Files
Paradise/code/controllers/ProcessScheduler/core/process.dm
tigercat2000 47cd4cb127 StonedMC
This commit ports the StonedMC from /tg/station, intended to replace the
Process Scheduler from goon. Currently, they exist simultaneously, as
it's very difficult to port our 22 processes to the SMC all at once.
Instead, we can make them work together until everything is converted
over at a later point, and then take the old PS out back and put a
couple of rounds into it's deformed skull.

Primary benefits of this new process controller include: Other people
that can actually maintain it, unlike the PS, pre-world-init
initialization for subsystems, ease of ports from /tg/station13, and
potential performance improvement (to be seen).
2018-03-02 21:12:49 -08:00

407 lines
11 KiB
Plaintext

// Process
/datum/controller/process
/**
* State vars
*/
// Main controller ref
var/tmp/datum/controller/processScheduler/main
// 1 if process is not running or queued
var/tmp/idle = 1
// 1 if process is queued
var/tmp/queued = 0
// 1 if process is running
var/tmp/running = 0
// 1 if process is blocked up
var/tmp/hung = 0
// 1 if process was killed
var/tmp/killed = 0
// Status text var
var/tmp/status
// Previous status text var
var/tmp/previousStatus
// 1 if process is disabled
var/tmp/disabled = 0
/**
* Config vars
*/
// Process schedule interval
// This controls how often the process would run under ideal conditions.
// If the process scheduler sees that the process has finished, it will wait until
// this amount of time has elapsed from the start of the previous run to start the
// process running again.
var/tmp/schedule_interval = PROCESS_DEFAULT_SCHEDULE_INTERVAL // run every 50 ticks
// Process sleep interval
// This controls how often the process will yield (call sleep(0)) while it is running.
// Every concurrent process should sleep periodically while running in order to allow other
// processes to execute concurrently.
var/tmp/sleep_interval = PROCESS_DEFAULT_SLEEP_INTERVAL
// Defer usage; the tick usage at which this process will defer until the next tick
var/tmp/defer_usage = PROCESS_DEFAULT_DEFER_USAGE
// hang_warning_time - this is the time (in 1/10 seconds) after which the server will begin to show "maybe hung" in the context window
var/tmp/hang_warning_time = PROCESS_DEFAULT_HANG_WARNING_TIME
// hang_alert_time - After this much time(in 1/10 seconds), the server will send an admin debug message saying the process may be hung
var/tmp/hang_alert_time = PROCESS_DEFAULT_HANG_ALERT_TIME
// hang_restart_time - After this much time(in 1/10 seconds), the server will automatically kill and restart the process.
var/tmp/hang_restart_time = PROCESS_DEFAULT_HANG_RESTART_TIME
// Number of deciseconds to delay before starting the process
var/start_delay = 0
/**
* recordkeeping vars
*/
// Records the time (1/10s timeofgame) at which the process last began running
var/tmp/run_start = 0
// Records the number of times this process has been killed and restarted
var/tmp/times_killed
// Tick count
var/tmp/ticks = 0
var/tmp/last_task = ""
var/tmp/last_object
// How many times in the current run has the process deferred work till the next tick?
var/tmp/cpu_defer_count = 0
// Counts the number of times an exception has occurred; gets reset after 10
var/tmp/list/exceptions = list()
// The next tick_usage the process will sleep at
var/tmp/next_sleep_usage
// Last run duration, in seconds
var/tmp/last_run_time = 0
// Last 20 run durations
var/tmp/list/last_twenty_run_times = list()
// Highest run duration, in seconds
var/tmp/highest_run_time = 0
// Tick usage at start of current run (updates upon deferring)
var/tmp/tick_usage_start
// Accumulated tick usage from before each deferral
var/tmp/tick_usage_accumulated = 0
/datum/controller/process/New(var/datum/controller/processScheduler/scheduler)
..()
main = scheduler
previousStatus = "idle"
idle()
name = "process"
run_start = 0
ticks = 0
last_task = 0
last_object = null
/datum/controller/process/Destroy()
..()
return QDEL_HINT_HARDDEL_NOW
/datum/controller/process/proc/started()
// Initialize run_start so we can detect hung processes.
