OpenBSD -current changelog
This selection is intended to include all important
and all user-visible changes.
For a complete record of all changes, please see the "source-changes"
mailing list, called "OpenBSD CVS"
in the archives,
or use CVS.
Note: Problems for which patches exist are marked in red.
Changes made between OpenBSD 5.4 and -current
- Stir PRNG after sshd(8) runs post-accept fork. The child gets a different PRNG state anyway via rexec and explicit privsep reseeds, but it's good to be sure.
- Do not let netstart(8) create a route to alias addresses via 127.0.0.1 (the stack is able to tell if the address is local or not).
- Correct format string mismatches in various code turned up by gcc(1) -Wformat=2.
- Added acpi(4) indicator sensor to acpibtnX, to reflect lid status when available. Useful for people not using machdep.lidsuspend.
- With wsdisplay(4), use the console locator to make sure wsdisplay0 is always the glass console on i386/amd64.
- Use int64_t in the spamdb(8) greylist db file, with backwards compatibility for records with 32-bit times.
- Fixed build of ix(4) without vlan.
- Added urtwn(4) to the sgi kernels where it makes sense.
- Added nc(1) -F flag to enabled fd-pass mode: establish connection and pass connected socket to stdout.
- Remove uscanner(4). Disabled in GENERIC long ago (when libusb became the preferred solution for USB scanners).
- Enabled mpii(4) on sparc64.
- Added ssh_config(5) ProxyUseFDPass option, to establish a connection and then pass file descriptor back to ssh(1).
- Make pf(4) "match quick" abort ruleset evaluation with the current block/pass state, rather than continuing on.
- Fixed kernel profiling on powerpc architectures (broken since the buffers were made per-cpu).
- Fixed prototypes of do_load_half() and do_load_byte() on m88k to stop obscure bugs in awk(1) and elsewhere.
- Disable appleagp(4). It doesn't really work yet, and hurts certain iBook G3 models (PowerMac4,3).
- Fixed xf86-video-ati shadowfb on big-endian machines.
- Updated to: xload 1.1.2, xprop 1.2.2, xset 1.2.3, libXfont 1.4.6 and xwd 1.0.6.
- Stop dhclient(8) creating a route to the bound address via 127.0.0.1, as the stack should be able to correctly short-circuit packet routing to local addresses without this.
- Make the vgafb(4) console correctly restore when exiting X(7).
- Stopped macppc pretending to support non-console devices, because vgafb(4) only matches the framebuffer device setup by the firmware.
- Use %lld and cast to long long when passwd(1) is printing time_t values, and atoll(3) when parsing them.
- For ikectl(8), snmpctl(8) and rdate(8) use %lld and cast to long long when printing time_t values.
- Enabled radeondrm(4) on macppc.
- Use mapiodev to do the initial mapping of vgafb(4). Allows radeondrm(4) to bus_space_map(9) it without panicking.
- Updated to xf86-video-ati 7.2.0. Fixes endianess problem with kms.
- Avoid integer overflow in sector calculation reading UDF DVDs, which broke reads past 2GB.
- On luna88k, check number of planes (1, 4, or 8bpp) in framebuffer by ourselves (PROM gives false values). Lets monochrome X(7) server work correctly with a 4bpp framebuffer.
- Enabled radeondrm(4) on sparc64.
- Align stack yp(8) buffers to 64-bit boundaries now, for they might contain a struct passwd(5).
- Blast uhts(4), as the code was merged into ums(4) last week.
- Stop defining SIOC{S,G}ETVLAN as SIOC{S,G}IFGENERIC. Lets vlan(4) handle pppoe(4)-specific ioctls and vice versa without smashing the stack of the caller.
- Make sure "drm.h" gets generated on all architectures that support pci(4), making alpha and loongson compile again.
- Set bgpd.conf(5) to filter the IPv6 Discard-Only prefix, because it should not be globally routable (see RFC6666).
- Rewrite the internals of the bgpd(8) RDE reload logic. First step to make reload non-blocking. Speeds up the reload time, should also fix a few edge cases on reloads.
- Enforce that the ls_id and the adv_rtr field in a type 1 router LSA are the same. This will drop updates trying to attack the ospfd(8) LSDB.
- Only skip dhclient(8) address deleting/adding and route deleting/adding on identical lease when we are in state S_RENEWING. Makes sure we end up with an IP address on the interface.
- Don't depend on a list of archs to determine endianess, unbreaks X(7) gallium build on sparc64.
- Let fw_update(1) handle firmware for radeondrm(4).
- Allow the user to specify which firmware to install with fw_update(1), or to install all of them using the -a switch.
