Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Hugo Sepulveda]

Greetings Luca,



Just had time today to test this again , sorry for the delay. I can
confirm the addition of the package into the repo's fixed the issue,
believing this bug report can thus be closed.
Thank you for your time and effort.
Kind regards,
Hugo


Op 01-06-17 om 12:09 schreef Luca Boccassi:
> On Thu, 2017-06-01 at 11:42 +0200, Hugo Sepulveda wrote:
>> Goodmorning  Luca,
>>
>>
>> Is it possible Andreas has uploaded it only to Stretch and sid? I've
>> yet
>> to see the update on jessie-backports.
>>
>> With kind regards,
>>
>>
>> Hugo.
> It's waiting in the queue:
>
> https://ftp-master.debian.org/backports-new.html
>
> Should be a matter of days, hopefully.
>
> Kind regards,
> Luca Boccassi
>
>> Op 31-05-17 om 11:25 schreef Luca Boccassi:
>>> On Mon, 2017-05-29 at 09:41 +0200, Hugo Sepulveda wrote:
 Op 28-05-17 om 14:18 schreef Luca Boccassi:
> On Fri, 2017-05-26 at 13:37 +0200, Hugo Sepulveda wrote:
>> I've isolated the problem, it seems the DisplayPort does not
>> get
>> initialized. Have no other OS to test this behaviour, not a
>> replacement
>> GPU card. However i really doubt the displayport just "died"
>> on
>> me,
>> neither is the error caused by hardware failure of the screen
>> itself.
>>
>> Admitted, it's a luxury problem (i cna still work with 2
>> display's
>> connected to the DVI ports), but it used to work fine with 3,
>> now
>> it's
>> all fubar, and X is tripping.
>>
>> Maybe that helps. If you need any additional info, i'm happy
>> to
>> provide it.
> If 375.66 fixes the problem, it might be just your lucky day!
>
> CVEs have been revealed and they are fixed by that version, so
> I
> have
> now uploaded it to unstable. Due to the security implications
> I've
> requested the release team to allow it to migrate to testing as
> well,
> and after that happens we will upload to jessie-backports too.
>
> It will take a few days to a couple of weeks to happen.
>
> Kind regards,
> Luca Boccassi
 That would be really wonderful! Great way to start of a monday.
 Thank you again for your efforts, I'll get back on this issue
 after
 the
 new version has been released on backports to give a status
 update.


 With kind regards,

 Hugo
>>> Andreas has just uploaded 375.66 to jessie-backports, it should be
>>> available in the repositories in a few hours.
>>>
>>> Kind regards,
>>> Luca Boccassi

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Luca Boccassi]

On Thu, 2017-06-01 at 11:42 +0200, Hugo Sepulveda wrote:
> Goodmorning  Luca,
> 
> 
> Is it possible Andreas has uploaded it only to Stretch and sid? I've
> yet
> to see the update on jessie-backports.
> 
> With kind regards,
> 
> 
> Hugo.

It's waiting in the queue:

https://ftp-master.debian.org/backports-new.html

Should be a matter of days, hopefully.

Kind regards,
Luca Boccassi

> Op 31-05-17 om 11:25 schreef Luca Boccassi:
> > On Mon, 2017-05-29 at 09:41 +0200, Hugo Sepulveda wrote:
> > > Op 28-05-17 om 14:18 schreef Luca Boccassi:
> > > > On Fri, 2017-05-26 at 13:37 +0200, Hugo Sepulveda wrote:
> > > > > I've isolated the problem, it seems the DisplayPort does not
> > > > > get
> > > > > initialized. Have no other OS to test this behaviour, not a
> > > > > replacement
> > > > > GPU card. However i really doubt the displayport just "died"
> > > > > on
> > > > > me,
> > > > > neither is the error caused by hardware failure of the screen
> > > > > itself.
> > > > > 
> > > > > Admitted, it's a luxury problem (i cna still work with 2
> > > > > display's
> > > > > connected to the DVI ports), but it used to work fine with 3,
> > > > > now
> > > > > it's
> > > > > all fubar, and X is tripping.
> > > > > 
> > > > > Maybe that helps. If you need any additional info, i'm happy
> > > > > to
> > > > > provide it.
> > > > 
> > > > If 375.66 fixes the problem, it might be just your lucky day!
> > > > 
> > > > CVEs have been revealed and they are fixed by that version, so
> > > > I
> > > > have
> > > > now uploaded it to unstable. Due to the security implications
> > > > I've
> > > > requested the release team to allow it to migrate to testing as
> > > > well,
> > > > and after that happens we will upload to jessie-backports too.
> > > > 
> > > > It will take a few days to a couple of weeks to happen.
> > > > 
> > > > Kind regards,
> > > > Luca Boccassi
> > > 
> > > That would be really wonderful! Great way to start of a monday.
> > > Thank you again for your efforts, I'll get back on this issue
> > > after
> > > the
> > > new version has been released on backports to give a status
> > > update.
> > > 
> > > 
> > > With kind regards,
> > > 
> > > Hugo
> > 
> > Andreas has just uploaded 375.66 to jessie-backports, it should be
> > available in the repositories in a few hours.
> > 
> > Kind regards,
> > Luca Boccassi
> 


