Bug 839354

Hi, mathias 

We only put the signed qxl driver into virtio-win package .And qxl driver is maintained by anotherteam belongs to qxl-win component . 

No qxl driver for win2k8  means we do not provide QXL driver for this platform even all the *Server* versions of Windows.

Let's move the right component (qxl-win) to get more info about this .

Mike

I have RHEV-M 3.2 (sf18.3) installed and have then installed a W2K8R2 sp1 server as a VM. I verified with the latest tools (rhev-guest-tools-iso-3.2-8.noarch.rpm) that the QXL driver does *not* get installed on the W2K8R2 sp1 server. I then added the qxl-win-0.1-17 Windows7x64 drivers (qxl_w7_x64.zip) manually and don't seem to have any apparent issues.

(In reply to Bill Sanford from comment #5)
> I have RHEV-M 3.2 (sf18.3) installed and have then installed a W2K8R2 sp1
> server as a VM. I verified with the latest tools
> (rhev-guest-tools-iso-3.2-8.noarch.rpm) that the QXL driver does *not* get
> installed on the W2K8R2 sp1 server. I then added the qxl-win-0.1-17
> Windows7x64 drivers (qxl_w7_x64.zip) manually and don't seem to have any
> apparent issues.

This break MSFT policy and Never try on this !
You can find in qxl.cat that the driver is not signed for win2k8 R2 . You must do whql certification on the win2k8R2 and submit it to MSFT .
Although the driver shows "signed" ,but if you run WHQL "signed driver check " job ,you will see the job with sth like "the driver is digital signed on XXX platform .but not for win2k8R2 platform"

Mike Cao, can you point to the policy this breaks, hopefully it's on msdn somewhere?

That's just curiosity - afaict Bill was just pointing out a workaround for installing the driver if you have to. The bug is still valid.

Alon

(In reply to Alon Levy from comment #7)
> Mike Cao, can you point to the policy this breaks, hopefully it's on msdn
> somewhere?

Tried "signed driver check" job in HCK ,will you see the job failed . It should be on MSDN ,I will find the MSFT policy to you later and what's more that's why virt-qe certify virtio-win driver on 11 platform as well.
> 
> That's just curiosity - afaict Bill was just pointing out a workaround for
> installing the driver if you have to. The bug is still valid.

Yes .That's what I usually do .but it is just a workaround ,we can not ship the driver w/o MSFT signed on the specified operation systems

Mike  
> 
> Alon

I cannot reproduce this bug with qxl-win-0.1-17 without having to set testsigning. This suggests it is a problem in packaging - either the driver in virtio is older (although I doubt that is a problem), or somehow the cat file is mangled to destroy the signature.

Could the original reporter verify that qxl-win-0.1-17 works for him, and then I can move this bug to virtio-win ?

Alon

(In reply to Alon Levy from comment #10)
> I cannot reproduce this bug with qxl-win-0.1-17 without having to set
> testsigning. This suggests it is a problem in packaging - either the driver
> in virtio is older (although I doubt that is a problem), or somehow the cat
> file is mangled to destroy the signature.
> 
> Could the original reporter verify that qxl-win-0.1-17 works for him, and
> then I can move this bug to virtio-win ?

This is MSFT policy issue referring to comment #9 .there is no choice but to run whql test over win2k8R2 and submit it to sysdev.microsoft.com first before we said we support it .
 

Mike
> 
> Alon

What's more ,the way to check OS info in the .cat file
http://msdn.microsoft.com/en-us/library/windows/hardware/gg463054.aspx

(In reply to Mike Cao from comment #11)
> (In reply to Alon Levy from comment #10)
> > I cannot reproduce this bug with qxl-win-0.1-17 without having to set
> > testsigning. This suggests it is a problem in packaging - either the driver
> > in virtio is older (although I doubt that is a problem), or somehow the cat
> > file is mangled to destroy the signature.
> > 
> > Could the original reporter verify that qxl-win-0.1-17 works for him, and
> > then I can move this bug to virtio-win ?
> 
> This is MSFT policy issue referring to comment #9 .there is no choice but to
> run whql test over win2k8R2 and submit it to sysdev.microsoft.com first
> before we said we support it .

