blob: 681e426e24821043301d549f25401d4c3f6b48e8 [file] [log] [blame]
Jesper Juhl754c7972005-09-09 13:10:12 -07001
2 Applying Patches To The Linux Kernel
3 ------------------------------------
4
5 (Written by Jesper Juhl, August 2005)
6
7
8
9A frequently asked question on the Linux Kernel Mailing List is how to apply
10a patch to the kernel or, more specifically, what base kernel a patch for
11one of the many trees/branches should be applied to. Hopefully this document
12will explain this to you.
13
14In addition to explaining how to apply and revert patches, a brief
15description of the different kernel trees (and examples of how to apply
16their specific patches) is also provided.
17
18
19What is a patch?
20---
21 A patch is a small text document containing a delta of changes between two
22different versions of a source tree. Patches are created with the `diff'
23program.
24To correctly apply a patch you need to know what base it was generated from
25and what new version the patch will change the source tree into. These
26should both be present in the patch file metadata or be possible to deduce
27from the filename.
28
29
30How do I apply or revert a patch?
31---
32 You apply a patch with the `patch' program. The patch program reads a diff
33(or patch) file and makes the changes to the source tree described in it.
34
35Patches for the Linux kernel are generated relative to the parent directory
36holding the kernel source dir.
37
38This means that paths to files inside the patch file contain the name of the
39kernel source directories it was generated against (or some other directory
40names like "a/" and "b/").
41Since this is unlikely to match the name of the kernel source dir on your
42local machine (but is often useful info to see what version an otherwise
43unlabeled patch was generated against) you should change into your kernel
44source directory and then strip the first element of the path from filenames
45in the patch file when applying it (the -p1 argument to `patch' does this).
46
47To revert a previously applied patch, use the -R argument to patch.
48So, if you applied a patch like this:
49 patch -p1 < ../patch-x.y.z
50
51You can revert (undo) it like this:
52 patch -R -p1 < ../patch-x.y.z
53
54
55How do I feed a patch/diff file to `patch'?
56---
57 This (as usual with Linux and other UNIX like operating systems) can be
58done in several different ways.
59In all the examples below I feed the file (in uncompressed form) to patch
60via stdin using the following syntax:
61 patch -p1 < path/to/patch-x.y.z
62
63If you just want to be able to follow the examples below and don't want to
64know of more than one way to use patch, then you can stop reading this
65section here.
66
67Patch can also get the name of the file to use via the -i argument, like
68this:
69 patch -p1 -i path/to/patch-x.y.z
70
71If your patch file is compressed with gzip or bzip2 and you don't want to
72uncompress it before applying it, then you can feed it to patch like this
73instead:
74 zcat path/to/patch-x.y.z.gz | patch -p1
75 bzcat path/to/patch-x.y.z.bz2 | patch -p1
76
77If you wish to uncompress the patch file by hand first before applying it
78(what I assume you've done in the examples below), then you simply run
79gunzip or bunzip2 on the file - like this:
80 gunzip patch-x.y.z.gz
81 bunzip2 patch-x.y.z.bz2
82
83Which will leave you with a plain text patch-x.y.z file that you can feed to
84patch via stdin or the -i argument, as you prefer.
85
86A few other nice arguments for patch are -s which causes patch to be silent
87except for errors which is nice to prevent errors from scrolling out of the
88screen too fast, and --dry-run which causes patch to just print a listing of
89what would happen, but doesn't actually make any changes. Finally --verbose
90tells patch to print more information about the work being done.
91
92
93Common errors when patching
94---
95 When patch applies a patch file it attempts to verify the sanity of the
96file in different ways.
97Checking that the file looks like a valid patch file, checking the code
98around the bits being modified matches the context provided in the patch are
99just two of the basic sanity checks patch does.
100
101If patch encounters something that doesn't look quite right it has two
102options. It can either refuse to apply the changes and abort or it can try
103to find a way to make the patch apply with a few minor changes.
