English Amiga Board

English Amiga Board (https://eab.abime.net/index.php)
-   support.WinUAE (https://eab.abime.net/forumdisplay.php?f=5)
-   -   WinUAE 3.6.0 beta series (https://eab.abime.net/showthread.php?t=87810)

Toni Wilen 15 August 2017 17:35

Quote:

Originally Posted by Shaytan (Post 1178347)
Thanks for the answer, it would be great if you could implement this option :)

Done now but I can't really see how this can be useful when 68020/30 and 68040 uses different cache flushing instructions that KS ROM use..

Quote:

Originally Posted by ross (Post 1178364)
Ok, the problem is easy replicable (apart my system there is various adf plagued).

Thanks. Should be fixed now.

Shaytan 15 August 2017 18:28

1 Attachment(s)
Quote:

Originally Posted by Toni Wilen (Post 1178373)
Done now but I can't really see how this can be useful when 68020/30 and 68040 uses different cache flushing instructions that KS ROM use..

Sorry to report that using the latest winuae.7z the problem still persists

Toni Wilen 15 August 2017 18:45

Quote:

Originally Posted by Shaytan (Post 1178387)
Sorry to report that using the latest winuae.7z the problem still persists

Attempt #2 and include log again.

Shaytan 15 August 2017 18:51

1 Attachment(s)
Quote:

Originally Posted by Toni Wilen (Post 1178391)
Attempt #2 and include log again.

Still the same

ross 15 August 2017 19:39

Quote:

Originally Posted by Toni Wilen (Post 1178373)
Thanks. Should be fixed now.

It works. Thanks :)

Toni Wilen 15 August 2017 21:21

Quote:

Originally Posted by Shaytan (Post 1178393)
Still the same

And again.. (This is normal when I can't test it..)

Quote:

Originally Posted by ross (Post 1178407)
It works. Thanks :)

Good. There was some mismatches, for example BSR and RTS used data cache path even when not enabled :)

Shaytan 15 August 2017 22:04

1 Attachment(s)
Quote:

Originally Posted by Toni Wilen (Post 1178433)
And again.. (This is normal when I can't test it..)

I've tried the latest dev build and by looking at the log i can see it triggers the fpu frame hack but then the FRESTORE error shows up resulting in a guru screen

Code:

07-879 [276 001-000]: CPU=68040, FPU=68040, MMU=0, JIT=0. fast
10-831 [451 227-017]: 6888x frame hack
10-831 [451 227-017]: FRESTORE 6888x (68881) invalid frame version 1f 1f380000 400171e6


Toni Wilen 15 August 2017 22:19

Getting closer... What about now?

Shaytan 15 August 2017 22:33

1 Attachment(s)
Quote:

Originally Posted by Toni Wilen (Post 1178446)
Getting closer... What about now?

I think you've made it!

No more gurus and everything seems working fine. :)

I'm attaching the logs anyway, thanks! :great

AMIGASYSTEM 15 August 2017 23:26

Ciao Toni, i was testing the game "Sixth Sense Investigation" and I noticed that with WinUAE 3.6.0 beta there are graphics problems.

With WinUAE 3.5.1 no problem, tested the same game with either OS 3.1 that with 3.9 (also standard not expanded), also tried all versions of Sixth Sense Investigation (German, English, Italian). To immediately get the problem to run icon "SixthSense" no "QuickStart"

Snake79 15 August 2017 23:53

Hi AMIGASYSTEM. I have had similar problem with f.e. Flink game using latest official beta (3.6.0). Try latest, as i call it 'alpha' http://www.winuae.net/files/b/winuae.7z.

AMIGASYSTEM 16 August 2017 00:10

Thank you Snake, today I have not been home and I did not notice Fix, usually check and verify new fixes.

Now with the new fix "Sixth Sense Investigation" no problem all OK.

Snake79 16 August 2017 00:28

You welcome. 'Thanks', should goes to amilo (reported the problem first), ransom and Toni especially (fixed it fast). Greetings.

michaljarz 16 August 2017 05:28

3 Attachment(s)
Hello.
I noticed that there was a problem with the kingcon-handler. In the startup-sequence file I have typed:
Assign con: dismount
Mount CON:


Up to version 3.5.0 works. With version 3.6.0 when loading Workbench I have the message: Can not open file 'DEVS:DOSDrivers/CON'.

