Version 1.1 is now available.

2006-07-13 15:19 Finished driver analysis, initial announcement
2006-07-14 10:43 Working on inserting new code
2006-07-15 09:56 First successful build
2006-07-15 22:03 Major change of plan
2006-07-20 18:10 Version 1.0 released
2006-07-21 04:51 Begin work on WinME version
2006-07-21 05:50 Finish WinME version
2006-07-21 06:22 Begin work on 4.10.2225
2006-07-21 06:37 Version 1.1 (4.10.2225) released
2006-07-22 07:34 Win98FE version (4.10.2001) released
2006-07-23 08:51 Win98FE version (4.10.2186) released
2006-07-29 20:05 Win98SE version (4.10.2226) released - only for replacing an existing v4.10.2226 file
2006-08-02 17:57 Win95 OSR2+ version (4.00.1111) released
2006-08-04 20:27 Win95 OSR2+ version (4.00.1119) released

For those that don't know, this is an attempt to implement 48-bit LBA support into Windows 98se's default IDE driver, enabling the use of hard drives in excess of 128Gb. This is not just another variant of Loew's "High Capacity Disk Patch" because

1] It's completely free (and of course, has no guarantee)
2] Addressing to 2048Gb is possible (limit of FAT32)
3] A large portion of driver code has been rewritten and optimised
4] LBA-48 commands implemented in separate routines instead of "patched into" existing 28-bit's code
5] Slightly increased performance over Loew's patch due to [4]

If Enable48BitLBA has worked for you, please post a link to a validation image containing the following:

1. Properties page of a drive filled past 137Gb
2. Information dialog, the results of a successful scandisk on the drive
3. System Properties -> Device Manager -> (your IDE controller) Properties -> Driver -> Driver File Details (must show ESDI_506.PDR being used)
Also specify your hardware (in particular, mobo chipset and HDD specs).

This will help to confirm that it works.

Reference: http://www.msfn.org/board/137GB-limit-ESDI-506PD-t46752.html

I am not responsible for any damage caused by the use of these drivers.

excellent

Congratulations.
Sounds like a major breakthrough for Win98 OSes.

Can't wait for 48-bit LBA on 98Se!

Having a little bit of a technical difficulty... the driver won't assemble! I'm getting "Out of Memory" errors from ML.

Might have to move the release date forward a bit...

take your time LLXX. remember that you have to create a 48-bit LBA driver NOT just for Win98se, but also for Win98 FE and WinME. you cant just create a driver just for Win98se and ignore the other versions of Windows.

I've solved the memory problem and it's assembling fine now...

Now I'll just need to figure out what link options were used so I can the exact binary as M$ original one (with unmodified source), then I'll start inserting the new code

@erpdude8 : Are there vast differences between the three versions you've mentioned? I'm only working on the one from 98se (file version 4.10.2222 - "ESDI_506 Virtual Device (Version 4.0)")

Keep in mind that the current esdi_506.pdr for 98SE is version 4.10.0.2225. When using Norton File Compare there is 7747 differences against version 4.10.0.2222.

22191 differences between 4.10.0.2222 and 4.90.0.3000 (Win ME)
22357 differences between 4.10.0.2225 and 4.90.0.3000 (Win ME)

See below for the Windows 98 Large IDE Update which contains the current files for both 98 SE (4.10.0.2225) and FE (4.10.0.2186).


http://support.microsoft.com/kb/q243450/
http://www.microsoft.com/windows98/downloa...ide/Default.asp

My esdi_506.pdr is 4.10.2225, also my Kernel32.dll is 4.10.2225. Running Win98SE of course. LLXX good luck with the build.

"vulnerable"... lol. All drives must support LBA according to the ATAPI-4 Standard, so this is a moot point. The driver automatically selects LBA mode when available.
If a device conforms to an earlier Standard (in which case LBA support is Optional), its capacity is unlikely to exceed 32Gb in any case. I doubt there were 32Gb IDE devices being produced in 1996.

Because of this, I doubt there will be any problems encountered in using the 4.10.2222 version of ESDI_506.PDR as a base. In any case, applying the patch to a newer version should prove just as easy.

I've chosen to abandon reassembling the driver completely, as it seems to have been built with a strange Linker that is apparently not available outside of Micro$oft. Instead, it will be more similar to a complex version of the Loew patch.

Let's not forget, there was also a version 4.10.2226 of esdi_506.pdr for Win98se [Q252187]. this hotfix was built for certain IBM laptops only that came bundled with pre-installed 98 SE. this version may cause startup problems (like hanging at startup before Win98se loads) on non-IBM machines.

Does anyone know what 4.10.2223 and 4.10.2224 were for?

Anyway, I'm planning to release a fixed 4.10.2222 and do other versions when necessary.

