Help Needed: Breaking Change with Drivers

[rbalsleyMSFT]

Prior to releasing 2511.1Preview, I need some help testing the commits that I'm syncing today. It's a big one with about a month's worth of commits, primarily dealing with changes to drivers.

The TLDR is that I need people to help with testing driver downloads and the automatic driver mapping as my HP and Dell devices have aged out. I've tested Lenovo. Surface should still work as there shouldn't have been any changes there.

If you don't use git to sync the UI_2510 branch, download the latest code.

Driver Mapping Changes

The DriverMapping.json file and ApplyFFU.ps1 file have undergone some major changes. While the model name mapping worked in most cases, there were still some issues with both how vendors store their names in WMI and how model names are written in the various places I retrieve them (HP Image Assistant, Dell Command Update, Lenovo System Update - Surface is just from the Surface website, so it's fairly straightforward).

ApplyFFU.ps1 now relies on SystemID (HP and Dell), or MachineType (Lenovo) values and grabs those values from WMI. In order to use those values, I also had to get those values. I had MachineType already for Lenovo, but for HP and Dell I wasn't using them from their XML files. So what you'll find now in the DriverMapping.json file are SystemID and MachineType.

Here's an example of the new DriverMapping.json format. Notice the SystemId and MachineType properties. Surface doesn't have this as this hasn't become a problem (and yes, I'm aware of System SKU - may introduce this in the future, but wanted to get HP and Dell done first)

[
  {
    "Manufacturer": "Dell",
    "Model": "Dell Pro Max Desktops Dell Pro Max Micro FCM2250,Dell Pro Max Micro XE FCM2250 (0D14)",
    "DriverPath": "Dell\\Dell Pro Max Desktops Dell Pro Max Micro FCM2250,Dell Pro Max Micro XE FCM2250 (0D14)",
    "SystemId": "0D14"
  },
  {
    "Manufacturer": "HP",
    "Model": "HP EliteBook 865 16 inch G11 Notebook PC (8d03)",
    "DriverPath": "HP\\HP EliteBook 865 16 inch G11 Notebook PC (8d03)",
    "SystemId": "8D03"
  },
  {
    "Manufacturer": "Lenovo",
    "Model": "Lenovo 300w Yoga Gen 4 (82VN)",
    "DriverPath": "Lenovo\\Lenovo 300w Yoga Gen 4 (82VN)",
    "MachineType": "82VN"
  },
  {
    "Manufacturer": "Microsoft",
    "Model": "Surface Pro for Business (11th Edition)",
    "DriverPath": "Microsoft\\Surface Pro for Business (11th Edition)"
  }
]

If you use an older DriverMapping.json file with the new code, it's likely to break. You'll need to use the new format.

This means you'll also need to know where to find the SystemId for Dell and HP in the BIOS/UEFI.

For HP, either under Main or System Information (it's going to be different depending on the model) you're looking for the System Board ID and it should be a four-character code.

Via PowerShell - (Get-CimInstance -Namespace 'root\WMI' -Class MS_SystemInformation).BaseboardProduct

For Dell, I'm not sure if it's possible to get the System ID from the BIOS/UEFI. I seem to recall in some BIOS screenshots that System SKU is listed in some BIOS/UEFI implementations, but it may not be consistent.

Via PowerShell you can run - (Get-CIMInstance -ClassName "MS_SystemInformation" -NameSpace "root\WMI").SystemSku or

[string]$OEMString = Get-WmiObject -Class "Win32_ComputerSystem" | Select-Object -ExpandProperty OEMStringArray
$ComputerDetails.FallbackSKU = [regex]::Matches($OEMString, '\[\S*]')[0].Value.TrimStart("[").TrimEnd("]")

Thanks to the MSEndPointMgr folks and Johan for the pointers - https://github.com/MSEndpointMgr/ModernDriverManagement/blob/master/Invoke-CMApplyDriverPackage.ps1

Dell and HP Driver changes

In switching to using the SystemID for both HP and Dell, you'll see a lot more models for each

Here are the Dell Models

And the HP Models

~1600 Dell and ~900 HP devices. The model list also has the SystemID in parenthesis at the end of the model name.

