1.. Copyright 2010 Nicolas Palix <npalix@diku.dk>
2.. Copyright 2010 Julia Lawall <julia@diku.dk>
3.. Copyright 2010 Gilles Muller <Gilles.Muller@lip6.fr>
4
5.. highlight:: none
6
7Coccinelle
8==========
9
10Coccinelle is a tool for pattern matching and text transformation that has
11many uses in kernel development, including the application of complex,
12tree-wide patches and detection of problematic programming patterns.
13
14Getting Coccinelle
15-------------------
16
17The semantic patches included in the kernel use features and options
18which are provided by Coccinelle version 1.0.0-rc11 and above.
19Using earlier versions will fail as the option names used by
20the Coccinelle files and coccicheck have been updated.
21
22Coccinelle is available through the package manager
23of many distributions, e.g. :
24
25 - Debian
26 - Fedora
27 - Ubuntu
28 - OpenSUSE
29 - Arch Linux
30 - NetBSD
31 - FreeBSD
32
33Some distribution packages are obsolete and it is recommended
34to use the latest version released from the Coccinelle homepage at
35http://coccinelle.lip6.fr/
36
37Or from Github at:
38
39https://github.com/coccinelle/coccinelle
40
41Once you have it, run the following commands::
42
43        ./autogen
44        ./configure
45        make
46
47as a regular user, and install it with::
48
49        sudo make install
50
51More detailed installation instructions to build from source can be
52found at:
53
54https://github.com/coccinelle/coccinelle/blob/master/install.txt
55
56Supplemental documentation
57---------------------------
58
59For supplemental documentation refer to the wiki:
60
61https://bottest.wiki.kernel.org/coccicheck
62
63The wiki documentation always refers to the linux-next version of the script.
64
65For Semantic Patch Language(SmPL) grammar documentation refer to:
66
67http://coccinelle.lip6.fr/documentation.php
68
69Using Coccinelle on the Linux kernel
70------------------------------------
71
72A Coccinelle-specific target is defined in the top level
73Makefile. This target is named ``coccicheck`` and calls the ``coccicheck``
74front-end in the ``scripts`` directory.
75
76Four basic modes are defined: ``patch``, ``report``, ``context``, and
77``org``. The mode to use is specified by setting the MODE variable with
78``MODE=<mode>``.
79
80- ``patch`` proposes a fix, when possible.
81
82- ``report`` generates a list in the following format:
83  file:line:column-column: message
84
85- ``context`` highlights lines of interest and their context in a
86  diff-like style.Lines of interest are indicated with ``-``.
87
88- ``org`` generates a report in the Org mode format of Emacs.
89
90Note that not all semantic patches implement all modes. For easy use
91of Coccinelle, the default mode is "report".
92
93Two other modes provide some common combinations of these modes.
94
95- ``chain`` tries the previous modes in the order above until one succeeds.
96
97- ``rep+ctxt`` runs successively the report mode and the context mode.
98  It should be used with the C option (described later)
99  which checks the code on a file basis.
100
101Examples
102~~~~~~~~
103
104To make a report for every semantic patch, run the following command::
105
106		make coccicheck MODE=report
107
108To produce patches, run::
109
110		make coccicheck MODE=patch
111
112
113The coccicheck target applies every semantic patch available in the
114sub-directories of ``scripts/coccinelle`` to the entire Linux kernel.
115
116For each semantic patch, a commit message is proposed.  It gives a
117description of the problem being checked by the semantic patch, and
118includes a reference to Coccinelle.
119
120As any static code analyzer, Coccinelle produces false
121positives. Thus, reports must be carefully checked, and patches
122reviewed.
123
124To enable verbose messages set the V= variable, for example::
125
126   make coccicheck MODE=report V=1
127
128Coccinelle parallelization
129---------------------------
130
131By default, coccicheck tries to run as parallel as possible. To change
132the parallelism, set the J= variable. For example, to run across 4 CPUs::
133
134   make coccicheck MODE=report J=4
135
136As of Coccinelle 1.0.2 Coccinelle uses Ocaml parmap for parallelization,
137if support for this is detected you will benefit from parmap parallelization.
138
139When parmap is enabled coccicheck will enable dynamic load balancing by using
140``--chunksize 1`` argument, this ensures we keep feeding threads with work
141one by one, so that we avoid the situation where most work gets done by only
142a few threads. With dynamic load balancing, if a thread finishes early we keep
143feeding it more work.
