| Message ID | 20170524092812.22937-1-daniel.vetter@ffwll.ch (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, May 24, 2017 at 11:28:12AM +0200, Daniel Vetter wrote: > We can't check this when applying (since r-b/a-b tags often get added > afterwards), but we can check this when pushing. This only looks at > patches authored by the pusher. > > Also update the docs to highlight that review requirements hold > especially also for bugfixes. > > Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com> > Cc: Lukas Wunner <lukas@wunner.de> > Cc: Alex Deucher <alexander.deucher@amd.com> > Cc: Christian König <deathsimple@vodafone.de> > Cc: Sean Paul <seanpaul@chromium.org> > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Reviewed-by: Lukas Wunner <lukas@wunner.de> Thanks, Lukas > --- > dim | 42 ++++++++++++++++++++++++++++++++++++++---- > drm-misc.rst | 4 +++- > 2 files changed, 41 insertions(+), 5 deletions(-) > > diff --git a/dim b/dim > index baa0b3832314..79738a1b37a0 100755 > --- a/dim > +++ b/dim > @@ -340,6 +340,15 @@ function git_branch_exists # branch > fi > } > > +function git_committer_email > +{ > + if ! commiter_email=$(git config --get user.email) ; then > + commiter_email=$EMAIL > + fi > + > + echo $commiter_email > +} > + > # get message id from file > # $1 = file > message_get_id () > @@ -632,11 +641,32 @@ function dim_rebuild_tip > exit 1 > fi > } > + > +# additional patch checks before pushing, e.g. for r-b tags > +function checkpatch_commit_push > +{ > + local sha1 > + > + sha1=$1 > + > + # check for a-b/r-b tag > + if git show -s $sha1 | grep -qi '\(reviewed\|acked\)\S*-by:' ; then > + return > + fi > + > + # check for commiter != author > + if [[ $(git show -s $sha1 --format="format:%ce") != $(git show -s $sha1 --format="format:%ae") ]]; then > + return > + fi > + > + warn_or_fail "$sha1 is lacking mandatory review" > +} > + > # push branch $1, rebuild drm-tip. the rest of the arguments are passed to git > # push. > function dim_push_branch > { > - local branch remote > + local branch remote commiter_email > > branch=${1:?$usage} > shift > @@ -645,6 +675,12 @@ function dim_push_branch > > remote=$(branch_to_remote $branch) > > + commiter_email=$(git_committer_email) > + > + for sha1 in $(git rev-list $branch@{u}..$branch --committer="$commiter_email" --no-merges) ; do > + checkpatch_commit_push $sha1 > + done > + > git push $DRY_RUN $remote $branch "$@" > > update_linux_next $branch drm-intel-next-queued drm-intel-next-fixes drm-intel-fixes > @@ -690,9 +726,7 @@ function dim_apply_branch > > message_id=$(message_get_id $file) > > - if ! commiter_email=$(git config --get user.email) ; then > - commiter_email=$EMAIL > - fi > + commiter_email=$(git_committer_email) > > patch_from=$(grep "From:" "$file" | head -1) > if [[ "$patch_from" != *"$commiter_email"* ]] ; then > diff --git a/drm-misc.rst b/drm-misc.rst > index caba8dc77696..d56c3c7f92a3 100644 > --- a/drm-misc.rst > +++ b/drm-misc.rst > @@ -90,7 +90,9 @@ Merge Criteria > Right now the only hard merge criteria are: > > * Patch is properly reviewed or at least Ack, i.e. don't just push your own > - stuff directly. > + stuff directly. This rule holds even more for bugfix patches - it would be > + embarrassing if the bugfix contains a small gotcha that review would have > + caught. > > * drm-misc is for drm core (non-driver) patches, subsystem-wide refactorings, > and small trivial patches all over (including drivers). For a detailed list of > -- > 2.11.0 >