Good point to first release a standard patch for Windows 98SE. The best would be to test on a clean Windows 98SE machine with harddiscs larger than 137 GB.

However we must not forget to test also HDD's smaller than 137 GB!

Especially HDD's <8GB and <32GB HDD's should get tested, since these use different specifications.

Esdi_506.pdr version 4.10.2223 is mentioned in MS article 239694:
http://support.microsoft.com/kb/239694/

not sure about v4.10.2224 of esdi_506.pdr



good points hp38guser made of testing the patch on HDDs of all sizes.

If I was following the original plan of fully rebuilding the driver, that would definitely be done before release. However, as the new plan of action I'm taking is not going to affect the existing code at all except for one change, I'm quite certain it will work fine at smaller sizes. Testing the new code, however, is a totally different problem.

Here's the only change to the existing code that I'll make:

Windows 98 SE ESDI_506.PDR versions:

* Microsoft Windows 98 SE Computer Hangs with Removable Device in Docking Station ESDI_506.PDR 4.10.2226 Fix:
http://support.microsoft.com/?id=252187
Direct download [154 KB, English]:
http://www.mdgx.com/files/Q252187.EXE
Install this ESDI_506.PDR Fix ONLY on IBM portables with removable disks!
Install ESDI_506.PDR 4.10.2225 Fix below on ALL other PCs/portables!

* Microsoft Windows 98/98 SE ScanDisk Errors With (E)IDE Hard Disks > 32 GB ESDI_506.PDR (4.10.2186 for Win98/98 SP1 + 4.10.2225 for Win98 SE) Fix:
http://support.microsoft.com/?id=243450
Direct download [158 KB, English]:
http://download.microsoft.com/download/win.../243450USA8.EXE
Install this ESDI_506.PDR Fix on ALL PCs/portables EXCEPT IBM portables with removable disks!
Install ESDI_506.PDR 4.10.2226 Fix above ONLY on IBM portables with removable disks!

Windows 98 FE (First Edition) ESDI_506.PDR version:

* Microsoft Windows 98 IBM ThinkPad Laptops ACPI Standby Error With IDE DMA Hard Disk ESDI_506.PDR 4.10.2001 + IOS.VXD 4.10.2001 Fix:
http://support.microsoft.com/?id=193473
Direct download [167 KB, English]:
http://www.mdgx.com/files/Q193473.EXE

HTH

LLXX:

Please tell Gape to make this thread a sticky after you post the driver, and also to merge the old one with this one [if possible], and then close your old one:
http://www.msfn.org/board/?showtopic=46752

Thanks.

Version 1.0 is now available.

I repeat once again, I am NOT responsible for any loss of data caused by this driver.

We do not have any hard drives supporting 48-bit LBA, so I am unable to test the extended commands.

There should be no problems with drives less than 128Gb. It's just the new code that needs testing.

I have two HDD's in my system. One 40GB and a 250GB Maxtor Diamond Max 10 ATA HDD. It's special for it's ATAPI-6 (133 MB/s) Interface. most HDD's use ATAPi-5 (100 MB/s).

I'll try to backup all stuff necessary to the 40GB HDD and DVD's. Than I'll be able to do some Windows 98 tests with the 250 GB HDD.

However I'm not sure about my mainboard. It's a VIA KT333 mainboard. Does this motherboard automaticly enable 48-bit lba support or only after I install the VIA ide drivers in Windows 98?

The point is to use my fixed driver, not VIA's own drivers. Nothing needs to be "enabled". The driver will automatically switch to using extended commands when reading past the limit (see doc for more info).

Congratulations + thanks for the driver.
Mr. R. Loew is not going to be happy his $10 "shareware" driver now has better freeware competition.

I can't test it either, all my drives are < or = to 120 GB.

Question:
Which ESDI_506.PDR version did you use, 4.10.2225 or 4.10.2226 ?
Best to use 4.10.2225 because this is the most compatible and up to date.

Suggestion:
No matter which version you used, file version should be changed to reflect the update and to install properly over any other older version that might exist in %windir%\system\iosubsys .
Your new driver should have 4.10.2227 [1 notch above the newest one by M$].

FYI:
I have created an iexpress installer with backup [original file (if any) renamed to ESDI_506.ORI], uninstall + restore [original file (if any) restored from ESDI_506.ORI back to ESDI_506.PDR] Control Panel Add/Remove Programs capabilities.
I took the liberty and changed file version to 4.10.2227 using XVI32 [freeware hex editor]:
http://www.chmaas.handshake.de/delphi/free...xvi32/xvi32.htm
I also took the liberty and added the links to your 2 forums into the README.TXT:
README.TXT opens for reading whenever one installs the driver.

