]>
Commit | Line | Data |
---|---|---|
c3e270f4 FB |
1 | --- |
2 | title: Coding Style | |
4cdca0af | 3 | category: Contributing |
b41a3f66 | 4 | layout: default |
0aff7b75 | 5 | SPDX-License-Identifier: LGPL-2.1-or-later |
c3e270f4 FB |
6 | --- |
7 | ||
c1d3483d | 8 | # Coding Style |
82143987 | 9 | |
8c9289e7 LP |
10 | ## Formatting |
11 | ||
3b69b18f ZJS |
12 | - 8ch indent, no tabs, except for files in `man/` which are 2ch indent, and |
13 | still no tabs, and shell scripts, which are 4ch indent, and no tabs either. | |
82143987 | 14 | |
8c9289e7 LP |
15 | - We prefer `/* comments */` over `// comments` in code you commit, |
16 | please. This way `// comments` are left for developers to use for local, | |
17 | temporary commenting of code for debug purposes (i.e. uncommittable stuff), | |
18 | making such comments easily discernible from explanatory, documenting code | |
19 | comments (i.e. committable stuff). | |
82143987 FA |
20 | |
21 | - Don't break code lines too eagerly. We do **not** force line breaks at 80ch, | |
22 | all of today's screens should be much larger than that. But then again, don't | |
c90ee834 | 23 | overdo it, ~109ch should be enough really. The `.editorconfig`, `.vimrc` and |
82143987 | 24 | `.dir-locals.el` files contained in the repository will set this limit up for |
3b69b18f ZJS |
25 | you automatically, if you let them (as well as a few other things). Please |
26 | note that emacs loads `.dir-locals.el` automatically, but vim needs to be | |
27 | configured to load `.vimrc`, see that file for instructions. | |
82143987 | 28 | |
7489ccc3 LB |
29 | - If you break a function declaration over multiple lines, do it like this: |
30 | ||
31 | ```c | |
32 | void some_function( | |
33 | int foo, | |
34 | bool bar, | |
35 | char baz) { | |
36 | ||
37 | int a, b, c; | |
38 | ``` | |
39 | ||
cf33b707 LP |
40 | (i.e. use double indentation — 16 spaces — for the parameter list.) |
41 | ||
8c9289e7 LP |
42 | - Try to write this: |
43 | ||
44 | ```c | |
45 | void foo() { | |
46 | } | |
47 | ``` | |
48 | ||
49 | instead of this: | |
50 | ||
51 | ```c | |
52 | void foo() | |
53 | { | |
54 | } | |
55 | ``` | |
56 | ||
57 | - Single-line `if` blocks should not be enclosed in `{}`. Write this: | |
58 | ||
59 | ```c | |
60 | if (foobar) | |
61 | waldo(); | |
62 | ``` | |
63 | ||
64 | instead of this: | |
65 | ||
66 | ```c | |
67 | if (foobar) { | |
68 | waldo(); | |
69 | } | |
70 | ``` | |
71 | ||
72 | - Do not write `foo ()`, write `foo()`. | |
4dbad977 AW |
73 | - `else` blocks should generally start on the same line as the closing `}`: |
74 | ```c | |
75 | if (foobar) { | |
76 | find(); | |
77 | waldo(); | |
78 | } else | |
79 | dont_find_waldo(); | |
80 | ``` | |
8c9289e7 | 81 | |
e28770e3 LP |
82 | - Please define flags types like this: |
83 | ||
84 | ```c | |
85 | typedef enum FoobarFlags { | |
86 | FOOBAR_QUUX = 1 << 0, | |
87 | FOOBAR_WALDO = 1 << 1, | |
88 | FOOBAR_XOXO = 1 << 2, | |
89 | … | |
90 | } FoobarFlags; | |
91 | ``` | |
92 | ||
93 | i.e. use an enum for it, if possible. Indicate bit values via `1 <<` | |
94 | expressions, and align them vertically. Define both an enum and a type for | |
95 | it. | |
96 | ||
97 | - If you define (non-flags) enums, follow this template: | |
98 | ||
99 | ```c | |
100 | typedef enum FoobarMode { | |
101 | FOOBAR_AAA, | |
102 | FOOBAR_BBB, | |
103 | FOOBAR_CCC, | |
104 | … | |
105 | _FOOBAR_MAX, | |
106 | _FOOBAR_INVALID = -EINVAL, | |
107 | } FoobarMode; | |
108 | ``` | |
109 | ||
110 | i.e. define a `_MAX` enum for the largest defined enum value, plus one. Since | |
111 | this is not a regular enum value, prefix it with `_`. Also, define a special | |
112 | "invalid" enum value, and set it to `-EINVAL`. That way the enum type can | |
113 | safely be used to propagate conversion errors. | |
114 | ||
115 | - If you define an enum in a public API, be extra careful, as the size of the | |
116 | enum might change when new values are added, which would break ABI | |
117 | compatibility. Since we typically want to allow adding new enum values to an | |
118 | existing enum type with later API versions, please use the | |
119 | `_SD_ENUM_FORCE_S64()` macro in the enum definition, which forces the size of | |
120 | the enum to be signed 64bit wide. | |
121 | ||
b4f12824 | 122 | ## Code Organization and Semantics |
8c9289e7 | 123 | |
40f55f69 LP |
124 | - For our codebase we intend to use ISO C11 *with* GNU extensions (aka |
