OpenBSD src changes summary for 2017-01-23
==========================================

bin/csh                                 bin/pax
distrib/arm64                           distrib/miniroot
distrib/sets                            distrib/special
etc/Makefile                            etc/etc.arm64/MAKEDEV
etc/etc.arm64/MAKEDEV.md                lib/libc
lib/libssl                              libexec/getty
libexec/ld.so                           libexec/spamd
regress/lib                             regress/usr.bin
sbin/dhclient                           sbin/pflogd
sbin/route                              share/man
sys/arch/arm/arm                        sys/arch/arm/simplebus
sys/arch/arm64/arm64                    sys/arch/arm64/conf
sys/arch/arm64/dev                      sys/arch/arm64/include
sys/arch/arm64/stand/efiboot            sys/arch/armv7/conf
sys/arch/landisk/landisk                sys/arch/loongson/dev
sys/dev/fdt                             sys/dev/ic
sys/dev/microcode                       sys/dev/pci
sys/dev/wscons                          sys/kern
sys/net                                 sys/netinet
sys/netinet6                            sys/sys
sys/uvm                                 usr.bin
usr.bin/tmux                            usr.sbin/acme-client
usr.sbin/bgpctl                         usr.sbin/bgpd
usr.sbin/eeprom                         usr.sbin/gpioctl
usr.sbin/hotplugd                       usr.sbin/httpd
usr.sbin/iscsid                         usr.sbin/ospf6d
usr.sbin/ospfd                          usr.sbin/tcpdump

== bin =============================================================== 01/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/bin

csh

  ~ lex.c                                 

  > Call isatty() before tcgetattr() in the lex.  This is a little redundant,
  > but avoids doing pledge "tty" operations against flat files.  The shell's
  > tty will be kind of special in the near future. isatty is also a lot less
  > kernel code to run through the loop. (deraadt@)

pax

  ~ ar_io.c                               ~ pax.c

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

  ~ ar_io.c                               ~ pax.c

  > Slow down the churn and continue using old pledge name "ioctl" instead
  > of "tape" for a week or so. (deraadt@)

== distrib =========================================================== 02/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/distrib

arm64

  + Makefile                              + ramdisk/Makefile
  + ramdisk/Makefile.inc                  + ramdisk/install.md
  + ramdisk/list                          

  > Add support for creating RAMDISK kernels for arm64.  Based on
  > the armv7 code, but with the u-boot, device tree and mkuboot
  > bits stripped out.
  > ok deraadt@ (patrick@)

miniroot

  ~ install.sub                           

  > Export AI_MAC and AI_SERVER in addition to AI_MODE so they're visible
  > in the install.site and upgrade.site scripts.
  > OK tb@ (rpe@)

  ~ install.sub                           

  > Clean up possible leftover prefetch area in case of installer restarts.
  > OK tb@ (rpe@)

sets

  ~ lists/base/mi                         ~ lists/man/mi

  > sync (deraadt@)

special

  ~ installboot/Makefile                  

  > Use stubs for arm64 as well.
  > ok phessler@ (patrick@)

== etc =============================================================== 03/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/etc

Makefile

  ~ Makefile                              

  > Interpolate a 'make cleandir' if KEEPKERNELS is set to a non-empty string,
  > right before building kernels. This should unbreak 'make release' for
  > people having this setting.
  > ok deraadt (tb@)

etc.arm64/MAKEDEV

  ~ etc.arm64/MAKEDEV                     

  > regen (kettenis@)

etc.arm64/MAKEDEV.md

  ~ etc.arm64/MAKEDEV.md                  

  > Add /dev/openprom. (kettenis@)

== lib =============================================================== 04/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/lib

libc

  ~ sys/pledge.2                          

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

  ~ sys/pledge.2                          

  > whitespace between macro args and punctuation, and zap a ton
  > of unneccessary quoting; ok deraadt (jmc@)

libssl

  ~ d1_clnt.c                             ~ d1_srvr.c
  ~ s23_clnt.c                            ~ s23_srvr.c
  ~ s3_clnt.c                             ~ s3_srvr.c
  ~ ssl.h                                 ~ ssl_lib.c
  ~ ssl_locl.h                            ~ ssl_sess.c