104
105One example of something that's not 'quite right' that patch will attempt to
106fix up is if all the context matches, the lines being changed match, but the
107line numbers are different. This can happen, for example, if the patch makes
108a change in the middle of the file but for some reasons a few lines have
109been added or removed near the beginning of the file. In that case
110everything looks good it has just moved up or down a bit, and patch will
111usually adjust the line numbers and apply the patch.
112
113Whenever patch applies a patch that it had to modify a bit to make it fit
114it'll tell you about it by saying the patch applied with 'fuzz'.
115You should be wary of such changes since even though patch probably got it
116right it doesn't /always/ get it right, and the result will sometimes be
117wrong.
118
119When patch encounters a change that it can't fix up with fuzz it rejects it
120outright and leaves a file with a .rej extension (a reject file). You can
121read this file to see exactely what change couldn't be applied, so you can
122go fix it up by hand if you wish.
123
124If you don't have any third party patches applied to your kernel source, but
125only patches from kernel.org and you apply the patches in the correct order,
126and have made no modifications yourself to the source files, then you should
127never see a fuzz or reject message from patch. If you do see such messages
128anyway, then there's a high risk that either your local source tree or the
129patch file is corrupted in some way. In that case you should probably try
130redownloading the patch and if things are still not OK then you'd be advised
131to start with a fresh tree downloaded in full from kernel.org.
132
133Let's look a bit more at some of the messages patch can produce.
134
135If patch stops and presents a "File to patch:" prompt, then patch could not
136find a file to be patched. Most likely you forgot to specify -p1 or you are
137in the wrong directory. Less often, you'll find patches that need to be
138applied with -p0 instead of -p1 (reading the patch file should reveal if
139this is the case - if so, then this is an error by the person who created
140the patch but is not fatal).
141
142If you get "Hunk #2 succeeded at 1887 with fuzz 2 (offset 7 lines)." or a
143message similar to that, then it means that patch had to adjust the location
144of the change (in this example it needed to move 7 lines from where it
145expected to make the change to make it fit).
146The resulting file may or may not be OK, depending on the reason the file
147was different than expected.
148This often happens if you try to apply a patch that was generated against a
149different kernel version than the one you are trying to patch.
150
151If you get a message like "Hunk #3 FAILED at 2387.", then it means that the
152patch could not be applied correctly and the patch program was unable to
153fuzz its way through. This will generate a .rej file with the change that
154caused the patch to fail and also a .orig file showing you the original
155content that couldn't be changed.
156
157If you get "Reversed (or previously applied) patch detected! Assume -R? [n]"
158then patch detected that the change contained in the patch seems to have
159already been made.
160If you actually did apply this patch previously and you just re-applied it
161in error, then just say [n]o and abort this patch. If you applied this patch
162previously and actually intended to revert it, but forgot to specify -R,
163then you can say [y]es here to make patch revert it for you.
164This can also happen if the creator of the patch reversed the source and
165destination directories when creating the patch, and in that case reverting
166the patch will in fact apply it.
167
168A message similar to "patch: **** unexpected end of file in patch" or "patch
169unexpectedly ends in middle of line" means that patch could make no sense of
170the file you fed to it. Either your download is broken or you tried to feed
171patch a compressed patch file without uncompressing it first.
172
173As I already mentioned above, these errors should never happen if you apply
174a patch from kernel.org to the correct version of an unmodified source tree.
175So if you get these errors with kernel.org patches then you should probably
176assume that either your patch file or your tree is broken and I'd advice you
177to start over with a fresh download of a full kernel tree and the patch you
178wish to apply.
179
180
181Are there any alternatives to `patch'?
182---
183 Yes there are alternatives. You can use the `interdiff' program
184(http://cyberelk.net/tim/patchutils/) to generate a patch representing the
185differences between two patches and then apply the result.
186This will let you move from something like 2.6.12.2 to 2.6.12.3 in a single
187step. The -z flag to interdiff will even let you feed it patches in gzip or
188bzip2 compressed form directly without the use of zcat or bzcat or manual
189decompression.