125 | "gnu11"). Public APIs (i.e. those we expose via `libsystemd.so` | |
126 | i.e. `systemd/sd-*.h`) should only use ISO C89 however (with a very limited | |
127 | set of conservative and common extensions, such as fixed size integer types | |
128 | from `<inttypes.h>`), so that we don't force consuming programs into C11 | |
129 | mode. (This discrepancy in particular means one thing: internally we use C99 | |
130 | `bool` booleans, externally C89-compatible `int` booleans which generally | |
131 | have different size in memory and slightly different semantics, also see | |
132 | below.) Both for internal and external code it's OK to use even newer | |
133 | features and GCC extension than "gnu11", as long as there's reasonable | |
134 | fallback #ifdeffery in place to ensure compatibility is retained with older | |
135 | compilers. | |
136 | ||
78e5b4d7 LP |
137 | - Please name structures in `PascalCase` (with exceptions, such as public API |
138 | structs), variables and functions in `snake_case`. | |
82143987 | 139 | |
b4f12824 LP |
140 | - Avoid static variables, except for caches and very few other cases. Think |
141 | about thread-safety! While most of our code is never used in threaded | |
142 | environments, at least the library code should make sure it works correctly | |
143 | in them. Instead of doing a lot of locking for that, we tend to prefer using | |
144 | TLS to do per-thread caching (which only works for small, fixed-size cache | |
145 | objects), or we disable caching for any thread that is not the main | |
146 | thread. Use `is_main_thread()` to detect whether the calling thread is the | |
147 | main thread. | |
82143987 | 148 | |
82143987 | 149 | - Do not write functions that clobber call-by-reference variables on |
b4f12824 | 150 | failure. Use temporary variables for these cases and change the passed in |
cf33b707 LP |
151 | variables only on success. The rule is: never clobber return parameters on |
152 | failure, always initialize return parameters on success. | |
153 | ||
154 | - Typically, function parameters fit into three categories: input parameters, | |
155 | mutable objects, and call-by-reference return parameters. Input parameters | |
156 | should always carry suitable "const" declarators if they are pointers, to | |
157 | indicate they are input-only and not changed by the function. Return | |
158 | parameters are best prefixed with "ret_", to clarify they are return | |
159 | parameters. (Conversely, please do not prefix parameters that aren't | |
160 | output-only with "ret_", in particular not mutable parameters that are both | |
161 | input as well as output). Example: | |
162 | ||
163 | ```c | |
164 | static int foobar_frobnicate( | |
165 | Foobar* object, /* the associated mutable object */ | |
166 | const char *input, /* immutable input parameter */ | |
167 | char **ret_frobnicated) { /* return parameter */ | |
168 | … | |
169 | return 0; | |
170 | } | |
171 | ``` | |
82143987 | 172 | |
82143987 | 173 | - The order in which header files are included doesn't matter too |
b4f12824 LP |
174 | much. systemd-internal headers must not rely on an include order, so it is |
175 | safe to include them in any order possible. However, to not clutter global | |
176 | includes, and to make sure internal definitions will not affect global | |
177 | headers, please always include the headers of external components first | |
178 | (these are all headers enclosed in <>), followed by our own exported headers | |
179 | (usually everything that's prefixed by `sd-`), and then followed by internal | |
180 | headers. Furthermore, in all three groups, order all includes alphabetically | |
82143987 FA |
181 | so duplicate includes can easily be detected. |
182 | ||
b4f12824 LP |
183 | - Please avoid using global variables as much as you can. And if you do use |
184 | them make sure they are static at least, instead of exported. Especially in | |
185 | library-like code it is important to avoid global variables. Why are global | |
186 | variables bad? They usually hinder generic reusability of code (since they | |
187 | break in threaded programs, and usually would require locking there), and as | |
188 | the code using them has side-effects make programs non-transparent. That | |
189 | said, there are many cases where they explicitly make a lot of sense, and are | |
190 | OK to use. For example, the log level and target in `log.c` is stored in a | |
