C Support in Clang
Clang implements the following published and upcoming ISO C standards:
Language Standard | Flag | Available in Clang? |
---|---|---|
C2y | -std=c2y | Partial |
C23 | -std=c23 | Partial |
C17 | -std=c17 | Partial |
C11 | -std=c11 | Partial |
C99 | -std=c99 | Clang 17 |
C89 | -std=c89 | Yes |
The implementation status for C11 and C23 are currently under investigation. Any proposal whose status in Clang is currently unknown will be marked in magenta.
The Clang community is continually striving to improve C standards compliance between releases by submitting and tracking C Defect Reports and implementing resolutions as they become available.
The LLVM bug tracker uses the "c", "c99", "c11", "c17", "c23", and "c2y" labels to track known bugs with Clang's language conformance.
C2y implementation status
Clang has support for some of the features of the C standard following C23, informally referred to as C2y.
You can use Clang in C2y mode with the -std=c2y
option (available in Clang 19 and later).
List of features and minimum Clang version with support
Language Feature | C2y Proposal | Available in Clang? |
---|---|---|
Sequential hexdigits | N3192 | Yes |
Generic selection expression with a type operand | N3260 | Clang 17 |
Round-trip rounding | N3232 | Yes |
Accessing byte arrays | N3254 | Yes |
Slay some earthly demons I | N3244 |
PartialClang does not document the implementation-defined behavior for decay of an array with the register storage class specifier. Clang does not diagnose anextern inline function with no definition in
the TU. Clang accepts and rejects redeclarations with/without an
alignment specifier, depending on the order of the declarations.
|
Support ++ and -- on complex values | N3259 | Yes |
Usability of a byte-wise copy of va_list | N3262 | Yes |
alignof of an incomplete array type | N3273 | Clang 3.5 |
Remove imaginary types | N3274 | Yes |
Floating-point exceptions for macro replacements | N3286 | Yes |
Obsolete implicitly octal literals and add delimited escape sequences | N3353 | No |
'if' declarations, v2 | N3356 | No |
Allowing stricter alignment for atomic types | N3312 | Yes |
Slay Some Earthly Demons III | N3341 | Unknown |
Slay Some Earthly Demons IV | N3342 | Unknown |
Slay Some Earthly Demons VI | N3344 | Unknown |
Slay Some Earthly Demons VII | N3345 | Yes |
Slay Some Earthly Demons VIII | N3346 | Unknown |
Introduce complex literals v. 2 | N3298 | Unknown |
Allow zero length operations on null pointers | N3322 | Yes |
How do you add one to something? | N3323 | Yes |
Give consistent wording for SNAN initialization v3 | N3364 | Unknown |
Case range expressions v3.1 | N3370 | Unknown |
New _Lengthof() operator (v4) | N3369 | No |
Named loops, v3 | N3355 | No |
C23 implementation status
Clang has support for some of the features of the C standard following C17, informally referred to as C23.
You can use Clang in C23 mode with the -std=c23
option (available in Clang 18 and later) or with the
-std=c2x
option (available in Clang 9 and later).
List of features and minimum Clang version with support
Language Feature | C23 Proposal | Available in Clang? |
---|---|---|
Evaluation formats | N2186 | Unknown |
Harmonizing static_assert with C++ | N2665 | Clang 9 |
nodiscard attribute | N2267 | Clang 9 |
maybe_unused attribute | N2270 | Clang 9 |
TS 18661 Integration | ||
N2314 | Unknown | |
N2341 | No | |
N2359 | No | |
N2546 | No | |
N2640 | Yes | |
N2755 | No | |
N2931 | No | |
Preprocessor line numbers unspecified | N2322 |
PartialThe line number associated with a macro invocation is not the line number of the first character of the macro name in the invocation. Additionally, Clang may not associate the line number of a pp-directive with the first# token. As these are recommended practices
and not normative requirements, Clang's behavior is still conforming.