190
191Here's how you'd go from 2.6.12.2 to 2.6.12.3 in a single step:
192 interdiff -z ../patch-2.6.12.2.bz2 ../patch-2.6.12.3.gz | patch -p1
193
194Although interdiff may save you a step or two you are generally advised to
195do the additional steps since interdiff can get things wrong in some cases.
196
197 Another alternative is `ketchup', which is a python script for automatic
198downloading and applying of patches (http://www.selenic.com/ketchup/).
199
200Other nice tools are diffstat which shows a summary of changes made by a
201patch, lsdiff which displays a short listing of affected files in a patch
202file, along with (optionally) the line numbers of the start of each patch
203and grepdiff which displays a list of the files modified by a patch where
204the patch contains a given regular expression.
205
206
207Where can I download the patches?
208---
209 The patches are available at http://kernel.org/
210Most recent patches are linked from the front page, but they also have
211specific homes.
212
213The 2.6.x.y (-stable) and 2.6.x patches live at
214 ftp://ftp.kernel.org/pub/linux/kernel/v2.6/
215
216The -rc patches live at
217 ftp://ftp.kernel.org/pub/linux/kernel/v2.6/testing/
218
219The -git patches live at
220 ftp://ftp.kernel.org/pub/linux/kernel/v2.6/snapshots/
221
222The -mm kernels live at
223 ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/
224
225In place of ftp.kernel.org you can use ftp.cc.kernel.org, where cc is a
226country code. This way you'll be downloading from a mirror site that's most
227likely geographically closer to you, resulting in faster downloads for you,
228less bandwidth used globally and less load on the main kernel.org servers -
229these are good things, do use mirrors when possible.
230
231
232The 2.6.x kernels
233---
234 These are the base stable releases released by Linus. The highest numbered
235release is the most recent.
236
237If regressions or other serious flaws are found then a -stable fix patch
238will be released (see below) on top of this base. Once a new 2.6.x base
239kernel is released, a patch is made available that is a delta between the
240previous 2.6.x kernel and the new one.
241
242To apply a patch moving from 2.6.11 to 2.6.12 you'd do the following (note
243that such patches do *NOT* apply on top of 2.6.x.y kernels but on top of the
244base 2.6.x kernel - if you need to move from 2.6.x.y to 2.6.x+1 you need to
245first revert the 2.6.x.y patch).
246
247Here are some examples:
248
249# moving from 2.6.11 to 2.6.12
250$ cd ~/linux-2.6.11 # change to kernel source dir
251$ patch -p1 < ../patch-2.6.12 # apply the 2.6.12 patch
252$ cd ..
253$ mv linux-2.6.11 linux-2.6.12 # rename source dir
254
255# moving from 2.6.11.1 to 2.6.12
256$ cd ~/linux-2.6.11.1 # change to kernel source dir
257$ patch -p1 -R < ../patch-2.6.11.1 # revert the 2.6.11.1 patch
258 # source dir is now 2.6.11
259$ patch -p1 < ../patch-2.6.12 # apply new 2.6.12 patch
260$ cd ..
261$ mv linux-2.6.11.1 inux-2.6.12 # rename source dir
262
263
264The 2.6.x.y kernels
265---
266 Kernels with 4 digit versions are -stable kernels. They contain small(ish)
267critical fixes for security problems or significant regressions discovered
268in a given 2.6.x kernel.
269
270This is the recommended branch for users who want the most recent stable
271kernel and are not interested in helping test development/experimental
272versions.
273
274If no 2.6.x.y kernel is available, then the highest numbered 2.6.x kernel is
275the current stable kernel.
276
277These patches are not incremental, meaning that for example the 2.6.12.3
278patch does not apply on top of the 2.6.12.2 kernel source, but rather on top
279of the base 2.6.12 kernel source.
