Mouse 'stuttering' in Amiberry - mostly due to acceleration bug?

[RodimusFVC]

Have been having an intermittent problem with Amiberry (MidWan's Amiga Emulator) mouse movement

Manjaro Linux - 6.14 kernel, but that doesn't matter in this instance as far as I can tell

Basically, within the emulated environment I can move the mouse only by slamming it across the screen - fine movement is impossible.

I have narrowed the problem to the sdl2-compat library:

v2.32.52 works great
v2.32.54 completely doesn't work

This has been going back and forth a while - where sometimes it works, sometimes it doesn't, so I assume some change keeps being made and reverted.

I've ensured all the other SDL packages are up to date with the latest versions.... so it's just this one compatibility layer that's having the issue.

[cgutman]

I can't reproduce this, but maybe I'm doing something wrong.

I built Amiberry from AUR on my Arch system (KDE desktop) and ran with sdl2-compat-2.32.54-2 installed as provided by Arch. I started Amiberry and clicked the Start button without changing any of the default options, then successfully moved the red mouse pointer when the gray screen appeared during boot. I tried both with SDL_VIDEODRIVER=wayland and SDL_VIDEODRIVER=x11 and they could both move the cursor as expected. I tried both fast and slow movements of the mouse and both seemed to work as expected.

[RodimusFVC]

Below is my setup (default.uae) for Amiberry that has the issue. It may have something specific to do with the resolution I run in?

This is not an Amiberry issue, as I have multiple versions compiled from v5.x through v7.x - and they all exhibit the behavior as I upgrade/downgrate the sdl2-compat package.

It's not that the mouse doesn't move - it WILL move, but any sort of 'fine grain' movement is impossible. slow mouse movement does nothing, it has to be over a certain threshold of movement before it will move.

Also - I've confirmed this problem AND the 'fix' of backing down this package on two different machines using 3 different mice.

Is there any way to track any differences in mouse handling between these two versions of the package to see what's different?

If there's anything I can log or extract to provide more detailed info, just ask. I'm happy to do whatever to help keep this from continuing to happen... this is like the third time this problem has cropped up in the past few months - sometimes it is magically fixed, and then some change breaks it again - so I suspect there's something janky here. :)

