Ninho

Still reinstalling+tweaking Win 98 SE on an older system. This has USB 1.x only, NOT USB 2, controllers (SiS 5595B chipset) Thinking of installing Maximus Decim's NUSB. Therefore before I let any USB storage device touch this computer, I will ask if this is the right road to go for getting universal support of USB mass storage devices (read : pocket flash drives, different brands), or is NUSB only for systems which have USB 2 ? If the latter, what is the proper thing to do instead ? Thank you ! -- Ninho

First, I must correct my above disgusted statement about windows update somewhat : That was using the original Internet explorer version bundled with windows 98 SE (aka IE5). However, a visit to Microsoft Download Center (not windows update!) let me find and install IE6 SP1, localised in French -> For whomever it might help, here's a handcrafted, direct URL - warning : URL may change in a not so far future : <http://download.microsoft.com/download/ie6sp1/finrel/6_sp1/W98NT42KMeXP/FR/ie6setup.exe> ... after installing the above, the windows Update, version 4, site does offer 24 critical and 42 non-critical updates for a start :=) So, it appears MS are still maintaining the updates for this unsupported OS - although the clueless who won't know to update their internet explorer first thing after (re)installing the OS will never know. Good to know ! Thanks Ponch ! Cheers, -- Ninho

Hi ! Due to a failed 40 Gig Maxtor diamond max (grr...) I've lost all my OS installations. Repairs will be long... Anyhoo, for a start, I've reloaded a fresh Windows 98 SE OEM (French) to an older computer. Went to the MS update, and behold! much to my distress, Windows Update says /zero/ update for my OS ! I knew they had stopped adding /new/ updates, but I /thought/ I would still get all the updates released until end of support ; how naive I was ! Hence my question, how could I update that system ? I mean, it's a French edition - I have no choice here - is there an "unfficial service pack" for this ? Or at least a list of applicable updates, with URLs to fetch them from ? Thank you ! __ Ninho

Good! Looks like we are a plethora of virtualisation adepts with differing scope and needs here. One plus point of VMware which has not been mentioned yet is the ability to run an installed OS from a physical hard disk (not for the faint of heart but so cool!) Not possible or at least not documented with MS VPC far as I know. Now I think I'll ask a question : among the virtual machine software products, is there a (free) one that will run on g'd old Windows 9x ? I know the old versions of Connectix VPC used to, but they are not free. Oh, I should add the product should not make use of hardware virtualisation facilities (IA or AMD64)... Cheers,

Replying myself... That post at the old VMware forum exposed the same problem, <http://communities.vmware.com/message/442873;jsessionid=966D1C1775C6CE1D3290BA89846D31BA#442873> and found a workaround (disable acceleration in VMware options... whatever that affects exactly remains to determine, but at least I can confirm it does allow Win32s applications to run) :=) -- Ninho

I presume this section isn't too inappropriate for discussing Windows for workgroups (3.11) questions... at least, none seems to exist that would be more adequate ;=) I run a fully tuned WfWG in VMware Player v1 - host sytem is Windows 2000 SP4. Everything's working great, incl. LAN, the internet (IE 5 !), mouse, sound, SVGA 1024*768*16 bit colours display, taming (i.e., halt when OS idles)... With Calmira added, the OS becomes very much usable and even pleasant. There is /one/ thing which I couldn't make work, however, and is the "Win 32s" add-on : as soon as I would launch a win32 app, the OS inside of the VM freezed - even the three finger salute has no effect (but the mouse pointer still reflecting physical mouse moves - looks like the core system is not dead and even taking mouse interrupts. No further user interaction possible however, clicks or key presses not operative in the guest system) ; on resetting the VM using VMware player's controls, Player - or Workstation - will invariably fail, giving rise to a bug # 4410 followed by guest memory dump Jun 27 23:15:36: vmx| ASSERT C:/ob/bora-19175/bora/devices/misc/dma.c:1539 bugNr=4410 Again, this happens /only/ with Win 32s installed in the VM, but then it happens consistently. Removing the win32s components restaurs full functionality (including correct shutdown of Windows inside the virtual machine). It is difficult to know if this fault is related to (something in my particular installation of) win 3.11, or a plain VMware bug, or a mix... Here are my questions to anyone has been running WfWG, within or without VMware : have you met, or heard of, something close to my problem ? Experienced bugs with the Win32 subset ? Incompatibilties with VMware ? Please report about specifics... and of course, known or proposed remedies... Might the problem go away if I were to install a later version of VMware (which I'm not in a hurry to do, fully satisfied with this oldie) ? Also, any info about VMware's bug number 4410 is welcome - couldn't find anything in the VMware database; indeed, been unable to access the database since they "upgraded" their website :=( Thanks all for any hints/evidence ... -- Ninho

