[OpenIndiana-discuss] Boot hangs after configuring, devices (last notice is of pseudo devices

[russell]

> On 24. Mar 2023, at 10:50, Udo Grabowski (IMK)<udo.grabowski at kit.edu>  wrote:
>
>
>
> On 24/03/2023 09:43, Marc Lobelle wrote:
>> On 3/24/23 08:16, Toomas Soome via openindiana-discuss wrote:
>>>> On 24. Mar 2023, at 07:58, Stephan Althaus<Stephan.Althaus at Duedinghausen.eu>  wrote:
>>>>
>>>> On 3/22/23 20:56, Richard Lowe wrote:
>>>>> The challenge of answering that is you often don't know the answer
>>>>> until you see it
>>>>>
>>>>> The ::stacks command will give you asummarized view of every kernel
>>>>> thread's stack
>>>>> the ::cpuinfo -v command will show you what the CPUs are doing now (or
>>>>> were doing, before you got into kmdb).
>>>>>
>>>>> Both or either might have a smoking gun of some kind, but it's hard to
>>>>> predict what it might look like.
>>>>>
>>>>> _______________________________________________
>>>>> openindiana-discuss mailing list
>>>>> openindiana-discuss at openindiana.org
>>>>> https://openindiana.org/mailman/listinfo/openindiana-discuss
>>>> Hello!
>>>>
>>>> In one of my earlier update-tests i had a problem with the first OI-nvidia-525 version in cooperation with Virtualbox,
>>>> so i had the idea to test this:
>>>>
>>>> I uninstalled 'userland-incorporation' and 'entire' as prerequisite to  install nvidia version 515,
>>>> now it boots!
>>>>
>>>> So, nvidia in the most recent version 525 stopped the boot process. (?)
>>>>
>>>> i checked with NVidia Solaris Driver Archive that my chipset (Quadro M1200) is on the supported list for this version:
>>>> https://www.nvidia.com/Download/driverResults.aspx/199661/en-us/
>>>>
>>>>
>>>> How/Why should the display driver be able to interrupt the boot process ?
>>>> How/Why should the display driver be able to interrupt the boot process without further notice?
>>>>
>>> That is something one can work out with debugger. But, this is binary only driver? Built *for Oracle Solaris 11.4*? With API interfaces changed inside Oracle Solaris? This means, the question is not ?if? but ?when? - it is only question of time when such drivers become unusable for us. Without gfx driver development, the desktop variant of illumos is dead end.
>> without desktop, it would unfortunately mean the dead of illumos. If Oracle solaris drivers become unusable shouldn't illumos move to another driver interface that is maintained ?
> Maybe we wait until the cause for the problem is known before
> declaring everything dead ?
>
It is a bit premature to declare everything dead, but, the fact is, we 
can not rely on binary only drivers forever. As soon as Oracle will 
decide to stop distributing Solaris, nvidia will have no reason to 
distribute those binary drivers. Note, we *still* do rely on few closed 
binaries to distribute illumos itself.

  rgds,

toomas


There was a feature request #8069 for AMDGPU support which is the latest 
open source AMD driver for AMD graphics cards, given how AMD has 
improved their CPUs and Graphics Cards under Lisa Su, would now be the 
time to consider adding better support for AMD cards.

Kind Regards


Russell