---

config_description=A1200 Default Setup
config_hardware=true
config_host=true
config_version=7.0.7
config_hardware_path=
config_host_path=
config_all_path=
amiberry.rom_path=./
amiberry.floppy_path=./
amiberry.hardfile_path=./
; host-specific
amiberry.middle_mouse=true
amiberry.soundcard=1
amiberry.soundcardname=Built-in Audio Analog Stereo
; common
use_gui=yes
use_debugger=false
kickstart_rom_file=/Games/Emulation/amiberry/roms/Kickstart 3.2.3 (47.115) (A1200)(Hyperion).rom
kickstart_ext_rom_file=
pcmcia_mb_rom_file=:ENABLED
ide_mb_rom_file=:ENABLED
flash_file=
cart_file=
rtc_file=
kickshifter=false
scsidevice_disable=false
floppy_volume=33
floppy0=
floppy0sound=1
floppy0soundvolume_disk=33
floppy0soundvolume_empty=100
floppy1=
floppy1sound=1
floppy1soundvolume_disk=33
floppy1soundvolume_empty=100
floppy2=
floppy2sound=1
floppy2soundvolume_disk=33
floppy2soundvolume_empty=100
floppy3=
floppy3sound=1
floppy3soundvolume_disk=33
floppy3soundvolume_empty=100
nr_floppies=2
floppy_speed=100
cd_speed=100
parallel_on_demand=false
serial_on_demand=false
serial_hardware_ctsrts=true
serial_status=true
serial_direct=false
scsi=true
uaeserial=false
sana2=false
sound_output=exact
sound_channels=stereo
sound_stereo_separation=7
sound_stereo_mixing_delay=0
sound_max_buff=8192
sound_frequency=44100
sound_interpol=anti
sound_filter=emulated
sound_filter_type=enhanced
sound_volume=0
sound_volume_paula=0
sound_volume_cd=0
sound_volume_ahi=0
sound_volume_midi=0
sound_volume_genlock=0
sound_auto=true
sound_stereo_swap_paula=false
sound_stereo_swap_ahi=false
comp_trustbyte=direct
comp_trustword=direct
comp_trustlong=direct
comp_trustnaddr=direct
comp_nf=true
comp_constjump=true
comp_flushmode=soft
compfpu=true
comp_catchfault=true
cachesize=0
joyport0=mouse
joyport0autofire=none
joyportfriendlyname0=System mouse
joyportname0=MOUSE0
joyport1=none
joyport1autofire=none
bsdsocket_emu=false
synchronize_clock=true
maprom=0x0
parallel_postscript_emulation=false
parallel_postscript_detection=false
ghostscript_parameters=
parallel_autoflush=5
gfx_display=0
gfx_display_rtg=0
gfx_framerate=1
gfx_width=720
gfx_height=568
gfx_x_windowed=2162
gfx_y_windowed=448
gfx_width_windowed=720
gfx_height_windowed=568
gfx_width_fullscreen=800
gfx_height_fullscreen=600
gfx_refreshrate=0
gfx_autoresolution=0
gfx_autoresolution_vga=true
gfx_backbuffers=2
gfx_backbuffers_rtg=1
gfx_vsync=false
gfx_vsyncmode=normal
gfx_vsync_picasso=false
gfx_vsyncmode_picasso=normal
gfx_lores=false
gfx_resolution=hires
gfx_lores_mode=normal
gfx_flickerfixer=false
gfx_linemode=double
gfx_fullscreen_amiga=false
gfx_fullscreen_picasso=false
gfx_center_horizontal=none
gfx_center_vertical=none
gfx_colour_mode=32bit
gfx_blacker_than_black=false
gfx_api=direct3d11
gfx_api_options=hardware
immediate_blits=false
multithreaded_drawing=true
ntsc=false
genlock=false
chipset=aga
chipset_refreshrate=49.920410
collision_level=playfields
chipset_compatible=A1200
rtc=MSM6242B
cia_overlay=false
ksmirror_a8=true
pcmcia=true
eclocksync=Gayle
ide=a600/a1200
z3mapping=real
fastmem_size=0
debugmem_start=0x0
debugmem_size=0
mem25bit_size=0
a3000mem_size=0
mbresmem_size=128
z3mem_size=128
z3mem_start=0x0
bogomem_size=0
gfxcard_hardware_vblank=false
gfxcard_hardware_sprite=false
gfxcard_multithread=false
gfxcard_size=128
gfxcard_type=ZorroIII
chipmem_size=16
cpu_speed=max
cpu_throttle=0.0
cpu_type=68040
cpu_model=68040
fpu_model=68040
cpu_compatible=true
cpu_24bit_addressing=false
cpu_data_cache=false
cpu_cycle_exact=false
cpu_memory_cycle_exact=false
blitter_cycle_exact=false
cycle_exact=false
fpu_strict=false
rtg_nocustom=true
rtg_modes=0x112
debug_mem=false
log_illegal_mem=false
kbd_lang=us
hardfile2=rw,DH0:/Games/Emulation/amiberry/harddrives/A1200Workbench32.hdf,0,0,0,512,0,,uae0
uaehf0=hdf,rw,DH0:/Games/Emulation/amiberry/harddrives/A1200Workbench32.hdf,0,0,0,512,0,,uae0
hardfile2=rw,DH1:/Games/Emulation/amiberry/harddrives/A1200MegaGames.hdf,0,0,0,512,0,,uae1
uaehf1=hdf,rw,DH1:/Games/Emulation/amiberry/harddrives/A1200MegaGames.hdf,0,0,0,512,0,,uae1
hardfile2=rw,DH2:/Games/Emulation/amiberry/harddrives/InstallationTools.hdf,32,1,2,512,0,,uae2
uaehf2=hdf,rw,DH2:/Games/Emulation/amiberry/harddrives/InstallationTools.hdf,32,1,2,512,0,,uae2
hardfile2=rw,DH3:/Games/Emulation/amiberry/harddrives/Music.hdf,32,1,2,512,0,,uae3
uaehf3=hdf,rw,DH3:/Games/Emulation/amiberry/harddrives/Music.hdf,32,1,2,512,0,,uae3
filesystem2=rw,Shared:Shared:/Games/Emulation/amiberry/Shared,-128
uaehf4=dir,rw,Shared:Shared:/Games/Emulation/amiberry/Shared,-128
input.config=0
input.joymouse_speed_analog=100
input.joymouse_speed_digital=10
input.joymouse_deadzone=33
input.joystick_deadzone=33
input.analog_joystick_multiplier=18
input.analog_joystick_offset=-5
input.mouse_speed=100
input.autofire_speed=600
input.autoswitch=true
; *** WHDLoad Booter. Options
whdload_filename=
whdload_game_name=
whdload_uuid=
whdload_slave_default=
whdload_slave_libraries=false
whdload_slave=
whdload_slave_data_path=
whdload_custom1=0
whdload_custom2=0
whdload_custom3=0
whdload_custom4=0
whdload_custom5=0
whdload_custom=
whdload_buttonwait=false
whdload_showsplash=true
whdload_writecache=false
whdload_quit_on_exit=false
`

