The Compiler Attributes series is an effort to disentangle
the include/linux/compiler*.h headers and bring them up to date.
The main idea behind the series is to use feature checking macros
(i.e. __has_attribute) instead of compiler version checks (e.g. GCC_VERSION),
which are compiler-agnostic (so they can be shared, reducing the size
of compiler-specific headers) and version-agnostic.
Other related improvements have been performed in the headers as well,
which on top of the use of __has_attribute it has amounted to a significant
simplification of these headers (e.g. GCC_VERSION is now only guarding 4
non-attribute macros).
This series should also help the efforts to support compiling the kernel
with clang and icc. A fair amount of documentation and comments have also
been added, clarified or removed; and the headers are now more readable,
which should help kernel developers in general.
The series was triggered due to the move to gcc >= 4.6. In turn, this series
has also triggered Sparse to gain the ability to recognize __has_attribute
on its own.
You can also fetch it from:
https://github.com/ojeda/linux/tree/compiler-attributes-v4
Enjoy!
Cheers,
Miguel
Cc: Jonathan Corbet <[email protected]>
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Miguel Ojeda <[email protected]>
v3 -> v4
This time it is an important fix I found while doing randconfigs, plus
the documentation sentence that Nick suggested and the MAINTAINERS file.
Compile-tested for a while on (x86_64, gcc-7.3) on a few randconfs.
New:
* Add MAINTAINERS entry for the new file (compiler_attributes.h)
so that we try to maintain that one as clean as possible.
(if someone else wants to join, feel free!)
Modified:
* Fix inline macro
__always_inline cannot be used in the inline macro,
because people marking a function as __always_inline would
expand to inline which in turn would expand back to __always_inline
(under some configs), which would not be expanded again.
Added a comment about this in the inline macro.
Also added another comment about __always_inline in compiler_attributes.h
to note that users expect to not have to write inline (as well as
the attribute). We cannot simply remove "inline" there (which would
solve the problem described above) because using the attribute is
not enough for gcc. A function marked as always_inline seems to require
to be marked as inline itself so that the attribute is applied:
"warning: always_inline function might not be inlinable [-Wattributes]"
From the gcc docs:
"For functions declared inline, this attribute inlines the function
independent of any restrictions that otherwise apply to inlining."
See https://godbolt.org/z/LpzUPj for an example.
From reviews:
* Add sentence about feature-detection in Docs (Nick)
Miguel Ojeda (13):
Compiler Attributes: remove unused attributes
Compiler Attributes: always use the extra-underscores syntax
Compiler Attributes: remove unneeded tests
Compiler Attributes: homogenize __must_be_array
Compiler Attributes: naked was fixed in gcc 4.6
Compiler Attributes: naked can be shared
Compiler Attributes: remove unneeded sparse (__CHECKER__) tests
Compiler Attributes: add missing SPDX ID in compiler_types.h
Compiler Attributes: use feature checks instead of version checks
Compiler Attributes: KENTRY used twice the "used" attribute
Compiler Attributes: remove uses of __attribute__ from compiler.h
Compiler Attributes: add Doc/process/programming-language.rst
Compiler Attributes: Add MAINTAINERS entry
Documentation/process/index.rst | 1 +
.../process/programming-language.rst | 45 ++++
MAINTAINERS | 5 +
include/linux/compiler-clang.h | 5 -
include/linux/compiler-gcc.h | 84 +-----
include/linux/compiler-intel.h | 9 -
include/linux/compiler.h | 19 +-
include/linux/compiler_attributes.h | 244 ++++++++++++++++++
include/linux/compiler_types.h | 105 ++------
9 files changed, 329 insertions(+), 188 deletions(-)
create mode 100644 Documentation/process/programming-language.rst
create mode 100644 include/linux/compiler_attributes.h
--
2.17.1
The naked attribute is supported by at least gcc >= 4.6 (for ARM,
which is the only current user), gcc >= 8 (for x86), clang >= 3.1
and icc >= 13. See https://godbolt.org/z/350Dyc
Therefore, move it out of compiler-gcc.h so that the definition
is shared by all compilers.
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Suggested-by: Arnd Bergmann <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-gcc.h | 8 --------
include/linux/compiler_types.h | 8 ++++++++
2 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 4cd5e9264bce..3b32bbfa5a49 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -72,14 +72,6 @@
#define __noretpoline __attribute__((__indirect_branch__("keep")))
#endif
-/*
- * it doesn't make sense on ARM (currently the only user of __naked)
- * to trace naked functions because then mcount is called without
- * stack and frame pointer being set up and there is no chance to
- * restore the lr register to the value before mcount was called.
- */
-#define __naked __attribute__((__naked__)) notrace
-
#define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
#define __compiletime_object_size(obj) __builtin_object_size(obj, 0)
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 83475515bc39..5ff9cda893f4 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -224,6 +224,14 @@ struct ftrace_likely_data {
#define notrace __attribute__((__no_instrument_function__))
#endif
+/*
+ * it doesn't make sense on ARM (currently the only user of __naked)
+ * to trace naked functions because then mcount is called without
+ * stack and frame pointer being set up and there is no chance to
+ * restore the lr register to the value before mcount was called.
+ */
+#define __naked __attribute__((__naked__)) notrace
+
#define __compiler_offsetof(a, b) __builtin_offsetof(a, b)
/*
--
2.17.1
__optimize and __deprecate_for_modules are unused in
the whole kernel tree. Simply drop them.
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-gcc.h | 2 --
include/linux/compiler.h | 4 ----
include/linux/compiler_types.h | 1 -
3 files changed, 7 deletions(-)
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 763bbad1e258..0a2d06677d83 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -95,8 +95,6 @@
#define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
-#define __optimize(level) __attribute__((__optimize__(level)))
-
#define __compiletime_object_size(obj) __builtin_object_size(obj, 0)
#ifndef __CHECKER__
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 681d866efb1e..7c0157d50964 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -301,10 +301,6 @@ static inline void *offset_to_ptr(const int *off)
#endif /* __ASSEMBLY__ */
-#ifndef __optimize
-# define __optimize(level)
-#endif
-
/* Compile time object size, -1 for unknown */
#ifndef __compiletime_object_size
# define __compiletime_object_size(obj) -1
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 3525c179698c..b6534292ea33 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -108,7 +108,6 @@ struct ftrace_likely_data {
/* Don't. Just don't. */
#define __deprecated
-#define __deprecated_for_modules
#endif /* __KERNEL__ */
--
2.17.1
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Suggested-by: Nick Desaulniers <[email protected]>
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler.h | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 17ee9165ca51..b5fb034fa6fa 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -23,8 +23,8 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
#define __branch_check__(x, expect, is_constant) ({ \
long ______r; \
static struct ftrace_likely_data \
- __attribute__((__aligned__(4))) \
- __attribute__((__section__("_ftrace_annotated_branch"))) \
+ __aligned(4) \
+ __section("_ftrace_annotated_branch") \
______f = { \
.data.func = __func__, \
.data.file = __FILE__, \
@@ -59,8 +59,8 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
({ \
int ______r; \
static struct ftrace_branch_data \
- __attribute__((__aligned__(4))) \
- __attribute__((__section__("_ftrace_branch"))) \
+ __aligned(4) \
+ __section("_ftrace_branch") \
______f = { \
.func = __func__, \
.file = __FILE__, \
@@ -146,7 +146,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
extern typeof(sym) sym; \
static const unsigned long __kentry_##sym \
__used \
- __attribute__((__section__("___kentry" "+" #sym ))) \
+ __section("___kentry" "+" #sym ) \
= (unsigned long)&sym;
#endif
@@ -287,7 +287,7 @@ unsigned long read_word_at_a_time(const void *addr)
* visible to the compiler.
