OS4 compatible UAE expansion development status

[Toni Wilen]

Originally Posted by Aegis

That's interesting since whilst SCSI plays the DVD .ISO, there's no audio

And yes, same configs - only differences are:

1) SCSI instead of UAE (duh )
2) Automount CDFS is turned on for UAE.

Then there is some unknown variable and this test becomes invalid

Which leads me to this question - *without* Limit size of directory drives to 1G WinUAE/OS4 is reporting the drive as 931G (705 used/225 free) - the actual formatted size of the drive is 1.81TB (1.37 used/451 free) - is that correct for WinUAE?

*Edit* just checked it in AF2016/WB3.X/3.2.2 - same deal - volume size/free space is half of what's actually there (different drive: actual size: 279GB, used: 249GB, free: 30GB / WinUAE size: 139GB, used: 124GB, free: 15GB).

I get halved size under OS41FE but correct in OS3.9. Probably block size related or calculation changed. Working in OS3.9 = everything seems to be correct

[Aegis]

Originally Posted by Toni Wilen

Then there is some unknown variable and this test becomes invalid

Hmm... maybe best to put this on the backburner until I can come up with a better test then? Aside from this issue, beta 6 has been really reliable

Originally Posted by Toni Wilen

I get halved size under OS41FE but correct in OS3.9. Probably block size related or calculation changed. Working in OS3.9 = everything seems to be correct

Weird. I just tried it out in 3.9 too and it's still reporting half the sizes it should be - oh well still - It's not like it's gonna burn down my house is it?

[Toni Wilen]

Got hang which means there is still some bug somewhere..

(btw, drive size reporting is fixed)

[Aegis]

Originally Posted by Toni Wilen

Got hang which means there is still some bug somewhere..

I'm back on b6 for now - maybe I'll find something repeatable in b7.

Originally Posted by Toni Wilen

(btw, drive size reporting is fixed)

Excellent btw, I noticed an issue with Blender if 'Hide all uae autoconfig boards' is enabled - it reports a minigl error and can't open its screen (haven't noticed it impacting anything else).

[Toni Wilen]

Can you also duplicate under OS4 that C:Info shows only blank lines in place of directory filesystem drive's information? Weird..

[Aegis]

Originally Posted by Toni Wilen

Can you also duplicate under OS4 that C:Info shows only blank lines in place of directory filesystem drive's information? Weird..

Huh. Yeah - that's weird! There should be a DH3: there...

*Edit* btw Toni, how does this work?

Originally Posted by winuae.net

Official CSPPC/BPPC flash updater can be used to install full ROM
image without having existing ROM image.

[Toni Wilen]

Originally Posted by Aegis

*Edit* btw Toni, how does this work?

Create empty file (zero size or file that only contains zero bytes) as CSPPC/BPPC ROM image. Boot, run official flash updater.

(btw, don't ask offtopic questions in this thread)

[Toni Wilen]

Originally Posted by Aegis

Huh. Yeah - that's weird! There should be a DH3: there...

It is. For example WB3.x C:Info under OS4.x works normally and all device and volume parameters look normal.

[Toni Wilen]

beta 7: warn reset loop fixed, free space (not really OS4 specific) is now correct. ACTION_FILESYSTEM_ATTR supported, Ranger shows filesystem details.

[mritter0]

When using uaegfx the boot anim for OS4 is not shown. Does it not "kick in" until WB is shown?

[Michael]

You get the boing only when no HDD, CD or Floppy present in drives
With latest updates you don't get the boot menu on the classic setup :-(

[Toni Wilen]

Originally Posted by mritter0

When using uaegfx the boot anim for OS4 is not shown. Does it not "kick in" until WB is shown?

I guess there is some early boot special case for some "popular" RTG boards because uaegfx is not initialized until few seconds before WB screen appears.

[Aegis]

b7's working really well for me - are you still planning bsdsocket.library support for 3.3.0?

[Toni Wilen]

Originally Posted by Aegis

b7's working really well for me - are you still planning bsdsocket.library support for 3.3.0?

I have been slowly converting it to new trap system but it is really complex and uses traps really heavily. It probably will be uselessly slow or does not work at all..

Also I am not sure if you can replace OS4 bsdsocket.library without side-effects.

[Toni Wilen]

There may be a bit hackish method to increase performance and enable near-direct memory access: Get PPC CPU MMU state from PPC emulator and use it to translate logical to physical address. If page's physical address is found (page exist in MMU tables): use physical address directly, if not found: use normal indirect method.

This is currently just an idea but should be possible, at least in theory...

[Aegis]

Originally Posted by Toni Wilen

This is currently just an idea but should be possible, at least in theory...

Please give us an incredibly unstable test build to try - that sounds awesome! (and I have lots of backups ).

[Toni Wilen]

winuae.7z updated. Experimental modified "busy-wait" mode traps. It appears to be faster. Maybe. (No Amiga-side interrupts needed for communication)

Make sure to test every part separately, it could make big filesystem reads slower but RTG blits faster etc..

[Aegis]

Originally Posted by Toni Wilen

Make sure to test every part separately, it could make big filesystem reads slower but RTG blits faster etc..

Having problems with this build - can't open any volumes upon Workbench booting - doesn't seem to be applying Workbench prefs either (no customizations to the title/menu bar) - I'll go through it in depth later to see if I can figure out where it's falling over.

[Toni Wilen]

Try switching off network emulations and clipboard sharing temporarily. There appears to be some unknown incompatibility..

[Aegis]

Originally Posted by Toni Wilen

Try switching off network emulations and clipboard sharing temporarily. There appears to be some unknown incompatibility..

Tried them both separately - networking is what's screwing with OS4 loading/running properly (both uaenet and A2065 Z2 tested) - disabling clipboard sharing seems to prevent some random lock-ups I was getting after I disabled networking - thanks Toni.

And yeah - initial impressions are tangibly improved RTG performance - it just feels overall a little snappier - more useful feedback to follow

*Edit* Getting a few random lock-ups just running various apps/games even with networking/clipboard sharing turned off - all most of the stuff I've tried works - no obvious incompatibilities, just some stability issues. Any info I can give you that'll help?

*Edit2* Damn! Some of these apps load/run *much* faster now OWB and pFTP are consistently hanging OS4 (though not WinUAE) - guess that's probably networking-related although Odyssey, iBrowse and NetSurf run fine (well, as best they can without a network connection )