Based on kernel version 3.9. Page generated on 2013-05-02 22:56 EST.
1 7: ADVANCED TOPICS 2 3 At this point, hopefully, you have a handle on how the development process 4 works. There is still more to learn, however! This section will cover a 5 number of topics which can be helpful for developers wanting to become a 6 regular part of the Linux kernel development process. 7 8 7.1: MANAGING PATCHES WITH GIT 9 10 The use of distributed version control for the kernel began in early 2002, 11 when Linus first started playing with the proprietary BitKeeper 12 application. While BitKeeper was controversial, the approach to software 13 version management it embodied most certainly was not. Distributed version 14 control enabled an immediate acceleration of the kernel development 15 project. In current times, there are several free alternatives to 16 BitKeeper. For better or for worse, the kernel project has settled on git 17 as its tool of choice. 18 19 Managing patches with git can make life much easier for the developer, 20 especially as the volume of those patches grows. Git also has its rough 21 edges and poses certain hazards; it is a young and powerful tool which is 22 still being civilized by its developers. This document will not attempt to 23 teach the reader how to use git; that would be sufficient material for a 24 long document in its own right. Instead, the focus here will be on how git 25 fits into the kernel development process in particular. Developers who 26 wish to come up to speed with git will find more information at: 27 28 http://git-scm.com/ 29 30 http://www.kernel.org/pub/software/scm/git/docs/user-manual.html 31 32 and on various tutorials found on the web. 33 34 The first order of business is to read the above sites and get a solid 35 understanding of how git works before trying to use it to make patches 36 available to others. A git-using developer should be able to obtain a copy 37 of the mainline repository, explore the revision history, commit changes to 38 the tree, use branches, etc. An understanding of git's tools for the 39 rewriting of history (such as rebase) is also useful. Git comes with its 40 own terminology and concepts; a new user of git should know about refs, 41 remote branches, the index, fast-forward merges, pushes and pulls, detached 42 heads, etc. It can all be a little intimidating at the outset, but the 43 concepts are not that hard to grasp with a bit of study. 44 45 Using git to generate patches for submission by email can be a good 46 exercise while coming up to speed. 47 48 When you are ready to start putting up git trees for others to look at, you 49 will, of course, need a server that can be pulled from. Setting up such a 50 server with git-daemon is relatively straightforward if you have a system 51 which is accessible to the Internet. Otherwise, free, public hosting sites 52 (Github, for example) are starting to appear on the net. Established 53 developers can get an account on kernel.org, but those are not easy to come 54 by; see http://kernel.org/faq/ for more information. 55 56 The normal git workflow involves the use of a lot of branches. Each line 57 of development can be separated into a separate "topic branch" and 58 maintained independently. Branches in git are cheap, there is no reason to 59 not make free use of them. And, in any case, you should not do your 60 development in any branch which you intend to ask others to pull from. 61 Publicly-available branches should be created with care; merge in patches 62 from development branches when they are in complete form and ready to go - 63 not before. 64 65 Git provides some powerful tools which can allow you to rewrite your 66 development history. An inconvenient patch (one which breaks bisection, 67 say, or which has some other sort of obvious bug) can be fixed in place or 68 made to disappear from the history entirely. A patch series can be 69 rewritten as if it had been written on top of today's mainline, even though 70 you have been working on it for months. Changes can be transparently 71 shifted from one branch to another. And so on. Judicious use of git's 72 ability to revise history can help in the creation of clean patch sets with 73 fewer problems. 74 75 Excessive use of this capability can lead to other problems, though, beyond 76 a simple obsession for the creation of the perfect project history. 77 Rewriting history will rewrite the changes contained in that history, 78 turning a tested (hopefully) kernel tree into an untested one. But, beyond 79 that, developers cannot easily collaborate if they do not have a shared 80 view of the project history; if you rewrite history which other developers 81 have pulled into their repositories, you will make life much more difficult 82 for those developers. So a simple rule of thumb applies here: history 83 which has been exported to others should generally be seen as immutable 84 thereafter. 