map server crash @ v16386

Originally posted by [b]jTynne[/b]
GDB nub here, didn't get the full crash dump. ;


#0 0x08131d9f in skill_castend_damage_id (src=0xeb348f8, bl=0xa8b243e4,
skillid=2230, skilllv=1, tick=3036849411, flag=0) at skill.c:4108
skill_lv = <value optimized out>
s = 0
skill_id = <value optimized out>
point = <value optimized out>
spell = {2230, 246631336, -1073745336, 135327648, -1073745288,
-1073745312, 135325962}
i = <value optimized out>
sd = 0xeb348f8
tstatus = <value optimized out>
sc = 0xeb34ba8
#1 0x0814796c in skill_castend_id (tid=-1, tick=3036849411, id=2009128,
data=0) at skill.c:8897
target = 0xa8b243e4
src = 0xeb348f8
sd = 0xeb348f8
md = 0x0
ud = 0xeb34910
sc = 0xa8b2462c
inf = <value optimized out>
inf2 = <value optimized out>
flag = 0

Originally posted by [b]jTynne[/b]
[14:04] <Llediarth> 16365 is my guess atm for breaking it
[14:07] <Llediarth> lol wait could it be
[14:07] <Llediarth> i = spell[s==1?0:rand()%s]
[14:07] <Llediarth> if s is 0 youd be dividing by 0?

Originally posted by [b]jTynne[/b]
Possible fix by Llediarth:

[font=arial, sans-serif][size=3]if(s==0) s = 1; // Prevent division by zero[/size][/font]
[font=arial, sans-serif][size=3]to Skill.c at line 4108[/size][/font]

[font=arial, sans-serif][size=3]Will test after next crash.[/size][/font]

Originally posted by [b]jTynne[/b]
[14:32] <Llediarth> i = spell[s<=1?0:rand()%s];
[14:32] <Llediarth> is better then

Originally posted by [b]jTynne[/b]
misc brainstorming


[14:57] <~jTynne> skill level used was 1, had five things stored
[14:57] <~jTynne> not sure what they are though
[14:57] <~jTynne> soooo
[14:57] <~jTynne> I'm guessing...
[14:57] <~jTynne> do the spells store for a specific amount of time?
[14:57] <~jTynne> and instead of ending at a countdown to 0
[14:57] <~jTynne> it continues past then
[14:57] <~jTynne> -tests-
[14:58] <Llediarth> probably the way they have them implemented yeah
[14:58] <Llediarth> as status changes

Originally posted by [b]jTynne[/b]
qq this needs to be severity high PLOX /kicksnadscreams

Thanks @ Trojal and Ind in IRC working on this. T_T

Also thanks to Llediarth

Originally posted by [b]jTynne[/b]
Issue has been fixed-- Thank you Trojal, Ind, and Llediarth!

Originally posted by [b]Mysterious[/b]
Eek.. Still? o_o

Originally posted by [b]jTynne[/b]
Still crashing. Reverted back to 16316, no probs there. =(

Originally posted by [b]malufett[/b]
how can this fix the problem??..
[quote]if(s==0) s = 1; // Prevent division by zero[/quote]
and [rev='16388']

since [b]i[/b] and [b]s[/b] will never be negative...

and you can't release a null spell since I already make check for it..
[CODE]if( s && sc->data[i] ){[/CODE]

and the crash will be cause by [CODE]if( sc->data[i] ) spell[s++] = i;[/CODE] if the [b]sc [/b]will be broke...
please try [CODE]if( sc && sc->data[i] ) spell[s++] = i;[/CODE] if it will work....

Originally posted by [b]malufett[/b]
BTW something is bugging me...why does 2230(WL_RELEASE) has been stored?
please check if your players are hacking the way it stores spells..

[quote]
[color=#282828][font=helvetica, arial, sans-serif][size=3]spell = {[b]2230[/b], 246631336, -1073745336, 135327648, -1073745288,[/size][/font][/color]
[color=#282828][font=helvetica, arial, sans-serif][size=3] -1073745312, 135325962}[/size][/font][/color]
[/quote]

and as I understand the crash dump that there is a redundant invocation of the skill...
notice the value of [b]s [/b]it is 0 which means '[color=#282828][font=helvetica, arial, sans-serif][size=3]skill.c:4108' is the bottle neck of the looping of the skill invocation..[/size][/font][/color]

Originally posted by [b]jTynne[/b]
Which would be done how? I've made no modifications to the code in question. =X We run Harmony, so..

Originally posted by [b]malufett[/b]
maybe there is some new tools for hacking which Harmony doesn't aware of...

anyway lets wait for more information regarding this crashing from other people that would confirmed..
and try to investigate what your players are doing...good luck!

Originally posted by [b]Ind[/b]
r16388 is wrong it is meant to be s == 0 not i. it crashes on
i = spell[s==1?0:rand()%s];// Random select of spell to be released.
rand()%0 = crash.

Originally posted by [b]Ind[/b]
should be fixed in 16396
[code]
[10:46:27] <CIA-7> RAthena: shennetsind * r16396 /trunk/src/map/skill.c: Follow up r16388, fix for bugreport:6200
[/code]