Clarify that tcp_tw_recycle will break communication with many
general-purpose remote Internet hosts (namely, remote NAT devices)
even when the Linux device itself is not behind NAT.
+
+
+==================== Changes in man-pages-3.78 ====================
+
+Released: 2015-01-22, Munich
+
+
+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:
+
+Akihiro Motoki <amotoki@gmail.com>
+Alexey Ishchuk <aishchuk@linux.vnet.ibm.com>
+Carlos O'Donell <carlos@redhat.com>
+Christian Seiler <christian@iwakd.de>
+Daniel J Blueman <daniel@quora.org>
+David Drysdale <drysdale@google.com>
+David Herrmann <dh.herrmann@gmail.com>
+Elie De Brauwer <eliedebrauwer@gmail.com>
+Elliot Hughes <enh@google.com>
+Jessica McKellar <jessica.mckellar@gmail.com>
+Kees Cook <keescook@chromium.org>
+Michael Hayes <mike@aiinc.ca>
+Michael Kerrisk <mtk.manpages@gmail.com>
+Rich Felker <dalias@aerifal.cx>
+Vince Weaver <vincent.weaver@maine.edu>
+
+Apologies if I missed anyone!
+
+
+New and rewritten pages
+-----------------------
+
+execveat.2
+ David Drysdale, Michael Kerrisk [Rich Felker]
+ New page for execveat(2)
+
+memfd_create.2
+ Michael Kerrisk, David Herrmann
+ New page for memfd_create() system call
+ Including notes about file sealing
+
+s390_pci_mmio_write.2
+ Alexey Ishchuk
+ New page for s390 s390_pci_mmio_write() and s390_pci_mmio_read()
+ New manual page for the new PCI MMIO memory access system
+ calls, s390_pci_mmio_write() and s390_pci_mmio_read(),
+ added for the s390 platform.
+
+
+Newly documented interfaces in existing pages
+---------------------------------------------
+
+fcntl.2
+ David Herrmann [Michael Kerrisk]
+ Document F_ADD_SEALS and F_GET_SEALS commands
+
+proc.5
+ Elie De Brauwer
+ Document /proc/sys/vm/compact_memory
+ Michael Kerrisk
+ Document /proc/sys/fs/nr_open
+
+
+New and changed links
+---------------------
+
+s390_pci_mmio_read.2
+ Michael Kerrisk
+ New link to new s390_pci_mmio_write(2) page
+
+
+Changes to individual pages
+---------------------------
+
+dup.2
+ Michael Kerrisk
+ Add reference to RLIMIT_NOFILE for EMFILE error
+ Michael Kerrisk
+ Add reference to RLIMIT_NOFILE for EBADF error on 'newfd'.
+
+execve.2
+fexecve.3
+ Michael Kerrisk
+ SEE ALSO: add execveat(2)
+
+fallocate.2
+mmap.2
+open.2
+truncate.2
+write.2
+ Michael Kerrisk
+ ERRORS: add EPERM for operation denied by file seal
+
+fcntl.2
+ Michael Kerrisk
+ ERRORS: add EBUSY case for F_SETPIPE_SZ
+ Michael Kerrisk
+ Add reference to RLIMIT_NOFILE for F_DUPFD EINVAL error on 'arg'.
+ Michael Kerrisk
+ ERRORS: add open file description lock error cases
+
+getrlimit.2
+ Michael Kerrisk
+ Update text on RLIMIT_NOFILE ceiling to refer to /proc/sys/fs/file-max
+
+mbind.2
+ Michael Kerrisk [Daniel J Blueman]
+ Clarify EFAULT text
+
+mmap.2
+shmget.2
+shm_open.3
+ Michael Kerrisk
+ SEE ALSO: add memfd_create(2)
+
+open.2
+ Michael Kerrisk
+ Refer to RLIMIT_NOFILE for explanation of EMFILE error
+ Michael Kerrisk
+ Add execveat(2) in system call list under "Rationale for openat()"
+
+perf_event_open.2
+ Vince Weaver
+ Clarify description of overflow events
+ Update the perf_event_open manpage to be more consistent when
+ discussing overflow events. It merges the discussion of
+ poll-type notifications with those generated by SIGIO
+ signal handlers.
