It was renamed to play_lock because it synchronized not just updates
but also shutdown. Since the previous commit, it again only
synchronizes updates. Rename it back.
Also move its initialization next to shutdown_lock's.
shutdwn() sets the EOF indicator, aborts the running command, if any,
forbids sleeping on I/O and wakes up the player thread, for all player
threads in state PS_PLAYING. It takes play_lock to prevent new
commands from running. It then waits up to 3s for player threads to
terminate, by polling player_next(), to let output buffers drain.
Issues:
1. Polling is lame.
2. New player threads can still enter state PS_PLAYING. They'll block
as soon as they try to run a command. Somehwat unclean.
3. We can exit before all player threads left state PS_PLAYING, losing
a treasury update, play time update, and log entries. Could happen
when player threads blocked on output until commit 90b3abc5 fixed
that; its commit message describes the bug's impact in more detail.
Since then, the bug shouldn't bite in practice, because player
threads should leave state PS_PLAYING quickly.
Fix by introducing shutdown_lock: player threads in state PS_PLAYING
hold it shared, shutdwn() takes it exclusive, instead of play_lock.
Takes care of the issues as follows:
3. shutdwn() waits until all player threads left state PS_PLAYING, no
matter how long it takes them.
2. New player threads block before entering state PS_PLAYING.
1. shutdwn() still polls up to 3s for player threads to terminate.
Still lame. Left for another day.
Why upgrade? I'm not a lawyer, but here's my take on the differences
to version 2:
* Software patents: better protection against abuse of patents to
prevent users from exercising the rights under the GPL. I doubt
we'll get hit with a patent suit, but it's a good move just on
general principles.
* License compatibility: compatible with more free licenses, i.e. can
"steal" more free software for use in Empire. I don't expect to steal
much, but it's nice to have the option.
* Definition of "source code": modernization of some details for today's
networked world, to make it easier to distribute the software. Not
really relevant to us now, as we normally distribute full source code.
* Tivoization: this is about putting GPL-licensed software in hardware,
then make the hardware refuse to run modified software. "Neat" trick
to effectively deny its users their rights under the GPL. Abuse was
"pioneered" by TiVo (popular digital video recorders). GPLv3 forbids
it. Unlikely to become a problem for us.
* Internationalization: more careful wording, to harden the license
outside the US. The lawyers tell us it better be done that way.
* License violations: friendlier way to deal with license violations.
This has come out of past experience enforcing the GPL.
* Additional permissions: Probably not relevant to us.
Also include myself in the list of principal authors.
A player thread may sleep on input or output, except:
(1) While it is executing a C_MOD command, it may only sleep on input.
(2) While it is being aborted by the update or shutdown, it may not
sleep at all.
To find out whether a player thread may sleep on input, code has to
check condition (2). It needs do to that in recvclient().
To find out whether it may sleep on output, it has to check both
conditions. It needs to do that in pr_player() and upr_player().
The code tracked condition (1) in global variable play_lock_wanted.
It checked condition (2) by examining struct player member command.
Replace all that by new struct player member may_sleep. Initialize it
in player_new(), update it in dispatch(), shutdwn() and update_run().
This makes the tests in recvclient(), pr_player() and upr_player()
obvious. play_wrlock_wanted() is now unused, remove it.
Commit fbf9f15b removed SO, SE, but left their declarations behind.
Ditto commit 3aea20e1 for bigcity_dchr, and commit 08b94556 for
player_kill_idle.
Commit afa65c8f accidentally added a declaration for xedit().
(update_run): Set it.
Replace incorrect uses of update_pending by update_running, fixing
buggy behavior while the update was trying to gain control:
(sct_do_upd_mob, shp_do_upd_mob, lnd_do_upd_mob, pln_do_upd_mob):
MOB_ACCESS mobility update was skipped.
(telegram_is_new): Fix recognition of new telegrams by timestamp.
(wu): Bulletins got misfiled as production reports.
(shp_nav): Sail path got ignored.
(setrel): Declaration of war failed silently with SLOW_WAR enabled.
Messages got suppressed.
(PR, mpr): Messages got misdirected to bulletins.
(lndstr, plnstr, sctstr, shpstr): Change timestamp members lnd_access,
pln_access, sct_access, shp_access from real time (time_t) to ETUs
within a turn (short).
(land_ca, plane_ca, sect_ca, ship_ca): Update accordingly.
(build_ship, build_land, build_bridge, build_plane, build_tower)
(explore, check_trade, bsanct, takeover, takeover_ship)
(takeover_land): Use game_tick_to_now() instead of time() to update
the timestamp. Change check_trade(), takeover_ship(), takeover_land()
to do that only when MOB_ACCESS is enabled, for consistency.
(lupgr, supgr, pupgr, takeover_ship): Don't touch the timestamp where
mobility isn't touched either.
