===================================================================
RCS file: /cvsrepo/anoncvs/cvs/www/plus.html,v
retrieving revision 1.1437
retrieving revision 1.1438
diff -u -r1.1437 -r1.1438
--- www/plus.html 2019/05/05 16:17:17 1.1437
+++ www/plus.html 2019/05/08 23:10:43 1.1438
@@ -83,6 +83,66 @@
+
+- Removed ld.so-local stack-protector handling functions on retguard systems, as hard-traps are used.
+
- Set the REQUEST_URI CGI variable to the requested URI and query string instead of the rewritten ones in httpd(8).
+
- Fixed integer overflow with an excessively large number of kqueue(2) events.
+
- Added support for more Intel 300 Series PCH to ichiic(4).
+
- Moved bgpd(8) pfkey socket to the parent process. The refreshing of the keys is done whenever the session state changes to IDLE or ACTIVE, which should behave better when reloading configs with auth changes.
+
- Fixed a bug in pkg_info(1) -S where version elements were ignored.
+
+
- Ensured mcx(4) completion queues are always rearmed to prevent rx or tx completion stalling.
+
+
- Adjusted unwind(8) to try to resolve the DNSSEC trust anchor only if we have a validating resolver context.
+
- To restore lockf(3) detection, introduced a list for all pending blocked locks to be scanned before waiting on a blocking lock in order to determine whether sleeping would cause a deadlock.
+
- Enforced store/load order when setting or clearing AST flag on mips64, preventing an unlikely case with inter-CPU ASTs where the receiving CPU uses stale state. Ensured that the clearing store is performed before other memory accesses, preventing potential loss of an AST request.
+
- Issued a write-write barrier before sending IPI on mips64, preventing a receiving CPU from observing an old state when processing the interrupt.
+
- Made the interrupt and trap return paths check for ASTs with interrupts disabled, fixing unintentional delay of ASTs on MP mips64.
+
- Fixed reloading of network statements that have no fixed prefix specification in bgpd(8).
+
- Fixed a logic error when configuring the alc(4) driver to use msi.
+
+
- Fixed the case where switchd(8) does not save a copy of a packet needing forwarding.
+
- Added support for the EFI Random Number Generator Protocol, using it to XOR random data into the buffer we feed the kernel for amd64.
+
+
- Relaxed a check in knote(9) to prevent a panic without KERNEL_LOCK().
+
- Began to assume eDP is present on port A when there is no VBT, making inteldrm(4) work on a Gemini Lake system (HP Stream Laptop 14-cb1XX) where finding the VBT fails.
+
- Improved the interaction among efifb(4), inteldrm(4) and radeondrm(4) for a serial console by introducing a 'primary' device, the one set up and used by firmware. This should help X to work correctly out of the box with multiple cards or with a serial console.
+
- Added mcx(4) driver for Mellanox ConnectX-4 (and later) Ethernet controllers.
+
+
- Began to allow panes to be empty in tmux(1), allowing output to be piped to them with split-window or display-message -I.
+
- Adjusted installer to remember 'autoconf' if install restarts.
+
- Added a 'forwarded' log format extending the 'combined' log format for httpd(8), allowing tracking of request origins behind a proxy. This format is compatible with log analyzers like GoAccess and Webalizer.
+
- Fixed endless loop / OOB write on 64 bit systems with large buffers.
+
- Improved man(1) matching for requests for a specific section.
+
- Renamed the -c option in sysupgrade(8) to -s to indicate snapshots, and implemented a -r option to force an upgrade to the next release.
+
- Added a -n option to sysupgrade(8) for a no-reboot option.
+
- Adjusted sysupgrade(8) to only fetch and upgrade if a new snapshot is available.
+
+
- Avoided opening httpd(8) log files on 'no log,' preventing startup failures where the log/directory is missing but logging is disabled anyway.
+
- Avoided running the activate function for a partially-attached ehci(4) driver, preventing a panic during suspend for the Realtek DASH ehci.
+
- Added support for the cryptographic coprocessor found on newer AMD Ryzen CPUs/APUs.
+
- Allowed non-root users to become master when they are the first to open a drm(4) device.
+
- Increased datasize and maxproc for the pbuild class in login.conf(5).
+
+
- Used the proper UAC-v2.0 request in uaudio(4) to read the device controls, fixing STALL errors in mixer requests causing the mixer to be unusable.
+
- Protected tun(4) wakeup with KERNEL_LOCK, rather than NET_LOCK.
+
+
- Pulled cdfs and cdrom production into the full ramdisk build for i386.
+
- Added a 'set_rate' method to the envy(4) codec API, allowing card-specific codec initialization code called when the sample rate is changed. Implemented this method for ESI Juli@ cards, switching the AK5385 and AK4358 chips among single-, double- and quad-speed modes depending on host sample rate and suppressing aliasing noise.
+
- Enabled mvmdio(4) and mvneta(4) on arm64.
+
- Enabled EnvyHT-specific sample rates (above 96kHz) on the host controller for envy(4) devices.
+
- Added support for the Armada 3720 pinctrl controller to mvpinctrl(4). This controller also includes GPIO controller functionality.
+
- Added mvuart(4) to support the Armada 3720's serial console.
+
- Added support for the Armada 3720 clocks to mvclock(4) and added mvuart(4) to support the serial console.
+
- Allowed enabling of regulators with the 'regulator-always-on' property.
+
- Added unveil(2) to chpass(1).
+
- Fixed user options crash in tmux(1).
+
- Began the process of making Makefiles more similar across architectures.
+
- Restricted hotplugd(8) filesystem access with unveil(2).
+
+
- Fixed a memory leak in window tree search in tmux(1).
+
+
- Added an equivalent test for master in drm_fb_helper_is_bound() for drm(4). This prevents black screens on hotplugging a new display with X(7) running, which required a vt switch.
- Added a check to ospfd(8) and ospf6d(8) that any "depend on" interfaces are in the same rdomain.
- Changed default Ruby version in ports(7) to 2.6.