WinUAE 4.0.0 beta series

[mark_k]

Yes, I meant Amiga lines. E.g. with line mode double and vertical scaling 3×, there are 12 black monitor pixels at the bottom of the Amiga display area. (Easy way to check is to enable End+F9 colour bars, press Pause then PrtSc, paste into MS Paint.)

[Toni Wilen]

It looks it does.. Fixed now but I am not yet sure if it causes side-effects.

[Toni Wilen]

http://www.winuae.net/files/b/winuae_4000b13.7z
http://www.winuae.net/files/b/winuae64_4000b13.7z

Beta 13: (Hopefully final next week)

- Toccata sound was not fully closed when reset/reset and caused crash if new config was loaded and started.
- Returning back to D3D11 fullscreen from GUI didn't automatically recapture mouse. (b12)
- Debugger f <start> <end> was broken.
- Corrupted OSD graphics (if OSD was enabled) and extra dark/bright line was drawn at the bottom of display in lagless D3D11 fullscreen mode after exiting GUI.
- Second to last line was not drawn in lagless mode. Very last line is invisible by design.
- FMOVE and FMOVEM to memory didn't check if destination addressing mode was invalid (PC relative) and didn't generate F-line.
- FMOVE to control register didn't check if more than 1 control register bit was set which is also invalid and should generate F-line.
- FMOVEM to (An)+ and from -(An) are invalid.
- Softfloat FABS and FNEG didn't handle infinity correctly.

[hexaae]

After some reboots with A+A+CTRL or using a reboot shell command WinUAE sometimes freezes with a grey screen, in boot phase, even before the screen goes black and Amiga bootload starts.
Logs: https://nofile.io/f/R8jTHYwPUFm/winuaelog.7z
Used cfg: https://nofile.io/f/QNw8ApnQbJV/Beta+4.0.0.uae

[Toni Wilen]

Originally Posted by hexaae

After some reboots with A+A+CTRL or using a reboot shell command WinUAE sometimes freezes with a grey screen, in boot phase, even before the screen goes black and Amiga bootload starts.
Logs: https://nofile.io/f/R8jTHYwPUFm/winuaelog.7z
Used cfg: https://nofile.io/f/QNw8ApnQbJV/Beta+4.0.0.uae

Question that needs answer: does it happen with 3.6.x official? This thread is beta only as usual.

[jdog320]

When booting up a fresh install of Mac OS 7.6.1, there seems to be an extension that's causing a HALT10, though I can't pinpoint which. Tried on both SS and Fusion but same issue persists.

[Toni Wilen]

Same reply as above.

EDIT: halt10 = supervisor stack pointer pointed to non-existing memory. Almost surely software side-conflict/bug and nothing to do with emulation (or config is something that software didn't expect).

[hexaae]

Originally Posted by Toni Wilen

Question that needs answer: does it happen with 3.6.x official? This thread is beta only as usual.

It starts with 4.0b1.
3.6.10 was ok and I could reboot 100 times without crashes.
Tell me what logs/dumps you need as the task crashes directly without WER requester. Grey screen after Amiga reboot and crashes. Always the same exception code:

Nome dell'applicazione che ha generato l'errore: winuae64.exe, versione: 4.0.0.0, timestamp: 0x5b2007a4
Nome del modulo che ha generato l'errore: ntdll.dll, versione: 10.0.17134.112, timestamp: 0x6529f37c
Codice eccezione: 0xc0000374
Offset errore 0x00000000000f4d1b
ID processo che ha generato l'errore: 0x49e8
Ora di avvio dell'applicazione che ha generato l'errore: 0x01d403cfbc72061f
Percorso dell'applicazione che ha generato l'errore: D:\WinUAE\winuae64.exe
Percorso del modulo che ha generato l'errore: C:\WINDOWS\SYSTEM32\ntdll.dll
ID segnalazione: 39f59d7c-c24b-40df-b7ef-6af3f8f75816
Nome completo pacchetto che ha generato l'errore:
ID applicazione relativo al pacchetto che ha generato l'errore:

[Toni Wilen]

Dump file is needed but only if you can duplicate it without JIT.

[hexaae]

there is no dump file saved in WinUAE dir after (silent) crash

[Toni Wilen]

Originally Posted by hexaae

there is no dump file saved in WinUAE dir after (silent) crash

Try in windowed mode. EDIT: Make sure nothing is disabled in windows error reporting.

