2005-04-17 05:20:36 +07:00
|
|
|
#ifndef __LINUX_COMPILER_H
|
|
|
|
#define __LINUX_COMPILER_H
|
|
|
|
|
|
|
|
#ifndef __ASSEMBLY__
|
|
|
|
|
|
|
|
#ifdef __CHECKER__
|
|
|
|
# define __user __attribute__((noderef, address_space(1)))
|
2009-10-29 20:34:15 +07:00
|
|
|
# define __kernel __attribute__((address_space(0)))
|
2005-04-17 05:20:36 +07:00
|
|
|
# define __safe __attribute__((safe))
|
|
|
|
# define __force __attribute__((force))
|
|
|
|
# define __nocast __attribute__((nocast))
|
|
|
|
# define __iomem __attribute__((noderef, address_space(2)))
|
2012-12-18 07:03:24 +07:00
|
|
|
# define __must_hold(x) __attribute__((context(x,1,1)))
|
2006-10-01 13:28:21 +07:00
|
|
|
# define __acquires(x) __attribute__((context(x,0,1)))
|
|
|
|
# define __releases(x) __attribute__((context(x,1,0)))
|
|
|
|
# define __acquire(x) __context__(x,1)
|
|
|
|
# define __release(x) __context__(x,-1)
|
2006-09-29 16:01:03 +07:00
|
|
|
# define __cond_lock(x,c) ((c) ? ({ __acquire(x); 1; }) : 0)
|
2009-10-29 20:34:15 +07:00
|
|
|
# define __percpu __attribute__((noderef, address_space(3)))
|
2010-04-29 04:39:09 +07:00
|
|
|
#ifdef CONFIG_SPARSE_RCU_POINTER
|
|
|
|
# define __rcu __attribute__((noderef, address_space(4)))
|
sparse: Add __private to privatize members of structs
In C programming language, we don't have a easy way to privatize a
member of a structure. However in kernel, sometimes there is a need to
privatize a member in case of potential bugs or misuses.
Fortunately, the noderef attribute of sparse is a way to privatize a
member, as by defining a member as noderef, the address-of operator on
the member will produce a noderef pointer to that member, and if anyone
wants to dereference that kind of pointers to read or modify the member,
sparse will yell.
Based on this, __private modifier and related operation ACCESS_PRIVATE()
are introduced, which could help detect undesigned public uses of
private members of structs. Here is an example of sparse's output if it
detect an undersigned public use:
| kernel/rcu/tree.c:4453:25: warning: incorrect type in argument 1 (different modifiers)
| kernel/rcu/tree.c:4453:25: expected struct raw_spinlock [usertype] *lock
| kernel/rcu/tree.c:4453:25: got struct raw_spinlock [noderef] *<noident>
Also, this patch improves compiler.h a little bit by adding comments for
"#else" and "#endif".
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
2015-12-29 11:18:46 +07:00
|
|
|
#else /* CONFIG_SPARSE_RCU_POINTER */
|
2010-05-12 06:13:14 +07:00
|
|
|
# define __rcu
|
sparse: Add __private to privatize members of structs
In C programming language, we don't have a easy way to privatize a
member of a structure. However in kernel, sometimes there is a need to
privatize a member in case of potential bugs or misuses.
Fortunately, the noderef attribute of sparse is a way to privatize a
member, as by defining a member as noderef, the address-of operator on
the member will produce a noderef pointer to that member, and if anyone
wants to dereference that kind of pointers to read or modify the member,
sparse will yell.
Based on this, __private modifier and related operation ACCESS_PRIVATE()
are introduced, which could help detect undesigned public uses of
private members of structs. Here is an example of sparse's output if it
detect an undersigned public use:
| kernel/rcu/tree.c:4453:25: warning: incorrect type in argument 1 (different modifiers)
| kernel/rcu/tree.c:4453:25: expected struct raw_spinlock [usertype] *lock
| kernel/rcu/tree.c:4453:25: got struct raw_spinlock [noderef] *<noident>
Also, this patch improves compiler.h a little bit by adding comments for
"#else" and "#endif".
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
2015-12-29 11:18:46 +07:00
|
|
|
#endif /* CONFIG_SPARSE_RCU_POINTER */
|
|
|
|
# define __private __attribute__((noderef))
|
2007-07-26 23:35:29 +07:00
|
|
|
extern void __chk_user_ptr(const volatile void __user *);
|
|
|
|
extern void __chk_io_ptr(const volatile void __iomem *);
|
sparse: Add __private to privatize members of structs
In C programming language, we don't have a easy way to privatize a
member of a structure. However in kernel, sometimes there is a need to
privatize a member in case of potential bugs or misuses.
Fortunately, the noderef attribute of sparse is a way to privatize a
member, as by defining a member as noderef, the address-of operator on
the member will produce a noderef pointer to that member, and if anyone
wants to dereference that kind of pointers to read or modify the member,
sparse will yell.
Based on this, __private modifier and related operation ACCESS_PRIVATE()
are introduced, which could help detect undesigned public uses of
private members of structs. Here is an example of sparse's output if it
detect an undersigned public use:
| kernel/rcu/tree.c:4453:25: warning: incorrect type in argument 1 (different modifiers)
| kernel/rcu/tree.c:4453:25: expected struct raw_spinlock [usertype] *lock
| kernel/rcu/tree.c:4453:25: got struct raw_spinlock [noderef] *<noident>
Also, this patch improves compiler.h a little bit by adding comments for
"#else" and "#endif".