191 | global variable, and that's OK and probably expected by most. Also in many | |
192 | cases we cache data in global variables. If you add more caches like this, | |
193 | please be careful however, and think about threading. Only use static | |
194 | variables if you are sure that thread-safety doesn't matter in your | |
195 | case. Alternatively, consider using TLS, which is pretty easy to use with | |
196 | gcc's `thread_local` concept. It's also OK to store data that is inherently | |
197 | global in global variables, for example data parsed from command lines, see | |
82143987 FA |
198 | below. |
199 | ||
82143987 FA |
200 | - Our focus is on the GNU libc (glibc), not any other libcs. If other libcs are |
201 | incompatible with glibc it's on them. However, if there are equivalent POSIX | |
202 | and Linux/GNU-specific APIs, we generally prefer the POSIX APIs. If there | |
203 | aren't, we are happy to use GNU or Linux APIs, and expect non-GNU | |
204 | implementations of libc to catch up with glibc. | |
205 | ||
78e5b4d7 LP |
206 | ## Using C Constructs |
207 | ||
756755d0 | 208 | - Allocate local variables where it makes sense: at the top of the block, or at |
9214f299 ZJS |
209 | the point where they can be initialized. Avoid huge variable declaration |
210 | lists at the top of the function. | |
211 | ||
212 | As an exception, `r` is typically used for a local state variable, but should | |
213 | almost always be declared as the last variable at the top of the function. | |
78e5b4d7 LP |
214 | |
215 | ```c | |
216 | { | |
9214f299 | 217 | uint64_t a; |
756755d0 | 218 | int r; |
78e5b4d7 | 219 | |
9214f299 ZJS |
220 | r = frobnicate(&a); |
221 | if (r < 0) | |
222 | … | |
223 | ||
224 | uint64_t b = a + 1, c; | |
756755d0 | 225 | |
9214f299 | 226 | r = foobarify(a, b, &c); |
756755d0 ZJS |
227 | if (r < 0) |
228 | … | |
9214f299 ZJS |
229 | |
230 | const char *pretty = prettify(a, b, c); | |
231 | … | |
78e5b4d7 LP |
232 | } |
233 | ``` | |
234 | ||
9214f299 ZJS |
235 | - Do not mix multiple variable definitions with function invocations or |
236 | complicated expressions: | |
78e5b4d7 LP |
237 | |
238 | ```c | |
239 | { | |
78e5b4d7 | 240 | uint64_t x = 7; |
756755d0 ZJS |
241 | int a; |
242 | ||
243 | a = foobar(); | |
78e5b4d7 LP |
244 | } |
245 | ``` | |
246 | ||
756755d0 | 247 | instead of: |
78e5b4d7 LP |
248 | |
249 | ```c | |
250 | { | |
756755d0 | 251 | int a = foobar(); |
78e5b4d7 | 252 | uint64_t x = 7; |
78e5b4d7 LP |
253 | } |
254 | ``` | |
255 | ||
756755d0 | 256 | - Use `goto` for cleaning up, and only use it for that. I.e. you may only jump |
78e5b4d7 LP |
257 | to the end of a function, and little else. Never jump backwards! |
258 | ||
259 | - To minimize strict aliasing violations, we prefer unions over casting. | |
260 | ||
261 | - Instead of using `memzero()`/`memset()` to initialize structs allocated on | |
262 | the stack, please try to use c99 structure initializers. It's short, prettier | |
263 | and actually even faster at execution. Hence: | |
264 | ||
265 | ```c | |
266 | struct foobar t = { | |
267 | .foo = 7, | |
268 | .bar = "bazz", | |
269 | }; | |
270 | ``` | |
271 | ||
272 | instead of: | |
273 | ||
274 | ```c | |
275 | struct foobar t; | |
276 | zero(t); | |
277 | t.foo = 7; | |
278 | t.bar = "bazz"; | |
279 | ``` | |
280 | ||
281 | - To implement an endless loop, use `for (;;)` rather than `while (1)`. The | |
282 | latter is a bit ugly anyway, since you probably really meant `while | |
283 | (true)`. To avoid the discussion what the right always-true expression for an | |
284 | infinite while loop is, our recommendation is to simply write it without any | |
285 | such expression by using `for (;;)`. | |
286 | ||
287 | - To determine the length of a constant string `"foo"`, don't bother with | |
288 | `sizeof("foo")-1`, please use `strlen()` instead (both gcc and clang optimize | |
289 | the call away for fixed strings). The only exception is when declaring an | |
9214f299 | 290 | array. In that case use `STRLEN()`, which evaluates to a static constant and |
78e5b4d7 LP |
291 | doesn't force the compiler to create a VLA. |
292 | ||
b5bd7a29 LP |
293 | - Please use C's downgrade-to-bool feature only for expressions that are |
294 | actually booleans (or "boolean-like"), and not for variables that are really | |
295 | numeric. Specifically, if you have an `int b` and it's only used in a boolean | |