The model names come directly from the XML files (productList.xml for HP; CatalogIndexPC.xml for Dell).

Dell has a lot of room for improvement (these are the nicest words I could possibly type) when it comes to consistency, which makes parsing their XML data pretty frustrating. So if you notice some really long and really short model names, those names are coming directly from their XML data (commas and slashes included!). There's further cleanup that can be done, but for now I think things work pretty well when downloading.

In most cases you should notice smaller driver packages for Dell devices. One of the issues I had with CatalogPC.xml was dealing with model numbers that are used across different brand names (e.g. Latitude, Inspiron, etc). CatalogIndexPC.xml kind of solves for this in that it provides a cab file to download for each systemID, but they don't do a good job in cleaning up the XML of the cab file, so you end up with a XML file that has a lot of old, outdated drivers in it. There's code that filters older drivers out, however you'll probably notice multiple video drivers per model download (e.g. multiple nvidia or AMD drivers; it's expected to have a single Intel and a single nvidia or AMD driver due to having an integrated and dedicated GPU). If you see this behavior, I'm aware and have tried a number of things to solve for it. This comes down to inconsistencies in their driver information in the XML files. I'll come back to this at some point to see if there's a better way to solve for it, but for now you can either remove the video drivers you don't need, or can manually download the drivers you want from Dell and bypass what I'm doing.

Let me know if you see any issues with Dell/HP drivers as these have undergone the most significant changes (in a good way).

Driver folder structure on the USB drive

Since the beginning, my guidance has been to put your drivers in model folders underneath the Drivers folder on the deploy partition of the USB drive (e.g. D:\Drivers\Lenovo 300w).

This worked well in the early days when we were doing a handful of devices and driver downloads via the BuildFFUVM.ps1 script were for a single model.

Now that the UI and BuildFFUVM.ps1 script have evolved to being able to download multiple models at a time, the Drivers folder needs better structure.

The new structure is Drivers\Make\Model (e.g. D:\Drivers\Lenovo\Lenovo 300w). This structure is consistent with how the UI and BuildFFUVM.ps1 script download and store drivers and automatically copy them. So if you've been following that, then no changes are required. But for those of you who have been manually copying your drivers to the USB drive and using the old D:\Drivers\<Model> format, you'll notice that when manually selecting the drivers to install from the menu during deployment that it shows the subfolders within your model folder. For example:

To fix this, just make sure use the Drivers\Make\Model folder structure.

Skip drivers when manually selecting drivers

You can now type 0 or just enter without selecting drivers to bypass driver installation (as per the screenshot above)

BITS Priority Setting in Build Tab

You can now set the BITS Priority (default is Normal) in the Build tab. This will set a new parameter named BitsPriority in the script. Prior, BITS downloads were done in Foreground since priority wasn't specified. Changing this to Normal actually caused an issue where parallel driver downloads stopped working since when not in Foreground priority BITS downloads happen sequentially. This meant that downloading multiple driver models went significantly slower.

So if you want downloads to happen really fast, set the BITS priority to Foreground. If you need to be a good network user, set them to Normal/High/Low.

HTTP Fallback for BITS Downloads as Standard User

As a standard user elevating PowerShell as admin, BITS downloads will now fail once and then fallback to HTTP. This works around the errors people were having when signed in as a standard user.

Those are the big changes. Let me know if things work or don't. Sorry for potentially breaking current processes, but these changes should help moving forward.

If you don't use git to sync the UI_2510 branch, download the latest code.

I plan to publish the 2511.1Preview Release first week of December.

