[solved] inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Questions about applications and software
Forum rules
Before you post please read how to get help
User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

[solved] inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sat Jun 30, 2018 5:32 am

Hello, folks.

Preface:
Running this:

Code: Select all

$ inxi -Sxx
System:    Host: voyager Kernel: 4.4.0-128-generic x86_64 bits: 64 compiler: gcc v: 5.4.0 
           Desktop: Cinnamon 3.2.7 dm: mdm Distro: Linux Mint 18.1 Serena base: Ubuntu 16.04 LTS Xenial
Having no issues with my nouveau driver and Cinnamon.

inxi 2.3.56 on graphics vs inxi 3.0.14 on graphics:

All the time, the old inxi 2.3.56 had been telling me this about my graphics card and driver:

Code: Select all

$ ./inxi -Fxz
[...]
Graphics:  Card: NVIDIA G86 [GeForce 8500 GT] bus-ID: 01:00.0
           Display Server: X.Org 1.18.4 drivers: nouveau (unloaded: modesetting,fbdev,vesa)
           Resolution: 1280x1024@60.02hz
           OpenGL: renderer: NV86 version: 3.3 Mesa 17.2.8 Direct Render: Yes
Which reads correct to me. nouveau is loaded and in use. This is why Cinnamon works correctly. No issues.

Now, inxi 3.0.14 comes along and tells me, however:

Code: Select all

$ inxi -Fxz
[...]
Graphics:  Card-1: NVIDIA G86 [GeForce 8500 GT] driver: nouveau v: kernel bus ID: 01:00.0 
           Display: server: X.Org 1.18.4 driver: none unloaded: fbdev,modesetting,nouveau,nvidia,vesa 
           resolution: 1280x1024~60Hz 
           OpenGL: renderer: NV86 v: 3.3 Mesa 17.2.8 direct render: Yes 
How can nouveau be unloaded? How would Cinnamon work, in case inxi 3.0.14 were right?

Summary question:

How comes that inxi 3.0.14 tells "nouveau" were "unloaded", although inxi 2.3.56 says "loaded"? And although Cinnamon is working correctly, which it would not do in case all drivers were unloaded.

Best regards,
Karl
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

gm10
Level 17
Level 17
Posts: 7080
Joined: Thu Jun 21, 2018 5:11 pm

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by gm10 » Sat Jun 30, 2018 7:13 am

inxi is parsing the xorg log file to try and find the driver in use.

Use this instead:

Code: Select all

xrandr --listproviders
Tune up your LM 19.x: ppa:gm10/linuxmint-tools

User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sat Jun 30, 2018 8:14 am

Hello, gm10.

Thanks for the reply and for the xandr commandline.

Code: Select all

$ xrandr --listproviders
Providers: number : 1
Provider 0: id: 0x63 cap: 0x1, Source Output crtcs: 2 outputs: 2 associated providers: 0 name:nouveau
I admit that this does not really appear any clearer to me than the inxi output.
On the contrary, the inxi output seems pretty straight forward.

What I am really after is the answer to the question, which inxi is right?
inxi 2.3.56, which says nouveau is loaded; or inxi 3.0.14, which says nouveau is unloaded?
As inxi 3.014 and inxi 2.3.56 are both present on my system currently, they read and parse exactly the same file /var/log/Xorg.0.log.
The xrandr output does not really help me answer my own question.

Regards,
Karl
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

gm10
Level 17
Level 17
Posts: 7080
Joined: Thu Jun 21, 2018 5:11 pm

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by gm10 » Sat Jun 30, 2018 8:15 am

karlchen wrote:
Sat Jun 30, 2018 8:14 am
The xrandr output does not really help me answer my own question.
Sure it does, it tells you that nouveau is the driver in use. So new inxi got it wrong, old one is correct.
Last edited by gm10 on Sat Jun 30, 2018 8:23 am, edited 2 times in total.
Tune up your LM 19.x: ppa:gm10/linuxmint-tools

User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sat Jun 30, 2018 8:23 am

It was this "provider 0" and "associated providers: 0" in the xandr output which managed to confuse me.
Should have remembered that it is an old weird programmer's habit to start counting at 0.
So the first and only provider is #0, not #1.

