HELP NEEDED! Win7 DWM.EXE issues upon install any m$ rollup of 2018

[Assurancetaurix]

Thanks for your replay. I did read all I could find about the Spectre-Meltdown registry settings and I did also try all possible combinations. But on my machine it did not make any difference. At 1.6 GHz (60%) everything runs fine. At 1.2 GHz (50%), no, a crash before you can say sh..

In the mean time, I took your suggestion and changed to Windows 7 x64. That meant a re-install from scratch. So the 32 bit environment that I had is now only in backup. Using 64 bit now. Same setup.

Please I meant no offense to anyone. I am not accusing MS of anything and I am not pretending that I have a better solution ! In fact, I don't have any solution myself. If it was not for your excellent work here I would never even have known what the problem was, let alone what the possible remedy would be. I am very grateful for your efforts. By myself I would have probably concluded that this machine should go in the garbage can. Even though I know it was working perfectly untill MS started to fix things ...

The whole story of Spectre/Meltdown seems a little exagerated to me, especially for my machine. It is mostly academic. I see the true solution only at hardware level. A job for the CPU manufacturers. In fact, I expect to hear a lot more about similar type of bugs in the near future. There may come a point when software fixing becomes intolerable.

What I don't understand, we seem to be able to switch off the Spectre-Meltdown fix with registry settings. But why then can I not lower the CPU speed if I do that ?

 

[laidbacktokyo]

oops, there is no any problem here...
but a minor problem @ your side seems to be your free interpretation of my words written here...

so pls forget all m$ academic words about spectre-meltdown and just do as follows:

1. change your win7 pagefile size to Recommended (the exact value as advised by your particular win7 in megabytes!) i.e. the visible by win7 RAM x ~1.5, would it be either win7x64 or x86!

2. set/create manually the Spectre-Meltdown protection win7 registry keys as:

[HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management]
"FeatureSettingsOverride"=dword:00000008
"FeatureSettingsOverrideMask"=dword:00000003

3. install the latest servicing stack update (SSU) KB3177467 = v2 - if not offered via win7 update center then download & install manually:
Microsoft Update Catalog

4. install November 13, 2018—KB4467107 (Monthly Rollup) support.microsoft.com/en-us/help/4467107 right over Dec2017 rollup!!!!

UPDATE: November 27, 2018—KB4467108 (Preview of Monthly Rollup)
support.microsoft.com/en-us/help/4467108

5. update your NET FW 4.5~4.7.2 if installed to the latest .NET Framework 4.8 Early Access build 3646 Announcing .NET Framework 4.8 Early Access build 3646 | .NET Blog

UPDATE: Announcing .NET Framework 4.8 Early Access build 3694 | .NET Blog

then set min state of cpu to 50% and report results here.

p.s. The Spectre-Meltdown protection registry keys are absolutely beyond any cpu speed adjustments! Naturally except the simple meaning of more or less loaded cpu!

p.p.s. you can easily check the Spectre + Meltdown protection status of your win7 by use of this tiny & portable utility GRC | InSpectre

p.p.p.s. pls don't worry! it was not you but me who's blaming m$ for idiotism and saying - f*ck m$!

 

[Assurancetaurix]

OK, I understan, I will try that. But it means that I have to uninstall until Dec 2017 level and then install Nov 2018 immediately after that. My visible ram is now 4G, so I will need to set the pagesize to 6G. I am not sure if I correctly understood the meaning of the registry values. I thought that when disabling the Spectre/Meltdown protection, I should have the same performance and CPU load as I had until Dec 2017. But that does not seem to be the case. I thought that I could just go back to minimum of 5% because the fixes are disabled and do not require more load than when not installed. Did I make any sense here?

Anyway it may take me a little while before I can do these operations. Please bear with me.

 

[laidbacktokyo]

My visible ram is now 4G, so I will need to set the pagesize to 6G.

Pls just set pagefle size to Recommended!!!!!! win7 will do the rest by itself!

I thought that when disabling the Spectre/Meltdown protection, I should have the same performance and CPU load as I had until Dec 2017.

It's a wrong guess. Unfortunately!!!!

I thought that I could just go back to minimum of 5% because the fixes are disabled and do not require more load than when not installed. Did I make any sense here?

Again. It's a wrong guess. A turning off Spectre or Meltdown or both in win7 registry never returns to all known CPUs (maybe except just introduced intel 9th gen ones) full performance. So if you wanna the full performance as before 2018 pls stay on Dec2017 rollup!

p.s. no problem here. I'm not in hurry.

c ya then

 

[Assurancetaurix]

Thanks for highlighting how wrong my assumptions are ! It was perhaps too naive of me. I was trying to test again, following all your directions, but I was interrupted and after retuning days later, I decided to give up. I am now using Win7 x64 for a while, with everything installed that I had in Win7 x86. I haven't had a crash in the nvidia driver yet. That makes me happy. And being on Dec2017 rollup, also no issue with DWM.

