OpenBSD src changes summary for 2016-04-03 to 2016-04-10 inclusive ==================================================================
distrib/miniroot distrib/sets etc/examples/pkg.conf lib/libc lib/libedit lib/libpcap lib/librthread lib/libsndio libexec/ftpd libexec/spamd regress/usr.bin sbin/bioctl sbin/ifconfig sbin/isakmpd share/man sys/arch/amd64/amd64 sys/arch/amd64/include sys/arch/arm/arm sys/arch/arm/cortex sys/arch/arm/include sys/arch/arm/xscale sys/arch/armv7/armv7 sys/arch/armv7/conf sys/arch/i386/i386 sys/arch/i386/include sys/arch/sparc64/conf sys/dev sys/dev/acpi sys/dev/i2c sys/dev/ofw sys/dev/pci sys/kern sys/net sys/netinet sys/netinet6 sys/sys sys/ufs/ffs sys/ufs/ufs sys/uvm usr.bin/bgplg usr.bin/grep usr.bin/netstat usr.bin/openssl usr.bin/rev usr.bin/ssh usr.bin/systat usr.bin/tcpbench usr.bin/tmux usr.sbin/arp usr.sbin/bgpd usr.sbin/dhcpd usr.sbin/iscsid usr.sbin/ndp usr.sbin/npppd usr.sbin/pppd usr.sbin/radiusd usr.sbin/rcctl usr.sbin/smtpd usr.sbin/tcpdump usr.sbin/vmd usr.sbin/ypldap == distrib =========================================================== 01/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/distrib miniroot ~ dot.profile > Directly assign device name to rootdisk with modified sed pattern. > With help from and OK halex@ > OK krw@ (rpe@) ~ install.sub > Rename getdevname() to get_dkdev_name() to better reflect the actual > purpose and simplify the logic to remove the partition information > from the device name. > With help from and OK halex@, krw@ (rpe@) ~ install.sub > Use _if instead of _ifs as variable name for single interface names. > OK krw@, halex@ (rpe@) ~ install.sub > Use _if for a single interface name and _ifs for a list of interface > names instead of _ifdev and _ifdevs. > OK krw@ halex@ (rpe@) ~ install.sub > Remove _tz as intermediate variable and use CGI_TZ directly. > Feedback from and OK halex@ > OK krw@ for a similar diff (rpe@) sets ~ lists/comp/mi > sync (jsg@) ~ lists/base/mi > sync (jsg@) == etc =============================================================== 02/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/etc examples/pkg.conf ~ examples/pkg.conf > sync (sthen@) ~ examples/pkg.conf > sync (sthen@) == lib =============================================================== 03/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/lib libc ~ stdlib/malloc.3 > Document ``use after free'' error message (otto@) ~ stdio/getdelim.3 > get* can change *lineptr on failure > ok sthen@ (guenther@) ~ include/namespace.h > Update example in comment: setlogin doesn't use {PROTO,DEF}_WRAP() now > (guenther@) ~ asr/res_init.c ~ net/res_random.c ~ stdio/findfp.c ~ stdlib/random.c > Prefer _MUTEX_*LOCK over _THREAD_PRIVATE_MUTEX_*LOCK() when thread-specific > data isn't necessary. > ok mpi@, ok&tweak natano@ (guenther@) ~ sys/munmap.2 > Document (near) standards conformance > based on diff by mmcc@ (guenther@) ~ stdlib/malloc.c > redundant memset(3), from Michael McConville, ok armani@ (otto@) ~ stdlib/malloc.c > tweak MALLOC_STATS printing (switched off by default), prodded by > Michael McConville (otto@) ~ sys/pledge.2 > Clean up and complete the lists of allowed syscalls. > With input from schwarze@ and semarie@. > "go ahead" schwarze@ > (documentation perspective only, without checking factual accuracy) > ok semarie@ on an earlier version (tb@) libedit ~ read.c > Move more fcntl(,F_GETFL,0) -> fcntl(,F_GETFL). > No functional change. > ok guenther@ (krw@) ~ tty.c ~ tty.h > Reset the terminal to its initial state before exiting a program > using libedit, using code from NetBSD tty.c rev. 1.42, 1.48, 1.49, > 1.58, and 1.59. Code mostly by Christos Zoulas, one bug report by > John Hein, one additional bugfix by me (rev. 1.59). > While here, switch t_mode to unsigned char, which is sufficient, > also from NetBSD. > OK