Guess this means I should file a bug on the inxi page, unless it has already been filed.
Or maybe I should have appended my initial post in that thread, started by h2-1.
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

gm10
Level 17
Level 17
Posts: 7080
Joined: Thu Jun 21, 2018 5:11 pm

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by gm10 » Sat Jun 30, 2018 8:25 am

karlchen wrote:
Sat Jun 30, 2018 8:23 am
Should have remembered that it is an old weird programmer's habit to start counting at 0.
Proof right there that Linux has gone mainstream. :lol:

PS: And yes, I'm sure h2-1 would be interested in your Xorg log.
Tune up your LM 19.x: ppa:gm10/linuxmint-tools

User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sat Jun 30, 2018 8:32 am

gm10 wrote:
Sat Jun 30, 2018 8:25 am
Proof right there that Linux has gone mainstream.
Or merely proves that I have not written a single line of C code for more than 15 years. :wink:
gm10 wrote:
Sat Jun 30, 2018 8:25 am
I'm sure h2-1 would be interested in your Xorg log.
We'll see. Providing it to him should not be an unsolvable technical challenge. :)
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

User avatar
Flemur
Level 16
Level 16
Posts: 6850
Joined: Mon Aug 20, 2012 9:41 pm
Location: Potemkin Village

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by Flemur » Sat Jun 30, 2018 12:01 pm

FWIW, I get different interesting results.
nVidia GS8400 card using, as far as I know, the nouveau driver (definitely NOT the nvidia driver).
Mint 18.3 xfce.

Code: Select all

$ xrandr --listproviders
Providers: number : 1
Provider 0: id: 0x43 cap: 0x9, Source Output, Sink Offload crtcs: 2 outputs: 2 associated providers: 0 name:modesetting
"name:modesetting" ?? "modesetting" is the name of
- the driver ?
- the "Provider" ?
- what has been provided ? (a setting for a mode)

Code: Select all

$ inxi --version
inxi 2.2.35-00 (2016-02-29)

Code: Select all

$ inxi -Fxz
...
Graphics:  Card: NVIDIA G98 [GeForce 8400 GS Rev. 2] bus-ID: 02:00.0
           Display Server: X.Org 1.18.4 drivers: (unloaded: fbdev,vesa) Resolution: 1680x1050@59.95hz
           GLX Renderer: NV98 GLX Version: 3.0 Mesa 17.2.8 Direct Rendering: Yes
My conclusion based on the output:
- No driver loaded,
- something, or perhaps a "Provider" of something, that's apparently not a driver is called "modesetting"
- "fbdev,vesa" were loaded and then unloaded ("unloaded" is not the same as "never loaded").

Even though "nouveau" is all over the place in Xorg.0.log, kern.log, and syslog.1 (but not syslog?)
Please edit your original post title to include [SOLVED] if/when it is solved!
Your data and OS are backed up....right?
Mint 19.1 Xfce/fluxbox
Manjaro openbox/fluxbox

gm10
Level 17
Level 17
Posts: 7080
Joined: Thu Jun 21, 2018 5:11 pm

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by gm10 » Sat Jun 30, 2018 1:50 pm

Flemur wrote:
Sat Jun 30, 2018 12:01 pm
"name:modesetting" ?? "modesetting" is the name of
- the driver ?
- the "Provider" ?
- what has been provided ? (a setting for a mode)

Code: Select all

man modesetting
Tune up your LM 19.x: ppa:gm10/linuxmint-tools

h2-1
Level 3
Level 3
Posts: 193
Joined: Sat Oct 16, 2010 4:02 pm
Contact:

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by h2-1 » Sat Jun 30, 2018 2:11 pm

You are correct, this question can only be answered by seeing the relevant /var/log/Xorg.0.log file.

However, with this said, a new inxi feature is to show per card drivers, where you see nouveau clearly listed as the active driver for that card. That comes from another source, which is fairly reliable I believe.

modesetting is a 'driver' technically speaking, just fyi. Many intel cards show modesetting for example as the primary driver, but it varies.