*/
#define __ADDRESSABLE(sym) \
- static void * __attribute__((__section__(".discard.addressable"), used)) \
+ static void * __section(".discard.addressable") __used \
__PASTE(__addressable_##sym, __LINE__) = (void *)&sym;
/**
--
2.17.1
Cc: Jonathan Corbet <[email protected]>
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Miguel Ojeda <[email protected]>
---
Documentation/process/index.rst | 1 +
.../process/programming-language.rst | 45 +++++++++++++++++++
2 files changed, 46 insertions(+)
create mode 100644 Documentation/process/programming-language.rst
diff --git a/Documentation/process/index.rst b/Documentation/process/index.rst
index 37bd0628b6ee..c56f24a22d2a 100644
--- a/Documentation/process/index.rst
+++ b/Documentation/process/index.rst
@@ -23,6 +23,7 @@ Below are the essential guides that every developer should read.
code-of-conflict
development-process
submitting-patches
+ programming-language
coding-style
maintainer-pgp-guide
email-clients
diff --git a/Documentation/process/programming-language.rst b/Documentation/process/programming-language.rst
new file mode 100644
index 000000000000..e5f5f065dc24
--- /dev/null
+++ b/Documentation/process/programming-language.rst
@@ -0,0 +1,45 @@
+.. _programming_language:
+
+Programming Language
+====================
+
+The kernel is written in the C programming language [c-language]_.
+More precisely, the kernel is typically compiled with ``gcc`` [gcc]_
+under ``-std=gnu89`` [gcc-c-dialect-options]_: the GNU dialect of ISO C90
+(including some C99 features).
+
+This dialect contains many extensions to the language [gnu-extensions]_,
+and many of them are used within the kernel as a matter of course.
+
+There is some support for compiling the kernel with ``clang`` [clang]_
+and ``icc`` [icc]_ for several of the architectures, although at the time
+of writing it is not completed, requiring third-party patches.
+
+Attributes
+----------
+
+One of the common extensions used throughout the kernel are attributes
+[gcc-attribute-syntax]_. Attributes allow to introduce
+implementation-defined semantics to language entities (like variables,
+functions or types) without having to make significant syntactic changes
+to the language (e.g. adding a new keyword) [n2049]_.
+
+In some cases, attributes are optional (i.e. a compiler not supporting them
+should still produce proper code, even if it is slower or does not perform
+as many compile-time checks/diagnostics).
+
+The kernel defines pseudo-keywords (e.g. ``__pure``) instead of using
+directly the GNU attribute syntax (e.g. ``__attribute__((__pure__))``)
+in order to feature detect which ones can be used and/or to shorten the code.
+
+Please refer to ``include/linux/compiler_attributes.h`` for more information.
+
+.. [c-language] http://www.open-std.org/jtc1/sc22/wg14/www/standards
+.. [gcc] https://gcc.gnu.org
+.. [clang] https://clang.llvm.org
+.. [icc] https://software.intel.com/en-us/c-compilers
+.. [gcc-c-dialect-options] https://gcc.gnu.org/onlinedocs/gcc/C-Dialect-Options.html
+.. [gnu-extensions] https://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html
+.. [gcc-attribute-syntax] https://gcc.gnu.org/onlinedocs/gcc/Attribute-Syntax.html
+.. [n2049] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2049.pdf
+
--
2.17.1
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
MAINTAINERS | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 9ad052aeac39..a7ef76d694d4 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -3721,6 +3721,11 @@ L: [email protected]
S: Maintained
F: drivers/platform/x86/compal-laptop.c
+COMPILER ATTRIBUTES
+M: Miguel Ojeda <[email protected]>
+S: Maintained
+F: include/linux/compiler_attributes.h
+
CONEXANT ACCESSRUNNER USB DRIVER
L: [email protected]
W: http://accessrunner.sourceforge.net/
--
2.17.1
Instead of using version checks per-compiler to define (or not)
each attribute, use __has_attribute to test for them, following
the cleanup started with commit 815f0ddb346c
("include/linux/compiler*.h: make compiler-*.h mutually exclusive"),
which is supported on gcc >= 5, clang >= 2.9 and icc >= 17.
In the meantime, to support 4.6 <= gcc < 5, we implement
__has_attribute by hand.
All the attributes that can be unconditionally defined and directly
map to compiler attribute(s) (even if optional) have been moved
to a new file include/linux/compiler_attributes.h
In an effort to make the file as regular as possible, comments
stating the purpose of attributes have been removed. Instead,
links to the compiler docs have been added (i.e. to gcc and,
if available, to clang as well). In addition, they have been sorted.
Finally, if an attribute is optional (i.e. if it is guarded
by __has_attribute), the reason has been stated for future reference.
Cc: Rasmus Villemoes <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Reviewed-by: Luc Van Oostenryck <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-clang.h | 4 -
include/linux/compiler-gcc.h | 51 ------
include/linux/compiler-intel.h | 6 -
include/linux/compiler_attributes.h | 244 ++++++++++++++++++++++++++++
include/linux/compiler_types.h | 74 ++-------
5 files changed, 254 insertions(+), 125 deletions(-)
create mode 100644 include/linux/compiler_attributes.h
diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h
index fa9532f8d885..3e7dafb3ea80 100644
--- a/include/linux/compiler-clang.h
+++ b/include/linux/compiler-clang.h
@@ -21,8 +21,6 @@
#define __SANITIZE_ADDRESS__
#endif
-#define __no_sanitize_address __attribute__((__no_sanitize__("address")))
-
/*
* Not all versions of clang implement the the type-generic versions
* of the builtin overflow checkers. Fortunately, clang implements
@@ -41,5 +39,3 @@
* compilers, like ICC.
*/
#define barrier() __asm__ __volatile__("" : : : "memory")
-#define __assume_aligned(a, ...) \
- __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 1ca6a51cfaa9..cfac027e1625 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -108,9 +108,6 @@
__builtin_unreachable(); \
} while (0)
-/* Mark a function definition as prohibited from being cloned. */
-#define __noclone __attribute__((__noclone__, __optimize__("no-tracer")))
-
#if defined(RANDSTRUCT_PLUGIN) && !defined(__CHECKER__)
#define __randomize_layout __attribute__((randomize_layout))
#define __no_randomize_layout __attribute__((no_randomize_layout))
@@ -119,32 +116,6 @@
#define randomized_struct_fields_end } __randomize_layout;
#endif
-/*
- * When used with Link Time Optimization, gcc can optimize away C functions or
- * variables which are referenced only from assembly code. __visible tells the
- * optimizer that something else uses this function or variable, thus preventing
- * this.
- */
-#define __visible __attribute__((__externally_visible__))
-
-/* gcc version specific checks */
-
-#if GCC_VERSION >= 40900
-/*
- * __assume_aligned(n, k): Tell the optimizer that the returned
- * pointer can be assumed to be k modulo n. The second argument is
- * optional (default 0), so we use a variadic macro to make the
- * shorthand.
- *
- * Beware: Do not apply this to functions which may return
- * ERR_PTRs. Also, it is probably unwise to apply it to functions
- * returning extra information in the low bits (but in that case the
- * compiler should see some alignment anyway, when the return value is
- * massaged by 'flags = ptr & 3; ptr &= ~3;').
- */
-#define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
-#endif
-
/*
* GCC 'asm goto' miscompiles certain code sequences:
*
@@ -176,32 +147,10 @@
#define KASAN_ABI_VERSION 3
#endif
-#if GCC_VERSION >= 40902
-/*
- * Tell the compiler that address safety instrumentation (KASAN)
- * should not be applied to that function.
- * Conflicts with inlining: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67368
- */
-#define __no_sanitize_address __attribute__((__no_sanitize_address__))
-#endif
-
#if GCC_VERSION >= 50100
-/*
- * Mark structures as requiring designated initializers.
- * https://gcc.gnu.org/onlinedocs/gcc/Designated-Inits.html
- */
-#define __designated_init __attribute__((__designated_init__))
#define COMPILER_HAS_GENERIC_BUILTIN_OVERFLOW 1
#endif
-#if !defined(__noclone)
-#define __noclone /* not needed */
-#endif
-
-#if !defined(__no_sanitize_address)
-#define __no_sanitize_address
-#endif
-
/*
* Turn individual warnings and errors on and off locally, depending
* on version.
diff --git a/include/linux/compiler-intel.h b/include/linux/compiler-intel.h
index 6004b4588bd4..517bd14e1222 100644
--- a/include/linux/compiler-intel.h
+++ b/include/linux/compiler-intel.h
@@ -34,9 +34,3 @@
/* icc has this, but it's called _bswap16 */
#define __HAVE_BUILTIN_BSWAP16__
#define __builtin_bswap16 _bswap16
-
-/* The following are for compatibility with GCC, from compiler-gcc.h,
- * and may be redefined here because they should not be shared with other
- * compilers, like clang.