martijn@. > Also checked by Christian Heckendorf <mbie at ulmus dot me>. (schwarze@) ~ Makefile ~ chartype.c ~ chartype.h ~ el.c ~ el.h ~ eln.c ~ hist.c ~ hist.h ~ history.c ~ keymacro.c ~ makelist ~ read.c ~ readline.c ~ search.c ~ sys.h ~ terminal.c ~ TEST/Makefile > Always compile with WIDECHAR on and delete that preprocessor switch. > OK martijn@. > Diff also proofread by Christian Heckendorf <mbie at ulmus dot me>. > (schwarze@) ~ chared.c ~ chartype.c ~ chartype.h ~ common.c ~ el.c ~ eln.c ~ emacs.c ~ hist.c ~ keymacro.c ~ map.c ~ read.c ~ refresh.c ~ search.c ~ terminal.c ~ tty.c ~ vi.c > Delete 26 wrapper macros and two wrapper functions that are no > longer needed now that we always compile with wide character support, > reducing code obfuscation. > OK czarkoff@ martijn@. > Diff also proofread by Christian Heckendorf <mbie at ulmus dot me>. > (schwarze@) ~ filecomplete.c ~ vi.c > For pointers, use NULL rather than 0; no functional change; > from Pedro Giffuni via Christos Zoulas. (schwarze@) libpcap ~ pcap.c > Move more fcntl(,F_GETFL,0) -> fcntl(,F_GETFL). > No functional change. > ok guenther@ (krw@) ~ pcap.3 ~ pcap.c ~ pcap.h ~ shlib_version > add two functions from libpcap-1.7.4 which are required by at least > gopacket > ok lteo@ "go for it" dlg@ (jasper@) ~ pcap.3 > bad spell of whether... (jmc@) librthread ~ arch/arm/_atomic_lock.c > replace 0xf57ff05f with dmb sy > spotted by guenther (jsg@) libsndio ~ sio_sun.c > Use the "new" audio(4) api and delete all the useless code to deal > with artificial complications caused by the old api. No behaviour > change. > ok armani, semarie (ratchov@) == libexec =========================================================== 04/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/libexec ftpd ~ ftpd.c > compare pointer to NULL instead of 0 > from fritjof ! alokat.org > ok ratchov@ natano@ (semarie@) spamd ~ spamd.c > per rfc, ehlo responses contain hostname, then options. from dyn+bsd > (tedu@) == regress =========================================================== 05/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/regress usr.bin ~ Makefile + rev/Makefile + rev/exp.ascii.txt + rev/exp.utf8.txt + rev/in.txt > Simple regression tests for rev(1), including UTF-8. > The program will soon be fixed by martijn@. (schwarze@) == sbin ============================================================== 06/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sbin bioctl ~ bioctl.8 > Remove caveat about only supporting 512-byte sectors. (krw@) ifconfig ~ ifconfig.c > move getting the vnetid out next to getting the ifparent > its now separate to getting the tunnel address. > ok mpi@ (dlg@) ~ ifconfig.8 > document autoconfprivacy being the default > ok stsp@ bluhm@ (jsg@) ~ ifconfig.c > move the parent and vnetid stuff around so it builds on ramdisks too. > ramdisk breakage found by jsg@ > ok jsg@ (dlg@) isakmpd ~ message.c > Fix memory leak. Call proto_free() always to free proto. > Diff from Yuuichi Someya. > ok reyk markus (yasuoka@) == share ============================================================= 07/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/share man ~ man4/ath.4 > ath(4) 11g support code is broken and not enabled. Document this > limitation. (stsp@) - man9/time.9 ~ man4/man4.octeon/octrtc.4 ~ man9/Makefile ~ man9/hardclock.9 ~ man9/inittodr.9 ~ man9/microtime.9 ~ man9/ppsratecheck.9 ~ man9/ratecheck.9 ~ man9/resettodr.9 + man9/time_second.9 > Long overdue updates for timecounter changes. > Feedback from jmc@, ok guenther@ for an earlier version. (jsg@) ~ man4/ath.4 > double word; (jmc@) ~ man4/ath.4 > Remove mention of Turbo Mode from ath(4) since the driver doesn't support > it > Discussed with and OK from sthen@ stsp@ kettenis@ jsg@ (mestre@) ~ man7/ports.7 > update the ports link; from solene rapenne (jmc@) ~ man7/ports.7 > Link to faq15.html#Ports instead of just faq15.html. This matches the > DESCRIPTION section that sthen@ fixed in November and reflects the > content of the former faq/ports/ports.html better. > discussed with jmc@ (tb@) ~ man9/mbuf.9 > document m_dup_pkt() (dlg@) ~ man9/mbuf.9 > m_freem returns an mbuf like m_free now > this steals the m_free words to describe what m_freem does now. (dlg@) ~ man9/mbuf.9 > document m_purge. (dlg@) ~ man9/bpf_mtap.9 > document bpf_filter and bpf_mfilter (dlg@) ~ man9/bpf_mtap.9 > may as well put words around bpf_validate while here (dlg@) ~ man9/bpf_mtap.9 > tweak previous; (jmc@) ~ man9/mbuf.9 > bare .Nm gets turned into m_copym2, which is wrong when describing mbufs. > use explicit .Nm mbuf where appropriate. > discussed with jmc@ (dlg@) ~ man4/com.4 > Add octeon. (visa@) == sys =============================================================== 08/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/sys arch/amd64/amd64 ~ db_disasm.c > Teach ddb's disassembler about {rd,wr}[fg]sbase > ok mpi@ (guenther@) ~ machdep.c ~ vm_machdep.c > Move the setting of pcb_fsbase out of reset_segs() into its callers > ok mpi@ (guenther@) ~ vm_machdep.c > Don't call savectx() from cpu_fork(): the registers saved are overwritten > ok kettenis@ (guenther@) ~ vmm.c > Deallocate guest vm_map when the guest gets terminated. > That way we no longer leak uvm data structures after a > guest VM was shut down. > ok mlarkin@ (stefan@) ~ vmm.c > Support processors without unrestricted guest capability. > ok stefan (mlarkin@) ~ vmm.c > define number of exit/entry save/load MSRs as a #define instead of a magic > number. (mlarkin@) arch/amd64/include ~ lock.h > Remove __lockbarrier() defines, unused since cpu_lock removal a few > years ago. > ok dlg@ (jsg@) ~ pcb.h > Move the setting of pcb_fsbase out of reset_segs() into its callers > ok mpi@ (guenther@) ~ vmmvar.h > define number of exit/entry save/load MSRs as a #define instead of a magic > number. (mlarkin@) arch/arm/arm ~ cpuswitch7.S > Remove XSCALE preprocessor checks from a file which is only for ARMv7. > ok jsg@ (patrick@) ~ cpufunc.c ~ cpufunc_asm.S > Add cpu_auxcontrol() to clear and set bits in the implementation/model > specific Auxiliary Control Register (ACTLR). > ok patrick@ (jsg@) ~ cpufunc.c > Set the SMP/coherency bit in ACTLR on Cortex A models it is documented > to exist on. This is required to use ldrex/strex in some cases. > ok patrick@ (jsg@) ~ cpufunc.c > use fixed size unsigned variables for reading/writing the control register > (jsg@) ~ cpufunc.c ~ cpufunc_asm_armv7.S > Read cache line sizes from CP15 Cache Type Register. > Previously we used the primary data cache's information on how big > the cache lines are. The CTR gives us better information about how > big the smallest cache line sizes (controlled by the CPU) are. > ok jsg@ (patrick@) ~ bcopyinout.S ~ copystr.S ~ cpuswitch7.S ~ irq_dispatch.S ~ locore.S > Store curcpu pointer in TPIDRPRW. > This will especially be helpful in future multiprocessor efforts. > ok jsg@ (patrick@) arch/arm/cortex ~ cortex.c > Fix match function so that the cortex bus only attaches if the attach > args are actually looking for the cortex bus. > ok kettenis@ (patrick@) arch/arm/include ~ lock.h > remove unused cpu_lock code (jsg@) ~ cpufunc.h > Add cpu_auxcontrol() to clear and set bits in the implementation/model > specific Auxiliary Control Register (ACTLR). > ok patrick@ (jsg@) ~ armreg.h > Set the SMP/coherency bit in ACTLR on Cortex A models it is documented > to exist on. This is required to use ldrex/strex in some cases. > ok patrick@ (jsg@) ~ armreg.h > Read cache line sizes from CP15 Cache Type Register. > Previously we used the primary data cache's information on how big > the cache lines are. The CTR gives us better