I found some fairly clear bugs in Xorg.0.log data during the inxi 3.0 testing, which I debated trying to follow through with by doing an xorg bug report to xorg, because it's quite common to see a driver get loaded, then unloaded, in the log file, which then trips the unloaded item for that driver, but I can't say more without seeing the entire log, use pastebin or something like it where I can get the real raw log file, don't copy or paste it because that can damage critical white space stuff.

The odds however are that actually 2.3.56 was wrong, and 3.0.14 is right, in terms of what the raw data shows, but I can't say for sure without seeing the raw data. Sometimes it's a slightly different syntax used.

User avatar
Flemur
Level 16
Level 16
Posts: 6850
Joined: Mon Aug 20, 2012 9:41 pm
Location: Potemkin Village

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by Flemur » Sat Jun 30, 2018 2:29 pm

gm10 wrote:
Sat Jun 30, 2018 1:50 pm

Code: Select all

man modesetting
Thanks!
I installed xserver-xorg-video-nouveau and now "inxi -Fxz" lists drivers: nouveau instead of drivers: <nothing>.
Please edit your original post title to include [SOLVED] if/when it is solved!
Your data and OS are backed up....right?
Mint 19.1 Xfce/fluxbox
Manjaro openbox/fluxbox

User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sat Jun 30, 2018 2:52 pm

Hello, h2-1.
h2-1 wrote:
Sat Jun 30, 2018 2:11 pm
You are correct, this question can only be answered by seeing the relevant /var/log/Xorg.0.log file.
As I started this thread, should I upload my logfile /var/log/Xorg.0.log? To termbin.com perhaps or to a different location?

Regards,
Karl
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

h2-1
Level 3
Level 3
Posts: 193
Joined: Sat Oct 16, 2010 4:02 pm
Contact:

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by h2-1 » Sat Jun 30, 2018 4:46 pm

yes, that would be helpful. Make sure to use the 'file upload' feature, not copy and paste because that will then preserve all critical whitespace etc.

User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sat Jun 30, 2018 5:07 pm

Hello, h2-1.

Please, find attached the unmodified logfile Xorg.0.log, zipped, so that it could be uploaded and added as an attachment here.
In case anything else is needed, please, let me know.

Regards,
Karl
Voyager_Serena_Xorg.0.log.zip
Host voyager, Mint 18.1 64-bit Serena, logfile /var/log/Xorg.0.log
(7.5 KiB) Downloaded 7 times
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

h2-1
Level 3
Level 3
Posts: 193
Joined: Sat Oct 16, 2010 4:02 pm
Contact:

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by h2-1 » Sat Jun 30, 2018 7:08 pm

thank you, that actually explains the issue, what is happening in some, but not all, Xorg.0.log cases, is that the Failed/unloaded are not being caused by failure per se, but because the module is already loaded.

As a rough attempt, I'm trying a new filter, when the Failed line includes:

Code: Select all

Failed to load module "nouveau" (already loaded, 0)
and the value was already set, I'm changing it to loaded, and changing the unloaded line to loaded.

this should take care of the middle block of loads when the modules are already loaded, basically I simply reset the value to loaded instead of failed or unloaded, if it has the already loaded error message for Failed, then at the very end, you'll notice they actually really do unload, which will then correctly set vesa, modesetting, and fbdev to unloaded, again.

But I believe this is a bug in Xorg or something around xorg, because it's not supposed to be doing this:

