linux/Documentation/applying-patches.txt
<<
>>
Prefs
   1.. _applying_patches:
   2
   3Applying Patches To The Linux Kernel
   4++++++++++++++++++++++++++++++++++++
   5
   6Original by:
   7        Jesper Juhl, August 2005
   8
   9Last update:
  10        2016-09-14
  11
  12
  13A frequently asked question on the Linux Kernel Mailing List is how to apply
  14a patch to the kernel or, more specifically, what base kernel a patch for
  15one of the many trees/branches should be applied to. Hopefully this document
  16will explain this to you.
  17
  18In addition to explaining how to apply and revert patches, a brief
  19description of the different kernel trees (and examples of how to apply
  20their specific patches) is also provided.
  21
  22
  23What is a patch?
  24================
  25
  26A patch is a small text document containing a delta of changes between two
  27different versions of a source tree. Patches are created with the ``diff``
  28program.
  29
  30To correctly apply a patch you need to know what base it was generated from
  31and what new version the patch will change the source tree into. These
  32should both be present in the patch file metadata or be possible to deduce
  33from the filename.
  34
  35
  36How do I apply or revert a patch?
  37=================================
  38
  39You apply a patch with the ``patch`` program. The patch program reads a diff
  40(or patch) file and makes the changes to the source tree described in it.
  41
  42Patches for the Linux kernel are generated relative to the parent directory
  43holding the kernel source dir.
  44
  45This means that paths to files inside the patch file contain the name of the
  46kernel source directories it was generated against (or some other directory
  47names like "a/" and "b/").
  48
  49Since this is unlikely to match the name of the kernel source dir on your
  50local machine (but is often useful info to see what version an otherwise
  51unlabeled patch was generated against) you should change into your kernel
  52source directory and then strip the first element of the path from filenames
  53in the patch file when applying it (the ``-p1`` argument to ``patch`` does
  54this).
  55
  56To revert a previously applied patch, use the -R argument to patch.
  57So, if you applied a patch like this::
  58
  59        patch -p1 < ../patch-x.y.z
  60
  61You can revert (undo) it like this::
  62
  63        patch -R -p1 < ../patch-x.y.z
  64
  65
  66How do I feed a patch/diff file to ``patch``?
  67=============================================
  68
  69This (as usual with Linux and other UNIX like operating systems) can be
  70done in several different ways.
  71
  72In all the examples below I feed the file (in uncompressed form) to patch
  73via stdin using the following syntax::
  74
  75        patch -p1 < path/to/patch-x.y.z
  76
  77If you just want to be able to follow the examples below and don't want to
  78know of more than one way to use patch, then you can stop reading this
  79section here.
  80
  81Patch can also get the name of the file to use via the -i argument, like
  82this::
  83
  84        patch -p1 -i path/to/patch-x.y.z
  85
  86If your patch file is compressed with gzip or xz and you don't want to
  87uncompress it before applying it, then you can feed it to patch like this
  88instead::
  89
  90        xzcat path/to/patch-x.y.z.xz | patch -p1
  91        bzcat path/to/patch-x.y.z.gz | patch -p1
  92
  93If you wish to uncompress the patch file by hand first before applying it
  94(what I assume you've done in the examples below), then you simply run
  95gunzip or xz on the file -- like this::
  96
  97        gunzip patch-x.y.z.gz
  98        xz -d patch-x.y.z.xz
  99
 100Which will leave you with a plain text patch-x.y.z file that you can feed to
 101patch via stdin or the ``-i`` argument, as you prefer.
 102
 103A few other nice arguments for patch are ``-s`` which causes patch to be silent
 104except for errors which is nice to prevent errors from scrolling out of the
 105screen too fast, and ``--dry-run`` which causes patch to just print a listing of
 106what would happen, but doesn't actually make any changes. Finally ``--verbose``
 107tells patch to print more information about the work being done.
 108
 109
 110Common errors when patching
 111===========================
 112
 113When patch applies a patch file it attempts to verify the sanity of the
 114file in different ways.