information about how > big the smallest cache line sizes (controlled by the CPU) are. > ok jsg@ (patrick@) ~ cpu.h > Store curcpu pointer in TPIDRPRW. > This will especially be helpful in future multiprocessor efforts. > ok jsg@ (patrick@) arch/arm/xscale ~ i80321_intr.c ~ pxa2x0_intr.c > Remove tests for "processing" which was never set. It attempted to > protect against multiple entries of a function that handled pending > soft interrupts. This seems to have been a mistake made when converting > code that used simple lock in the 80321 code which got reused in armv7. > arm softintr_dispatch() already has mutexes around invoked callbacks > which should be enough. Make the pxa2x0 code work the same way which > removes the last remaining simple lock use on arm. > ok patrick@ (jsg@) arch/armv7/armv7 ~ intr.c > Remove tests for "processing" which was never set. It attempted to > protect against multiple entries of a function that handled pending > soft interrupts. This seems to have been a mistake made when converting > code that used simple lock in the 80321 code which got reused in armv7. > arm softintr_dispatch() already has mutexes around invoked callbacks > which should be enough. Make the pxa2x0 code work the same way which > removes the last remaining simple lock use on arm. > ok patrick@ (jsg@) ~ armv7_machdep.c > Map the bootconfig/FDT area passed by u-boot and try to init FDT on it. > If we were passed an FDT, we can completely map the tree and later on > copy it somewhere in our control, so it's not overwritten by anything. > If we are passed a device tree, we need to grab the information we > otherwise would get from the bootinfo struct in another way. We can use > the /memory node to parse the machine's memory information. Bootargs are > stored in /chosen's bootargs property. > This allows us to boot up like before using the board id, but now also > allows us to make use of device tree information. > ok jsg@ (patrick@) ~ autoconf.c ~ armv7.c > Don't try to get a board name string before attaching mainbus with the > platform abstraction. The string is only displayed when attaching the > soc abstractions so this is not needed. When arm mainbus becomes aware > of fdt it can show the /model property itself. (jsg@) ~ autoconf.c > remove an uneeded include (jsg@) arch/armv7/conf ~ files.armv7 > Map the bootconfig/FDT area passed by u-boot and try to init FDT on it. > If we were passed an FDT, we can completely map the tree and later on > copy it somewhere in our control, so it's not overwritten by anything. > If we are passed a device tree, we need to grab the information we > otherwise would get from the bootinfo struct in another way. We can use > the /memory node to parse the machine's memory information. Bootargs are > stored in /chosen's bootargs property. > This allows us to boot up like before using the board id, but now also > allows us to make use of device tree information. > ok jsg@ (patrick@) arch/i386/i386 ~ vm_machdep.c > Don't call savectx() from cpu_fork(): the registers saved are overwritten > and the PCB_SAVECTX flag would be a lie. > ok kettenis@ mpi@ (guenther@) arch/i386/include ~ lock.h > Remove __lockbarrier() defines, unused since cpu_lock removal a few > years ago. > ok dlg@ (jsg@) arch/sparc64/conf ~ GENERIC > For now, disable ahc(4) such that GENERIC.MP kernels don't overflow the 8M > reserved for .text and .rodata. Hopefully I can get rid of this limit > soon. > propmpted by guenther@ (kettenis@) dev ~ softraid.c ~ softraid_concat.c ~ softraid_crypto.c ~ softraid_raid0.c ~ softraid_raid1.c ~ softraid_raid5.c ~ softraid_raid6.c ~ softraidvar.h > Enable creation of softraid volumes using disks with non-512 byte > sectors. Volumes created will present a sector size equal to the > largest sector size of the constituent disks. > Softraid