I decided to keep this system as it is. I am making here the assumption that in all the 2018 (and later) rollups are not going to be an significant improvement in anything. There are probably only security fixes on security fixes for Spectre/Meltdown type of vulnerabilities. I do not rate these as high risks.

Even if in future I would buy a new recent CPU, such as Coffee Lake, I would still want to use Win7 and I would also stop at Dec2017 rollups even though there should not be DWM crashes. But in my view Win7 is only getting more and more crippled ...

But then again, my assumptions could all be wrong again ?!

 

[laidbacktokyo]

1. As for your both mobile videocards 8600M & 8400M and 342.01 driver crashes, so you'd better do the clean install of a bit older driver GeForce 309.08 Driver Graphics Driver - GeForce 309.08 Driver | NVIDIA It contains a bit less of that damn security and performs much better than 342.01. Also pls install driver only. A technique has been discussed in this thread earlier on page#2.
2. You can't upgrade your C2D of 2008 mobile CPUs to Coffee Lake ones. It's just physically impossible! Also if you'd buy a new coffee lake laptop and install win7, then you'll meet another problem concerning m$ updating - it's another m$ crap - and you'll need this utility to be installed: Releases . zeffy/wufuc . GitHub Also now there is a sense to buy not 8th gen intel cpus but 9th gen ones, which are designed with spectre-meltdown protection integrated.
3. As for your current decision to keep staying on Dec2017 rollup, so I assume it as a bit odd but anyhow it's your choice. Also you can return to further updating at any moment.

rgds,

 

[Assurancetaurix]

Thanks for all your help. I am coming back now on this topic, since it is not over. I have been using my laptop all this time without any issues. I am still on 2017-12 update level, with further updates only for other stuff (.NET Framework, etc), but no more roll-ups. However, they got me now. There is a new update KB4474419 for SHA-2 code signing support. It seems to be mandatory (in future). I installed it along with another update, can't remember the number. And the the DWM crashes started to appear again every few minutes or so.

This update KB4474419 contains an update to the processor driver (intelppm.sys). Remembering what you mentioned above, I strongly suspect this to be the culprit. Is there a way around it? Is this processor update really required for SHA-2 code signing, since the processor really did not change ? Can the MS driver still be used instead ?

If this issue can be solved without raising processor frequencies, perhaps I could continue to install the rollups as well ?

(I did in fact buy a second home PC, type all-in-one. It has an intel 300 series chipset. Impossible to use Windows 7 on it, since there seem to be no drivers for USB 3 for this chipset for WIndows 7 And since it is all-in-one, I cannot add another USB controller ! Argh, only choice here is to go to Windows 10. This all-in-one PC just makes a beautiful wall decoration at present. What a pitty !)

 

[laidbacktokyo]

1. yep, you are right - fortunately or unfortunately. KB4474419 is the culprit of your dwm crashes return. KB4474419 is not rollup but something like security only version of rollup and it contains win7 core files update, not cpu driver only! its intelppm.sys + Processr.sys + Ntoskrnl.exe = updated to V6.1.7601.24384. but you'll need it to be installed by some Aug-Sep2019 to get further updating...
so you still have same 2 basic options - to remove KB4474419 and wait for some news here until Aug-Sep2019 (if any) or set your cpu min state to 50 or 60 or 100% (pls refer to all above info again) aimed to be on safe side to install any update you want including all rollups.
also you can use the same purpose utility as mentioned by me somewhere above (it's upgraded now to a new name) Quick CPU - CPU Core Parking and Performance Optimization where its Frequency slider must be set to same 50 or 60 or 100%

2. Intel 300 series chipset + Win7 + USB3 no driver - sorry you comment sounds like incorrect = your lack of info. pls post here device IDs of USB3 controller(s) and I'm almost sure that I'll be able to find a decent driver! I mean the required device IDs info should looks something like:
PCI\VEN_8086&DEV_283A&SUBSYS_01FF1028&REV_02
PCI\VEN_8086&DEV_283A&SUBSYS_01FF1028
PCI\VEN_8086&DEV_283A&CC_0C0320
PCI\VEN_8086&DEV_283A&CC_0C03

also pls post the exact mobo model name of your new all-in-one pc...

rgds,

 

[Assurancetaurix]

Thanks that's clear. For now I uninstalled this update. I wonder whether is would still be worth while to have new updates since updates are supposed to end anyway somewhere in January 2020 when extended support for Windows 7 runs out. My laptop is working just fine now. Especially with the new (old) nVidia driver that you mentioned above.