144
145When parmap is enabled, if an error occurs in Coccinelle, this error
146value is propagated back, the return value of the ``make coccicheck``
147captures this return value.
148
149Using Coccinelle with a single semantic patch
150---------------------------------------------
151
152The optional make variable COCCI can be used to check a single
153semantic patch. In that case, the variable must be initialized with
154the name of the semantic patch to apply.
155
156For instance::
157
158	make coccicheck COCCI=<my_SP.cocci> MODE=patch
159
160or::
161
162	make coccicheck COCCI=<my_SP.cocci> MODE=report
163
164
165Controlling Which Files are Processed by Coccinelle
166---------------------------------------------------
167
168By default the entire kernel source tree is checked.
169
170To apply Coccinelle to a specific directory, ``M=`` can be used.
171For example, to check drivers/net/wireless/ one may write::
172
173    make coccicheck M=drivers/net/wireless/
174
175To apply Coccinelle on a file basis, instead of a directory basis, the
176following command may be used::
177
178    make C=1 CHECK="scripts/coccicheck"
179
180To check only newly edited code, use the value 2 for the C flag, i.e.::
181
182    make C=2 CHECK="scripts/coccicheck"
183
184In these modes, which works on a file basis, there is no information
185about semantic patches displayed, and no commit message proposed.
186
187This runs every semantic patch in scripts/coccinelle by default. The
188COCCI variable may additionally be used to only apply a single
189semantic patch as shown in the previous section.
190
191The "report" mode is the default. You can select another one with the
192MODE variable explained above.
193
194Debugging Coccinelle SmPL patches
195---------------------------------
196
197Using coccicheck is best as it provides in the spatch command line
198include options matching the options used when we compile the kernel.
199You can learn what these options are by using V=1, you could then
200manually run Coccinelle with debug options added.
201
202Alternatively you can debug running Coccinelle against SmPL patches
203by asking for stderr to be redirected to stderr, by default stderr
204is redirected to /dev/null, if you'd like to capture stderr you
205can specify the ``DEBUG_FILE="file.txt"`` option to coccicheck. For
206instance::
207
208    rm -f cocci.err
209    make coccicheck COCCI=scripts/coccinelle/free/kfree.cocci MODE=report DEBUG_FILE=cocci.err
210    cat cocci.err
211
212You can use SPFLAGS to add debugging flags, for instance you may want to
213add both --profile --show-trying to SPFLAGS when debugging. For instance
214you may want to use::
215
216    rm -f err.log
217    export COCCI=scripts/coccinelle/misc/irqf_oneshot.cocci
218    make coccicheck DEBUG_FILE="err.log" MODE=report SPFLAGS="--profile --show-trying" M=./drivers/mfd/arizona-irq.c
219
220err.log will now have the profiling information, while stdout will
221provide some progress information as Coccinelle moves forward with
222work.
223
224DEBUG_FILE support is only supported when using coccinelle >= 1.0.2.
225
226.cocciconfig support
227--------------------
228
229Coccinelle supports reading .cocciconfig for default Coccinelle options that
230should be used every time spatch is spawned, the order of precedence for
231variables for .cocciconfig is as follows:
232
233- Your current user's home directory is processed first
234- Your directory from which spatch is called is processed next
235- The directory provided with the --dir option is processed last, if used
236
237Since coccicheck runs through make, it naturally runs from the kernel
238proper dir, as such the second rule above would be implied for picking up a
239.cocciconfig when using ``make coccicheck``.
240
241``make coccicheck`` also supports using M= targets. If you do not supply
242any M= target, it is assumed you want to target the entire kernel.
243The kernel coccicheck script has::
244
245    if [ "$KBUILD_EXTMOD" = "" ] ; then
246        OPTIONS="--dir $srctree $COCCIINCLUDE"
247    else
248        OPTIONS="--dir $KBUILD_EXTMOD $COCCIINCLUDE"
249    fi
250
251KBUILD_EXTMOD is set when an explicit target with M= is used. For both cases
252the spatch --dir argument is used, as such third rule applies when whether M=
253is used or not, and when M= is used the target directory can have its own
254.cocciconfig file. When M= is not passed as an argument to coccicheck the
255target directory is the same as the directory from where spatch was called.