signature.asc
Description: This is a digitally signed message part

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Hugo Sepulveda]

Op 28-05-17 om 14:18 schreef Luca Boccassi:
> On Fri, 2017-05-26 at 13:37 +0200, Hugo Sepulveda wrote:
>> I've isolated the problem, it seems the DisplayPort does not get
>> initialized. Have no other OS to test this behaviour, not a
>> replacement
>> GPU card. However i really doubt the displayport just "died" on me,
>> neither is the error caused by hardware failure of the screen itself.
>>
>> Admitted, it's a luxury problem (i cna still work with 2 display's
>> connected to the DVI ports), but it used to work fine with 3, now
>> it's
>> all fubar, and X is tripping.
>>
>> Maybe that helps. If you need any additional info, i'm happy to
>> provide it.
> If 375.66 fixes the problem, it might be just your lucky day!
>
> CVEs have been revealed and they are fixed by that version, so I have
> now uploaded it to unstable. Due to the security implications I've
> requested the release team to allow it to migrate to testing as well,
> and after that happens we will upload to jessie-backports too.
>
> It will take a few days to a couple of weeks to happen.
>
> Kind regards,
> Luca Boccassi
That would be really wonderful! Great way to start of a monday.
Thank you again for your efforts, I'll get back on this issue after the
new version has been released on backports to give a status update.


With kind regards,

Hugo

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Luca Boccassi]

On Fri, 2017-05-26 at 13:37 +0200, Hugo Sepulveda wrote:
> I've isolated the problem, it seems the DisplayPort does not get
> initialized. Have no other OS to test this behaviour, not a
> replacement
> GPU card. However i really doubt the displayport just "died" on me,
> neither is the error caused by hardware failure of the screen itself.
> 
> Admitted, it's a luxury problem (i cna still work with 2 display's
> connected to the DVI ports), but it used to work fine with 3, now
> it's
> all fubar, and X is tripping.
> 
> Maybe that helps. If you need any additional info, i'm happy to
> provide it.

If 375.66 fixes the problem, it might be just your lucky day!

CVEs have been revealed and they are fixed by that version, so I have
now uploaded it to unstable. Due to the security implications I've
requested the release team to allow it to migrate to testing as well,
and after that happens we will upload to jessie-backports too.

It will take a few days to a couple of weeks to happen.

Kind regards,
Luca Boccassi

signature.asc
Description: This is a digitally signed message part

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Hugo Sepulveda]

I've isolated the problem, it seems the DisplayPort does not get
initialized. Have no other OS to test this behaviour, not a replacement
GPU card. However i really doubt the displayport just "died" on me,
neither is the error caused by hardware failure of the screen itself.

Admitted, it's a luxury problem (i cna still work with 2 display's
connected to the DVI ports), but it used to work fine with 3, now it's
all fubar, and X is tripping.

Maybe that helps. If you need any additional info, i'm happy to provide it.

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Hugo Sepulveda]

On Thu, 25 May 2017 16:10:44 +0100 Luca Boccassi
 wrote:

> Control: severity -1 normal
>
> On Thu, 2017-05-25 at 09:18 +0200, Harald Dunkel wrote:
> > I am still running nvidia-graphics-drivers 375.26-1~bpo8+1 together
> > with the
> > new linux-image-4.9.0-0.bpo.3-amd64:amd64. This combination seems to
> > work
> > fine.
> >
> > Other interesting packages in this context:
> >
> > dkms 2.3-2~bpo8+1
> > nvidia-modprobe 358.09-1~bpo8+1
> > nvidia-settings 340.93-1~bpo8+1
> > nvidia-support 20151021+1~bpo8+1
> >
> >
> > Hope this helps
> > Harri
>
> Same here.
>
> And anyway a new version cannot be uploaded to backports until it's in
> stretch, and it cannot be uploaded to stretch as it's in pre-release
> freeze.
>
> It will be uploaded after Stretch ships.
>
> Kind regards,
> Luca Boccassi

Hello Luca, thanks for taking the time to answer to this bugreport.

