Jesper Juhl | 8b7ecb1 | 2007-07-15 23:40:51 -0700 | [diff] [blame] | 1 | Linux Kernel patch submission checklist |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 2 | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
| 3 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 4 | Here are some basic things that developers should do if they want to see their |
| 5 | kernel patch submissions accepted more quickly. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 6 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 7 | These are all above and beyond the documentation that is provided in |
| 8 | Documentation/SubmittingPatches and elsewhere regarding submitting Linux |
| 9 | kernel patches. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 10 | |
| 11 | |
Randy Dunlap | b5247c8 | 2010-03-10 15:21:59 -0800 | [diff] [blame] | 12 | 1: If you use a facility then #include the file that defines/declares |
| 13 | that facility. Don't depend on other header files pulling in ones |
| 14 | that you use. |
| 15 | |
| 16 | 2: Builds cleanly with applicable or modified CONFIG options =y, =m, and |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 17 | =n. No gcc warnings/errors, no linker warnings/errors. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 18 | |
Randy Dunlap | b5247c8 | 2010-03-10 15:21:59 -0800 | [diff] [blame] | 19 | 2b: Passes allnoconfig, allmodconfig |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 20 | |
Randy Dunlap | 85f9642 | 2010-05-23 17:02:30 -0700 | [diff] [blame] | 21 | 2c: Builds successfully when using O=builddir |
| 22 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 23 | 3: Builds on multiple CPU architectures by using local cross-compile tools |
Randy Dunlap | 7de3369 | 2009-12-15 16:46:59 -0800 | [diff] [blame] | 24 | or some other build farm. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 25 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 26 | 4: ppc64 is a good architecture for cross-compilation checking because it |
| 27 | tends to use `unsigned long' for 64-bit quantities. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 28 | |
J. Bruce Fields | df24d9a | 2008-02-13 15:03:20 -0800 | [diff] [blame] | 29 | 5: Check your patch for general style as detailed in |
| 30 | Documentation/CodingStyle. Check for trivial violations with the |
| 31 | patch style checker prior to submission (scripts/checkpatch.pl). |
| 32 | You should be able to justify all violations that remain in |
| 33 | your patch. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 34 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 35 | 6: Any new or modified CONFIG options don't muck up the config menu. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 36 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 37 | 7: All new Kconfig options have help text. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 38 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 39 | 8: Has been carefully reviewed with respect to relevant Kconfig |
| 40 | combinations. This is very hard to get right with testing -- brainpower |
| 41 | pays off here. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 42 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 43 | 9: Check cleanly with sparse. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 44 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 45 | 10: Use 'make checkstack' and 'make namespacecheck' and fix any problems |
| 46 | that they find. Note: checkstack does not point out problems explicitly, |
| 47 | but any one function that uses more than 512 bytes on the stack is a |
| 48 | candidate for change. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 49 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 50 | 11: Include kernel-doc to document global kernel APIs. (Not required for |
| 51 | static functions, but OK there also.) Use 'make htmldocs' or 'make |
| 52 | mandocs' to check the kernel-doc and fix any issues. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 53 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 54 | 12: Has been tested with CONFIG_PREEMPT, CONFIG_DEBUG_PREEMPT, |
