On Thu, 24 Aug 2023, Jani Nikula wrote:> On Thu, 24 Aug 2023, Lee Jones <lee at kernel.org> wrote: > > This set is part of a larger effort attempting to clean-up W=1 > > kernel builds, which are currently overwhelmingly riddled with > > niggly little warnings. > > The next question is, how do we keep it W=1 clean going forward?My plan was to fix them all, then move each warning to W=0. Arnd recently submitted a set doing just that for a bunch of them. https://lore.kernel.org/all/20230811140327.3754597-1-arnd at kernel.org/ I like to think a bunch of this is built on top of my previous efforts. GPU is a particularly tricky though - the warnings seem to come in faster than I can squash them. Maybe the maintainers can find a way to test new patches on merge?> Most people don't use W=1 because it's too noisy, so it's a bit of a > catch-22. > > In i915, we enable a lot of W=1 warnings using subdir-ccflags-y in our > Makefile. For CI/developer use we also enable kernel-doc warnings by > default. > > Should we start enabling some of those warning flags in drm/Makefile to > to keep the entire subsystem warning free?That would we awesome! We'd just need buy-in. -- Lee Jones [???]
On Thu, 24 Aug 2023, Lee Jones wrote:> On Thu, 24 Aug 2023, Jani Nikula wrote: > > > On Thu, 24 Aug 2023, Lee Jones <lee at kernel.org> wrote: > > > This set is part of a larger effort attempting to clean-up W=1 > > > kernel builds, which are currently overwhelmingly riddled with > > > niggly little warnings. > > > > The next question is, how do we keep it W=1 clean going forward? > > My plan was to fix them all, then move each warning to W=0.Some history: - Starting with v5.8-rc1: 18867 - 2020-07-01: 18089 - 2020-07-07: 17288 - 2020-07-17: 15762 - 2020-07-20: 15724 - 2020-07-23: 15116 - 2020-08-12: 15184 - 2020-10-19: 10909 - 2020-11-04: 9385 - 2021-01-04: 5478 - 2021-01-12 4749 - 2021-01-29 4911 - 2021-04-07 3594 - 2021-05-20 2938 - 2021-07-01 2587 - 2023-02-10 2587 - 2023-08-22 1650> Arnd recently submitted a set doing just that for a bunch of them. > > https://lore.kernel.org/all/20230811140327.3754597-1-arnd at kernel.org/ > > I like to think a bunch of this is built on top of my previous efforts. > > GPU is a particularly tricky though - the warnings seem to come in faster > than I can squash them. Maybe the maintainers can find a way to test > new patches on merge? > > > Most people don't use W=1 because it's too noisy, so it's a bit of a > > catch-22. > > > > In i915, we enable a lot of W=1 warnings using subdir-ccflags-y in our > > Makefile. For CI/developer use we also enable kernel-doc warnings by > > default. > > > > Should we start enabling some of those warning flags in drm/Makefile to > > to keep the entire subsystem warning free? > > That would we awesome! We'd just need buy-in.-- Lee Jones [???]
Hamza Mahfooz
2023-Aug-24 12:14 UTC
[Nouveau] [PATCH (set 1) 00/20] Rid W=1 warnings from GPU
On 8/24/23 08:07, Lee Jones wrote:> On Thu, 24 Aug 2023, Jani Nikula wrote: > >> On Thu, 24 Aug 2023, Lee Jones <lee at kernel.org> wrote: >>> This set is part of a larger effort attempting to clean-up W=1 >>> kernel builds, which are currently overwhelmingly riddled with >>> niggly little warnings. >> >> The next question is, how do we keep it W=1 clean going forward? > > My plan was to fix them all, then move each warning to W=0. > > Arnd recently submitted a set doing just that for a bunch of them. > > https://lore.kernel.org/all/20230811140327.3754597-1-arnd at kernel.org/ > > I like to think a bunch of this is built on top of my previous efforts. > > GPU is a particularly tricky though - the warnings seem to come in faster > than I can squash them. Maybe the maintainers can find a way to test > new patches on merge?I guess on that note, do you know if there is a way to run `scripts/kernel-doc` on patches instead of whole files? That would make much easier to block new kernel-doc issues from appearing.> >> Most people don't use W=1 because it's too noisy, so it's a bit of a >> catch-22. >> >> In i915, we enable a lot of W=1 warnings using subdir-ccflags-y in our >> Makefile. For CI/developer use we also enable kernel-doc warnings by >> default. >> >> Should we start enabling some of those warning flags in drm/Makefile to >> to keep the entire subsystem warning free? > > That would we awesome! We'd just need buy-in. >-- Hamza
On Thu, 24 Aug 2023, Hamza Mahfooz wrote:> > On 8/24/23 08:07, Lee Jones wrote: > > On Thu, 24 Aug 2023, Jani Nikula wrote: > > > > > On Thu, 24 Aug 2023, Lee Jones <lee at kernel.org> wrote: > > > > This set is part of a larger effort attempting to clean-up W=1 > > > > kernel builds, which are currently overwhelmingly riddled with > > > > niggly little warnings. > > > > > > The next question is, how do we keep it W=1 clean going forward? > > > > My plan was to fix them all, then move each warning to W=0. > > > > Arnd recently submitted a set doing just that for a bunch of them. > > > > https://lore.kernel.org/all/20230811140327.3754597-1-arnd at kernel.org/ > > > > I like to think a bunch of this is built on top of my previous efforts. > > > > GPU is a particularly tricky though - the warnings seem to come in faster > > than I can squash them. Maybe the maintainers can find a way to test > > new patches on merge? > > I guess on that note, do you know if there is a way to run > `scripts/kernel-doc` on patches instead of whole files? That would make > much easier to block new kernel-doc issues from appearing.Not off hand. When I run builds on patches I author, I run them twice concurrently. Once on the commit I'm basing on and once on the HEAD of my patchset. I then diff the two. So as long as the number of errors and warnings stay the same or reduce, we're golden. Perhaps the same method could be used with `kernel-doc`? -- Lee Jones [???]
Michel Dänzer
2023-Aug-28 16:11 UTC
[Nouveau] [PATCH (set 1) 00/20] Rid W=1 warnings from GPU
On 8/24/23 14:07, Lee Jones wrote:> On Thu, 24 Aug 2023, Jani Nikula wrote: >> On Thu, 24 Aug 2023, Lee Jones <lee at kernel.org> wrote: >>> This set is part of a larger effort attempting to clean-up W=1 >>> kernel builds, which are currently overwhelmingly riddled with >>> niggly little warnings. >> >> The next question is, how do we keep it W=1 clean going forward? > > My plan was to fix them all, then move each warning to W=0. > > Arnd recently submitted a set doing just that for a bunch of them. > > https://lore.kernel.org/all/20230811140327.3754597-1-arnd at kernel.org/ > > I like to think a bunch of this is built on top of my previous efforts. > > GPU is a particularly tricky though - the warnings seem to come in faster > than I can squash them. Maybe the maintainers can find a way to test > new patches on merge?One approach for this which has proved effective in Mesa and other projects is to make warnings fatal in CI which must pass for any changes to be merged. There is ongoing work toward introducing this for the DRM subsystem, using gitlab.freedesktop.org CI. -- Earthling Michel D?nzer | https://redhat.com Libre software enthusiast | Mesa and Xwayland developer