Posted here:
http://www.mdgx.com/web.htm#9SU
and here:
http://www.msfn.org/board/?showtopic=46581

* Unofficial Windows 98 SE > 137 GB Hard Disk IDE Driver ESDI_506.PDR 4.10.2227 Fix:
http://www.msfn.org/board/?showtopic=78592
Direct download [80 KB, English]:
http://www.mdgx.com/files/48BITLBA.EXE

Keep up the good work.

Could you please make one for Windows Millenium ? Please please... ? , I have a 160 GB hard drive but don't have 98SE only Millenium...

Here's my ESDI_506.PDR (attached):

As stated above, this is based on 4.10.2222. If you think it's necessary, I'll patch the newer versions as well.
I don't know if you've noticed, but I already changed the version information slightly - it now reads "ESDI Virtual Device (Version 4.0b)" instead of just 4.0. However, according to your suggested versioning, I think the following may be appropriate:

4.10.2222 -> 4.10.2227
4.10.2223 -> 4.10.2228
4.10.2224 -> 4.10.2229
4.10.2225 -> 4.10.2230
4.10.2226 -> 4.10.2231
Thanks, I'll do the patching and leave the packaging up to you

MDGX, for your information original ESDI_506.PDR 4.10.2222 was used. So numbering this LLXX patch 4.10.2227 would people think the ESDi_506.PDR 4.10.2225 and 4.10.2226 hotfixes are included.

However I have an idea how to create custom version numbering for special patches. Why just not increase 2nd digit?

With this method you also overwrite newer hotfix files. However in this case I think it's necessary, since it solves a very bad problem: file corruption.

For example LLXX patch for esdi_506.pdr 4.10.2222 would become 4.10.2232 and LLXX patch for esdi_506.pdr 4.10.2225 would become 4.10.2235. This clearly shows a difference between the official Microsoft hotfixes and LLXX patch.

Microsoft won't bring out any new patches anyway

Windows ME was going to be next. It's coming

so whats this do for my 40G drive?

Will it speed things up?

ATA133

Thanks for the info, guys.

I did notice the 4.0b change, but that's only for internal use, the actual iexpress INF update won't take in account the product version, only the file version.
Edit:
The product version should not be changed, 4.0 means the Win95/98/ME/NT4 line of OSes.
There is no such thing as 4.0b WinOS.
2000 is product line 5.0, XP/2003 is 5.1 and Vista will probably be 5.2 or 6.0 [?].

And the file version must always be newer [larger number] than the previous release [M$ came up with that rule], so the INF installer can overwrite the older file with the newer one.

IMHO, I strongly recommend to use 4.10.2225 [the most compatible so far, and already has all up-to-date patches] to create the 48-bit LBA driver.
The original 4.10.2222 has bugs, which were already fixed by M$ in 4.10.2223, 4.10.2225 + 4.10.2226 [you can leave the last one aside, because that fix applies strictly to IBM portables].
If you patched the buggy 4.10.2222, the fixes implemented by M$ in 2223 + 2225 would be lost, and that's not a good thing.
And, besides, proper patching should be cumulative.

So, LLXX, if possible, at your convenience, let me know if you can patch 4.10.2225.
And you can name it whatever you wish, as long as it's above 2226.
Thanks.

excellent work llxx and mdgx it just goes to show there no end to the 98se still alive drive.

4.90.3000 is now fixed and ready for download. That didn't take long
Will do. You can manage the versions when you package it, though I recommend 4.10.2230.

Version 1.1 is now available.

Done

Thank you so much, going to test now !!!

LLXX, You've amazed me - not only did you have one version done, but managed to get three different releases working in a short time. Good work. Now if only I had a way to test it, but alas, all my systems run XP or have small hdd's (I have a 98SE box with a bunch of 4-18GB SCSI drives, but not anything over that size).

Doing the first one was the difficult part, since I had to analyze the existing code and figure out how to integrate the new code. After that, it was mostly copy+paste with a hex editor.

bump

Kartel, no, this is only for drives larger than 132gb and preventing data corruption.
LLXX, great piece of work, now I can get drives bigger 120gb

Amazing work. I tested the WIN ME Edition of ESDI_506.pdr on a 6.4GB drive (HP Vectra VL 6/400 with PHOENIX BIOS 4.0), it doesn't cause corruption on drives <128GB as it seems. Everything works fine until now. But i have to try with something bigger than 128GB...

I know I'm probably too late here but I think it would be better to change the version numbers like this...

4.10.2222 -> 4.10.2232
4.10.2223 -> 4.10.2233
4.10.2224 -> 4.10.2234
4.10.2225 -> 4.10.2235
4.10.2226 -> 4.10.2236

Two reasons:

1) there might actually be a 2227 or higher build in the wild that we don't know about so jumping by 10 leaves a gap for safety.
2) the last digit stays the same so you can easily tell what the original version was

Just for reference, there's a wiki page on microsoft version numbering...
http://en.wikipedia.org/wiki/Microsoft_Version_Number

I personally agree that a 1 version increment is all that is required. (2225 should be changed to 2227 as a 2226 version already exists).