[vangelis-agileactors]

Hello! I tested the new driver mapping version, but still there are a lot of Intel updates available for installation after installing the FFU on a laptop. I think the driver mapping is correct and at least I had network drivers during the OOBE (in order to do Autopilot pre-provisioning).

In the UI settings I disabled "Install drivers in the FFU", is there any chance this is why all the Intel drivers weren't installed?

In any case it's a much faster imaging process, will check for the next FFU builder updates!

[rbalsleyMSFT]

Can you attach the scriptlog.txt file from your USB drive?

[vangelis-agileactors]

Sure, here it is
ScriptLog.txt

[rbalsleyMSFT]

Thanks. That all looks good.

The reason Windows Update is providing those Intel drivers is likely due to Dell not providing the URL to download the chipset drivers in the XML file for the Latitude 5530 (it's also missing the Genesys memory card driver). This is one of the many inconsistencies Dell has between their XMLs, their website, and even their enterprise Driver CABs.

On Dell's website they flag the chipset driver as a utility.

In the Latitude_0B06.xml file, they don't list the chipset "utility." You can check this yourself in your C:\FFUDevelopment\Drivers\Dell\Latitude_0B06.xml file.

However, they do include the chipset in their driver pack but have a lot of older drivers or are missing drivers that aren't listed on their website for the model.

FFU Builder is downloading the latest of exactly what Dell provides in the Latitude_0B06.xml file. I can't explain why the Genesys drivers aren't there, but if I were to guess on the Intel Chipset drivers, my guess would be that since they flag them as a utility, their automation might just skip adding that category/classification to the XML file.

I wonder how Dell Command Update handles this. If they show Intel chipset drivers for this model, what are they using to query the URL?

[vangelis-agileactors]

That's very strange. However when running Windows Update for that laptop there were more than 10 Intel Updates, but there aren't many "utility" updates listed in Dell's website. I didn't keep the list of the updates that were missing after the FFU installation, will do on the next one and let you know.

[vangelis-agileactors]

I did another clean install from the FFU builder USB, this is the status of the laptop right after first login:

First image, 2 updates that probably happened during Autopilot (I guess)

And these are the available updates during the first "check for updates" in Settings

(it's the same Latitude 5530)

Hope it helps somehow

[Kofl]

Hi,

we have issues with two models so far, where some downloads fail.

Additionally FFU does not detect the drivers for "HP Elitebook 830 G9" although it was included and the folder exists:
E:\Drivers\HP\HP EliteBook 830 13.3 inch G9 Notebook PC

04/12/2025 16:09:59 Dispatching INTERMEDIATE status for 'HP EliteBook 830 13 inch G10 Notebook PC (8b41)': '13/30 Extracting Intel Innovation Platform Framework'
04/12/2025 16:09:59 Attempt 1 of 3 failed to download https://ftp.hp.com/pub/softpaq/sp160001-160500/sp160379.exe. Error: The resource loader cache doesn't have loaded MUI entry. (0x80073B01).
04/12/2025 16:10:00 Driver extracted to: C:\FFU-UI_2510\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\Intel Innovation Platform Framework_2.2.10002.4_sp157496

