OpenBSD src changes summary for 2017-02-09 ==========================================
lib/libcrypto regress/usr.sbin sys/kern sys/net sys/netinet sys/netinet6 usr.bin/mandoc usr.bin/nc usr.bin/tmux usr.sbin/pstat usr.sbin/relayd == lib =============================================================== 01/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/lib libcrypto ~ des/set_key.c ~ man/DES_set_key.3 > DES keys are not 7 days long. ok jsing@ (dtucker@) == regress =========================================================== 02/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/regress usr.sbin ~ ospfd/Makefile > Add defines that are needed to compile XS code with Perl 5.24.1. (bluhm@) == sys =============================================================== 03/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sys kern ~ sched_bsd.c > Do no select a CPU to execute the current thread when being preempt()ed. > Calling sched_choosecpu() at this moment often result in moving the thread > to a different CPU. This does not help the scheduler and creates a domino > effect, resulting in kernel thread moving to other CPUs. > Tested by many without performance impact. Simon Mages measured a small > performance improvement and a smaller variance with an http proxy. > Discussed with kettenis@, ok martijn@, beck@, visa@ (mpi@) ~ uipc_usrreq.c > Release the NET_LOCK() before calling unp_detach(), fix a recursion > found by dtucker@ (mpi@) ~ uipc_usrreq.c > Temporarily grab the NET_LOCK() around soisdisconnected(). > This makes a (wrong) assert disappear and makes sure we hack to avoid a > recursion in the upcall case still work. > The real solution to this problem is to not grab the NET_LOCK() before > entering uipc_usrreq(). > Issue reported by dtucker@ (mpi@) ~ vfs_cache.c > Use TAILQ_FOREACH_SAFE in cache_purgevfs(). Fix whitespaces. > No binary change. (bluhm@) ~ kern_tc.c > remove a dead variable; ok millert, guenther (mikeb@) net ~ pf_table.c > Replace a custom loop calling yield() by the idiom to check if the > current process is hogging a CPU. > ok mikeb@, visa@, tedu@ (mpi@) ~ pf.c > percpu counters for TCP stats > ok mpi@ bluhm@ (jca@) netinet ~ ip_ah.c ~ ip_ipcomp.c > Put back a return that I have removed by accident. (bluhm@) ~ ip_output.c ~ tcp_input.c ~ tcp_output.c ~ tcp_subr.c ~ tcp_timer.c ~ tcp_usrreq.c ~ tcp_var.h > percpu counters for TCP stats > ok mpi@ bluhm@ (jca@) ~ icmp6.h > percpu counters for raw ipv6 and icmp6 stats > ok mpi@ (jca@) ~ ip_divert.c ~ ip_divert.h > percpu counters for divert(4) stats > ok dlg@ (jca@) ~ ip_mroute.c > Unbreak 'netstat -g' and make multicast route stats sysctl more robust. > ok mpi@ (rzalamena@) netinet6 ~ ip6_output.c > percpu counters for TCP stats > ok mpi@ bluhm@ (jca@) ~ icmp6.c ~ in6_var.h ~ mld6.c ~ nd6.c ~ nd6_nbr.c ~ nd6_rtr.c ~ raw_ip6.c ~ raw_ip6.h > percpu counters for raw ipv6 and icmp6 stats > ok mpi@ (jca@) ~ ip6_divert.c ~ ip6_divert.h > percpu counters for divert(4) stats > ok dlg@ (jca@) ~ icmp6.c > Dedup calls to icmp6stat_inc in icmp6_errcount, should save a few bytes > ok bluhm@ (jca@) == usr.bin =========================================================== 04/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin mandoc ~ libmandoc.h ~ main.c ~ mandoc_aux.h ~ mandocdb.c ~ tag.c ~ term_ps.c > Be consistent in protecting __attribute__ attributes with __; > from Christos Zoulas <christos @ NetBSD>. (schwarze@) ~ dba.c ~ dbm_map.c > Improve clarity of some casts; > from Christos Zoulas <christos at NetBSD>. (schwarze@) ~ mandocdb.c > No need to cast NULL when assigning it to a variable; > from Christos Zoulas <christos at NetBSD>. (schwarze@) nc ~ netcat.c > When getaddrinfo fails, print the requested host and port. > Should make debugging easier, especially when using -x literal_ipv6_address > (jca@) ~ nc.1 > Document that -x can take an ipv6 address enclosed in square brackets. > (jca@) ~ netcat.c > When netcat was started with -Uz, the exit status was always 1. If > the unix connect is successful, let nc -z close the socket and exit > with 0. > OK jca@ (bluhm@) tmux ~ screen-write.c > When an ordinary (not collected) cell is received, we need to flush any > delayed scrolling before drawing it. (nicm@) ~ screen-write.c > Combining characters need a full flush. (nicm@) ~ tmux.1 > Don't lie about the default size in the man page. (nicm@) ~ server-client.c ~ status.c ~ tmux.h > Break the message storage function into its own function, useful for > debugging. (nicm@) ~ tmux.1 > Document refresh-client -C. (nicm@) ~ cmd-new-session.c ~ cmd-swap-window.c ~ format.c ~ server-fn.c ~ server.c ~ session.c ~ tmux.1 ~ tmux.h > Instead of numbering session groups, give them a name which may be given > to -t instead of a target session. Also allow them to contain only one > session. (nicm@) == usr.sbin ========================================================== 05/05 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.sbin pstat ~ pstat.c > Teach pstat -d to also print untyped kernel symbols. > Storage defined in assembly code often doesn't define its type and size, > thus the nlist interface can't present it as N_DATA. Instead of fixing > all the assembly by adding .type pseudo ops, work around the problem in > pstat. Let's hope that allowing N_COMM is sufficient. This makes > ''pstat -d u cpuid_level'' useful. > Reported by mikeb@, "looks sane" mikeb@ guenther@ (jca@) relayd ~ check_tcp.c > Do not loose the send/expect validation error on timeout. > When a read timeout is reached, the code also has to check if there is > already anything in the input buffer and verify it again. This fixes > relayd and relayctl to show "send/expect failed" instead of "tcp read > timeout". > Found by Michael W. Lucas (while working on the Relayd Mastery book). > Thanks. > OK benno@ (reyk@) =============================================================================== _______________________________________________ odc mailing list [email protected] http://www.squish.net/mailman/listinfo/odc