  > Move the stats struct from SSL_CTX to internal.
  > ok beck@ (jsing@)

  ~ ssl.h                                 

  > Remove heartbeat related fields from the SSL_CTX, which are unused
  > (thankfully). (jsing@)

  ~ s3_clnt.c                             ~ s3_lib.c
  ~ s3_srvr.c                             ~ ssl.h
  ~ ssl_lib.c                             ~ ssl_locl.h
  ~ ssl_sess.c                            

  > Move not_resumable and sess_cert from SSL_SESSION to internal.
  > ok beck@ (jsing@)

  ~ d1_clnt.c                             ~ d1_pkt.c
  ~ d1_srvr.c                             ~ s23_clnt.c
  ~ s23_srvr.c                            ~ s3_clnt.c
  ~ s3_lib.c                              ~ s3_pkt.c
  ~ s3_srvr.c                             ~ ssl.h
  ~ ssl_cert.c                            ~ ssl_lib.c
  ~ ssl_locl.h                            ~ ssl_rsa.c
  ~ ssl_sess.c                            ~ t1_lib.c

  > Move callback function pointers and argument pointers from SSL_CTX to
  > internal.
  > ok beck@ (jsing@)

  ~ bio_ssl.c                             ~ d1_both.c
  ~ d1_clnt.c                             ~ d1_pkt.c
  ~ d1_srvr.c                             ~ s23_clnt.c
  ~ s23_lib.c                             ~ s23_srvr.c
  ~ s3_both.c                             ~ s3_clnt.c
  ~ s3_lib.c                              ~ s3_pkt.c
  ~ s3_srvr.c                             ~ ssl.h
  ~ ssl_cert.c                            ~ ssl_lib.c
  ~ ssl_locl.h                            ~ ssl_sess.c
  ~ t1_lib.c                              

  > move the callbacks from ssl_st to internal
  > ok jsing@ (beck@)

  ~ d1_srtp.c                             ~ s3_clnt.c
  ~ s3_lib.c                              ~ s3_srvr.c
  ~ ssl.h                                 ~ ssl_cert.c
  ~ ssl_lib.c                             ~ ssl_locl.h
  ~ ssl_rsa.c                             ~ ssl_sess.c
  ~ t1_lib.c                              

  > Move most of the fields in SSL_CTX to internal - the ones that remain are
  > known to be in use.
  > ok beck@ (jsing@)

  ~ ssl.h                                 ~ ssl_locl.h
  ~ ssl_sess.c                            

  > Move ex_data, next and prev from SSL_SESSION to internal.
  > ok beck@ (jsing@)

  ~ d1_both.c                             ~ d1_clnt.c
  ~ d1_enc.c                              ~ d1_pkt.c
  ~ d1_srtp.c                             ~ d1_srvr.c
  ~ s23_clnt.c                            ~ s23_pkt.c
  ~ s23_srvr.c                            ~ s3_both.c
  ~ s3_cbc.c                              ~ s3_clnt.c
  ~ s3_lib.c                              ~ s3_pkt.c
  ~ s3_srvr.c                             ~ ssl.h
  ~ ssl_cert.c                            ~ ssl_lib.c
  ~ ssl_locl.h                            ~ ssl_sess.c
  ~ t1_enc.c                              ~ t1_lib.c

  > Move a large part of ssl_st into internal, so we can see what squeals.
  > ok jsing@ (beck@)

  ~ d1_enc.c                              ~ d1_pkt.c
  ~ s3_cbc.c                              ~ s3_pkt.c
  ~ ssl.h                                 ~ ssl_lib.c
  ~ ssl_locl.h                            ~ t1_enc.c

  > move back read_hash and enc_read_ctx into ssl_st. wpa_supplicant and
  > other perversions touches them sickly and unnaturally. (beck@)

  ~ d1_both.c                             ~ d1_clnt.c
  ~ d1_pkt.c                              ~ d1_srvr.c
  ~ s23_clnt.c                            ~ s23_srvr.c
  ~ s3_both.c                             ~ s3_clnt.c
  ~ s3_lib.c                              ~ s3_pkt.c
  ~ s3_srvr.c                             ~ ssl.h
  ~ ssl_lib.c                             ~ ssl_locl.h
  ~ ssl_stat.c                            ~ t1_lib.c