[hexaae]

No help from window mode for both native & rtg.
WER is ok, but this looks like a Windows vs WinUAE bug, so WinUAE (silently) crashes before the dump, closing also the -log window. From the last log lines:

54-308 [2600 000-000]: -> -360.000000 270.341614 360.000000 -270.341614 0.957447 0.941957
54-309 [2600 000-000]: Buffer size (752*574) Native
54-311 [2600 000-000]: ActiveHeight: 1080 TotalHeight: 1111 VFreq=75002/1000=75.00Hz HFreq=173320000/2080=83.327KHz
54-311 [2600 000-000]: ActiveHeight: 1080 TotalHeight: 1111 VFreq=75002/1000=75.00Hz HFreq=173320000/2080=83.327KHz
54-311 [2600 000-000]: RTGFREQ: 312*49.9204 = 15575.1680 / 75.0 = 207
54-311 [2600 000-000]: D3D11 Shader and extra textures restored
54-318 [2600 000-001]: TOD HACK reset 50,60 2615,2615
54-322 [2601 169-002]: CPU reset PC=1fffac48 (Zorro III Fast RAM)..
54-322 [2601 169-002]: clipboard: reset (1008c5c0)

I tried to disable shared clipboard but nothing changed.
0xc0000374 from the net looks like an heap corruption...

[hexaae]

@Toni
Found the culprit!
I use in my WB env: Devs/DOSDrivers/CD0 loaded on startup, containing the following CD0: mountlist:

/***************************************************************/
/*  CacheCDFS mountlist entry © 1993 Elaborate Bytes, O. Kastl */
/***************************************************************/
	FileSystem     = L:CacheCDFS /* The name of the game */
	Device         = "uaescsi.device" /* Name of exec device driver */
	Unit           = 0 /* exec device unit */
	Flags          = 0 /* OpenDevice flags */
	BlocksPerTrack = 1 /* Unused */
	BlockSize      = 2048 /* True, but unused */
	Mask           = 0xffffffff /* Memory mask for direct read */
    MaxTransfer    = 0xffffffff /* Maximum amount of bytes for direct read */
	Reserved       = 0 /* Unused */
	Interleave     = 0 /* Unused */
	LowCyl         = 0 /* Unused */
	HighCyl        = 0 /* Unused */
	Surfaces       = 1 /* Unused */
	Buffers        = 8 /* Number of cache lines */
	BufMemType     = 0 /* MEMF_PUBLIC|MEMF_24BITDMA */
	BootPri        = 2 /* Boot priority for MountCD */
	GlobVec        = -1 /* Do not change! */
	Mount          = 1 /* Mount it immediately */
	Priority       = 10 /* Priority of FileSystem task */
	DosType        = 0x43443031 /* Currently unused */
	StackSize      = 4096 /* Minimum stack required is 3000! */
	Control        = "MD=16 LC=1 DC=1 S L LV AL LFC=1 HR=.rsrc Q AUDIO=S:BatchFiles/StartCDAudioProgRX"
   /* The Control field is for special adjustments */
   /* L/S convert all file/volume names to lowercase */
   /* LV/S convert volume names to lowercase */
   /* AL/S Auto-Lower converts only non-Amiga CDs */
   /* LFC/N start converting at this character */
   /* LC/N/A number of blocks per cache line */
   /* DC/N/A number of cache lines for the data cache */
   /* MD/N/A number of blocks, when starting direct read, not using */
   /*        the cache. 0 will be a reasonable default (LC*DC+1) */
   /* S/S Do SCSI direct commands, no Trackdisk like commands! */
   /* NC/S Do NOT use TD_ADDCHANGEINT, poll for DiskChange! */
   /* M/S Issue a TD_MOTOR (OFF) command after read */

For some reason this causes big issues here, but only on subsequent WB reboots (never on very first boot). At reboot and in normal Windows usage, sometimes, I can actually hear the drive like "reactivating": usual Intel power state hack…
Maybe this causes some problems to WinUAE and device emulation.


Attached CD0 and CacheCDFS required for CD0: https://nofile.io/f/fYvRAV88o9n/xToni.zip

[Toni Wilen]

I can't duplicate. Does it still happen if you remove dosdriver mounting from startup-sequence and manually mount it after WB has booted? (shell "mount cd0:"?)

Run with -log -logflush -scsilog -scsiemulog command line parameters and attach the log after crash.

[hexaae]

No, can't reproduce if I mount it manually by double-click every time after WB boot and I don't leave it in DEVS:DOSDrivers.
The log window output is the same saved in the winuaelog.txt file? Because I don't have the time to collect it since WinUAE suddenly crashes silently closing that log with no .dmp file either…


EDIT:
Adding a Wait 1 in my sseq before the mount is a valid workaround and WinUAE won't hang anymore: seems a too fast loading can cause the issue, even with JIT off.

