Stuff Summit and Sean noticed while testing the new dim:
- Run git fetch in setup to make sure we can create tracking branches.
- Don't fall over the lack of .git/rr-cache.
- Handle drm-intel-next specially in dim checkout since it's a dim
  branch, but not included in drm-tip.

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

Gustovo asked whether there's docs for drm-misc, so took the
opportunity to flesh out the very sparse existing write-up. Also
adjust it to reality, since we've started to take some pretty big
features in through drm-misc (like explicit fencing for atomic).

v2: Be more clear on scope of drm-misc (Jani).

Acked-by:  <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

I never knew this even was a thing, and plain git am cuts everything
below the scissors line (since it matches the --- of a diffstat
start).

Cc: Robert Bragg <robert@sixbynine.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

- dim update-branches was entirely broken, fix it.
- dim checkout's auto-branch-create mode needs to pick the right
  remote
- check for drm-tip remote only in the drm-nightly repo (in case of
  git without worktree support).

v2: Also run git fetch $remote in dim_checkout when creating a new
branch, to avoid uncompresible error message when the branch (or the
remote name) is super-new.

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

If DIM_DRM_INTEL has .. in it (i.e. relative path) not everything
worked as expected, this fixes it. Allows you to use an existing git
checkout, while stashing all the dim stuff into some subdir somewhere
out of sight.

v2: Fixup paths. Oops.

v3: Fixup misplace hunk from rebasing.

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

I had a bug in dim for short while which resulted in the rr-cache dir
getting copied into itself recursively, each time dim rebuild-nightly
was run. I fixed that, but unfortunately it spread to the real repo
already, and a bunch of folks have it in their local checkout.

Stop this infection from spreading by explicitly deleting the
recursive rr-cache dir.

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

With the automatic remote lookup this is no longer needed.

v2: Wrap strings in "" (Jani).

v3: Rebasing, yay!

v4: Fixup rebase fallout and bugs I've spotted.

v5: s/remote_url/url/

v6: Remove get_remote_name, now superseeded by url_to_remote.

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

With the remotes stored in nightly.conf and git worktrees we can avoid
hard-coding them in even more places.

v2: Rebase.

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

- Only update the for-linux-next branches in the repo for the branch
  just pushed.
- Roll out the same logic (to allow us to always merge feature
  patches) for drm-misc.

v2: Bikesheds, bikesheds ... oh dear is this going to be fun to
re-test and then re-re-rebase out all the bugfixes in the right places
again.

v3: Fixup fallout - get_remote_for_branch wasn't working correctly.

v4: Don't use the nightly_ prefix, this is not about the integration
tree. Also sprinkle local annotations for the other read variables.

v5: s/map_nightly_remote/repo_to_remote/ to align with url_to_remote.

v6: s/get_nightly_repo_for_branch/branch_to_repo/ to align with
url_to_remote.  Also move next to url_to_remote.

v7: Split out branch_to_repo, already needed in an earlier patch.

v8: Really try to get rid of rebase fail, somehow a nightly_origin
got resurrected!

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

Looking up the remote for deleting the branch is a bit an exercise.

v2: Correctly keep the fake remote name from nightly.conf and the real
(local) remote name apart.

v3: More untested rebasing. This removes the FIXME comments and compat
code from Jani, since these two functions should work now properly.

v4: Now tested, and with fallout fixed - map_nightly_remote was
was missing, and also accidently left some old garbage in.

v5: s/map_nightly_remote/repo_to_remote/ to align with url_to_remote.
Also move next to url_to_remote.

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

This first patch just handles push and pull-request. Since the code
now auto-transfroms ssh:// git urls to git:// urls we can remove
$drm_intel_git.

Still left to do is fixing up create-branch and remove-branch.
Everything else should work here now.

v2: Rebase&fixup fallout.

v3: Squash in follow-up patch to rectify pull-request more.

v4: Also make it work properly again for drm-intel-next (which is not
in $integration_configuration and hence a special case again).

v5: s/get_remote_for_branch/branch_to_remote/ to align with
url_to_remote.  Also move next to url_to_remote.

v6: Move branch_to_repo to the right patch.

v7: Split out branch_to_remote, needed in an earlier patch.