For example:
2186 (98FE)-2187
2225/2226 (98SE)-2227
3000 (ME)-3001

Can you edit the version for 98FE (4.10.2186, same hotfix as 4.10.2225)?

the_guy

yea LLXX, let's not forget Win98 FE systems. there are still a few users out there stuck with Win98 first edition.

I agree with you
Too many versions gonna get confusing

I've searched the internets, there is no official ESDI_506.PDR version 4.10.2227. That version number is currently used for the fixed 4.10.2222. Fixed version of 4.10.2225 will be 4.10.2230. Adding 5 to the version number isn't that confusing...

BTW I've also fixed 4.10.2001 (Windows 98FE). I haven't found 4.10.2186 yet.

(Someone may want to provide more information on First Edition versioning so an appropriate scheme for the new files can be implemented.)

version 4.10.2186 is included with kb243450. Direct Download Link here.

Also, I know 4.10.2222 patched is 4.10.2227. Why not make 4.10.2225 patched 4.10.2228? Also, the 98FE version should be made 4.10.2187. Just my opinion.

the_guy

It is extremely confusing because all Microsoft hotfixes are cumulative, it means higher minor version number contains all fixes from the lower version number. The proposed numbering breaks this rule. See
General information about Windows 98 and Windows 98 Second Edition hotfixes

General information about Windows Millennium Edition hotfixes
contains the same statement.

My suggestion is to modify version 4.10.2225 only (forget about 4.10.2222, nobody needs it), and name it 4.10.1.2225. 4.10.2226 could be 4.10.1.2226 if anybody needs it. 4.10.2186 could be 4.10.1.2186 and 4.90.3000 could be 4.90.1.3000.

This would clearly indicate different versions branch for drivers with LLXX's patch.

But - it is nice to discuss about the version numbers but nobody verified 100% functionality on different disks and with different chipsets yet. This is much more important.

Petr

I don't agree about Version numbering not being first priority. We must assist that we don't get any confusing numbering in testing phase.

I didn't know about the extra digit that could also get placed before the buildnumber. I definetely vote for this numbering method

About testing. I'm already backing up my stuff right now, before I can run some tests on my system with 250 GB UDMA-6 HDD.

Couple of stupid question but here goes:

#1 What driver version is for 98se?

#2 Is this meant to be integrated into the source?

#3 How can I address the issue in fdisk where there is a limit of @8gigs?

There are two occurences of the vesrion number in the version resource.

The first is string type and there may be written anything, including any text details about the build, for example hhctrl.ocx has "5.2.3790.1830 (srv03_sp1_rtm.050324-1447)" in this field, and standard Windows 98 SE files have 4.10.2222 here. Maybe we can have here something like "4.10.2227 (LLXX 060722-0001)" to identify clearly the origin and detail version of the file - I think there may be many different builds of the same file.

The second is binary version number and it consists from 4 16-bit number, it means this version number may be in the range 0.0.0.0 - 65535.65535.65335.65335. Standard Windows 98 SE files have 4.10.0.2222 here.


Strange thing is with Windows Me files, most files have binary version number 4.90.0.3000 but some of them don't follow this rule:
but most of the files in mesp follow the a.m. rule:

There in no similar confusion for Windows 98 and 98SE - with one exception only:

Petr

Never mind I figured out all 3

#1 had to download 3 packs but I got the right one.
#2 Yup it sure is integrate-able.
#3 SuperFdisk!!! w00t!

I now have a win98 installed on a 200 gig drive from the beginning!

What would be the QUICKEST way to fill 130 gigs of space?

Copy stuff over until you reach this point. I usually do tar nearly 2GB of files and then copy it over until I reach the limit I want.

Unfortunately I can't test myself ATM.

FWIW, I too vote for this method of numbering LLXX's patch. Of all the schemes suggested so far, it is the least presumptuous, the least complicated and the least confusing, given the number of versions ESDI_506.PDR being patched and renumbered (when I say the "least presumptuous", I mean that if an official 4.10.2227 did show up [not likely, I know], then this method easily deals with it and avoids replicated numbering).

At the same time, this method clearly sets apart LLXX's patches from the official versions, but also produces no uncertainty about which patch corresponds to which official version. As Petr again pointed out, maybe the string type field can be used by LLXX to add some kind of identification for his patches, too.

LLXX - thanks, I can't wait to try it. You've done 9x a great turn.

Here you can find corrected version of FDISK http://support.microsoft.com/Default.aspx?kbid=263044. I have tried it with 250 GB disk, it works but has a display bug for disks and partitions bigger than 99 GB.

Or you can use any 3rd part too, I use Ranish Partition Manager.

Petr