- */
-#define __visible __attribute__((__externally_visible__))
diff --git a/include/linux/compiler_attributes.h b/include/linux/compiler_attributes.h
new file mode 100644
index 000000000000..f0f9fc398440
--- /dev/null
+++ b/include/linux/compiler_attributes.h
@@ -0,0 +1,244 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __LINUX_COMPILER_ATTRIBUTES_H
+#define __LINUX_COMPILER_ATTRIBUTES_H
+
+/*
+ * The attributes in this file are unconditionally defined and they directly
+ * map to compiler attribute(s) -- except those that are optional.
+ *
+ * Any other "attributes" (i.e. those that depend on a configuration option,
+ * on a compiler, on an architecture, on plugins, on other attributes...)
+ * should be defined elsewhere (e.g. compiler_types.h or compiler-*.h).
+ *
+ * This file is meant to be sorted (by actual attribute name,
+ * not by #define identifier). Use the __attribute__((__name__)) syntax
+ * (i.e. with underscores) to avoid future collisions with other macros.
+ * If an attribute is optional, state the reason in the comment.
+ */
+
+/*
+ * To check for optional attributes, we use __has_attribute, which is supported
+ * on gcc >= 5, clang >= 2.9 and icc >= 17. In the meantime, to support
+ * 4.6 <= gcc < 5, we implement __has_attribute by hand.
+ *
+ * sparse does not support __has_attribute (yet) and defines __GNUC_MINOR__
+ * depending on the compiler used to build it; however, these attributes have
+ * no semantic effects for sparse, so it does not matter. Also note that,
+ * in order to avoid sparse's warnings, even the unsupported ones must be
+ * defined to 0.
+ */
+#ifndef __has_attribute
+# define __has_attribute(x) __GCC4_has_attribute_##x
+# define __GCC4_has_attribute___assume_aligned__ (__GNUC_MINOR__ >= 9)
+# define __GCC4_has_attribute___designated_init__ 0
+# define __GCC4_has_attribute___externally_visible__ 1
+# define __GCC4_has_attribute___noclone__ 1
+# define __GCC4_has_attribute___optimize__ 1
+# define __GCC4_has_attribute___no_sanitize_address__ (__GNUC_MINOR__ >= 8)
+#endif
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-alias-function-attribute
+ */
+#define __alias(symbol) __attribute__((__alias__(#symbol)))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-aligned-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-aligned-type-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-aligned-variable-attribute
+ */
+#define __aligned(x) __attribute__((__aligned__(x)))
+#define __aligned_largest __attribute__((__aligned__))
+
+/*
+ * Note: users of __always_inline currently do not write "inline" themselves,
+ * which seems to be required by gcc to apply the attribute according
+ * to its docs (and also "warning: always_inline function might not be
+ * inlinable [-Wattributes]" is emitted).
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-always_005finline-function-attribute
+ * clang: mentioned
+ */
+#define __always_inline inline __attribute__((__always_inline__))
+
+/*
+ * The second argument is optional (default 0), so we use a variadic macro
+ * to make the shorthand.
+ *
+ * Beware: Do not apply this to functions which may return
+ * ERR_PTRs. Also, it is probably unwise to apply it to functions
+ * returning extra information in the low bits (but in that case the
+ * compiler should see some alignment anyway, when the return value is
+ * massaged by 'flags = ptr & 3; ptr &= ~3;').
+ *
+ * Optional: only supported since gcc >= 4.9
+ * Optional: not supported by icc
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-assume_005faligned-function-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#assume-aligned
+ */
+#if __has_attribute(__assume_aligned__)
+# define __assume_aligned(a, ...) __attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
+#else
+# define __assume_aligned(a, ...)
+#endif
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-cold-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Label-Attributes.html#index-cold-label-attribute
+ */
+#define __cold __attribute__((__cold__))
+
+/*
+ * Note the long name.
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-const-function-attribute
+ */
+#define __attribute_const__ __attribute__((__const__))
+
+/*
+ * Don't. Just don't. See commit 771c035372a0 ("deprecate the '__deprecated'
+ * attribute warnings entirely and for good") for more information.
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-deprecated-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-deprecated-type-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-deprecated-variable-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Enumerator-Attributes.html#index-deprecated-enumerator-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#deprecated
+ */
+#define __deprecated
+
+/*
+ * Optional: only supported since gcc >= 5.1
+ * Optional: not supported by clang
+ * Optional: not supported by icc
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-designated_005finit-type-attribute
+ */
+#if __has_attribute(__designated_init__)
+# define __designated_init __attribute__((__designated_init__))
+#else
+# define __designated_init
+#endif
+
+/*
+ * Optional: not supported by clang
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-externally_005fvisible-function-attribute
+ */
+#if __has_attribute(__externally_visible__)
+# define __visible __attribute__((__externally_visible__))
+#else
+# define __visible
+#endif
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-format-function-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#format
+ */
+#define __printf(a, b) __attribute__((__format__(printf, a, b)))
+#define __scanf(a, b) __attribute__((__format__(scanf, a, b)))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-gnu_005finline-function-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#gnu-inline
+ */
+#define __gnu_inline __attribute__((__gnu_inline__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-malloc-function-attribute
+ */
+#define __malloc __attribute__((__malloc__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-mode-type-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-mode-variable-attribute
+ */
+#define __mode(x) __attribute__((__mode__(x)))
+
+/*
+ * Optional: not supported by clang
+ * Note: icc does not recognize gcc's no-tracer
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-noclone-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-optimize-function-attribute
+ */
+#if __has_attribute(__noclone__)
+# if __has_attribute(__optimize__)
+# define __noclone __attribute__((__noclone__, __optimize__("no-tracer")))
+# else
+# define __noclone __attribute__((__noclone__))
+# endif
+#else
+# define __noclone
+#endif
+
+/*
+ * Note the missing underscores.
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-noinline-function-attribute
+ * clang: mentioned
+ */
+#define noinline __attribute__((__noinline__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-noreturn-function-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#noreturn
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#id1
+ */
+#define __noreturn __attribute__((__noreturn__))
+
+/*
+ * Optional: only supported since gcc >= 4.8
+ * Optional: not supported by icc
+ *
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-no_005fsanitize_005faddress-function-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#no-sanitize-address-no-address-safety-analysis
+ */
+#if __has_attribute(__no_sanitize_address__)
+# define __no_sanitize_address __attribute__((__no_sanitize_address__))
+#else
+# define __no_sanitize_address
+#endif
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-packed-type-attribute
+ * clang: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-packed-variable-attribute
+ */
+#define __packed __attribute__((__packed__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-pure-function-attribute
+ */
+#define __pure __attribute__((__pure__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-section-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-section-variable-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#section-declspec-allocate
+ */
+#define __section(S) __attribute__((__section__(#S)))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-unused-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Type-Attributes.html#index-unused-type-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-unused-variable-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Label-Attributes.html#index-unused-label-attribute
+ * clang: https://clang.llvm.org/docs/AttributeReference.html#maybe-unused-unused
+ */
+#define __always_unused __attribute__((__unused__))
+#define __maybe_unused __attribute__((__unused__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-used-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-used-variable-attribute
+ */
+#define __used __attribute__((__used__))
+
+/*
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-weak-function-attribute
+ * gcc: https://gcc.gnu.org/onlinedocs/gcc/Common-Variable-Attributes.html#index-weak-variable-attribute
+ */
+#define __weak __attribute__((__weak__))
+
+#endif /* __LINUX_COMPILER_ATTRIBUTES_H */
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index ed7c0e4a180e..3439d7d0249a 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -55,6 +55,9 @@ extern void __chk_io_ptr(const volatile void __iomem *);
#ifdef __KERNEL__
+/* Attributes */
+#include <linux/compiler_attributes.h>
+
/* Compiler specific macros. */
#ifdef __clang__
#include <linux/compiler-clang.h>
@@ -79,12 +82,6 @@ extern void __chk_io_ptr(const volatile void __iomem *);
#include <asm/compiler.h>
#endif
-/*
- * Generic compiler-independent macros required for kernel
- * build go below this comment. Actual compiler/compiler version
- * specific implementations come from the above header files
- */
-
struct ftrace_branch_data {
const char *func;
const char *file;
@@ -107,9 +104,6 @@ struct ftrace_likely_data {
unsigned long constant;
};
-/* Don't. Just don't. */
-#define __deprecated
-
#endif /* __KERNEL__ */
#endif /* __ASSEMBLY__ */
@@ -119,10 +113,6 @@ struct ftrace_likely_data {
* compilers. We don't consider that to be an error, so set them to nothing.