I understand the version issue concenrning the pre-release of stretch.

First a bit of background, i have a Gainward GTX960 with 3 screens
connected, Two on the dvi ports and one on the displayport

Still , I'm having the same issue, even when removing the proprietary
driver from site (NVidia... . sh --uninstall), rebooting and installing
the nvidia-driver from backports. It really hangs X and it also gives me
the following new messages in the kernel. (attached to this message)

If i interpret the messages correctly, it is an issue with the following
components of nvidia:

>>>
May 26 09:35:01 Agnes kernel: [  155.454239] NVRM: RmInitAdapter failed!
(0x26:0x40:1097)
May 26 09:35:01 Agnes kernel: [  155.454291] NVRM: rm_init_adapter
failed for device bearing minor number 0
<<<

I've did some further troubleshooting, even cleaned out the hardware
itnernally (just for the case that dust would be the culprit)  then
i slowly disconnected all display's and started to connect them one by one.

Turns out the Displayport does not get properly initialized/detected.
I've excluded cable issues (replaced the displayport cable with a
brandnew one). When i click 'detect dispays' with the nvidia server
settings, the system seems to freeze whenever the displays are probed
for the displayport. They also report not being detected.

>>>
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): Samsung
SMS24A450 (CRT-0): connected
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): Samsung
SMS24A450 (CRT-0): 400.0 MHz maximum pixel clock
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0):
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-0:
disconnected
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-0:
Internal TMDS
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-0:
330.0 MHz maximum pixel clock
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0):
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-1:
disconnected
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-1:
Internal DisplayPort
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-1:
960.0 MHz maximum pixel clock
May 26 10:43:26 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0):
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-2:
disconnected
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-2:
Internal TMDS
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-2:
165.0 MHz maximum pixel clock
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0):
==> here it's where it hangs for 15 seconds or so, then continues
reporting disconnected, while everything is there.
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-3:
disconnected
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-3:
Internal TMDS
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): DFP-3:
330.0 MHz maximum pixel clock
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0):
<==
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): Samsung
SMS24A450 (DFP-4): connected
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): Samsung
SMS24A450 (DFP-4): Internal TMDS
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0): Samsung
SMS24A450 (DFP-4): 330.0 MHz maximum pixel clock
May 26 10:43:34 Agnes gdm-Xorg-:0[1399]: (--) NVIDIA(GPU-0):
<<<

My suspicion is that DisplayPort is the culprit here. this is also
something new in /var/log/messages when connecting the said displayport

>>>
May 26 11:03:57 Agnes kernel: [  339.437215] NVRM: GPU at
PCI::03:00: GPU-2e2f0b82-ed57-fa37-b7f2-a957266b69f0
May 26 11:04:00 Agnes kernel: [  342.439369] pcieport :00:02.0: AER:
Uncorrected (Non-Fatal) error received: id=0010
May 26 11:04:00 Agnes kernel: [  342.439402] pcieport :00:02.0: AER:
Device recovery failed
May 26 11:04:09 Agnes kernel: [  351.591094] NVRM: os_schedule:
Attempted to yield the CPU while in atomic or interrupt context
May 26 11:04:11 Agnes kernel: [  353.591132] NVRM: os_schedule:
Attempted to 

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[Luca Boccassi]

Control: severity -1 normal

On Thu, 2017-05-25 at 09:18 +0200, Harald Dunkel wrote:
> I am still running nvidia-graphics-drivers 375.26-1~bpo8+1 together
> with the
> new linux-image-4.9.0-0.bpo.3-amd64:amd64. This combination seems to
> work
> fine.
> 
> Other interesting packages in this context:
> 
>   dkms2.3-2~bpo8+1
>   nvidia-modprobe 358.09-1~bpo8+1
>   nvidia-settings 340.93-1~bpo8+1
>   nvidia-support  20151021+1~bpo8+1
> 
> 
> Hope this helps
> Harri

Same here.

And anyway a new version cannot be uploaded to backports until it's in
stretch, and it cannot be uploaded to stretch as it's in pre-release
freeze.

It will be uploaded after Stretch ships.

Kind regards,
Luca Boccassi

signature.asc
Description: This is a digitally signed message part

Bug#863276: nvidia-driver: nvidia fails to initialize Xorg with linux-image-4.9.0-0.bpo.3-amd64

[NeatNerdPrime]

Package: nvidia-driver
Version: 375.39-1~bpo8+1
Severity: grave
File: nvidia-driver
Tags: upstream
Justification: causes non-serious data loss

Dear Maintainer,

TL;DR: please update nvidia driver to version 375.66 in backports, so it works
with backports kernel available.