 115
 116Checking that the file looks like a valid patch file and checking the code
 117around the bits being modified matches the context provided in the patch are
 118just two of the basic sanity checks patch does.
 119
 120If patch encounters something that doesn't look quite right it has two
 121options. It can either refuse to apply the changes and abort or it can try
 122to find a way to make the patch apply with a few minor changes.
 123
 124One example of something that's not 'quite right' that patch will attempt to
 125fix up is if all the context matches, the lines being changed match, but the
 126line numbers are different. This can happen, for example, if the patch makes
 127a change in the middle of the file but for some reasons a few lines have
 128been added or removed near the beginning of the file. In that case
 129everything looks good it has just moved up or down a bit, and patch will
 130usually adjust the line numbers and apply the patch.
 131
 132Whenever patch applies a patch that it had to modify a bit to make it fit
 133it'll tell you about it by saying the patch applied with **fuzz**.
 134You should be wary of such changes since even though patch probably got it
 135right it doesn't /always/ get it right, and the result will sometimes be
 136wrong.
 137
 138When patch encounters a change that it can't fix up with fuzz it rejects it
 139outright and leaves a file with a ``.rej`` extension (a reject file). You can
 140read this file to see exactly what change couldn't be applied, so you can
 141go fix it up by hand if you wish.
 142
 143If you don't have any third-party patches applied to your kernel source, but
 144only patches from kernel.org and you apply the patches in the correct order,
 145and have made no modifications yourself to the source files, then you should
 146never see a fuzz or reject message from patch. If you do see such messages
 147anyway, then there's a high risk that either your local source tree or the
 148patch file is corrupted in some way. In that case you should probably try
 149re-downloading the patch and if things are still not OK then you'd be advised
 150to start with a fresh tree downloaded in full from kernel.org.
 151
 152Let's look a bit more at some of the messages patch can produce.
 153
 154If patch stops and presents a ``File to patch:`` prompt, then patch could not
 155find a file to be patched. Most likely you forgot to specify -p1 or you are
 156in the wrong directory. Less often, you'll find patches that need to be
 157applied with ``-p0`` instead of ``-p1`` (reading the patch file should reveal if
 158this is the case -- if so, then this is an error by the person who created
 159the patch but is not fatal).
 160
 161If you get ``Hunk #2 succeeded at 1887 with fuzz 2 (offset 7 lines).`` or a
 162message similar to that, then it means that patch had to adjust the location
 163of the change (in this example it needed to move 7 lines from where it
 164expected to make the change to make it fit).
 165
 166The resulting file may or may not be OK, depending on the reason the file
 167was different than expected.
 168
 169This often happens if you try to apply a patch that was generated against a
 170different kernel version than the one you are trying to patch.
 171
 172If you get a message like ``Hunk #3 FAILED at 2387.``, then it means that the
 173patch could not be applied correctly and the patch program was unable to
 174fuzz its way through. This will generate a ``.rej`` file with the change that
 175caused the patch to fail and also a ``.orig`` file showing you the original
 176content that couldn't be changed.
 177
 178If you get ``Reversed (or previously applied) patch detected!  Assume -R? [n]``
 179then patch detected that the change contained in the patch seems to have
 180already been made.
 181
 182If you actually did apply this patch previously and you just re-applied it
 183in error, then just say [n]o and abort this patch. If you applied this patch
 184previously and actually intended to revert it, but forgot to specify -R,
 185then you can say [**y**]es here to make patch revert it for you.
 186
 187This can also happen if the creator of the patch reversed the source and
 188destination directories when creating the patch, and in that case reverting
 189the patch will in fact apply it.
 190
 191A message similar to ``patch: **** unexpected end of file in patch`` or
 192``patch unexpectedly ends in middle of line`` means that patch could make no
 193sense of the file you fed to it. Either your download is broken, you tried to
 194feed patch a compressed patch file without uncompressing it first, or the patch
 195file that you are using has been mangled by a mail client or mail transfer
 196agent along the way somewhere, e.g., by splitting a long line into two lines.