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

Much shorter and less ugly than &> /dev/null. Suggested by Jani.

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

This is a prep step to eventually split out the drm-misc branches all
into their own repo. To get there we first need to split out the
integration tree mangling.

Since dim doesn't auto-update and since the transition is a good
reason to switch over to worktrees and it's tricky it's not scripted.
The following needs to be done when dim rebuild-nightly fails:

- Make sure you have the latest dim version.

- Good to also check you have a recent-ish git. Latest dim supports
  git worktree to save on disk space needed for .git metadata. It's
  much more robust than the git alternates storage trick we've used
  thus far. Note that git worktree is still optional (but that's not a
  recommended setup).

- Delete $DIM_PREFIX/{drm-intel-rerere,drm-intel-nightly,maintainer-tools}
  It's not strictly needed to kill the maintainer-tools checkout too,
  but might as well switch it over to git worktree too.

- Make sure you don't have any branches left over for drm-intel-rerere
  and drm-intel-nightly anywhere (since they need to switch their
  remotes).

- run dim setup and enjoy the new awesomeness.

If it breaks, I'm guilty ;-)

v2: don't call get_remote_name before switching directories.

v3: Rebase prep patches to be at the start of the series, which also
allows this patch to be squashed together with the one that renames
-nightly to -tip.

v4: Jumping over bikesheds.

v5: One more jump in read_integration_config.

v6: Use remote instead of origin in local variables, picked up from
Jani.

v7: Split out misplaced hunk.

v8: s/nightly_remote/remote, nightly_ should be only used for the
integration branch remotes. While at it also use grep -q instead of &>
/dev/null

v9: With nightly_remote free use that to denote the drm-tip remote in
rebuild-nightly.

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

Joonas rightly complained that the current output is useless and just
confuses.

v2: Bikeshed stuff some more.

v3: Split out misplaced hunk.

Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Needs new url-mapping information in nightly.conf to work properly. Only
leftover bit from my original patch after rebasing is to abstract away
the remote for the integration tree.

v2:
- Fix typo when calling rr_cache_dir. Oops.
- Use cut -f 1 instead of sed/regex horrors (jani).

v3: Fix bug that resulted in nesting rr-cache directories.

v4: Rebasing.

v5: Resurrect $repo (as $remote) printing when fetching branches,
originally from my patch.

v6: Use remote instead of origin local variables, to be more
consistent. Picked up from Jani.

v7: Move misplaced hunks to right place.

v8: For the nightly remote use nightly_remote instead of anything
called origin - it's confusing.

v9: Create a separate rerere_remote for the rerere-cache branch. It
might be in a difference local git repository (Jani).

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

Needs new url-mapping information in nightly.conf to work properly. Only
leftover bit from my original patch after rebasing is to abstract away
the remote for the integration tree.

v2:
- Fix typo when calling rr_cache_dir. Oops.
- Use cut -f 1 instead of sed/regex horrors (jani).

v3: Fix bug that resulted in nesting rr-cache directories.

v4: Rebasing.

v5: Resurrect $repo (as $remote) printing when fetching branches,
originally from my patch.

v6: Use remote instead of origin local variables, to be more
consistent. Picked up from Jani.

v7: Move misplaced hunks to right place.

v8: For the nightly remote use nightly_remote instead of anything
called origin - it's confusing.

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

If available by default. This saves quite a pile of disk-space that
imo making it the default is justified.

Note that this will break the rebuild-nightly script for now,
follow-up patches will fix that.

v2: Rebase to put the dim rework at the end of the series.

v3: s/remote_url/url/

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

The goals here are multiple:
- simpler configuration through autodetection
- allows seamless upgrading to git worktree for the aux checkouts
- eventually I want to split up drm-misc into a separate remote ...

