From e2f6918e7e7e6a96f81b49f65bfb7333004b926d Mon Sep 17 00:00:00 2001 From: Michael Kerrisk Date: Tue, 31 Mar 2009 10:07:27 +1300 Subject: [PATCH] Changes: Update for 3.20 changes Signed-off-by: Michael Kerrisk --- Changes | 408 +++++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 405 insertions(+), 3 deletions(-) diff --git a/Changes b/Changes index 7f3e92401..f9c2dd36c 100644 --- a/Changes +++ b/Changes @@ -9,6 +9,30 @@ Contributors The following people contributed notes, ideas, or patches that have been incorporated in changes in this release: +Alan Curry +Américo Wang +Andi Kleen +Carlos O'Donell +Chunming Chang +Colin Watson +Eelco Dolstra +Jan Engelhardt +Jens Thoms Toerring +Johannes Stezenbach +Leandro A. F. Pereira +Martin Gebert +Michael Kerrisk +Miike O'Connor +Mike Frysinger +Nikanth Karthikesan +Reuben Thomas +Reuben Thomas +Roland McGrath +Sam Varshavchik +Simon Gomizelj +Tanaka Akira +Teddy Hogeborn +Walter Jontofsohn Apologies if I missed anyone! @@ -16,19 +40,397 @@ Apologies if I missed anyone! New and rewritten pages ----------------------- +cpuid.4 + Andi Kleen + New page for cpuid access device + +msr.4 + Andi Kleen + New page documenting x86 CPU MSR access device + Newly documented interfaces in existing pages --------------------------------------------- - -New and changed links ---------------------- +proc.5 + Américo Wang + Document /proc/sys/vm/swappiness + Michael Kerrisk + Document /proc/sysrq-trigger Global changes -------------- +timer_create.2 +timer_delete.2 +timer_getoverrun.2 +timer_settime.2 +numa.7 + Michael Kerrisk + Make source layout of 'Link with' text consistent with other pages + No actual change to formatted output, but this makes the + page sources more consistent for the purpose of grepping, etc. + +mempcpy.3 +signbit.3 +significand.3 + Michael Kerrisk + Global fix: acknowledge FSF in copyright + These pages are heavily based on original material in + glibc info pages, but the comments in the source of the pages + did not acknowledge the FRF copyright on the original material. + Fix that. + +accept.2 +read.2 +recv.2 +send.2 +write.2 + Michael Kerrisk + Fix discussion of EAGAIN/EWOULDBLOCK errors + For a non-blocking socket, POSIX.1-2001/2008 allow either + EAGAIN or EWOULDBLOCK to be returned in cases where a call + would have blocked. Although these constants are defined + with the same value on most Linux architectures (PA-RISC + is the exception), POSIX.1 does not require them to have + the same value. Therefore, a portable application using + the sockets API should test for both errors when checking + this case. + + (NB POSIX.1 only mentions EWOULDBLOCK in the context of + the sockets interfaces.) + + Change made after a note cross-posted on linux-arch@vger, + http://thread.gmane.org/gmane.linux.debian.ports.hppa/5615 + and a suggestion for write(2) from Carlos O'Donell + +basename.3 +getgrent.3 +getgrnam.3 +getpwent.3 +getpwnam.3 +readdir.3 + Michael Kerrisk + Note that returned pointer should not be given to free() + +armscii-8.7 +cp1251.7 +iso_8859-10.7 +iso_8859-11.7 +iso_8859-13.7 +iso_8859-14.7 +iso_8859-15.7 +iso_8859-16.7 +iso_8859-2.7 +iso_8859-3.7 +iso_8859-4.7 +iso_8859-5.7 +iso_8859-6.7 +iso_8859-7.7 +iso_8859-8.7 +iso_8859-9.7 +koi8-r.7 +koi8-u.7 + Michael Kerrisk + Add explicit character set encoding to first line of source + Nowadays mandb has provision to understand a character set + encoding that is explicitly indicated in the first line + of the source. As pointed out by Colin Watson, including + such an explicit indication on pages encoded in anything + other than ISO 8859-1 or UTF-8 is useful for man-pages + that aren't shipped in UTF-8. + + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=519209 + and for some other background (responded to by Colin Watson + in the above report): + http://thread.gmane.org/gmane.comp.internationalization.linux/6040 + ("man page encoding", 5 Jul 2005) + Changes to individual pages --------------------------- +fallocate.2 + Michael Kerrisk + VERSIONS: glibc support is provided since version 2.10 + +fcntl.2 + Michael Kerrisk + Remove mention of EWOULDBLOCK from discussion of mandatory locking + In the kernel, the error on encountering a mandatory lock is + EAGAIN. Although EAGAIN and EWOULDBLOCK are the same on + most Linux architectures, on some they are not, so don't + mention EWOULDBLOCK as it is misleading. (Mea culpa.) + +getcontext.2 + Michael Kerrisk + Note that POSIX.1-2008 removes the specification of getcontext() + +getitimer.2 + Michael Kerrisk + Note that POSIX.1-2008 recommends POSIX timers API instead of this API + +gettimeofday.2 + Michael Kerrisk + Note that POSIX.1-2008 recommends clock_gettime() instead of this API + +ptrace.2 + Michael Kerrisk + Note use of 'data' for PTRACE_SYS{CALL,EMU} and PTRACE_*_SINGLESTEP + These operations use the 'data' argument as a signal number, + like PTRACE_CONT. + +ptrace.2 + Mike Frysinger + only reference + The kernel no longer installs linux/user.h, so update + references to sys/user.h. + +recv.2 + Michael Kerrisk + Add 'iovec' defn to defn of 'msghdr' structure + The 'msghdr' structure includes a field of type 'iovec', + so show the definition of that structure in this page. + +rename.2 + Michael Kerrisk + Make ENOENT description consistent with POSIX.1-2008 + +timerfd_create.2 + Michael Kerrisk + ERRORS: add EINVAL for invalid 'flags' for timer_settime() + +truncate.2 + Michael Kerrisk + SYNOPSIS: Fix description of feature test macro requirements + After a report by Arvid Norlander. + +bcopy.3 + Michael Kerrisk + Note that POSIX.1-2008 removes specification of bcopy() + +bsd_signal.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends sigaction(2) instead of this API + +ctime.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends strftime(3) instead of these functions + +ecvt.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends sprintf(3) instead of these functions + +gcvt.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends sprintf(3) instead of this function + +getcwd.3 + Michael Kerrisk + Note that getcwd() should be used instead of the obsolete getwd() + +getgrent.3 + Michael Kerrisk + Returned buffer may be statically allocated and overwritten by + later calls + +gethostbyname.3 + Michael Kerrisk + POSIX.1-2008 recommends getaddrinfo(3) and getnameinfo(3) instead + +getnetent_r.3 + Michael Kerrisk + Fix function name in text: s/getnetbynumber_r/getnetbyaddr_r/ + The SYNOPSIS showed the right function name (getnetbyaddr_r), + but the text repeatedly used the wrong name (getnetbynumber_r). + Probably, this was a cut-and-paste error. + +getpwent.3 + Michael Kerrisk + Returned buffer may be statically allocated and overwritten by + later calls + +index.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends strchr(3) and strrchr(3) instead + +isalpha.3 + Michael Kerrisk + Explain why POSIX.1-2008 marks isascii(3) obsolete + +lockf.3 + Nikanth Karthikesan + Update pointer to documentation in kernel source + +makecontext.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends the use of POSIX threads instead + +mq_notify.3 + Michael Kerrisk + Document the POSIX.1-2008 optional EINVAL error + POSIX.1-2008 allows an optional EINVAL error if + notification==NULL and the caller is not currently + registered to receive notifications. + +posix_fallocate.3 + Michael Kerrisk + Clarify that EINVAL also occurs of 'len' *equals* zero + See http://bugzilla.kernel.org/show_bug.cgi?id=12919 + +posix_fallocate.3 + Michael Kerrisk + Document POSIX.1-2001 and POSIX.1-2008 specifications for EINVAL error + See http://bugzilla.kernel.org/show_bug.cgi?id=12919 + +posix_memalign.3 + Michael Kerrisk + Document handling of size==0 case for posix_memalign() + +pthread_exit.3 + Michael Kerrisk + Fix error in DESCRIPTION: s/pthread_create/pthread_exit/ + +realpath.3 + Michael Kerrisk + Rework resolved_path==NULL discussion w.r.t. POSIX.1-200[18] + Although the page already mentioned the resolved_path==NULL + feature, and that this feature was added in POSIX.1-2008, there + was still some crufty text in BUGS that hadn't been updated to + reflect the POSIX.1-2008 changes. + + Also, some other minor wording and grammar fixes. + +scalb.