* For example, some of them are for compiler specific plugins.
*/
-#ifndef __designated_init
-# define __designated_init
-#endif
-
#ifndef __latent_entropy
# define __latent_entropy
#endif
@@ -140,17 +130,6 @@ struct ftrace_likely_data {
# define randomized_struct_fields_end
#endif
-#ifndef __visible
-#define __visible
-#endif
-
-/*
- * Assume alignment of return value.
- */
-#ifndef __assume_aligned
-#define __assume_aligned(a, ...)
-#endif
-
/* Are two types/vars the same type (ignoring qualifiers)? */
#define __same_type(a, b) __builtin_types_compatible_p(typeof(a), typeof(b))
@@ -159,10 +138,6 @@ struct ftrace_likely_data {
(sizeof(t) == sizeof(char) || sizeof(t) == sizeof(short) || \
sizeof(t) == sizeof(int) || sizeof(t) == sizeof(long))
-#ifndef __noclone
-#define __noclone
-#endif
-
/* Helpers for emitting diagnostics in pragmas. */
#ifndef __diag
#define __diag(string)
@@ -182,37 +157,6 @@ struct ftrace_likely_data {
#define __diag_error(compiler, version, option, comment) \
__diag_ ## compiler(version, error, option)
-/*
- * From the GCC manual:
- *
- * Many functions have no effects except the return value and their
- * return value depends only on the parameters and/or global
- * variables. Such a function can be subject to common subexpression
- * elimination and loop optimization just as an arithmetic operator
- * would be.
- * [...]
- */
-#define __pure __attribute__((__pure__))
-#define __attribute_const__ __attribute__((__const__))
-#define __aligned(x) __attribute__((__aligned__(x)))
-#define __aligned_largest __attribute__((__aligned__))
-#define __printf(a, b) __attribute__((__format__(printf, a, b)))
-#define __scanf(a, b) __attribute__((__format__(scanf, a, b)))
-#define __maybe_unused __attribute__((__unused__))
-#define __always_unused __attribute__((__unused__))
-#define __mode(x) __attribute__((__mode__(x)))
-#define __malloc __attribute__((__malloc__))
-#define __used __attribute__((__used__))
-#define __noreturn __attribute__((__noreturn__))
-#define __packed __attribute__((__packed__))
-#define __weak __attribute__((__weak__))
-#define __alias(symbol) __attribute__((__alias__(#symbol)))
-#define __cold __attribute__((__cold__))
-#define __section(S) __attribute__((__section__(#S)))
-#define __always_inline inline __attribute__((__always_inline__))
-#define __gnu_inline __attribute__((__gnu_inline__))
-
-
#ifdef CONFIG_ENABLE_MUST_CHECK
#define __must_check __attribute__((__warn_unused_result__))
#else
@@ -246,18 +190,20 @@ struct ftrace_likely_data {
* semantics rather than c99. This prevents multiple symbol definition errors
* of extern inline functions at link time.
* A lot of inline functions can cause havoc with function tracing.
+ * Do not use __always_inline here, since currently it expands to inline again
+ * (which would break users of __always_inline).
*/
#if !defined(CONFIG_ARCH_SUPPORTS_OPTIMIZED_INLINING) || \
!defined(CONFIG_OPTIMIZE_INLINING)
-#define inline \
- inline __attribute__((__always_inline__, __unused__)) notrace __gnu_inline
+#define inline inline __attribute__((__always_inline__)) __gnu_inline \
+ __maybe_unused notrace
#else
-#define inline inline __attribute__((__unused__)) notrace __gnu_inline
+#define inline inline __gnu_inline \
+ __maybe_unused notrace
#endif
#define __inline__ inline
-#define __inline inline
-#define noinline __attribute__((__noinline__))
+#define __inline inline
/*
* Rather then using noinline to prevent stack consumption, use
--
2.17.1
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 4030a2940d6b..17ee9165ca51 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -146,7 +146,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
extern typeof(sym) sym; \
static const unsigned long __kentry_##sym \
__used \
- __attribute__((__section__("___kentry" "+" #sym ), used)) \
+ __attribute__((__section__("___kentry" "+" #sym ))) \
= (unsigned long)&sym;
#endif
--
2.17.1
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler_types.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index a3eceb3ad1b3..ed7c0e4a180e 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -1,3 +1,4 @@
+/* SPDX-License-Identifier: GPL-2.0 */
#ifndef __LINUX_COMPILER_TYPES_H
#define __LINUX_COMPILER_TYPES_H
--
2.17.1
Sparse knows about a few more attributes now, so we can remove
the __CHECKER__ conditions from them (which, in turn, allow us
to move some of them later on to compiler_attributes.h).
* assume_aligned: since sparse's commit ffc860b ("sparse:
ignore __assume_aligned__ attribute"), included in 0.5.1
* error: since sparse's commit 0a04210 ("sparse: Add 'error'
to ignored attributes"), included in 0.5.0
* hotpatch: since sparse's commit 6043210 ("sparse/parse.c:
ignore hotpatch attribute"), included in 0.5.1
* warning: since sparse's commit 977365d ("Avoid "attribute
'warning': unknown attribute" warning"), included in 0.4.2
On top of that, __must_be_array does not need it either because:
* Even ancient versions of sparse do not have a problem
* BUILD_BUG_ON_ZERO() is currently disabled for __CHECKER__
Cc: Rasmus Villemoes <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Acked-by: Luc Van Oostenryck <[email protected]>
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-gcc.h | 6 ++----
include/linux/compiler.h | 4 ----
include/linux/compiler_types.h | 2 +-
3 files changed, 3 insertions(+), 9 deletions(-)
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 3b32bbfa5a49..1ca6a51cfaa9 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -76,14 +76,12 @@
#define __compiletime_object_size(obj) __builtin_object_size(obj, 0)
-#ifndef __CHECKER__
#define __compiletime_warning(message) __attribute__((__warning__(message)))
#define __compiletime_error(message) __attribute__((__error__(message)))
-#ifdef LATENT_ENTROPY_PLUGIN
+#if defined(LATENT_ENTROPY_PLUGIN) && !defined(__CHECKER__)
#define __latent_entropy __attribute__((latent_entropy))
#endif
-#endif /* __CHECKER__ */
/*
* calling noreturn functions, __builtin_unreachable() and __builtin_trap()
@@ -131,7 +129,7 @@
/* gcc version specific checks */
-#if GCC_VERSION >= 40900 && !defined(__CHECKER__)
+#if GCC_VERSION >= 40900
/*
* __assume_aligned(n, k): Tell the optimizer that the returned
* pointer can be assumed to be k modulo n. The second argument is
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 165b1d5683ed..4030a2940d6b 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -357,11 +357,7 @@ static inline void *offset_to_ptr(const int *off)
compiletime_assert(__native_word(t), \
"Need native word sized stores/loads for atomicity.")
-#ifdef __CHECKER__
-#define __must_be_array(a) 0
-#else
/* &a[0] degrades to a pointer: a different type from an array */
#define __must_be_array(a) BUILD_BUG_ON_ZERO(__same_type((a), &(a)[0]))
-#endif
#endif /* __LINUX_COMPILER_H */
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 5ff9cda893f4..a3eceb3ad1b3 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -218,7 +218,7 @@ struct ftrace_likely_data {
#define __must_check
#endif
-#if defined(CC_USING_HOTPATCH) && !defined(__CHECKER__)
+#if defined(CC_USING_HOTPATCH)
#define notrace __attribute__((hotpatch(0, 0)))
#else
#define notrace __attribute__((__no_instrument_function__))
--
2.17.1
Commit 9c695203a7dd ("compiler-gcc.h: gcc-4.5 needs noclone
and noinline on __naked functions") added noinline and noclone
as a workaround for a gcc 4.5 bug, which was resolved in 4.6.0.
Since now the minimum gcc supported version is 4.6,
we can clean it up.
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=44290
and https://godbolt.org/z/h6NMIL
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-gcc.h | 8 +-------
1 file changed, 1 insertion(+), 7 deletions(-)
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 76f4907ef707..4cd5e9264bce 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -77,14 +77,8 @@
* to trace naked functions because then mcount is called without
* stack and frame pointer being set up and there is no chance to
* restore the lr register to the value before mcount was called.