256
257If not using the kernel's coccicheck target, keep the above precedence
258order logic of .cocciconfig reading. If using the kernel's coccicheck target,
259override any of the kernel's .coccicheck's settings using SPFLAGS.
260
261We help Coccinelle when used against Linux with a set of sensible defaults
262options for Linux with our own Linux .cocciconfig. This hints to coccinelle
263git can be used for ``git grep`` queries over coccigrep. A timeout of 200
264seconds should suffice for now.
265
266The options picked up by coccinelle when reading a .cocciconfig do not appear
267as arguments to spatch processes running on your system, to confirm what
268options will be used by Coccinelle run::
269
270      spatch --print-options-only
271
272You can override with your own preferred index option by using SPFLAGS. Take
273note that when there are conflicting options Coccinelle takes precedence for
274the last options passed. Using .cocciconfig is possible to use idutils, however
275given the order of precedence followed by Coccinelle, since the kernel now
276carries its own .cocciconfig, you will need to use SPFLAGS to use idutils if
277desired. See below section "Additional flags" for more details on how to use
278idutils.
279
280Additional flags
281----------------
282
283Additional flags can be passed to spatch through the SPFLAGS
284variable. This works as Coccinelle respects the last flags
285given to it when options are in conflict. ::
286
287    make SPFLAGS=--use-glimpse coccicheck
288
289Coccinelle supports idutils as well but requires coccinelle >= 1.0.6.
290When no ID file is specified coccinelle assumes your ID database file
291is in the file .id-utils.index on the top level of the kernel, coccinelle
292carries a script scripts/idutils_index.sh which creates the database with::
293
294    mkid -i C --output .id-utils.index
295
296If you have another database filename you can also just symlink with this
297name. ::
298
299    make SPFLAGS=--use-idutils coccicheck
300
301Alternatively you can specify the database filename explicitly, for
302instance::
303
304    make SPFLAGS="--use-idutils /full-path/to/ID" coccicheck
305
306See ``spatch --help`` to learn more about spatch options.
307
308Note that the ``--use-glimpse`` and ``--use-idutils`` options
309require external tools for indexing the code. None of them is
310thus active by default. However, by indexing the code with
311one of these tools, and according to the cocci file used,
312spatch could proceed the entire code base more quickly.
313
314SmPL patch specific options
315---------------------------
316
317SmPL patches can have their own requirements for options passed
318to Coccinelle. SmPL patch specific options can be provided by
319providing them at the top of the SmPL patch, for instance::
320
321	// Options: --no-includes --include-headers
322
323SmPL patch Coccinelle requirements
324----------------------------------
325
326As Coccinelle features get added some more advanced SmPL patches
327may require newer versions of Coccinelle. If an SmPL patch requires
328at least a version of Coccinelle, this can be specified as follows,
329as an example if requiring at least Coccinelle >= 1.0.5::
330
331	// Requires: 1.0.5
332
333Proposing new semantic patches
334-------------------------------
335
336New semantic patches can be proposed and submitted by kernel
337developers. For sake of clarity, they should be organized in the
338sub-directories of ``scripts/coccinelle/``.
339
340
341Detailed description of the ``report`` mode
342-------------------------------------------
343
344``report`` generates a list in the following format::
345
346  file:line:column-column: message
347
348Example
349~~~~~~~
350
351Running::
352
353	make coccicheck MODE=report COCCI=scripts/coccinelle/api/err_cast.cocci
354
355will execute the following part of the SmPL script::
356
357   <smpl>
358   @r depends on !context && !patch && (org || report)@
359   expression x;
360   position p;
361   @@
362
363     ERR_PTR@p(PTR_ERR(x))
364
365   @script:python depends on report@
366   p << r.p;
367   x << r.x;
368   @@
369
370   msg="ERR_CAST can be used with %s" % (x)
371   coccilib.report.print_report(p[0], msg)
372   </smpl>
373
374This SmPL excerpt generates entries on the standard output, as
375illustrated below::
376
377    /home/user/linux/crypto/ctr.c:188:9-16: ERR_CAST can be used with alg
378    /home/user/linux/crypto/authenc.c:619:9-16: ERR_CAST can be used with auth
379    /home/user/linux/crypto/xts.c:227:9-16: ERR_CAST can be used with alg
380
381
382Detailed description of the ``patch`` mode
383------------------------------------------
384
385When the ``patch`` mode is available, it proposes a fix for each problem
386identified.
