Luma3DS
Luma3DS copied to clipboard
LumaTeam
Screen init has been fixed
I have spend some time reverse engineering GSP code to finally fix this. Corgi3DS was also a huge help. It mostly boils down to waiting for MCU events 24-29 and waiting for the LCDs to indicate they are ready. Not doing so causes random screen init problems like we have been plagued by for years. The fix only works if all homebrew bootloaders respect it because turning screens off for example and booting a FIRM too early can make screen init fail in the booted FIRM.
Because i'm not sure if/how IRQ stuff should be integrated into Luma (which is needed to wait for MCU events) i decided to make an issue instead of a PR. Hopefully that doesn't come across as lazy or something.
Here is the code: https://github.com/profi200/open_agb_firm/blob/master/source/arm11/hardware/gfx.c#L118
Wow, I certainly didn't expect that, thank you very much!
It's not going into this release but into the next one, for which I have planned to have an IRQ handler for the Arm11 (to prevent MCU i2c polling)
No worries. It will do for now. When writing that GBA stuff i had screen init problems once more on my N3DS XL so i thought it's time to finally fix this.
The line you are searching for is now here: https://github.com/profi200/libn3ds/blob/master/source/arm11/drivers/gfx.c#L66
This new driver design has been stable for me and the ported version in GodMode9 is doing well too.
And what i have to do with this line? i have the same problem with the screen flicking with the splash screen on
And what i have to do with this line? i have the same problem with the screen flicking with the splash screen on
These lines of code making up a "block of code" is more so pointed here for the developers of the software, who would port over the code from libn3ds/GodMode9 to Luma3DS. It is not beneficial to an end-user.
The code base Luma uses right now is missing some things like interrupts that are needed for this so i doubt libn3ds will help here (it's in a pretty messy state so i would assume they don't want to use it anyway). If i recall correctly an attempt was made to port this years ago but it didn't work as expected. GodMode9 has some basic interrupt support at least on ARM11 so it works good enough.