- *
- * The asm() bodies of naked functions often depend on standard calling
- * conventions, therefore they must be noinline and noclone.
- *
- * GCC 4.[56] currently fail to enforce this, so we must do so ourselves.
- * See GCC PR44290.
*/
-#define __naked __attribute__((__naked__)) noinline __noclone notrace
+#define __naked __attribute__((__naked__)) notrace
#define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
--
2.17.1
Attributes const and always_inline have tests around them
which are unneeded, since they are supported by gcc >= 4.6,
clang >= 3 and icc >= 13. https://godbolt.org/z/DFPq37
In the case of gnu_inline, we do not need to test for
__GNUC_STDC_INLINE__ because, regardless of the current
inlining behavior, we can simply always force the old
GCC inlining behavior by using the attribute in all cases.
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler_types.h | 23 +++--------------------
1 file changed, 3 insertions(+), 20 deletions(-)
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 2bc0f94df38e..83475515bc39 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -158,10 +158,6 @@ struct ftrace_likely_data {
(sizeof(t) == sizeof(char) || sizeof(t) == sizeof(short) || \
sizeof(t) == sizeof(int) || sizeof(t) == sizeof(long))
-#ifndef __attribute_const__
-#define __attribute_const__ __attribute__((__const__))
-#endif
-
#ifndef __noclone
#define __noclone
#endif
@@ -196,6 +192,7 @@ struct ftrace_likely_data {
* [...]
*/
#define __pure __attribute__((__pure__))
+#define __attribute_const__ __attribute__((__const__))
#define __aligned(x) __attribute__((__aligned__(x)))
#define __aligned_largest __attribute__((__aligned__))
#define __printf(a, b) __attribute__((__format__(printf, a, b)))
@@ -211,6 +208,8 @@ struct ftrace_likely_data {
#define __alias(symbol) __attribute__((__alias__(#symbol)))
#define __cold __attribute__((__cold__))
#define __section(S) __attribute__((__section__(#S)))
+#define __always_inline inline __attribute__((__always_inline__))
+#define __gnu_inline __attribute__((__gnu_inline__))
#ifdef CONFIG_ENABLE_MUST_CHECK
@@ -227,18 +226,6 @@ struct ftrace_likely_data {
#define __compiler_offsetof(a, b) __builtin_offsetof(a, b)
-/*
- * Feature detection for gnu_inline (gnu89 extern inline semantics). Either
- * __GNUC_STDC_INLINE__ is defined (not using gnu89 extern inline semantics,
- * and we opt in to the gnu89 semantics), or __GNUC_STDC_INLINE__ is not
- * defined so the gnu89 semantics are the default.
- */
-#ifdef __GNUC_STDC_INLINE__
-# define __gnu_inline __attribute__((__gnu_inline__))
-#else
-# define __gnu_inline
-#endif
-
/*
* Force always-inline if the user requests it so via the .config.
* GCC does not warn about unused static inline functions for
@@ -263,10 +250,6 @@ struct ftrace_likely_data {
#define __inline inline
#define noinline __attribute__((__noinline__))
-#ifndef __always_inline
-#define __always_inline inline __attribute__((__always_inline__))
-#endif
-
/*
* Rather then using noinline to prevent stack consumption, use
* noinline_for_stack instead. For documentation reasons.
--
2.17.1
Different definitions of __must_be_array:
* gcc: disabled for __CHECKER__
* clang: same definition as gcc's, but without __CHECKER__
* intel: the comment claims __builtin_types_compatible_p()
is unsupported; but icc seems to support it since 13.0.1
(released in 2012). See https://godbolt.org/z/S0l6QQ
Therefore, we can remove all of them and have a single definition
in compiler.h
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-clang.h | 1 -
include/linux/compiler-gcc.h | 7 -------
include/linux/compiler-intel.h | 3 ---
include/linux/compiler.h | 7 +++++++
4 files changed, 7 insertions(+), 11 deletions(-)
diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h
index d11cad61ba5c..fa9532f8d885 100644
--- a/include/linux/compiler-clang.h
+++ b/include/linux/compiler-clang.h
@@ -41,6 +41,5 @@
* compilers, like ICC.
*/
#define barrier() __asm__ __volatile__("" : : : "memory")
-#define __must_be_array(a) BUILD_BUG_ON_ZERO(__same_type((a), &(a)[0]))
#define __assume_aligned(a, ...) \
__attribute__((__assume_aligned__(a, ## __VA_ARGS__)))
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 371b6fa2d074..76f4907ef707 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -68,13 +68,6 @@
*/
#define uninitialized_var(x) x = x
-#ifdef __CHECKER__
-#define __must_be_array(a) 0
-#else
-/* &a[0] degrades to a pointer: a different type from an array */
-#define __must_be_array(a) BUILD_BUG_ON_ZERO(__same_type((a), &(a)[0]))
-#endif
-
#ifdef RETPOLINE
#define __noretpoline __attribute__((__indirect_branch__("keep")))
#endif
diff --git a/include/linux/compiler-intel.h b/include/linux/compiler-intel.h
index fef8bb3e53ef..6004b4588bd4 100644
--- a/include/linux/compiler-intel.h
+++ b/include/linux/compiler-intel.h
@@ -29,9 +29,6 @@
*/
#define OPTIMIZER_HIDE_VAR(var) barrier()
-/* Intel ECC compiler doesn't support __builtin_types_compatible_p() */
-#define __must_be_array(a) 0
-
#endif
/* icc has this, but it's called _bswap16 */
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index ec4a28bad2c6..165b1d5683ed 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -357,4 +357,11 @@ static inline void *offset_to_ptr(const int *off)
compiletime_assert(__native_word(t), \
"Need native word sized stores/loads for atomicity.")
+#ifdef __CHECKER__
+#define __must_be_array(a) 0
+#else
+/* &a[0] degrades to a pointer: a different type from an array */
+#define __must_be_array(a) BUILD_BUG_ON_ZERO(__same_type((a), &(a)[0]))
+#endif
+
#endif /* __LINUX_COMPILER_H */
--
2.17.1
The attribute syntax optionally allows to surround attribute names
with "__" in order to avoid collisions with macros of the same name
(see https://gcc.gnu.org/onlinedocs/gcc/Attribute-Syntax.html).
This homogenizes all attributes to use the syntax with underscores.
While there are currently only a handful of cases of some TUs defining
macros like "error" which may collide with the attributes,
this should prevent futures surprises.
This has been done only for "standard" attributes supported by
the major compilers. In other words, those of third-party tools
(e.g. sparse, plugins...) have not been changed for the moment.
Cc: Rasmus Villemoes <[email protected]>
Cc: Luc Van Oostenryck <[email protected]>
Cc: Eli Friedman <[email protected]>
Cc: Christopher Li <[email protected]>
Cc: Kees Cook <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Geert Uytterhoeven <[email protected]>
Cc: Arnd Bergmann <[email protected]>
Cc: Greg Kroah-Hartman <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Joe Perches <[email protected]>
Cc: Dominique Martinet <[email protected]>
Cc: Linus Torvalds <[email protected]>
Cc: [email protected]
Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Miguel Ojeda <[email protected]>
---
include/linux/compiler-clang.h | 2 +-
include/linux/compiler-gcc.h | 14 ++++++------
include/linux/compiler-intel.h | 2 +-
include/linux/compiler.h | 8 +++----
include/linux/compiler_types.h | 40 +++++++++++++++++-----------------
5 files changed, 33 insertions(+), 33 deletions(-)
diff --git a/include/linux/compiler-clang.h b/include/linux/compiler-clang.h
index b1ce500fe8b3..d11cad61ba5c 100644
--- a/include/linux/compiler-clang.h
+++ b/include/linux/compiler-clang.h
@@ -21,7 +21,7 @@
#define __SANITIZE_ADDRESS__
#endif
-#define __no_sanitize_address __attribute__((no_sanitize("address")))
+#define __no_sanitize_address __attribute__((__no_sanitize__("address")))
/*
* Not all versions of clang implement the the type-generic versions
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 0a2d06677d83..371b6fa2d074 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -76,7 +76,7 @@
#endif
#ifdef RETPOLINE
-#define __noretpoline __attribute__((indirect_branch("keep")))
+#define __noretpoline __attribute__((__indirect_branch__("keep")))
#endif
/*
@@ -91,15 +91,15 @@
* GCC 4.[56] currently fail to enforce this, so we must do so ourselves.