I don't understand your response. What I was saying is that the current *signed* driver *installs* correctly on win2k8r2sp1, so it already has whatever is required. Where is the problem then?

>  
> 
> Mike
> > 
> > Alon

(In reply to Alon Levy from comment #13)
> (In reply to Mike Cao from comment #11)
> > (In reply to Alon Levy from comment #10)
> > > I cannot reproduce this bug with qxl-win-0.1-17 without having to set
> > > testsigning. This suggests it is a problem in packaging - either the driver
> > > in virtio is older (although I doubt that is a problem), or somehow the cat
> > > file is mangled to destroy the signature.
> > > 
> > > Could the original reporter verify that qxl-win-0.1-17 works for him, and
> > > then I can move this bug to virtio-win ?
> > 
> > This is MSFT policy issue referring to comment #9 .there is no choice but to
> > run whql test over win2k8R2 and submit it to sysdev.microsoft.com first
> > before we said we support it .
> 
> I don't understand your response. What I was saying is that the current
> *signed* driver *installs* correctly on win2k8r2sp1, so it already has
> whatever is required. Where is the problem then?
Just run signed driver check job in HCk ,you will find the problem ,it will say something that "the driver is digital signed ,but not for win2k8R2 platform" ,what's more ,the cat file will show the OS arch is win7 x64 instead of win2k8R2 
> 
> >  
> > 
> > Mike
> > > 
> > > Alon

(In reply to Alon Levy from comment #13)
> (In reply to Mike Cao from comment #11)
> > (In reply to Alon Levy from comment #10)
> > > I cannot reproduce this bug with qxl-win-0.1-17 without having to set
> > > testsigning. This suggests it is a problem in packaging - either the driver
> > > in virtio is older (although I doubt that is a problem), or somehow the cat
> > > file is mangled to destroy the signature.
> > > 
> > > Could the original reporter verify that qxl-win-0.1-17 works for him, and
> > > then I can move this bug to virtio-win ?
> > 
> > This is MSFT policy issue referring to comment #9 .there is no choice but to
> > run whql test over win2k8R2 and submit it to sysdev.microsoft.com first
> > before we said we support it .
> 
> I don't understand your response. What I was saying is that the current
> *signed* driver *installs* correctly on win2k8r2sp1, so it already has
> whatever is required. Where is the problem then?

Here is the error logs:

Start Test 8/22/2013 4:38:31.843 PM PCI\VEN_1B36&DEV_0100&SUBSYS_11001AF4&REV_04\3&13C0B0C5&0&10 
Message 8/22/2013 4:38:31.843 PM Name: Red Hat QXL GPU 
Message 8/22/2013 4:38:31.843 PM Device Description: Red Hat QXL GPU 
Message 8/22/2013 4:38:31.843 PM Problem Code: 0 
Message 8/22/2013 4:38:31.843 PM Installed Driver File: c:\windows\inf\oem2.inf in catalog C:\Windows\system32\CatRoot\{F750E6C3-38EE-11D1-85E5-00C04FC295EE}\oem2.CAT. 
Message 8/22/2013 4:38:31.843 PM Certification Submission ID: 1598591 
Message 8/22/2013 4:38:31.843 PM Checking: "DriverTrustLevel". Has value: "Trusted: Certified". 
Message 8/22/2013 4:38:31.843 PM Expected Certification Level: 1000 4000 5000 , Actual Certification Level: 5000 
Message 8/22/2013 4:38:31.843 PM Expected OS Attribute: Server2008R2X64, Actual OS Attribute: XPX86,7X64 
Error 8/22/2013 4:38:31.843 PM Problem: "Not certified for this version of Windows": This file, which is part of a driver package, is certified, but for a different operating system..
Solution: Contact the driver vendor for the correctly signed driver package for the installed operating system and install that driver package..
 