 197Often these warnings can easily be fixed by joining (concatenating) the
 198two lines that had been split.
 199
 200As I already mentioned above, these errors should never happen if you apply
 201a patch from kernel.org to the correct version of an unmodified source tree.
 202So if you get these errors with kernel.org patches then you should probably
 203assume that either your patch file or your tree is broken and I'd advise you
 204to start over with a fresh download of a full kernel tree and the patch you
 205wish to apply.
 206
 207
 208Are there any alternatives to ``patch``?
 209========================================
 210
 211
 212Yes there are alternatives.
 213
 214You can use the ``interdiff`` program (http://cyberelk.net/tim/patchutils/) to
 215generate a patch representing the differences between two patches and then
 216apply the result.
 217
 218This will let you move from something like 4.7.2 to 4.7.3 in a single
 219step. The -z flag to interdiff will even let you feed it patches in gzip or
 220bzip2 compressed form directly without the use of zcat or bzcat or manual
 221decompression.
 222
 223Here's how you'd go from 4.7.2 to 4.7.3 in a single step::
 224
 225        interdiff -z ../patch-4.7.2.gz ../patch-4.7.3.gz | patch -p1
 226
 227Although interdiff may save you a step or two you are generally advised to
 228do the additional steps since interdiff can get things wrong in some cases.
 229
 230Another alternative is ``ketchup``, which is a python script for automatic
 231downloading and applying of patches (http://www.selenic.com/ketchup/).
 232
 233Other nice tools are diffstat, which shows a summary of changes made by a
 234patch; lsdiff, which displays a short listing of affected files in a patch
 235file, along with (optionally) the line numbers of the start of each patch;
 236and grepdiff, which displays a list of the files modified by a patch where
 237the patch contains a given regular expression.
 238
 239
 240Where can I download the patches?
 241=================================
 242
 243The patches are available at http://kernel.org/
 244Most recent patches are linked from the front page, but they also have
 245specific homes.
 246
 247The 4.x.y (-stable) and 4.x patches live at
 248
 249        ftp://ftp.kernel.org/pub/linux/kernel/v4.x/
 250
 251The -rc patches live at
 252
 253        ftp://ftp.kernel.org/pub/linux/kernel/v4.x/testing/
 254
 255In place of ``ftp.kernel.org`` you can use ``ftp.cc.kernel.org``, where cc is a
 256country code. This way you'll be downloading from a mirror site that's most
 257likely geographically closer to you, resulting in faster downloads for you,
 258less bandwidth used globally and less load on the main kernel.org servers --
 259these are good things, so do use mirrors when possible.
 260
 261
 262The 4.x kernels
 263===============
 264
 265These are the base stable releases released by Linus. The highest numbered
 266release is the most recent.
 267
 268If regressions or other serious flaws are found, then a -stable fix patch
 269will be released (see below) on top of this base. Once a new 4.x base
 270kernel is released, a patch is made available that is a delta between the
 271previous 4.x kernel and the new one.
 272
 273To apply a patch moving from 4.6 to 4.7, you'd do the following (note
 274that such patches do **NOT** apply on top of 4.x.y kernels but on top of the
 275base 4.x kernel -- if you need to move from 4.x.y to 4.x+1 you need to
 276first revert the 4.x.y patch).
 277
 278Here are some examples::
 279
 280        # moving from 4.6 to 4.7
 281
 282        $ cd ~/linux-4.6                # change to kernel source dir
 283        $ patch -p1 < ../patch-4.7      # apply the 4.7 patch
 284        $ cd ..
 285        $ mv linux-4.6 linux-4.7        # rename source dir
 286
 287        # moving from 4.6.1 to 4.7
 288
 289        $ cd ~/linux-4.6.1              # change to kernel source dir
 290        $ patch -p1 -R < ../patch-4.6.1 # revert the 4.6.1 patch
 291                                        # source dir is now 4.6
 292        $ patch -p1 < ../patch-4.7      # apply new 4.7 patch
 293        $ cd ..