Metadata version cranks to 6 due to new field. > ok jsing@ with tweaks that will follow soon. (krw@) dev/acpi ~ dwiic.c > Fix layer violation in the ihidev(4) code by implementing a generic > mechanism > that allows the i2c controller implementation to establish interrupts on > behalf of i2c slave device drivers. Use this mechanism in dwiic(4) to let > it configure the right acpi interrupt (global or gpio). Change the level > to > IPL_TTY as this is the appropriate level to use for keyboards and other > input devices. > ok jsg@ (kettenis@) dev/i2c ~ i2cvar.h ~ ihidev.c > Fix layer violation in the ihidev(4) code by implementing a generic > mechanism > that allows the i2c controller implementation to establish interrupts on > behalf of i2c slave device drivers. Use this mechanism in dwiic(4) to let > it configure the right acpi interrupt (global or gpio). Change the level > to > IPL_TTY as this is the appropriate level to use for keyboards and other > input devices. > ok jsg@ (kettenis@) dev/ofw ~ fdt.c > The callers of fdt_next_node() expect it to return NULL if a next node > cannot be found. Due to implementation details, it used to return > a pointer to the next node, without checking if it actually exists. > Modify fdt_next_node() to only return a next node if it actually exists > and implement an internal skip_node() function that keeps the previous > behaviour. > ok mpi@ (patrick@) ~ fdt.c ~ fdt.h > Implement a way to extract memory information from a node. This can be > used to extract information about how much RAM is available or where a > controller is in the address space and how big its IO space is. > Add a way to get the device tree's size. This is helpful to find out > how much you need to copy if you want to move the tree somewhere else. > Also add helpers to easily retrieve integers from a property. > ok jsg@ (patrick@) ~ fdt.c > Remove erroneous whitespace added in previous change. (patrick@) ~ fdt.c > Add support for bus space translation. > The ranges property can be used to translate from one address space > to another. This means that we cannot just use a controllers's reg > property to get the physical address, but have to use its parent's > ranges attribute to translate the address to another. > ok kettenis@ (patrick@) ~ fdt.c > Convert memory extract routine to return errno-based errors. > Also check that mem is not NULL, as its part of the FDT API > and should make sure it's not accessing a null pointer. > ok bmercer@ (patrick@) dev/pci + drm/drm_linux_rbtree.h > Add a linux-compatible rbtree interface on top of our implementation in > <sys/tree.h>. From FreeBSD/OFED via DragonFlyBSD. > ok jsg@ (kettenis@) ~ drm/drm_atomic.h > Add atomic_add_unless(), another strange linux interface. > ok jsg@ (kettenis@) ~ drm/drm_linux.h + drm/drm_vma_manager.c + drm/drm_vma_manager.h > Add vma offset manager code. > ok jsg@ (kettenis@) ~ drm/drmP.h ~ drm/drm_drv.c ~ drm/drm_linux.c ~ drm/drm_modes.c ~ drm/files.drm ~ drm/i915/i915_gem.c ~ drm/ttm/ttm_bo_vm.c + drm/drm_gem.c > Split out the generic GEM code (like Linux did) and switch it over to > the vma offset manager. This brings us a little bit more isolation between > applications as GEM buffers are now tied to a specific /dev/drmX clone. > ok jsg@ (kettenis@) ~ drm/drmP.h > Remove completely out-of-date comment about locking protocol. (kettenis@) ~ drm/drm_linux.c ~ drm/drm_linux.h > Add an implementation of the Linux "idr" API. (kettenis@) ~ drm/drmP.h ~ drm/drm_gem.c ~ drm/i915/i915_gem_execbuffer.c > Now that we the "idr" API, we can get rid of a lot of > #ifdef __linux__ > ... > #else > ... > #endif > code in the generic drm GEM code. (kettenis@) ~ drm/drm_gem.c ~ drm/drm_vma_manager.c ~ drm/drm_vma_manager.h > Add $OpenBSD$ markers. (kettenis@) ~ drm/i915/i915_drv.h ~ drm/i915/i915_gem.c ~ drm/i915/i915_gem_context.c > Use the "idr" API here as well, reducing the diffs to Linux. (kettenis@) ~ drm/drm_crtc.c ~ drm/drm_crtc.h > Another "idr" conversion. (kettenis@) ~ drm/drmP.h > Remove some pre-idr leftovers. (kettenis@) ~ drm/drm_crtc.c ~ drm/drm_crtc.h ~ drm/drmP.h > Revert the drm_crt.c "idr" conversion as it breaks X on (at least) the > T430s. (kettenis@) ~ drm/drm_linux.c > Return -ENOSPC if idr_alloc() fails to allocate an unused id instead of > spinning forever. (kettenis@) ~ drm/drm_crtc.c ~ drm/drm_crtc.h ~ drm/drmP.h > Retry the drm_crtc.c "idr"conversion. Turns out the xf86-video-intel > driver > is buggy and trucates the ids to 8 bits. So specifymaximum in the > idr_alloc() > call until that gets fixed. (kettenis@) ~ drm/drmP.h ~ drm/drm_crtc.c ~ drm/drm_drv.c ~ drm/i915/i915_drv.c ~ drm/i915/i915_gem_execbuffer.c ~ drm/radeon/radeon_kms.c > Get rid of some infrastrcuture that is now obsolete and synchronize some of > the data structures in drmP.h with Linux 3.14. > ok jsg@ (kettenis@) ~ pcidevs > Add Intel Bay Trail EHCI (as found on 1st gen Compute Stick) (reyk@) ~ pcidevs.h ~ pcidevs_data.h > sync (reyk@) kern ~ vfs_subr.c ~ spec_vnops.c > Increase size of the clone bitmap (revised diff after revert). I have > tested this with fuse _and_ drm on amd64 and macppc. Also tested with > cloning bpf (not in the tree) on macppc. > ok mikeb > "looks correct to me" millert > The original commit message is as follows: > Increase size of the clone bitmap. A limit of only 64 device clones > turned out to be too low for the upcoming work on cloning bpf. The new > limit is 1024 device clones. As part of the size increase, the bitmap > has been changed to be allocated separately to avoid bloating all device > nodes, as suggested by guenther, millert and deraadt. > ok millert mikeb (natano@) ~ uipc_mbuf.c > correct the order of arguments to m_get in m_dup_pkt (dlg@) ~ vfs_subr.c > Share clone bitmap between aliased vnodes. This prevents duplicate clone > instance numbers being handed out for the same minor device. > ok mikeb (natano@) ~ uipc_mbuf.c > add m_purge for freeing a list of mbufs linked via m_nextpkt > this tweaks m_freem so it returns the m_nextpkt from the mbuf it freed, > like how m_free returns the m_next from the mbuf it frees. > ok mpi@ (dlg@) net ~ bpf.h > mark the packet buffer in bpf_filter as bounded by the pktlen arg > ok guenther@ (dlg@) ~ if_vlan.c > turns out EVL_ENCAPLEN is 4, not 18 or sizeof(*evl). (dlg@) ~ pf_osfp.c > Add sys/queue.h where it's needed. Unbreak userland following recent > removal from mbuf.h. ok mpi@ (sthen@) ~ pf.c > Instead of panicking if an mbuf(9) already has a statekey dump its > content and unlink the statekey. > This should allow us to find the reminding corner cases of packets > looped back in the stack. > ok dlg@ (mpi@) netinet ~ in_pcb.c ~ in_pcb.h > Move inp_laddr assignment after in_pcbpickport(), extend in_pcbpickport() > as needed. > Ok bluhm@ (vgross@) ~ in_pcb.c > Move reserved port checks from in(6)_pcbaddrisavail() to in_pcbbind(). > Kill old comments while at it. > Ok mpi@ bluhm@ (vgross@) ~ in_pcb.c > We must always check if the socket is already bound, and always set > INPLOOKUP_IPV6 when applicable. Bug introduced in revision 1.197 by > yours truly. > Ok bluhm@ (vgross@) netinet6 ~ in6_pcb.c > Move reserved port checks from in(6)_pcbaddrisavail() to in_pcbbind(). > Kill old comments while at it. > Ok mpi@ bluhm@ (vgross@) sys ~ mbuf.h > mbuf.h doenst need queue.h. > move malloc.h down to where its used, and inside _KERNEL. > ok deraadt@ mpi@ (dlg@) ~ specdev.h > Increase size of the clone bitmap (revised diff after revert). I have > tested this with fuse _and_ drm on amd64 and macppc. Also tested with > cloning bpf (not in the tree) on macppc. > ok mikeb > "looks correct to me" millert > The