VMware is more potent, esp. for the networking options, USB support, and installing Linux in VMs; I've run VMware both in windows and Linux hosts and fully satisfied. Recently though, I started using also VPC in order to run their windows XP-SP2-with-IE7 free from MSDN (time bombed demo, but for me a way to explore Windows XP). I'm not displeased with it, but it's not on par with VMware. FWIW, cheers -- Ninho edit: now seing this poll is very old. Sorry, or rather, not, for resurrecting it. If others want to take the poll we'll see how VPC is doing vs. VMware in these days...

As you say, it's crazy ! Did you patch your copy of VMware ? What product/verssion ? Asking cause in principle, VMware will check and refuse to run inside a VM - using the backdoor API. In any case you must take a very serious penalty hit for the inner VMs. I assume your physical system has processor virtualisation extensions available ? Amusing sight anyway -- Ninho

It certainly did help. Unsetting the "KernelModuleAuth" let fwdrv.sys launch itself without blue-screening. However this is not the end of the story, as I first thought. Instead it uncovered a further problem, leading to another Stop-D1 blue screen, this time accusing NDIS.SYS ! This seems to occur whnever the firewall engine, fwdrv.exe, is launched as a service (as it should, really). OTOH no blue screen occurs and the firewall /seems/ to be working if fwdrv is launched manually from the desktop (with admin privilege of course). However I believe it's not practical - and maybe not secure - to launch the FW engine from the desktop. Well, I surely am not going to struggle with undocumented settings in an otherwise excellent but sadly unsupported FW ;=) Just consider this report a pebble added to the online collection of experience about KPF 2.1... Thanks again

GrofLuigi ! I have to apologise for not quite believing, and thank you! Indeed, changing the KernelModuleAuth value to 0 (from 1) and simultaneously doubliing MaxBuffer up to 32768 made Kerio happy! :=) I have yet to try the mods separately, to acertain which the crucial one is (unless both are). Just guessing and ICBW, from the name of the value, "KernelModuleAuth" seems to be the more effective cause for the observed behavior. [Edited :] Yep, the KernelModuleAuth is the one that matters ! [Edit end] Do you know exactly what this is supposed to do, and what I lose without it ? I googled a bit to no avail... Cheers...

Hi! > The key is HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\fwdrv > Interesting values are: KernelModuleAuth.... It is well configured, since it runs OK outside a VM; I run the VM out of the same physical, raw disk, hence the Kerio in particular is the same. I don't think tweaking that registry key should have an influence; which value are you suggesting to be adjusted anyway, and why ? Furthermore, the crash occurs on launching the kernel driver, not the service executable. It is not even sure if fwdrv.sys examines the registry at all, or crashes before it has a chance (I could try to check using sysinternals Registry monitor, ISTR it has a way to be launched very early in the boot sequence, however... not sure it can be active before a /kernel driver/ such as fwdrv.sys is activated) Thank you, GrofLuigi ! More ideas ? Others ?