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
2015-12-29 11:18:46 +07:00
|
|
|
# define ACCESS_PRIVATE(p, member) (*((typeof((p)->member) __force *) &(p)->member))
|
|
|
|
#else /* __CHECKER__ */
|
2017-01-14 02:14:39 +07:00
|
|
|
# ifdef STRUCTLEAK_PLUGIN
|
|
|
|
# define __user __attribute__((user))
|
|
|
|
# else
|
|
|
|
# define __user
|
|
|
|
# endif
|
2005-04-17 05:20:36 +07:00
|
|
|
# define __kernel
|
|
|
|
# define __safe
|
|
|
|
# define __force
|
|
|
|
# define __nocast
|
|
|
|
# define __iomem
|
|
|
|
# define __chk_user_ptr(x) (void)0
|
|
|
|
# define __chk_io_ptr(x) (void)0
|
|
|
|
# define __builtin_warning(x, y...) (1)
|
2012-12-18 07:03:24 +07:00
|
|
|
# define __must_hold(x)
|
2005-04-17 05:20:36 +07:00
|
|
|
# define __acquires(x)
|
|
|
|
# define __releases(x)
|
|
|
|
# define __acquire(x) (void)0
|
|
|
|
# define __release(x) (void)0
|
2006-09-29 16:01:03 +07:00
|
|
|
# define __cond_lock(x,c) (c)
|
2009-10-29 20:34:15 +07:00
|
|
|
# define __percpu
|
2010-05-12 06:13:14 +07:00
|
|
|
# define __rcu
|
sparse: Add __private to privatize members of structs
In C programming language, we don't have a easy way to privatize a
member of a structure. However in kernel, sometimes there is a need to
privatize a member in case of potential bugs or misuses.
Fortunately, the noderef attribute of sparse is a way to privatize a
member, as by defining a member as noderef, the address-of operator on
the member will produce a noderef pointer to that member, and if anyone
wants to dereference that kind of pointers to read or modify the member,
sparse will yell.
Based on this, __private modifier and related operation ACCESS_PRIVATE()
are introduced, which could help detect undesigned public uses of
private members of structs. Here is an example of sparse's output if it
detect an undersigned public use:
| kernel/rcu/tree.c:4453:25: warning: incorrect type in argument 1 (different modifiers)
| kernel/rcu/tree.c:4453:25: expected struct raw_spinlock [usertype] *lock
| kernel/rcu/tree.c:4453:25: got struct raw_spinlock [noderef] *<noident>
Also, this patch improves compiler.h a little bit by adding comments for
"#else" and "#endif".
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
2015-12-29 11:18:46 +07:00
|
|
|
# define __private
|
|
|
|
# define ACCESS_PRIVATE(p, member) ((p)->member)
|
|
|
|
#endif /* __CHECKER__ */
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2012-11-22 09:00:25 +07:00
|
|
|
/* Indirect macros required for expanded argument pasting, eg. __LINE__. */
|
|
|
|
#define ___PASTE(a,b) a##b
|
|
|
|
#define __PASTE(a,b) ___PASTE(a,b)
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#ifdef __KERNEL__
|
|
|
|
|
2009-01-03 00:23:03 +07:00
|
|
|
#ifdef __GNUC__
|
|
|
|
#include <linux/compiler-gcc.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#endif
|
|
|
|
|
2015-10-07 15:54:36 +07:00
|
|
|
#if defined(CC_USING_HOTPATCH) && !defined(__CHECKER__)
|
2015-01-18 22:45:42 +07:00
|
|
|
#define notrace __attribute__((hotpatch(0,0)))
|
|
|
|
#else
|
2008-08-15 09:47:18 +07:00
|
|
|
#define notrace __attribute__((no_instrument_function))
|
2015-01-18 22:45:42 +07:00
|
|
|
#endif
|
2008-08-15 09:47:18 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/* Intel compiler defines __GNUC__. So we will overwrite implementations
|
|
|
|
* coming from above header files here
|
|
|
|
*/
|
|
|
|
#ifdef __INTEL_COMPILER
|
|
|
|
# include <linux/compiler-intel.h>
|
|
|
|
#endif
|
|
|
|
|
2012-11-21 04:13:10 +07:00
|
|
|
/* Clang compiler defines __GNUC__. So we will overwrite implementations
|
|
|
|
* coming from above header files here
|
|
|
|
*/
|
|
|
|
#ifdef __clang__
|
|
|
|
#include <linux/compiler-clang.h>
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* Generic compiler-dependent macros required for kernel
|
|
|
|
* build go below this comment. Actual compiler/compiler version
|
|
|
|
* specific implementations come from the above header files
|
|
|
|
*/
|
|
|
|
|
2008-11-13 03:24:24 +07:00
|
|
|
struct ftrace_branch_data {
|
2008-11-12 12:14:39 +07:00
|
|
|
const char *func;
|
|
|
|
const char *file;
|
|
|
|
unsigned line;
|
2008-11-21 13:30:54 +07:00
|
|
|
union {
|
|
|
|
struct {
|
|
|
|
unsigned long correct;
|
|
|
|
unsigned long incorrect;
|
|
|
|
};
|
|
|
|
struct {
|
|
|
|
unsigned long miss;
|
|
|
|
unsigned long hit;
|
|
|
|
};
|
2009-03-18 03:15:44 +07:00
|
|
|
unsigned long miss_hit[2];
|
2008-11-21 13:30:54 +07:00
|
|
|
};
|
2008-11-12 12:14:39 +07:00
|
|
|
};
|
2008-11-13 03:24:24 +07:00
|
|
|
|
2017-01-19 20:57:14 +07:00
|
|
|
struct ftrace_likely_data {
|
|
|
|
struct ftrace_branch_data data;
|
|
|
|
unsigned long constant;
|
|
|
|
};
|
|
|
|
|
2008-11-13 03:24:24 +07:00
|
|
|
/*
|
|
|
|
* Note: DISABLE_BRANCH_PROFILING can be used by special lowlevel code
|
|
|
|
* to disable branch tracing on a per file basis.
