5 SPDX-License-Identifier: LGPL-2.1-or-later
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.
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).
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
23 overdo it, ~109ch should be enough really. The `.editorconfig`, `.vimrc` and
24 `.dir-locals.el` files contained in the repository will set this limit up for
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.
29 - If you break a function declaration over multiple lines, do it like this:
40 (i.e. use double indentation — 16 spaces — for the parameter list.)
57 - Single-line `if` blocks should not be enclosed in `{}`. Write this:
72 - Do not write `foo ()`, write `foo()`.
74 - `else` blocks should generally start on the same line as the closing `}`:
83 - Please define flags types like this:
86 typedef enum FoobarFlags {
88 FOOBAR_WALDO = 1 << 1,
94 i.e. use an enum for it, if possible. Indicate bit values via `1 <<`
95 expressions, and align them vertically. Define both an enum and a type for
98 - If you define (non-flags) enums, follow this template:
101 typedef enum FoobarMode {
107 _FOOBAR_INVALID = -EINVAL,
111 i.e. define a `_MAX` enum for the largest defined enum value, plus one. Since
112 this is not a regular enum value, prefix it with `_`. Also, define a special
113 "invalid" enum value, and set it to `-EINVAL`. That way the enum type can
114 safely be used to propagate conversion errors.
116 - If you define an enum in a public API, be extra careful, as the size of the
117 enum might change when new values are added, which would break ABI
118 compatibility. Since we typically want to allow adding new enum values to an
119 existing enum type with later API versions, please use the
120 `_SD_ENUM_FORCE_S64()` macro in the enum definition, which forces the size of
121 the enum to be signed 64-bit wide.
123 - Empty lines to separate code blocks are a good thing, please add them
124 abundantly. However, please stick to one at a time, i.e. multiple empty lines
125 immediately following each other are not OK. Also, we try to keep function
126 calls and their immediate error handling together. Hence:
129 /* → empty line here is good */
130 r = some_function(…);
131 /* → empty line here would be bad */
133 return log_error_errno(r, "Some function failed: %m");
134 /* → empty line here is good */
137 - In shell scripts, do not use whitespace after the redirection operator
138 (`>some/file` instead of `> some/file`, `<<EOF` instead of `<< EOF`).
140 ## Code Organization and Semantics
142 - For our codebase we intend to use ISO C11 *with* GNU extensions (aka
143 "gnu11"). Public APIs (i.e. those we expose via `libsystemd.so`
144 i.e. `systemd/sd-*.h`) should only use ISO C89 however (with a very limited
145 set of conservative and common extensions, such as fixed size integer types
146 from `<inttypes.h>`), so that we don't force consuming programs into C11
147 mode. (This discrepancy in particular means one thing: internally we use C99
148 `bool` booleans, externally C89-compatible `int` booleans which generally
149 have different size in memory and slightly different semantics, also see
150 below.) Both for internal and external code it's OK to use even newer
151 features and GCC extension than "gnu11", as long as there's reasonable
152 fallback #ifdeffery in place to ensure compatibility is retained with older
155 - Please name structures in `PascalCase` (with exceptions, such as public API
156 structs), variables and functions in `snake_case`.
158 - Avoid static variables, except for caches and very few other cases. Think
159 about thread-safety! While most of our code is never used in threaded
160 environments, at least the library code should make sure it works correctly
161 in them. Instead of doing a lot of locking for that, we tend to prefer using
162 TLS to do per-thread caching (which only works for small, fixed-size cache
163 objects), or we disable caching for any thread that is not the main
164 thread. Use `is_main_thread()` to detect whether the calling thread is the
167 - Do not write functions that clobber call-by-reference variables on
168 failure. Use temporary variables for these cases and change the passed in
169 variables only on success. The rule is: never clobber return parameters on
170 failure, always initialize return parameters on success.
