Arkadiusz Hiler authored 6 years ago and Simona Vetter committed 6 years ago

So we have any clues in case of getting different results.

Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The migration broke dim setup :-/

Reported-by: Shawn Guo <shawnguo@kernel.org>
Cc: Shawn Guo <shawnguo@kernel.org>
Tested-by: Shawn Guo <shawnguo@kernel.org>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

I broke auto-update by applying a patch to the old tree. To fix this, we
need to merge it into Gitlab so ff works.

Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Acked-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>

Still pointed at ~airlied/linux.

Reported-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20180918152333.217692-1-sean@poorly.run

Of course I forgot the imporant part.

Note: Never push this, or anything after it, to the old place in
drm-tip/maintainer-tools.

Given our track record of forgetting a critical part in migrations I
think it'd be good if we hold off for a week or so until we apply this
one though. It's just a few wasted cpu cycles when running dim
update-branches.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Makes a lot easier to discover because unfortunately gitlab doesn't
have such a link built-in.

Cc: Jani Nikula <jani.nikula@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

sphinx-build and rst2man binaries on Fedora 28 have a -3 suffix when
they are installed for python3 in order to be able to be installed in
parallel to the python2's package.

For python3-sphinx, there's a bash module that plugs into /etc/profile.d
to add /usr/libexec/python3-sphinx to PATH. That however doesn't work
if you don't reload the profile after you installed the package.

So this makes it simpler by stating that we are compatible with any
of the tools (and we prefer the one with -3 suffix if available).
While at it, also allow people to override RST2MAN in the command line
in case we have even more creative scenarios.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

This gives us a nice

https://drm.pages.freedesktop.org/maintainer-tools



Which means we can simply change the old place to a web redirect, and
it'll all work out with the new location. Plus disable the
intel-gfx-ci job for maintainer-tools ofc.

v2: Update url in documentation.

Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Cc: Daniel Stone <daniels@collabora.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

- Use the https url so we don't require everyone to have their gitlab
  accounts ready already. Otherwise we'd need to gate migrating
  maintainer-tools on migrating all the drm kernel repos, and I'd
  really like to partition the migration. Also, we want to reduce
  maintainer-tools committers anyway, to shrink the attack surface a
  lot.

  Committers need to either set up the http access tokens, or set up
  ssh certificates and change their remote for the maintainer-tools
  repo.

- For testing, you can undo this auto-update using

$ git remote remove maintainer-tools ; git branch -u drm-tip/maintainer-tools

- My plan is that we push an immediate revert of this code to the
  gitlab repo (and only there) so it doesn't stick around.

- fd.o admins recommended that we don't do a read-only copy of
  maintaienr-tools at the old-place, since it's not a 1:1 match. For
  everything else we're going to migrate there will be a read-only
  copy with all urls working nicely, maintainer-tools is the only
  exception here.

v2: Don't forget about dim_setup.

v3: Update all the urls in docs!

v4: Also update the new maintainer-tools repo in dim_uptodate (Jani).

Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Daniel Stone <daniels@collabora.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com> (v2)
Reviewed-by: Daniel Stone <daniels@collabora.com> (v2)
Reviewed-by: Sean Paul <seanpaul@chromium.org> (v3)
Acked-by: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

This gives us neat little CI integration. No more "you have a
different version of shellcheck" - we just pick the one everyone can
run with docker. Also, no more "oops, forgot to run make check", that
is, if we adopt a merge request based flow.

Even without this this is useful, since if you do a fork and test
there, gitlab CI will run stuff for you. Example:

https://gitlab.freedesktop.org/danvet/maintainer-tools/blob/master/.gitlab-ci.yml



Observe the awesome green checkmark in the top-left corner!

v2: Use python3 (Lucas). And fix whitespace.

v3: Go back to python2, on fedora the python3 packages have a -3
suffix. Don't ask.

Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Again this is just to make things look neater on gitlab. For the
curious, I pushed this series to

https://gitlab.freedesktop.org/drm/maintainer-tools/tree/master



Long-term we might want o polish this more, especially in case we
decide to embrace gitlab merge requests and issues for
maintainer-tools. But this should be good enough for now.

Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

This way gitlab will automatically spot it and show it when people try
to do a merge request or file an issue.

Also shrink the title a bit, it looks terrible in the sidebar.

Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Since -fixes and -fixes-next (to a lesser extent) are rebasing trees in
drm-misc, add a dim rebase command that sanity checks the upstream and
adds SoB for the committer.

Changes in v2:
- s/validate_upstream_branch/validate_upstream_baseline/ (Daniel)
- Use check_conflicts instead of hand rolling (Daniel)

Cc: Boris Brezillon <boris.brezillon@bootlin.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20180828183100.189905-1-sean@poorly.run

Shashank wanted to reuse his drm-tip repo for DIM_REPO, which doesn't
work great. Catch this.

Also group the various check functions all together for a bit of OCD.

Cc: Shashank Sharma <shashank.sharma@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Some notes I took while ramping up Shashank on dim.

Aside: Requiring that DIM_REPO is already checked out seems to confuse
too. Maybe we should just fall back to initializing an empty git repo
- we'll all all the branches and remotes anyway.

Cc: Shashank Sharma <shashank.sharma@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Requested by Dave Airlie, so that we automatically record the pull
request submitter somewhere. After quick irc chat we decided to put it
at the bottom, to avoid confusion with the From: git format-patch
inserts at the top for authors not matching the patch submitter.

Requested-by: Dave Airlie <airlied@redhat.com>
Cc: Dave Airlie <airlied@redhat.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Tested-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

We are not writing anything after that message.

Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

Linus prefers that we backmerge a specific tag instead of a random
point in his branch. Allow that.

I guess it'd be nice to somehow figure out where a tag came from, but
git doesn't namespace tags. So that idea is out of the window
unfortunately.

v2: Unlazy and also update the docs.

v3: Make the check work from anywhere, we need to move the cd a bit
up (Dave).

Requested-by: Dave Airlie <airlied@redhat.com>
Cc: Dave Airlie <airlied@redhat.com>
Tested-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

As a rule of thumb, don't change patches while committing.

v2: Added note about conflicts (Daniel)

Cc: Imre Deak <imre.deak@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Lots has happened in the CI front since the first version was added.

v2: Add Piglit test requirement (Daniel)

Acked-by: Daniel Vetter <daniel@ffwll.ch>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Liviu Dudau authored 6 years ago and Simona Vetter committed 6 years ago

'dim status' is intended to be used by maintainers and not by commiters.
Move the description of the subcommand into the 'COMMANDS FOR MAINTAINERS'
area and clarify in the COMMITERS area that they should use vanilla
'git status' for checking the health of the branch.

v2: Fix duplicate 'status' warnings by merging the text into the 'COMMANDS
FOR MAINTAINERS' description.

Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The one case in dim_status is entirely unused.

The usage in dim_setup seems still in use, but this is code for
backwards compat hacks predating the original creating of the
maintainer-tools branch - i.e. no maintainer or committer except me
ever needed this.

There's also no need to make sure drm-tip has all the right remotes
nowadays, because that's all encoded in nightly.conf, and will be
checked (if the remotes are missing) in dim_rebuild_tip.

Reported-by: Daniel Stone <daniels@collabora.com>
Cc: Daniel Stone <daniels@collabora.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

... instead of hard-coding them. Will allow us to move the drm
subsystem trees around without also having to update dim.

Inspired by a patch from Jani.

I did test dim status and dim list-upstreams, but not the pull request
related commands (for lack of having pull requests to do, yay for
being a maintainer, no longer).

v2: Rebase, I had a patch in my local tree which isn't needed with
this one here anymore.

v3: Don't forget the drm-next branch name I accidentally dropped
(Jani).

Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

The old version relied on branch@{upstream}, which requires that
the branch is checked out. Instead use the indirection through the
abstract drm-tip repo.

v2: Questions from Jani:

- We still need the fallback path for non-managed branches like
  rerere-cache or drm-intel-next.

- Also this change removes some of the implicit validation that the
  branch has a local tracking branch. I only spotted one place where
  an assert_branch was missing.

Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Just checks that merges aren't done without minimal thought.

Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

We have some documents floating around internally, imo better if we
track at least the simple ideas here ...

v2: Alternative approach for getting rid of the patch rediffing noise.

v3: We have TODO.rst (Rodrigo).

Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

In

commit c0c4dc1c
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Wed May 2 15:08:24 2018 +0200

    dim: check all commits in dim apply-pull and push-branch

I broke the managed_branch logic which made sure we don't check for
Link: tags for pulled branches (since external pulls are most likely
not managed by dim complaining about the lack of Link: tags is just
noise).

Fix this.

Also fix some missing local variables while at it.

v2: Drop now unecessary {} (Jani).

Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

For trees which integrate other trees it's a bit much to show all the
commits pulled in through those other trees - that should all be
summarized already in the merge commit.

Instead only show commits done in the local tree, whether that's
merges or normal commits. --first-parent achieves that.

Cc: Dave Airlie <airlied@gmail.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Until now, the drm-intel commit access have been handed out ad hoc,
without transparency, consistency, or fairness. With pressure to add
more committers, this is no longer tenable, if it ever was. Document the
requirements and expectations around becoming a drm-intel committer.

The drm-intel maintainers believe that a reasonable level of experience
and track record of working on the driver, as well as actively engaging
in the community upstream, are necessary before becoming a committer.

While the requirements outlined here may seem strict in contrast with
many projects, it seems easier to start strict and relax the
requirements later on as needed than the other way round.

v2: Address some of the concerns brought up by Daniel, and try to align
    the structure with the proposed igt rules.

v3: Update commit message.

Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: dim-tools@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Cc: intel-gfx@lists.freedesktop.org
Acked-by: Dave Airlie <airlied@gmail.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

Obvious oversight.

Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Sean Paul authored 6 years ago and Jani Nikula committed 6 years ago

Using --dry-run for git push on a non-existant branch (since the command
above doesn't actually create the branch) fails with:

error: src refspec topic/dim_test does not match any.
error: failed to push some refs to <remote>

So instead of using --dry-run for git push, just echo the command.

Changes in v2:
- Added comment (Daniel)

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Sean Paul authored 6 years ago and Jani Nikula committed 6 years ago

repo is re-assigned 2 lines lower

Changes in v2:
 - None

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Lucas De Marchi authored 6 years ago and Jani Nikula committed 6 years ago

A single sed can do the job of taking the second line after a match and
it looks simpler.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

This could be a parsing error of the pull (for all the people who
don't use a script like dim pull-request to make sure all the silly
details are right), or some confusion on the part of the sender, or
something else.

Either way not good to continue.

v2: Use -z and drop the wc -l (Jani).

Cc: Dave Airlie <airlied@gmail.com>
Acked-by: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Lots of things we didn't mark up. Reported by Dave.

Cc: Dave Airlie <airlied@gmail.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

This way there's no need for a dim apply-pull-continue, plain old git
commit is enough.

Aside: We might want to do the same trick for dim apply-branch, but
git am is a bit harder to script.

v2: Drop cat, use shell redirects to appeas shellcheck.

Acked-by: Jani Nikula <jani.nikula@intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Looks funny, but let's allow it to be overriden. Also move it
up before we commit to the merge.

v2: Remeber to remove the old message_id parsing (Jani).

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Now that we experiment with dim for drm-next it's much more likely
that pull requests have conflicts. But also that dim already knows
about them, in the recent drm-intel-next pull it resolve 7/8
conflicts.

If it solves them all then just go ahead an commit.

v2: Conflict howto is now in drm-tip.rst (Jani).

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

When merging a pull requests there's potentially a long list of
problematic patches. But currently we only report the first issue and
then bail out.

The upside here is that without this patch the workflow when
processing a pull is:

$ dim apply-pull ...
-> dim refuses, only reports first problem
$ dim -d apply-pull
-> you see all the issues, make up your mind to merge or not
$ dim -f apply-pull

With this patch we have one step less:

$ dim apply-pull
-> refuses pull, but prints all the issues
$ dim -f apply-pull

On the implementation:
- new helper function to check all the patches, that's the only place
  we now call warn_or_fail
- rework the check function to check for everything and return the
  final verdict

v2: Complete rework on Jani's suggestion.

v3: Wrap $@ in "" (Jani).

Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>