296 | sense, by all means check its state with `if (b) …` — but if `b` can actually | |
297 | have more than two semantic values, and you want to compare for non-zero, | |
d238709c | 298 | then please write that explicitly with `if (b != 0) …`. This helps readability |
b5bd7a29 LP |
299 | as the value range and semantical behaviour is directly clear from the |
300 | condition check. As a special addition: when dealing with pointers which you | |
301 | want to check for non-NULL-ness, you may also use downgrade-to-bool feature. | |
302 | ||
303 | - Please do not use yoda comparisons, i.e. please prefer the more readable `if | |
304 | (a == 7)` over the less readable `if (7 == a)`. | |
305 | ||
c159efe3 LP |
306 | ## Destructors |
307 | ||
308 | - The destructors always deregister the object from the next bigger object, not | |
309 | the other way around. | |
310 | ||
311 | - For robustness reasons, destructors should be able to destruct | |
312 | half-initialized objects, too. | |
313 | ||
314 | - When you define a destructor or `unref()` call for an object, please accept a | |
315 | `NULL` object and simply treat this as NOP. This is similar to how libc | |
316 | `free()` works, which accepts `NULL` pointers and becomes a NOP for them. By | |
317 | following this scheme a lot of `if` checks can be removed before invoking | |
318 | your destructor, which makes the code substantially more readable and robust. | |
319 | ||
320 | - Related to this: when you define a destructor or `unref()` call for an | |
321 | object, please make it return the same type it takes and always return `NULL` | |
322 | from it. This allows writing code like this: | |
323 | ||
324 | ```c | |
325 | p = foobar_unref(p); | |
326 | ``` | |
327 | ||
2d1b9281 | 328 | which will always work regardless if `p` is initialized or not, and |
c159efe3 LP |
329 | guarantees that `p` is `NULL` afterwards, all in just one line. |
330 | ||
b065e1f1 LP |
331 | ## Error Handling |
332 | ||
333 | - Error codes are returned as negative `Exxx`. e.g. `return -EINVAL`. There are | |
334 | some exceptions: for constructors, it is OK to return `NULL` on OOM. For | |
335 | lookup functions, `NULL` is fine too for "not found". | |
336 | ||
337 | Be strict with this. When you write a function that can fail due to more than | |
338 | one cause, it *really* should have an `int` as the return value for the error | |
339 | code. | |
340 | ||
947796ea LP |
341 | - libc system calls typically return -1 on error (with the error code in |
342 | `errno`), and >= 0 on success. Use the RET_NERRNO() helper if you are looking | |
343 | for a simple way to convert this libc style error returning into systemd | |
344 | style error returning. e.g. | |
345 | ||
346 | ```c | |
347 | … | |
348 | r = RET_NERRNO(unlink(t)); | |
349 | … | |
350 | ``` | |
351 | ||
352 | or | |
353 | ||
354 | ```c | |
355 | … | |
356 | r = RET_NERRNO(open("/some/file", O_RDONLY|O_CLOEXEC)); | |
357 | … | |
358 | ``` | |
359 | ||
b065e1f1 LP |
360 | - Do not bother with error checking whether writing to stdout/stderr worked. |
361 | ||
362 | - Do not log errors from "library" code, only do so from "main program" | |
363 | code. (With one exception: it is OK to log with DEBUG level from any code, | |
364 | with the exception of maybe inner loops). | |
365 | ||
366 | - In public API calls, you **must** validate all your input arguments for | |
367 | programming error with `assert_return()` and return a sensible return | |
368 | code. In all other calls, it is recommended to check for programming errors | |
369 | with a more brutal `assert()`. We are more forgiving to public users than for | |
370 | ourselves! Note that `assert()` and `assert_return()` really only should be | |
371 | used for detecting programming errors, not for runtime errors. `assert()` and | |
be7148eb | 372 | `assert_return()` by usage of `_likely_()` inform the compiler that it should |
b065e1f1 LP |
373 | not expect these checks to fail, and they inform fellow programmers about the |
374 | expected validity and range of parameters. | |
375 | ||
376 | - When you invoke certain calls like `unlink()`, or `mkdir_p()` and you know it | |
377 | is safe to ignore the error it might return (because a later call would | |
378 | detect the failure anyway, or because the error is in an error path and you | |
379 | thus couldn't do anything about it anyway), then make this clear by casting | |
380 | the invocation explicitly to `(void)`. Code checks like Coverity understand | |