172 - Typically, function parameters fit into three categories: input parameters,
173 mutable objects, and call-by-reference return parameters. Input parameters
174 should always carry suitable "const" declarators if they are pointers, to
175 indicate they are input-only and not changed by the function. Return
176 parameters are best prefixed with "ret_", to clarify they are return
177 parameters. (Conversely, please do not prefix parameters that aren't
178 output-only with "ret_", in particular not mutable parameters that are both
179 input as well as output). Example:
182 static int foobar_frobnicate(
183 Foobar* object, /* the associated mutable object */
184 const char *input, /* immutable input parameter */
185 char **ret_frobnicated) { /* return parameter */
191 - The order in which header files are included doesn't matter too
192 much. systemd-internal headers must not rely on an include order, so it is
193 safe to include them in any order possible. However, to not clutter global
194 includes, and to make sure internal definitions will not affect global
195 headers, please always include the headers of external components first
196 (these are all headers enclosed in <>), followed by our own exported headers
197 (usually everything that's prefixed by `sd-`), and then followed by internal
198 headers. Furthermore, in all three groups, order all includes alphabetically
199 so duplicate includes can easily be detected.
201 - Please avoid using global variables as much as you can. And if you do use
202 them make sure they are static at least, instead of exported. Especially in
203 library-like code it is important to avoid global variables. Why are global
204 variables bad? They usually hinder generic reusability of code (since they
205 break in threaded programs, and usually would require locking there), and as
206 the code using them has side-effects make programs non-transparent. That
207 said, there are many cases where they explicitly make a lot of sense, and are
208 OK to use. For example, the log level and target in `log.c` is stored in a
209 global variable, and that's OK and probably expected by most. Also in many
210 cases we cache data in global variables. If you add more caches like this,
211 please be careful however, and think about threading. Only use static
212 variables if you are sure that thread-safety doesn't matter in your
213 case. Alternatively, consider using TLS, which is pretty easy to use with
214 gcc's `thread_local` concept. It's also OK to store data that is inherently
215 global in global variables, for example, data parsed from command lines, see
218 - Our focus is on the GNU libc (glibc), not any other libcs. If other libcs are
219 incompatible with glibc it's on them. However, if there are equivalent POSIX
220 and Linux/GNU-specific APIs, we generally prefer the POSIX APIs. If there
221 aren't, we are happy to use GNU or Linux APIs, and expect non-GNU
222 implementations of libc to catch up with glibc.
224 ## Using C Constructs
226 - Allocate local variables where it makes sense: at the top of the block, or at
227 the point where they can be initialized. Avoid huge variable declaration
228 lists at the top of the function.
230 As an exception, `int r` is typically used for a local state variable, but
231 should almost always be declared as the last variable at the top of the
243 uint64_t b = a + 1, c;
245 r = foobarify(a, b, &c);
249 const char *pretty = prettify(a, b, c);
254 - Do not mix multiple variable definitions with function invocations or
255 complicated expressions:
275 - Use `goto` for cleaning up, and only use it for that. I.e. you may only jump
276 to the end of a function, and little else. Never jump backwards!
278 - To minimize strict aliasing violations, we prefer unions over casting.
280 - Instead of using `memzero()`/`memset()` to initialize structs allocated on
281 the stack, please try to use c99 structure initializers. It's short, prettier
282 and actually even faster at execution. Hence:
300 - To implement an endless loop, use `for (;;)` rather than `while (1)`. The
301 latter is a bit ugly anyway, since you probably really meant `while
302 (true)`. To avoid the discussion what the right always-true expression for an
303 infinite while loop is, our recommendation is to simply write it without any
304 such expression by using `for (;;)`.
306 - To determine the length of a constant string `"foo"`, don't bother with
307 `sizeof("foo")-1`, please use `strlen()` instead (both gcc and clang optimize
308 the call away for fixed strings). The only exception is when declaring an
309 array. In that case use `STRLEN()`, which evaluates to a static constant and
310 doesn't force the compiler to create a VLA.
312 - Please use C's downgrade-to-bool feature only for expressions that are
313 actually booleans (or "boolean-like"), and not for variables that are really
314 numeric. Specifically, if you have an `int b` and it's only used in a boolean
315 sense, by all means check its state with `if (b) …` — but if `b` can actually
316 have more than two semantic values, and you want to compare for non-zero,
317 then please write that explicitly with `if (b != 0) …`. This helps readability
318 as the value range and semantical behaviour is directly clear from the
319 condition check. As a special addition: when dealing with pointers which you
320 want to check for non-NULL-ness, you may also use downgrade-to-bool feature.