Code: Select all

cat Xorg.0-voyager-serena.log | grep -E '(Failed|Unload|Loading)'
[    49.796] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[    50.518] (II) UnloadModule: "nvidia"
[    50.518] (II) Unloading nvidia
[    50.518] (EE) Failed to load module "nvidia" (module does not exist, 0)
[    50.518] (II) Loading /usr/lib/xorg/modules/drivers/nouveau_drv.so
[    50.606] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[    50.606] (II) Loading /usr/lib/xorg/modules/drivers/fbdev_drv.so
[    50.606] (II) Loading /usr/lib/xorg/modules/drivers/vesa_drv.so
[    50.607] (II) UnloadModule: "nvidia"
[    50.607] (II) Unloading nvidia
[    50.607] (EE) Failed to load module "nvidia" (module does not exist, 0)
[    50.607] (II) Loading /usr/lib/xorg/modules/drivers/nouveau_drv.so
[    50.607] (II) UnloadModule: "nouveau"
[    50.607] (II) Unloading nouveau
[    50.607] (II) Failed to load module "nouveau" (already loaded, 0)
[    50.607] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[    50.607] (II) UnloadModule: "modesetting"
[    50.607] (II) Unloading modesetting
[    50.607] (II) Failed to load module "modesetting" (already loaded, 0)
[    50.608] (II) Loading /usr/lib/xorg/modules/drivers/fbdev_drv.so
[    50.608] (II) UnloadModule: "fbdev"
[    50.608] (II) Unloading fbdev
[    50.608] (II) Failed to load module "fbdev" (already loaded, 0)
[    50.608] (II) Loading /usr/lib/xorg/modules/drivers/vesa_drv.so
[    50.608] (II) UnloadModule: "vesa"
[    50.608] (II) Unloading vesa
[    50.608] (II) Failed to load module "vesa" (already loaded, 0)
[    50.610] (II) Loading sub module "fbdevhw"
[    50.610] (II) Loading /usr/lib/xorg/modules/libfbdevhw.so
[    50.610] (II) Loading sub module "dri2"
[    50.677] (II) Loading sub module "fb"
[    50.677] (II) Loading /usr/lib/xorg/modules/libfb.so
[    50.693] (II) Loading sub module "shadowfb"
[    50.694] (II) Loading /usr/lib/xorg/modules/libshadowfb.so
[    50.694] (II) UnloadModule: "modesetting"
[    50.694] (II) Unloading modesetting
[    50.694] (II) UnloadModule: "fbdev"
[    50.694] (II) Unloading fbdev
[    50.694] (II) UnloadSubModule: "fbdevhw"
[    50.694] (II) Unloading fbdevhw
[    50.694] (II) UnloadModule: "vesa"
[    50.694] (II) Unloading vesa
[    50.775] (II) Loading sub module "exa"
[    50.775] (II) Loading /usr/lib/xorg/modules/libexa.so
[    53.441] (II) Loading /usr/lib/xorg/modules/input/evdev_drv.so
[ 18243.064] (II) UnloadModule: "evdev"
[ 18243.064] (II) UnloadModule: "evdev"
[ 18243.064] (II) UnloadModule: "evdev"
[ 18243.064] (II) UnloadModule: "evdev"
but rather something more like this:

Code: Select all

cat /var/log/Xorg.0.log | grep -E '(Failed|Unload|Loading)'
[     9.112] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[     9.122] (II) Loading /usr/lib/xorg/modules/drivers/radeon_drv.so
[     9.124] (II) Loading /usr/lib/xorg/modules/drivers/ati_drv.so
[     9.125] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[     9.126] (II) Loading /usr/lib/xorg/modules/drivers/fbdev_drv.so
[     9.126] (II) Loading /usr/lib/xorg/modules/drivers/vesa_drv.so
[     9.137] (II) Loading sub module "fbdevhw"
[     9.138] (II) Loading /usr/lib/xorg/modules/libfbdevhw.so
[     9.139] (II) Loading sub module "fb"
[     9.139] (II) Loading /usr/lib/xorg/modules/libfb.so
[     9.140] (II) Loading sub module "dri2"
[     9.140] (II) Loading sub module "glamoregl"
[     9.140] (II) Loading /usr/lib/xorg/modules/libglamoregl.so
[     9.516] (II) Loading sub module "ramdac"
[     9.516] (II) UnloadModule: "modesetting"
[     9.516] (II) Unloading modesetting
[     9.516] (II) UnloadModule: "fbdev"
[     9.516] (II) Unloading fbdev
[     9.516] (II) UnloadSubModule: "fbdevhw"
[     9.516] (II) Unloading fbdevhw
[     9.516] (II) UnloadModule: "vesa"
[     9.516] (II) Unloading vesa
[     9.684] (II) Loading /usr/lib/xorg/modules/input/libinput_drv.so
[  2015.550] (II) UnloadModule: "libinput"
[  2015.550] (II) UnloadModule: "libinput"
[  2015.550] (II) UnloadModule: "libinput"
[  2015.550] (II) UnloadModule: "libinput"
[  2015.550] (II) UnloadModule: "libinput"
[  2015.550] (II) UnloadModule: "libinput"
[  2015.550] (II) UnloadModule: "libinput"
As you can see with this shortened format of output, it loads, then correctly unloads the modules, without then trying to load them a second time, yielding the resultant 'unloaded,failed', status.