  > send state and rstate from ssl_st into internal. There are accessors
  > so these should not be diddled with directly
  > ok jsing@ (beck@)

  ~ d1_clnt.c                             ~ d1_meth.c
  ~ d1_srvr.c                             ~ ssl.h
  ~ ssl_lib.c                             ~ t1_clnt.c
  ~ t1_meth.c                             ~ t1_srvr.c

  > Remove ssl_ctrl, ssl_ctx_ctrl, ssl_callback_ctrl and ssl_ctx_callback_ctrl
  > from SSL_METHOD, replacing usage with direct calls to the appropriate
  > functions.
  > ok beck@ (jsing@)

  ~ bio_ssl.c                             ~ d1_both.c
  ~ d1_clnt.c                             ~ d1_lib.c
  ~ d1_meth.c                             ~ d1_pkt.c
  ~ d1_srvr.c                             ~ s23_clnt.c
  ~ s23_srvr.c                            ~ s3_both.c
  ~ s3_clnt.c                             ~ s3_lib.c
  ~ s3_pkt.c                              ~ s3_srvr.c
  ~ ssl.h                                 ~ ssl_lib.c
  ~ ssl_locl.h                            ~ ssl_sess.c
  ~ t1_clnt.c                             ~ t1_enc.c
  ~ t1_lib.c                              ~ t1_meth.c
  ~ t1_srvr.c                             

  > Split most of SSL_METHOD out into an internal variant, which is opaque.
  > Discussed with beck@ (jsing@)

  ~ d1_pkt.c                              ~ s23_clnt.c
  ~ s23_srvr.c                            ~ s3_both.c
  ~ s3_lib.c                              ~ s3_pkt.c
  ~ s3_srvr.c                             ~ ssl.h
  ~ ssl_lib.c                             ~ ssl_locl.h
  ~ t1_enc.c                              ~ t1_lib.c

  > Move options and mode from SSL_CTX and SSL to internal, since these can be
  > set and cleared via existing functions. (jsing@)

  ~ ssl.h                                 ~ ssl_lib.c
  ~ ssl_locl.h                            ~ ssl_rsa.c

  > move default_passwd_cb and default_passwd_cb_userdata back into
  > the ssl_ctx from internal - these are used directly by python
  > and openvpn and a few other things - we have the set accessors
  > but the get accessors were added in 1.1 and these roll their
  > own caveat OPENSSL_VERSION chickenpluckery (beck@)

== libexec =========================================================== 05/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/libexec

getty

  ~ main.c                                

  > pledge a tiny bit later.  This results in the specified tty being
  > opened before the first pledge call.
  > testing by tb and benno and others (deraadt@)

ld.so

  ~ aarch64/rtld_machine.c                

  > Fix format string mismatches
  > testing assistance patrick@ (guenther@)

  ~ amd64/rtld_machine.c                  ~ arm/rtld_machine.c
  ~ mips64/rtld_machine.c                 ~ powerpc/rtld_machine.c

  > Fix format string mismatches (guenther@)

  ~ dir.h                                 ~ path.h
  ~ sod.h                                 ~ util.h

  > Wrap function declarations with __{BEGIN,END}_HIDDEN_DECLS to reduce
  > relocations inside ld.so that it has to bootstrap
  > ok kettenis@ (guenther@)

  ~ tib.c                                 

  > The functionality of _dl_allocate_tib() is used internally, so rename
  > it to allocate_tib(), mark that hidden, then make the exported function
  > an alias of that.
  > ok kettenis@ (guenther@)

  ~ dl_printf.c                           ~ dlfcn.c
  ~ resolve.c                             

  > Mark a bunch of stuff static
  > ok kettenis@ (guenther@)

  ~ resolve.c                             

  > Whoops, lost the adjustment of DT_RUNPATH value as a DT_STRTAB value.
  > problem reported by semarie@ (guenther@)

  ~ resolve.c                             

  > remove line that slipped in with rev 1.77 but doesn't belong there; ok
  > sthen@ (naddy@)

spamd

  ~ spamd.c                               

  > Switch spamd to nonblocking descriptors so we can count on getting a
  > short write when the socket buffers are full, instead of blocking.
  > ok benno@ claudio@ (beck@)