|
|
|
|
*/
|
2009-04-05 21:20:02 +07:00
|
|
|
#if defined(CONFIG_TRACE_BRANCH_PROFILING) \
|
|
|
|
&& !defined(DISABLE_BRANCH_PROFILING) && !defined(__CHECKER__)
|
2017-01-19 20:57:14 +07:00
|
|
|
void ftrace_likely_update(struct ftrace_likely_data *f, int val,
|
tracing: Process constants for (un)likely() profiler
When running the likely/unlikely profiler, one of the results did not look
accurate. It noted that the unlikely() in link_path_walk() was 100%
incorrect. When I added a trace_printk() to see what was happening there, it
became 80% correct! Looking deeper into what whas happening, I found that
gcc split that if statement into two paths. One where the if statement
became a constant, the other path a variable. The other path had the if
statement always hit (making the unlikely there, always false), but since
the #define unlikely() has:
#define unlikely() (__builtin_constant_p(x) ? !!(x) : __branch_check__(x, 0))
Where constants are ignored by the branch profiler, the "constant" path
made by the compiler was ignored, even though it was hit 80% of the time.
By just passing the constant value to the __branch_check__() function and
tracing it out of line (as always correct, as likely/unlikely isn't a factor
for constants), then we get back the accurate readings of branches that were
optimized by gcc causing part of the execution to become constant.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2017-01-18 00:29:35 +07:00
|
|
|
int expect, int is_constant);
|
2008-11-12 12:14:39 +07:00
|
|
|
|
|
|
|
#define likely_notrace(x) __builtin_expect(!!(x), 1)
|
|
|
|
#define unlikely_notrace(x) __builtin_expect(!!(x), 0)
|
|
|
|
|
tracing: Process constants for (un)likely() profiler
When running the likely/unlikely profiler, one of the results did not look
accurate. It noted that the unlikely() in link_path_walk() was 100%
incorrect. When I added a trace_printk() to see what was happening there, it
became 80% correct! Looking deeper into what whas happening, I found that
gcc split that if statement into two paths. One where the if statement
became a constant, the other path a variable. The other path had the if
statement always hit (making the unlikely there, always false), but since
the #define unlikely() has:
#define unlikely() (__builtin_constant_p(x) ? !!(x) : __branch_check__(x, 0))
Where constants are ignored by the branch profiler, the "constant" path
made by the compiler was ignored, even though it was hit 80% of the time.
By just passing the constant value to the __branch_check__() function and
tracing it out of line (as always correct, as likely/unlikely isn't a factor
for constants), then we get back the accurate readings of branches that were
optimized by gcc causing part of the execution to become constant.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2017-01-18 00:29:35 +07:00
|
|
|
#define __branch_check__(x, expect, is_constant) ({ \
|
2008-11-12 12:14:39 +07:00
|
|
|
int ______r; \
|
2017-01-19 20:57:14 +07:00
|
|
|
static struct ftrace_likely_data \
|
2008-11-12 12:14:39 +07:00
|
|
|
__attribute__((__aligned__(4))) \
|
2008-11-21 12:40:40 +07:00
|
|
|
__attribute__((section("_ftrace_annotated_branch"))) \
|
2008-11-12 12:14:39 +07:00
|
|
|
______f = { \
|
2017-01-19 20:57:14 +07:00
|
|
|
.data.func = __func__, \
|
|
|
|
.data.file = __FILE__, \
|
|
|
|
.data.line = __LINE__, \
|
2008-11-12 12:14:39 +07:00
|
|
|
}; \
|
tracing: Process constants for (un)likely() profiler
When running the likely/unlikely profiler, one of the results did not look
accurate. It noted that the unlikely() in link_path_walk() was 100%
incorrect. When I added a trace_printk() to see what was happening there, it
became 80% correct! Looking deeper into what whas happening, I found that
gcc split that if statement into two paths. One where the if statement
became a constant, the other path a variable. The other path had the if
statement always hit (making the unlikely there, always false), but since
the #define unlikely() has:
#define unlikely() (__builtin_constant_p(x) ? !!(x) : __branch_check__(x, 0))
Where constants are ignored by the branch profiler, the "constant" path
made by the compiler was ignored, even though it was hit 80% of the time.
By just passing the constant value to the __branch_check__() function and
tracing it out of line (as always correct, as likely/unlikely isn't a factor
for constants), then we get back the accurate readings of branches that were
optimized by gcc causing part of the execution to become constant.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2017-01-18 00:29:35 +07:00
|
|
|
______r = __builtin_expect(!!(x), expect); \
|
|
|
|
ftrace_likely_update(&______f, ______r, \
|
|
|
|
expect, is_constant); \
|
2008-11-12 12:14:39 +07:00
|
|
|
______r; \
|
|
|
|
})
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Using __builtin_constant_p(x) to ignore cases where the return
|
|
|
|
* value is always the same. This idea is taken from a similar patch
|
|
|
|
* written by Daniel Walker.
|
|
|
|
*/
|
|
|
|
# ifndef likely
|
tracing: Process constants for (un)likely() profiler
When running the likely/unlikely profiler, one of the results did not look
accurate. It noted that the unlikely() in link_path_walk() was 100%
incorrect. When I added a trace_printk() to see what was happening there, it
became 80% correct! Looking deeper into what whas happening, I found that
gcc split that if statement into two paths. One where the if statement
became a constant, the other path a variable. The other path had the if
statement always hit (making the unlikely there, always false), but since
the #define unlikely() has:
#define unlikely() (__builtin_constant_p(x) ? !!(x) : __branch_check__(x, 0))
Where constants are ignored by the branch profiler, the "constant" path
made by the compiler was ignored, even though it was hit 80% of the time.