This latter case yields the correct data:

Code: Select all

pinxi -Gxxxy80
Graphics:
  Card-1: ASPEED Graphics Family driver: N/A bus ID: 12:00.0 
  chip ID: 1a03:2000 
  Card-2: AMD Cape Verde PRO [Radeon HD 7750/8740 / R7 250E] driver: radeon 
  v: kernel bus ID: 21:00.0 chip ID: 1002:683f 
  Display: server: X.org 1.20.0 driver: ati,radeon 
  unloaded: fbdev,modesetting,vesa tty: 125x40 
  Message: Advanced graphics data unavailable in console. Try -G --display
Your case, if I correct the Xorg bug, will now yield this:

Code: Select all

driver: nouveau unloaded: modesetting,fbdev,vesa
I'm pretty sure this is an Xorg bug, and it may have been fixed in newer Xorgs, as you can see, the version that works is 1.20

As with all cases where you're dealing with another tool's bugs, the results are going to be a touch off but maybe slightly better than not fixing it to work around the bug, since in fact, nouveau, modesetting, vesa, and fbdev were all loaded successfully, then Xorg pretended to reload them again, but that resulted in a Failed message because it's already been loaded.

This patch is in pinxi 3.0.14-08

Since this is a fairly pressing fix, I'm probably going to do an inxi 3.0.15 quite soon.

User avatar
karlchen
Level 20
Level 20
Posts: 10674
Joined: Sat Dec 31, 2011 7:21 am
Location: Germany

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by karlchen » Sun Jul 01, 2018 4:54 am

Hello, h2-1.

Thank your very much for looking into this little issue so quickly.
Indeed, when going through the Xorg.0.log file, I had noticed as well that something tried to load nouveau a second time, although it had been loaded already.

Up to now I have not spotted the reported behaviour on my little Acer Aspire One netbook, which does not have an old nVidia card or chip, but an equally old integrated Intel graphics chip. Here inxi 3.0.14 reports: driver: i915, and loaded: Intel.

Best regards,
Karl
Image
Linux Mint 18.1 64-bit Cinnamon Desktop, Total Commander 9.22a 64-bit
Ubuntu 18.04.2 32-bit Mate Desktop, Total Commander 9.22a 32-bit
Windows? - 1 window in every room

gm10
Level 17
Level 17
Posts: 7080
Joined: Thu Jun 21, 2018 5:11 pm

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by gm10 » Sun Jul 01, 2018 5:41 am

h2-1 wrote:
Sat Jun 30, 2018 7:08 pm
thank you, that actually explains the issue, what is happening in some, but not all, Xorg.0.log cases, is that the Failed/unloaded are not being caused by failure per se, but because the module is already loaded.
Actually they are being caused by a failure, namely this one:

Code: Select all

[    50.518] (EE) Failed to load module "nvidia" (module does not exist, 0)
You always get those multiple loads when a driver that was autodetected

Code: Select all

[    50.607] (==) Matched nvidia as autoconfigured driver 02
could not be found. It then eventually fails over and starts anew:

Code: Select all

[    50.607] (==) Matched nvidia as autoconfigured driver 0
[    50.607] (==) Matched nouveau as autoconfigured driver 1
[    50.607] (==) Matched nvidia as autoconfigured driver 2
[    50.607] (==) Matched nouveau as autoconfigured driver 3
It's possibly a bug or just the lazy way to handle this, either way it's a bit ugly for you. You might have to check further down the log which driver is actually providing the output. Maybe run something like

Code: Select all

cat /var/log/Xorg.0.log | grep -i ': Output'
over your collection of data, my guess is it'll always provide the correct driver only.
Tune up your LM 19.x: ppa:gm10/linuxmint-tools

h2-1
Level 3
Level 3
Posts: 193
Joined: Sat Oct 16, 2010 4:02 pm
Contact:

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by h2-1 » Sun Jul 01, 2018 2:10 pm

gm10, it may appear that is the case, but it's not, I just confirmed.