run_start = TimeOfGame
// Initialize defer count
cpu_defer_count = 0
// Prepare usage tracking (defer() updates these)
tick_usage_start = world.tick_usage
tick_usage_accumulated = 0
running()
main.processStarted(src)
onStart()
/datum/controller/process/proc/finished()
ticks++
recordRunTime()
idle()
main.processFinished(src)
onFinish()
/datum/controller/process/proc/recordRunTime()
// Convert from tick usage (100/tick) to seconds of CPU time used
var/total_usage = (tick_usage_accumulated + (world.tick_usage - tick_usage_start)) / 1000 * world.tick_lag
last_run_time = total_usage
if(total_usage > highest_run_time)
highest_run_time = total_usage
if(last_twenty_run_times.len == 20)
last_twenty_run_times.Cut(1, 2)
last_twenty_run_times += total_usage
/datum/controller/process/proc/doWork()
/datum/controller/process/proc/setup()
/datum/controller/process/proc/process()
started()
doWork()
finished()
/datum/controller/process/proc/running()
idle = 0
queued = 0
running = 1
hung = 0
setStatus(PROCESS_STATUS_RUNNING)
/datum/controller/process/proc/idle()
queued = 0
running = 0
idle = 1
hung = 0
setStatus(PROCESS_STATUS_IDLE)
/datum/controller/process/proc/queued()
idle = 0
running = 0
queued = 1
hung = 0
setStatus(PROCESS_STATUS_QUEUED)
/datum/controller/process/proc/hung()
hung = 1
setStatus(PROCESS_STATUS_HUNG)
/datum/controller/process/proc/handleHung()
var/datum/lastObj = last_object
var/lastObjType = "null"
if(istype(lastObj))
lastObjType = lastObj.type
var/msg = "[name] process hung at tick #[ticks]. Process was unresponsive for [(TimeOfGame - run_start) / 10] seconds and was restarted. Last task: [last_task]. Last Object Type: [lastObjType]"
log_debug(msg)
message_admins(msg)
log_runtime(EXCEPTION(msg), src)
main.restartProcess(src.name)
/datum/controller/process/proc/kill()
if(!killed)
var/msg = "[name] process was killed at tick #[ticks]."
log_debug(msg)
message_admins(msg)
log_runtime(EXCEPTION(msg), src)
//finished()
// Allow inheritors to clean up if needed
onKill()
qdel(src)
// Do not call this directly - use SHECK
/datum/controller/process/proc/defer()
if(killed)
// The kill proc is the only place where killed is set.
// The kill proc should have deleted this datum, and all sleeping procs that are
// owned by it.
CRASH("A killed process is still running somehow...")
if(hung)
// This will only really help if the doWork proc ends up in an infinite loop.
handleHung()
CRASH("Process [name] hung and was restarted.")
tick_usage_accumulated += (world.tick_usage - tick_usage_start)
if(world.tick_usage < defer_usage)
sleep(0)
else
sleep(world.tick_lag)
cpu_defer_count++
tick_usage_start = world.tick_usage
next_sleep_usage = min(world.tick_usage + sleep_interval, defer_usage)
/datum/controller/process/proc/update()
// Clear delta
if(previousStatus != status)
setStatus(status)
var/elapsedTime = getElapsedTime()
if(hung)
handleHung()
return
else if(elapsedTime > hang_restart_time)
hung()
else if(elapsedTime > hang_alert_time)
setStatus(PROCESS_STATUS_PROBABLY_HUNG)
else if(elapsedTime > hang_warning_time)
setStatus(PROCESS_STATUS_MAYBE_HUNG)
/datum/controller/process/proc/getElapsedTime()
return TimeOfGame - run_start
/datum/controller/process/proc/tickDetail()
return
/datum/controller/process/proc/getContext()
return "<tr><td>[name]</td><td>[getAverageRunTime()]</td><td>[last_run_time]</td><td>[highest_run_time]</td><td>[ticks]</td></tr>\n"
/datum/controller/process/proc/getContextData()
return list(
"name" = name,
"averageRunTime" = getAverageRunTime(),
"lastRunTime" = last_run_time,
"highestRunTime" = highest_run_time,
"ticks" = ticks,
"schedule" = schedule_interval,
"status" = getStatusText(),
"disabled" = disabled
)
/datum/controller/process/proc/getStatus()
