[my2001]

Here's a list of available methods dealing with integrating (inf-based) device drivers (SATA,RAID,etc.)

M1.a described by raskren, see "Unattended install, SATA drivers, and NO Floppy"
M1.b described by RyanVM, see "So I finally decided to integrate RAID drivers" (no longer active) and "Bâshrat the Sneaky's DriverPacks - overview"
M1.c described by GreenMachine, see "Slipstreaming SATA/RAID Drivers"

M2.a described by trainee, see "TEXTMODE MassStorageDrivers Method" and "RASOR - Textmode Driver Utility"
M2.b described by Pure Knome, see "XP Basic CD Unattended SATA/RAID Install , How to avoid hitting F6 for CD Install"

M3 by Pyron, see "SOLVED: drivers from CD"

Questions:
Q1) Is this little summary correct?
Q2) Are there any other methods (threads)? What else to mention?
Q3) M2 doesn't need editing of TXTSETUP.SIF? [for answer see N3 below]
Q4) (deleted)

Notes: (independent of question numbering)
N1) Pure Knome offers in his posting (link see above at M2.b) a zipped manual where he describes his M2.b-method. In that document he also mentions M1 and refers to it as "method short". His own method is referred to as "method long" there.
N2) I try to update this posting from time to time as far as needed
N3) M1 does require editing of TXTSETUP.SIF; M2, i.e. M2.a and M2.b, do NOT require editing of TXTSETUP.SIF


Thanks for ya help!


edited_1: divided M1 into M1.a and M1.b; M2: new link added
edited_2: divided M2 into M2.a and M2.b according to trainee's reply to this posting; introduced new section "notes"
edited_3: added M1.c
edited_4: deleted Question Q4
edited_5: updated this topic with new links, and renamed some outdated linknames

This post has been edited by my2001: 16 July 2005 - 06:25 AM

[Bilou_Gateux]

M2 method is used by HP/Compaq SmartStart and IBM ServerGuide.

SmartStart is a bootable CD with wizard html code and scripts
ServerGuide is a bootable CD with wizard written in Java

Both asks the end-user wich OS to install on the server, the target disk is partitionned, formated bootable. Source I386 is copied from the user source CD to target disk.
and the most important, the OEM build specific TEXTMODE folder and UNATTEND.TXT file with the necessary infos related to the mass storage hardware.
See my post http://www.msfn.org/...showtopic=19792 : HP/Compaq example.

With the M2 method, you can install Windows unattended with your specific mass storage hardware with a standard M$ source CD.

[trainee]

Hey! I am still on a long trip so please excuse any errors in this post, I'm just trying to give my 'quick' 2cents.

Assuming that you are trying to get a specific driver working
It is usually easiest to find an example of someone who already has that driver working and then go with that. Obviously I am a big fan of M2 and I have been hoping that other people will try it out, but it is the least verified of the 3 methods and therefore the least help is available for it. If you just want to get it done, I suggest M1.

My answers to your questions:
Q1 & Q2. This summary seems correct, however Pure Krome's method evolved significantly into XP Basic CD Unattended SATA/RAID Install , How to avoid hitting F6 for CD Install. This is effectively M2 to the 20th power which is to say that it is somewhat complex (in my opinion anyway). However this is not to say it is incorrect, and may work for some drivers that the standard M2 method does not work for. As for my credit for M2, I certainly did not 'create it' however I did write the most recent and (hopefully) most complete guide. This newer Pure Krome link is more appropriate for something like M2.b and should be used instead of the other Pure Krome link. For those concerned that M1&M3 involve too much direct hacking of driver setup files, or have drivers like the 3ware's with more than 1 driver per Hardware ID, M2 may be the best alternative.

You may also want to reference the M1-style integration of certain drivers into GM's XPCREATE, which is very easy for people who have those drivers. It's only a subset of the drivers that have been shown to work with M1 however.
I would try to include only the most current/complete topics for each method since there is a lot of old/incorrect information that has been talked about while people were simply trying to figure things out. Actually your current list seems very good.
I have no personal experince with Pyron's method so I can't say much about it. I like the M1, M2 etc. nomencalture so M3 gets my vote, but I will leave it to someone more 'senior' than me to confirm this naming.

Q3. M2 does NOT require editing of TXTSETUP.SIF (well M2.a? anyway, I have not looked at M2.b in some time so it may or may not).