381 | that, and will not complain about ignored error codes. Hence, please use | |
382 | this: | |
383 | ||
384 | ```c | |
385 | (void) unlink("/foo/bar/baz"); | |
386 | ``` | |
387 | ||
388 | instead of just this: | |
389 | ||
390 | ```c | |
391 | unlink("/foo/bar/baz"); | |
392 | ``` | |
393 | ||
800d0802 AZ |
394 | When returning from a `void` function, you may also want to shorten the error |
395 | path boilerplate by returning a function invocation cast to `(void)` like so: | |
396 | ||
397 | ```c | |
398 | if (condition_not_met) | |
399 | return (void) log_tests_skipped("Cannot run ..."); | |
400 | ``` | |
401 | ||
b065e1f1 LP |
402 | Don't cast function calls to `(void)` that return no error |
403 | conditions. Specifically, the various `xyz_unref()` calls that return a | |
404 | `NULL` object shouldn't be cast to `(void)`, since not using the return value | |
405 | does not hide any errors. | |
406 | ||
407 | - When returning a return code from `main()`, please preferably use | |
408 | `EXIT_FAILURE` and `EXIT_SUCCESS` as defined by libc. | |
409 | ||
96f6cfbf LP |
410 | ## Logging |
411 | ||
412 | - For every function you add, think about whether it is a "logging" function or | |
cf33b707 | 413 | a "non-logging" function. "Logging" functions do (non-debug) logging on their |
f223fd6a | 414 | own, "non-logging" functions never log on their own (except at debug level) |
cf33b707 LP |
415 | and expect their callers to log. All functions in "library" code, i.e. in |
416 | `src/shared/` and suchlike must be "non-logging". Every time a "logging" | |
417 | function calls a "non-logging" function, it should log about the resulting | |
418 | errors. If a "logging" function calls another "logging" function, then it | |
419 | should not generate log messages, so that log messages are not generated | |
420 | twice for the same errors. (Note that debug level logging — at syslog level | |
421 | `LOG_DEBUG` — is not considered logging in this context, debug logging is | |
422 | generally always fine and welcome.) | |
96f6cfbf LP |
423 | |
424 | - If possible, do a combined log & return operation: | |
425 | ||
426 | ```c | |
427 | r = operation(...); | |
428 | if (r < 0) | |
429 | return log_(error|warning|notice|...)_errno(r, "Failed to ...: %m"); | |
430 | ``` | |
431 | ||
432 | If the error value is "synthetic", i.e. it was not received from | |
433 | the called function, use `SYNTHETIC_ERRNO` wrapper to tell the logging | |
434 | system to not log the errno value, but still return it: | |
435 | ||
436 | ```c | |
437 | n = read(..., s, sizeof s); | |
438 | if (n != sizeof s) | |
439 | return log_error_errno(SYNTHETIC_ERRNO(EIO), "Failed to read ..."); | |
440 | ``` | |
441 | ||
04858240 LP |
442 | ## Memory Allocation |
443 | ||
444 | - Always check OOM. There is no excuse. In program code, you can use | |
445 | `log_oom()` for then printing a short message, but not in "library" code. | |
446 | ||
447 | - Avoid fixed-size string buffers, unless you really know the maximum size and | |
756755d0 | 448 | that maximum size is small. It is often nicer to use dynamic memory, |
e3bde912 LP |
449 | `alloca_safe()` or VLAs. If you do allocate fixed-size strings on the stack, |
450 | then it is probably only OK if you either use a maximum size such as | |
451 | `LINE_MAX`, or count in detail the maximum size a string can | |
452 | have. (`DECIMAL_STR_MAX` and `DECIMAL_STR_WIDTH` macros are your friends for | |
453 | this!) | |
04858240 LP |
454 | |
455 | Or in other words, if you use `char buf[256]` then you are likely doing | |
456 | something wrong! | |
457 | ||
458 | - Make use of `_cleanup_free_` and friends. It makes your code much nicer to | |
459 | read (and shorter)! | |
460 | ||
e3bde912 LP |
461 | - Do not use `alloca()`, `strdupa()` or `strndupa()` directly. Use |
462 | `alloca_safe()`, `strdupa_safe()` or `strndupa_safe()` instead. (The | |
463 | difference is that the latter include an assertion that the specified size is | |
464 | below a safety threshold, so that the program rather aborts than runs into | |
465 | possible stack overruns.) | |
466 | ||
467 | - Use `alloca_safe()`, but never forget that it is not OK to invoke | |
468 | `alloca_safe()` within a loop or within function call | |
469 | parameters. `alloca_safe()` memory is released at the end of a function, and | |
470 | not at the end of a `{}` block. Thus, if you invoke it in a loop, you keep | |