- Updated iceauth to 1.0.6 and xfwp to 1.0.3.
- Let the Xserver(1) handle more /dev/drmN devices.
- Improved pci(4)'s VGA arbiter support now the kernel can give us information about the resources required by a particular device.
- Moved ray(4) to the attic.
- When handling puts from a client, tftpd(8) will now close the file once downloaded, instead of waiting until after we clean up the client .
- Remove bogus refcounting of usb(4) pipes, to avoid memory leaks and potentially aborting a pipe when requests are still pending.
- Made the tree compile again on architectures without drm(4).
- Switch time_t, ino_t, clock_t, and struct kevent's ident and data members to 64bit types; build perl(1) with -DBIG_TIME.
- Add the tcp(4) socket option TCP_NOPUSH to delay sending the stream. Useful to aggregate data in the kernel from multiple sources, avoids sending small packets.
- ps(1) flags are split between thread and process but it is useful for ddb(4) to show both. Show both in ps overview, list them explicitly for the specific thread.
- Match macppc framebuffer device against its OpenFirmware node, to enable sharing of console detection logic with sparc64 in radeondrm(4).
- Build mesa radeon(4) drivers on sparc64; switch to gallium drivers for r300 and r600.
- Enabled xf86-video-ati kernel modesetting support.
- Added TTM and radeon(4) DRM code. Includes kernel modesetting, framebuffer console and support for newer hardware.
- Fixes for as(1) on mvme88k: provide 88110 syntactic sugar for the control register names; correctly handle -m options; allow the register prefix to correctly be recognised in front of the condition codes.
- Cancel timeouts on pflow(4) interface destruction.
- Plugged file descriptor and memory leak in fuse(4)'s fuse_open() error path.
- Fixed a potential race on luna88k, where a secondary cpu incorrectly resets variables on the primary cpu.
- Merged uhts(1) into ums(1).
- Added rc.d(8) scripts for ipropd-master and ipropd-slave.
- Added urtwn(4) for upgrades to macppc ramdisk.
- Enabled ctrl-left-arrow and ctrl-right-arrow to move forward/back a word in ssh(1); matching ksh(1)'s recent change.
- Do getopt parsing for all sftp(1) commands, to ensure consistent behaviour.
- Implemented the roff(7) font-escape sequence \f(BI "bold+italic".
- Made it possible for graphics drivers to opt-out of VGA arbitration.
- Neither the flow control diff nor the RX checksum offload diff is responsible for gem(4) hangs, so put them back.
- Bugfix update to nginx(8) version 1.4.2.
- Build libdrm_radeon on sparc64.
- Create drm(4) device nodes on sparc64 and create additional entries on amd64/i386/macppc for running with multiple cards.
- Fixed sftp(1) regression: symlinking a file would incorrectly canonicalise the target path (bz#2129).
- Disabled libexec/identd and enable usr.sbin/identd(8).
- After a panic(9), do not fill up the dmesg(8) with splassert messages. They scroll away the important information.
- Another major overhaul of the inteldrm(4) GEM code.
- Make sure states learnt via pfsync(4) from a peer with the same ruleset checksum get properly assigned to rules.
- Stopped gcc(1) warning on missing newline at EOF. Matches the behaviour of gcc >= 4.3 and clang.
- Moved check in mktemp(1) for template length, so error message is also shown in the -t case instead of EINVAL.
- Make user(8) write UIDs/GIDs as unsigned in the passwd file and error messages. Matches what pw_scan(3) expects.
- Switched vax to gcc(1) version 3.3.6; removed gcc(1) version 2.95 from the tree.
- Fixed mandoc(1) bug by reverting r1.10 of term.c: after a leading blank on an output line, don't generate a premature line break.
- Updated ix(4) to the Intel/FreeBSD current version. Brings support for the flow control and additional (untested) bits for 1G fibre versions.
- Enabled MSI on re(4) RT810xE.
- Unbreak PMTU-discovery for ipsec(4) AES-GCM.
- Don't flush the cache on page inflate for hibernate on amd64 for faster resume on some machines.
- Fixed corner cases in the implementation of pow(3) to make it compatible with C99.
- Implement if_nametoindex(3) and if_indextoname(3) for pfctl(8). 20x speedup factor with some complex rulesets.
- Create more inodes on the sparc64 installation ramdisk. Prevents problems during install/upgrade when files or nodes need to be added.
- Added ral(4) support for the Ralink RT3060, which uses the RT3572 MAC/BBP.
- Remove the sysmerge(8) glue from the installer (it has never been used and is being replaced).
www@openbsd.org
$OpenBSD: plus.html,v 1.1300 2013/08/30 00:03:12 brett Exp $