original commit message is as follows: > Increase size of the clone bitmap. A limit of only 64 device clones > turned out to be too low for the upcoming work on cloning bpf. The new > limit is 1024 device clones. As part of the size increase, the bitmap > has been changed to be allocated separately to avoid bloating all device > nodes, as suggested by guenther, millert and deraadt. > ok millert mikeb (natano@) ~ mbuf.h > add m_purge for freeing a list of mbufs linked via m_nextpkt > this tweaks m_freem so it returns the m_nextpkt from the mbuf it freed, > like how m_free returns the m_next from the mbuf it frees. > ok mpi@ (dlg@) ufs/ffs ~ ffs_vnops.c > Always call bread_cluster() instead of calling it only if the current > logical block is contiguous to the previous one. > This logic is a left-over of the pre-bread_cluster() area. When the > read-ahead version of bread(9) was used to prefetch blocks. Nowadays > bread_cluster() do the right thing (tm). > ok stefan@ (mpi@) ufs/ufs ~ ufs_dirhash.c > Remove sparc special-casing from ufsdirhash_init(). This is not required > anymore since the kernel VM space increase work done in sparc about one > year ago. > from Miod Vallat; thanks! > ok tobiasu (natano@) uvm ~ uvm_amap.c ~ uvm_amap.h ~ uvm_extern.h ~ uvm_map.c ~ uvm_unix.c > UVM_FLAG_AMAPPAD has no effect anymore, nuke it. > This flag caused amaps to be allocated with additional spare slots, to > make extending them cheaper. However, the kernel never extends amaps, > so allocating spare slots is pointless. Also UVM_FLAG_AMAPPAD only > has an effect in combination with UVM_FLAG_OVERLAY. The only function > that used both flags was sys_obreak, but that function had the use of > UVM_FLAG_OVERLAY removed recently. > While there, kill the unused prototypes amap_flags and amap_refs. > They're defined as macros already. > ok mlarkin@ kettenis@ mpi@ (stefan@) == usr.bin =========================================================== 09/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.bin bgplg ~ bgplg.c > Use SERVER_NAME for bgplg's hostname (displayed on the page), this is > passed > from the webserver so is more likely to be the correct name to expose, and > is > easier to adjust if needed. OK claudio@ florian@ (sthen@) grep ~ util.c > Reverse search optimization makes no sense (and doesn't work) if > we are looking for all matches in a line; ok natano@ miilert@ tedu@ (otto@) netstat ~ mbuf.c > Add sys/queue.h where it's needed. Unbreak userland following recent > removal from mbuf.h. ok mpi@ (sthen@) openssl ~ enc.c > hexidecimal->hexadecimal; from mmcc > ok beck (jmc@) rev ~ rev.c > Enable UTF-8 support in rev. > Some minor cleanups while here. > OK schwarze@ (martijn@) ssh ~ misc.c ~ misc.h ~ readconf.c > don't record duplicate LocalForward and RemoteForward entries; > fixes failure with ExitOnForwardFailure+hostname canonicalisation > where the same forwards are added on the second pass through > the configuration file. bz#2562; ok dtucker@ (djm@) ~ PROTOCOL > We accidentally send an empty string and a zero uint32 with every > [email protected] channel open, in contravention of > our own spec. > Fixing this is too hard wrt existing versions that expect these > fields to be present and fatal() if they aren't, so document them > as "reserved" fields in the PROTOCOL spec as though we always > intended this and let us never speak of it again. > bz#2529, reported by Ron Frederick (djm@) ~ sftp.c ~ sftp-client.c > whitespace at EOL (djm@) ~ authfile.c ~ sshkey.c > make private key loading functions consistently handle NULL > key pointer arguments; ok markus@ (djm@) systat ~ mbufs.c > Add sys/queue.h where it's needed. Unbreak userland following recent > removal from mbuf.h. ok mpi@ (sthen@) tcpbench ~ tcpbench.c > Lose the pointless third paramater to fcntl(F_GETFL). Using