85 86 So, once you push a set of changes to your publicly-available server, those 87 changes should not be rewritten. Git will attempt to enforce this rule if 88 you try to push changes which do not result in a fast-forward merge 89 (i.e. changes which do not share the same history). It is possible to 90 override this check, and there may be times when it is necessary to rewrite 91 an exported tree. Moving changesets between trees to avoid conflicts in 92 linux-next is one example. But such actions should be rare. This is one 93 of the reasons why development should be done in private branches (which 94 can be rewritten if necessary) and only moved into public branches when 95 it's in a reasonably advanced state. 96 97 As the mainline (or other tree upon which a set of changes is based) 98 advances, it is tempting to merge with that tree to stay on the leading 99 edge. For a private branch, rebasing can be an easy way to keep up with 100 another tree, but rebasing is not an option once a tree is exported to the 101 world. Once that happens, a full merge must be done. Merging occasionally 102 makes good sense, but overly frequent merges can clutter the history 103 needlessly. Suggested technique in this case is to merge infrequently, and 104 generally only at specific release points (such as a mainline -rc 105 release). If you are nervous about specific changes, you can always 106 perform test merges in a private branch. The git "rerere" tool can be 107 useful in such situations; it remembers how merge conflicts were resolved 108 so that you don't have to do the same work twice. 109 110 One of the biggest recurring complaints about tools like git is this: the 111 mass movement of patches from one repository to another makes it easy to 112 slip in ill-advised changes which go into the mainline below the review 113 radar. Kernel developers tend to get unhappy when they see that kind of 114 thing happening; putting up a git tree with unreviewed or off-topic patches 115 can affect your ability to get trees pulled in the future. Quoting Linus: 116 117 You can send me patches, but for me to pull a git patch from you, I 118 need to know that you know what you're doing, and I need to be able 119 to trust things *without* then having to go and check every 120 individual change by hand. 121 122 (http://lwn.net/Articles/224135/). 123 124 To avoid this kind of situation, ensure that all patches within a given 125 branch stick closely to the associated topic; a "driver fixes" branch 126 should not be making changes to the core memory management code. And, most 127 importantly, do not use a git tree to bypass the review process. Post an 128 occasional summary of the tree to the relevant list, and, when the time is 129 right, request that the tree be included in linux-next. 130 131 If and when others start to send patches for inclusion into your tree, 132 don't forget to review them. Also ensure that you maintain the correct 133 authorship information; the git "am" tool does its best in this regard, but 134 you may have to add a "From:" line to the patch if it has been relayed to 135 you via a third party. 136 137 When requesting a pull, be sure to give all the relevant information: where 138 your tree is, what branch to pull, and what changes will result from the 139 pull. The git request-pull command can be helpful in this regard; it will 140 format the request as other developers expect, and will also check to be 141 sure that you have remembered to push those changes to the public server. 142 143 144 7.2: REVIEWING PATCHES 145 146 Some readers will certainly object to putting this section with "advanced 147 topics" on the grounds that even beginning kernel developers should be 148 reviewing patches. It is certainly true that there is no better way to 149 learn how to program in the kernel environment than by looking at code 150 posted by others. In addition, reviewers are forever in short supply; by 151 looking at code you can make a significant contribution to the process as a 152 whole. 153 154 Reviewing code can be an intimidating prospect, especially for a new kernel 155 developer who may well feel nervous about questioning code - in public - 156 which has been posted by those with more experience. Even code written by 157 the most experienced developers can be improved, though. Perhaps the best 158 piece of advice for reviewers (all reviewers) is this: phrase review 159 comments as questions rather than criticisms. Asking "how does the lock 160 get released in this path?" will always work better than stating "the 161 locking here is wrong." 162 163 Different developers will review code from different points of view. Some 164 are mostly concerned with coding style and whether code lines have trailing 165 white space. Others will focus primarily on whether the change implemented 166 by the patch as a whole is a good thing for the kernel or not. Yet others 167 will check for problematic locking, excessive stack usage, possible 168 security issues, duplication of code found elsewhere, adequate 169 documentation, adverse effects on performance, user-space ABI changes, etc. 170 All types of review, if they lead to better code going into the kernel, are 171 welcome and worthwhile. 172