WinUAE 4.0.0 beta series

[hexaae]

Originally Posted by Toni Wilen

Crash dumps from winuae.7z are useless. But redownload, run with -log window open and attach last 10 lines when it crashes. When inserting the device and also when removing the device.

Redownloaded... it is stable now: no more crashes plugging in or removing the USB joystick!

[Toni Wilen]

Originally Posted by hexaae

Redownloaded... it is stable now: no more crashes plugging in or removing the USB joystick!

Great. But I suspect you got some cached copy of winuae.7z because today's update shouldn't have made any difference.

Originally Posted by Amiga68k

Yes, lots of mouseupdate 1 x*y log entries.

Ok, next step. Redownload winuae2.7z. Does cursor now appear as single white line? Does log now keep outputting different lines? Attach last 20 lines or so.

[Amiga68k]

Originally Posted by Toni Wilen

Ok, next step. Redownload winuae2.7z. Does cursor now appear as single white line? Does log now keep outputting different lines? Attach last 20 lines or so.

At first I didn't see the white line. I suspect I didn't see it because the pointer starts at the top left of the screen when WB starts up, a part where the background is white. I moved it to the middle of the screen (remember, I still see it moving when left mouse button is pressed) and indeed see a white line then (see picture), the cursor is also still there though.

Couple of entries:

setcursor 0 618x341 720x576 1 0
618.0x341.0 0x0
-> 618.0x341.0 1.0 1.0
setcursor 0 593x363 720x576 1 0
593.0x363.0 0x0
-> 593.0x363.0 1.0 1.0
setcursor 0 558x374 720x576 1 0
558.0x374.0 0x0
-> 558.0x374.0 1.0 1.0
setcursor 0 535x374 720x576 1 0
535.0x374.0 0x0
-> 535.0x374.0 1.0 1.0
setcursor 0 512x368 720x576 1 0
512.0x368.0 0x0
-> 512.0x368.0 1.0 1.0
setcursor 0 491x353 720x576 1 0
491.0x353.0 0x0
-> 491.0x353.0 1.0 1.0
setcursor 0 478x335 720x576 1 0
478.0x335.0 0x0
-> 478.0x335.0 1.0 1.0
setcursor 0 470x320 720x576 1 0
470.0x320.0 0x0
-> 470.0x320.0 1.0 1.0
setcursor 0 466x316 720x576 1 0
466.0x316.0 0x0
-> 466.0x316.0 1.0 1.0
setcursor 0 464x316 720x576 1 0
464.0x316.0 0x0
-> 464.0x316.0 1.0 1.0
setcursor 0 464x316 720x576 1 0
464.0x316.0 0x0
-> 464.0x316.0 1.0 1.0
setcursor 0 464x316 720x576 1 0
464.0x316.0 0x0
-> 464.0x316.0 1.0 1.0

[Toni Wilen]

Originally Posted by Amiga68k

the cursor is also still there though.

Does the line still appear when normal cursor graphics is missing? Or did the cursor stop disappearing now?

[Amiga68k]

Originally Posted by Toni Wilen

Does the line still appear when normal cursor graphics is missing? Or did the cursor stop disappearing now?

The cursor is *always* there

[Toni Wilen]

Originally Posted by Amiga68k

The cursor is *always* there

Your original report says:

Sometimes when WB starts up I don't see a mouse at all, sometimes I see it, but it is 'stuck'.

So did something change if it is now always visible? "stuck" part is still unchanged?

[Amiga68k]

Originally Posted by Toni Wilen

So did something change if it is now always visible? "stuck" part is still unchanged?

Just retried around 20 times, and I can't reproduce it not showing... stuck though: always (except when holding left mouse button)

[hexaae]

Originally Posted by Toni Wilen

Great. But I suspect you got some cached copy of winuae.7z because today's update shouldn't have made any difference.

I think so Glad it's now stable for me.

[Kalamatee]

Originally Posted by Amiga68k

Just retried around 20 times, and I can't reproduce it not showing... stuck though: always (except when holding left mouse button)

I get the same behaviour testing AROS in WinUAE beta 8.

winuae.7z -> Doesnt update when moved unless the mouse buttons are being pressed.

winuae2.7z -> the first row of the pointer is a white line, and it doesnt update when moved unless the mouse buttons are being pressed.

[Toni Wilen]

Just wondering... does the mouse start moving if you press middle mouse button? (and middle mouse is not uncapture mouse button)

[Amiga68k]

Originally Posted by Toni Wilen

Just wondering... does the mouse start moving if you press middle mouse button? (and middle mouse is not uncapture mouse button)

Nope, no activity with middle mouse button.

