logs archiveBotHelp.net / Freenode / #3dsdev / 2015 / September / 13 / 1
endrift
2 and a half hours with no freeze
I think aliaspider was right
profi200
Don't say that until you are 100% sure.
endrift
I said "I think"
not "turns out"
I'm running the version that removes that call
gonna leave it running for like 5 hours
would be nice if I had a second 3DS charger
Nintendo is weird
profi200
Their reason not to include chargers for the XL model is stupid.
I know no battery driven handheld which does not come with a charger.
Except the XLs from Nintendo of course.
endrift
o3DS froze
don't remember what build it's runnin
3+ hours
with sound
the fix is really gross though
smea
what is it ?
endrift
I removed some cache flushing ;_;
and just sort of hoped that the cache would be flushed in time
smea
...and that fixes things ? :|
endrift
yes :(
or so it seems
it's been going for 3 hours which is longer than I've seen it go before
aliaspider
what the cache flushing was the source of the lockup ?
endrift
seems that way
http://pastie.org/private/nm5zzbhrykpeslqzxoazxw
note that doing everything EXCEPT removing the cache flushes did not fix the crashing
which is really strange
aliaspider
hmm
endrift
that function is only called when csnd is properly initialized
aliaspider
removing csndIsPlaying(0x8, &playing); alone isn't enough ?
endrift
correct
so I wonder if it's somewhere else
aliaspider
i do use that cache flushing and Im not guetting any lockups
endrift
but this is the only function that's called repeatedly in the sound-only path
aliaspider
it could be that the cache flush makes another function act badly
endrift
yeah but what?
also yuriks said that tweaking the timing a bit caused it to not crash
aliaspider
ah wait , you call the flush on random addresses
endrift
ohhh
does it need to be aligned/
aliaspider
i only flush the whole buffer
just flush the whole thing
it will only affect what you just wrote there anyway
endrift
oop nevermind
it did crash
it just took 3 and a half hours
aliaspider
:/
endrift
so I guess that wasn't it
so I didn't leave the audioless version going for 3 hours
smea
does the crash only happen with csnd endrift ?
endrift
I thought it did
but now I'm not sure
I know removing a gspWaitForPPF0 drastically improved stability
which is ~really weird~
smea
oh
is it though
do you have a timeout ?
because gsp event waiting causing random freezes has been an issue since forever
the dirty fix was to add a timeout
endrift
oh
does that have an arg for a timeout?
smea
i think there were reports that the timeout might no longer be necessary but...
probably not in ctrulib no
https://github.com/smealum/portal3DS/blob/master/source/gfx/gs.c#L573-L583
endrift
g r e a t
smea
one day our gpu code will actually be good
i swear
:P
endrift
where IS yuriks today anyway
so it crashed at track 166 in the Mother 3 sound player last time
it's still going at 56 right now
130
still going
watch it die at 166 again just to show it's not actually random
watch there be exactly 166 tracks and I didn't realize it hadn't frozen
so I just shut it off
159
the suspense is killing me
it's been over 3 hours though
167
I guess it had frozen
oh dang
there it goes
so...3 and a half hours, huh
profi200
endrift: Could you include some option to force mono sound on 3DS? That would at least make it sound less bad. It sounds very strange if the center of the audio is shifted to the left side like now because of the unsychron audio :|
The original GBA just had 1 speaker :p
aliaspider
but stereo headphones
endrift
I left a tweaked version running overnight
the graphics were corrupted when I woke up (looked like an sf2dlib heap scribble or something) but it hadn't frozen
\o/
aliaspider
yay !
what did you tweak this time ?
endrift
I turned off vsync and made a version that had vsync
er
timeout
that was slightly more than 1 frame
aliaspider
ah that is why you got the corruption ?
because from my experience, even if I use timed gsp syncs, I end up with an unuseable gsp anyway
« prev 1 2 next »