322 - Please do not use yoda comparisons, i.e. please prefer the more readable `if
323 (a == 7)` over the less readable `if (7 == a)`.
327 - The destructors always deregister the object from the next bigger object, not
328 the other way around.
330 - For robustness reasons, destructors should be able to destruct
331 half-initialized objects, too.
333 - When you define a destructor or `unref()` call for an object, please accept a
334 `NULL` object and simply treat this as NOP. This is similar to how libc
335 `free()` works, which accepts `NULL` pointers and becomes a NOP for them. By
336 following this scheme a lot of `if` checks can be removed before invoking
337 your destructor, which makes the code substantially more readable and robust.
339 - Related to this: when you define a destructor or `unref()` call for an
340 object, please make it return the same type it takes and always return `NULL`
341 from it. This allows writing code like this:
347 which will always work regardless if `p` is initialized or not, and
348 guarantees that `p` is `NULL` afterwards, all in just one line.
350 ## Common Function Naming
352 - Name destructor functions that destroy an object in full freeing all its
353 memory and associated resources (and thus invalidating the pointer to it)
354 `xyz_free()`. Example: `strv_free()`.
356 - Name destructor functions that destroy only the referenced content of an
357 object but leave the object itself allocated `xyz_done()`. If it resets all
358 fields so that the object can be reused later call it `xyz_clear()`.
360 - Functions that decrease the reference counter of an object by one should be
361 called `xyz_unref()`. Example: `json_variant_unref()`. Functions that
362 increase the reference counter by one should be called `xyz_ref()`. Example:
367 - Error codes are returned as negative `Exxx`. e.g. `return -EINVAL`. There are
368 some exceptions: for constructors, it is OK to return `NULL` on OOM. For
369 lookup functions, `NULL` is fine too for "not found".
371 Be strict with this. When you write a function that can fail due to more than
372 one cause, it *really* should have an `int` as the return value for the error
375 - libc system calls typically return -1 on error (with the error code in
376 `errno`), and >= 0 on success. Use the RET_NERRNO() helper if you are looking
377 for a simple way to convert this libc style error returning into systemd
378 style error returning. e.g.
382 r = RET_NERRNO(unlink(t));
390 r = RET_NERRNO(open("/some/file", O_RDONLY|O_CLOEXEC));
394 - Do not bother with error checking whether writing to stdout/stderr worked.
396 - Do not log errors from "library" code, only do so from "main program"
397 code. (With one exception: it is OK to log with DEBUG level from any code,
398 with the exception of maybe inner loops).
400 - In public API calls, you **must** validate all your input arguments for
401 programming error with `assert_return()` and return a sensible return
402 code. In all other calls, it is recommended to check for programming errors
403 with a more brutal `assert()`. We are more forgiving to public users than for
404 ourselves! Note that `assert()` and `assert_return()` really only should be
405 used for detecting programming errors, not for runtime errors. `assert()` and
406 `assert_return()` by usage of `_likely_()` inform the compiler that it should
407 not expect these checks to fail, and they inform fellow programmers about the
408 expected validity and range of parameters.
410 - When you invoke certain calls like `unlink()`, or `mkdir_p()` and you know it
411 is safe to ignore the error it might return (because a later call would
412 detect the failure anyway, or because the error is in an error path and you
413 thus couldn't do anything about it anyway), then make this clear by casting
414 the invocation explicitly to `(void)`. Code checks like Coverity understand
415 that, and will not complain about ignored error codes. Hence, please use
419 (void) unlink("/foo/bar/baz");
422 instead of just this:
425 unlink("/foo/bar/baz");
428 When returning from a `void` function, you may also want to shorten the error
429 path boilerplate by returning a function invocation cast to `(void)` like so:
432 if (condition_not_met)
433 return (void) log_tests_skipped("Cannot run ...");
436 Don't cast function calls to `(void)` that return no error
437 conditions. Specifically, the various `xyz_unref()` calls that return a
438 `NULL` object shouldn't be cast to `(void)`, since not using the return value
439 does not hide any errors.