* See GCC PR44290.
*/
-#define __naked __attribute__((naked)) noinline __noclone notrace
+#define __naked __attribute__((__naked__)) noinline __noclone notrace
#define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
#define __compiletime_object_size(obj) __builtin_object_size(obj, 0)
#ifndef __CHECKER__
-#define __compiletime_warning(message) __attribute__((warning(message)))
-#define __compiletime_error(message) __attribute__((error(message)))
+#define __compiletime_warning(message) __attribute__((__warning__(message)))
+#define __compiletime_error(message) __attribute__((__error__(message)))
#ifdef LATENT_ENTROPY_PLUGIN
#define __latent_entropy __attribute__((latent_entropy))
@@ -148,7 +148,7 @@
* optimizer that something else uses this function or variable, thus preventing
* this.
*/
-#define __visible __attribute__((externally_visible))
+#define __visible __attribute__((__externally_visible__))
/* gcc version specific checks */
@@ -205,7 +205,7 @@
* should not be applied to that function.
* Conflicts with inlining: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67368
*/
-#define __no_sanitize_address __attribute__((no_sanitize_address))
+#define __no_sanitize_address __attribute__((__no_sanitize_address__))
#endif
#if GCC_VERSION >= 50100
@@ -213,7 +213,7 @@
* Mark structures as requiring designated initializers.
* https://gcc.gnu.org/onlinedocs/gcc/Designated-Inits.html
*/
-#define __designated_init __attribute__((designated_init))
+#define __designated_init __attribute__((__designated_init__))
#define COMPILER_HAS_GENERIC_BUILTIN_OVERFLOW 1
#endif
diff --git a/include/linux/compiler-intel.h b/include/linux/compiler-intel.h
index 4c7f9befa9f6..fef8bb3e53ef 100644
--- a/include/linux/compiler-intel.h
+++ b/include/linux/compiler-intel.h
@@ -42,4 +42,4 @@
* and may be redefined here because they should not be shared with other
* compilers, like clang.
*/
-#define __visible __attribute__((externally_visible))
+#define __visible __attribute__((__externally_visible__))
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 7c0157d50964..ec4a28bad2c6 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -24,7 +24,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
long ______r; \
static struct ftrace_likely_data \
__attribute__((__aligned__(4))) \
- __attribute__((section("_ftrace_annotated_branch"))) \
+ __attribute__((__section__("_ftrace_annotated_branch"))) \
______f = { \
.data.func = __func__, \
.data.file = __FILE__, \
@@ -60,7 +60,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
int ______r; \
static struct ftrace_branch_data \
__attribute__((__aligned__(4))) \
- __attribute__((section("_ftrace_branch"))) \
+ __attribute__((__section__("_ftrace_branch"))) \
______f = { \
.func = __func__, \
.file = __FILE__, \
@@ -146,7 +146,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
extern typeof(sym) sym; \
static const unsigned long __kentry_##sym \
__used \
- __attribute__((section("___kentry" "+" #sym ), used)) \
+ __attribute__((__section__("___kentry" "+" #sym ), used)) \
= (unsigned long)&sym;
#endif
@@ -287,7 +287,7 @@ unsigned long read_word_at_a_time(const void *addr)
* visible to the compiler.
*/
#define __ADDRESSABLE(sym) \
- static void * __attribute__((section(".discard.addressable"), used)) \
+ static void * __attribute__((__section__(".discard.addressable"), used)) \
__PASTE(__addressable_##sym, __LINE__) = (void *)&sym;
/**
diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index b6534292ea33..2bc0f94df38e 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h
@@ -195,26 +195,26 @@ struct ftrace_likely_data {
* would be.
* [...]
*/
-#define __pure __attribute__((pure))
-#define __aligned(x) __attribute__((aligned(x)))
-#define __aligned_largest __attribute__((aligned))
-#define __printf(a, b) __attribute__((format(printf, a, b)))
-#define __scanf(a, b) __attribute__((format(scanf, a, b)))
-#define __maybe_unused __attribute__((unused))
-#define __always_unused __attribute__((unused))
-#define __mode(x) __attribute__((mode(x)))
+#define __pure __attribute__((__pure__))
+#define __aligned(x) __attribute__((__aligned__(x)))
+#define __aligned_largest __attribute__((__aligned__))
+#define __printf(a, b) __attribute__((__format__(printf, a, b)))
+#define __scanf(a, b) __attribute__((__format__(scanf, a, b)))
+#define __maybe_unused __attribute__((__unused__))
+#define __always_unused __attribute__((__unused__))
+#define __mode(x) __attribute__((__mode__(x)))
#define __malloc __attribute__((__malloc__))
#define __used __attribute__((__used__))
-#define __noreturn __attribute__((noreturn))
-#define __packed __attribute__((packed))
-#define __weak __attribute__((weak))
-#define __alias(symbol) __attribute__((alias(#symbol)))
-#define __cold __attribute__((cold))
+#define __noreturn __attribute__((__noreturn__))
+#define __packed __attribute__((__packed__))
+#define __weak __attribute__((__weak__))
+#define __alias(symbol) __attribute__((__alias__(#symbol)))
+#define __cold __attribute__((__cold__))
#define __section(S) __attribute__((__section__(#S)))
#ifdef CONFIG_ENABLE_MUST_CHECK
-#define __must_check __attribute__((warn_unused_result))
+#define __must_check __attribute__((__warn_unused_result__))
#else
#define __must_check
#endif
@@ -222,7 +222,7 @@ struct ftrace_likely_data {
#if defined(CC_USING_HOTPATCH) && !defined(__CHECKER__)
#define notrace __attribute__((hotpatch(0, 0)))
#else
-#define notrace __attribute__((no_instrument_function))
+#define notrace __attribute__((__no_instrument_function__))
#endif
#define __compiler_offsetof(a, b) __builtin_offsetof(a, b)
@@ -234,7 +234,7 @@ struct ftrace_likely_data {
* defined so the gnu89 semantics are the default.
*/
#ifdef __GNUC_STDC_INLINE__
-# define __gnu_inline __attribute__((gnu_inline))
+# define __gnu_inline __attribute__((__gnu_inline__))
#else
# define __gnu_inline
#endif
@@ -254,17 +254,17 @@ struct ftrace_likely_data {
#if !defined(CONFIG_ARCH_SUPPORTS_OPTIMIZED_INLINING) || \
!defined(CONFIG_OPTIMIZE_INLINING)
#define inline \
- inline __attribute__((always_inline, unused)) notrace __gnu_inline
+ inline __attribute__((__always_inline__, __unused__)) notrace __gnu_inline
#else
-#define inline inline __attribute__((unused)) notrace __gnu_inline
+#define inline inline __attribute__((__unused__)) notrace __gnu_inline
#endif
#define __inline__ inline
#define __inline inline
-#define noinline __attribute__((noinline))
+#define noinline __attribute__((__noinline__))
#ifndef __always_inline
-#define __always_inline inline __attribute__((always_inline))
+#define __always_inline inline __attribute__((__always_inline__))
#endif
/*
--
2.17.1
On Sat, Sep 08, 2018 at 11:24:46PM +0200, Miguel Ojeda wrote:
> The Compiler Attributes series is an effort to disentangle
> the include/linux/compiler*.h headers and bring them up to date.
>
> The main idea behind the series is to use feature checking macros
> (i.e. __has_attribute) instead of compiler version checks (e.g. GCC_VERSION),
> which are compiler-agnostic (so they can be shared, reducing the size
> of compiler-specific headers) and version-agnostic.
>
> Other related improvements have been performed in the headers as well,
> which on top of the use of __has_attribute it has amounted to a significant
> simplification of these headers (e.g. GCC_VERSION is now only guarding 4
> non-attribute macros).
>
> This series should also help the efforts to support compiling the kernel
> with clang and icc. A fair amount of documentation and comments have also
> been added, clarified or removed; and the headers are now more readable,
> which should help kernel developers in general.
>
> The series was triggered due to the move to gcc >= 4.6. In turn, this series
> has also triggered Sparse to gain the ability to recognize __has_attribute
> on its own.
>
> You can also fetch it from:
>
> https://github.com/ojeda/linux/tree/compiler-attributes-v4
>
> Enjoy!
It's a nice and welcomed cleanup!