I'm running Windows 2000 SP4 inside of a VMware (Player 1) virtual machine, hosted on SUSE Linux. Had to disable the Kerio firewall 2.1.5, because else the virtual machine repeatably blue-screens at loading fwdrv.sys (the infamous STOP 0xD1, driver_irql_not_less_or_equal). When running the same config natively instead of in a VM, Kerio loads and works flawlessly. Kerio also loads and works flawlessly in a VM, but in Windows 98 instead of win 2k (this uses a VxD rather than the NT style .sys driver). All this would point to some incompatibility between Kerio's driver and VMware Player, and I would accept it as one of those sad facts of life, except - someone I don't distrust told me, on another forum, that he is able to run KPF 2.15 in a Windows 2k VM in Linux, without the driver startup crashing the VM. This if confirmed could justify some more debugging : therefore I'm asking for more evidence of (in)compatibility between VMware Player on the one hand, Kerio's fwdrv.sys OTOH. If it works for some, the reason it doesn't for me might be some subtle interference with other drivers, driver load order, the amount of memory or other resources, the fact that my VM runs from a "raw" disk rather than virtual... who knows. Anyway, I could as well submit here, in case someone had a bright idea or two... Cheers -- Ninho

Yes, I've already mentioned (and provided links) for both "NTFS for Windows 98" and "NTFSDOS" (for DOS). However, again, neither of these program solutions will provide what I'm looking for. ...Even the commercial products do not provide full integration. Sorry, I am listening, you are not ! NTFS for Windows 9x, by Winternals (Mark Russinovich, now with Microsoft) - was a full IFS, read/write. (The free demo was crippled to read-only). DOSNTFS, from the same author, is a different hack by the way : a rescue system, not a Windows IFS. I am well aware these are no more supported, but they are available from the web - the free versions, and (if you want to check my assertions, for educational purposes only) the for-pay versions can be found also. I did test years ago, and deleted the 'warez' after testing. But I kept NTFSDOS (the free trial, hacked for read-write by myself for my own use) on diskettes as a souvenir. We have Bart PE nowadays, and NTFS access from Linux... Anyway, my point is full NTFS support in windows 95 is feasible, as demonstrated by the ex- Winternals. If you want to redo it from scratch, great ! My other point, which you may disagree, is that apart for the beauty of the exercise, that effort would be rather moot in the 21st century. If you multi-boot, access your NTFS partition(s) from the OSes that can do so natively, and/or from Linux ; if you have only Win 9x installed on that machine, why would you want an NTFS to begin with ? Cheers, -- Ninho

Yes of course, I've used such apps, but they're not Windows installable file systems. Apples and oranges. Precisely. Sometimes useful, but don't expect full integration. -- Ninho

Well, installable file systems have existed since Windows for Workgroups 3.11 ! The possibility exists to implement protected mode support for virtually any file system type, including of course the NTFS (without advanced functions, security etc... which are not covered by the Windows 9x API). Small problem : since MS did not want to do it for obvious reasons (differentiate NT "professional grade" OS from home OS), you have to doo it yourself (and with very little documention at that). Third party support for NTFS in Windows 9x /exists/, it is (was) a commercial product from Wininternals/Sysinternals. This said, like the other responder said, if FAT32 does not satisfy your needs, why would you want a proprietary file system like NTFS ? Much more interesting and useful as an IFS would be Ext2FS (Ext3FS without the journaling). But don't hold your breath, I don't think anybody's going to implement, code and test a full IFS on Win 9x in this century. Cheers -- Ninho

@dencorso : "Windows remaps all XMS memory already alocated before starting Windows to addresses in the fourth MB (downwards from 4 to 3 MB, starting at the highest free address) just as if that memory was owned by some hardware device, " "And then, there still remain the AGP aperture and VCache to consider, but such memory is mapped in the 4th megabyte, anyway..." I guess you mean GiB, not Megs... Surely a typo, though an unfortunate one since the subj is already clear as mud... Also for things to be as clear as possible, I guess your are talking in terms of virtual, not physical addresses. This may be self evident, but I'm sure more than one has been confused over such subtleties. With this in mind, let's see if I understood well : dare I ask, what you are saying amounts to Windows/VMM at startup remapping to system memory space (the 3rd virtual gigabyte) any XMS memory blocks that were allocated (in DOS presumably) before Win started ? Makes sense ! -- Ninho