Interestingly, if I correct the above entry on:
Assign con: dismount
Mount CON: from devs:mountlist

then version 3.6.0 also works.

PS.
- Disks are mounted as a windows directory.
- It looks a bit like in version 3.6.0 after checking for a file 'DEVS:DOSDrivers/CON' no longer checked file 'DEVS:mountlist'.
- Tested Kingcon-handler 1.3 & 1.8
- There is no CON file in the DEVS:DOSDrivers directory (because Windows can not create a file named CON). Description to CON: is only in Mountlist file.

EDIT:
Logs from 3.5.0 (this version works)
46-171 [37 226-082]: FILESYS: 'CON' illegal filename
46-250 [40 226-254]: FILESYS: 'CON' illegal filename


Logs form 3.6.0 last beta
12-562 [36 224-182]: H=ffffffff TYPE=ffffffff ERR=00000005 'f:\Programy\WinUAE\Dyski\Workbench_czysty_uae\Devs\DOSDrivers\CON'
12-562 [36 225-193]: failed to open 'f:\Programy\WinUAE\Dyski\Workbench_czysty_uae\Devs\DOSDrivers\CON' c0000000 3 err=2
12-640 [40 223-309]: failed to open 'f:\Programy\WinUAE\Dyski\Workbench_czysty_uae\Devs\DOSDrivers\CON' c0000000 3 err=2

Toni Wilen 16 August 2017 09:37

Quote:

Originally Posted by michaljarz (Post 1178515)
Hello.
I noticed that there was a problem with the kingcon-handler.

Should be fixed. (Disabled that extra check if Windows is too old..)

michaljarz 16 August 2017 15:32

Thanks :) It works now.

Toni Wilen 16 August 2017 20:45

http://www.winuae.net/files/b/winuae_3600b2.7z
http://www.winuae.net/files/b/winuae64_3600b2.7z

Beta 2:

- Read Identity and floppy drive bootblock font size ignored DPI setting.
- Multiple 68030 non-MMU and MMU mode fixes.
- 68030 data cache emulation was not compatible with indirect boot ROM. Note that this requires either 68030 CACR Write Allocate bit off or MMU setup that inhibits caching. (If write allocate is enabled and MMU does not inhibit caching: long aligned writes to cache inhibited regions will allocate data cache lines!)
- Previously accidentally working 6888x <> 68040 on the fly switch is now a feature. If unimplemented fpu emu is ticked and FRESTORE frame matches other FPU, use matching FPU emulator to execute it and do not generate exception.
- L debugger command fixed, also length is now optional (=read until end of file)
- AGA bitplane special case missed a check and broke many AGA programs.
- Windows illegal name check fix.

ross 16 August 2017 21:39

Quote:

Originally Posted by Toni Wilen (Post 1178660)
- 68030 data cache emulation was not compatible with indirect boot ROM. Note that this requires either 68030 CACR Write Allocate bit off or MMU setup that inhibits caching. (If write allocate is enabled and MMU does not inhibit caching: long aligned writes to cache inhibited regions will allocate data cache lines!)

Hi Toni, so how to boot AmigaOS (EDIT: through UAE FS Rom) with 68030 and data cache emulation active?
(AmigaOS require Write Allocate because FCx bits is used in cache logic and memory is shared in Supervisor and User mode...).

I tried it and in fact nothing works :sad

[EDIT:well, a modified ROM with an MMU setup for inhibit caching can work, but not too practical ;)]

Cheers,
ross

Toni Wilen 16 August 2017 21:43

Quote:

Originally Posted by ross (Post 1178675)
Hi Toni, so how to boot AmigaOS with 68030 and data cache emulation active? (AmigaOS require Write Allocate because FCx bits is used in cache logic and memory is shared in Supervisor and User mode...).

I tried it and in fact nothing works :sad

- UAE boot ROM must 128k indirect
- MMU emulation must be enabled
- Use Thor's Mu680x0Libs.lha (from Aminet)

Unfortunately emulation needs to be accurate (and 68030 write-allocate is stupid) and indirect boot rom does not have space for separate read and write regions.

Or use any emulated real world controller :)

mark_k 16 August 2017 21:55

One suggested workaround for (virtual in this case :)) hardware designers was to have longword-wide read/write registers not longword-aligned. So register at $xxxx2 instead of $xxxx0 for example.


All times are GMT +2. The time now is 07:10.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.

Page generated in 0.05883 seconds with 11 queries