BSD Commit Log Search

   Do not walk off end or beginning of menu item list if every item is
   disabled, GitHub issue 5255 from aizumusheer2 at gmail dot com.

   include a comment in the tm_wday = -1 codesample, since we are
   trying to encourage this to become an idiomatic use pattern.
   Everyone doing this, please copy the comment also.
   ok tb

   Avoid netlock dance in cad(4) ioctl.

   While interface ioctl(8) is moving towards better parallel access,
   drivers assuming exclusive netlock and unlocking explicitly cause
   trouble.  Use the drivers own configuration lock for protection.
   Do not wait for task barriers while potentially holding the netlock.

   Panic reported by Mark McBride and Thomas Pries.
   Tested by dlg@; OK dlg@ mvs@

   sys/videoio.h: sync with Linux

   1. media: v4l2-controls: Add control for background detection
   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5d801b59633f6af60bb0e18d3bbb18b7b040a6d9

   OK: mglocker@

   timegm(3) is in C23

   from Christian Schulte

   Replace all printf with DPRINTF so that kernel messages are only printed
   if the FUSE_DEBUG option is set at compile time. Each debug message on
   the console is prefexed with "fusefs: "

   Remove the useless fuse_dump_buff() function while we are at it.

   ok mlarkin

   setup the radeondrm wscons burner task, lost in 6.18 drm changes

   avoids a NULL deref when wscons screen burner is enabled
   reported by al and Philippe on bugs@

   incorrect path checking in ypbindproc_domain_2x
   found by Christian Schwabl
   ok millert

   sys/octeon: add em to GENERIC and RAMDISK kernels

   OK: visa@

   Error if key is unknown rather than just carry on, GitHub issue 5248.

   check that make rejects empty variable names, from espie

   Fix {} keys in tmux.1, pointed out by tb.

   refuse empty variable names, from espie
   has been through multiple mkr and at least one ports bulk

   dhcpd: use error check per timegm(3) manpage

   ok florian

   httpd: error check timegm(3) per the manual

   ok job rsadowski

   Calculate the msg_copyin() `xfer' as min(available space, ...) not
   min(total space, ...). Otherwise we read outside of userland buffer.

   Also, I like to use sizeof(variable) instead of sizeof(type).

   ok cludwig

   popup_write can be deleted now also.

   Use a floating pane for the buffer mode editor instead of a popup.

   Do not loop forever if a pane ends up with borders the wrong way round.

   Draw help directly instead of using a popup.

   Add a loop_index variable and use it to show last used 5 sessions on the
   session menu instead of next/previous.

   Correctly report theme changes to panes, based on a fix from someone in
   GitHub issue 5246.

   Make default M-C-Up/Down bound to prompt movement commands in copy mode.

   sys/cnmac: read next RX pointer before release

   OK: visa@

   Allocate a buffer for the line cache instead of reaching into db data.

   Previously, vi used the db internal data buffer directly.  Fixes a
   problem where the line cache could become corrupted when the db
   data changes out from under it.  From Walter Alejandro Iglesias.

   bin/ksh: restore source after interactive error recovery

   unwind() may clear source while popping execution environments. The
   interactive shell then expands PS1 before compile() restores source, so
   a prompt command substitution using read -s can dereference NULL while
   saving history. Restore source to the current interactive input source
   before continuing to the prompt loop.

   Reported by Luigi Vianello

   OK: millert@

   knfmt

   Use a helper and skip some setup for empty cells.

   Tidy up a bit in tty_draw_line.

   Do not dereference the current window if it is NULL (can be if the
   notification fires after session is destroyed). GitHub issue 5244 from
   Ben Maurer.