387
388Example
389~~~~~~~
390
391Running::
392
393	make coccicheck MODE=patch COCCI=scripts/coccinelle/api/err_cast.cocci
394
395will execute the following part of the SmPL script::
396
397    <smpl>
398    @ depends on !context && patch && !org && !report @
399    expression x;
400    @@
401
402    - ERR_PTR(PTR_ERR(x))
403    + ERR_CAST(x)
404    </smpl>
405
406This SmPL excerpt generates patch hunks on the standard output, as
407illustrated below::
408
409    diff -u -p a/crypto/ctr.c b/crypto/ctr.c
410    --- a/crypto/ctr.c 2010-05-26 10:49:38.000000000 +0200
411    +++ b/crypto/ctr.c 2010-06-03 23:44:49.000000000 +0200
412    @@ -185,7 +185,7 @@ static struct crypto_instance *crypto_ct
413 	alg = crypto_attr_alg(tb[1], CRYPTO_ALG_TYPE_CIPHER,
414 				  CRYPTO_ALG_TYPE_MASK);
415 	if (IS_ERR(alg))
416    -		return ERR_PTR(PTR_ERR(alg));
417    +		return ERR_CAST(alg);
418
419 	/* Block size must be >= 4 bytes. */
420 	err = -EINVAL;
421
422Detailed description of the ``context`` mode
423--------------------------------------------
424
425``context`` highlights lines of interest and their context
426in a diff-like style.
427
428      **NOTE**: The diff-like output generated is NOT an applicable patch. The
429      intent of the ``context`` mode is to highlight the important lines
430      (annotated with minus, ``-``) and gives some surrounding context
431      lines around. This output can be used with the diff mode of
432      Emacs to review the code.
433
434Example
435~~~~~~~
436
437Running::
438
439	make coccicheck MODE=context COCCI=scripts/coccinelle/api/err_cast.cocci
440
441will execute the following part of the SmPL script::
442
443    <smpl>
444    @ depends on context && !patch && !org && !report@
445    expression x;
446    @@
447
448    * ERR_PTR(PTR_ERR(x))
449    </smpl>
450
451This SmPL excerpt generates diff hunks on the standard output, as
452illustrated below::
453
454    diff -u -p /home/user/linux/crypto/ctr.c /tmp/nothing
455    --- /home/user/linux/crypto/ctr.c	2010-05-26 10:49:38.000000000 +0200
456    +++ /tmp/nothing
457    @@ -185,7 +185,6 @@ static struct crypto_instance *crypto_ct
458 	alg = crypto_attr_alg(tb[1], CRYPTO_ALG_TYPE_CIPHER,
459 				  CRYPTO_ALG_TYPE_MASK);
460 	if (IS_ERR(alg))
461    -		return ERR_PTR(PTR_ERR(alg));
462
463 	/* Block size must be >= 4 bytes. */
464 	err = -EINVAL;
465
466Detailed description of the ``org`` mode
467----------------------------------------
468
469``org`` generates a report in the Org mode format of Emacs.
470
471Example
472~~~~~~~
473
474Running::
475
476	make coccicheck MODE=org COCCI=scripts/coccinelle/api/err_cast.cocci
477
478will execute the following part of the SmPL script::
479
480    <smpl>
481    @r depends on !context && !patch && (org || report)@
482    expression x;
483    position p;
484    @@
485
486      ERR_PTR@p(PTR_ERR(x))
487
488    @script:python depends on org@
489    p << r.p;
490    x << r.x;
491    @@
492
493    msg="ERR_CAST can be used with %s" % (x)
494    msg_safe=msg.replace("[","@(").replace("]",")")
495    coccilib.org.print_todo(p[0], msg_safe)
496    </smpl>
497
498This SmPL excerpt generates Org entries on the standard output, as
499illustrated below::
500
501    * TODO [[view:/home/user/linux/crypto/ctr.c::face=ovl-face1::linb=188::colb=9::cole=16][ERR_CAST can be used with alg]]
502    * TODO [[view:/home/user/linux/crypto/authenc.c::face=ovl-face1::linb=619::colb=9::cole=16][ERR_CAST can be used with auth]]
503    * TODO [[view:/home/user/linux/crypto/xts.c::face=ovl-face1::linb=227::colb=9::cole=16][ERR_CAST can be used with alg]]
504