And yes this is just a start.

v2: Print errors to stderr, otherwise they can't be seen when directly
assigning the result of get_remote_name to a variable.

v3: Rebased to pull the cleanup patches ahead of the dim rework.

v4: Rebase to put the dim rework at the end of the series.

v5: s/get_remote_name/url_to_remote/ to align with Jani.

v6: s/remote_url/url/ plus add a variable naming convention.

v7: Move the naming convention next to the foo_to_bar mapping
functions.

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

When there's a working-tree checkout we need to remove that first,
before removing the branch. git complains otherwise. Not sure,
but this might be a recently added more strict test for the git
worktree support, at least I believe the "branch still checkout out
in $dir" warning is new-ish.

v2: Move to the front, also include the git worktree prune call to
clean up the potential mess.

Fixes: f507c923 ("dim: abstract dim_{create,remove}_branch")
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Rodrigo recently asked me about this, and I totally failed to properly
document it!

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

NOTE: This change depends on the nightly.conf commit 0d0e6d4e7dcc
("nightly.conf: prepare for remote agnostic configuration") in the
rerere-cache branch of the drm-intel repo. Looking at that first makes
this change more sensible.

Use two arrays to configure the repos and branches to be merged to the
integration branch:

drm_tip_repos

	An associative array that maps repo names to urls. This is
	mostly a convenience for defining the other array. The repo
	names are symbolic, and not related to actual git remote names.
	It's also helpful for implementing dim create-branch and
	remove-branch.

drm_tip_config

	An array of strings which describes the repos and branches to be
	used to generate the integration branch. The repos are listed
	using the symbolic repo names from the drm_tip_repos array. It's
	also possible to list an override sha, in case there's a need to
	hold back updating to the tip of the branch for some reason.

dim as a whole still remains dependent on specific (and configured)
remote names, but this change detaches nightly.conf from the remote
names.

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

We'll change the name at some point, add some indirection, with a
generic variable name.

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

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

echoerr for error output, and start them with an explanation for
what's amiss. Also adjust the check to look for a .git directory, not
just any directory.

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

Chris Wilson authored 8 years ago and Jani Nikula committed 8 years ago

Along with the DRIVER_DATE, also update the DRIVER_TIMESTAMP as measured
in seconds from the start of the epoch.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

For even more Cc! Result is sometimes a bit excessive and will
need some trimming sometimes.

v2: Don't sort, since that makes a mess of the list. Use magic awk
command found through googling instead.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[Jani: fix some indentation, add --pattern-depth 1 option]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

Encodes the best practices for undoing a botched-up conflict
resolution.

v2: Rebased to pull prep work before the dim reorg.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
[Jani: fix some typos]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

... instead of git branch | grep. Handles submatches and branches
starting with - reliably. Motivated by a review from Jani.

v2: Rebase ahead of the bigger dim reworks (requested by Jani).

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

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

With our proliferation of branches it's become long useless. Nowadays
my MO is to revert the offending commit in the rerere-cache branch
explicitly, remove drm-intel-nightly/.git/rr-cache and then re-run
rebuild-nightly. That works much better, hence nuke this helper.

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

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

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

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

Exits script to annoy people roughly every 100th time ...

Also switch to the magic @{upstream} reference, in case the remote is
not called origin (which is pretty normal in case of using git
worktree).

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

Simona Vetter authored 8 years ago and Jani Nikula committed 8 years ago

Just maybe a bit more visibility, the scripts are growing.

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

Add a command to check if the user is running an up-to-date version of
dim.

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

Run sparse only on files that have changed in the range.

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

Defaults to -j20.

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

The -f option does what the doctor orders.

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

The usage has become a bit unwieldy with the command list.

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

Add $dim for the basename of the script and use it.

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

The user can eyeball the list of Cc's while git churns through history
to figure out whether the fix belongs in stable or not.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Add 'dim cite' subcommand to cite the supplied committish in format
'sha1 ("commit subject")'.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>