Q4. I have no experience with this either, however Bâshrat's is usually right on top of things and you should get an answer from him pretty soon I would think

Compaq/IBM Tools
I think that Bilou_Gateux is showing these 2 tools as proof that the M2 method is the 'proper' way to install drivers (since these large companies appear to use it). I personally feel that the 'proper' way to install drivers would be for Microsoft to let us drop the driver files in a folder and let-er-rip without any setup file hacking. I don't think this will happen any time soon. I am willing to bet that there is a whole lot of 'other' stuff giong on in these 2 tools that makes the M2 method work so well for them, I mean c'mon they want you to buy a service contract to use them.

That's It
Okay, I will end by saying that if you anyone has a chance or inclination to try out M2 for your hardware please do so! Right now the only confirmed M2 drivers (that I know of) are the 3ware 7k/8k drivers.

Edit: Yes! I am aware that I need to work on my conciseness

[my2001]

@ Trainee

Thanks a lot for your answer! I really appreciate it!

I edited my posting to reflect your issues.

But:

Quote

You may also want to reference the M1-style integration of certain drivers into GM's XPCREATE

I'm not quite sure if I know what you're talking about. Maybe you can gimme a link or so?

[Bilou_Gateux]

Another example of the M2 method

Intel® Application Accelerator RAID Edition
Unattended Installation Instructions Under Windows* XP

[GreenMachine]

my2001, on Jun 13 2004, 05:48 AM, said:

... I'm not quite sure if I know what you're talking about.  Maybe you can gimme a link or so?

http://greenmachine....ING/addraid.htm

As for all the fuss about "hacking" TXTSETUP.SIF: that is a load of rubbish ... the changes made are both documented by Microsoft, and quite similar to the "hacks" you will otherwise need to make to another Setup Information File, a.k.a. SIF file, WINNT.SIF. Six of one, half dozen of another ...

http://support.microsoft.com/default.aspx?...kb;en-us;817616

[bilemke]

Interesting summary page.. I note menioned by trainee...

Quote

I personally feel that the 'proper' way to install drivers would be for Microsoft to let us drop the driver files in a folder and let-er-rip without any setup file hacking.

This would be ideal, right now even the best integration requires some hack job of a unattended install.. Even chaning registry entries to point towards folders full of drivers.

Before trying using a unattended install almost a year ago I had done enough modifying to the windows install to install promise fasttrak tx2000 drivers. It was so seamless that windows did not notice the files were different then the rest of the stock/retail files(note, were WHQL drivers, so no issues with unsigned drivers). I did not nead a floppy or any unattended file. No modified registry entries in hive files or anything. Involved editing txtsetup.sif and one of the main stock windows driver inf files.. like scsipnp.inf I think... Anyway, besides that, just added primary .sys files and I was off.

It worked regardless if any unattend option was used or even if a unattend file was present. Problem was when I tried to expand the technique, it partially failed on drivers from almost any other manufacturer except promise?

To this day, from when I switched to OEM/Unattend methods described here, I have searched for a method close to the one I described above. One that would work with or without unattended setups.

Further note, I could hit f6 and put a floppy in with a driver for the same hardware, and setup would ask if I wanted to use the "integrated" drivers in the windows setup or use the ones from the oem floppy.

[bilemke]

Hmm. this thread got me searching..

GreenMachine pointed to http://support.microsoft.com/default.aspx?...kb;en-us;817616

Microsoft documented modifications to txtsetup.sif. I beleive they intended this for only adding support to the recovery console. This shows a relationship to the M1 methods.

However, another thread in this forum points to http://support.microsoft.com/default.aspx?...kb;en-us;817616

Now Microsoft describe driver integration (not the recovery console) in particulair. This is almost exactly how both M2 methods work. Thus I beleive (like Compaq/IBM example mentioned) this is the official way to integrate the drivers. The M1 methods produced here, will work, I don't see a problem with most drivers as the difference in the M1/M2 methods is very minor. A added benefit to the M1 method is if you use the cd produced from it to install the recovery console on your system, it should have support for the raid/sata controller already. A cd produced using the M2 method would fail as I do beleive the unnatend info is not called in to play when install the recovery console.

I rarely use the recovery conolse so I can not veryify this (use windows pe, BartPE).

[Ropera]

MY 2 CENTS: http://www.msfn.org/board/index.php?showtopic=23683&hl=

[RogueSpear]

Boilerplate Disclaimer: Yes I searched but didn't find anything about this exactly.

Is is possible / beneficial to put drivers in the drivers.cab file? Especially if some sort of script procedure could be made to the different infs and sifs.

My unattended project has been growing so complex lately that when something doesn't work I often just resort to using nLite on a virgin install source and plop all my files back in again.

It's getting to be a little much, all of the hacks, mods, and scripts we engross ourselves with for the sake of making the "perfect" unattended source. Sorry for the rant.