 294        $ mv linux-4.6.1 linux-4.7      # rename source dir
 295
 296
 297The 4.x.y kernels
 298=================
 299
 300Kernels with 3-digit versions are -stable kernels. They contain small(ish)
 301critical fixes for security problems or significant regressions discovered
 302in a given 4.x kernel.
 303
 304This is the recommended branch for users who want the most recent stable
 305kernel and are not interested in helping test development/experimental
 306versions.
 307
 308If no 4.x.y kernel is available, then the highest numbered 4.x kernel is
 309the current stable kernel.
 310
 311.. note::
 312
 313 The -stable team usually do make incremental patches available as well
 314 as patches against the latest mainline release, but I only cover the
 315 non-incremental ones below. The incremental ones can be found at
 316 ftp://ftp.kernel.org/pub/linux/kernel/v4.x/incr/
 317
 318These patches are not incremental, meaning that for example the 4.7.3
 319patch does not apply on top of the 4.7.2 kernel source, but rather on top
 320of the base 4.7 kernel source.
 321
 322So, in order to apply the 4.7.3 patch to your existing 4.7.2 kernel
 323source you have to first back out the 4.7.2 patch (so you are left with a
 324base 4.7 kernel source) and then apply the new 4.7.3 patch.
 325
 326Here's a small example::
 327
 328        $ cd ~/linux-4.7.2              # change to the kernel source dir
 329        $ patch -p1 -R < ../patch-4.7.2 # revert the 4.7.2 patch
 330        $ patch -p1 < ../patch-4.7.3    # apply the new 4.7.3 patch
 331        $ cd ..
 332        $ mv linux-4.7.2 linux-4.7.3    # rename the kernel source dir
 333
 334The -rc kernels
 335===============
 336
 337These are release-candidate kernels. These are development kernels released
 338by Linus whenever he deems the current git (the kernel's source management
 339tool) tree to be in a reasonably sane state adequate for testing.
 340
 341These kernels are not stable and you should expect occasional breakage if
 342you intend to run them. This is however the most stable of the main
 343development branches and is also what will eventually turn into the next
 344stable kernel, so it is important that it be tested by as many people as
 345possible.
 346
 347This is a good branch to run for people who want to help out testing
 348development kernels but do not want to run some of the really experimental
 349stuff (such people should see the sections about -git and -mm kernels below).
 350
 351The -rc patches are not incremental, they apply to a base 4.x kernel, just
 352like the 4.x.y patches described above. The kernel version before the -rcN
 353suffix denotes the version of the kernel that this -rc kernel will eventually
 354turn into.
 355
 356So, 4.8-rc5 means that this is the fifth release candidate for the 4.8
 357kernel and the patch should be applied on top of the 4.7 kernel source.
 358
 359Here are 3 examples of how to apply these patches::
 360
 361        # first an example of moving from 4.7 to 4.8-rc3
 362
 363        $ cd ~/linux-4.7                        # change to the 4.7 source dir
 364        $ patch -p1 < ../patch-4.8-rc3          # apply the 4.8-rc3 patch
 365        $ cd ..
 366        $ mv linux-4.7 linux-4.8-rc3            # rename the source dir
 367
 368        # now let's move from 4.8-rc3 to 4.8-rc5
 369
 370        $ cd ~/linux-4.8-rc3                    # change to the 4.8-rc3 dir
 371        $ patch -p1 -R < ../patch-4.8-rc3       # revert the 4.8-rc3 patch
 372        $ patch -p1 < ../patch-4.8-rc5          # apply the new 4.8-rc5 patch
 373        $ cd ..
 374        $ mv linux-4.8-rc3 linux-4.8-rc5        # rename the source dir
 375
 376        # finally let's try and move from 4.7.3 to 4.8-rc5
 377
 378        $ cd ~/linux-4.7.3                      # change to the kernel source dir
 379        $ patch -p1 -R < ../patch-4.7.3         # revert the 4.7.3 patch
 380        $ patch -p1 < ../patch-4.8-rc5          # apply new 4.8-rc5 patch
 381        $ cd ..
