223 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3====================4Rebasing and merging5====================6 7Maintaining a subsystem, as a general rule, requires a familiarity with the8Git source-code management system. Git is a powerful tool with a lot of9features; as is often the case with such tools, there are right and wrong10ways to use those features. This document looks in particular at the use11of rebasing and merging. Maintainers often get in trouble when they use12those tools incorrectly, but avoiding problems is not actually all that13hard.14 15One thing to be aware of in general is that, unlike many other projects,16the kernel community is not scared by seeing merge commits in its17development history. Indeed, given the scale of the project, avoiding18merges would be nearly impossible. Some problems encountered by19maintainers result from a desire to avoid merges, while others come from20merging a little too often.21 22Rebasing23========24 25"Rebasing" is the process of changing the history of a series of commits26within a repository. There are two different types of operations that are27referred to as rebasing since both are done with the ``git rebase``28command, but there are significant differences between them:29 30 - Changing the parent (starting) commit upon which a series of patches is31 built. For example, a rebase operation could take a patch set built on32 the previous kernel release and base it, instead, on the current33 release. We'll call this operation "reparenting" in the discussion34 below.35 36 - Changing the history of a set of patches by fixing (or deleting) broken37 commits, adding patches, adding tags to commit changelogs, or changing38 the order in which commits are applied. In the following text, this39 type of operation will be referred to as "history modification"40 41The term "rebasing" will be used to refer to both of the above operations.42Used properly, rebasing can yield a cleaner and clearer development43history; used improperly, it can obscure that history and introduce bugs.44 45There are a few rules of thumb that can help developers to avoid the worst46perils of rebasing:47 48 - History that has been exposed to the world beyond your private system49 should usually not be changed. Others may have pulled a copy of your50 tree and built on it; modifying your tree will create pain for them. If51 work is in need of rebasing, that is usually a sign that it is not yet52 ready to be committed to a public repository.53 54 That said, there are always exceptions. Some trees (linux-next being55 a significant example) are frequently rebased by their nature, and56 developers know not to base work on them. Developers will sometimes57 expose an unstable branch for others to test with or for automated58 testing services. If you do expose a branch that may be unstable in59 this way, be sure that prospective users know not to base work on it.60 61 - Do not rebase a branch that contains history created by others. If you62 have pulled changes from another developer's repository, you are now a63 custodian of their history. You should not change it. With few64 exceptions, for example, a broken commit in a tree like this should be65 explicitly reverted rather than disappeared via history modification.66 67 - Do not reparent a tree without a good reason to do so. Just being on a68 newer base or avoiding a merge with an upstream repository is not69 generally a good reason.70 71 - If you must reparent a repository, do not pick some random kernel commit72 as the new base. The kernel is often in a relatively unstable state73 between release points; basing development on one of those points74 increases the chances of running into surprising bugs. When a patch75 series must move to a new base, pick a stable point (such as one of76 the -rc releases) to move to.77 78 - Realize that reparenting a patch series (or making significant history79 modifications) changes the environment in which it was developed and,80 likely, invalidates much of the testing that was done. A reparented81 patch series should, as a general rule, be treated like new code and82 retested from the beginning.83 84A frequent cause of merge-window trouble is when Linus is presented with a85patch series that has clearly been reparented, often to a random commit,86shortly before the pull request was sent. The chances of such a series87having been adequately tested are relatively low - as are the chances of88the pull request being acted upon.89 90If, instead, rebasing is limited to private trees, commits are based on a91well-known starting point, and they are well tested, the potential for92trouble is low.93 94Merging95=======96 97Merging is a common operation in the kernel development process; the 5.198development cycle included 1,126 merge commits - nearly 9% of the total.99Kernel work is accumulated in over 100 different subsystem trees, each of100which may contain multiple topic branches; each branch is usually developed101independently of the others. So naturally, at least one merge will be102required before any given branch finds its way into an upstream repository.103 104Many projects require that branches in pull requests be based on the105current trunk so that no merge commits appear in the history. The kernel106is not such a project; any rebasing of branches to avoid merges will, most107likely, lead to trouble.108 109Subsystem maintainers find themselves having to do two types of merges:110from lower-level subsystem trees and from others, either sibling trees or111the mainline. The best practices to follow differ in those two situations.112 113Merging from lower-level trees114------------------------------115 116Larger subsystems tend to have multiple levels of maintainers, with the117lower-level maintainers sending pull requests to the higher levels. Acting118on such a pull request will almost certainly generate