441 - When returning a return code from `main()`, please preferably use
442 `EXIT_FAILURE` and `EXIT_SUCCESS` as defined by libc.
446 - For every function you add, think about whether it is a "logging" function or
447 a "non-logging" function. "Logging" functions do (non-debug) logging on their
448 own, "non-logging" functions never log on their own (except at debug level)
449 and expect their callers to log. All functions in "library" code, i.e. in
450 `src/shared/` and suchlike must be "non-logging". Every time a "logging"
451 function calls a "non-logging" function, it should log about the resulting
452 errors. If a "logging" function calls another "logging" function, then it
453 should not generate log messages, so that log messages are not generated
454 twice for the same errors. (Note that debug level logging — at syslog level
455 `LOG_DEBUG` — is not considered logging in this context, debug logging is
456 generally always fine and welcome.)
458 - If possible, do a combined log & return operation:
463 return log_(error|warning|notice|...)_errno(r, "Failed to ...: %m");
466 If the error value is "synthetic", i.e. it was not received from
467 the called function, use `SYNTHETIC_ERRNO` wrapper to tell the logging
468 system to not log the errno value, but still return it:
471 n = read(..., s, sizeof s);
473 return log_error_errno(SYNTHETIC_ERRNO(EIO), "Failed to read ...");
478 - Always check OOM. There is no excuse. In program code, you can use
479 `log_oom()` for then printing a short message, but not in "library" code.
481 - Avoid fixed-size string buffers, unless you really know the maximum size and
482 that maximum size is small. It is often nicer to use dynamic memory,
483 `alloca_safe()` or VLAs. If you do allocate fixed-size strings on the stack,
484 then it is probably only OK if you either use a maximum size such as
485 `LINE_MAX`, or count in detail the maximum size a string can
486 have. (`DECIMAL_STR_MAX` and `DECIMAL_STR_WIDTH` macros are your friends for
489 Or in other words, if you use `char buf[256]` then you are likely doing
492 - Make use of `_cleanup_free_` and friends. It makes your code much nicer to
495 - Do not use `alloca()`, `strdupa()` or `strndupa()` directly. Use
496 `alloca_safe()`, `strdupa_safe()` or `strndupa_safe()` instead. (The
497 difference is that the latter include an assertion that the specified size is
498 below a safety threshold, so that the program rather aborts than runs into
499 possible stack overruns.)
501 - Use `alloca_safe()`, but never forget that it is not OK to invoke
502 `alloca_safe()` within a loop or within function call
503 parameters. `alloca_safe()` memory is released at the end of a function, and
504 not at the end of a `{}` block. Thus, if you invoke it in a loop, you keep
505 increasing the stack pointer without ever releasing memory again. (VLAs have
506 better behavior in this case, so consider using them as an alternative.)
507 Regarding not using `alloca_safe()` within function parameters, see the BUGS
508 section of the `alloca(3)` man page.
510 - If you want to concatenate two or more strings, consider using `strjoina()`
511 or `strjoin()` rather than `asprintf()`, as the latter is a lot slower. This
512 matters particularly in inner loops (but note that `strjoina()` cannot be
517 - Avoid leaving long-running child processes around, i.e. `fork()`s that are
518 not followed quickly by an `execv()` in the child. Resource management is
519 unclear in this case, and memory CoW will result in unexpected penalties in
520 the parent much, much later on.
522 - Don't block execution for arbitrary amounts of time using `usleep()` or a
523 similar call, unless you really know what you do. Just "giving something some
524 time", or so is a lazy excuse. Always wait for the proper event, instead of
525 doing time-based poll loops.
527 - Whenever installing a signal handler, make sure to set `SA_RESTART` for it,
528 so that interrupted system calls are automatically restarted, and we minimize
529 hassles with handling `EINTR` (in particular as `EINTR` handling is pretty
532 - When applying C-style unescaping as well as specifier expansion on the same
533 string, always apply the C-style unescaping first, followed by the specifier
534 expansion. When doing the reverse, make sure to escape `%` in specifier-style
535 first (i.e. `%` → `%%`), and then do C-style escaping where necessary.