471 | increasing the stack pointer without ever releasing memory again. (VLAs have | |
472 | better behavior in this case, so consider using them as an alternative.) | |
473 | Regarding not using `alloca_safe()` within function parameters, see the BUGS | |
474 | section of the `alloca(3)` man page. | |
04858240 LP |
475 | |
476 | - If you want to concatenate two or more strings, consider using `strjoina()` | |
477 | or `strjoin()` rather than `asprintf()`, as the latter is a lot slower. This | |
478 | matters particularly in inner loops (but note that `strjoina()` cannot be | |
479 | used there). | |
480 | ||
4467d393 LP |
481 | ## Runtime Behaviour |
482 | ||
483 | - Avoid leaving long-running child processes around, i.e. `fork()`s that are | |
484 | not followed quickly by an `execv()` in the child. Resource management is | |
485 | unclear in this case, and memory CoW will result in unexpected penalties in | |
486 | the parent much, much later on. | |
487 | ||
488 | - Don't block execution for arbitrary amounts of time using `usleep()` or a | |
489 | similar call, unless you really know what you do. Just "giving something some | |
490 | time", or so is a lazy excuse. Always wait for the proper event, instead of | |
491 | doing time-based poll loops. | |
492 | ||
493 | - Whenever installing a signal handler, make sure to set `SA_RESTART` for it, | |
494 | so that interrupted system calls are automatically restarted, and we minimize | |
495 | hassles with handling `EINTR` (in particular as `EINTR` handling is pretty | |
496 | broken on Linux). | |
497 | ||
498 | - When applying C-style unescaping as well as specifier expansion on the same | |
c90b6abc | 499 | string, always apply the C-style unescaping first, followed by the specifier |
4467d393 LP |
500 | expansion. When doing the reverse, make sure to escape `%` in specifier-style |
501 | first (i.e. `%` → `%%`), and then do C-style escaping where necessary. | |
502 | ||
503 | - Be exceptionally careful when formatting and parsing floating point | |
504 | numbers. Their syntax is locale dependent (i.e. `5.000` in en_US is generally | |
505 | understood as 5, while in de_DE as 5000.). | |
506 | ||
507 | - Make sure to enforce limits on every user controllable resource. If the user | |
508 | can allocate resources in your code, your code must enforce some form of | |
509 | limits after which it will refuse operation. It's fine if it is hard-coded | |
510 | (at least initially), but it needs to be there. This is particularly | |
511 | important for objects that unprivileged users may allocate, but also matters | |
6ae11e12 | 512 | for everything else any user may allocate. |
4467d393 | 513 | |
f42c1cd4 LP |
514 | ## Types |
515 | ||
516 | - Think about the types you use. If a value cannot sensibly be negative, do not | |
517 | use `int`, but use `unsigned`. | |
518 | ||
519 | - Use `char` only for actual characters. Use `uint8_t` or `int8_t` when you | |
520 | actually mean a byte-sized signed or unsigned integers. When referring to a | |
521 | generic byte, we generally prefer the unsigned variant `uint8_t`. Do not use | |
522 | types based on `short`. They *never* make sense. Use `int`, `long`, `long | |
523 | long`, all in unsigned and signed fashion, and the fixed-size types | |
524 | `uint8_t`, `uint16_t`, `uint32_t`, `uint64_t`, `int8_t`, `int16_t`, `int32_t` | |
525 | and so on, as well as `size_t`, but nothing else. Do not use kernel types | |
526 | like `u32` and so on, leave that to the kernel. | |
527 | ||
528 | - Stay uniform. For example, always use `usec_t` for time values. Do not mix | |
529 | `usec` and `msec`, and `usec` and whatnot. | |
530 | ||
531 | - Never use the `off_t` type, and particularly avoid it in public APIs. It's | |
532 | really weirdly defined, as it usually is 64-bit and we don't support it any | |
533 | other way, but it could in theory also be 32-bit. Which one it is depends on | |
534 | a compiler switch chosen by the compiled program, which hence corrupts APIs | |
535 | using it unless they can also follow the program's choice. Moreover, in | |
536 | systemd we should parse values the same way on all architectures and cannot | |
537 | expose `off_t` values over D-Bus. To avoid any confusion regarding conversion | |
538 | and ABIs, always use simply `uint64_t` directly. | |
539 | ||
540 | - Unless you allocate an array, `double` is always a better choice than | |