return status
/datum/controller/process/proc/getStatusText(var/s = 0)
if(!s)
s = status
switch(s)
if(PROCESS_STATUS_IDLE)
return "idle"
if(PROCESS_STATUS_QUEUED)
return "queued"
if(PROCESS_STATUS_RUNNING)
return "running"
if(PROCESS_STATUS_MAYBE_HUNG)
return "maybe hung"
if(PROCESS_STATUS_PROBABLY_HUNG)
return "probably hung"
if(PROCESS_STATUS_HUNG)
return "HUNG"
else
return "UNKNOWN"
/datum/controller/process/proc/getPreviousStatus()
return previousStatus
/datum/controller/process/proc/getPreviousStatusText()
return getStatusText(previousStatus)
/datum/controller/process/proc/setStatus(var/newStatus)
previousStatus = status
status = newStatus
/datum/controller/process/proc/setLastTask(var/task, var/object)
last_task = task
last_object = object
/datum/controller/process/proc/_copyStateFrom(var/datum/controller/process/target)
main = target.main
name = target.name
schedule_interval = target.schedule_interval
sleep_interval = target.sleep_interval
run_start = 0
times_killed = target.times_killed
ticks = target.ticks
last_task = target.last_task
last_object = target.last_object
copyStateFrom(target)
/datum/controller/process/proc/copyStateFrom(var/datum/controller/process/target)
/datum/controller/process/proc/onKill()
/datum/controller/process/proc/onStart()
/datum/controller/process/proc/onFinish()
/datum/controller/process/proc/disable()
disabled = 1
/datum/controller/process/proc/enable()
disabled = 0
/datum/controller/process/proc/getAverageRunTime()
var/t = 0
var/c = 0
for(var/time in last_twenty_run_times)
t += time
c++
if(c > 0)
return t / c
return c
/datum/controller/process/proc/getLastRunTime()
return last_run_time
/datum/controller/process/proc/getHighestRunTime()
return highest_run_time
/datum/controller/process/proc/getTicks()
return ticks
/datum/controller/process/proc/statProcess()
var/averageRunTime = round(getAverageRunTime(), 0.001)
var/lastRunTime = round(last_run_time, 0.001)
var/highestRunTime = round(highest_run_time, 0.001)
var/deferTime = round(cpu_defer_count / 10 * world.tick_lag, 0.01)
if(!statclick)
statclick = new /obj/effect/statclick/debug(src)
stat("[name]", statclick.update("T#[getTicks()] | AR [averageRunTime] | LR [lastRunTime] | HR [highestRunTime] | D [deferTime]"))
/datum/controller/process/proc/catchException(var/exception/e, var/thrower)
if(istype(e)) // Real runtimes go to the real error handler
log_runtime(e, thrower, "Caught by process: [name]")
return
var/etext = "[e]"
var/eid = "[e]" // Exception ID, for tracking repeated exceptions
var/ptext = "" // "processing..." text, for what was being processed (if known)
if(istype(e))
etext += " in [e.file], line [e.line]"
eid = "[e.file]:[e.line]"
if(eid in exceptions)
if(exceptions[eid]++ >= 10)
return
else
exceptions[eid] = 1
if(istype(thrower, /datum))
var/datum/D = thrower
ptext = " processing [D.type]"
if(istype(thrower, /atom))
var/atom/A = thrower
ptext += " ([A]) ([A.x],[A.y],[A.z])"
log_to_dd("\[[time_stamp()]\] Process [name] caught exception[ptext]: [etext]")
if(exceptions[eid] >= 10)
log_to_dd("This exception will now be ignored for ten minutes.")
spawn(6000)
exceptions[eid] = 0
/datum/controller/process/proc/catchBadType(var/datum/caught)
if(isnull(caught) || !istype(caught) || !isnull(caught.gcDestroyed))
return // Only bother with types we can identify and that don't belong
catchException("Type [caught.type] does not belong in process' queue")
// This is called to make a controller THE global authority on a global variable - like `air_master`
/datum/controller/process/proc/assertGlobality()
return
// Called to let another instance of the controller have the fun
/datum/controller/process/proc/releaseGlobality()
return