537 - Be exceptionally careful when formatting and parsing floating point
538 numbers. Their syntax is locale dependent (i.e. `5.000` in en_US is generally
539 understood as 5, while in de_DE as 5000.).
541 - Make sure to enforce limits on every user controllable resource. If the user
542 can allocate resources in your code, your code must enforce some form of
543 limits after which it will refuse operation. It's fine if it is hard-coded
544 (at least initially), but it needs to be there. This is particularly
545 important for objects that unprivileged users may allocate, but also matters
546 for everything else any user may allocate.
550 - Think about the types you use. If a value cannot sensibly be negative, do not
551 use `int`, but use `unsigned`. We prefer `unsigned` form to `unsigned int`.
553 - Use `char` only for actual characters. Use `uint8_t` or `int8_t` when you
554 actually mean a byte-sized signed or unsigned integers. When referring to a
555 generic byte, we generally prefer the unsigned variant `uint8_t`. Do not use
556 types based on `short`. They *never* make sense. Use `int`, `long`, `long
557 long`, all in unsigned and signed fashion, and the fixed-size types
558 `uint8_t`, `uint16_t`, `uint32_t`, `uint64_t`, `int8_t`, `int16_t`, `int32_t`
559 and so on, as well as `size_t`, but nothing else. Do not use kernel types
560 like `u32` and so on, leave that to the kernel.
562 - Stay uniform. For example, always use `usec_t` for time values. Do not mix
563 `usec` and `msec`, and `usec` and whatnot.
565 - Never use the `off_t` type, and particularly avoid it in public APIs. It's
566 really weirdly defined, as it usually is 64-bit and we don't support it any
567 other way, but it could in theory also be 32-bit. Which one it is depends on
568 a compiler switch chosen by the compiled program, which hence corrupts APIs
569 using it unless they can also follow the program's choice. Moreover, in
570 systemd we should parse values the same way on all architectures and cannot
571 expose `off_t` values over D-Bus. To avoid any confusion regarding conversion
572 and ABIs, always use simply `uint64_t` directly.
574 - Unless you allocate an array, `double` is always a better choice than
575 `float`. Processors speak `double` natively anyway, so there is no speed
576 benefit, and on calls like `printf()` `float`s get promoted to `double`s
577 anyway, so there is no point.
579 - Use the bool type for booleans, not integers. One exception: in public
580 headers (i.e those in `src/systemd/sd-*.h`) use integers after all, as `bool`
581 is C99 and in our public APIs we try to stick to C89 (with a few extensions;
586 - Do not issue NSS requests (that includes user name and hostname lookups)
587 from PID 1 as this might trigger deadlocks when those lookups involve
588 synchronously talking to services that we would need to start up.
590 - Do not synchronously talk to any other service from PID 1, due to risk of
595 - When you allocate a file descriptor, it should be made `O_CLOEXEC` right from
596 the beginning, as none of our files should leak to forked binaries by
597 default. Hence, whenever you open a file, `O_CLOEXEC` must be specified,
598 right from the beginning. This also applies to sockets. Effectively, this
599 means that all invocations to:
601 - `open()` must get `O_CLOEXEC` passed,
602 - `socket()` and `socketpair()` must get `SOCK_CLOEXEC` passed,
603 - `recvmsg()` must get `MSG_CMSG_CLOEXEC` set,
604 - `F_DUPFD_CLOEXEC` should be used instead of `F_DUPFD`, and so on,
605 - invocations of `fopen()` should take `e`.
607 - It's a good idea to use `O_NONBLOCK` when opening 'foreign' regular files,
608 i.e. file system objects that are supposed to be regular files whose paths
609 were specified by the user and hence might actually refer to other types of
610 file system objects. This is a good idea so that we don't end up blocking on
611 'strange' file nodes, for example, if the user pointed us to a FIFO or device
612 node which may block when opening. Moreover even for actual regular files
613 `O_NONBLOCK` has a benefit: it bypasses any mandatory lock that might be in
614 effect on the regular file. If in doubt consider turning off `O_NONBLOCK`
617 - These days we generally prefer `openat()`-style file APIs, i.e. APIs that
618 accept a combination of file descriptor and path string, and where the path
619 (if not absolute) is considered relative to the specified file
620 descriptor. When implementing library calls in similar style, please make
621 sure to imply `AT_EMPTY_PATH` if an empty or `NULL` path argument is
622 specified (and convert that latter to an empty string). This differs from the
623 underlying kernel semantics, where `AT_EMPTY_PATH` must always be specified
624 explicitly, and `NULL` is not accepted as path.