541 | `float`. Processors speak `double` natively anyway, so there is no speed | |
542 | benefit, and on calls like `printf()` `float`s get promoted to `double`s | |
543 | anyway, so there is no point. | |
544 | ||
545 | - Use the bool type for booleans, not integers. One exception: in public | |
546 | headers (i.e those in `src/systemd/sd-*.h`) use integers after all, as `bool` | |
40f55f69 LP |
547 | is C99 and in our public APIs we try to stick to C89 (with a few extensions; |
548 | also see above). | |
f42c1cd4 | 549 | |
3b75e079 LP |
550 | ## Deadlocks |
551 | ||
38b38500 | 552 | - Do not issue NSS requests (that includes user name and hostname lookups) |
3b75e079 LP |
553 | from PID 1 as this might trigger deadlocks when those lookups involve |
554 | synchronously talking to services that we would need to start up. | |
555 | ||
556 | - Do not synchronously talk to any other service from PID 1, due to risk of | |
557 | deadlocks. | |
558 | ||
25553cd9 LP |
559 | ## File Descriptors |
560 | ||
561 | - When you allocate a file descriptor, it should be made `O_CLOEXEC` right from | |
562 | the beginning, as none of our files should leak to forked binaries by | |
563 | default. Hence, whenever you open a file, `O_CLOEXEC` must be specified, | |
564 | right from the beginning. This also applies to sockets. Effectively, this | |
565 | means that all invocations to: | |
566 | ||
567 | - `open()` must get `O_CLOEXEC` passed, | |
568 | - `socket()` and `socketpair()` must get `SOCK_CLOEXEC` passed, | |
569 | - `recvmsg()` must get `MSG_CMSG_CLOEXEC` set, | |
570 | - `F_DUPFD_CLOEXEC` should be used instead of `F_DUPFD`, and so on, | |
571 | - invocations of `fopen()` should take `e`. | |
572 | ||
573 | - It's a good idea to use `O_NONBLOCK` when opening 'foreign' regular files, | |
574 | i.e. file system objects that are supposed to be regular files whose paths | |
d0515a28 | 575 | were specified by the user and hence might actually refer to other types of |
25553cd9 LP |
576 | file system objects. This is a good idea so that we don't end up blocking on |
577 | 'strange' file nodes, for example if the user pointed us to a FIFO or device | |
578 | node which may block when opening. Moreover even for actual regular files | |
579 | `O_NONBLOCK` has a benefit: it bypasses any mandatory lock that might be in | |
580 | effect on the regular file. If in doubt consider turning off `O_NONBLOCK` | |
581 | again after opening. | |
582 | ||
996f119d LP |
583 | ## Command Line |
584 | ||
585 | - If you parse a command line, and want to store the parsed parameters in | |
586 | global variables, please consider prefixing their names with `arg_`. We have | |
587 | been following this naming rule in most of our tools, and we should continue | |
588 | to do so, as it makes it easy to identify command line parameter variables, | |
589 | and makes it clear why it is OK that they are global variables. | |
590 | ||
591 | - Command line option parsing: | |
592 | - Do not print full `help()` on error, be specific about the error. | |
593 | - Do not print messages to stdout on error. | |
594 | - Do not POSIX_ME_HARDER unless necessary, i.e. avoid `+` in option string. | |
595 | ||
56380761 LP |
596 | ## Exporting Symbols |
597 | ||
598 | - Variables and functions **must** be static, unless they have a prototype, and | |
599 | are supposed to be exported. | |
600 | ||
601 | - Public API calls (i.e. functions exported by our shared libraries) | |
602 | must be marked `_public_` and need to be prefixed with `sd_`. No | |
603 | other functions should be prefixed like that. | |
604 | ||
605 | - When exposing public C APIs, be careful what function parameters you make | |
606 | `const`. For example, a parameter taking a context object should probably not | |
607 | be `const`, even if you are writing an otherwise read-only accessor function | |
608 | for it. The reason is that making it `const` fixates the contract that your | |
609 | call won't alter the object ever, as part of the API. However, that's often | |
610 | quite a promise, given that this even prohibits object-internal caching or | |
611 | lazy initialization of object variables. Moreover, it's usually not too | |
612 | useful for client applications. Hence, please be careful and avoid `const` on | |
613 | object parameters, unless you are very sure `const` is appropriate. | |
614 | ||
971dfffa LP |
615 | ## Referencing Concepts |
616 | ||
82143987 FA |