 382        $ mv linux-4.7.3 linux-4.8-rc5          # rename the kernel source dir
 383
 384
 385The -git kernels
 386================
 387
 388These are daily snapshots of Linus' kernel tree (managed in a git
 389repository, hence the name).
 390
 391These patches are usually released daily and represent the current state of
 392Linus's tree. They are more experimental than -rc kernels since they are
 393generated automatically without even a cursory glance to see if they are
 394sane.
 395
 396-git patches are not incremental and apply either to a base 4.x kernel or
 397a base 4.x-rc kernel -- you can see which from their name.
 398A patch named 4.7-git1 applies to the 4.7 kernel source and a patch
 399named 4.8-rc3-git2 applies to the source of the 4.8-rc3 kernel.
 400
 401Here are some examples of how to apply these patches::
 402
 403        # moving from 4.7 to 4.7-git1
 404
 405        $ cd ~/linux-4.7                        # change to the kernel source dir
 406        $ patch -p1 < ../patch-4.7-git1         # apply the 4.7-git1 patch
 407        $ cd ..
 408        $ mv linux-4.7 linux-4.7-git1           # rename the kernel source dir
 409
 410        # moving from 4.7-git1 to 4.8-rc2-git3
 411
 412        $ cd ~/linux-4.7-git1                   # change to the kernel source dir
 413        $ patch -p1 -R < ../patch-4.7-git1      # revert the 4.7-git1 patch
 414                                                # we now have a 4.7 kernel
 415        $ patch -p1 < ../patch-4.8-rc2          # apply the 4.8-rc2 patch
 416                                                # the kernel is now 4.8-rc2
 417        $ patch -p1 < ../patch-4.8-rc2-git3     # apply the 4.8-rc2-git3 patch
 418                                                # the kernel is now 4.8-rc2-git3
 419        $ cd ..
 420        $ mv linux-4.7-git1 linux-4.8-rc2-git3  # rename source dir
 421
 422
 423The -mm patches and the linux-next tree
 424=======================================
 425
 426The -mm patches are experimental patches released by Andrew Morton.
 427
 428In the past, -mm tree were used to also test subsystem patches, but this
 429function is now done via the
 430:ref:`linux-next <https://www.kernel.org/doc/man-pages/linux-next.html>`
 431tree. The Subsystem maintainers push their patches first to linux-next,
 432and, during the merge window, sends them directly to Linus.
 433
 434The -mm patches serve as a sort of proving ground for new features and other
 435experimental patches that aren't merged via a subsystem tree.
 436Once such patches has proved its worth in -mm for a while Andrew pushes
 437it on to Linus for inclusion in mainline.
 438
 439The linux-next tree is daily updated, and includes the -mm patches.
 440Both are in constant flux and contains many experimental features, a
 441lot of debugging patches not appropriate for mainline etc., and is the most
 442experimental of the branches described in this document.
 443
 444These patches are not appropriate for use on systems that are supposed to be
 445stable and they are more risky to run than any of the other branches (make
 446sure you have up-to-date backups -- that goes for any experimental kernel but
 447even more so for -mm patches or using a Kernel from the linux-next tree).
 448
 449Testing of -mm patches and linux-next is greatly appreciated since the whole
 450point of those are to weed out regressions, crashes, data corruption bugs,
 451build breakage (and any other bug in general) before changes are merged into
 452the more stable mainline Linus tree.
 453
 454But testers of -mm and linux-next should be aware that breakages are
 455more common than in any other tree.
 456
 457
 458This concludes this list of explanations of the various kernel trees.
 459I hope you are now clear on how to apply the various patches and help testing
 460the kernel.
 461
 462Thank you's to Randy Dunlap, Rolf Eike Beer, Linus Torvalds, Bodo Eggert,
 463Johannes Stezenbach, Grant Coady, Pavel Machek and others that I may have
 464forgotten for their reviews and contributions to this document.
 465
 466