|
deprecated attribute | N2334 | Clang 9 |
Attributes | ||
N2335 | Clang 9 | |
N2554 | Clang 9 | |
Defining new types in offsetof | N2350 | Yes |
fallthrough attribute | N2408 | Clang 9 |
Two's complement sign representation | N2412 | Clang 14 |
Adding the u8 character prefix | N2418 | Clang 15 |
Remove support for function definitions with identifier lists | N2432 | Clang 15 |
Annex F.8 update for implementation extensions and rounding | N2384 | Unknown |
_Bool definitions for true and false | N2393 | Subsumed by N2935 |
[[nodiscard("should have a reason")]] | N2448 | Clang 10 |
Allowing unnamed parameters in function definitions | N2480 | Clang 11 |
Free positioning of labels inside compound statements | N2508 | Clang 18 |
Querying attribute support | N2553 | Clang 9 |
Binary literals | N2549 | Clang 9 |
Allow duplicate attributes | N2557 | Clang 13 |
Character encoding of diagnostic text | N2563 | Yes |
What we think we reserve | N2572 | Partial |
Remove mixed wide string literal concatenation | N2594 | Clang 9 |
Update to IEC 60559:2020 | N2600 | Unknown |
Compatibility of Pointers to Arrays with Qualifiers | N2607 |
PartialMuch of the proposal is implemented, but Clang lacks pedantic diagnostics in C17 and earlier regarding use of incompatible pointer types as an extension. Further, Clang does not properly compute the correct result type for the?: operator when the result type should be a
qualified array type.
|
Unclear type relationship between a format specifier and its argument | N2562 | Clang 16 |
String functions for freestanding implementations | N2524 | No |
Digit separators | N2626 | Clang 13 |
Missing +(x) in table | N2641 | Yes |
Add support for preprocessing directives elifdef and elifndef | N2645 | Clang 13 |
[[maybe_unused]] for labels | N2662 | Clang 16 |
Zeros compare equal | N2670 | Yes |
Negative values | N2671 | Yes |
5.2.4.2.2 cleanup | N2672 | Yes |
Towards Integer Safety | N2683 | Clang 18 |
Adding Fundamental Type for N-bit Integers | ||
N2763 | Clang 15 | |
N2775 | Clang 15 | |
N2969 | Clang 15 | |
N3035 | Clang 15 | |
#warning directive | N2686 | Yes |
Sterile characters | N2686 | Yes |
Numerically equal | N2716 | Yes |
char16_t & char32_t string literals shall be UTF-16 & UTF-32 | N2728 | Yes |
IEC 60559 binding | N2749 | Unknown |
__has_include for C | N2799 | Yes |
Annex F overflow and underflow | N2747 | Yes |
Remove UB from Incomplete Types in Function Parameters | N2770 | Yes |
Variably-modified types | N2778 | Yes |
Types do not have types | N2781 | Yes |
5.2.4.2.2 cleanup (N2672 update) | N2806 | Yes |
Allow 16-bit ptrdiff_t | N2808 | Yes |
Proposal to update CFP freestanding requirements | N2823 | Unknown |
Types and sizes | N2838 | Yes |
Clarifying integer terms | N2837 | Yes |
Clarification for max exponent macros | N2843 | Subsumed by N2882 |
Clarification about expression transformations | N2846 | Unknown |
Contradiction about INFINITY macro | N2848 | Clang 19 |
Require exact-width integer type interfaces | N2872 | Yes |
@, $, and ‘ in the source/execution character set | N2701 | Yes |
Quantum exponent of NaN (version 2) | N2754 | Unknown |
The noreturn attribute | N2764 | Clang 15 |
*_HAS_SUBNORM==0 implies what? | N2797 | Yes |
Disambiguate the storage class of some compound literals | N2819 | No |
Add annotations for unreachable control flow v2 | N2826 | Clang 17 |
Unicode Sequences More Than 21 Bits are a Constraint Violation r0 | N2828 | Clang 3.6 |
Identifier Syntax using Unicode Standard Annex 31 | N2836 | Clang 15 |
No function declarators without prototypes | N2841 | Clang 15 |
Remove default argument promotions for _FloatN types | N2844 | No |
Revised Suggestions of Change for Numerically Equal/Equivalent | N2847 | Yes |
5.2.4.2.2 Cleanup, Again Again (N2806 update) | N2879 | Yes |
char8_t: A type for UTF-8 characters and strings | N2653 | Clang 19 |
Clarification for max exponent macros-update | N2882 | Unknown |
Consistent, Warningless, and Intuitive Initialization with {} | ||
N2900 | Clang 17 | |
N3011 | Clang 17 | |
Not-so-magic: typeof | ||
N2927 | Clang 16 | |
N2930 | Clang 16 | |
Revise spelling of keywords v7 | N2934 | Clang 17 |
Make false and true first-class language features v8 | N2935 | Clang 15 |