617 | - When referring to a configuration file option in the documentation and such, |
618 | please always suffix it with `=`, to indicate that it is a configuration file | |
619 | setting. | |
620 | ||
621 | - When referring to a command line option in the documentation and such, please | |
622 | always prefix with `--` or `-` (as appropriate), to indicate that it is a | |
623 | command line option. | |
624 | ||
625 | - When referring to a file system path that is a directory, please always | |
626 | suffix it with `/`, to indicate that it is a directory, not a regular file | |
627 | (or other file system object). | |
628 | ||
2d0dce2a LP |
629 | ## Functions to Avoid |
630 | ||
631 | - Use `memzero()` or even better `zero()` instead of `memset(..., 0, ...)` | |
632 | ||
633 | - Please use `streq()` and `strneq()` instead of `strcmp()`, `strncmp()` where | |
634 | applicable (i.e. wherever you just care about equality/inequality, not about | |
635 | the sorting order). | |
636 | ||
637 | - Never use `strtol()`, `atoi()` and similar calls. Use `safe_atoli()`, | |
638 | `safe_atou32()` and suchlike instead. They are much nicer to use in most | |
639 | cases and correctly check for parsing errors. | |
640 | ||
641 | - `htonl()`/`ntohl()` and `htons()`/`ntohs()` are weird. Please use `htobe32()` | |
642 | and `htobe16()` instead, it's much more descriptive, and actually says what | |
643 | really is happening, after all `htonl()` and `htons()` don't operate on | |
644 | `long`s and `short`s as their name would suggest, but on `uint32_t` and | |
645 | `uint16_t`. Also, "network byte order" is just a weird name for "big endian", | |
646 | hence we might want to call it "big endian" right-away. | |
647 | ||
a5b28b77 ZJS |
648 | - Use `typesafe_inet_ntop()`, `typesafe_inet_ntop4()`, and |
649 | `typesafe_inet_ntop6()` instead of `inet_ntop()`. But better yet, use the | |
650 | `IN_ADDR_TO_STRING()`, `IN4_ADDR_TO_STRING()`, and `IN6_ADDR_TO_STRING()` | |
74223cbe | 651 | macros which allocate an anonymous buffer internally. |
a5b28b77 | 652 | |
2d0dce2a | 653 | - Please never use `dup()`. Use `fcntl(fd, F_DUPFD_CLOEXEC, 3)` instead. For |
c1495f8e | 654 | two reasons: first, you want `O_CLOEXEC` set on the new `fd` (see |
2d0dce2a LP |
655 | above). Second, `dup()` will happily duplicate your `fd` as 0, 1, 2, |
656 | i.e. stdin, stdout, stderr, should those `fd`s be closed. Given the special | |
657 | semantics of those `fd`s, it's probably a good idea to avoid | |
658 | them. `F_DUPFD_CLOEXEC` with `3` as parameter avoids them. | |
659 | ||
82143987 FA |
660 | - Don't use `fgets()`, it's too hard to properly handle errors such as overly |
661 | long lines. Use `read_line()` instead, which is our own function that handles | |
c90b6abc | 662 | this much more nicely. |
2d0dce2a LP |
663 | |
664 | - Don't invoke `exit()`, ever. It is not replacement for proper error | |
665 | handling. Please escalate errors up your call chain, and use normal `return` | |
666 | to exit from the main function of a process. If you `fork()`ed off a child | |
667 | process, please use `_exit()` instead of `exit()`, so that the exit handlers | |
668 | are not run. | |
669 | ||
6ae11e12 | 670 | - We never use the POSIX version of `basename()` (which glibc defines in |
2d0dce2a LP |
671 | `libgen.h`), only the GNU version (which glibc defines in `string.h`). The |
672 | only reason to include `libgen.h` is because `dirname()` is needed. Every | |
673 | time you need that please immediately undefine `basename()`, and add a | |
674 | comment about it, so that no code ever ends up using the POSIX version! | |
831781b9 | 675 | |
bcef0f33 ZJS |
676 | - Never use `FILENAME_MAX`. Use `PATH_MAX` instead (for checking maximum size |
677 | of paths) and `NAME_MAX` (for checking maximum size of filenames). | |
678 | `FILENAME_MAX` is not POSIX, and is a confusingly named alias for `PATH_MAX` | |
6ae11e12 | 679 | on Linux. Note that `NAME_MAX` does not include space for a trailing `NUL`, |
bcef0f33 | 680 | but `PATH_MAX` does. UNIX FTW! |
b775b182 | 681 | |
ff2c2d08 | 682 | ## Committing to git |
831781b9 LP |
683 | |
684 | - Commit message subject lines should be prefixed with an appropriate component | |
685 | name of some kind. For example "journal: ", "nspawn: " and so on. | |
686 | ||
687 | - Do not use "Signed-Off-By:" in your commit messages. That's a kernel thing we | |
688 | don't do in the systemd project. |