OpenBSD src changes summary for 2016-05-31 ==========================================
gnu/usr.bin/binutils-2.17 lib/libc lib/libedit lib/libexpat regress/lib regress/usr.bin sbin/newfs sys/arch sys/dev sys/kern sys/net sys/netinet sys/netinet6 sys/sys usr.bin/aucat usr.bin/ssh usr.sbin/httpd usr.sbin/installboot usr.sbin/rebound == gnu =============================================================== 01/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/gnu usr.bin/binutils-2.17 ~ bfd/bfd-in2.h ~ bfd/elf-bfd.h ~ bfd/elf.c ~ bfd/elflink.c ~ bfd/libbfd-in.h ~ bfd/libbfd.h ~ bfd/linker.c ~ bfd/targets.c ~ include/bfdlink.h ~ ld/ld.h ~ ld/ldlang.c ~ ld/ldmain.c ~ ld/lexsup.c > Port over binutils fix for PR ld/3111. > This greatly speeds up up linking of object files that contain > lots of dwarf2 symbols by caching symbol tables. > Original patch before binutils was switched to GPLv3 from: > https://sourceware.org/ml/binutils/2006-08/msg00334.html > Aaron Miller made us aware of that diff and adapted it to our > in-tree binutils. Many thanks! > Ports build testing on amd64 by krw@ > ok deraadt@ krw@ (stefan@) == lib =============================================================== 02/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/lib libc ~ sys/ptrace.2 > correct documentation for PT_WRITE_I. it may be necessary to flush icache. > after explanation from kettenis (tedu@) libedit ~ readline.c > Fix a crash in stifle_history() that occurred when the history > contained more than twice as many entries as the new limit: The > argument of remove_history() only counts existing elements, not > those previously deleted. > Crash found by bentley@ when exiting sqlite3(1). (schwarze@) libexpat ~ lib/xmlparse.c ~ lib/xmltok.c ~ lib/xmltok.h ~ lib/xmltok_impl.c > fix CVE-2016-0718. (rpointel@) == regress =========================================================== 03/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/regress lib ~ libedit/readline/history.c > Style unification and whitespace cleanup, in part suggested by > Christian Heckendorf. No functional change. (schwarze@) ~ libedit/readline/history.c > The argument of remove_history() only counts existing entries, > not those that were previously deleted. (schwarze@) usr.bin ~ sdiff/Iflag12.out ~ sdiff/Iflag21.out ~ sdiff/Ilflag12.out ~ sdiff/Ilflag21.out ~ sdiff/Ilsflag12.out ~ sdiff/Ilsflag21.out ~ sdiff/Isflag12.out ~ sdiff/Isflag21.out ~ sdiff/Makefile ~ sdiff/file1 ~ sdiff/file2 ~ sdiff/lflag.out ~ sdiff/same.out ~ sdiff/sflag.out ~ sdiff/short.out ~ sdiff/wflag.out + sdiff/short.in > Replace most of the tests based on systrace output; from Ray. (okan@) == sbin ============================================================== 04/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sbin newfs ~ newfs.c > permit wxallowed on mfs > discussed with naddy a bit, ok millert (deraadt@) == sys =============================================================== 05/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sys arch ~ armish/Makefile ~ armv7/Makefile ~ landisk/Makefile ~ loongson/Makefile ~ luna88k/Makefile ~ macppc/Makefile ~ octeon/Makefile ~ sgi/Makefile ~ socppc/Makefile ~ zaurus/Makefile > fix arch/mach confusion in make tags. from frantisek holop (tedu@) dev ~ softraid_crypto.c > Ensure that softraid crypto is not run on the crypto taskq. > When softraid crypto was written, it was intended that all crypto > operations would be run by softraid. As such, it called crypto_invoke() > directly, rather than crypto_dispatch() which would queue the task for > later processing. However, r1.53 of crypto/crypto.c started running > completion callbacks as a separate task. This leads to deadlocks with the > crypto taskq blocking in biowait, when softraid crypto volumes are stacked > (often with vnds). > In order to avoid task queueing and ensure that the operations are run > within softraid, set the CRYPTO_F_NOQUEUE flag. With r1.62 of > crypto/crypto.c crypto_dispatch() also started respecting this flag, hence > we can also switch back to calling the standard crypto_dispatch() > interface, rather