  ~ spamd.c                               

  > check EAGAIN in the correct place,
  > noticed and ok phessler@ (beck@)

== regress =========================================================== 06/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/regress

lib

  ~ libssl/unit/cipher_list.c             

  > Fix regress to handle movement of ssl structures to internal (beck@)

  + libc/malloc/malloc_general/Makefile   
  + libc/malloc/malloc_general/malloc_general.c

  > test malloc/realloc/free with some flag combo's (otto@)

  ~ libc/malloc/Makefile                  

  > hook malloc_general to the tests (otto@)

usr.bin

  ~ ftp/redirect.sh                       

  > When run as root, fstat prints the pcb pointer instead of 0x0.
  > Force netcat and ftp to use IPv4.  Unset ftp proxy environment. (bluhm@)

  ~ ftp/redirect.sh                       

  > The ftp program is only verbose by default, if stdin is a tty.  As
  > the test greps for messages that are in the verbose output, add a
  > ftp -v option so that make regress can be run by cron(8). (bluhm@)

  ~ ftp/redirect.sh                       

  > For testing http redirect unset http_proxy, not ftp_proxy. (bluhm@)

== sbin ============================================================== 07/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sbin

dhclient

  ~ dhclient.c                            ~ dispatch.c

  > Keep socket open from startup, and reuse for SIOCGIFFLAGS later on,
  > rather than opening a fresh socket on the fly.
  > ok krw (deraadt@)

pflogd

  ~ privsep.c                             

  > pflogd will need pledge(proc), still disabled because of bfd
  > ok deraadt@ (benno@)

  ~ privsep.c                             

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

route

  ~ route.c                               

  > Zap some bad whitespace. (krw@)

== share ============================================================= 08/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/share

man

  ~ man4/gpio.4                           

  > add sxipio(4) (jsg@)

  ~ man5/bsd.port.mk.5                    

  > quick blurb about the upcoming CLANG_ARCHS variable (phessler@)

  ~ man5/bsd.port.mk.5                    

  > Add LLVM_ARCHS description as well, to make more clear the distinction
  > between it and CLANG_ARCHS.
  > Clang comes before GCC in English, so sort the header itself, and move
  > and move the whole item up. (zhuk@)

== sys =============================================================== 09/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sys

arch/arm/arm

  ~ conf.c                                

  > remove #defines present in MI conf.h; ok kettenis (deraadt@)

arch/arm/simplebus

  ~ simplebus.c                           

  > No need to explicitly use config_activate_children in the cfattach struct.
  > (kettenis@)

arch/arm64/arm64

  ~ conf.c                                

  > Add gpio(4) and switch(4) to cdevsw.
  > ok patrick@ (jsg@)

  ~ pmap.c                                

  > Avoid mapping memory outside kernel VA space.  This happens on machines
  > where we don't load the kernel at the start of physical memory such as
  > the Raspberry Pi 3.
  > ok patrick@ (kettenis@)

  ~ mem.c                                 

  > Synchronize with the arm version of this file. (kettenis@)

  ~ conf.c                                + openprom.c

  > openprom(4) (kettenis@)

  ~ locore.S                              

  > In comparison to _start, where we only need the address of the
  > function, we need another indirection for esym.  The address of
  > esym is simply the address of the variable.  But in the case of
  > esym we want to read and modify its value.
  > ok kettenis@ (patrick@)

  ~ machdep.c                             

  > Initialize DDB if compiled with option DDB.
  > ok kettenis@ (patrick@)

  ~ locore.S                              

  > Create a mapping for the FDT if it happens to be on a different 1 GiB
  > mapping than the kernel. (patrick@)

  ~ machdep.c                             

  > Enter UKC when built with BOOT_CONFIG and -c is specified.
  > ok kettenis@ patrick@ (jsg@)

arch/arm64/conf

  ~ GENERIC                               + RAMDISK

  > Add RAMDISK config for arm64.  While there, clean up spaces in
  > the GENERIC config. (patrick@)

  ~ GENERIC                               ~ files.arm64

  > Add drivers for the Raspberry Pi 3. (kettenis@)

