channel1 down checkpoint shutdown gamerev 34083

[servan12 0]

channel1 down checkpoint shutdown gamerev 34083 syserr ;

CHECKPOINT shutdown: tics did not updated.

Analyze: Handshake phase does not handle packet 6 (fd 30)

Thank you in advance I would be glad if the diff

[servan12 0]

up to date?

[servan12 0]

Does anyone have information?

[servan12 0]

Does anyone have information?

[servan12 0]

Update

[servan12 0]

update

[servan12 0]

update?

[Zonni 224]

Try to run sequence patcher and apply to your client & server. Honestly, i was look for this errror message in 40k sf source and i cant find it.

[TheSLZ 104]

Because it's in the libthecore folder

 

/usr/src/mainline/Srcs/Server/libthecore/src/signal.c

 

RETSIGTYPE checkpointing(int sig)
{
    if (!tics)
    {
	sys_err("CHECKPOINT shutdown: tics did not updated.");
	abort();
    }
    else
	tics = 0;
}

I think it's not the gamefile's problem.

 

[servan12 0]

Try to run sequence patcher and apply to your client & server. Honestly, i was look for this errror message in 40k sf source and i cant find it.

 

sequence was

 

 

Because it's in the libthecore folder

 

/usr/src/mainline/Srcs/Server/libthecore/src/signal.c

 

RETSIGTYPE checkpointing(int sig)
{
    if (!tics)
    {
	sys_err("CHECKPOINT shutdown: tics did not updated.");
	abort();
    }
    else
	tics = 0;
}

I think it's not the gamefile's problem.

 

 

where is problem?

[servan12 0]

update_?

[martysama0134 6966]

update_?

Ymir enabled an anti-starvation system:

    signal_timer_enable(30);

    signal(SIGVTALRM, checkpointing);

They set a SIGVTALRM signal that calls (via kernel) the checkpointing function every 30 seconds of effective execution time.

Considering this:

int idle()
{
...
	while (passed_pulses--) {
		heartbeat(thecore_heart, ++thecore_heart->pulse);

		// To reduce the possibility of abort() in checkpointing
		thecore_tick();
	}
	CHARACTER_MANAGER::instance().Update(thecore_heart->pulse);
	db_clientdesc->Update(t);
...
}

The bug will appear if such functions will spend more than 30 seconds to be completely executed. (still talking about execution time and not real one)

A quest containing a "while true begin end" will also cause such error and the game will be crashed after 30 seconds.

If you still have the game.core generated by that crash and the relative game binary, I'll try to analyze it. (it should contain where the presumed "starvation" is)

Otherwise, you'll be fucked up.

You can even try to remove the abort() call via dif, but you could still get a freezed game if it's a real starvation.