> -----Original Message----- > From: Daniel Vetter [mailto:daniel.vetter@ffwll.ch] > Sent: Wednesday, May 24, 2017 5:28 AM > To: DRI Development > Cc: Intel Graphics Development; Daniel Vetter; Patrik Jakobsson; Lukas > Wunner; Deucher, Alexander; Christian König; Sean Paul; Daniel Vetter > Subject: [PATCH] dim: Enforce review requirements > > We can't check this when applying (since r-b/a-b tags often get added > afterwards), but we can check this when pushing. This only looks at > patches authored by the pusher. > > Also update the docs to highlight that review requirements hold > especially also for bugfixes. > > Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com> > Cc: Lukas Wunner <lukas@wunner.de> > Cc: Alex Deucher <alexander.deucher@amd.com> > Cc: Christian König <deathsimple@vodafone.de> > Cc: Sean Paul <seanpaul@chromium.org> > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Acked-by: Alex Deucher <alexander.deucher@amd.com> > --- > dim | 42 ++++++++++++++++++++++++++++++++++++++---- > drm-misc.rst | 4 +++- > 2 files changed, 41 insertions(+), 5 deletions(-) > > diff --git a/dim b/dim > index baa0b3832314..79738a1b37a0 100755 > --- a/dim > +++ b/dim > @@ -340,6 +340,15 @@ function git_branch_exists # branch > fi > } > > +function git_committer_email > +{ > + if ! commiter_email=$(git config --get user.email) ; then > + commiter_email=$EMAIL > + fi > + > + echo $commiter_email > +} > + > # get message id from file > # $1 = file > message_get_id () > @@ -632,11 +641,32 @@ function dim_rebuild_tip > exit 1 > fi > } > + > +# additional patch checks before pushing, e.g. for r-b tags > +function checkpatch_commit_push > +{ > + local sha1 > + > + sha1=$1 > + > + # check for a-b/r-b tag > + if git show -s $sha1 | grep -qi '\(reviewed\|acked\)\S*-by:' ; then > + return > + fi > + > + # check for commiter != author > + if [[ $(git show -s $sha1 --format="format:%ce") != $(git show -s > $sha1 --format="format:%ae") ]]; then > + return > + fi > + > + warn_or_fail "$sha1 is lacking mandatory review" > +} > + > # push branch $1, rebuild drm-tip. the rest of the arguments are passed to > git > # push. > function dim_push_branch > { > - local branch remote > + local branch remote commiter_email > > branch=${1:?$usage} > shift > @@ -645,6 +675,12 @@ function dim_push_branch > > remote=$(branch_to_remote $branch) > > + commiter_email=$(git_committer_email) > + > + for sha1 in $(git rev-list $branch@{u}..$branch -- > committer="$commiter_email" --no-merges) ; do > + checkpatch_commit_push $sha1 > + done > + > git push $DRY_RUN $remote $branch "$@" > > update_linux_next $branch drm-intel-next-queued drm-intel-next- > fixes drm-intel-fixes > @@ -690,9 +726,7 @@ function dim_apply_branch > > message_id=$(message_get_id $file) > > - if ! commiter_email=$(git config --get user.email) ; then > - commiter_email=$EMAIL > - fi > + commiter_email=$(git_committer_email) > > patch_from=$(grep "From:" "$file" | head -1) > if [[ "$patch_from" != *"$commiter_email"* ]] ; then > diff --git a/drm-misc.rst b/drm-misc.rst > index caba8dc77696..d56c3c7f92a3 100644 > --- a/drm-misc.rst > +++ b/drm-misc.rst > @@ -90,7 +90,9 @@ Merge Criteria > Right now the only hard merge criteria are: > > * Patch is properly reviewed or at least Ack, i.e. don't just push your own > - stuff directly. > + stuff directly. This rule holds even more for bugfix patches - it would be > + embarrassing if the bugfix contains a small gotcha that review would have > + caught. > > * drm-misc is for drm core (non-driver) patches, subsystem-wide > refactorings, > and small trivial patches all over (including drivers). For a detailed list of > -- > 2.11.0