By just passing the constant value to the __branch_check__() function and
tracing it out of line (as always correct, as likely/unlikely isn't a factor
for constants), then we get back the accurate readings of branches that were
optimized by gcc causing part of the execution to become constant.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2017-01-18 00:29:35 +07:00
|
|
|
# define likely(x) (__branch_check__(x, 1, __builtin_constant_p(x)))
|
2008-11-12 12:14:39 +07:00
|
|
|
# endif
|
|
|
|
# ifndef unlikely
|
tracing: Process constants for (un)likely() profiler
When running the likely/unlikely profiler, one of the results did not look
accurate. It noted that the unlikely() in link_path_walk() was 100%
incorrect. When I added a trace_printk() to see what was happening there, it
became 80% correct! Looking deeper into what whas happening, I found that
gcc split that if statement into two paths. One where the if statement
became a constant, the other path a variable. The other path had the if
statement always hit (making the unlikely there, always false), but since
the #define unlikely() has:
#define unlikely() (__builtin_constant_p(x) ? !!(x) : __branch_check__(x, 0))
Where constants are ignored by the branch profiler, the "constant" path
made by the compiler was ignored, even though it was hit 80% of the time.
By just passing the constant value to the __branch_check__() function and
tracing it out of line (as always correct, as likely/unlikely isn't a factor
for constants), then we get back the accurate readings of branches that were
optimized by gcc causing part of the execution to become constant.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2017-01-18 00:29:35 +07:00
|
|
|
# define unlikely(x) (__branch_check__(x, 0, __builtin_constant_p(x)))
|
2008-11-12 12:14:39 +07:00
|
|
|
# endif
|
2008-11-21 13:30:54 +07:00
|
|
|
|
|
|
|
#ifdef CONFIG_PROFILE_ALL_BRANCHES
|
|
|
|
/*
|
|
|
|
* "Define 'is'", Bill Clinton
|
|
|
|
* "Define 'if'", Steven Rostedt
|
|
|
|
*/
|
2009-04-07 21:59:41 +07:00
|
|
|
#define if(cond, ...) __trace_if( (cond , ## __VA_ARGS__) )
|
|
|
|
#define __trace_if(cond) \
|
tracing: Fix freak link error caused by branch tracer
In my randconfig tests, I came across a bug that involves several
components:
* gcc-4.9 through at least 5.3
* CONFIG_GCOV_PROFILE_ALL enabling -fprofile-arcs for all files
* CONFIG_PROFILE_ALL_BRANCHES overriding every if()
* The optimized implementation of do_div() that tries to
replace a library call with an division by multiplication
* code in drivers/media/dvb-frontends/zl10353.c doing
u32 adc_clock = 450560; /* 45.056 MHz */
if (state->config.adc_clock)
adc_clock = state->config.adc_clock;
do_div(value, adc_clock);
In this case, gcc fails to determine whether the divisor
in do_div() is __builtin_constant_p(). In particular, it
concludes that __builtin_constant_p(adc_clock) is false, while
__builtin_constant_p(!!adc_clock) is true.
That in turn throws off the logic in do_div() that also uses
__builtin_constant_p(), and instead of picking either the
constant- optimized division, and the code in ilog2() that uses
__builtin_constant_p() to figure out whether it knows the answer at
compile time. The result is a link error from failing to find
multiple symbols that should never have been called based on
the __builtin_constant_p():
dvb-frontends/zl10353.c:138: undefined reference to `____ilog2_NaN'
dvb-frontends/zl10353.c:138: undefined reference to `__aeabi_uldivmod'
ERROR: "____ilog2_NaN" [drivers/media/dvb-frontends/zl10353.ko] undefined!
ERROR: "__aeabi_uldivmod" [drivers/media/dvb-frontends/zl10353.ko] undefined!
This patch avoids the problem by changing __trace_if() to check
whether the condition is known at compile-time to be nonzero, rather
than checking whether it is actually a constant.
I see this one link error in roughly one out of 1600 randconfig builds
on ARM, and the patch fixes all known instances.
Link: http://lkml.kernel.org/r/1455312410-1058841-1-git-send-email-arnd@arndb.de
Acked-by: Nicolas Pitre <nico@linaro.org>
Fixes: ab3c9c686e22 ("branch tracer, intel-iommu: fix build with CONFIG_BRANCH_TRACER=y")
Cc: stable@vger.kernel.org # v2.6.30+
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
2016-02-13 04:26:42 +07:00
|
|
|
if (__builtin_constant_p(!!(cond)) ? !!(cond) : \
|
2008-11-21 13:30:54 +07:00
|
|
|
({ \
|
|
|
|
int ______r; \
|
|
|
|
static struct ftrace_branch_data \
|
|
|
|
__attribute__((__aligned__(4))) \
|
|
|
|
__attribute__((section("_ftrace_branch"))) \
|
|
|
|
______f = { \
|
|
|
|
.func = __func__, \
|
|
|
|
.file = __FILE__, \
|
|
|
|
.line = __LINE__, \
|
|
|
|
}; \
|
|
|
|
______r = !!(cond); \
|
2009-03-18 03:15:44 +07:00
|
|
|
______f.miss_hit[______r]++; \
|
2008-11-21 13:30:54 +07:00
|
|
|
______r; \
|
|
|
|
}))
|
|
|
|
#endif /* CONFIG_PROFILE_ALL_BRANCHES */
|
|
|
|
|
2008-11-12 12:14:39 +07:00
|
|
|
#else
|
|
|
|
# define likely(x) __builtin_expect(!!(x), 1)
|
|
|
|
# define unlikely(x) __builtin_expect(!!(x), 0)
|
|
|
|
#endif
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/* Optimization barrier */
|
|
|
|
#ifndef barrier
|
|
|
|
# define barrier() __memory_barrier()
|
|
|
|
#endif
|
|
|
|
|
lib: make memzero_explicit more robust against dead store elimination
In commit 0b053c951829 ("lib: memzero_explicit: use barrier instead
of OPTIMIZER_HIDE_VAR"), we made memzero_explicit() more robust in
case LTO would decide to inline memzero_explicit() and eventually
find out it could be elimiated as dead store.