a merge commit; that119is as it should be. In fact, subsystem maintainers may want to use120the --no-ff flag to force the addition of a merge commit in the rare cases121where one would not normally be created so that the reasons for the merge122can be recorded. The changelog for the merge should, for any kind of123merge, say *why* the merge is being done. For a lower-level tree, "why" is124usually a summary of the changes that will come with that pull.125 126Maintainers at all levels should be using signed tags on their pull127requests, and upstream maintainers should verify the tags when pulling128branches. Failure to do so threatens the security of the development129process as a whole.130 131As per the rules outlined above, once you have merged somebody else's132history into your tree, you cannot rebase that branch, even if you133otherwise would be able to.134 135Merging from sibling or upstream trees136--------------------------------------137 138While merges from downstream are common and unremarkable, merges from other139trees tend to be a red flag when it comes time to push a branch upstream.140Such merges need to be carefully thought about and well justified, or141there's a good chance that a subsequent pull request will be rejected.142 143It is natural to want to merge the master branch into a repository; this144type of merge is often called a "back merge". Back merges can help to make145sure that there are no conflicts with parallel development and generally146gives a warm, fuzzy feeling of being up-to-date. But this temptation147should be avoided almost all of the time.148 149Why is that? Back merges will muddy the development history of your own150branch. They will significantly increase your chances of encountering bugs151from elsewhere in the community and make it hard to ensure that the work152you are managing is stable and ready for upstream. Frequent merges can153also obscure problems with the development process in your tree; they can154hide interactions with other trees that should not be happening (often) in155a well-managed branch.156 157That said, back merges are occasionally required; when that happens, be158sure to document *why* it was required in the commit message. As always,159merge to a well-known stable point, rather than to some random commit.160Even then, you should not back merge a tree above your immediate upstream161tree; if a higher-level back merge is really required, the upstream tree162should do it first.163 164One of the most frequent causes of merge-related trouble is when a165maintainer merges with the upstream in order to resolve merge conflicts166before sending a pull request. Again, this temptation is easy enough to167understand, but it should absolutely be avoided. This is especially true168for the final pull request: Linus is adamant that he would much rather see169merge conflicts than unnecessary back merges. Seeing the conflicts lets170him know where potential problem areas are. He does a lot of merges (382171in the 5.1 development cycle) and has gotten quite good at conflict172resolution - often better than the developers involved.173 174So what should a maintainer do when there is a conflict between their175subsystem branch and the mainline? The most important step is to warn176Linus in the pull request that the conflict will happen; if nothing else,177that demonstrates an awareness of how your branch fits into the whole. For178especially difficult conflicts, create and push a *separate* branch to show179how you would resolve things. Mention that branch in your pull request,180but the pull request itself should be for the unmerged branch.181 182Even in the absence of known conflicts, doing a test merge before sending a183pull request is a good idea. It may alert you to problems that you somehow184didn't see from linux-next and helps to understand exactly what you are185asking upstream to do.186 187Another reason for doing merges of upstream or another subsystem tree is to188resolve dependencies. These dependency issues do happen at times, and189sometimes a cross-merge with another tree is the best way to resolve them;190as always, in such situations, the merge commit should explain why the191merge has been done. Take a moment to do it right; people will read those192changelogs.193 194Often, though, dependency issues indicate that a change of approach is195needed. Merging another subsystem tree to resolve a dependency risks196bringing in other bugs and should almost never be done. If that subsystem197tree fails to be pulled upstream, whatever problems it had will block the198merging of your tree as well. Preferable alternatives include agreeing199with the maintainer to carry both sets of changes in one of the trees or200creating a topic branch dedicated to the prerequisite commits that can be201merged into both trees. If the dependency is related to major202infrastructural changes, the right solution might be to hold the dependent203commits for one development cycle so that those changes have time to204stabilize in the mainline.205 206Finally207=======208 209It is relatively common to merge with the mainline toward the beginning of210the development cycle in order to pick up changes and fixes done elsewhere211in the tree. As always, such a merge should pick a well-known release212point rather than some random spot. If your upstream-bound branch has213emptied entirely into the mainline during the merge window, you can pull it214forward with a command like::215 216 git merge --ff-only v5.2-rc1217 218The guidelines laid out above are just that: guidelines. There will always219be situations that call out for a different solution, and these guidelines220should not prevent developers from doing the right thing when the need221arises. But one should always think about whether the need has truly222arisen and be prepared to explain why something abnormal needs to be done. 223