File:    Line: 0 
Error Type:    
Error Code:   0x0 
Error Text:   Error 0x00000000 
Message 8/22/2013 4:38:31.843 PM Installed Driver File: c:\windows\system32\drivers\qxl.sys in catalog C:\Windows\system32\CatRoot\{F750E6C3-38EE-11D1-85E5-00C04FC295EE}\oem2.CAT. 
Message 8/22/2013 4:38:31.843 PM Certification Submission ID: 1598591 
Message 8/22/2013 4:38:31.843 PM Checking: "DriverTrustLevel". Has value: "Trusted: Certified". 
Message 8/22/2013 4:38:31.843 PM Expected Certification Level: 1000 4000 5000 , Actual Certification Level: 5000 
Message 8/22/2013 4:38:31.843 PM Expected OS Attribute: Server2008R2X64, Actual OS Attribute: XPX86,7X64 
Error 8/22/2013 4:38:31.843 PM Problem: "Not certified for this version of Windows": This file, which is part of a driver package, is certified, but for a different operating system..
Solution: Contact the driver vendor for the correctly signed driver package for the installed operating system and install that driver package..
 
File:    Line: 0 
Error Type:    
Error Code:   0x0 
Error Text:   Error 0x00000000 
Message 8/22/2013 4:38:31.843 PM Installed Driver File: c:\windows\system32\qxldd.dll in catalog C:\Windows\system32\CatRoot\{F750E6C3-38EE-11D1-85E5-00C04FC295EE}\oem2.CAT. 
Message 8/22/2013 4:38:31.843 PM Certification Submission ID: 1598591 
Message 8/22/2013 4:38:31.843 PM Checking: "DriverTrustLevel". Has value: "Trusted: Certified". 
Message 8/22/2013 4:38:31.843 PM Expected Certification Level: 1000 4000 5000 , Actual Certification Level: 5000 
Message 8/22/2013 4:38:31.843 PM Expected OS Attribute: Server2008R2X64, Actual OS Attribute: XPX86,7X64 
Error 8/22/2013 4:38:31.843 PM Problem: "Not certified for this version of Windows": This file, which is part of a driver package, is certified, but for a different operating system..
Solution: Contact the driver vendor for the correctly signed driver package for the installed operating system and install that driver package..
 
File:    Line: 0 
Error Type:    
Error Code:   0x0 
Error Text:   Error 0x00000000 
End Test 8/22/2013 4:38:31.843 PM PCI\VEN_1B36&DEV_0100&SUBSYS_11001AF4&REV_04\3&13C0B0C5&0&10 
Result:   Fail 
Repro:   driverassessment.exe /CheckLogo

Sorry, I'm all confused - there isn't any change required to the inf file for this bug as far as I can tell. I just took the new latest (qxl-win-0.1-19) driver and installed it with testsigning on on 2008R2SP1 and it works, it does say that the signature is invalid but that is clearly outlined above in comment 15 by Mike Cao as happening because the cat file from the WHQL process doesn't include 2008R2, but that's a process problem, which will be solved if we run the WHQL for 2008R2 as well - it can work with the existing drivers.

Where am I wrong? Bill? Asking you since you are doing the WHQL.

Alon

(In reply to Alon Levy from comment #16)
> Sorry, I'm all confused - there isn't any change required to the inf file
> for this bug as far as I can tell. I just took the new latest
> (qxl-win-0.1-19) driver and installed it with testsigning on on 2008R2SP1
> and it works, it does say that the signature is invalid but that is clearly
> outlined above in comment 15 by Mike Cao as happening because the cat file
> from the WHQL process doesn't include 2008R2, but that's a process problem,
> which will be solved if we run the WHQL for 2008R2 as well - it can work
> with the existing drivers.
> 
> Where am I wrong? Bill? Asking you since you are doing the WHQL.
> 
> Alon

We are getting into loop we need to break, Afaik we tried the whql with what we have in the zip under w7/amd64 folder and we got one fail - https://bugzilla.redhat.com/show_bug.cgi?id=1005962 We need to resolve this fail to get through WHQL (whatever the reason of the error is).
Any suggestions?