  ~ kern.ldscript                         

  > Remove final align as it confuses lld.
  > ok patrick@ (kettenis@)

  ~ RAMDISK                               

  > Add Raspberry Pi 3 drivers to RAMDISK as well.
  > ok kettenis@ (jsg@)

  ~ GENERIC                               ~ files.arm64

  > openprom(4) (kettenis@)

  ~ GENERIC                               

  > Enable simplefb(4). (kettenis@)

  ~ kern.ldscript                         

  > _end should be on a 64-bit boundary, so make sure to align
  > the end of BSS by 64-bit and not 32-bit.
  > With and ok kettenis@ (patrick@)

  ~ files.arm64                           ~ RAMDISK

  > Make the RAMDISK kernel actually try to boot the ramdisk. (patrick@)

  ~ GENERIC                               ~ RAMDISK

  > Enable USBVERBOSE on both the GENERIC and RAMDISK kernel. (patrick@)

  ~ kern.ldscript                         

  > Remove duplicate $OpenBSD$ tag. (kettenis@)

  ~ GENERIC                               

  > Add bcmdog(4), a watchdog timer for the Raspberry Pi.
  > Based on code written by patrick@ (kettenis@)

  ~ RAMDISK                               

  > Enable bcmdog on RAMDISK as well so it can reboot.
  > ok kettenis@ patrick@ (jsg@)

arch/arm64/dev

  ~ simplebus.c                           

  > No need to explicitly use config_activate_children in the cfattach struct.
  > (kettenis@)

  + bcm2835_dwctwo.c                      + bcm2836_intr.c

  > Add drivers for the Raspberry Pi 3. (kettenis@)

  ~ agtimer.c                             

  > Also attach to "arm,armv7-timer".
  > ok patrick@ (kettenis@)

  ~ simplebus.c                           ~ simplebusvar.h

  > Parse and use "dma-ranges" property to do the appropriate dma address
  > translations on machines like the Raspberry Pi.
  > ok jsg@ (kettenis@)

  ~ simplebus.c                           

  > Add (temporary) workaround for the missing "dma-ranges" property in the
  > device tree provided by the current Raspberry Pi firmware.
  > ok jsg@ (kettenis@)

arch/arm64/include

  ~ conf.h                                + openpromio.h

  > openprom(4) (kettenis@)

arch/arm64/stand/efiboot

  ~ exec.c                                

  > We are supposed to be mapped on a 128 MB block, so skip the first
  > few bits to calculate a valid virtual address for esym.
  > Found with kettenis@ (patrick@)

arch/armv7/conf

  ~ RAMDISK                               ~ GENERIC

  > enable bcmdog on armv7 as well (jsg@)

arch/landisk/landisk

  ~ conf.c                                

  > remove #defines present in MI conf.h; ok kettenis (deraadt@)

arch/loongson/dev

  ~ kb3310.c                              

  > Nuke extraneous whitespace and tabs. (fcambus@)

dev/fdt

  ~ files.fdt                             + bcm2835_dog.c

  > Add bcmdog(4), a watchdog timer for the Raspberry Pi.
  > Based on code written by patrick@ (kettenis@)

dev/ic

  ~ mfireg.h                              ~ mfi.c

  > represent the mbox layout with union instead of an array of bytes.
  > memcpying uint16_ts into inconsistently addressed offsets is hard
  > to read, and this makes future work easier to implement.
  > tested on mfi(4) and mfii(4) (dlg@)

  ~ mfireg.h                              

  > massage some event related details.
  > turn the enum of event argument types into a series of #defines so
  > when you printf the arg type you can search for the number and get
  > an answer without having to count lines.
  > add a struct for extended physical address arguments in event
  > messages.
  > add a list of event types. (dlg@)

dev/microcode

  ~ atmel/Makefile                        ~ kue/Makefile
  ~ rum/Makefile                          ~ tusb3410/Makefile
  ~ udl/Makefile                          ~ zydas/Makefile

  > Build firmware for usb devices on arm64.
  > ok patrick@ (jsg@)

dev/pci

  ~ mfii.c                                

  > represent the mbox layout with union instead of an array of bytes.
  > memcpying uint16_ts into inconsistently addressed offsets is hard
  > to read, and this makes future work easier to implement.
  > tested on mfi(4) and mfii(4) (dlg@)