+ This addresses the remaining FIXMEs is the document.
+ Vince Weaver
+ Remove inaccurate paragraph describing attr.config
+ Remove an inaccurate paragraph about values in the attr.config
+ field. This information was never true in any released kernel;
+ it somehow snuck into the manpage because it is still described
+ this way in tools/perf/design.txt in the kernel source tree.
+ Michael Kerrisk
+ Correct the kernel version number for PERF_COUNT_HW_CACHE_NODE
+ Michael Kerrisk
+ Add some kernel version numbers to various fields and constants
+
+ptrace.2
+sigaction.2
+seccomp.2
+ Kees Cook
+ Ptrace and siginfo details
+ While writing some additional seccomp tests, I realized
+ PTRACE_EVENT_SECCOMP wasn't documented yet. Fixed this, and added
+ additional notes related to ptrace events SIGTRAP details.
+
+readv.2
+ Michael Kerrisk
+ Update details on glibc readv()/writev() wrapper behavior
+ And add a historical detail about Linux 2.0.
+
+select.2
+ Michael Kerrisk
+ Mention RLIMIT_NOFILE as a possible cause of EINVAL error
+
+syscall.2
+ Kees Cook
+ Add arm64 and mips
+ Add mips and arm64 to tables, along with some further
+ details on these architectures,
+
+syscalls.2
+ Michael Kerrisk
+ Add s390_pci_mmio_read(2) and s390_pci_mmio_write(2)
+ Michael Kerrisk
+ Note kernel() version that introduced get_kernel_syms()
+ Note kernel version that introduced ppc_rtas()
+ Note kernel version that introduced create_module()
+ Note kernel version that added setup()
+ Michael Kerrisk
+ Remove some details for sync_file_range2()
+ Make the table a bit simpler. The details can anyway be
+ found in the system call man page.
+
+utimensat.2
+ Michael Kerrisk [Elliot Hughes]
+ If both tv_sec fields are UTIME_OMIT, the file need not exist
+ As noted by Elliot, if both tv_sec fields are UTIME_OMIT,
+ utimensat() will return success even if the file does not exist.
+
+errno.3
+ Michael Kerrisk
+ The RLIMIT_NOFILE resource limit is a common cause of EMFILE
+
+exec.3
+ Michael Kerrisk
+ SEE ALSO: add execveat(2)
+
+fclose.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+ Harmonize all the manual pages to use "stream" for FILE*
+ instead of randomly using "fp" or "stream." Choosing something
+ and being consistent helps users scan the man pages quickly
+ and understand what they are looking at.
+
+fexecve.3
+ Michael Kerrisk
+ Rewrite the script+close-on-exec problem as a BUG
+ Also, add one or two details about this scenario.
+ Michael Kerrisk
+ The natural idiom when using fexecve() is to use the close-on-exec flag
+
+fmemopen.3
+ Michael Kerrisk
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+fopencookie.3
+ Michael Kerrisk
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+getgrent_r.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+getline.3
+ Michael Kerrisk
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+getmntent.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+getpw.3
+ Michael Kerrisk [Carlos O'Donell]
+ Describe return value when 'uid' is not found
+
+getpwent_r.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+getspnam.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+malloc_info.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+
+posix_fallocate.3
+ Michael Kerrisk
+ Note that posix_fallocate() is implemented using fallocate(2)
+
+putgrent.3
+ Carlos O'Donell
+ Consistency fix: use "stream" as name for "FILE *" argument
+ Harmonize all the manual pages to use "stream" for FILE*
+ instead of randomly using "fp" or "stream." Choosing something
+ and being consistent helps users scan the man pages quickly
+ and understand what they are looking at.
+
+locale.5
+ Akihiro Motoki
+ Correct variable name
+
+proc.5
+ Michael Kerrisk
+ Remove bogus statement about NR_OPEN being a ceiling for file-max
OSDN Git Service