Seems a bit strange to say the least. Not discussing your choices, you are willing to limit the mem used by Windows to slightly under 1 gigabyte - if my quick math is correct. The MaxFileCache value IIRC is decimal kilobytes, and 30000 seems acceptable if a tad small for your situation. As a consequence, it /should/ work... Please make absolutely certain you entered the required parameter lines in the respective right sections of your %windir%\system.ini file. The MaxPhysPage= should be under [386enh], while MaxFileCache= goes under [Vcache]. I know you must have been careful, but please... double-check before we try to reproduce or understand your problem ! Cheers -- Ninho

A very bad solution IMNSHO. You are in effect removing part of your memory from the system. The problem you experienced is very well known, and it's surprising someone who is never shy on giving advice was not aware of the right way of correcting it. You should ditch the MaxPhysPage altogether, and instead add (or modify, if it exists) an appropriate [VCache] section to your System.ini : _________________________ [VCache] MaxFileCache=65536 _________________________ HTH. And please don't bother to thank me. -- Ninho Edited, to explain : the actual figure, '65536' is not critical. It is what I use on a system similar to Benoit's - same MoBo & installed memory. I could as well, say, double it. In Windows 98 as opposed to Win 95, programs can be run directly from the VCache if they happen to be there already, so having a larger VCache is more appealing. The critical point is to set a definite limit, and one which is /not too high/, because of a bug or oversight on the part of Microsoft.

It is all too true that Firefox's rendering of HTTP is very slow, maybe five to ten times slower than IE's. It's only because our PCs are so fast that we usually do not notice it. When I was forced to use a 486 DX2 for some time, a few months ago, I couldn't but notice it : Firefox was functional in theory, yet in practice unusable. My guess is a large part of FF is programmed in interpreted, or semi-interpreted, languages, and not optimised at that. Someone who looked at the code base might confirm (or not). Anyway I'm not bashing FF or Moz, I love Firefox, but the truth is it is not optimised or even designed with speed in mind. Cheers -- Ninho

Hello, 98 ! The VMM and VxDs comprising the Operating System proper underneath Windows 95/98/ME are 32-bit code. These provide memory and device management, ressource allocation etc. As for the main Windows modules (Windows is a presentation layer, not an OS!): The Windows 95/98 kernel is 32-bit. Of course there is a 16-bit kernel module for use by 16-bit code, but this one calls ('thunks') into the 32-bit Kernel32. The GDI is 16-bit. The above for compatibility applies, only in the reverse direction ;=) Finally the User module is mixed, about half 16 and half 32 bit code with thunking in both directions. This is well known and easily verifiable by anyone armed with a (kernel) debugger and some programming skills. HTH -- Ninho

Hmm, not a sign of interest here! Nobody cares, or - I guess - those who know don't care & those who care don't know ;=) Well, I found my answer, thanks to gentleman Laszlo at Sandboxie's forum. Caring ones can look it up there : <http://sandboxie.com/phpbb/viewtopic.php?t=2094> -- Ninho

One more, big big thank you, Maximus Decim ! I've just installed NUSB version 3.1 (Fr) on Win 98 SE. No frills, just works out of the box, like magic ! I'm the most amazed, as my MoBo has the SiS USB chipset for which previously /nothing/ could make *USB2* work in windows 98, from the manufacturer or otherwise, and not even a previous version 2.x of NUSB. This is some miracle ! I would advise any reader with SiS chipsets to try it immediately. With regards, -- Ninho

Hey TravisO, why dare you write : MakeMeAdmin (cf. Aaron Margosis's at MSDN blogs) and similar work fully in Win2k, indeed they were developped on it. I run as a restricted user all the time & use MakeMEAdmin and never have to logon as Admin in 2k, not even for running and installing Windows Updates. Please do not sow FUD. Cheers -- Ninho

Hi! Having different hardware profiles. How do you manage to tell Windows (2000) to load/not load a specific kernel driver according to the selected HW profile ? You can define which services to load selectively using the 'services.msc' console, what I need to achieve is similar but for .sys drivers... Thanks ! -- Ninho

With the MAxPhysPage value you are advising he will be using only 1 Gigabyte however; 2 Gigabytes would be : MaxPhysPage=80000 I'd advise to leave the MAxPhysPage line out altogether to begin with, and try playing with the other settings. HTH -- Ninho