I have entered a bug that the "WHQL "PCI Hardware Compliance Test" fails on W2K8R2 server with Win7x64 QXL driver."

The "test129" is looking for a Win7 device and it should be looking for a W2K8R2 device.

(In reply to Marian Krcmarik from comment #18)
> (In reply to Alon Levy from comment #16)
> > Sorry, I'm all confused - there isn't any change required to the inf file
> > for this bug as far as I can tell. I just took the new latest
> > (qxl-win-0.1-19) driver and installed it with testsigning on on 2008R2SP1
> > and it works, it does say that the signature is invalid but that is clearly
> > outlined above in comment 15 by Mike Cao as happening because the cat file
> > from the WHQL process doesn't include 2008R2, but that's a process problem,
> > which will be solved if we run the WHQL for 2008R2 as well - it can work
> > with the existing drivers.
> > 
> > Where am I wrong? Bill? Asking you since you are doing the WHQL.
> > 
> > Alon
> 
> We are getting into loop we need to break, Afaik we tried the whql with what
> we have in the zip under w7/amd64 folder and we got one fail -
> https://bugzilla.redhat.com/show_bug.cgi?id=1005962 We need to resolve this
> fail to get through WHQL (whatever the reason of the error is).
> Any suggestions?

The error is in the xml file that is in bug 1005962.

(In reply to Alon Levy from comment #16)
> Sorry, I'm all confused - there isn't any change required to the inf file
> for this bug as far as I can tell. I just took the new latest
> (qxl-win-0.1-19) driver and installed it with testsigning on on 2008R2SP1
> and it works, it does say that the signature is invalid but that is clearly
> outlined above in comment 15 by Mike Cao as happening because the cat file


The cat file in w7/amd64 folder says that It's for OS type VistaX64 and 7X64, nothing about 2k8r2, isn't it the problem?
That's what we are waiting for, a build with correct inf, cat files or whatever is needed.

(In reply to Marian Krcmarik from comment #21)
> (In reply to Alon Levy from comment #16)
> > Sorry, I'm all confused - there isn't any change required to the inf file
> > for this bug as far as I can tell. I just took the new latest
> > (qxl-win-0.1-19) driver and installed it with testsigning on on 2008R2SP1
> > and it works, it does say that the signature is invalid but that is clearly
> > outlined above in comment 15 by Mike Cao as happening because the cat file
> 
> 
> The cat file in w7/amd64 folder says that It's for OS type VistaX64 and
> 7X64, nothing about 2k8r2, isn't it the problem?
> That's what we are waiting for, a build with correct inf, cat files or
> whatever is needed.

I've been looking for a magical inf incantation in msdn, no luck so far. So I tried to load the driver - it works, hence comment 16, and like Mike said in comment 17 I expected we will try WHQL on Win2k8r2sp1. And you're right, we've been here before, so I'll look at bug 1005962 and hope Bill can supply the breaking test in standalone form or I'll be forced to setup the whole WHQL infrastructure.

Regardless about cat: I don't make a cat, that's the signature that WHQL produces. So let's not have another loop ;) The inf I have no clue about still, hopefully this test will enlighten me.

Just to clarify: The cat file will say w2kr8 when the WHQL is succeful, not before, unless you want a self signed one.

Oops: The "Fixed-in" actually specifies a qxl-win package, not a virtio-win package. So I'm changing to POST until there is a virtio-win package.

qxl-win-0.1-21
https://brewweb.devel.redhat.com/taskinfo?taskID=6730736

Moving back to MODIFIED state, as this component is qxl-win.
If needed, another bug is to be created for virtio-win.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHEA-2014-0053.html