628 - If you parse a command line, and want to store the parsed parameters in
629 global variables, please consider prefixing their names with `arg_`. We have
630 been following this naming rule in most of our tools, and we should continue
631 to do so, as it makes it easy to identify command line parameter variables,
632 and makes it clear why it is OK that they are global variables.
634 - Command line option parsing:
635 - Do not print full `help()` on error, be specific about the error.
636 - Do not print messages to stdout on error.
637 - Do not POSIX_ME_HARDER unless necessary, i.e. avoid `+` in option string.
641 - Variables and functions **must** be static, unless they have a prototype, and
642 are supposed to be exported.
644 - Public API calls (i.e. functions exported by our shared libraries)
645 must be marked `_public_` and need to be prefixed with `sd_`. No
646 other functions should be prefixed like that.
648 - When exposing public C APIs, be careful what function parameters you make
649 `const`. For example, a parameter taking a context object should probably not
650 be `const`, even if you are writing an otherwise read-only accessor function
651 for it. The reason is that making it `const` fixates the contract that your
652 call won't alter the object ever, as part of the API. However, that's often
653 quite a promise, given that this even prohibits object-internal caching or
654 lazy initialization of object variables. Moreover, it's usually not too
655 useful for client applications. Hence, please be careful and avoid `const` on
656 object parameters, unless you are very sure `const` is appropriate.
658 ## Referencing Concepts
660 - When referring to a configuration file option in the documentation and such,
661 please always suffix it with `=`, to indicate that it is a configuration file
664 - When referring to a command line option in the documentation and such, please
665 always prefix with `--` or `-` (as appropriate), to indicate that it is a
668 - When referring to a file system path that is a directory, please always
669 suffix it with `/`, to indicate that it is a directory, not a regular file
670 (or other file system object).
672 ## Functions to Avoid
674 - Use `memzero()` or even better `zero()` instead of `memset(..., 0, ...)`
676 - Please use `streq()` and `strneq()` instead of `strcmp()`, `strncmp()` where
677 applicable (i.e. wherever you just care about equality/inequality, not about
680 - Never use `strtol()`, `atoi()` and similar calls. Use `safe_atoli()`,
681 `safe_atou32()` and suchlike instead. They are much nicer to use in most
682 cases and correctly check for parsing errors.
684 - `htonl()`/`ntohl()` and `htons()`/`ntohs()` are weird. Please use `htobe32()`
685 and `htobe16()` instead, it's much more descriptive, and actually says what
686 really is happening, after all `htonl()` and `htons()` don't operate on
687 `long`s and `short`s as their name would suggest, but on `uint32_t` and
688 `uint16_t`. Also, "network byte order" is just a weird name for "big endian",
689 hence we might want to call it "big endian" right-away.
691 - Use `typesafe_inet_ntop()`, `typesafe_inet_ntop4()`, and
692 `typesafe_inet_ntop6()` instead of `inet_ntop()`. But better yet, use the
693 `IN_ADDR_TO_STRING()`, `IN4_ADDR_TO_STRING()`, and `IN6_ADDR_TO_STRING()`
694 macros which allocate an anonymous buffer internally.
696 - Please never use `dup()`. Use `fcntl(fd, F_DUPFD_CLOEXEC, 3)` instead. For
697 two reasons: first, you want `O_CLOEXEC` set on the new `fd` (see
698 above). Second, `dup()` will happily duplicate your `fd` as 0, 1, 2,
699 i.e. stdin, stdout, stderr, should those `fd`s be closed. Given the special
700 semantics of those `fd`s, it's probably a good idea to avoid
701 them. `F_DUPFD_CLOEXEC` with `3` as parameter avoids them.