Elaborate explanation:

When updating today to the newest kernel release available on jessie backports
(linux-image-4.9.0-0.bpo.3-amd64:amd64 and it's dependencies), after reboot,
the system refuses to start X. This is in conjunction with the Nvidia driver ,
also available from jessie backports nvidia-driver:amd64 (375.39-1~bpo8+1).
Restarting with a prior kernel version proved ineffective (DKMS modules were
installed), this seemed to be caused by various kernel components being
replaced, what the older kernel is not suited for.

When inspecting the logfile /var/log/messages it shows these lines that cause
the X server to constantly re-initialize and fail, /var/log/xorg.0.log shows
nothing useful. Logfile shown below steps of reproduction and possible
resolution (attached with this bugreport):


   * What led up to the situation?
- Upgrading to newest kernel release available from jessie-backports
(linux-image-4.9.0-0.bpo.3-amd64:amd64)
- dependencies installed with this kernel update:
- linux-headers-4.9.0-0.bpo.3-amd64:amd64
- linux-headers-4.9.0-0.bpo.3-common:amd64
- dependencies updated with this upgrade:
- linux-kbuild-4.9:amd64 (4.9.18-1~bpo8+1, 4.9.25-1~bpo8+1),
- linux-image-amd64:amd64 (4.9+79~bpo8+1, 4.9+80~bpo8+1),
- linux-headers-amd64:amd64 (4.9+79~bpo8+1, 4.9+80~bpo8+1),
- linux-compiler-gcc-4.9-x86:amd64 (4.9.18-1~bpo8+1,
4.9.25-1~bpo8+1)
   * What exactly did you do (or not do) that was effective (or
 ineffective)?
- What was ineffective:
- booting from older kernels, DKMS modules ensured installed
- reinstalling/building modules with "sudo dkms install
module/version"
- reverting to prior packages, they are not available anymore
in backports
- Effective solution:
- remove all nvidia related packages from APT &&
- installing the nvidia driver from nvidia.com manually (.run
script from site): http://us.download.nvidia.com/XFree86/Linux-x86_64/375.66
/NVIDIA-Linux-x86_64-375.66.run

- Nvidia driver version that works: 375.66
   * What was the outcome of this action?
- After UNINSTALLING the nvidia software from APT and INSTALLING the
Nvidia driver from site script, the system started X again
   * What outcome did you expect instead?
- expected normal reboot after upgrading to newest kernel without
uninstalling nvidia software pacakges.


As for now it is unclear what really caused the breakage. Is it the kernel? Is
it the driver? My guts tell me it's the driver since it works with the newest
release from their site.

Additional information can be supplied on request.



-- System Information:
Debian Release: 8.8
  APT prefers stable-updates
  APT policy: (500, 'stable-updates'), (500, 'stable')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 4.9.0-0.bpo.3-amd64 (SMP w/8 CPU cores)
Locale: LANG=nl_BE.UTF-8, LC_CTYPE=nl_BE.UTF-8 (charmap=UTF-8)
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: X.Org X Server 1.16.4
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Release Date: 2014-12-20
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: X Protocol Version 11, Revision 0
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Build Operating System: Linux 
3.16.0-4-amd64 x86_64 Debian
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Current Operating System: Linux Agnes 
4.9.0-0.bpo.3-amd64 #1 SMP Debian 4.9.25-1~bpo8+1 (2017-05-19) x86_64
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Kernel command line: 
BOOT_IMAGE=/vmlinuz-4.9.0-0.bpo.3-amd64 root=/dev/mapper/agnes-root ro quiet
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Build Date: 11 February 2015  
12:32:02AM
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: xorg-server 2:1.16.4-1 
(http://www.debian.org/support)
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Current version of pixman: 0.32.6
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Before reporting problems, check 
http://wiki.x.org
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: to make sure that you have the latest 
version.
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: Markers: (--) probed, (**) from config 
file, (==) default setting,
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: (++) from command line, (!!) notice, 
(II) informational,
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: (WW) warning, (EE) error, (NI) not 
implemented, (??) unknown.
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: (==) Log file: "/var/log/Xorg.0.log", 
Time: Wed May 24 19:43:07 2017
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: (==) Using config file: 
"/etc/X11/xorg.conf"
May 24 19:43:07 Agnes gdm-Xorg-:0[1370]: (==) Using system config directory