SYSCONF's IPL.CB should be signed, but leads to other issues about libogc HOT 15 CLOSED

Sorry for the radio silence, I am waiting for my friend to come back from a work trip. I can't test this myself since I wasn't able to replicate the issue on any of my Wiis. Hopefully he'll come back soon and test it. I will make sure to let you know as soon as I have any feedback from him

from libogc.

@Aurelio92 : new libogc released. it'll need the new devkitPPC too btw

from libogc.

Hey Aurelio,

thanks for the report!
this is a ... interesting issue. from what i remember from the PAD interface, it works pure on the registers and the VI clock, so no idea how changing that would cause it to fail.
do you get video out with the change? (to check if VI is up and running and it's clock is ok)

did you create a test app by any chance to display time and test pad and stuff?

did some looking in pad, and it uses the SI interface and SI_Transfer, which in turn does some timing stuff (gettime, udelay etc), which might be broken like you mentioned.
i assume the other stuff are broken for the same reason

from libogc.

Thanks for your quick answer! Video out works just fine. I don't really have a test app, but I can write that later tonight. The only thing I noticed that didn't work so far was PAD, but more stuff could be broken and I might have not noticed

from libogc.

np! i try to keep communication here positive haha

if the SI transfer is broken because of the time changes, im sure other time based communications are broken as well tbh.
i'd need to look at how the timebase works in libogc, but kinda busy this evening

from libogc.

@Aurelio92 : just poking you to see if you have an app to test with. there is a possible fix i want to test and make sure before i merge it into libogc

from libogc.

Sorry for the delay. Here's the code: https://gist.github.com/Aurelio92/d13a74788f09c95e8658a3c5f8945362
I had a friend run some tests since I can't easily replicate the issue on any of my Wiis, though I should probably try setting a fake date to force that haha. Anyways he said that basically half the time he run this code his Wii doesn't detect any input when pressing A on his GCC.

from libogc.

ok so, i think we have 2 bugs here.
1 bug is internal timing stuff should not use gettime, but rather use system time.
this system time should have a signed bias, and be kept seperate from the rest

SO, i suggest your code to be build with my potential fix and see if the gamecube controller works then.
if it does, that means separating the times works, and we can look at the bias bug.

for me this works btw, but since all changes shows me the correct time (to the second) i can't validate its correctness hehe.
could you send this to your friend? (code changes are included)
rtcTest.zip

from libogc.

@Aurelio92 : sorry to poke you again. youre probably a busy person, just like i am haha.
but have you gotten any results from the test using extrems' fix?

from libogc.

@DacoTaco I just got the results from my friend. Your fix does work indeed! Thank you for this

from libogc.

its not my fix, but extrem's fix hehe

ok, ill merge the time bias into libogc then !

then there is the second bug that needs to be addressed : is the change of CONF_GetCounterBias(u32 *bias) still needed ? or was it fixed all together?

from libogc.

I can confirm both times coming from the new libogc and from my_SYS_Time are the same, so I guess making the bias signed is not needed anymore now. Thank you and Extrems again!

from libogc.

thank you for reporting this!
i hope we can release a new libogc update soon ! ^^;

from libogc.

Awesome, thank you!

from libogc.

Just saw that few minutes ago, that's great to hear! Thank you for reaching out!

from libogc.