[Toni Wilen]

Ok, lets try something else..

- does it happen in D3D9 mode?
- boot to RTG mode, mouse is not moving, run some program that opens native mode (for example sysinfo), does mouse move? if you switch back to RTG screen, what happens?
- does mouse really work normally if "hardware sprite" option is not ticked? I can't see how it can affect mouse movement. (It can make mouse look different or disappear completely)
- both left and right mouse buttons still work normally? (can select icons, can open menu
- weird option: load your config, change CPU panel FPU option to "softfloat", start emulation. does anything change?

[Amiga68k]

Originally Posted by Toni Wilen

Ok, lets try something else..

- does it happen in D3D9 mode?

Yes

- boot to RTG mode, mouse is not moving, run some program that opens native mode (for example sysinfo), does mouse move? if you switch back to RTG screen, what happens?

I used Screenmode to get a High Res screen: mouse behaves as it should, reverting to RTG mode: mouse stuck.

Sysinfo: mouse behaves as it should, after quitting: mouse stuck.

Strange result in PPaint, it opens a UAE 1024x768 (8 bit) screen as default. The area where you can draw, the mouse behaves correctly (!), when going to the left, the pointer gets 'stuck' as soon as it enters the icon area and it gets stuck at the top as soon as it enters the 'right-menu' area... weird!! (picture with area's attached)

That got me thinking, does it have to do with screen depth perhaps? But after choosing the 8 bit variant of the RTG WB screen, mouse was still stuck...

- does mouse really work normally if "hardware sprite" option is not ticked? I can't see how it can affect mouse movement. (It can make mouse look different or disappear completely)

Yes, then all is good...

- both left and right mouse buttons still work normally? (can select icons, can open menu

Yes. You don't see the mouse moving, but it actually goes where you want it to go. Opening drawers, right menu, double clicking utilities, it all works...

- weird option: load your config, change CPU panel FPU option to "softfloat", start emulation. does anything change?

I see that option, but I cannot select it...!? When clicking it, it just displays Host (64-bit). When choosing 64-bit (instead of 80-bit default) it is still stuck.

[Toni Wilen]

Originally Posted by Amiga68k

I see that option, but I cannot select it...!? When clicking it, it just displays Host (64-bit). When choosing 64-bit (instead of 80-bit default) it is still stuck.

Right, you have JIT FPU enabled which is not softfloat compatible.

Does 64-bit option change anything? Or if you disable JIT and then select softfloat?

[Amiga68k]

Right, you have JIT FPU enabled which is not softfloat compatible.

Originally Posted by Toni Wilen

Does 64-bit option change anything? Or if you disable JIT and then select softfloat?

Unfortunately, same (stuck) result with both options.

[Toni Wilen]

It really should be fixed now. (winuae.7z).

It was as simple as if nothing changed (cursor does not count because it is not part of bitmap if hardware sprite), screen was not rendered.. I always had RTG on screen leds enabled: it forced screen rendering every frame.. (and when not, I had multithreaded on which does the same..)

[Amiga68k]

Originally Posted by Toni Wilen

It really should be fixed now. (winuae.7z).

Pff, confirmed!

It was as simple as if nothing changed (cursor does not count because it is not part of bitmap if hardware sprite), screen was not rendered.. I always had RTG on screen leds enabled: it forced screen rendering every frame.. (and when not, I had multithreaded on which does the same..)

Urgh, nasty one ... but cracked

[Toni Wilen]

http://www.winuae.net/files/b/winuae_4000b9.7z
http://www.winuae.net/files/b/winuae64_4000b9.7z

Beta 9:

- D3D11 mode screen was not fully refreshed when returning back to fullscreen RTG mode with ALT-TAB/GUI exit
- Inserting or removing USB device with very long internal device name caused a crash.
- Inserting USB device may have also logged some other already connected device as newly inserted.
- Fixed JIT FPU FETOXM1.
- If D3D11 mode with level 9.x hardware: test in init phase if main texture creation succeeds, fallback to D3D9 if it fails. It may unexpectedly fail with some old and broken drivers. Previously it would have caused blank screen.
- RTG screen was not updated if only hardware mouse cursor moved and RTG OSD was not enabled or multithreaded RTG was not enabled. (b8)

[freddy]

Toni, can you activate the string for translation ?

[vagrant]

Originally Posted by Toni Wilen

I'll make some adjustments to make debugging easier in next beta..

Toni I just want to report on-the-fly switching still results in a freeze.
Though it could be pointless information as I'm not sure anything has changed yet in last 2 betas