SOCK_NONBLOCK > instead of fcntl() dance would significantly change designed semantics. > ok djm@ (krw@) tmux ~ xmalloc.c > fatalx() not fatal(). (nicm@) == usr.sbin ========================================================== 10/10 == http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/usr.sbin arp ~ arp.8 > Describe the format used by -s once, and refer to it when documenting -f. > Discussed with jmc@ (jca@) bgpd ~ kroute.c > When bgp receives an RTM_CHANGE message for an RTF_MPATH route[1], and > it finds it actually on an interface, bgpd can get confused. In this > case, just log it and treat it as an RTM_ADD so bgpd doesn't suicide. > [1] includes all ospfd installed routes, btw. > OK benno@ sthen@ (phessler@) dhcpd ~ udpsock.c > In udpsockmode codepath dhcpd.c had the promise route for pledge(2) but it > was > missing from udpsock_startup() (udpsock.c) which happens earlier and > therefore > would abort the program due to SIOCGIFADDR ioctl(2) on udpsock_handler > Reported by Philip Higgins <phil ! unita.com.au> > OK tb@ (mestre@) iscsid ~ control.c > Remove the iterator for control connections unfinished and unused. > ok claudio (yasuoka@) ndp ~ ndp.8 ~ ndp.c > Hook up ndp -f. > Initial diff from Dimitris Papastamos. Support from mikeb@, ok benno@. > (jca@) ~ ndp.8 ~ ndp.c > etheraddr -> ether_addr, as in arp(8); discussed with jmc@ (jca@) ~ ndp.8 > Pointless commented-out lines; discussed with jmc@ (jca@) ~ ndp.8 > Consistent v6-speak: hostname -> nodename, proxy NDP -> ND proxy > Discussed with jmc@ (jca@) ~ ndp.8 > Describe the format used by -s once, and refer to it when documenting -f. > Discussed with jmc@ (jca@) npppd ~ npppd/npppd_subr.c > Move more fcntl(,F_GETFL,0) -> fcntl(,F_GETFL). > No functional change. > ok guenther@ (krw@) pppd ~ chat/chat.c > Move more fcntl(,F_GETFL,0) -> fcntl(,F_GETFL). > No functional change. > ok guenther@ (krw@) radiusd ~ radiusd_module.c > Move more fcntl(,F_GETFL,0) -> fcntl(,F_GETFL). > No functional change. > ok guenther@ (krw@) rcctl ~ rcctl.sh > Use builtin. (ajacoutot@) smtpd ~ smtpd.conf.5 > arguments to "chain" are space separated, not comma; verified by jung > diff from david+bsd > i also removed Op, since ... is enough/ (jmc@) tcpdump ~ print-enc.c ~ print-pflog.c > Add sys/queue.h where it's needed. Unbreak userland following recent > removal from mbuf.h. ok mpi@ (sthen@) vmd ~ loadfile_elf.c ~ virtio.c ~ vmd.h ~ vmm.c > Directly use physical addresses from ELF header for kernel loading. > This allows us to remove the 'do_mask' parameters in read_mem and > write_mem as well as the address mask operaton itself. > ok mlarkin@ (stefan@) ~ loadfile.h ~ loadfile_elf.c ~ vmm.c > Support processors without unrestricted guest capability. > ok stefan (mlarkin@) ~ vmm.c > Remove headers associated with code that's been moved to other .c files > ok mlarkin@ (guenther@) ~ loadfile_elf.c > Place a BOOTARG_END section at the end of the boot arguments list pushed > to the VM during boot. i386 guest kernels need this marker to boot, and > with this diff, it is possible to install and run an OpenBSD i386 guest > VM using vmm(4). Note that i386 guests require a host CPU that supports > unrestricted guest mode (eg, post-Nehalem) for the time being, hopefully > this can be addressed later. (mlarkin@) ypldap ~ ber.c > Apply revisions 1.4 and 1.5 from ldapd's ber.c to ypldap's copy, so it can > deal with messages that haven't been fully read from the server yet. > Not needed yet, but will be soon. > ok dlg@ (jmatthew@) ~ ypldap_dns.c ~ ypldap.h ~ ldapclient.c > convert ypldap_addr list to a tailq > ok dlg@ (jmatthew@) =============================================================================== _______________________________________________ owc mailing list [email protected] http://www.squish.net/mailman/listinfo/owc