280So, in order to apply the 2.6.12.3 patch to your existing 2.6.12.2 kernel
281source you have to first back out the 2.6.12.2 patch (so you are left with a
282base 2.6.12 kernel source) and then apply the new 2.6.12.3 patch.
283
284Here's a small example:
285
286$ cd ~/linux-2.6.12.2 # change into the kernel source dir
287$ patch -p1 -R < ../patch-2.6.12.2 # revert the 2.6.12.2 patch
288$ patch -p1 < ../patch-2.6.12.3 # apply the new 2.6.12.3 patch
289$ cd ..
290$ mv linux-2.6.12.2 linux-2.6.12.3 # rename the kernel source dir
291
292
293The -rc kernels
294---
295 These are release-candidate kernels. These are development kernels released
296by Linus whenever he deems the current git (the kernel's source management
297tool) tree to be in a reasonably sane state adequate for testing.
298
299These kernels are not stable and you should expect occasional breakage if
300you intend to run them. This is however the most stable of the main
301development branches and is also what will eventually turn into the next
302stable kernel, so it is important that it be tested by as many people as
303possible.
304
305This is a good branch to run for people who want to help out testing
306development kernels but do not want to run some of the really experimental
307stuff (such people should see the sections about -git and -mm kernels below).
308
309The -rc patches are not incremental, they apply to a base 2.6.x kernel, just
310like the 2.6.x.y patches described above. The kernel version before the -rcN
311suffix denotes the version of the kernel that this -rc kernel will eventually
312turn into.
313So, 2.6.13-rc5 means that this is the fifth release candidate for the 2.6.13
314kernel and the patch should be applied on top of the 2.6.12 kernel source.
315
316Here are 3 examples of how to apply these patches:
317
318# first an example of moving from 2.6.12 to 2.6.13-rc3
319$ cd ~/linux-2.6.12 # change into the 2.6.12 source dir
320$ patch -p1 < ../patch-2.6.13-rc3 # apply the 2.6.13-rc3 patch
321$ cd ..
322$ mv linux-2.6.12 linux-2.6.13-rc3 # rename the source dir
323
324# now let's move from 2.6.13-rc3 to 2.6.13-rc5
325$ cd ~/linux-2.6.13-rc3 # change into the 2.6.13-rc3 dir
326$ patch -p1 -R < ../patch-2.6.13-rc3 # revert the 2.6.13-rc3 patch
327$ patch -p1 < ../patch-2.6.13-rc5 # apply the new 2.6.13-rc5 patch
328$ cd ..
329$ mv linux-2.6.13-rc3 linux-2.6.13-rc5 # rename the source dir
330
331# finally let's try and move from 2.6.12.3 to 2.6.13-rc5
332$ cd ~/linux-2.6.12.3 # change to the kernel source dir
333$ patch -p1 -R < ../patch-2.6.12.3 # revert the 2.6.12.3 patch
334$ patch -p1 < ../patch-2.6.13-rc5 # apply new 2.6.13-rc5 patch
335$ cd ..
336$ mv linux-2.6.12.3 linux-2.6.13-rc5 # rename the kernel source dir
337
338
339The -git kernels
340---
341 These are daily snapshots of Linus' kernel tree (managed in a git
342repository, hence the name).
343
344These patches are usually released daily and represent the current state of
345Linus' tree. They are more experimental than -rc kernels since they are
346generated automatically without even a cursory glance to see if they are
347sane.
348
349-git patches are not incremental and apply either to a base 2.6.x kernel or
350a base 2.6.x-rc kernel - you can see which from their name.
351A patch named 2.6.12-git1 applies to the 2.6.12 kernel source and a patch
352named 2.6.13-rc3-git2 applies to the source of the 2.6.13-rc3 kernel.
353
354Here are some examples of how to apply these patches:
355
356# moving from 2.6.12 to 2.6.12-git1
357$ cd ~/linux-2.6.12 # change to the kernel source dir
358$ patch -p1 < ../patch-2.6.12-git1 # apply the 2.6.12-git1 patch
359$ cd ..