Properly define blocks as part of the grammar v3 | N2937 | Yes |
Annex X (replacing Annex H) for IEC 60559 interchange | N2601 | No |
Indeterminate Values and Trap Representations | N2861 | Yes |
Remove ATOMIC_VAR_INIT v2 | N2886 | Clang 17 |
Require exact-width integer type interfaces v2 | N2888 | Yes |
Wording Change for Variably-Modified Types | N2992 | Yes |
Identifier syntax fixes | N2939 | Clang 15 |
Remove trigraphs??! | N2940 | Clang 18 |
Improved normal enumerations | N3029 | Clang 20 |
Relax requirements for va_start | N2975 | Clang 16 |
Enhanced enumerations | N3030 | Clang 20 |
Freestanding C and IEC 60559 conformance scope reduction | N2951 | Unknown |
Unsequenced functions | N2956 | No |
Comma ommission and deletion (__VA_OPT__) | N3033 | Clang 12 |
Underspecified object definitions | N3006 | No |
Type inference for object declarations | N3007 | Clang 18 |
constexpr for object definitions | N3018 | Clang 19 |
Introduce storage class specifiers for compound literals | N3038 | No |
Identifier primary expressions | N3034 | Yes |
Introduce the nullptr constant | N3042 | Clang 17 |
Memory layout of unions | N2929 | Yes |
Improved tag compatibility | N3037 | No |
#embed | N3017 | Clang 19 |
C17 implementation status
There are no major changes in this edition, only technical corrections and clarifications that are tracked by Defect Report.
You can use Clang in C17 mode with the -std=c17
or -std=c18
options (available in Clang 6 and later).
C11 implementation status
Clang implements a significant portion of the ISO 9899:2011 (C11) standard, but the status of individual proposals is still under investigation.
You can use Clang in C11 mode with the -std=c11
option (use -std=c1x
in Clang 3.0 and earlier).
List of features and minimum Clang version with support
Language Feature | C11 Proposal | Available in Clang? |
---|---|---|
A finer-grained specification for sequencing | N1252 | Unknown |
Clarification of expressions | N1282 | Yes |
Extending the lifetime of temporary objects (factored approach) | N1285 | No |
Requiring signed char to have no padding bits | N1310 | Yes |
Initializing static or external variables | N1311 | Yes |
Conversion between pointers and floating types | N1316 | Yes |
Adding TR 19769 to the C Standard Library | N1326 | Clang 3.3 |
Static assertions | N1330 | Yes |
Parallel memory sequencing model proposal | N1349 | Unknown |
_Bool bit-fields | N1356 | Yes |
Technical corrigendum for C1X | N1359 | Yes |
Benign typedef redefinition | N1360 | Clang 3.1 |
Thread-local storage | N1364 | Clang 3.3 |
Constant expressions | N1365 | Clang 16 |
Contractions and expression evaluation methods | N1367 | Unknown |
Floating-point to int/_Bool conversions | N1391 | Yes |
Wide function returns (alternate proposal) | N1396 |
Yes*Clang conforms to this paper on all targets except 32-bit x86 without SSE2. However, Clang does not claim conformance to Annex F on any target and does not intend to ever conform to Annex F on that specific target, so no changes are needed to conform to this paper. |
Alignment | ||
N1397 | Clang 3.2 | |
N1447 | Clang 3.2 | |
Anonymous member-structures and unions (modulo "name lookup") | N1406 | Yes |
Completeness of types | N1439 | Yes |
Generic macro facility | N1441 | Yes |
Dependency ordering for C memory model | N1444 | Unknown |
Subsetting the standard | N1460 | Yes |
Assumed types in F.9.2 | N1468 | Unknown |
Supporting the 'noreturn' property in C1x | N1478 | Clang 3.3 |
Updates to C++ memory model based on formalization | N1480 | Unknown |
Explicit initializers for atomics | N1482 | Clang 4 |
Atomics proposal (minus ternary op) | N1485 | Yes |
UTF-8 string literals | N1488 | Clang 3.3 |
Optimizing away infinite loops | N1509 | Yes |
Conditional normative status for Annex G | N1514 | Yes (1) |
Creation of complex value | N1464 | Clang 12 |
Recommendations for extended identifier characters for C and C++ | N1518 | Unknown |
Atomic C1x/C++0x compatibility refinements (1st part only) | N1526 | Yes |
Atomic bitfields implementation defined | N1530 | Yes |
Small fix for the effect of alignment on struct/union type compatibility | N1532 | Yes |
Clarification for wide evaluation | N1531 | Unknown |
__STDC_IEC_559_COMPLEX__
macro.
