This document describes the policy, process and criteria for updating nVidia proprietary drivers in a stable supported distro, including LTS.
nVidia proprietary GPU drivers are broadly used by gamers or for GPU compute activities (AI/ML). Support for new GPUs as well as bug fixes are regularly released by the vendor (on average 2 driver updates a month). For an improved experience or simply to support new GPUs without installing drivers from unofficial sources (for example PPAs), it is essential to release the latest version of the drivers regularly to the stable releases of Ubuntu.
What sort of updates
There are 3 levels of supported drivers:
Long lived branch: This series is to provide critical or low risk fixes and other minor updates for non-legacy hardware without exposing any new functionalities. These releases are intended for users of non-legacy GPUs who don't necessarily need the latest and greatest features. Support for new GPUs can be backported to this series.
Short lived branch: As opposed to the long lived branch, this series provides introduces new features.
Legacy drivers: Legacy GPUs are older-generation nVidia GPUs which are no longer supported in the regular nVidia Unified UNIX Graphics Driver. Instead, these GPUs will continue to be supported through special "Legacy GPU" drivers that will be updated periodically to add support for new versions of Linux system components (e.g., new Linux kernels, new versions of the X server, etc). Support timeframe for legacy drivers is documented on the official nVidia site support page
The release targeted by the SRU are:
- Long lived branch,
- Short lived branch,
- Long lived branch,
Short lived branches are not SRUed to non-lts releases to align the policy on the HWE stacks however this decision can be revisited depending on the demand.
When a new major version of a driver is available and uploaded, the versioning and naming scheme of the packages is so that users won’t be automatically upgraded to the latest version. They will only have updates for minor releases of a driver.
For example if version 100.1 is in the LTS.
- Version 100.2 is uploaded to the LTS, users will be upgraded.
- Version 110.1 is uploaded to the LTS, a transitional package is created to upgrade the user to the latest version of the branch. Although we won’t upgrade a long lived branch to a short lived one. However we might upgrade users from a short lived branch to a long lived branch.
If several versions of the driver support the GPU, ubuntu-driver will expose them all.
Week 0: New version of a driver released by the vendor
Week 1: Packaging (porting patches against new kernel and tool chain)
Week 2: QA (smoke test, installs, boot, shell comes up)
Week 3: Upload to the development release
Week 2: Packaging for stable releases
Week 4: QA (Full test suite)
Week 6: Upload to stable releases (-proposed pocket)
Week 8: Promotion to the -updates pocket.
Requesting the SRU
The SRU should be done with a single process bug, instead of individual bug reports for individual bug fixes. The one bug should have the following:
The SRU should be requested per the StableReleaseUpdates documented process
- The template at the end of this document should be used and all ‘TODO’ filled out
- The change log will contain a reference to the single SRU process bug, not all bugs fixed by the SRU. However, if there are very important bugs that are deemed worthy of reference they too should be included in the change log.
- Major changes should be called out in the SRU template, especially where changed behavior is not backward compatible.
- For each release that is proposed to be updated by the SRU a link to the results of the automated tests so that anyone can verify that they have been executed successfully.
- Additionally, the SRU bug should be verbose in documenting any manual testing that occurred.
- Any architecture specific fixes need to be noted and architecture specific test results included.
- Any packaging changes (e.g. a dependency change) need to be stated
The objective of the separate packaging QA is to test:
- Package installation from scratch
- Package upgrades
- Distribution upgrade
The resulting package, with all the changes in place, must undergo and pass the following additional QA procedures:
Installation from scratch in the current distribution by the recommended way to install Nvidia driver:
- If the above recommended way failed, please comment on the SRU bug then install the driver by using:
- Upgrade test from previous version of the package. This test must be performed with:
apt-get <install or upgrade>
- Upgrade test from previous distribution to the current one. If the current distribution is an LTS one, the upgrade path from the previous LTS distribution must also be exercised.
Certification test suite must pass on a range of hardware.
Call for testing is sent to the community via the community hub as soon as the drivers are available in the staging PPA.
[Impact] This release provides both bug fixes and new features and we would like to make sure all of our users have access to these improvements. The notable ones are: * <TODO: Create list with LP: # included > See the changelog entry below for a full list of changes and bugs. [Test Case] The following development and SRU process was followed: https://wiki.ubuntu.com/NVidiaUpdates <TODO Document any QA done, automated and manual> The QA team that executed the tests will be in charge of attaching the artifacts and console output of the appropriate run to the bug. nVidia maintainers team members will not mark ‘verification-done’ until this has happened. [Regression Potential] In order to mitigate the regression potential, the results of the aforementioned system level tests are attached to this bug. <TODO: attach nvidia-proposed test artifacts for every SRU release, not a link as links expire> [Discussion] <TODO: other background> <TODO: Paste in change log entry from nVidia for this version of the driver>
If an nVidia driver is updated then all nVidia user space components will stop working immediately after the respective package updates as the loaded kernel module and the user space components have a version mismatch. The consequences are not immediately visible to the user as nVidia components in memory are still properly matched and hence still work. The real issue is with new processes as for an instance no OpenGL applications or CUDA workloads can be launched anymore.
The way to fix this is to reboot immediately after an nVidia driver has been updated, including for minor version updates.
This particular issue is discussed in Debian Bug 889669