  ~ mfii.c                                

  > add the mfii opcode for passthru commands (dlg@)

  ~ mfii.c                                

  > store the full 64bits of the sense bufers dva.
  > this is in preparation of reusing the sense buffer for passthru dcmds
  > (dlg@)

  ~ mfii.c                                

  > rework how we address physical disks on the second scsibus.
  > to talk to a physical disk on mfii you need to know its physical
  > disk id and its current dev handle.
  > you can fetch a list of physical disks with a MR_DCMD_PD_GET_LIST
  > command, and you get the dev handle by fetching the associated info
  > from a MR_DCMD_LD_MAP_GET_INFO command.
  > i made a mistake and thought that the index into the PD_GET_LIST
  > info was the target id of a disk, and that should also be used as
  > the index into the LD_MAP_GET_INFO structures. it turns out that
  > the PD_GET_LIST is a dense array of info that points at the target
  > id, and the target id is used as the index into the LD_MAP_GET_INFO.
  > it also turns out that the dev handles from LD_MAP_GET_INFO can
  > change at runtime, so we should update them all when the bus topology
  > changes.
  > this change fetches a list of dev handles via LD_MAP_GET_INFO up
  > front, and makes them availble to the passthru bus scsi_cmd handler
  > via SRP. in the future hotplug events will update the map concurrently
  > with disk accesses.
  > also get rid of the probe for pd disks since the id from PD_GET_LIST
  > and what the scsibus automatically probes are the same. we just let
  > the midlayre blindly probe all possible targets on the passthru
  > bus. the pd scsibus probe handler still fetches specific disk info
  > to ensure we only provide access to disks marked as PASSTHRU in the
  > firmware. (dlg@)

dev/wscons

  ~ wsdisplay_compat_usl.c                

  > Only allow change-setting ioctls when FWRITE is set on the descriptor.
  > ok I forget who (deraadt@)

kern

  ~ kern_pledge.c                         

  > whitespace (deraadt@)

  ~ kern_pledge.c                         

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

  ~ kern_pledge.c                         

  > Backwards compat for pledge "ioctl" for about a week (deraadt@)

  ~ uipc_proto.c                          

  > The function raw_input() has not been called since netiso has been
  > removed in 2004.  The comment about raw_input() above rip_input()
  > was added in 1981, but it is wrong since 1992.  After that it has
  > been copied to rip6_input().  (*pr_input)() is never called with
  > the parameters (mbuf, sockproto, sockaddr, sockaddr).
  > So retire raw_input().
  > OK guenther@ deraadt@ (bluhm@)

  ~ kern_descrip.c                        ~ vfs_syscalls.c

  > Avoid curproc dance in dupfdopen(), by passing a struct proc *
  > ok guenther mpi (deraadt@)

  ~ kern_descrip.c                        

  > Allocate all memory chunks, and potentially sleeping, before freeing
  > the old array of open files.
  > Fix a race for multi-threaded processes reported by [email protected]
  > on bugs@ and analyzed with bluhm@.
  > ok deraadt@, bluhm@ (mpi@)

  ~ uipc_usrreq.c                         

  > File descriptor passing internalizes fd's into an mbuf-stored array of
  > struct file *, and then externalizes back to fd upon delivery.
  > Convert storage to array of struct fdpass, containing struct file *
  > (and soon something else).  memcpy originally intended for alignment
  > purposes can be removed because CMSG_ALIGN is _ALIGN.
  > There is some anxiety over changing this code, but it reads easier.
  > ok mpi guenther kettenis (deraadt@)

net

  ~ rtsock.c                              

  > Zap some bad whitespace. (krw@)

  ~ route.h                               

  > Typo in comment. It's 'rtm_addrs', not 'rtm_addr' (krw@)

  ~ route.h                               

  > Add comment above rtm_flags #define's. (krw@)

  ~ route.h                               

  > Add comment above rtm_type #define's. (krw@)

  ~ art.c                                 ~ art.h

  > Make the art interface a bit more generic by not depending on sockaddr
  > in the functions. This way it can be used for other trees as well.
  > OK mpi@ phessler@ (claudio@)