than calling crypto_invoke() directly. > Tested by Peter Wens (peter at wenka dot nl) - thanks! > ok krw@ mlarkin@ mikeb@ tedu@ (jsing@) ~ softraid.c ~ softraid_raid5.c ~ softraidvar.h > Provide a function for calculting the rebuild percentage, rather than > having five copies of the same code. > ok krw@ (jsing@) kern ~ syscalls.master > sys_o58_kill is no longer needed for compat. > ok guenther sthen (deraadt@) ~ syscalls.master > sys_osendsyslog can go away; we have transitioned to the new version with > a flags argument > ok guenther sthen (deraadt@) ~ init_sysent.c ~ syscalls.c > sync (deraadt@) ~ kern_pledge.c ~ subr_log.c > sys_osendsyslog can go away; we have transitioned to the new version with > a flags argument > ok guenther sthen (deraadt@) ~ kern_pledge.c ~ kern_sig.c > sys_o58_kill is no longer needed for compat. > ok guenther sthen (deraadt@) ~ sys_process.c > Fix ptrace PT_WRITE_D that returned EFAULT. > Broken in r.1.33. After discussion from kettenis@, don't attempt to > make PT_WRITE_D and PT_WRITE_I equivalent again. > From Mathieu (naabed at poolp dot org) (jca@) net ~ route.c > Plug a route entry leak triggered under memory pressure. > Help to track the leak from Hrvoje Popovski, ok bluhm@ (mpi@) ~ pf.c ~ pf_norm.c > Do not call nd6_output() without route entry argument. > ok sthen@, bluhm@ (mpi@) ~ route.c > Flush dynamic route entries attached to an interface when its link state > becomes DOWN. > This follows the same reasonning as for L2 (cloned) entries. > Hopefully enough to fix tedu@'s stale RTF_DYNAMIC routes when switching > WiFi network during suspend/resume. > ok sthen@ (mpi@) ~ if_ethersubr.c > Ensure that a valid route entry is passed to ether_output() if L2 > resolution is required. > This will allow us to enforce that no route entry is inserted in > the routing table after ether_output(). > This is now possible because if_output() is no longer called with > a NULL route argument. > Tested by Hrvoje Popovski, ok visa@, bluhm@ (mpi@) netinet ~ ip_output.c > Inverse two conditions to not grabe the KERNEL_LOCK for every multicast > packet. > ok visa@, stsp@, sthen@ (mpi@) ~ if_ether.c > Stop creating and inserting a route entry for ARP and ND automagically. > Callers MUST do a route lookup before sending a packet. > Tested by Hrvoje Popovski, ok visa@, bluhm@ (mpi@) netinet6 ~ nd6.c > Stop creating and inserting a route entry for ARP and ND automagically. > Callers MUST do a route lookup before sending a packet. > Tested by Hrvoje Popovski, ok visa@, bluhm@ (mpi@) sys ~ syscall.h ~ syscallargs.h > sync (deraadt@) ~ sockio.h > remove OSIOCGIFCONF. it's dead. ok deraadt guenther mpi (tedu@) == usr.bin =========================================================== 06/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin aucat ~ aucat.c > To store MMC position, use samples (at device rate) as time units > instead of 2400-ths of second. No behaviour change. (ratchov@) ~ aucat.c ~ defs.h > Pass hr:min:sec:frame.cent to dev_mmcloc() directly and remove the > useless intermediate conversion from and to 2400-th of seconds. (ratchov@) ssh ~ kexgexs.c > Ensure that the client's proposed DH-GEX max value is at least as big > as the minimum the server will accept. ok djm@ (dtucker@) == usr.sbin ========================================================== 07/07 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.sbin httpd ~ config.c > Unbreak compilation with -DDEBUG. > From Fabian Raetz <fabian dot raetz at gmail dot com> (jsing@) installboot ~ Makefile ~ i386_installboot.c + armv7_installboot.c > Add support for armv7. Fix an error message in the i386 support before > creating more copies. > ok deraadt@, jsing@ (kettenis@) rebound ~ rebound.c > with the kernel perm check fixed, we can do this kevent after setuid, > but still needs to be before pledge. also check for failure. (tedu@) =============================================================================== _______________________________________________ odc mailing list [email protected] http://www.squish.net/mailman/listinfo/odc
