[PROBLEMCHYLD]

rloew, on 04 December 2011 - 10:38 PM, said:

@PROBLEMCHYLD: If you use this new Version,. please change your comment on IO.SYS to reflect the change so there is no confusion.

Thanks for the fix. It has been added and SP3 has been updated.

This post has been edited by PROBLEMCHYLD: 05 December 2011 - 05:28 AM

[submix8c]

dencorso, on 04 December 2011 - 08:14 PM, said:

You can use MS CRC.EXE and MS FCIV.EXE to confirm the CRC32 and MD5 hashes, but you must extract the latter from inside KB841290, since the installer refuses to run on Win 9x, but FCIV.EXE itself runs OK on Win 9x.

If you prepare this on an NT-type system (2K and up), HashMyFiles is excellent (stumbled across on MSFN - worth re-mentioning...).

Thx. Will try when time permits. rloew ROX!!!

[Prozactive]

I patched IO.SYS per dencorso's instructions then updated my system from DOS, but upon reboot I got a persistent Windows registry error message:

Windows encountered an error accessing the system registry.
Windows will restart and repair the registry for you.

The system would reboot, run a DOS registry checker app and fix some registry error, then boot up to the desktop where the same registry error would occur again. I finally gave up and restored the original IO.SYS which resolved the problem. I suspect I need to run the Microsoft 311561USA8.EXE update first before updating to the latest version of IO.SYS as I still have the original 23 April 1999 version on my system. I'll report back if that solves (or does not solve) the problem.

UPDATE: Installing the 311561USA8.EXE update seems to have resolved the problem.

This post has been edited by Prozactive: 06 December 2011 - 02:35 PM

[rloew]

Prozactive, on 06 December 2011 - 01:50 PM, said:

I patched IO.SYS per dencorso's instructions then updated my system from DOS, but upon reboot I got a persistent Windows registry error message:

Windows encountered an error accessing the system registry.
Windows will restart and repair the registry for you.

The system would reboot, run a DOS registry checker app and fix some registry error, then boot up to the desktop where the same registry error would occur again. I finally gave up and restored the original IO.SYS which resolved the problem. I suspect I need to run the Microsoft 311561USA8.EXE update first before updating to the latest version of IO.SYS as I still have the original 23 April 1999 version on my system. I'll report back if that solves (or does not solve) the problem.

UPDATE: Installing the 311561USA8.EXE update seems to have resolved the problem.

The KB311561 Update was specified in my Instructions. The original would not have matched the "old" data shown in the second block. A slightly different Patch is needed for the original Version of IO.SYS.

[dencorso]

I'm glad you've swiftly solved the issue you've found.
However, allow me to nitpick:

dencorso, on 05 December 2011 - 12:07 AM, said:

Here is RLoew's complete patch, as a patch pattern, so that those interested may create the patched IO.SYS.
Thanks are due to RLoew for creating it and for sharing it with us all. You do rock, RLoew!
In order to use the enclosed patch pattern you need to grab [...] the file winboot.98s from inside the Q311561USA8.EXE hotfix, [...]

Prozactive, on 06 December 2011 - 01:50 PM, said:

I patched IO.SYS per dencorso's instructions then updated my system from DOS, but upon reboot I got a persistent Windows registry error message:[...]

Well, no. You actually didn't follow my instructions, at first, despite your claims...
Then, when you finally did follow my instructions (not exactly, but sort of):

Prozactive, on 06 December 2011 - 01:50 PM, said:

UPDATE: Installing the 311561USA8.EXE update seems to have resolved the problem.

Surprise: It worked!

Hence, when one follows instructions, one ought to follow them in full, not to decide what to leave behind...

Patch patterns are matched to the file (the target) they intend to patch. When one gives them a different target they may patch it nonetheless, however the result will be a suffusion of yellow, to quote jaclaz.

[Prozactive]

No, you misunderstood or I didn't explain it clearly. I did indeed patch the winboot.98s file extracted from the 311561USA8.EXE update, then I replaced my original IO.SYS from DOS, renaming the older file with another extension.

Ah, I see. It was misleading of me to say I "patched" IO.SYS. I should've said I updated IO.SYS by patching winboot.98s per your instructions.

Hopefully this cleared things up.

This post has been edited by Prozactive: 06 December 2011 - 09:20 PM

[dencorso]

Prozactive, on 06 December 2011 - 09:10 PM, said:

No, you misunderstood or I didn't explain it clearly. I did indeed patch the winboot.98s file extracted from the 311561USA8.EXE update, then I replaced my original IO.SYS from DOS, renaming the older file with another extension.