Everything look good to me, so feel free to add my
Reviewed-by: Luc Van Oostenryck <[email protected]>
Cheers,
-- Luc
Hi Luc,
On Sun, Sep 9, 2018 at 10:02 AM, Luc Van Oostenryck
<[email protected]> wrote:
> On Sat, Sep 08, 2018 at 11:24:46PM +0200, Miguel Ojeda wrote:
>> The Compiler Attributes series is an effort to disentangle
>> the include/linux/compiler*.h headers and bring them up to date.
>>
>> The main idea behind the series is to use feature checking macros
>> (i.e. __has_attribute) instead of compiler version checks (e.g. GCC_VERSION),
>> which are compiler-agnostic (so they can be shared, reducing the size
>> of compiler-specific headers) and version-agnostic.
>>
>> Other related improvements have been performed in the headers as well,
>> which on top of the use of __has_attribute it has amounted to a significant
>> simplification of these headers (e.g. GCC_VERSION is now only guarding 4
>> non-attribute macros).
>>
>> This series should also help the efforts to support compiling the kernel
>> with clang and icc. A fair amount of documentation and comments have also
>> been added, clarified or removed; and the headers are now more readable,
>> which should help kernel developers in general.
>>
>> The series was triggered due to the move to gcc >= 4.6. In turn, this series
>> has also triggered Sparse to gain the ability to recognize __has_attribute
>> on its own.
>>
>> You can also fetch it from:
>>
>> https://github.com/ojeda/linux/tree/compiler-attributes-v4
>>
>> Enjoy!
>
> It's a nice and welcomed cleanup!
> Everything look good to me, so feel free to add my
> Reviewed-by: Luc Van Oostenryck <[email protected]>
Thanks a lot for the review! :-)
Cheers,
Miguel
Hi all,
On Sat, Sep 8, 2018 at 11:24 PM, Miguel Ojeda
<[email protected]> wrote:
>
> You can also fetch it from:
>
> https://github.com/ojeda/linux/tree/compiler-attributes-v4
In case it helps, I have also rebased it onto next-20180907 at:
https://github.com/ojeda/linux/tree/compiler-attributes-for-next
The only conflict was the removal of __aligned_largest in f3569fd613f6
("crypto: shash - Remove VLA usage in unaligned hashing").
(Luc's Reviewed-by is not included in that version)
Cheers,
Miguel
On Sat, 8 Sep 2018 23:24:58 +0200
Miguel Ojeda <[email protected]> wrote:
> Documentation/process/index.rst | 1 +
> .../process/programming-language.rst | 45 +++++++++++++++++++
> 2 files changed, 46 insertions(+)
> create mode 100644 Documentation/process/programming-language.rst
>
So I have some overall thoughts on the documentation; my apologies for
not getting to this until you got to v4...
1) I think the document is mistitled. It's not really about the language
that the kernel used, it's about compiler attributes. So I would make
both the name of the document and it introduction reflect that.
2) This is an ideal opportunity to document what all of those attributes
actually mean. I would guess that is the information many developers
will come here looking for, and they'll go away frustrated. The ideal
thing to do, IMO, would be do say what each attribute means (rather
than just which compilers support it) in a DOC section in the new
compiler_attributes.h header, then use RST directives to pull all that
information into this document.
Thanks,
jon
Hi Jonathan,
On Sun, Sep 9, 2018 at 8:19 PM, Jonathan Corbet <[email protected]> wrote:
> On Sat, 8 Sep 2018 23:24:58 +0200
> Miguel Ojeda <[email protected]> wrote:
>
>> Documentation/process/index.rst | 1 +
>> .../process/programming-language.rst | 45 +++++++++++++++++++
>> 2 files changed, 46 insertions(+)
>> create mode 100644 Documentation/process/programming-language.rst
>>
> So I have some overall thoughts on the documentation; my apologies for
> not getting to this until you got to v4...
No problem at all! :-)
>
> 1) I think the document is mistitled. It's not really about the language
> that the kernel used, it's about compiler attributes. So I would make
> both the name of the document and it introduction reflect that.
The idea here is to create a document explaining the programming
language that the kernel uses (which we hadn't), taking the chance to
describe the attributes (which in a way are extensions to the C
language); in the future expanding it with other (non-attribute)
extensions that we use. Of course, that is not done, but I thought
starting it was a nice idea (meant for people coming from a C
background that do not know the "dialect" used in Linux).
Also other kind of very commonly used macros and functions used
throughout the kernel could be summarized there as a summary for
newcomers (e.g. stuff like `printk`/`pr_*`, `likely`/`unlikely`,
`panic`, `EXPORT_SYMBOL`, `BUG_ON`/`WARN_ON`...), without writing full
documentation there (but pointing to where they are described, either
in the Docs or a source code if not).
What do you think?
>
> 2) This is an ideal opportunity to document what all of those attributes
> actually mean. I would guess that is the information many developers
I thought about that, but the issue is that compilers already describe
them: in the case of GCC, all of them are documented; some are in
clang; icc is not well documented --- see the previous threads about
this), so we would be duplicating that information (and it will become
outdated as compilers are released and improve the documentation).
This is the reason why the series removes most of the copy-pasted
documentation from gcc and instead supplies a link for each attribute
compiler_attributes.h to gcc & clang online docs.
> will come here looking for, and they'll go away frustrated. The ideal
> thing to do, IMO, would be do say what each attribute means (rather
> than just which compilers support it) in a DOC section in the new
> compiler_attributes.h header, then use RST directives to pull all that
> information into this document.
Initially my idea was to provide a table with the name and the links
to each compiler docs; so that it could be easily used. However, when
thinking about it, it seems that most people consulting such file
would come from the actual source code, so they would have to move
from there to the Doc/ file; so I did it the other way around (IIRC
Nick also mentioned his preference for keeping them in the source
code).
Now, the idea of keeping them in the header but pulling them to the
RST can be a good idea but I was unsure how to do it best. I took a
look at how other RST files did it (the pull), but I thought (maybe
wrongly) that I wouldn't be able to create a nice table (i.e. instead
it would be a long list of attributes, which most people will not use
-- and in my idea of a document describing all the extensions, it
would be taking most of the space), so I discarded it. I am not sure
about the future goals for the Docs/, so excuse me if I have the
completely wrong idea, but in a way, in the current state, I see the
kernel docs as articles/chapters/book on high-level concepts about the
kernel, rather than a technical reference (i.e. the source code with a
better formatting).
Thanks a lot for reviewing!
Cheers,
Miguel
On Sat, Sep 8, 2018 at 2:25 PM Miguel Ojeda
<[email protected]> wrote:
>
> The Compiler Attributes series is an effort to disentangle
> the include/linux/compiler*.h headers and bring them up to date.
>
> The main idea behind the series is to use feature checking macros
> (i.e. __has_attribute) instead of compiler version checks (e.g. GCC_VERSION),
> which are compiler-agnostic (so they can be shared, reducing the size
> of compiler-specific headers) and version-agnostic.
>
> Other related improvements have been performed in the headers as well,
> which on top of the use of __has_attribute it has amounted to a significant
> simplification of these headers (e.g. GCC_VERSION is now only guarding 4
> non-attribute macros).
>
> This series should also help the efforts to support compiling the kernel
> with clang and icc. A fair amount of documentation and comments have also
> been added, clarified or removed; and the headers are now more readable,
> which should help kernel developers in general.
>
> The series was triggered due to the move to gcc >= 4.6. In turn, this series
> has also triggered Sparse to gain the ability to recognize __has_attribute
> on its own.
>
> You can also fetch it from:
>
> https://github.com/ojeda/linux/tree/compiler-attributes-v4
>
> Enjoy!
>
> Cheers,
> Miguel
>
> Cc: Jonathan Corbet <[email protected]>
> Cc: Rasmus Villemoes <[email protected]>
> Cc: Luc Van Oostenryck <[email protected]>
> Cc: Eli Friedman <[email protected]>
> Cc: Christopher Li <[email protected]>
> Cc: Kees Cook <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Geert Uytterhoeven <[email protected]>
> Cc: Arnd Bergmann <[email protected]>
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: Masahiro Yamada <[email protected]>
> Cc: Joe Perches <[email protected]>
> Cc: Dominique Martinet <[email protected]>
> Cc: Nick Desaulniers <[email protected]>
> Cc: Linus Torvalds <[email protected]>
> Cc: [email protected]
> Cc: [email protected]
> Signed-off-by: Miguel Ojeda <[email protected]>
>
> v3 -> v4
>
> This time it is an important fix I found while doing randconfigs, plus
> the documentation sentence that Nick suggested and the MAINTAINERS file.
