rtl8812au 4.3.8.12175.20140902+dfsg-0ubuntu2 ADT test failure with linux-hwe-edge 4.13.0-16.19~16.04.3
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
rtl8812au (Ubuntu) |
In Progress
|
Undecided
|
Marcelo Cerri | ||
Xenial |
Won't Fix
|
Undecided
|
Unassigned |
Bug Description
[Impact]
Currently the DKMS package fails to install in supported custom kernels that are based on 4.11 or 4.13. That includes the current 4.11 hwe-edge and the upcoming 4.13 hwe-edge kernels and some of the custom and cloud kernels as well.
[Test Case]
Install the DKMS package with the 4.13 hwe-edge kernel from -proposed. The package installation should proceed without any errors.
[Regression Potential]
Although new patches were added, the regression risk is very low since the new changes are conditionally compiled based on the kernel version.
Besides that the new package was tested with the following kernels in an amd64 environment:
- linux-generic 4.4
- linux-hwe 4.10
- linux-hwe-edge 4.11
- linux-hwe-edge 4.13 (from xenial-proposed)
- linux-azure 4.11
- linux-azure-edge 4.13 (which is in process of getting promoted to
xenial-proposed)
[Original description]
Testing failed on:
i386: https:/
tags: |
added: kernel-adt-failure-hwe-edge-4.13 removed: kernel-adt-failure |
description: | updated |
Changed in rtl8812au (Ubuntu): | |
status: | New → In Progress |
assignee: | nobody → Marcelo Cerri (mhcerri) |
> Although new patches were added, the regression risk is very low since the new changes are conditionally compiled based on the kernel version.
Is this statement still true for this upload? I see a bunch of changes that don't appear to be conditional on kernel version.
Please read and understand what the Regression Potential section is supposed to cover (documented in https:/ /wiki.ubuntu. com/StableRelea seUpdates# Procedure) as I think you have misunderstood it.
I appreciate your efforts in enabling this package for HWE users, particularly as this is the default in point release installers. But it's also important (perhaps more important) that we don't accidentally regress users already on this driver on a non-HWE kernel. How have you tested that behaviour for users not on an HWE kernel has not regressed? Did you use real hardware? Or did you just test that the build succeeded?