While using barrier() works well for the case of gcc, recent efforts
from LLVMLinux people suggest to use llvm as an alternative to gcc,
and there, Stephan found in a simple stand-alone user space example
that llvm could nevertheless optimize and thus elimitate the memset().
A similar issue has been observed in the referenced llvm bug report,
which is regarded as not-a-bug.
Based on some experiments, icc is a bit special on its own, while it
doesn't seem to eliminate the memset(), it could do so with an own
implementation, and then result in similar findings as with llvm.
The fix in this patch now works for all three compilers (also tested
with more aggressive optimization levels). Arguably, in the current
kernel tree it's more of a theoretical issue, but imho, it's better
to be pedantic about it.
It's clearly visible with gcc/llvm though, with the below code: if we
would have used barrier() only here, llvm would have omitted clearing,
not so with barrier_data() variant:
static inline void memzero_explicit(void *s, size_t count)
{
memset(s, 0, count);
barrier_data(s);
}
int main(void)
{
char buff[20];
memzero_explicit(buff, sizeof(buff));
return 0;
}
$ gcc -O2 test.c
$ gdb a.out
(gdb) disassemble main
Dump of assembler code for function main:
0x0000000000400400 <+0>: lea -0x28(%rsp),%rax
0x0000000000400405 <+5>: movq $0x0,-0x28(%rsp)
0x000000000040040e <+14>: movq $0x0,-0x20(%rsp)
0x0000000000400417 <+23>: movl $0x0,-0x18(%rsp)
0x000000000040041f <+31>: xor %eax,%eax
0x0000000000400421 <+33>: retq
End of assembler dump.
$ clang -O2 test.c
$ gdb a.out
(gdb) disassemble main
Dump of assembler code for function main:
0x00000000004004f0 <+0>: xorps %xmm0,%xmm0
0x00000000004004f3 <+3>: movaps %xmm0,-0x18(%rsp)
0x00000000004004f8 <+8>: movl $0x0,-0x8(%rsp)
0x0000000000400500 <+16>: lea -0x18(%rsp),%rax
0x0000000000400505 <+21>: xor %eax,%eax
0x0000000000400507 <+23>: retq
End of assembler dump.
As gcc, clang, but also icc defines __GNUC__, it's sufficient to define
this in compiler-gcc.h only to be picked up. For a fallback or otherwise
unsupported compiler, we define it as a barrier. Similarly, for ecc which
does not support gcc inline asm.
Reference: https://llvm.org/bugs/show_bug.cgi?id=15495
Reported-by: Stephan Mueller <smueller@chronox.de>
Tested-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Stephan Mueller <smueller@chronox.de>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Cc: mancha security <mancha1@zoho.com>
Cc: Mark Charlebois <charlebm@gmail.com>
Cc: Behan Webster <behanw@converseincode.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
2015-04-30 09:13:52 +07:00
|
|
|
#ifndef barrier_data
|
|
|
|
# define barrier_data(ptr) barrier()
|
|
|
|
#endif
|
|
|
|
|
2009-12-05 08:44:50 +07:00
|
|
|
/* Unreachable code */
|
|
|
|
#ifndef unreachable
|
|
|
|
# define unreachable() do { } while (1)
|
|
|
|
#endif
|
|
|
|
|
kbuild: allow archs to select link dead code/data elimination
Introduce LD_DEAD_CODE_DATA_ELIMINATION option for architectures to
select to build with -ffunction-sections, -fdata-sections, and link
with --gc-sections. It requires some work (documented) to ensure all
unreferenced entrypoints are live, and requires toolchain and build
verification, so it is made a per-arch option for now.
On a random powerpc64le build, this yelds a significant size saving,
it boots and runs fine, but there is a lot I haven't tested as yet, so
these savings may be reduced if there are bugs in the link.
text data bss dec filename
11169741 1180744 1923176 14273661 vmlinux
10445269 1004127 1919707 13369103 vmlinux.dce
~700K text, ~170K data, 6% removed from kernel image size.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.com>
2016-08-24 19:29:20 +07:00
|
|
|
/*
|
|
|
|
* KENTRY - kernel entry point
|
|
|
|
* This can be used to annotate symbols (functions or data) that are used
|
|
|
|
* without their linker symbol being referenced explicitly. For example,
|
|
|
|
* interrupt vector handlers, or functions in the kernel image that are found
|
|
|
|
* programatically.
|
|
|
|
*
|
|
|
|
* Not required for symbols exported with EXPORT_SYMBOL, or initcalls. Those
|
|
|
|
* are handled in their own way (with KEEP() in linker scripts).
|
|
|
|
*
|
|
|
|
* KENTRY can be avoided if the symbols in question are marked as KEEP() in the
|
|
|
|
* linker script. For example an architecture could KEEP() its entire
|
|
|
|
* boot/exception vector code rather than annotate each function and data.