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends scalbln*(3) instead + +seekdir.3 + Michael Kerrisk + SYNOPSIS: Fix type of 'offset' argument: s/off_t/long/ + And add a NOTES section pointing out that 'off_t' + was indeed used in glibc 2.1.1 and earlier. + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=519230 + +sem_post.3 + Michael Kerrisk + Document EOVERFLOW error + +shm_open.3 + Michael Kerrisk + Recast discussion on name length to exclude terminating NULL byte + Probably it's clearer to describe the length of the IPC object + name as a count that excludes the null terminator. + +siginterrupt.3 + Michael Kerrisk + Note that POSIX.1-2008 recommends sigaction() instead + +sigset.3 + Michael Kerrisk + Note APIs that POSIX.1-2008 recommends instead of these obsolete APIs + +strftime.3 + Michael Kerrisk + Small fix to description of %G + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=516677 + +strftime.3 + Michael Kerrisk + Add details on ISO 8601 week-based dates + ISO 8602 week-based dates are relevant for %G, %g, and %V, + and the existing details on these dates are a little thin. + +strftime.3 + Michael Kerrisk + Remove mention of year from ISO 8601 standard + The text mentioned the 1988 8601 standard, but there have + already been two revisions of the standard since then, so + simply remove mention of the year. + +telldir.3 + Michael Kerrisk + SYNOPSIS: Fix return type: s/off_t/long/ + And add a NOTES section pointing out that 'off_t' + was indeed used in glibc 2.1.1 and earlier. + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=519230 + +timeradd.3 + Michael Kerrisk + Note that on some systems, <=, >=, == don't work for timercmp() + +timeradd.3 + Michael Kerrisk + SYNOPSIS: Fix return types of timerisset() and timercmp() + +toascii.3 + Michael Kerrisk + Note why POSIX.1-2008 marks this function obsolete + +console_ioctl.4 + Alan Curry + Fix 'argp' type for KDGETLED description + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=517485 + +group.5 + Michael Kerrisk + Various minor rewordings and improvements + +resolv.conf.5 + Michael Kerrisk + Document 'ip6-bytestring' option + +resolv.conf.5 + Michael Kerrisk + Document 'edns0' option + +resolv.conf.5 + Michael Kerrisk + Document 'ip6-dotint' / 'no-ip6-dotint' option + +resolv.conf.5 + Michael Kerrisk + Note that maximum value of 'ndots' option is capped to 15 + +resolv.conf.5 + Michael Kerrisk + Note that maximum value of 'timeout' option is capped to 30 + +hier.7 + Michael Kerrisk + Add description of /srv + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=520904 + +ip.7 + Michael Kerrisk + Fix type used to declare sin6_port + The page should use the type specified by POSIX, + rather than the (equivalent) type used in the kernel + +ipv6.7 + Teddy Hogeborn + Fix types used to declare sin6_family and sin6_port + The page should use the types specified by POSIX, + rather than the (equivalent) types used in the kernel. + + See http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=517074 + +mq_overview.7 + Michael Kerrisk + Recast discussion on name length to exclude terminating NULL byte + Probably it's clearer to describe the length of the IPC object + name as a count that excludes the null terminator. + +rtld-audit.7 + Michael Kerrisk + Note that multiple libraries in LD_AUDIT doesn't work + This is reportedly fixed in glibc 2.10. + See http://sourceware.org/bugzilla/show_bug.cgi?id=9733 + +sem_overview.7 + Michael Kerrisk + Fix discussion of length of semaphore names + Because of the "sem." prefix added by glibc to a semaphore + name, the limit on the length of the name (excluding the + terminating null byte) is 251 characters.