Ah, I see. It was misleading of me to say I "patched" IO.SYS. I should've said I updated IO.SYS by patching winboot.98s per your instructions.

Hopefully this cleared things up.

If so, my apologies.
Then again, applying 311561USA8.EXE previously to replacing the extant IO.SYS by the patched one should make no difference at all. Yet it did...

Then again, many things that make perfect sense just don't work... and, conversely, there's also a lot of things that make no sense at all, and yet work, including the Chewbacca Defense.

[Prozactive]

No problem. After reading your reply (and rloew's, which I missed before my initial response), I could finally see how my statements were misleading. Yes, I thought I could go straight from the original 23 Apr 1999 version of IO.SYS to the latest patched version from rloew but that's apparently not the case. It's a prerequisite to install the 311561USA8.EXE update first to update IO.SYS to that version before the new patched version is manually installed. I assumed that the 311561USA8.EXE update made some required registry changes during installation. Also, the file size of IO.SYS is different for the two versions (original vs. 311561USA8.EXE updated).

This post has been edited by Prozactive: 06 December 2011 - 10:00 PM

[jaclaz]

dencorso, on 06 December 2011 - 07:50 PM, said:

When one gives them a different target they may patch it nonetheless, however the result will be a suffusion of yellow, to quote jaclaz.

Who was actually quoting Douglas Adams
Just in case of need :
http://www.thateden.co.uk/dirk/

jaclaz

[dencorso]

jaclaz, on 07 December 2011 - 04:54 AM, said:

dencorso, on 06 December 2011 - 07:50 PM, said:

When one gives them a different target they may patch it nonetheless, however the result will be a suffusion of yellow, to quote jaclaz.

Who was actually quoting Douglas Adams
Just in case of need :
http://www.thateden.co.uk/dirk/

Yeah, I'm aware it actually is a matryoshka of quotations!

[egrabrych]

Prozactive, on 06 December 2011 - 09:55 PM, said:

It's a prerequisite to install the 311561USA8.EXE update first to update IO.SYS to that version before the new patched version is manually installed.

Then I would like to add that it does not need to be the 311561USA8.EXE hotfix, because RLoew's patch also introduces the CORRECT changes to the Io.sys file installed by the 311561POL8.EXE hotfix and probably also by all the other updates for Windows 98SE still available for download at: http://support.microsoft.com/kb/311561
In general, I think the Microsoft wouldn't have released 27 different language versions of this update without a reason (even more so, because, for example, Polish version doesn't have messages translated to Polish language at all).

So I recommend preinstalling the 311561 hotfix in the correct language version, and then using the RLoew's patch on the Io.sys file installed this way.

One more thing: at least I have exclusively positive experience with overwriting the Io.sys file (ever since MS-DOS 3.3 till now; naturally on the FAT/FAT32 partition), PROVIDED the volumes of these two files are EXACTLY THE SAME: new and old (the one being replaced). So the Io.sys file copied this way by me was always placed in the right place on a disk, the place occupied by the previous Io.sys file.

update:
Well, I wrote these words, and only THEN I decided to see what REALLY differences exist between the various language versions of the amendment 311561 hotfix. Well, it turned out that virtually .... NO. Winboot.98s files (and Winboot.98g) are IDENTICAL, entries in the Windows Registry also practically the same, only the files involved in the installation are in separate languages.
But that does not interfere with using proper language patch for installed in Windows

This post has been edited by egrabrych: 07 December 2011 - 02:30 PM

[Prozactive]

@egrabrych:
Yes of course, one should install the correct language version of the KB311561 update. My apologies for being US-centric.

I don't understand your last paragraph at all, however. Do you mean that the old and updated versions of IO.SYS has to physically occupy the exact same location on the HDD? I don't think that's the case AFAIK, as I'm pretty sure my updated IO.SYS did not exactly replace the older file's location. I must be misunderstanding you.

[jaclaz]

Prozactive, on 07 December 2011 - 01:27 PM, said:

Do you mean that the old and updated versions of IO.SYS has to physically occupy the exact same location on the HDD? I don't think that's the case AFAIK, as I'm pretty sure my updated IO.SYS did not exactly replace the older file's location. I must be misunderstanding you.

Naah, it is an OLD requisite, up to DOS 6.22 IO.SYS had to be FIRST file in FAT.

Compare with the bootpart.txt attached to bootpart:
http://www.winimage.com/bootpart.htm

This requisite is not needed anymore with DOS 7.x/8.0.

