The call to cprng_fast32() requires that per-cpu data has been initialized
by corng_fast_init(), which doesn't get called until after the first part
of auto-configuration is done, long after pfattach() calls cprng_fast32().
Fixed PR kern/52620
XXX This needs pull-up to the -8 branch.
- Use "inline.sh" instead of "foo.sh".
- Clarify that "inline.sh" scripts will terminate the boot if
the script causes trhe current shell process to terminate.
- Use "service" instead of "bar.
- Use "sub-shell" instead of "subshell".
Inspired by feedback from kre@
group fails. This happens with empty vlan interfaces
- make syscalls checks against -1
- fix errors to print %s: instead of <%s>
XXX: if_vlan is the only pseudo interface in net/ that returns such an
error..
polled, so that the logic for AT_POLL matches how e.g. ata_dmaerr() is
called; this was the original intent of the change in 1.428.2.25,
to make the error handling safe wrt. polled xfers
this is stopgap fix for ATA channel wedge after DMA error, as reported
by Martin Husemann in PR kern/52606, and PR kern/52605
problem happened due to ata_reset_channel() being called once in ata_dmaerr()
with flags == 0, which freezed channel and set flag to reset via thread,
then ata_reset_channel() was called via wdc_drive_reset() with AT_POLL, which
just executed the reset and cleared the flag, without clearing the extra
freeze; that logic will be refactored in separate commit
if_baudrate atomatically by mii_phy_statusmsg(). ixg(4) and ixv(4) don't
use mii(4), so calling ifmedia_baudrate() directly is required.
XXX need the same same work for wm(4)'s TBI and SERDEV devices.
* Add select-pane -T to set pane title.
* Fix memory leak when lines with BCE are removed from history.
* Fix (again) the "prefer unattached" behaviour of attach-session.
* Reorder how keys are checked to allow keys to be specified that have a
leading escape. GitHub issue 1048.
* Support REP escape sequence (\033[b).
* Run alert hooks based on options rather than always, and allow further bells
even if there is an existing bell.
* Add -d flag to display-panes to override display-panes-time.
* Add selection_present format when in copy mode (allows key bindings that do
something different if there is a selection).
* Add pane_at_left, pane_at_right, pane_at_top and pane_at_bottom formats.
* Make bell, activity and silence alerting more consistent by: removing the
bell-on-alert option; adding activity-action and silence-action options with
the same possible values as the existing bell-action; adding a "both" value
for the visual-bell, visual-activity and visual-silence options to trigger
both a bell and a message.
* Add a pane_pipe format to show if pipe-pane is active.
* Block signals between forking and resetting signal handlers so that the
libevent signal handler doesn't get called in the child and incorrectly write
into the signal pipe that it still shares with the parent. GitHub issue 1001.
* Allow punctuation in pane_current_command.
* Add -c for respawn-pane and respawn-window.
* Wait for any remaining data to flush when a pane is closed while pipe-pane is
in use.
* Fix working out current client with no target. GitHub issue 995.
* Try to fallback to C.UTF-8 as well as en_US.UTF-8 when looking for a UTF-8
locale.
* Add user-keys option for user-defined key escape sequences (mapped to User0
to User999 keys).
* Add pane-set-clipboard hook.
* FAQ file has moved out of repository to online.
* Fix problem with high CPU usage when a client dies unexpectedly. GitHub issue
941.
* Do a dance on OS X 10.10 and above to return tmux to the user namespace,
allowing access to the clipboard.
* Do not allow escape sequences which expect a specific terminator (APC, DSC,
OSC) to wait for forever - use a small timeout. This reduces the chance of
the pane locking up completely when sent garbage (cat /dev/random or
similar).
* Support SIGUSR2 to toggle logging on a running server, also generate the
"out" log file with -vv not -vvvv.
* Make set-clipboard a three state option: on (tmux both sends to outside
terminal and accepts from applications inside); external (tmux sends outside
but does not accept inside); and off.
* Fix OSC 4 palette setting for bright foreground colours. GitHub issue 954.
* Use setrgbf and setrgbb terminfo(5) capabilities to set RGB colours, if they
are available. (Tc is still supported as well.)