My Startup-Sequence

...
BindDrivers
Wait 1
C:Mount >NIL: DEVS:DOSDrivers/~(#?.info)

;IF EXISTS DEVS:Monitors
;  IF EXISTS DEVS:Monitors/VGAOnly
;    DEVS:Monitors/VGAOnly
;  EndIF
;  C:List >NIL:  DEVS:Monitors/~(#?.info) TO T:M LFORMAT "DEVS:Monitors/%s"
;  Execute T:M
;  C:Delete >NIL: T:M
;EndIF

LoadMonDrvs  
...

EDIT2:
never mind, WinUAE crashed on A+A+CTRL again from WB (but never using the shell cmd!?) :(

EDIT3:
aargh! 1/30+ crashed after reboot even when CD0: gets started manually (checked with a loop script)

EDIT4:
in short: it won't EVER crash only when NOT using CD0:, I'm almost sure it's a Intel hack's powersaving side effect. Maybe rebooting WinUAE finds the device in a strange status or simply slow to kick-in and crashes...

[Mad-Matt]

Originally Posted by jdog320

When booting up a fresh install of Mac OS 7.6.1, there seems to be an extension that's causing a HALT10, though I can't pinpoint which. Tried on both SS and Fusion but same issue persists.

I believe its the Appleshare Extension that needs removing to boot 7.6.1 properly. It does happen on real hardware as well so not really emu related.

[Toni Wilen]

Originally Posted by hexaae

No, can't reproduce if I mount it manually by double-click every time after WB boot and I don't leave it in DEVSOSDrivers.
The log window output is the same saved in the winuaelog.txt file? Because I don't have the time to collect it since WinUAE suddenly crashes silently closing that log with no .dmp file either…

-logflush flushed file to disk after each file so crash won't cause lost log messages.

It could be driver/drive issue too but lets try some more debugging:

- does it happen if you have CD in drive?
- does it happen if you use CD image? (Harddrives/CD panel: select image file, not CD drive letter)
- If you have some virtual CD program (that simulates true CD drive, Windows' own isn't good enough): does same happen if you use it instead of real drive?
- when you added delay to startup-sequence, does it reduce crashes if delay is at the end of s-s or just before loadwb? (Trying to find out if it is WB that does something with the CD or some other program in s-s)

400b1 does not have any CD access related changes which makes this as weird as your older sound/slowdown issue..

[hexaae]

Using from shell or WHDLoad installer very small values for JIT cache (e.g. uae-configuration cachesize 8) makes WinUAE 4.0 beta freeze and 3.6.1 popup error "Couldn't allocate pool size!"... Maybe it's better to add some safe checks? Must be a multiply of <n> (?) or the whole range between 0-max is allowed?

EDIT: what about the single line option to control direct/indirect JIT CACHE (e.g. uae-configuration comp_trust direct/indirect)?

[hexaae]

Originally Posted by Toni Wilen

-logflush flushed file to disk after each file so crash won't cause lost log messages.

It could be driver/drive issue too but lets try some more debugging:

1 does it happen if you have CD in drive?
2 does it happen if you use CD image? (Harddrives/CD panel: select image file, not CD drive letter)
3 If you have some virtual CD program (that simulates true CD drive, Windows' own isn't good enough): does same happen if you use it instead of real drive?
4 when you added delay to startup-sequence, does it reduce crashes if delay is at the end of s-s or just before loadwb? (Trying to find out if it is WB that does something with the CD or some other program in s-s)

400b1 does not have any CD access related changes which makes this as weird as your older sound/slowdown issue..

Before we start, notice I use these settings in WinUAE:



1: yes, still happens
2: yes, still happens
3: I don't use them. Before Win7/10 I used WinCDEmu… Should I try creating a virtual CD drive letter, then in WinUAE assign ("Add SCSI/IDE CD Drive" button) a CD drive to it, and mount it from Amiga mountlist?
4: only slightly alleviate the problem but after many attempts still happens. Reduces the chances a little bit, doesn't matter where I put it in my SSeq.

Another strange thing, it ALWAYS crashes if:
- after very first boot with CD0: in DEVS:DOSDrivers
- I use A+A+CTRL to reboot
In this case WB looks freezed for a couple of secs and WinUAE silently crashes, not even reboot + grey screen attempt. A C:reboot cmd from shell instead will trigger reboot + the grey screen etc.

[Toni Wilen]

Ok, so it is not CD drive related if CD image does the same. Log using above command line (-log -logflush -scsilog -scsiemulog) would be really useful because it is most likely some SCSI command that does something bad.