04/12/2025 16:20:44 Dequeued progress update: {"Status":"Error: The resource loader cache doesn't have loaded MUI entry. (0x80073B01)","Identifier":"HP EliteBook 850 G7 Notebook PC (8724)"}
04/12/2025 16:20:44 Dispatching INTERMEDIATE status for 'HP EliteBook 850 G7 Notebook PC (8724)': 'Error: The resource loader cache doesn't have loaded MUI entry. (0x80073B01)'
04/12/2025 16:20:52 Update-SelectAllHeaderCheckBoxState: Selected count is 0 for ListView 'lstUSBDrives
```'.

[rbalsleyMSFT]

For the HP EliteBook 830 13 inch G10 Notebook PC (8b41)

I just tried downloading that model and it was successful. What was interesting is that I see 28 items, not 30. Can you try again? Here's what I see in the FFUDevelopment_UI.log. It's possible HP had an issue and updated their XML for that model.

12/9/2025 10:49:59 AM Downloading driver to: C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\sp157496.exe
12/9/2025 10:49:59 AM Dequeued progress update: {"Status":"13/28 Downloading Intel Innovation Platform Framework","Identifier":"HP EliteBook 830 13 inch G10 Notebook PC (8b41)"}
12/9/2025 10:49:59 AM Dispatching INTERMEDIATE status for 'HP EliteBook 830 13 inch G10 Notebook PC (8b41)': '13/28 Downloading Intel Innovation Platform Framework'
12/9/2025 10:49:59 AM Successfully transferred https://ftp.hp.com/pub/softpaq/sp157001-157500/sp157496.exe to C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\sp157496.exe.
12/9/2025 10:49:59 AM Driver downloaded: C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\sp157496.exe
12/9/2025 10:49:59 AM Creating extraction folder: C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\Intel Innovation Platform Framework_2.2.10002.4_sp157496
12/9/2025 10:49:59 AM Extracting driver C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\sp157496.exe with args: /s /e /f "C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\Intel Innovation Platform Framework_2.2.10002.4_sp157496"
12/9/2025 10:49:59 AM Running HP Driver Extraction Command: C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\sp157496.exe /s /e /f "C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\Intel Innovation Platform Framework_2.2.10002.4_sp157496"
12/9/2025 10:49:59 AM Dequeued progress update: {"Status":"13/28 Extracting Intel Innovation Platform Framework","Identifier":"HP EliteBook 830 13 inch G10 Notebook PC (8b41)"}
12/9/2025 10:49:59 AM Dispatching INTERMEDIATE status for 'HP EliteBook 830 13 inch G10 Notebook PC (8b41)': '13/28 Extracting Intel Innovation Platform Framework'
12/9/2025 10:50:00 AM Driver extracted to: C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\Intel Innovation Platform Framework_2.2.10002.4_sp157496
12/9/2025 10:50:00 AM Deleted driver installer: C:\FFUDevelopment\Drivers\HP\HP EliteBook 830 13 inch G10 Notebook PC (8b41)\Driver - Chipset\sp157496.exe

I was also able to successfully download the 'HP EliteBook 850 G7 Notebook PC (8724)'

[Kofl]

Thanks for verification, G10 works fine now, G7 still fails. Maybe a HP mirror server issue.

[ihatecreatingusernames]

I'm pretty new to FFU and had just run into issues with the drivers not applying properly for some of our HP desktops. I was so happy to see this post. I applied the latest code zip above on 12/23. It worked perfectly for the HP EliteDesk 800 G3 SFF that I was testing with.

The only minor issue I had was with models that use multiple IDs. One example is our HP EliteBook 840 G8s. I've found three IDs for that model in our environment (880d, 8ab3, 8ab8). FFU Builder downloads pretty much the same thing for all of them. One difference is the 8ab3 pack didn't include AMD or Nvidia graphics drivers. I manually modified the jsons to point to the same WIM for testing, but as soon as I clicked Build FFU, it reverted my json and downloaded them all again. I COULD ignore it all, but if I add any more models, I'm going to need a larger USB drive.

[rbalsleyMSFT]

It reverted the DriverMapping.json I assume?

[ihatecreatingusernames]

It reverted the DriverMapping.json I assume?

Correct. DriverMapping.json was modified. But I'm now thinking that I clicked "Download Selected" again and that's when it reverted the json, not when I clicked Build FFU. That was a week ago and I can't remember exactly. I'll need to test again. Maybe modifying DriverMapping.json without clicking Download will be fine.

[rbalsleyMSFT]

It should only modify once a download happens (whether from the UI or BuildFFUVM.ps1). If you've manually modified it after downloading/copying in your own drivers, then it shouldn't be modified again.