On Wed, May 24, 2017 at 11:28:12AM +0200, Daniel Vetter wrote: > We can't check this when applying (since r-b/a-b tags often get added > afterwards), but we can check this when pushing. This only looks at > patches authored by the pusher. BTW, can we also have a rule that large drivers (i.e. with a large user base) are required to provide at least one person at all times (or at least outside the merge window) who is able to review fixes, push them to the driver's repo and send a timely pull to Dave? On multiple occasions I've seen Dave send a fixes pull with an apology along the lines of "a bit more than usual at this time of the cycle because it includes two weeks worth of driver XYZ fixes". Thanks, Lukas
On Thu, May 25, 2017 at 7:37 AM, Lukas Wunner <lukas@wunner.de> wrote: > On Wed, May 24, 2017 at 11:28:12AM +0200, Daniel Vetter wrote: >> We can't check this when applying (since r-b/a-b tags often get added >> afterwards), but we can check this when pushing. This only looks at >> patches authored by the pusher. > > BTW, can we also have a rule that large drivers (i.e. with a large user > base) are required to provide at least one person at all times (or at > least outside the merge window) who is able to review fixes, push them > to the driver's repo and send a timely pull to Dave? > > On multiple occasions I've seen Dave send a fixes pull with an apology > along the lines of "a bit more than usual at this time of the cycle > because it includes two weeks worth of driver XYZ fixes". AMD is already planning to swtich to a group/committer model, I just discussed that with Alex. That takes care of the huge drivers (intel/amd) and the small ones (if they are in drm-misc). No idea what we'll end up doing with the medium sized drivers, but I guess that's a discussion for the next months/years ... -Daniel
On Wed, May 24, 2017 at 11:28:12AM +0200, Daniel Vetter wrote: > We can't check this when applying (since r-b/a-b tags often get added > afterwards), but we can check this when pushing. This only looks at > patches authored by the pusher. > > Also update the docs to highlight that review requirements hold > especially also for bugfixes. > > Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com> > Cc: Lukas Wunner <lukas@wunner.de> > Cc: Alex Deucher <alexander.deucher@amd.com> > Cc: Christian König <deathsimple@vodafone.de> > Cc: Sean Paul <seanpaul@chromium.org> I think I acked this on IRC, but just in case, Acked-by: Sean Paul <seanpaul@chromium.org> > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> > --- > dim | 42 ++++++++++++++++++++++++++++++++++++++---- > drm-misc.rst | 4 +++- > 2 files changed, 41 insertions(+), 5 deletions(-) > > diff --git a/dim b/dim > index baa0b3832314..79738a1b37a0 100755 > --- a/dim > +++ b/dim > @@ -340,6 +340,15 @@ function git_branch_exists # branch > fi > } > > +function git_committer_email > +{ > + if ! commiter_email=$(git config --get user.email) ; then > + commiter_email=$EMAIL > + fi > + > + echo $commiter_email > +} > + > # get message id from file > # $1 = file > message_get_id () > @@ -632,11 +641,32 @@ function dim_rebuild_tip > exit 1 > fi > } > + > +# additional patch checks before pushing, e.g. for r-b tags > +function checkpatch_commit_push > +{ > + local sha1 > + > + sha1=$1 > + > + # check for a-b/r-b tag > + if git show -s $sha1 | grep -qi '\(reviewed\|acked\)\S*-by:' ; then > + return > + fi > + > + # check for commiter != author > + if [[ $(git show -s $sha1 --format="format:%ce") != $(git show -s $sha1 --format="format:%ae") ]]; then > + return > + fi > + > + warn_or_fail "$sha1 is lacking mandatory review" > +} > + > # push branch $1, rebuild drm-tip. the rest of the arguments are passed to git > # push. > function dim_push_branch > { > - local branch remote > + local branch remote commiter_email > > branch=${1:?$usage} > shift > @@ -645,6 +675,12 @@ function dim_push_branch > > remote=$(branch_to_remote $branch) > > + commiter_email=$(git_committer_email) > + > + for sha1 in $(git rev-list $branch@{u}..