search

kbeckmann / game-and-watch-retro-go

Emulator collection for Nintendo® Game & Watch™
GNU General Public License v2.0
440 stars 133 forks source link

GG (Sega Game Gear) doesn't work #97

Closed olderzeus closed 3 years ago

olderzeus commented 3 years ago

All gg rom run an error occurred

FATAL EXCEPTION: Hardfault 5244218-dirty PC=0x* LR=0x

kbeckmann commented 3 years ago

What are your local changes? The -dirty tells us that there are changes.

Make sure that you've run git pull --recurse-submodules to update the submodules.

When you see the problem again, take a photo of the screen and post it please as it contains helpful information.

olderzeus commented 3 years ago

I pull a new version agin， same error. rom download from romhustler.org.

see the Img: https://imgtu.com/i/RWifhR

kbeckmann commented 3 years ago

I just tested a few GG roms together with NES and GB (saw that you have that setup) and it works for me. Can you post a callstack from your system?

First make sure to run make flash, then trigger the bug so you see the bluescreen. Now, in one console, run make openocd. In another, run make gdb and then type bt in the console and post the result.

Should look like this: https://asciinema.org/a/rgIy0jcXkbaHUhzX4K8L22uf4

olderzeus commented 3 years ago

(gdb) bt

0 0x08000730 in buttons_get () at Core/Src/gw_buttons.c:11

1 0x08001324 in BSOD (fault=65, fault@entry=BSOD_HARDFAULT, pc=0x100da2f, lr=0x800d796 <memset+10>) at Core/Src/main.c:178

2 0x080041a2 in HardFault_Handler () at Core/Src/stm32h7xx_it.c:96

3 \

4 0x0800d796 in memset ()

5 0x08005422 in emulator_start (file=, load_state=, start_paused=) at Core/Src/retro-go/rg_emulators.c:404

6 0x08005638 in emulator_show_file_menu (file=0x80119a8 ) at Core/Src/retro-go/rg_emulators.c:354

7 0x080056a2 in event_handler (event=, tab=0x2000ce7c) at Core/Src/retro-go/rg_emulators.c:75

8 0x080050ba in retro_loop () at Core/Src/retro-go/rg_main.c:337

9 0x08005178 in app_main () at Core/Src/retro-go/rg_main.c:385

10 0x08002044 in main () at Core/Src/main.c:534

kbeckmann commented 3 years ago

Ah thank you. I can reproduce it now by having a lot of roms.

olderzeus commented 3 years ago

Thank you for the all

kbeckmann commented 3 years ago

I am looking into this, but the fix will just be a better validation of the memory usage, i.e. it will complain and tell you that you are using too many roms. If you look at the build output you probably see something like this

ram_emu_nes 103408 / 671744 (568336 bytes free (0.542 MB))
ram_emu_gb  242348 / 671744 (429396 bytes free (0.410 MB))
ram_emu_sms 673172 / 671744 (-1428 bytes free (-0.001 MB))
ram_emu_pce 0 / 671744  (671744 bytes free (0.641 MB))

note the negative value. I will add proper validation now, but in the longer term we will try to use less ram for the rom information.

kbeckmann commented 3 years ago

So what you can do right now is to remove roms until you don't see any negative size values.

kbeckmann commented 3 years ago

Alright, I have now fully understood the issue and added asserts in 52187019472cf91efdeb47a30dbfd7610529f8ff. It is not related to number of roms.

On my system where I use GCC 11.1.0, I reproduced it by enabling debug information which increased the RAM utilization. However, it seems that the current code will not fit when built with GCC 10.2.0. I'll see if I can squeeze it down a bit.

kbeckmann commented 3 years ago

It's fixed now. Please run git pull --recurse-submodules and try again.