What happens in both the good and bad case is that the set of to be autoconfigured modules is loaded if they exist.

this trips the inxi xorg driver per driver. drivers that do not exist are ignored.

Then, in the buggy instance, a SECOND instance of the autoloading block appears, and xorg AGAIN tries to load them, only this time, with a failure message for the drivers that have already been loaded. This reset driver status to first unloaded, then failed. Including the actually loaded ones.

In both cases, there is a closing block of unloads, where the unneeded drivers are unloaded, that is the same in both cases. In the bug case, this would reset the failed status to unloaded, and the actually loaded drivers would remain as failed since they are not unloaded in the end.

The fix here was relatively simple, simply check that for actually loaded drivers that give the 'already loaded' fail message, reset it to loaded, then the bottom set of correct unloads will reset the values for the drivers to be unloaded, and for the actually valid driver(s), they will correctly show as loaded. This is in pinxi latest now, and should work fine, since really all that inxi had to do to fix this bug was essentially ignore the middle second set of autoloads that tripped the already loaded message. The bug, to be clear, and this is quite obvious when you compare an xorg log from an xorg and system with this second set of autoload bug against an xorg log without it, is the second repeated set of autoloads, which then trip the failed message.

Since the last unloads happen in both cases at the end, all the 'failed' already loaded are switched to unloaded, and the active drivers remained at failed switch state, since they were not being reset to loaded.

This bug has been around a long time, I don't know if it's specific to a certain set of xorg versions, if it's a configuration thing that accidentally sets xorg to autoload twice, I never really found the pattern, but it is definitely a bug somewhere, but in this case, it proved to be a bug that is actually quite easy to work around and resolve without messing up anything else for inxi.
Last edited by h2-1 on Sun Jul 01, 2018 2:39 pm, edited 2 times in total.

gm10
Level 17
Level 17
Posts: 7080
Joined: Thu Jun 21, 2018 5:11 pm

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by gm10 » Sun Jul 01, 2018 2:38 pm

h2-1 wrote:
Sun Jul 01, 2018 2:10 pm
gm10, it may appear that is the case, but it's not, I just confirmed.
[...]
This is in pinxi latest now, and should work fine, since really all that inxi had to do to fix this bug was essentially ignore the middle second set of autoloads that tripped the already loaded message.
Well, I don't mind if I'm wrong but I can get a 100% repro of the issue that way, and latest pinxi (3.0.14-08) does not fix it.

Test case:
  1. Code: Select all

    apt remove xserver-xorg-video-fbdev
  2. restart X whichever way you prefer
  3. Code: Select all

    $ pinxi -G
    Graphics:  Card-1: Intel driver: i915 v: kernel 
               Display: x11 server: X.Org 1.19.6 driver: none unloaded: fbdev,modesetting,vesa
  4. Code: Select all

    $ cat /var/log/Xorg.0.log | grep -im 1 ': Output'
    [ 13792.006] (II) modeset(0): Output eDP-1 has no monitor section
    
Can get you the log but see if you can't repro yourself first.
Tune up your LM 19.x: ppa:gm10/linuxmint-tools

h2-1
Level 3
Level 3
Posts: 193
Joined: Sat Oct 16, 2010 4:02 pm
Contact:

Re: inxi 3.0.14 tells "nouveau" unloaded, inxi 2.3.56 says "loaded"

Post by h2-1 » Sun Jul 01, 2018 2:42 pm

yes, I'll need to see the full log, because this is a bug, and I may have only fixed one single instance of it. When stuff gets super specific, I tend to inject the actual user file or data directly into pinxi to confirm fixes or remaining issues.

If this bug had been easy or obvious to work around or fix, it would have been fixed a long time ago.

I'm confirming by using the direct data that the bug isn't fixed yet, lol. I should have tested it directly. Will need a few more fine tunings, but the cause is now understood so I can debug it. Should be done in 3.0.14.-09
Last edited by h2-1 on Sun Jul 01, 2018 2:48 pm, edited 1 time in total.

Post Reply

Return to “Software & Applications”