|
|
|
|
*/
|
|
|
|
#ifndef KENTRY
|
|
|
|
# define KENTRY(sym) \
|
|
|
|
extern typeof(sym) sym; \
|
|
|
|
static const unsigned long __kentry_##sym \
|
|
|
|
__used \
|
|
|
|
__attribute__((section("___kentry" "+" #sym ), used)) \
|
|
|
|
= (unsigned long)&sym;
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#ifndef RELOC_HIDE
|
|
|
|
# define RELOC_HIDE(ptr, off) \
|
|
|
|
({ unsigned long __ptr; \
|
|
|
|
__ptr = (unsigned long) (ptr); \
|
|
|
|
(typeof(ptr)) (__ptr + (off)); })
|
|
|
|
#endif
|
|
|
|
|
2013-11-26 07:00:41 +07:00
|
|
|
#ifndef OPTIMIZER_HIDE_VAR
|
|
|
|
#define OPTIMIZER_HIDE_VAR(var) barrier()
|
|
|
|
#endif
|
|
|
|
|
2012-11-22 09:00:25 +07:00
|
|
|
/* Not-quite-unique ID. */
|
|
|
|
#ifndef __UNIQUE_ID
|
|
|
|
# define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __LINE__)
|
|
|
|
#endif
|
|
|
|
|
2014-11-25 16:01:16 +07:00
|
|
|
#include <uapi/linux/types.h>
|
|
|
|
|
2015-10-19 15:37:17 +07:00
|
|
|
#define __READ_ONCE_SIZE \
|
|
|
|
({ \
|
|
|
|
switch (size) { \
|
|
|
|
case 1: *(__u8 *)res = *(volatile __u8 *)p; break; \
|
|
|
|
case 2: *(__u16 *)res = *(volatile __u16 *)p; break; \
|
|
|
|
case 4: *(__u32 *)res = *(volatile __u32 *)p; break; \
|
|
|
|
case 8: *(__u64 *)res = *(volatile __u64 *)p; break; \
|
|
|
|
default: \
|
|
|
|
barrier(); \
|
|
|
|
__builtin_memcpy((void *)res, (const void *)p, size); \
|
|
|
|
barrier(); \
|
|
|
|
} \
|
|
|
|
})
|
|
|
|
|
|
|
|
static __always_inline
|
|
|
|
void __read_once_size(const volatile void *p, void *res, int size)
|
2014-11-25 16:01:16 +07:00
|
|
|
{
|
2015-10-19 15:37:17 +07:00
|
|
|
__READ_ONCE_SIZE;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_KASAN
|
|
|
|
/*
|
|
|
|
* This function is not 'inline' because __no_sanitize_address confilcts
|
|
|
|
* with inlining. Attempt to inline it may cause a build failure.
|
|
|
|
* https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67368
|
|
|
|
* '__maybe_unused' allows us to avoid defined-but-not-used warnings.
|
|
|
|
*/
|
|
|
|
static __no_sanitize_address __maybe_unused
|
|
|
|
void __read_once_size_nocheck(const volatile void *p, void *res, int size)
|
|
|
|
{
|
|
|
|
__READ_ONCE_SIZE;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static __always_inline
|
|
|
|
void __read_once_size_nocheck(const volatile void *p, void *res, int size)
|
|
|
|
{
|
|
|
|
__READ_ONCE_SIZE;
|
2014-11-25 16:01:16 +07:00
|
|
|
}
|
2015-10-19 15:37:17 +07:00
|
|
|
#endif
|
2014-11-25 16:01:16 +07:00
|
|
|
|
2015-01-13 16:46:42 +07:00
|
|
|
static __always_inline void __write_once_size(volatile void *p, void *res, int size)
|
2014-11-25 16:01:16 +07:00
|
|
|
{
|
|
|
|
switch (size) {
|
|
|
|
case 1: *(volatile __u8 *)p = *(__u8 *)res; break;
|
|
|
|
case 2: *(volatile __u16 *)p = *(__u16 *)res; break;
|
|
|
|
case 4: *(volatile __u32 *)p = *(__u32 *)res; break;
|
|
|
|
case 8: *(volatile __u64 *)p = *(__u64 *)res; break;
|
|
|
|
default:
|
|
|
|
barrier();
|
|
|
|
__builtin_memcpy((void *)p, (const void *)res, size);
|
|
|
|
barrier();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Prevent the compiler from merging or refetching reads or writes. The
|
|
|
|
* compiler is also forbidden from reordering successive instances of
|
2015-01-13 16:46:42 +07:00
|
|
|
* READ_ONCE, WRITE_ONCE and ACCESS_ONCE (see below), but only when the
|
2014-11-25 16:01:16 +07:00
|
|
|
* compiler is aware of some particular ordering. One way to make the
|
|
|
|
* compiler aware of ordering is to put the two invocations of READ_ONCE,
|
2015-01-13 16:46:42 +07:00
|
|
|
* WRITE_ONCE or ACCESS_ONCE() in different C statements.
|
2014-11-25 16:01:16 +07:00
|
|
|
*
|
|
|
|
* In contrast to ACCESS_ONCE these two macros will also work on aggregate
|
|
|
|
* data types like structs or unions. If the size of the accessed data
|
|
|
|
* type exceeds the word size of the machine (e.g., 32 bits or 64 bits)
|
2016-01-26 04:33:20 +07:00
|
|
|
* READ_ONCE() and WRITE_ONCE() will fall back to memcpy(). There's at
|
|
|
|
* least two memcpy()s: one for the __builtin_memcpy() and then one for
|
|
|
|
* the macro doing the copy of variable - '__u' allocated on the stack.
|
2014-11-25 16:01:16 +07:00
|
|
|
*
|
|
|
|
* Their two major use cases are: (1) Mediating communication between
|
|
|
|
* process-level code and irq/NMI handlers, all running on the same CPU,
|
|
|
|
* and (2) Ensuring that the compiler does not fold, spindle, or otherwise
|
|
|
|
* mutilate accesses that either do not require ordering or that interact
|
|
|
|
* with an explicit memory barrier or atomic instruction that provides the
|
|
|
|
* required ordering.
|
|
|
|
*/
|
|
|
|
|
2015-10-19 15:37:17 +07:00
|
|
|
#define __READ_ONCE(x, check) \
|
|
|
|
({ \
|
|
|
|
union { typeof(x) __val; char __c[1]; } __u; \
|
|
|
|
if (check) \
|
|
|
|
__read_once_size(&(x), __u.__c, sizeof(x)); \
|
|
|
|
else \
|
|
|
|
__read_once_size_nocheck(&(x), __u.__c, sizeof(x)); \
|
|
|
|
__u.__val; \
|
|
|
|
})
|
|
|
|
#define READ_ONCE(x) __READ_ONCE(x, 1)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Use READ_ONCE_NOCHECK() instead of READ_ONCE() if you need
|
|
|
|
* to hide memory access from KASAN.