Cannot say if - in "peculiar" cases - the IO.SYS is put beyond some "limit" (say beyond 512 Mb or 8 Gb or 128 Gb ) something bad can happen .

jaclaz

[rloew]

egrabrych, on 07 December 2011 - 12:07 PM, said:

Prozactive, on 06 December 2011 - 09:55 PM, said:

It's a prerequisite to install the 311561USA8.EXE update first to update IO.SYS to that version before the new patched version is manually installed.

Then I would like to add that it does not need to be the 311561USA8.EXE hotfix, because RLoew's patch also introduces the CORRECT changes to the Io.sys file installed by the 311561POL8.EXE hotfix and probably also by all the other updates for Windows 98SE still available for download at: http://support.microsoft.com/kb/311561
In general, I think the Microsoft wouldn't have released 27 different language versions of this update without a reason (even more so, because, for example, Polish version doesn't have messages translated to Polish language at all).
So I recommend preinstalling the 311561 hotfix in the correct language version, and then using the RLoew's patch on the Io.sys file installed this way.

The Pad area I used is just above a text block. If the text for a particular language is longer than the English Version than the Patch will fail. Some languages will work but others need a different Patch. When Patching using my Patch, verify that the OLD data shown matches what is in the File. Dencorso's Version of my Patch is locked to the USA Version or exact equivalents.

Quote

One more thing: at least I have exclusively positive experience with overwriting the Io.sys file (ever since MS-DOS 3.3 till now; naturally on the FAT/FAT32 partition), PROVIDED the volumes of these two files are EXACTLY THE SAME: new and old (the one being replaced). So the Io.sys file copied this way by me was always placed in the right place on a disk, the place occupied by the previous Io.sys file.

DOS 7 and it's associated Boot Sectors only require that the first four Sectors of IO.SYS be contiguous. It is no longer necessary that IO.SYS start on Cluster 2 or that it be the first entry in the Root Directory.

[egrabrych]

Prozactive, on 07 December 2011 - 01:27 PM, said:

@egrabrych:
Yes of course, one should install the correct language version of the KB311561 update. My apologies for being US-centric.

Hmm... And I just noticed that the Winboot.98s files are in all languages IDENTICAL

egrabrych said:

Well, I wrote these words, and only THEN I decided to see what REALLY differences exist between the various language versions of the amendment 311561 hotfix. Well, it turned out that virtually .... NO. Winboot.98s files (and Winboot.98g) are IDENTICAL, entries in the Windows Registry also practically the same, only the files involved in the installation are in separate languages.
But that does not interfere with using proper language patch for installed in Windows

Prozactive, on 07 December 2011 - 01:27 PM, said:

I don't understand your last paragraph at all, however. Do you mean that the old and updated versions of IO.SYS has to physically occupy the exact same location on the HDD? I don't think that's the case AFAIK, as I'm pretty sure my updated IO.SYS did not exactly replace the older file's location. I must be misunderstanding you.

It seems that I am a "dinosaur".
From Wikipedia:
IO.SYS must be the first file stored in the data storage area for files (...) must be contiguous.
- So it was in earlier versions of DOS, but in DOS 7.0 and 7.10 probably does not need to be. So please forgive me ... though what I wrote was (formerly) true.

[Prozactive]

Thanks jaclaz, rloew, and egrabrych. Just for technical education, how would you accomplish that (copying IO.SYS to a specific cluster(s)/location on the HDD)? And now that you mention it, I do recall reading that Wikipedia article a long time ago.

This post has been edited by Prozactive: 07 December 2011 - 03:57 PM

[dencorso]

egrabrych, on 07 December 2011 - 03:31 PM, said:

Hmm... And I just noticed that the Winboot.98s files are in all languages IDENTICAL

egrabrych said:

Well, I wrote these words, and only THEN I decided to see what REALLY differences exist between the various language versions of the amendment 311561 hotfix. Well, it turned out that virtually .... NO. Winboot.98s files (and Winboot.98g) are IDENTICAL, entries in the Windows Registry also practically the same, only the files involved in the installation are in separate languages.
But that does not interfere with using proper language patch for installed in Windows

Sure. Did I ever rant about nobody ever searching the forum anymore yet?
Localized Q311561... It's too bad, however that

http://old.winpack.org/petr/98/311561/*

doesn't exist anymore, and it's a part of the contents of "old.winpack.org" that wasn't preserved by the Wayback Machine.
However, I'm sure some of our members must have saved copies of the POL8 version, and it's possible MDGx may have saved a copy of them all. Then again, in order to be able to use Petr's localized version with RLoew's complete patch, a special version of it would be required, and, since most users never noticed Q311561 IO.SYS isn't really localized, I'd say it's not worth the effort.
This is just my 2 ¢, of course.

