summaryrefslogtreecommitdiff
path: root/lib/libc
AgeCommit message (Collapse)Author
2015-11-14Split the non-syscall ASM bits from SYS.h into DEFS.h and use that in thePhilip Guenther
non-syscall .S source ok millert@ miod@
2015-11-14Give clear directions on how to declare, PROTO_*() and DEF_*() new symbolsPhilip Guenther
prodded by deraadt@
2015-11-10update NAME section to include all documented functions,Jason McIntyre
or otherwise change Dt to reflect the name of an existing function; feedback/ok schwarze
2015-11-10Document tzsetwall(). OK schwarze@ jmc@Todd C. Miller
2015-11-10Split the intra-thread functionality from kill(2) into its own syscallPhilip Guenther
thrkill(2), rolling the kill(2) syscall number with the ABI change to avoid breaking binaries during during the transition. thrkill(2) includes a 'tcb' argument that eliminates the need for locking in pthread_kill() and simplifies pthread_cancel(). Switch __stack_smash_handler() to use thrkill(2) and explicitly unblock SIGABRT. Minor bump to both libc and libpthread: make sure you install a new kernel! ok semarie@
2015-11-10libc.so can't be unloaded, so move the hidden atexit() and pthread_atfork()Philip Guenther
stubs for the executable from crtbegin.o into libc, which lets them be excluded from static links that don't use them. For this, drop the normal crt{begin,end}S.o from libc.so: the .init and .fini sections for libc aren't called at the right times anyway, so it's good that they're unused. libc.so just needs __guard_local and the .note.openbsd.ident section, so add them to stack_protector.c for now (this will be improved) "good time" deraadt@
2015-11-08inet(4), not inet(3);Jason McIntyre
2015-11-07add missing NAME entries;Jason McIntyre
2015-11-06Remove needless comma.mmcc
2015-11-05Set the name server counter in __res_state correctly if a nameserverAlexander Bluhm
from asr_ctx was skiped. Missed in previous commit. OK deraadt@
2015-11-05After removing National Language Support (NLS) from base, stopAlexander Bluhm
creating the directory /usr/share/nls. Having a non-existing default path in catopen(3) does not make sense, so remove it. If the user does not specify a NLS path, better fail early than fail because of an empty directory. Remove path form hier(7). OK stsp@ schwarze@ jmc@
2015-11-05When filling the __res_state compatibiliy struct, a long list ofAlexander Bluhm
nameservers could overflow the dns search pointers. Restrict the number, size and address family of nameservers in res_init(3). This fixes a crash in sendmail. Only programs that use the bind resolver internals directly are affected. OK deraadt@ millert@
2015-11-05document the net.inet6.ip6.ifq variables in sysctl(3) as well, referring backStuart Henderson
to the 'struct ifqueue' description for net.inet.ip.ifq to reduce chance of them getting out of sync. ok logan, ok/tweak mikeb
2015-11-04replace setbuf with setvbuf, from Frederic NowakTed Unangst
2015-11-04clean up HISTORY;Jason McIntyre
2015-11-03Improve explanations throughoutTheo de Raadt
2015-11-03systrace should work nowTed Unangst
2015-11-02kill 0 has worked for ages, no idea how the man page could be so innacurate ↵Theo de Raadt
so long; ok guenther
2015-11-02(struct timezone *)NULL -> NULLmmcc
2015-11-02(struct timezone *)0 -> NULLmmcc
2015-11-01replace "can not" with "cannot";Jason McIntyre
2015-11-01delete old lint ARGSUSED commentsPhilip Guenther
2015-11-01delete old lint ARGSUSED commentsPhilip Guenther
2015-10-31Invoke the _HIDDEN() sigprocmask label rather than the public one.Miod Vallat
2015-10-31This is sigprocmask, not sigpending.Miod Vallat
2015-10-31Do not include a timestamp in the syslog message. There is no need --Theo de Raadt
syslogd will fill it in immediately upon reception on the other side of sendsyslog(2). Our libc only talks to our syslogd, which will fix the timestamp before forwarding. syslog_r has done this for a long time already. ok tedu bluhm
2015-10-30For getgrent.3:Philip Guenther
- delete the commented out setgrfile bits, as well as the MLINK - rename getgr{nam,gid}_r()'s 'buffer' parameters to 'buf' to be consistent with getpwnam(3) and practically all other section 3 manpages - we have sysconf(_SC_GETGR_R_SIZE_MAX) now, so uncomment and revise the text For getpwnam.3: - rename buflen to bufsize for consistency - rewrite the description of the getpw{nam,uid}_r functions to match the better getgr{nam,gid}_r() description, adding sysconf() bits Add sysconf(3) to SEE ALSO for both encouraged by schwarze@ ok millert@