|
|
|
|
*/
|
|
|
|
#define READ_ONCE_NOCHECK(x) __READ_ONCE(x, 0)
|
2014-11-25 16:01:16 +07:00
|
|
|
|
2015-01-13 16:46:42 +07:00
|
|
|
#define WRITE_ONCE(x, val) \
|
2015-08-04 14:55:48 +07:00
|
|
|
({ \
|
|
|
|
union { typeof(x) __val; char __c[1]; } __u = \
|
|
|
|
{ .__val = (__force typeof(x)) (val) }; \
|
|
|
|
__write_once_size(&(x), __u.__c, sizeof(x)); \
|
|
|
|
__u.__val; \
|
|
|
|
})
|
2014-11-25 16:01:16 +07:00
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#endif /* __KERNEL__ */
|
|
|
|
|
|
|
|
#endif /* __ASSEMBLY__ */
|
|
|
|
|
2006-05-02 16:41:25 +07:00
|
|
|
#ifdef __KERNEL__
|
2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* Allow us to mark functions as 'deprecated' and have gcc emit a nice
|
|
|
|
* warning for each use, in hopes of speeding the functions removal.
|
|
|
|
* Usage is:
|
|
|
|
* int __deprecated foo(void)
|
|
|
|
*/
|
|
|
|
#ifndef __deprecated
|
|
|
|
# define __deprecated /* unimplemented */
|
|
|
|
#endif
|
|
|
|
|
2005-05-01 22:59:03 +07:00
|
|
|
#ifdef MODULE
|
|
|
|
#define __deprecated_for_modules __deprecated
|
|
|
|
#else
|
|
|
|
#define __deprecated_for_modules
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#ifndef __must_check
|
|
|
|
#define __must_check
|
|
|
|
#endif
|
|
|
|
|
2006-08-15 12:43:18 +07:00
|
|
|
#ifndef CONFIG_ENABLE_MUST_CHECK
|
|
|
|
#undef __must_check
|
|
|
|
#define __must_check
|
|
|
|
#endif
|
2007-10-25 15:06:13 +07:00
|
|
|
#ifndef CONFIG_ENABLE_WARN_DEPRECATED
|
|
|
|
#undef __deprecated
|
|
|
|
#undef __deprecated_for_modules
|
|
|
|
#define __deprecated
|
|
|
|
#define __deprecated_for_modules
|
|
|
|
#endif
|
2006-08-15 12:43:18 +07:00
|
|
|
|
2016-05-20 07:10:52 +07:00
|
|
|
#ifndef __malloc
|
|
|
|
#define __malloc
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* Allow us to avoid 'defined but not used' warnings on functions and data,
|
|
|
|
* as well as force them to be emitted to the assembly file.
|
|
|
|
*
|
2007-05-09 16:35:27 +07:00
|
|
|
* As of gcc 3.4, static functions that are not marked with attribute((used))
|
|
|
|
* may be elided from the assembly file. As of gcc 3.4, static data not so
|
2005-04-17 05:20:36 +07:00
|
|
|
* marked will not be elided, but this may change in a future gcc version.
|
|
|
|
*
|
2007-05-09 16:35:27 +07:00
|
|
|
* NOTE: Because distributions shipped with a backported unit-at-a-time
|
|
|
|
* compiler in gcc 3.3, we must define __used to be __attribute__((used))
|
|
|
|
* for gcc >=3.3 instead of 3.4.
|
|
|
|
*
|
2005-04-17 05:20:36 +07:00
|
|
|
* In prior versions of gcc, such functions and data would be emitted, but
|
|
|
|
* would be warned about except with attribute((unused)).
|
2007-05-09 16:35:27 +07:00
|
|
|
*
|
|
|
|
* Mark functions that are referenced only in inline assembly as __used so
|
|
|
|
* the code is emitted even though it appears to be unreferenced.
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
2007-05-09 16:35:27 +07:00
|
|
|
#ifndef __used
|
|
|
|
# define __used /* unimplemented */
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifndef __maybe_unused
|
|
|
|
# define __maybe_unused /* unimplemented */
|
2005-04-17 05:20:36 +07:00
|
|
|
#endif
|
|
|
|
|
2009-11-02 07:50:52 +07:00
|
|
|
#ifndef __always_unused
|
|
|
|
# define __always_unused /* unimplemented */
|
|
|
|
#endif
|
|
|
|
|
2006-05-04 06:41:02 +07:00
|
|
|
#ifndef noinline
|
|
|
|
#define noinline
|
|
|
|
#endif
|
|
|
|
|
2008-03-05 05:28:40 +07:00
|
|
|
/*
|
|
|
|
* Rather then using noinline to prevent stack consumption, use
|
2012-02-23 19:42:30 +07:00
|
|
|
* noinline_for_stack instead. For documentation reasons.
|
2008-03-05 05:28:40 +07:00
|
|
|
*/
|
|
|
|
#define noinline_for_stack noinline
|
|
|
|
|
2006-05-04 06:41:02 +07:00
|
|
|
#ifndef __always_inline
|
|
|
|
#define __always_inline inline
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#endif /* __KERNEL__ */
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/*
|
|
|
|
* From the GCC manual:
|
|
|
|
*
|
|
|
|
* Many functions do not examine any values except their arguments,
|
|
|
|
* and have no effects except the return value. Basically this is
|
|
|
|
* just slightly more strict class than the `pure' attribute above,
|
|
|
|
* since function is not allowed to read global memory.