360$ mv linux-2.6.12 linux-2.6.12-git1 # rename the kernel source dir
361
362# moving from 2.6.12-git1 to 2.6.13-rc2-git3
363$ cd ~/linux-2.6.12-git1 # change to the kernel source dir
364$ patch -p1 -R < ../patch-2.6.12-git1 # revert the 2.6.12-git1 patch
365 # we now have a 2.6.12 kernel
366$ patch -p1 < ../patch-2.6.13-rc2 # apply the 2.6.13-rc2 patch
367 # the kernel is now 2.6.13-rc2
368$ patch -p1 < ../patch-2.6.13-rc2-git3 # apply the 2.6.13-rc2-git3 patch
369 # the kernel is now 2.6.13-rc2-git3
370$ cd ..
371$ mv linux-2.6.12-git1 linux-2.6.13-rc2-git3 # rename source dir
372
373
374The -mm kernels
375---
376 These are experimental kernels released by Andrew Morton.
377
378The -mm tree serves as a sort of proving ground for new features and other
379experimental patches.
380Once a patch has proved its worth in -mm for a while Andrew pushes it on to
381Linus for inclusion in mainline.
382
383Although it's encouraged that patches flow to Linus via the -mm tree, this
384is not always enforced.
385Subsystem maintainers (or individuals) sometimes push their patches directly
386to Linus, even though (or after) they have been merged and tested in -mm (or
387sometimes even without prior testing in -mm).
388
389You should generally strive to get your patches into mainline via -mm to
390ensure maximum testing.
391
392This branch is in constant flux and contains many experimental features, a
393lot of debugging patches not appropriate for mainline etc and is the most
394experimental of the branches described in this document.
395
396These kernels are not appropriate for use on systems that are supposed to be
397stable and they are more risky to run than any of the other branches (make
398sure you have up-to-date backups - that goes for any experimental kernel but
399even more so for -mm kernels).
400
401These kernels in addition to all the other experimental patches they contain
402usually also contain any changes in the mainline -git kernels available at
403the time of release.
404
405Testing of -mm kernels is greatly appreciated since the whole point of the
406tree is to weed out regressions, crashes, data corruption bugs, build
407breakage (and any other bug in general) before changes are merged into the
408more stable mainline Linus tree.
409But testers of -mm should be aware that breakage in this tree is more common
410than in any other tree.
411
412The -mm kernels are not released on a fixed schedule, but usually a few -mm
413kernels are released in between each -rc kernel (1 to 3 is common).
414The -mm kernels apply to either a base 2.6.x kernel (when no -rc kernels
415have been released yet) or to a Linus -rc kernel.
416
417Here are some examples of applying the -mm patches:
418
419# moving from 2.6.12 to 2.6.12-mm1
420$ cd ~/linux-2.6.12 # change to the 2.6.12 source dir
421$ patch -p1 < ../2.6.12-mm1 # apply the 2.6.12-mm1 patch
422$ cd ..
423$ mv linux-2.6.12 linux-2.6.12-mm1 # rename the source appropriately
424
425# moving from 2.6.12-mm1 to 2.6.13-rc3-mm3
426$ cd ~/linux-2.6.12-mm1
427$ patch -p1 -R < ../2.6.12-mm1 # revert the 2.6.12-mm1 patch
428 # we now have a 2.6.12 source
429$ patch -p1 < ../patch-2.6.13-rc3 # apply the 2.6.13-rc3 patch
430 # we now have a 2.6.13-rc3 source
431$ patch -p1 < ../2.6.13-rc3-mm3 # apply the 2.6.13-rc3-mm3 patch
432$ cd ..
433$ mv linux-2.6.12-mm1 linux-2.6.13-rc3-mm3 # rename the source dir
434
435
436This concludes this list of explanations of the various kernel trees and I
437hope you are now crystal clear on how to apply the various patches and help
438testing the kernel.
439