  ~ if.c                                  ~ if.h
  ~ rtsock.c                              

  > merge the ifnet and ifqueue stats together when userland wants them.
  > a new if_data() function takes a pointer to ifnet and merges its
  > if_data and ifq statistics. it takes the ifq mutex around the reads
  > of the ifq stats so they get a consistent copy.
  > the ifnet and ifq stats are merged because some parts of the stack
  > still update the ifnet counters.
  > ok visa@ (on an earlier diff) mpi@ claudio@ (dlg@)

  ~ if.h                                  ~ if.c
  ~ rtsock.c                              

  > i botched the copyout to ifr->ifr_data in SIOCGIFDATA.
  > this lets pflogd run again.
  > rename if_data() to if_getdata() while here to make grepping for
  > things less noisy.
  > reported by jsg@
  > worked through with deraadt@ (dlg@)

  ~ pf_table.c                            

  > Kill unecessary splsoftnet()/splx() dances, what's protecting radix
  > globals is the KERNEL_LOCK().
  > ok claudio@ (mpi@)

  ~ if.h                                  ~ if_bridge.c
  ~ if_enc.c                              ~ if_pflog.c
  ~ if_gif.c                              ~ if_gre.c
  ~ if_loop.c                             ~ if_mpe.c
  ~ if_mpw.c                              ~ if_pair.c
  ~ if_pfsync.c                           ~ if_pflow.c
  ~ if_etherip.c                          ~ if_vlan.c
  ~ if.c                                  

  > Flag pseudo-interfaces as such in order to call add_net_randomness()
  > only once per packet.
  > Fix a regression introduced when if_input() started to be called by
  > every pseudo-driver.
  > ok claudio@, dlg@ (mpi@)

  ~ pfkey.c                               ~ raw_cb.h
  ~ raw_usrreq.c                          

  > The function raw_input() has not been called since netiso has been
  > removed in 2004.  The comment about raw_input() above rip_input()
  > was added in 1981, but it is wrong since 1992.  After that it has
  > been copied to rip6_input().  (*pr_input)() is never called with
  > the parameters (mbuf, sockproto, sockaddr, sockaddr).
  > So retire raw_input().
  > OK guenther@ deraadt@ (bluhm@)

netinet

  ~ ipsec_input.c                         

  > Assert for IPL_SOFTNET rather than raising SPL recursively.
  > ok benno@ (mpi@)

  ~ ip_carp.c                             

  > Flag pseudo-interfaces as such in order to call add_net_randomness()
  > only once per packet.
  > Fix a regression introduced when if_input() started to be called by
  > every pseudo-driver.
  > ok claudio@, dlg@ (mpi@)

  ~ raw_ip.c                              

  > The function raw_input() has not been called since netiso has been
  > removed in 2004.  The comment about raw_input() above rip_input()
  > was added in 1981, but it is wrong since 1992.  After that it has
  > been copied to rip6_input().  (*pr_input)() is never called with
  > the parameters (mbuf, sockproto, sockaddr, sockaddr).
  > So retire raw_input().
  > OK guenther@ deraadt@ (bluhm@)

netinet6

  ~ raw_ip6.c                             

  > The function raw_input() has not been called since netiso has been
  > removed in 2004.  The comment about raw_input() above rip_input()
  > was added in 1981, but it is wrong since 1992.  After that it has
  > been copied to rip6_input().  (*pr_input)() is never called with
  > the parameters (mbuf, sockproto, sockaddr, sockaddr).
  > So retire raw_input().
  > OK guenther@ deraadt@ (bluhm@)

sys

  ~ pledge.h                              

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

  ~ filedesc.h                            

  > Avoid curproc dance in dupfdopen(), by passing a struct proc *
  > ok guenther mpi (deraadt@)

  ~ unpcb.h                               

  > File descriptor passing internalizes fd's into an mbuf-stored array of
  > struct file *, and then externalizes back to fd upon delivery.
  > Convert storage to array of struct fdpass, containing struct file *
  > (and soon something else).  memcpy originally intended for alignment
  > purposes can be removed because CMSG_ALIGN is _ALIGN.
  > There is some anxiety over changing this code, but it reads easier.
  > ok mpi guenther kettenis (deraadt@)

uvm

  ~ uvm_addr.c                            

  > Move static function to the bottom of the file, to the only user
  > of that function, to guard it with SMALL_KERNEL.
  > ok mpi@ (patrick@)

