Tracked Linux kernel regression
You likely landed here via a link in a message adding a Linux kernel issue to the list of tracked regression. Don't worry about the tracking performed using regzbot, it shouldn't bother you. There are just two hardly special things you want to keep in mind:
- If the person adding the issue to the tracking got anything wrong or missed something important, please reply to the mail that brought you here and set the public record straight.
- If you are a developer addressing the issue, remember to add 'Link:' tags pointing to any reports, as explained in more detail by the Linux kernel's documentation and the second to last section of this document.
Additionally, if you want help regression tracking and thus Linux kernel development, consider doing a small extra duty:
- If you need to set the public record straight, want to provide additional details about a tracked regression or change some of its properties, please tell regzbot directly about it.
The latter is quite easy: send a reply to the mail with the report or the mail that brought you here and include a paragraph with one or multiple of the following '#regzbot commands'; you in fact can use them in any direct or indirect reply to those mails. A few examples:
Mark a regression as resolved
Tell regzbot the commit-id of the change that already fixed this or is going to:
1#regzbot fixed-by: 1f2e3d4c5d
Stop tracking the regression
Reply with a short statement explaining why the regression doesn't need to be tracked anymore:
1#regzbot invalid: not a regression, it never worked
Write the statement like a subject of an email; if you want to provide more details, just do so in the mail where you use this command, regzbot will link to it.
Specify when the regressions started to happen
When the issue was added to the tracking, regzbot was told about a version range or commit when the regression started to happen. You can update this information easily by writing a reply with something like this:
1#regzbot introduced v5.15..v5.16-rc1
or
1#regzbot introduced 1f2e3d4c5d
Update the title stored by regzbot
Regzbot stores a title for every regression that was either specified manually or derived from the mails title. You can update it like this:
1#regzbot title: foo
Point regzbot to related discussions
In case the issue or a fix for it is already discussed somewhere, let regzbot know like this:
1#regzbot monitor: https://lore.kernel.org/all/30th.anniversary.repost@klaava.Helsinki.FI/
Reminder for developers: use 'Link:' tags
When fixing an issue someone reported, always add a 'Link:' tag pointing to the report in the patch description.
This is expected from kernel developers for reasons Linux' documentation explains in Documentation/process/submitting-patches.rst and Documentation/process/5.Posting.rst. In short: these links are often incredibly helpful, if you or someone else needs to look into the change and its backgrounds weeks, months, or years later. To ensure mailed reports are still available then, link to them in the kernel's official mailing list archives on lore.kernel.org whenever possible.
As developers are expected to place 'Link:' tags anyway, the regression tracking efforts started to rely on them. Regzbot in fact relies heavily on them, as they allow connecting reports with any patches posted or applied to fix the reported issue. That is the legwork of regression tracking and thus enables it without causing any additional work for Linux kernel developers.
More on Linux kernel regression tracking with rezbot
The current acting #Linux kernel's regression tracker, Thorsten Leemhuis, created and runs regzbot to make the regression tracking work feasible at a scale.
Above is just a very brief explanation of how regzbot operates and the commands it understands. For a more detailed elaboration of regzbot, see its getting started guide and its reference documentation.
See also: list of tracked Linux kernel regressions