf(1) pkey_alloc.2 Michael Kerrisk Clarify description of pkey_alloc() 'flags' argument prctl.2 Michael Kerrisk Defer to capabilities(7) for discussion of the "keep capabilities" flag recvmmsg.2 sendmmsg.2 Nikola Forró Point out that error handling is unreliable seccomp.2 Michael Kerrisk Clarify that SECCOMP_RET_TRAP SIGSYS signal is thread-directed syscalls.2 Michael Kerrisk Add s390-specific s390_sthyi(2) to syscall list unshare.2 Michael Kerrisk Clarify the EUSERS occurred only until kernel 4.8 errno.3 Michael Kerrisk 'errno -s' can be used to search for errors by string in description Michael Kerrisk Add Linux error text corresponding to ENOMEM fgetpwent.3 Michael Kerrisk Add missing ATTRIBUTES preamble fts.3 Michael Kerrisk [Pradeep Kumar] fts_pathlen = strlen(fts_path) + strlen(fts_name) fuse.4 Michael Kerrisk Places errors in alphabetical order (no content changes) veth.4 Michael Kerrisk Add network_namespaces(7) sysfs.5 Michael Kerrisk Refer to cgroups(7) for information about files in /sys/kernel/cgroup capabilities.7 Michael Kerrisk Note which capability sets are affected by SECBIT_NO_SETUID_FIXUP Note explicitly that SECBIT_NO_SETUID_FIXUP is relevant for the permitted, effective, and ambient capability sets. Michael Kerrisk Deemphasize the ancient prctl(2) PR_SET_KEEPCAPS command The modern approach is SECBITS_KEEP_CAPS. Michael Kerrisk Clarify effect of CAP_SETFCAP Make it clear that CAP_SETFCAP allows setting arbitrary capabilities on a file. Michael Kerrisk Clarify which capability sets are effected by SECBIT_KEEP_CAPS This flag has relevance only for the process permitted and effective sets. Michael Kerrisk Rephrase CAP_SETPCAP description * Mention kernel versions. * Place current kernel behavior first Michael Kerrisk SECBIT_KEEP_CAPS is ignored if SECBIT_NO_SETUID_FIXUP is set Michael Kerrisk Ambient set is also cleared when UIDs are set to nonzero value cgroups.7 Michael Kerrisk Add a more complete description of cgroup v1 named hierarchies Michael Kerrisk Add a section on unmounting cgroup v1 filesystems Michael Kerrisk Add subsection describing cgroups v2 subtree delegation Michael Kerrisk Mention ENOENT error that can occur when writing to subtree_control file Michael Kerrisk Add list of currently available version 2 controllers Nikolay Borisov Add information about RDMA controller Michael Kerrisk Rewrite the description of cgroup v2 subtree control Michael Kerrisk [Tejun Heo] Note Linux 4.11 changes to cgroup v2 delegation containment rules Michael Kerrisk systemd(1) nowadays automatically mounts the cgroup2 filesystem Michael Kerrisk Clarify that cgroup.controllers is read-only Michael Kerrisk Elaborate a little on problems of splitting threads across cgroups in v1 Michael Kerrisk [Tejun Heo] Tweak the description of delegation of cgroup.subtree_control ip.7 Ricardo Biehl Pasquali INADDR_* values cannot be assigned directly to 's_addr' Michael Kerrisk s/INADDR_ANY/INADDR_LOOPBACK/ in discussion of htonl() INADDR_LOOPBACK is a better example, since it is not byte-order neutral. namespaces.7 network_namespaces.7 Michael Kerrisk Move content from namespaces(7) to network_namespaces(7) pid_namespaces.7 Michael Kerrisk SEE ALSO: add mount_namespaces(7) sched.7 Michael Kerrisk [Andrea Parri] Correctly describe effect of priority changes for RT threads The placement of a thread in the run queue for its new priority depends on the direction of movement in priority. (This appears to contradict POSIX, except in the case of pthread_setschedprio().) user_namespaces.7 Michael Kerrisk Mention NS_GET_OWNER_UID ioctl() operation