| 55 | CONFIG_DEBUG_SLAB, CONFIG_DEBUG_PAGEALLOC, CONFIG_DEBUG_MUTEXES, |
| 56 | CONFIG_DEBUG_SPINLOCK, CONFIG_DEBUG_SPINLOCK_SLEEP all simultaneously |
| 57 | enabled. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 58 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 59 | 13: Has been build- and runtime tested with and without CONFIG_SMP and |
| 60 | CONFIG_PREEMPT. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 61 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 62 | 14: If the patch affects IO/Disk, etc: has been tested with and without |
Bartlomiej Zolnierkiewicz | 90c699a | 2009-06-19 08:08:50 +0200 | [diff] [blame] | 63 | CONFIG_LBDAF. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 64 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 65 | 15: All codepaths have been exercised with all lockdep features enabled. |
Randy Dunlap | 915a56d | 2006-06-23 02:05:49 -0700 | [diff] [blame] | 66 | |
Andrew Morton | e54695a | 2006-07-10 04:45:42 -0700 | [diff] [blame] | 67 | 16: All new /proc entries are documented under Documentation/ |
| 68 | |
| 69 | 17: All new kernel boot parameters are documented in |
| 70 | Documentation/kernel-parameters.txt. |
| 71 | |
| 72 | 18: All new module parameters are documented with MODULE_PARM_DESC() |
Randy Dunlap | 1a036cd | 2006-09-29 02:01:26 -0700 | [diff] [blame] | 73 | |
| 74 | 19: All new userspace interfaces are documented in Documentation/ABI/. |
| 75 | See Documentation/ABI/README for more information. |
Michael Kerrisk | 1d992ce | 2008-10-03 15:23:44 -0700 | [diff] [blame] | 76 | Patches that change userspace interfaces should be CCed to |
| 77 | linux-api@vger.kernel.org. |
Andrew Morton | ce584f9 | 2006-09-30 23:27:39 -0700 | [diff] [blame] | 78 | |
| 79 | 20: Check that it all passes `make headers_check'. |
Andrew Morton | 0dd4e5b | 2006-12-10 02:18:56 -0800 | [diff] [blame] | 80 | |
| 81 | 21: Has been checked with injection of at least slab and page-allocation |
James Bowes | 80abe55 | 2007-10-20 03:13:09 +0200 | [diff] [blame] | 82 | failures. See Documentation/fault-injection/. |
Andrew Morton | 0dd4e5b | 2006-12-10 02:18:56 -0800 | [diff] [blame] | 83 | |
| 84 | If the new code is substantial, addition of subsystem-specific fault |
| 85 | injection might be appropriate. |
Andrew Morton | a517b9f | 2007-01-22 20:40:36 -0800 | [diff] [blame] | 86 | |
Andrew Morton | 9809169 | 2007-05-10 22:22:49 -0700 | [diff] [blame] | 87 | 22: Newly-added code has been compiled with `gcc -W' (use "make |
| 88 | EXTRA_CFLAGS=-W"). This will generate lots of noise, but is good for |
| 89 | finding bugs like "warning: comparison between signed and unsigned". |
Randy Dunlap | 244474b | 2007-02-28 20:12:35 -0800 | [diff] [blame] | 90 | |
| 91 | 23: Tested after it has been merged into the -mm patchset to make sure |
| 92 | that it still works with all of the other queued patches and various |
| 93 | changes in the VM, VFS, and other subsystems. |
Randy Dunlap | 8033fe6 | 2008-10-15 22:01:59 -0700 | [diff] [blame] | 94 | |
| 95 | 24: All memory barriers {e.g., barrier(), rmb(), wmb()} need a comment in the |
| 96 | source code that explains the logic of what they are doing and why. |
Randy Dunlap | 7de3369 | 2009-12-15 16:46:59 -0800 | [diff] [blame] | 97 | |
| 98 | 25: If any ioctl's are added by the patch, then also update |
| 99 | Documentation/ioctl/ioctl-number.txt. |
Randy Dunlap | 85f9642 | 2010-05-23 17:02:30 -0700 | [diff] [blame] | 100 | |
| 101 | 26: If your modified source code depends on or uses any of the kernel |
| 102 | APIs or features that are related to the following kconfig symbols, |
| 103 | then test multiple builds with the related kconfig symbols disabled |
| 104 | and/or =m (if that option is available) [not all of these at the |
| 105 | same time, just various/random combinations of them]: |
| 106 | |
| 107 | CONFIG_SMP, CONFIG_SYSFS, CONFIG_PROC_FS, CONFIG_INPUT, CONFIG_PCI, |
| 108 | CONFIG_BLOCK, CONFIG_PM, CONFIG_HOTPLUG, CONFIG_MAGIC_SYSRQ, |
| 109 | CONFIG_NET, CONFIG_INET=n (but latter with CONFIG_NET=y) |