[astavitsky]

Not having any luck with automatic driver matching for at least 2 Surface models: 2038 and 2103. I know I can manually modify the DriverMapping.json file, but I expected it to work out of the box. I added ALL Surface drivers to the USB via 2511.1Preview UI. Attaching the ScriptLog.txt for 2038 and will attach for 2103 shortly.
ScriptLog_2038.txt
ScriptLog_2103.txt

Edit: DriverMapping.json file for reference (auto-generated, not manually edited)
DriverMapping.json

[astavitsky]

Drivers.json

[rbalsleyMSFT]

This is expected as Surface currently does an exact match on the model name. For the Pro 9 and Pro 11 devices, they have multiple drivers for Intel/ARM but the model name is generic for the Pro 9. For the Pro 11, the normalized name contains most of the model string, but isn't exact.

I'll need to think of a good way to map what the script pulls from the Surface Drivers Page and the Surface System SKU Reference. There are more nuanced models from the System SKU reference page that need to be mapped to what is available to download. I'm trying to avoid having to manage a list of models when Microsoft releases new models and have the script reach out to the page and just figure it out.

[rbalsleyMSFT]

@astavitsky can you try this branch? https://github.com/rbalsleyMSFT/FFU/tree/SurfaceMapping. I think this should work. If you can test with as many surface models as you can, that'd be really helpful as I don't have many myself to test with.

[astavitsky]

It worked correctly with Surface Pro 11 and 9, both with Intel CPU. I will test with more devices during the work week and report back.
ScriptLog.txt

[rbalsleyMSFT]

Awesome, thanks for the quick test! The log looks good. Looking forward to the other test results

[rolfhh]

I've been testing on our Dell Latitude laptops with v2601.1Preview. The only issue so far is on a Latitude 5410, where the deployment does not complete because a driver fails to inject. Looking at issue #345, this does not appear to be the expected behavior when injecting from the USB drive.

Here's the log file:
ScriptLog.txt

[rolfhh]

Looks good! I tested it on the problematic 5410 and the deployment completed successfully, with a warning about the failed driver.

Here are the logs, including a big, new one that seems related to the driver injection:
dism.log
ScriptLog.txt
setupapi.offline.log

[rbalsleyMSFT]

The setupapi.offline.log gets copied when driver errors happen. It allows me to see what happened (dism provides an error, but setupapi.offline.log provides more details).

[rolfhh]

As a final note on this I got absolutely nowhere with Dell's support because we have no 5410s that are under warranty. Then again, the driver in question was for a sensor that helps park hard drive heads if the laptop is physically dropped.

We haven't used actual hard drives in laptops for many years, but the 5410s do indeed support 2,5" SATA drives of the spinning rust variety in some configurations.

[rbalsleyMSFT]

Dell has a broader problem with keeping their tools consistent. They're the hardest OEM to work with due to the inconsistent nature of how they manage drivers. I'd say even if your 5410s are out of support, you probably could still talk to them about wanting to have more consistency between Dell Command Update and what they post on their website. Even using DCU, Support Assist, etc you're unlikely to get the latest drivers they have published on their site.

[rolfhh]

I'll document some cases of driver jank and have a go at our customer contact at Dell. We're a tiny account in the grand scale, but it's worth a shot.

And yes, the driver inconsistency is horrendous. I mentioned I found it messy earlier, but then spent some time digging after you commented on it, and it's way worse than I'd realized. It can be terrible even for recent models.

[PixelbeardQc]

I have been running a couple version of the UI build since that change and its been pretty solid here.

Can you comment one one thing, We have Lenovo deployed, is there a reason to download both version for a single model? I belive that when you use update retreiver they are bundled together. I have my tech use the existing one if we ever receive a stray laptop from the other branch.

Exemple ThinkPad P16s Gen 2 has 21HK and 21HL

[rbalsleyMSFT]

For the automatic driver matching to work, you'll need to have both versions downloaded if you have both machine types. If you know you only have one version, then you just need to download that single version.