2015-10-29update text to be more like posix, in particular wrt length = 0Ted Unangst
2015-10-29(struct timezone *)0 -> NULL, while I'm here.mmcc
2015-10-28We are always using _PATH_RESCONF, so no need to remember the path onEric Faurot
the resolver. ok millert@ deraadt@
2015-10-28Remove support for [addr]:port syntax from the "nameserver" line.Theo de Raadt
This extension never made it to other systems. (pledge is also happy with this. The idea of DNS @ any port collides with pledge encouraring differentiation between DNS and non-DNS sockets) ok phessler jung sthen kettenis
2015-10-28Rewrite the mbtowc(3) page for clarity. Explain what needs to be done on error.Stefan Sperling
With input from jmc, zhuk, schwarze, and bentley. ok jmc zhuk bentley
2015-10-27Wrap the remaining math functions in libc: __fpclassify*(), __flt_rounds(),Philip Guenther
and ldexp(). ok millert@
2015-10-27The fd of the passwd database is marked close-on-exec nowPhilip Guenther
Incorrect text pointed out by tedu@ ok deraadt@ millert@ tedu@
2015-10-26don't keep the passwd database open; callers do not expect that they willTed Unangst
become responsible for calling endpwent. ok deraadt
2015-10-25Wrap waitpid() so calls go direct; weaken wait() and wait3().Philip Guenther
Strip out unnecessary #includes and use NULL instead of (struct rusage *)0
2015-10-25Move the _atfork_list definition to atexit.c so that the fork syscall stubPhilip Guenther
doesn't get pulled into all static executables ok millert@ jca@
2015-10-25Hide __atexit and __atexit_register_cleanup()Philip Guenther
Wrap __cxa_{atexit,finalize}() so the call from exit() goes direct Switch regress/lib/libc/atexit/ to be built with -static so that it can still access __atexit* ok millert@ jca@
2015-10-25Nuke union of d_packname with un_d_boot0 & un_d_boot1, keepingKenneth R Westerback
d_packname as a simple field. Also nuke the access #defines, and support for b0 and b1 capabilities in disktab. ok deraadt@ miod@
2015-10-25The only thing that was translated into multiple languages in OpenBSDAlexander Bluhm
are the errno messages and signal names. Everything else is in English. We are not planning to translate more text. Running a mixed system with less than 1% of the text in native language makes no sense. So remove the NLS support from libc messages. The catopen(3) functions stay as they are. OK stsp@ mpi@
2015-10-25Use sigaction() instead of signal() to avoid pulling in unnecessaryPhilip Guenther
wrappers. To keep uses from crawling back in, mark signal() as deprecated inside libc. ok deraadt@
2015-10-24typommcc
2015-10-24Move #includes from private.h to the .c files that need them, getting rid ofPhilip Guenther
several. Switch from FILENAME_MAX to PATH_MAX (it's for open(), not fopen()). ok deraadt@ tedu@ krw@
2015-10-24nl_langinfo(3) conforms to POSIX.Anthony J. Bentley
ok jmc@
2015-10-24Unify all the errno names inAlexander Bluhm
- include comment - libc errlist - nls C msg - man page OK tedu@
2015-10-24remove a paste error and get section numbers right for sysctl(3);Jason McIntyre
2015-10-23Verify that opened message catalog is valid, i.e. avoid integer overflowsTobias Stoeckmann
and out of boundary accesses. with input by miod, ok stsp
2015-10-23Switch if_nameindex(3) to use the new NET_RT_IFNAMES sysctl to get theClaudio Jeker
list of interface names. At the same time switch if_nametoindex(3) and if_indextoname(3) to use if_nameindex(3) instead of getifaddrs(3). if_nameindex(3) exposes much less then getifaddrs(3) and is allowed by pledge(2). With and OK deraadt@
2015-10-23Fix waitpid() loop again: do the errno check only if waitpid() returns -1Philip Guenther
and check WIFEXITED() only if it returns != -1. Delete the logging of errors other than ECHILD: EFAULT and EINVAL are impossible here. ok deraadt@ millert@
2015-10-23Use waitpid() instead of wait() to avoid returning early from another childPhilip Guenther
exiting, and loop the waitpid() on EINTR ok deraadt@ millert@