(sct_do_upd_mob, shp_do_upd_mob, lnd_do_upd_mob, pln_do_upd_mob): Use
game_tick_to_now() instead of increase_mob() to compute ETUs since
the timestamp and update the timestamp. Closes#1012699.
(increase_mob): Remove.
(mob_sect, mob_ship, mob_land, mob_plane): sct_do_upd_mob() & friends
no longer do the right thing at the update. Use game_reset_tick() and
pass its result directly to do_mob_sect() & friends. This is only
correct when argument is etu_per_update, which it always is. Remove
parameter. Callers changed.
(do_mob_sect, do_mob_ship, do_mob_land, do_mob_plane): Oops on
negative argument.
(mob_acc_globals, timestampfil, mobupdate, updating_mob)
(update_all_mob, timestamp_fixing, update_timestamps, mobility_check):
The mobupdate command was important to let deities manually
synchronize mobility updating with updates. That's no longer needed.
The code behind it is somewhat hairy and ugly, and updating it to work
with the Empire clock is just not worth it. Remove. Users changed.
(player_coms): Update accordingly.
(upda): Remove display of mobility updating state.
(mobility_init): No need to fix up mobility on startup, as the Empire
clock runs normally even when the server is down. Remove. Caller
changed.
(schedulefil): New.
(set_dirs, set_paths): Rename. Initialize schedulfil.
(read_schedule): New. Can read several updates, which will be used in
later changesets.
(update_time): Change to array. Will be used in later changesets.
(update_schedule_anchor): New.
(update_init): Initialize it.
(update_get_schedule): New.
(update_init): Call it to initialize update_time[].
(update_sched): Rewrite.
(update_forced, update_wanted): Replace.
(update_reschedule): New.
(main): Call it on SIGHUP to reload the schedule.
(update_trigger, update_force, force, player_coms): Drop force's
capability to schedule updates in the future, because it's not worth
the trouble to implement again. Deities can simply edit the schedule
file to schedule updates. Remove update_force() and
update_trigger()'s parameter.
(upda): Update for new scheduler. Take care to keep output the same
as far as possible, even though it's ugly, to avoid breaking clients.
(update_policy, adj_update, update_times, hourslop, blitz_time):
econfig keys removed.
(update_demand, UPD_DEMAND_NONE, UPD_DEMAND_SCHED, UPD_DEMAND_ASYNC)
(update_demandpolicy, UDP_NORMAL, UDP_TIMES, UDP_NORMAL, UDP_BLITZ)
(UDP_MAX, UDP_DEFAULT, UDDEM_TMCHECK, UDDEM_COMSET, UDDEM_DISABLE)
(UDDEM_MAX, UDDEM_DEFAULT): econfig key and values replaced. Users
changed. wantupd.h is now empty, remove.
(demand_check): External linkage.
(update_policy_check): Now pointless, remove.
(is_daytime_near, min_to_next_daytime, regular_update_time)
(scheduled_update_time, next_scheduled_time, updatetime)
(next_update_time, next_update_check_time): Unused, Remove.
(demand_check, demandupdatecheck): Move call of demand_update_time()
from demand_check(), which controls all demand updates, to
demandupdatecheck(), which controls only unscheduled ones. Fixes
update command not to lie about the next scheduled demand update.
(demandupdatecheck): Check updates_disabled() so that zdone no longer
claims to trigger an update when it can't.
(shutdown_initiate): New.
(shut): Use it. Shutdown in zero minutes no longer cancels the
shutdown, it just works. Use negative argument to cancel. Logging is
less detailed.
(shutdown_sequence): Internal linkage.
(pr_wall): All callers prefix text it with the same header. Move it
into the function.
(update_trigger): New.
(force, zdon): Use it.
(update_sem, update_force): Internal linkage.
(update_trigger, update_force): Passing a pointer to static storage as
thread argument is racy. Use dynamic allocation.
(update_main): No longer a thread entry point. Remove the parameter.
Remove *player setup. Don't terminate the thread when done.
(update_pending): Move to src/server/update.c.
(update_wait): Call update_main() instead of running it in its own
thread. Set up *player for it.
(update_init): Create the fused update thread instead of the update
wait thread.
resolving #1458175 and #1504036):
(update_lock): New.
(update_shed): Initialize it.
(update_wait, dispatch): Take exclusive update_lock around the update,
shared update_lock around commands. This makes the update block until
all aborted commands terminated and gave up their lock. Remove the
cheesy and unsafe two second wait for commands to finish. It also
makes player threads block before executing commands while the update
is pending, removing the need to fail commands then. So don't.
other. Ensure headers in include/ can be included in any order
(except for econfig-spec.h, which is special). New header types.h to
help avoid inclusion cycles. Sort include directives. Remove some
superflous includes.
between headers. Code is now fully prototyped and compiles cleanly
with gcc -Wstrict-prototypes -Wmissing-prototypes -Wnested-externs
-Wredundant-decls. Closes#723788.