C99 implementation status
Clang implements all of the ISO 9899:1999 (C99) standard.
Note, the list of C99 features comes from the C99 committee draft. Not all C99 documents are publicly available, so the documents referenced in this section may be inaccurate, unknown, or not linked.
You can use Clang in C99 mode with the -std=c99
option.
List of features and minimum Clang version with support
Language Feature | C99 Proposal | Available in Clang? |
---|---|---|
restricted character set support via digraphs and <iso646.h> | Unknown | Yes |
more precise aliasing rules via effective type | Unknown | Yes |
restricted pointers | N448 |
PartialClang's support forrestrict is fully conforming but
considered only partially implemented. Clang implements all of the
constraints required for restrict support, but LLVM only
supports restrict optimization semantics for restricted
pointers used as function parameters; it does not fully support the
semantics for restrict on local variables or data members.
|
variable length arrays | N683 | Yes |
flexible array members | Unknown | Yes |
static and type qualifiers in parameter array declarators | Unknown | Yes |
complex and imaginary support in <complex.h> | N693 |
PartialClang supports_Complex type specifiers but does not
support _Imaginary type specifiers. Support for
_Imaginary is optional in C99 and Clang does not claim
conformance to Annex G._Complex support requires an underlying support library
such as compiler-rt to provide functions like __divsc3 .
Compiler-rt isn't linked in automatically in MSVC style environments.
|
type-generic math macros in <tgmath.h> | N693 | Yes |
the long long int type | N601 | Yes |
increase minimum translation limits | N590 | Clang 3.2 |
additional floating-point characteristics in <float.h> | Unknown | Clang 16 |
remove implicit int | ||
N635 | Yes | |
N692 | Yes | |
N722 | Yes | |
reliable integer division | N617 | Yes |
universal character names (\u and \U) | Unknown | Yes |
extended identifiers | N717 | Clang 17 |
hexadecimal floating-point constants | N308 | Yes |
compound literals | N716 | Yes |
designated initializers | N494 | Yes |
// comments | N644 | Yes |
extended integer types and library functions in <inttypes.h> and <stdint.h> | Unknown | Yes |
remove implicit function declaration | N636 | Yes |
preprocessor arithmetic done in intmax_t/uintmax_t | N736 | Yes |
mixed declarations and code | N740 | Yes |
new block scopes for selection and iteration statements | Unknown | Yes |
integer constant type rules | N629 | Yes |
integer promotion rules | N725 | Yes |
macros with a variable number of arguments | N707 | Yes |
IEC 60559 support | Unknown |
PartialClang supports much of the language requirements for Annex F, but full conformance is only possible to determine when considering the compiler's language support, the C runtime library's math library support, and the target system's floating-point environment support. Clang does not currently raise an "invalid" floating-point exception on certain conversions, does not raise floating-point exceptions for arithmetic constant expressions, and other corner cases. Note, Clang does not define__STDC_IEC_559__ because the compiler
does not fully conform. However, some C standard library
implementations
(
glibc,
musl will define the macro regardless of compiler support unless
the compiler defines __GCC_IEC_559 , which Clang does not
currently define.Additionally, Clang intentionally will not conform to Annex F on 32-bit x86 without SSE2 due to the behavior of floating-point operations in x87. |
trailing comma allowed in enum declaration | Unknown | Yes |
inline functions | N741 | Yes |
boolean type in <stdbool.h> | N815 | Yes |
idempotent type qualifiers | N505 | Yes |
empty macro arguments | N570 | Yes |
additional predefined macro names | Unknown | Yes |
_Pragma preprocessing operator | N634 | Yes |
standard pragmas | ||
N631 | Yes | |
N696 | Yes | |
__func__ predefined identifier | N611 | Yes |
va_copy macro | N671 | Yes |
remove deprecation of aliased array parameters | Unknown | Yes |
conversion of array to pointer not limited to lvalues | N835 | Yes |
relaxed constraints on aggregate and union initialization | N782 | Clang 3.4 |
relaxed restrictions on portable header names | N772 | Yes |
return without an expression not permitted in function that returns a value | Unknown | Yes |
C89 implementation status
Clang implements all of the ISO 9899:1990 (C89) standard.
You can use Clang in C89 mode with the -std=c89
or -std=c90
options.