I checked the devices in my new all-in-one wall picture. The USB controller is:

PCI\VEN_8086&DEV_A36D&SUBSYS_84EE103C&REV_10\3&11583659&2&A0
Intel(R) USB 3.1 eXtensible Host Controller - 1.10 (Microsoft)

As far as I understand Intel never made a driver for it for Windows 7. Neither did MS. Would be great if you had any news on it !

 

[laidbacktokyo]

1. Pls post motherboard full model name!
2. Pls post all compatible device IDs as my above sample!

 

[Assurancetaurix]

The All-in-one is HP model XA0006NG. Motherboard is proprietary (DiamondStar, code 84EE). ID's:

Compatible ID's:
PCI\VEN_8086&DEV_A36D&REV_10
PCI\VEN_8086&DEV_A36D
PCI\VEN_8086&CC_0C0330
PCI\VEN_8086&CC_0C03
PCI\VEN_8086
PCI\CC_0C0330
PCI\CC_0C03

Hardware ID's:
PCI\VEN_8086&DEV_A36D&REV_10
PCI\VEN_8086&DEV_A36D
PCI\VEN_8086&CC_0C0330
PCI\VEN_8086&CC_0C03
PCI\VEN_8086
PCI\CC_0C0330
PCI\CC_0C03

 

[laidbacktokyo]

1. This one?
Access Denied
Access Denied
Access Denied
If yes then its mobo's HP name isn't DiamondStar but - HP name: Diamaster SSID: 84EE

2. Yep, it seems there are the well known driver problems of this usb3 controller & win7. Anyhow pls first try this Gigabyte/Intel beta driver of 2016 fall:
Intel(R) USB 3.0 eXtensible Host Controller driver Download for WIN7/64bits - Gigabyte B360M-D3H
It's worth to try when its published compatible HW IDs include:
Compatible IDs:
PCI\VEN_8086&DEV_A36D&REV_10
PCI\VEN_8086&DEV_A36D
PCI\VEN_8086&CC_0C0330
PCI\VEN_8086&CC_0C03
PCI\VEN_8086
PCI\CC_0C0330
PCI\CC_0C03
also it's for win7x64 + win8x64 = [Intel.NTAMD64.6.1] + [Intel.NTAMD64.6.2]
however I don't know other way to try if it's ok or not for sure as to install win7 on this pc first.

Thus pls first locate the 3 below info files of this driver controller + hub + Switch inside of your currently installed win10x64 - iusb3xhc.inf + iusb3hub.inf + iusb3hcs.inf and upload them to any web sharing space, and drop me a link here.

 

[Assurancetaurix]

Actually, you are right. It's the Diamaster 84EE. You see, cannot trust my memory. Some bits are flipping because no ECC checking ! Indeed the doc you found is correct. The chipset is H370. The B360 seems to be an older generation (200 series). There is also a B365 that is a 300 series and has Windows 7 support, but no USB driver ! A very weird situation. I did this research a while ago, so all is from memory. Some bits could have flipped in my brain !

DriverIdentifier shows wrong information. I already tested this driver. It is a 200 series driver. Not for the H370 chipset. You can also see that in the driver date (2016) which is way before the release of the 300 series chipset.

I have the 2 files you wanted to see, but I don't know anymore where to upload. Instead I have attached them here. BTW the file names are different now.

 

[laidbacktokyo]

Did you really take both files from your current live win10x64?
If so I don't see any OS version limitations inside - it's just winNTx64 DriverVer = 09/14/2018,10.0.17763.1, and it seems to be native win10 driver (that's why it's n/a for win7!)

1. Then another your obvious option is to backup all 3(!) parts of this driver from your current win10x64 by any driver backup software (I personally know that Driver Genius Pro performs drivers backup quite well) to some safe location beyond your current win10x64, and install them to Win7 manually!

2. Another option seems to be here - modded drivers by people doing such modding on everyday basis. Pls read this forum Forum - USB 3.0/3.1 Drivers (original and modded) or at least 1 page here Forum - Coffee Lake with W7 USB /AHCI /MEI Drivers S**TSTORM and @ the bottom post download the mod pack. It's safe at least!



Fernando has attached files to this post

• pure 64bit Intel USB 3.0+3.1 Drivers v5.0.4.43 mod+signed by Fernando with DPInst option.rar

 

[Assurancetaurix]

Yes, these files are from the actual machine. Checked in Device Manager which .inf loads the usb driver and hub. Then copied the .inf's. The file names you mentioned (iusb...) do not exist on the machine.

Agreed, the drivers do not mention any OS limitation. But they don't work on Windows 7 ! Neither do any Windows 8/8.1 drivers.