>
> Compile-tested for a while on (x86_64, gcc-7.3) on a few randconfs.
>
> New:
>
> * Add MAINTAINERS entry for the new file (compiler_attributes.h)
> so that we try to maintain that one as clean as possible.
> (if someone else wants to join, feel free!)
>
> Modified:
>
> * Fix inline macro
>
> __always_inline cannot be used in the inline macro,
> because people marking a function as __always_inline would
> expand to inline which in turn would expand back to __always_inline
> (under some configs), which would not be expanded again.
> Added a comment about this in the inline macro.
>
> Also added another comment about __always_inline in compiler_attributes.h
> to note that users expect to not have to write inline (as well as
> the attribute). We cannot simply remove "inline" there (which would
> solve the problem described above) because using the attribute is
> not enough for gcc. A function marked as always_inline seems to require
> to be marked as inline itself so that the attribute is applied:
>
> "warning: always_inline function might not be inlinable [-Wattributes]"
>
> From the gcc docs:
>
> "For functions declared inline, this attribute inlines the function
> independent of any restrictions that otherwise apply to inlining."
>
> See https://godbolt.org/z/LpzUPj for an example.
>
> From reviews:
>
> * Add sentence about feature-detection in Docs (Nick)
>
> Miguel Ojeda (13):
> Compiler Attributes: remove unused attributes
> Compiler Attributes: always use the extra-underscores syntax
> Compiler Attributes: remove unneeded tests
> Compiler Attributes: homogenize __must_be_array
> Compiler Attributes: naked was fixed in gcc 4.6
> Compiler Attributes: naked can be shared
> Compiler Attributes: remove unneeded sparse (__CHECKER__) tests
> Compiler Attributes: add missing SPDX ID in compiler_types.h
> Compiler Attributes: use feature checks instead of version checks
> Compiler Attributes: KENTRY used twice the "used" attribute
> Compiler Attributes: remove uses of __attribute__ from compiler.h
> Compiler Attributes: add Doc/process/programming-language.rst
> Compiler Attributes: Add MAINTAINERS entry
>
> Documentation/process/index.rst | 1 +
> .../process/programming-language.rst | 45 ++++
> MAINTAINERS | 5 +
> include/linux/compiler-clang.h | 5 -
> include/linux/compiler-gcc.h | 84 +-----
> include/linux/compiler-intel.h | 9 -
> include/linux/compiler.h | 19 +-
> include/linux/compiler_attributes.h | 244 ++++++++++++++++++
> include/linux/compiler_types.h | 105 ++------
> 9 files changed, 329 insertions(+), 188 deletions(-)
> create mode 100644 Documentation/process/programming-language.rst
> create mode 100644 include/linux/compiler_attributes.h
>
> --
> 2.17.1
>
For the ones I didn't already review:
Reviewed-by: Nick Desaulniers <[email protected]>
--
Thanks,
~Nick Desaulniers
Hi,
On 08.09.2018 14:24, Miguel Ojeda wrote:
> Commit 9c695203a7dd ("compiler-gcc.h: gcc-4.5 needs noclone
> and noinline on __naked functions") added noinline and noclone
> as a workaround for a gcc 4.5 bug, which was resolved in 4.6.0.
>
> Since now the minimum gcc supported version is 4.6,
> we can clean it up.
>
> See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=44290
> and https://godbolt.org/z/h6NMIL
>
> Cc: Rasmus Villemoes <[email protected]>
> Cc: Luc Van Oostenryck <[email protected]>
> Cc: Eli Friedman <[email protected]>
> Cc: Christopher Li <[email protected]>
> Cc: Kees Cook <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Geert Uytterhoeven <[email protected]>
> Cc: Arnd Bergmann <[email protected]>
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: Masahiro Yamada <[email protected]>
> Cc: Joe Perches <[email protected]>
> Cc: Dominique Martinet <[email protected]>
> Cc: Nick Desaulniers <[email protected]>
> Cc: Linus Torvalds <[email protected]>
> Cc: [email protected]
> Signed-off-by: Miguel Ojeda <[email protected]>
> ---
> include/linux/compiler-gcc.h | 8 +-------
> 1 file changed, 1 insertion(+), 7 deletions(-)
>
> diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
> index 76f4907ef707..4cd5e9264bce 100644
> --- a/include/linux/compiler-gcc.h
> +++ b/include/linux/compiler-gcc.h
> @@ -77,14 +77,8 @@
> * to trace naked functions because then mcount is called without
> * stack and frame pointer being set up and there is no chance to
> * restore the lr register to the value before mcount was called.
I think this is also no longer required since GCC 4.6 and newer only
have the new mcount interface:
https://lore.kernel.org/lkml/[email protected]/T/#u
That said, I am not sure whether it is a good idea to enable tracing for
naked functions even with the mcount interface.
--
Stefan
> - *
> - * The asm() bodies of naked functions often depend on standard calling
> - * conventions, therefore they must be noinline and noclone.
> - *
> - * GCC 4.[56] currently fail to enforce this, so we must do so ourselves.
> - * See GCC PR44290.
> */
> -#define __naked __attribute__((__naked__)) noinline __noclone notrace
> +#define __naked __attribute__((__naked__)) notrace
>
> #define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
On 08.09.2018 14:24, Miguel Ojeda wrote:
> The naked attribute is supported by at least gcc >= 4.6 (for ARM,
> which is the only current user), gcc >= 8 (for x86), clang >= 3.1
> and icc >= 13. See https://godbolt.org/z/350Dyc
>
> Therefore, move it out of compiler-gcc.h so that the definition
> is shared by all compilers.
>
This actually fixes Clang support for ARM32, it fixes: 815f0ddb346c
("include/linux/compiler*.h: make compiler-*.h mutually exclusive").
So this should go in v4.19. Is this patch set (patches 5/6 in
particular) planned for v4.19? If not, we probably have to split out
this fix...
--
Stefan
> Cc: Rasmus Villemoes <[email protected]>
> Cc: Luc Van Oostenryck <[email protected]>
> Cc: Eli Friedman <[email protected]>
> Cc: Christopher Li <[email protected]>
> Cc: Kees Cook <[email protected]>
> Cc: Ingo Molnar <[email protected]>
> Cc: Geert Uytterhoeven <[email protected]>
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: Masahiro Yamada <[email protected]>
> Cc: Joe Perches <[email protected]>
> Cc: Dominique Martinet <[email protected]>
> Cc: Nick Desaulniers <[email protected]>
> Cc: Linus Torvalds <[email protected]>
> Cc: [email protected]
> Suggested-by: Arnd Bergmann <[email protected]>
> Signed-off-by: Miguel Ojeda <[email protected]>
> ---
> include/linux/compiler-gcc.h | 8 --------
> include/linux/compiler_types.h | 8 ++++++++
> 2 files changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
> index 4cd5e9264bce..3b32bbfa5a49 100644
> --- a/include/linux/compiler-gcc.h
> +++ b/include/linux/compiler-gcc.h
> @@ -72,14 +72,6 @@
> #define __noretpoline __attribute__((__indirect_branch__("keep")))
> #endif
>
> -/*
> - * it doesn't make sense on ARM (currently the only user of __naked)
> - * to trace naked functions because then mcount is called without
> - * stack and frame pointer being set up and there is no chance to
> - * restore the lr register to the value before mcount was called.
> - */
> -#define __naked __attribute__((__naked__)) notrace
> -
> #define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
>
> #define __compiletime_object_size(obj) __builtin_object_size(obj, 0)
> diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
> index 83475515bc39..5ff9cda893f4 100644
> --- a/include/linux/compiler_types.h
> +++ b/include/linux/compiler_types.h
> @@ -224,6 +224,14 @@ struct ftrace_likely_data {
> #define notrace __attribute__((__no_instrument_function__))
> #endif
>
> +/*
> + * it doesn't make sense on ARM (currently the only user of __naked)
> + * to trace naked functions because then mcount is called without
> + * stack and frame pointer being set up and there is no chance to
> + * restore the lr register to the value before mcount was called.
> + */
> +#define __naked __attribute__((__naked__)) notrace
> +
> #define __compiler_offsetof(a, b) __builtin_offsetof(a, b)
>
> /*