703 - Don't use `fgets()`, it's too hard to properly handle errors such as overly
704 long lines. Use `read_line()` instead, which is our own function that handles
705 this much more nicely.
707 - Don't invoke `exit()`, ever. It is not replacement for proper error
708 handling. Please escalate errors up your call chain, and use normal `return`
709 to exit from the main function of a process. If you `fork()`ed off a child
710 process, please use `_exit()` instead of `exit()`, so that the exit handlers
713 - Do not use `basename()` or `dirname()`. The semantics in corner cases are
714 full of pitfalls, and the fact that there are two quite different versions of
715 `basename()` (one POSIX and one GNU, of which the latter is much more useful)
716 doesn't make it better either. Use path_extract_filename() and
717 path_extract_directory() instead.
719 - Never use `FILENAME_MAX`. Use `PATH_MAX` instead (for checking maximum size
720 of paths) and `NAME_MAX` (for checking maximum size of filenames).
721 `FILENAME_MAX` is not POSIX, and is a confusingly named alias for `PATH_MAX`
722 on Linux. Note that `NAME_MAX` does not include space for a trailing `NUL`,
723 but `PATH_MAX` does. UNIX FTW!
727 - Commit message subject lines should be prefixed with an appropriate component
728 name of some kind. For example, "journal: ", "nspawn: " and so on.
730 - Do not use "Signed-Off-By:" in your commit messages. That's a kernel thing we
731 don't do in the systemd project.
735 - The best place for code comments and explanations is in the code itself. Only
736 the second best is in git commit messages. The worst place is in the GitHub
737 PR cover letter. Hence, whenever you type a commit message consider for a
738 moment if what you are typing there wouldn't be a better fit for an in-code
739 comment. And if you type the cover letter of a PR, think hard if this
740 wouldn't be better as a commit message or even code comment. Comments are
741 supposed to be useful for somebody who reviews the code, and hence hiding
742 comments in git commits or PR cover letters makes reviews unnecessarily
743 hard. Moreover, while we rely heavily on GitHub's project management
744 infrastructure we'd like to keep everything that can reasonably be kept in
745 the git repository itself in the git repository, so that we can theoretically
746 move things elsewhere with the least effort possible.
748 - It's OK to reference GitHub PRs, GitHub issues and git commits from code
749 comments. Cross-referencing code, issues, and documentation is a good thing.
751 - Reasonable use of non-ASCII Unicode UTF-8 characters in code comments is
752 welcome. If your code comment contains an emoji or two this will certainly
753 brighten the day of the occasional reviewer of your code. Really! 😊
757 - We generally avoid using threads, to the level this is possible. In
758 particular in the service manager/PID 1 threads are not OK to use. This is
759 because you cannot mix memory allocation in threads with use of glibc's
760 `clone()` call, or manual `clone()`/`clone3()` system call wrappers. Only
761 glibc's own `fork()` call will properly synchronize the memory allocation
762 locks around the process clone operation. This means that if a process is
763 cloned via `clone()`/`clone3()` and another thread currently has the
764 `malloc()` lock taken, it will be cloned in locked state to the child, and
765 thus can never be acquired in the child, leading to deadlocks. Hence, when
766 using `clone()`/`clone3()` there are only two ways out: never use threads in the
767 parent, or never do memory allocation in the child. For our uses we need
768 `clone()`/`clone3()` and hence decided to avoid threads. Of course, sometimes the
769 concurrency threads allow is beneficial, however we suggest forking off
770 worker *processes* rather than worker *threads* for this purpose, ideally
771 even with an `execve()` to remove the CoW trap situation `fork()` easily
774 - A corollary of the above is: never use `clone()` where a `fork()` would do
775 too. Also consider using `posix_spawn()` which combines `clone()` +
776 `execve()` into one and has nice properties since it avoids becoming a CoW
777 trap by using `CLONE_VORK` and `CLONE_VM` together.
779 - While we avoid forking off threads on our own, writing thread-safe code is a
780 good idea where it might end up running inside of libsystemd.so or
781 similar. Hence, use TLS (i.e. `thread_local`) where appropriate, and maybe
782 the occasional `pthread_once()`.