Merge ~gary-wzl77/ubuntu/+source/nvidia-prime:fix_undefined_variable into ubuntu/+source/nvidia-prime:ubuntu/devel

[Gary.Wang]

fix(prime-select): fix the undefined variable in the _has_integrated_gpu function

Fix the undefined variable when detecting the integrated gpu.

[Sebastien Bacher (seb128)]

Thanks, a similar fix has been submitted upstream on https://github.com/tseliot/nvidia-prime/pull/25

[Robie Basak (racb)]

Looks like that pull request was submitted in November, with no response from upstream yet.

This merge proposal is still in the sponsorship queue. What do we need to do with it? Can it be closed - just effectively replaced with the upstream PR? Or does this need to land in an Ubuntu delta in the meantime?

What's the actual problem being fixed, please? How important is it?

[Gary.Wang (gary-wzl77)]

> Looks like that pull request was submitted in November, with no response from
> upstream yet.
>
> This merge proposal is still in the sponsorship queue. What do we need to do
> with it? Can it be closed - just effectively replaced with the upstream PR? Or
> does this need to land in an Ubuntu delta in the meantime?
A: I am not the maintainer of this package and not quite sure what's process/policy I need to go through to get this landed or just close it in favor of the upstream PR. I am fine with either way.
>
> What's the actual problem being fixed, please? How important is it?
A: My setup is a bit different than others. I have Ampere arm64 server[1] workstation + NVIDIA RTX A4500 GPU. After installing the nvidia-prime package and run

$ sudo prime-select nvidia
[sudo] password for gary:
Traceback (most recent call last):
  File "/usr/bin/prime-select", line 458, in <module>
    switcher.enable_profile(arg)
  File "/usr/bin/prime-select", line 223, in enable_profile
    if not self._has_integrated_gpu():
  File "/usr/bin/prime-select", line 198, in _has_integrated_gpu
    if t.find('8086') != -1 or t.lower().find('1002') != -1:
UnboundLocalError: local variable 't' referenced before assignment

To me, generally, this bug would affect the use case where the arm64 server + NVIDIA GPU both are in use.

[1] https://www.ipi.wiki/pages/comhpc-docs?page=index.html

[Dan Bungert (dbungert)]

Uploaded

diff --git a/prime-select b/prime-select
index 1d46ffe..572e17e 100755
--- a/prime-select
+++ b/prime-select
@@ -194,7 +194,7 @@ class Switcher(object):
         else:
             card = self._get_bootvga_card()
             if card:
-                vendor = self._get_card_vendor(card)
+                t = self._get_card_vendor(card)
                 if t.find('8086') != -1 or t.lower().find('1002') != -1:
                     status = True