On Sat, Jul 24, 2021, 6:43 AM Eric Auer <e.a...@jpberlin.de> wrote: > > Hi Jeremy, > > does that mean the unstable kernel already supported Win 3.1 386enh? >
Yes, but it didn't work as well for running a dos prompt, 1st character typed was lost and possibly other issues > Cool to know :-) How about Windows for Workgroups in 386 mode, which > is "non safe mode" there, so features are lost without it in WfW 3.11? > I have not yet tested it yet > Thanks for cherry-picking all the relevant patches! I guess the FDPP > kernel of DOSEMU2 will have some more of those for you ;-) > > Checking your video, it says kernel 2043 build Jul 24 2021, > but the copyright messages says 1995-2012, probably a typo. > I guess that date hasn't been updated in a while :-) > Do you have a link to the relevant patchsets for proof-reading > in case there is a risk of regressions? > I will post them later today, but yes more work is needed to ensure no regressions > I see you are using the Microsoft HIMEM 3.07 (02/14/92), Microsoft > EMM386 4.44 (1991) and Microsoft SMARTDRV, are all of those actually > necessary? I would expect things to also work with HIMEMX or XMGR, > as long as no free EMM386 is loaded For my initial work I wanted to limit possible issues, so I started with versions I know work with Windows 3 since they came with it. These specific versions are not a requirement. at all? Why do you use 4DOS in > the DOS window inside Windows? Any I believe it is because I have 4dos as my c:/command.com so that is being spawned, but I may have set it explicitly at some point in the past. I originally tested with 4dos just like other programs, because it is known to work. special system.ini [386enh] items? > I don't recall, this is a virtual machine I set up years ago. I am going to create a new one from scratch and document the steps so anyone (with win3) can reproduce, just haven't had time yet See my notes below :-) Is setting VERSION=6.0 required, too? > No, just minimizing possible issues for initial testing > Cool that WIN, WIN /3 and WIN /S apparently all work :-) > Some custom (un-)settings from an old system.ini [386enh] of mine: > > ; device=lanman10.386 > ; mouse=lvmd.386 > ; network=*dosnet,*vnetbios > ; old version: device=*vtd new version: device=vtda.386 (in "WW0981" fix) > device=vtda.386 > FileSysChange=0 > PagingFile=C:\WINDOWS\WIN386.SWP MaxPagingFileSize=1024 > ; also: PermSwapDOSDrive=... PermSwapSizeK=... > ; disable swapfile stuff: > Paging=0 > ; prepare for more than 200 breakpoints, 150 is minimum useful: > MaxBPs=768 > ; better if lots of RAM: > PageOverCommit=1 > ; equivalent of /D:FSVX > 32BitDiskAccess=No > SystemROMBreakPoint=No VirtualHDIrq=No > ; *** EMMEXclude=A000-FFFF > NoEmmDriver=1 > IgnoreInstalledEMM=1 > WinExclusive=1 > ; > TimerCriticalSection=10000 > DMABufferSize=64 > XlatBufferSize=128 > KeyBoostTime=.005 > MinUserDiskSpace=5120 > PageBuffer=32 > Com1Buffer=512 > ComBoostTime=20 > Com1AutoAssign=-1 > ScreenLines=50 > ; > InDOSPolling=1 > ; P.V.F.: 10, or 0 if share installed > PerVMFILES=0 > ReflectDosInt2a=1 > INT28Critical=1 > ; I.V.W.U.T.: 1/2/4/*8*... seconds: how often to pump int 8/1c into idle > VMs > IdleVMWakeUpTime=1 > ; D.P.E.I.: enable to get explanation how to leave DOS box when starting > one > DOSPromptExitInstruc=0 > ; force DMA buffers to be in 1st 1MB range: > DMABufferIn1MB=1 > > For standard mode, I also had those settings: > > [standard] > ; Stacks=12 (8..64) StackSize=384 - Settings for DOSX DOS Extender > ; Int28Filter=0,1..*10*..: only let through every Nth int 28 ... > Stacks=16 > StackSize=512 > Int28Filter=1 > DOSPromptExitInstruc=0 > > Note that Windows 3.x all have issues when you have too much > RAM. There are binary patch files for that, but the obvious > other workaround is telling your EMM386 or HIMEM or similar > to not let Windows know how much RAM you really have ;-) > I went the simple way and limited the memory to my VM and it's disk size. > Cheers, Eric > > > https://youtu.be/35OQjLYdvJ0 > > > For the technical aspect - the changes are minimal to the kernel, > > added support for a few int 2F function calls that were never merged > > in was about all it took. All significant changes behind a > > WIN31SUPPORT #ifdef so doesn't need to be compiled in if unwanted. > > I will post an update with better details as soon as I can, hopefully this evening. Jeremy
_______________________________________________ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