== usr.bin =========================================================== 10/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin

usr.bin

  ~ Makefile                              

  > Relink opencvs to the build.
  > sure deraadt@ (joris@)

tmux

  ~ Makefile                              ~ client.c
  ~ tmux.c                                ~ tmux.h
  ~ window.c                              + pty.c

  > Open /dev/ptm before pledge() and save it to be used for PTMGET later
  > (this means inlining forkpty()).
  > ok deraadt (nicm@)

== usr.sbin ========================================================== 11/11 ==

  http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.sbin

acme-client

  ~ parse.y                               

  > whitespace (deraadt@)

bgpctl

  ~ bgpctl.c                              

  > cleanup log* functions that break with bgpd reacharound
  > ok claudio@ (benno@)

  ~ Makefile                              

  > Kill the timer.c reacharound it is not needed. (claudio@)

  ~ bgpctl.c                              

  > Mission accomplished, bgpctl is now fatal free. (claudio@)

bgpd

  ~ log.c                                 

  > whitespace
  > ok claudio@ (benno@)

  ~ rde_update.c                          

  > Refactor a bit to make it look a bit more like the rde_update_update code.
  > No functional change, less recursion and indents. (claudio@)

  ~ rde.c                                 ~ rde.h
  ~ rde_rib.c                             

  > Rename struct rib to struct rib_desc. Mechanical change, OK gcc (claudio@)

  ~ rde.h                                 ~ rde_rib.c

  > Now rename struct rib_tree to struct rib. Again OK gcc (claudio@)

  ~ rde.c                                 ~ rde.h
  ~ rde_rib.c                             

  > More rototilling, make rib_new and rib_find return a point to struct
  > rib_desc (claudio@)

  ~ rde.h                                 ~ rde_rib.c

  > Revert the struct rib_tree rename. I need a struct in between because of
  > how struct rib_entry is used. (claudio@)

  ~ rde.c                                 ~ rde.h
  ~ rde_decide.c                          ~ rde_rib.c

  > Introduce a struct rib sitting between struct rib_desc and struct rib_tree.
  > This way the tree becomes a bit better decoupled. (claudio@)

  ~ rde.c                                 ~ rde.h
  ~ rde_rib.c                             

  > Rename rib pointer in struct prefix to re since it points to a rib_entry.
  > While there also remove a comment that is since a few years at least.
  > OK gcc (claudio@)

  ~ util.c                                

  > Make util.c fatal() free by allowing undefined behaviour in prefix_compare.
  > If you pass in crap then you will not get gold back. (claudio@)

eeprom

  ~ Makefile                              

  > Build on arm64. (kettenis@)

gpioctl

  ~ Makefile                              

  > Build gpioctl(8) and hotplugd(8) on arm64.
  > ok patrick@ (jsg@)

hotplugd

  ~ Makefile                              

  > Build gpioctl(8) and hotplugd(8) on arm64.
  > ok patrick@ (jsg@)

httpd

  ~ httpd.c                               

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

iscsid

  ~ iscsid.c                              ~ log.c
  ~ log.h                                 

  > copy log.c/h from bgpd.
  > ok claudio (benno@)

ospf6d

  ~ log.c                                 ~ log.h

  > copy log.c/h from bgpd.
  > ok claudio (benno@)

ospfd

  ~ ospfd.h                               ~ rde.c

  > Remove incomplete "forced nexthop" support.
  > Discussed with claudio@ (jca@)

  ~ log.c                                 ~ log.h

  > copy log.c/h from bgpd.
  > ok claudio (benno@)

tcpdump

  ~ privsep.c                             

  > Split pledge "ioctl" into "tape" and "bpf", and allow SIOCGIFGROUP only
  > upon "inet".  Adjust the 4 programs that care about this. (deraadt@)

===============================================================================
_______________________________________________
odc mailing list
[email protected]
http://www.squish.net/mailman/listinfo/odc

Reply via email to