BSD Commit Log Search

   sys/uvm: fix amap lock order during swapoff

   amap_wipeout() removed the amap from the global list while holding
   amap->am_lock. swapoff walks that list as amaplstlk -> amaplk, while the
   last unref path could establish the reverse order through
   amap_wipeout(), producing a witness lock order reversal when disabling
   swap.

   OK: renaud@

   Consistently use convenience variables rather then referencing
   members of the input structs in fuse_lookup(). No functional change.

   OK claudio@

   Unescape arguments before passing to format_expand1 so that escaping :s
   etc actually works.

   Check time periodically in loops rather than every one.

   Use evbuffer for building S, W, L lists, also add an xmemdup helper
   instead of misusing asprintf.

   Move log_imsg(), proc_title(), proc_name(), and imsg_to_str() into util.c.
   This is prepwork to give all processes their own main().

   OK deraadt@

   Copy over iked's proc.c, with the addition that proc_run()'s run argument
   is called after IMSG_CTL_PROCREADY is handled. This addition makes sure
   that run() can drop recvfd pledge if the process itself doesn't need it,
   and data can be send to peers immediately.

   Help from, and ok jmatthew@

   Add a dim= style attribute to dim the colours (as best as tmux is able
   to). GitHub issue 4842.

   Add functions to work out cell neighbours, and to remove a tiled cell.
   From Dane Jensen.

   sys/vio: align receive buffers

   vio removes the virtio net header before handing packets to the Ethernet
   input path. With the 12 byte modern virtio net header this leaves the
   Ethernet frame at an address where the following IP header is not 32 bit
   aligned; armv7 traps on the resulting unaligned access while processing
   DHCP traffic.

   OK: dlg@

   Check the panes again if deferred redraw to make sure the flag is always
   set. Also add a helper for the loop.

   Drop the incoming packet containing 16 MPLS labels with no BoS bit.

   ok claudio

   openssl-ruby-tests: prefer 40 over 34

   openssl-ruby-tests: work with ruby34 and ruby40

   add one of the microsoft branded Quectel EM12-G modems.

   i have a few i tested with, and it works fine.

   regen

   add one of the microsoft branded Quectel EM12-G modems

   i've tested that they work with umsm(4) and umb(4)

   Update ruby-module.5 for default Ruby version switch to 4.0

   To work with bounce buffers, xen_bus_dmamap_sync has to call
   bus_dmamap_sync using the parent dma tag.

   ok kettenis@ dlg@

   pppoe(4): leverage if_hardmtu for RFC 4638 instead of parent if_mtu

   Interfaces operate at their maximum supported packet size at all times,
   independent of the software IP stack MTU.

   Enforce MTU ceilings during PPPOESETPARMS and SIOCSIFMTU based on
   if_hardmtu instead of the parent's software if_mtu. This allows pppoe(4)
   devices to use "mtu 1500" without requiring manual MTU manipulation on
   the physical parent interface. Update pppoe.4 to match.

   "if it works for you then it's ok by me" dlg

   Implement bgscan, add hooks for set and delete key and improve mwx_stop
   so that there is a better chance to recover after down/up.
   On top of this implement mwx_activate() so that suspend/resume works.
   None of this is perfect but hopefully good enough.

   Tested on MT7921

   Everything needed to bring the FUSE kernel protocol in line with
   the Linux implementation is now in place. With this update, the
   kernel can support ports that talk directly to /dev/fuse0 rather
   than relying on libfuse.

   sys/fusebuf.h is retained rather than introducing fuse_kernel.h

   OK claudio@

   Tidy up server_client_check_redraw. Get rid of the bitmask tracking
   panes which want redraw for deferred clients - if they are deferred then
   they can just have a full redraw instead. Also return earlier if no
   redraw is actually needed, and improve the comments.

   Fix shifts to 1ULL and check scrollbar even if pane is not being redrawn.

   rpki-client: indent some out labels for consistency

   rpki-client: unify CMS signed object parsers

   This commit starts unifying various CMS signed object parsers in
   preparation for a lot more refactoring of this long-accumulated
   copy-paste mess.

   The signed object is passed as a void * object to various handlers, so
   these handlers all have the same signature.

   In this step, *_parse() is essentially split into allocation plus:

   1. *_cert_info(), which checks some basic things on the EE cert (usually
      inheritance, presence or absence of RFC 3779 extensions). For MFTs
      it also extracts some info and hangs that off mft.

   2. the already existing *_parse_econtent()

   3. *_validate() that does some validation steps and sets ->valid.
      In most signed object handlers the validate step can't currently

    [10 lines not shown]

   Add -M to move-pane and default bindings for M-drag. From Michael Grant.

   FIx a couple of minor errors in use of layout_cell_is_tiled.

   Make the behavior of the '(' command match the ')' command.

   Previously, the ')' command required a sentence to end with two
   blanks, but '(' only required a single space.  Historic vi (and the
   documentation) requires two spaces in both directions.

   Based on a diff from Walter Alejandro Iglesias.

   Factor out hex2bin into a shared helper function.  Replace sscanf %hhx
   (which is C99) with plain %x for better compatibility in -portable.
   ok djm@