I did indeed check all of the discussion and attempts on Fernando's forum. None of the modded drivers actually work. Fernando himself also admits that in several places. But it is a long read. Latest comment from a few days ago:

Forum - RE: USB 3.0/3.1 Drivers (original and modded) - 135

Considering the huge interest, whenever a solution will be found we can be sure it makes HEADLINES. The latest views are to take advantage of leaked sources/ Linux drivers, etc. Pretty desparate ...

The MS drivers are quite generic. It seems Intel has completely stopped driver efforts and leaves the game to MS.

 

[laidbacktokyo]

Agreed, the drivers do not mention any OS limitation. But they don't work on Windows 7 ! Neither do any Windows 8/8.1 drivers.

Again: I can't check this by myself due to a lack of hardware but the backuped drivers from win10 - drivers only! - not just driver files copied manually but backuped by some dedicated drivers-backup-software will likely work on win7! And there are many options like modding inf files and/or forcing driver install... etc. Sorry I won't do this w/o hardware to test near me.

rgds,

p.s. if you hate to use win10 then install win8.1 (with win7 GUI software optionally). as far as I know win8.1 also has native usb3 drivers integrated

 

[Assurancetaurix]

Understand. The driver problem is not in the backup or the modding. It is the actual driver code. Indeed, a modded install can be forced. But then the driver still does not work. Alas. Until now no solution seems to have been found by anyone. Also not the mainboard manufacturers that support Windows 7.

Yes, Win8.1 could be an option. But that is also such a 2-legged system. And the work on GUI mod's has stopped long ago. Windows 10 is an on-going construction site. Changes and instabilities. Will it ever be finished ? I am beginning to believe that MS has a new generation of programmers with little knowledge of the code. It is easier to throw away than to support what you don't understand.

At least Windows 7 is a fairly consistent system. Any missing func's can be added easily with 3rd party developments. So sorry I cannot use it on this new hardware ! I had not anticipated this when I bought it. I checked all major Windows 7 drivers and never thought the USB would be a problem. Until the previous chipset (200 series) Windows 7 is fine. But as from series 300: No Windows 7.

MS have been fairly unsuccessful in suppressing Windows 7. But Intel has.

 

[laidbacktokyo]

Understand. The driver problem is not in the backup or the modding. It is the actual driver code.

well... I'm sorry to say that you are a bit lost in your own self-contradictions...
your current win10 native usb3 drivers are performing ok... right?
then these drivers contain code for intel 300 series chipset you are looking for!
sorry I can't say it many times again & again - try some driver-backup software.

p.s. win7 GUI software development stopped due to win10 way of development. however the latest version of this well-known GUI software should perform ok on win8.1...
p.p.s. I guess next time you'll stay away from buying all-in-one like-apple-style PCs due to a lack of the external and/or internal extension slots - PCI+PCI-E mainly. It's a a fun when even a decade ago laptops had pci extension slots....

 

[Assurancetaurix]

The native Windows 10 drivers work fine on Windows 10, of course. The problem is to get them to work on Windows 7. I have not been able to do that, neither has anyone else in as far as I can see. The mod's that are floating around do not really work (in a stable way). But my own talents are quite limited as you correctly noticed. I just hope that one day you will have yourself an opportunity to test on a PC with this 300 series chipset.

All-in-one PC's are a design choice but have disadvantage. Perhaps I should go back to custom build. But I do not have a lot of floor space here. That was why the all-in-one ... My mistake.

 

[laidbacktokyo]

Pls don't listen anyone, even real experts. Always first make os restore point and then try! when any test always depends on the real hardware+os.
Aside in your case I insist that it's worth to try win10 native usb3 drivers backup (pls all 3 of them = controller+hub+switch) and then reinstall all to win7 of same x64. Manually thru .inf files or thru that original-driver-backuping software. z 2nd option looks pretty much better, at least for backing-up drivers!

well here is just a reference of m$ (!) drivers available in general:
Intel(R) USB 3.0 eXtensible Host Controller - 1.0 (Microsoft) Driver Download - PC Pitstop Driver Library
Intel(R) USB 3.1 eXtensible Host Controller - 1.0 (Microsoft) Driver Download - PC Pitstop Driver Library
Intel(R) USB 3.1 eXtensible Host Controller - 1.10 (Microsoft) Driver Download - PC Pitstop Driver Library

p.s. I'd appreciate if you'd just follow my manual/request if any and then report result to estimate any next step. thanks.
p.p.s. dejavue. is this spelling correct? it seems I've discussed same or similar issue long ago.

btw, folks are saying that Z370 ONLY(!) performs ok on 5.0.4.43 v2 drivers for older chipsets.... Download Intel(R) USB 3.0 eXtensible Host Controller Driver for Intel(R) 8/9/100 Series and Intel(R) C220/C610 Chipset Family
but other ones like B3XX, H3XX, Z3XX don't.

pls request HP to replace H370 mobo to the based on Z370 one