[slouken]

Can you try with the latest main SDL and sdl2-compat? I just tested here and the mouse was equally responsive using both SDL2 and sdl2-compat.

[slouken]

I believe this was related to mouse motion accumulation that was fixed since this bug was reported.

[RodimusFVC]

Ok, I'm currently using 2.32.52.1 - moving to 2.32.56.2 - Will report back in a few. :)

[RodimusFVC]

Nope, still busted. It's smooth when I slam the mouse around, but slow speed movement goes absolutely nowhere. :(

[RodimusFVC]

Just fyi, downgraded back to 2.32.52.1 and it works fine again.

[slouken]

Any chance you can pack up your install with specific steps so I can try to repro here? I've never run amiberry or done Amiga emulation, so noob instructions would be great. :)

[RodimusFVC]

Um, not likely that I can do that as 1) it's 23.5GB in size and 2) it contains the copyrighted rom images from my actual chips...

I can walk you through an install, but you'd need to buy the latest AmigaOS (3.2.3) or at a minimum, v3.1 (from amiga forever is the best way) ... I haven't tried to see if older versions of the OS have the problem - I just assume they all do. (Or do a google search for it and find a less reputable copy)

Also happy to do a discord screenshare and show you what's going on and do some live troubleshooting. I want this darn thing fixed, too.

Once you have amiberry installed, the os up and running, the problem presents itself - and it's drop dead obvious...

The steps are 'install amiberry, install the amiga os on amiberry, and boot it'.... As you move the mouse around, it just won't move at all unless you hit a certain point of acceleration, and then it moves ok.... if you drop under that threshold again, it stops dead.

So whatever was fixed isn't fully fixed? :)

I'm not sure what else to say about it.

I don't know if it matters that I'm on Arch/Manjaro. I can post some system stats, etc.

[RodimusFVC]

Well, ok - I can say this... when amiberry is 'paused' - the emulation is frozen - the mouse works fully normal speed within the GUI that controls it. the moment you start the emulation back up and get back into the emulated environment, back to sludge mouse speeds.

So it's definitely something that rears its head 'in use' as part of the emulation layer interaction with SDL. But it's definitely caused by the sdl2-compat 'layer' because I can use the older package and it works fine vs the newer one which does not.

[slouken]

Well, since I don't have AmigaOS, it sounds like I'm observing the emulation paused case, which as you mentioned is fine.

Can you use git bisect to identify which commit breaks it?