$branch --committer="$commiter_email" --no-merges) ; do > + checkpatch_commit_push $sha1 > + done > + > git push $DRY_RUN $remote $branch "$@" > > update_linux_next $branch drm-intel-next-queued drm-intel-next-fixes drm-intel-fixes > @@ -690,9 +726,7 @@ function dim_apply_branch > > message_id=$(message_get_id $file) > > - if ! commiter_email=$(git config --get user.email) ; then > - commiter_email=$EMAIL > - fi > + commiter_email=$(git_committer_email) > > patch_from=$(grep "From:" "$file" | head -1) > if [[ "$patch_from" != *"$commiter_email"* ]] ; then > diff --git a/drm-misc.rst b/drm-misc.rst > index caba8dc77696..d56c3c7f92a3 100644 > --- a/drm-misc.rst > +++ b/drm-misc.rst > @@ -90,7 +90,9 @@ Merge Criteria > Right now the only hard merge criteria are: > > * Patch is properly reviewed or at least Ack, i.e. don't just push your own > - stuff directly. > + stuff directly. This rule holds even more for bugfix patches - it would be > + embarrassing if the bugfix contains a small gotcha that review would have > + caught. > > * drm-misc is for drm core (non-driver) patches, subsystem-wide refactorings, > and small trivial patches all over (including drivers). For a detailed list of > -- > 2.11.0
On 05/24/2017 11:28 AM, Daniel Vetter wrote: > We can't check this when applying (since r-b/a-b tags often get added > afterwards), but we can check this when pushing. This only looks at > patches authored by the pusher. > > Also update the docs to highlight that review requirements hold > especially also for bugfixes. > > Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com> > Cc: Lukas Wunner <lukas@wunner.de> > Cc: Alex Deucher <alexander.deucher@amd.com> > Cc: Christian König <deathsimple@vodafone.de> > Cc: Sean Paul <seanpaul@chromium.org> > Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> > --- > dim | 42 ++++++++++++++++++++++++++++++++++++++---- > drm-misc.rst | 4 +++- > 2 files changed, 41 insertions(+), 5 deletions(-) > > diff --git a/dim b/dim > index baa0b3832314..79738a1b37a0 100755 > --- a/dim > +++ b/dim > @@ -340,6 +340,15 @@ function git_branch_exists # branch > fi > } > > +function git_committer_email > +{ > + if ! commiter_email=$(git config --get user.email) ; then > + commiter_email=$EMAIL > + fi > + > + echo $commiter_email > +} > + > # get message id from file > # $1 = file > message_get_id () > @@ -632,11 +641,32 @@ function dim_rebuild_tip > exit 1 > fi > } > + > +# additional patch checks before pushing, e.g. for r-b tags > +function checkpatch_commit_push > +{ > + local sha1 > + > + sha1=$1 > + > + # check for a-b/r-b tag > + if git show -s $sha1 | grep -qi '\(reviewed\|acked\)\S*-by:' ; then > + return > + fi > + > + # check for commiter != author > + if [[ $(git show -s $sha1 --format="format:%ce") != $(git show -s $sha1 --format="format:%ae") ]]; then > + return > + fi > + > + warn_or_fail "$sha1 is lacking mandatory review" > +} > + > # push branch $1, rebuild drm-tip. the rest of the arguments are passed to git > # push. > function dim_push_branch > { > - local branch remote > + local branch remote commiter_email > > branch=${1:?$usage} > shift > @@ -645,6 +675,12 @@ function dim_push_branch > > remote=$(branch_to_remote $branch) > > + commiter_email=$(git_committer_email) > + > + for sha1 in $(git rev-list $branch@{u}..