Prozactive, on 07 December 2011 - 03:55 PM, said:

how would you accomplish that (copying IO.SYS to a specific cluster(s)/location on the HDD)?

Given a partition that has just been formatted, if you just copy IO.SYS alone to it, using DOS, it'll be copied as a single continuous image, starting at cluster 2 (the first cluster). Now, to put in a continuous image of any file, starting at an arbitrary cluster number, especially when the partition in not empty anymore, that requires a lot more calisthenics and patience, but can be acomplished by hand, with a good hexeditor/sector-copier, by anyone who has a really strong reason to do it (like learning how-to, for instance).

[rloew]

I have written a Patching Program that will add the Partition Offset Bug Fix to IO.SYS. This program will work with the original IO.SYS as well as the recommended Updated IO.SYS Version.
This version should work with any of the Localized Versions as well.

Use the following commands to perform the Patch.

C:
CD \
REN IO.SYS IO.OLD
PATCHPAR IO.OLD IO.SYS

This post has been edited by rloew: 07 December 2011 - 11:43 PM

[jds]

Prozactive, on 06 December 2011 - 09:55 PM, said:

Yes, I thought I could go straight from the original 23 Apr 1999 version of IO.SYS to the latest patched version from rloew but that's apparently not the case. It's a prerequisite to install the 311561USA8.EXE update first to update IO.SYS to that version before the new patched version is manually installed.

This seems strange. I'm pretty sure I've done this (with my version of phelum's patches) without any dramas. Did you immediately reboot after changing 'IO.SYS' (just in case some copy of 'IO.SYS' in memory gets reloaded and then code addresses have changed, or some such confusion)?

dencorso, on 07 December 2011 - 08:10 PM, said:

Prozactive, on 07 December 2011 - 03:55 PM, said:

how would you accomplish that (copying IO.SYS to a specific cluster(s)/location on the HDD)?

Given a partition that has just been formatted, if you just copy IO.SYS alone to it, using DOS, it'll be copied as a single continuous image, starting at cluster 2 (the first cluster). Now, to put in a continuous image of any file, starting at an arbitrary cluster number, especially when the partition in not empty anymore, that requires a lot more calisthenics and patience, but can be acomplished by hand, with a good hexeditor/sector-copier, by anyone who has a really strong reason to do it (like learning how-to, for instance).

Well, according to Mr Loew, only the first 2K of 'IO.SYS' needs to be contiguous, so the problem shouldn't normally arise for FAT16 partitions of 64M (or greater) or FAT32 partitions of 128M (or greater), see : http://support.microsoft.com/kb/140365

Joe.

PS. Earlier O/S seemed to start using 2K cluster size with smaller partitions : http://support.microsoft.com/kb/67321

This post has been edited by jds: 08 December 2011 - 01:15 AM

[dencorso]

jds, on 08 December 2011 - 12:47 AM, said:

Prozactive, on 06 December 2011 - 09:55 PM, said:

Yes, I thought I could go straight from the original 23 Apr 1999 version of IO.SYS to the latest patched version from rloew but that's apparently not the case. It's a prerequisite to install the 311561USA8.EXE update first to update IO.SYS to that version before the new patched version is manually installed.

This seems strange. I'm pretty sure I've done this (with my version of phelum's patches) without any dramas. Did you immediately reboot after changing 'IO.SYS' (just in case some copy of 'IO.SYS' in memory gets reloaded and then code addresses have changed, or some such confusion)?

I'm positive I did it numerous times. I usually populate my unusual disk images by hand, not using sys.com, because it *requires* physical media to work, since it relies on DOS Absolute Disk Read/Write (INT 25H/26H) services to create/modify the boot sector. In fact, it'll fail silently on any image mounted by Grub4DOS, under DOS, or mounted by VDM/VDK, under the NT-OSes. Hence I use a hexeditor to transfer and edit the boot sector (nowadays using jaclaz's excel worksheet to actually create the boot sector), then copy first IO.SYS, then COMMAND.COM, then whatever else I want in the image. It just works. And that's why I said it makes no sense. I do not doubt Prozactive had a problem and solved it by applying KB311561, but I do doubt that the applying of that hotfix was really the factor leading to the actual solution of the issue.