* Fix redrawing panes when they are resized several times but end up with the
size unchanged (for example, splitw/resizep -Z/breakp).
* Major rewrite of choose mode. Now includes preview, sorting, searching and
tagging; commands that can be executed directly from the mode (for example,
to delete one or more buffers); and filtering in tree mode.
* choose-window and choose-session are now aliases of choose-tree (in the
command-alias option).
* Support OSC 10 and OSC 11 to set foreground and background colours.
* Check the U8 capability to determine whether to use UTF-8 line drawing
characters for ACS.
* Some missing notifications for layout changes.
* Control mode clients now do not affect session sizes until they issue
refresh-client -C. new-session -x and -y works with control clients even if
the session is not detached.
* All new sessions that are unattached (whether with -d or started with no
terminal) are now created with size 80 x 24. Whether the status line is on or
off does not affect the size of new sessions until they are attached.
* Expand formats in option names and add -F flag to expand them in option values.
* Remember the search string for a pane even if copy mode is exited and entered
again.
* Some further BCE fixes (scroll up, reverse index).
* Improvements to how terminals are cleared (entirely or partially).
CHANGES FROM 2.4 TO 2.5, 09 May 2017
* Reset updated flag when restarting #() command so that new output is properly
recognised. GitHub issue 922.
* Fix ECH with a background colour.
* Do not rely on the terminal not moving the cursor after DL or EL.
* Fix send-keys and send-prefix in copy-mode (so C-b C-b works). GitHub issue
905.
* Set the current pane for rotate-window so it works in command sequences.
* Add pane_mode format.
* Differentiate M-Up from Escape+Up when possible (that is, in terminals with
xterm(1) style function keys). GitHub issue 907.
* Add session_stack and window_stack_index formats.
* Some new control mode notifications and corresponding hooks:
pane-mode-changed, window-pane-changed, client-session-changed,
session-window-changed.
* Format pane_search_string for last search term while in copy mode (useful
with command-prompt -I).
* Fix a problem with high CPU usage and multiple clients with #(). GitHub issue
889.
* Fix UTF-8 combining characters in column 0.
* Fix reference counting so that panes are properly destroyed and their
processes killed.
* Clamp SU (CSI S) parameter to work around a bug in Konsole.
* Tweak line wrapping in full width panes to play more nicely with terminal
copy and paste.
* Fix when we emit SGR 0 in capture-pane -e.
* Do not change TERM until after config file parsing has finished, so that
commands run inside the config file can use it to make decisions (typically
about default-terminal).
* Make the initial client wait until config file parsing has finished to avoid
racing with commands.
* Fix core when if-shell fails.
* Only use ED to clear screen if the pane is at the bottom.
* Fix multibyte UTF-8 output.
* Code improvements around target (-t) resolution.
* Change how the default target (for commands without -t) is managed across
command sequences: now it is set up at the start and commands are required
to update it if needed. Fixes binding command sequences to mouse keys.
* Make if-shell from the config file work correctly.
* Change to always check the root key table if no binding is found in the
current table (prefix table or copy-mode table or whatever). This means that
root key bindings will take effect even in copy mode, if not overridden by a
copy mode key binding.
* Fix so that the history file works again.
* Run config file without a client rather than using the first client, restores
previous behaviour.
* If a #() command doesn't exit, continue to read from it and use its last full
line of output.
* Handle slow terminals and fast output better: when the amount of data
outstanding gets too large, discard output until it is drained and we are
able to do a full redraw. Prevents tmux sitting on a huge buffer that the
terminal will take forever to consume.
* Do not redraw a client unless we realistically think it can accept the data -
defer redraws until the client has nothing else waiting to write.
PPPOE_SESSION_LOCK protects variables used in PPP packet
processing, on the other hand PPPOE_PARAM_LOCK protects
the other variables used to establish a PPPoE session id.
Those locks isn't acquired in the same time because the
PPP packet processing doesn't work without PPPoE session id.
By the reason, the locks can be integrated into PPPOE_LOCK.
Add locking notes later.
skrll