$branch --committer="$commiter_email" --no-merges) ; do > + checkpatch_commit_push $sha1 > + done > + > git push $DRY_RUN $remote $branch "$@" > > update_linux_next $branch drm-intel-next-queued drm-intel-next-fixes drm-intel-fixes > @@ -690,9 +726,7 @@ function dim_apply_branch > > message_id=$(message_get_id $file) > > - if ! commiter_email=$(git config --get user.email) ; then > - commiter_email=$EMAIL > - fi > + commiter_email=$(git_committer_email) > > patch_from=$(grep "From:" "$file" | head -1) > if [[ "$patch_from" != *"$commiter_email"* ]] ; then > diff --git a/drm-misc.rst b/drm-misc.rst > index caba8dc77696..d56c3c7f92a3 100644 > --- a/drm-misc.rst > +++ b/drm-misc.rst > @@ -90,7 +90,9 @@ Merge Criteria > Right now the only hard merge criteria are: > > * Patch is properly reviewed or at least Ack, i.e. don't just push your own > - stuff directly. > + stuff directly. This rule holds even more for bugfix patches - it would be > + embarrassing if the bugfix contains a small gotcha that review would have > + caught. > > * drm-misc is for drm core (non-driver) patches, subsystem-wide refactorings, > and small trivial patches all over (including drivers). For a detailed list of > Reviewed-by: Neil Armstrong <narmstrong@baylibre.com>
diff --git a/dim b/dim index baa0b3832314..79738a1b37a0 100755 --- a/dim +++ b/dim @@ -340,6 +340,15 @@ function git_branch_exists # branch fi } +function git_committer_email +{ + if ! commiter_email=$(git config --get user.email) ; then + commiter_email=$EMAIL + fi + + echo $commiter_email +} + # get message id from file # $1 = file message_get_id () @@ -632,11 +641,32 @@ function dim_rebuild_tip exit 1 fi } + +# additional patch checks before pushing, e.g. for r-b tags +function checkpatch_commit_push +{ + local sha1 + + sha1=$1 + + # check for a-b/r-b tag + if git show -s $sha1 | grep -qi '\(reviewed\|acked\)\S*-by:' ; then + return + fi + + # check for commiter != author + if [[ $(git show -s $sha1 --format="format:%ce") != $(git show -s $sha1 --format="format:%ae") ]]; then + return + fi + + warn_or_fail "$sha1 is lacking mandatory review" +} + # push branch $1, rebuild drm-tip. the rest of the arguments are passed to git # push. function dim_push_branch { - local branch remote + local branch remote commiter_email branch=${1:?$usage} shift @@ -645,6 +675,12 @@ function dim_push_branch remote=$(branch_to_remote $branch) + commiter_email=$(git_committer_email) + + for sha1 in $(git rev-list $branch@{u}..$branch --committer="$commiter_email" --no-merges) ; do + checkpatch_commit_push $sha1 + done + git push $DRY_RUN $remote $branch "$@" update_linux_next $branch drm-intel-next-queued drm-intel-next-fixes drm-intel-fixes @@ -690,9 +726,7 @@ function dim_apply_branch message_id=$(message_get_id $file) - if ! commiter_email=$(git config --get user.email) ; then - commiter_email=$EMAIL - fi + commiter_email=$(git_committer_email) patch_from=$(grep "From:" "$file" | head -1) if [[ "$patch_from" != *"$commiter_email"* ]] ; then diff --git a/drm-misc.rst b/drm-misc.rst index caba8dc77696..d56c3c7f92a3 100644 --- a/drm-misc.rst +++ b/drm-misc.rst @@ -90,7 +90,9 @@ Merge Criteria Right now the only hard merge criteria are: * Patch is properly reviewed or at least Ack, i.e. don't just push your own - stuff directly. + stuff directly. This rule holds even more for bugfix patches - it would be + embarrassing if the bugfix contains a small gotcha that review would have + caught. * drm-misc is for drm core (non-driver) patches, subsystem-wide refactorings, and small trivial patches all over (including drivers). For a detailed list of
We can't check this when applying (since r-b/a-b tags often get added afterwards), but we can check this when pushing. This only looks at patches authored by the pusher. Also update the docs to highlight that review requirements hold especially also for bugfixes. Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com> Cc: Lukas Wunner <lukas@wunner.de> Cc: Alex Deucher <alexander.deucher@amd.com> Cc: Christian König <deathsimple@vodafone.de> Cc: Sean Paul <seanpaul@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> --- dim | 42 ++++++++++++++++++++++++++++++++++++++---- drm-misc.rst | 4 +++- 2 files changed, 41 insertions(+), 5 deletions(-)