|
|
|
|
*
|
|
|
|
* Note that a function that has pointer arguments and examines the
|
|
|
|
* data pointed to must _not_ be declared `const'. Likewise, a
|
|
|
|
* function that calls a non-`const' function usually must not be
|
|
|
|
* `const'. It does not make sense for a `const' function to return
|
|
|
|
* `void'.
|
|
|
|
*/
|
|
|
|
#ifndef __attribute_const__
|
|
|
|
# define __attribute_const__ /* unimplemented */
|
|
|
|
#endif
|
|
|
|
|
2017-04-05 23:49:19 +07:00
|
|
|
#ifndef __designated_init
|
|
|
|
# define __designated_init
|
|
|
|
#endif
|
|
|
|
|
2016-06-21 01:42:34 +07:00
|
|
|
#ifndef __latent_entropy
|
|
|
|
# define __latent_entropy
|
|
|
|
#endif
|
|
|
|
|
2017-05-06 13:37:45 +07:00
|
|
|
#ifndef __randomize_layout
|
|
|
|
# define __randomize_layout __designated_init
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifndef __no_randomize_layout
|
|
|
|
# define __no_randomize_layout
|
|
|
|
#endif
|
|
|
|
|
2017-04-06 12:43:33 +07:00
|
|
|
#ifndef randomized_struct_fields_start
|
|
|
|
# define randomized_struct_fields_start
|
|
|
|
# define randomized_struct_fields_end
|
|
|
|
#endif
|
|
|
|
|
2007-07-21 22:10:00 +07:00
|
|
|
/*
|
|
|
|
* Tell gcc if a function is cold. The compiler will assume any path
|
|
|
|
* directly leading to the call is unlikely.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef __cold
|
|
|
|
#define __cold
|
|
|
|
#endif
|
|
|
|
|
2008-01-21 00:54:48 +07:00
|
|
|
/* Simple shorthand for a section definition */
|
|
|
|
#ifndef __section
|
|
|
|
# define __section(S) __attribute__ ((__section__(#S)))
|
|
|
|
#endif
|
|
|
|
|
2012-09-18 04:09:15 +07:00
|
|
|
#ifndef __visible
|
|
|
|
#define __visible
|
|
|
|
#endif
|
|
|
|
|
2015-11-06 09:45:02 +07:00
|
|
|
/*
|
|
|
|
* Assume alignment of return value.
|
|
|
|
*/
|
|
|
|
#ifndef __assume_aligned
|
|
|
|
#define __assume_aligned(a, ...)
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
2009-06-13 10:46:56 +07:00
|
|
|
/* Are two types/vars the same type (ignoring qualifiers)? */
|
|
|
|
#ifndef __same_type
|
|
|
|
# define __same_type(a, b) __builtin_types_compatible_p(typeof(a), typeof(b))
|
|
|
|
#endif
|
|
|
|
|
2013-11-06 20:57:36 +07:00
|
|
|
/* Is this type a native word size -- useful for atomic operations */
|
|
|
|
#ifndef __native_word
|
compiler: Allow 1- and 2-byte smp_load_acquire() and smp_store_release()
CPUs without single-byte and double-byte loads and stores place some
"interesting" requirements on concurrent code. For example (adapted
from Peter Hurley's test code), suppose we have the following structure:
struct foo {
spinlock_t lock1;
spinlock_t lock2;
char a; /* Protected by lock1. */
char b; /* Protected by lock2. */
};
struct foo *foop;
Of course, it is common (and good) practice to place data protected
by different locks in separate cache lines. However, if the locks are
rarely acquired (for example, only in rare error cases), and there are
a great many instances of the data structure, then memory footprint can
trump false-sharing concerns, so that it can be better to place them in
the same cache cache line as above.
But if the CPU does not support single-byte loads and stores, a store
to foop->a will do a non-atomic read-modify-write operation on foop->b,
which will come as a nasty surprise to someone holding foop->lock2. So we
now require CPUs to support single-byte and double-byte loads and stores.
Therefore, this commit adjusts the definition of __native_word() to allow
these sizes to be used by smp_load_acquire() and smp_store_release().
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
2014-09-06 01:14:48 +07:00
|
|
|
# define __native_word(t) (sizeof(t) == sizeof(char) || sizeof(t) == sizeof(short) || sizeof(t) == sizeof(int) || sizeof(t) == sizeof(long))
|
2013-11-06 20:57:36 +07:00
|
|
|
#endif
|
|
|
|
|
2009-09-26 19:33:01 +07:00
|
|
|
/* Compile time object size, -1 for unknown */
|
|
|
|
#ifndef __compiletime_object_size
|
|
|
|
# define __compiletime_object_size(obj) -1
|
|
|
|
#endif
|
2009-09-30 18:05:23 +07:00
|
|
|
#ifndef __compiletime_warning
|
|
|
|
# define __compiletime_warning(message)
|
|
|
|
#endif
|
2009-10-02 21:50:50 +07:00
|
|
|
#ifndef __compiletime_error
|
|
|
|
# define __compiletime_error(message)
|
2014-06-05 06:11:16 +07:00
|
|
|
/*
|
|
|
|
* Sparse complains of variable sized arrays due to the temporary variable in
|
|
|
|
* __compiletime_assert. Unfortunately we can't just expand it out to make
|
|
|
|
* sparse see a constant array size without breaking compiletime_assert on old
|
|
|
|
* versions of GCC (e.g. 4.2.4), so hide the array from sparse altogether.
|
|
|
|
*/
|
|
|
|
# ifndef __CHECKER__
|
|
|
|
# define __compiletime_error_fallback(condition) \
|
bug.h, compiler.h: introduce compiletime_assert & BUILD_BUG_ON_MSG
Introduce compiletime_assert to compiler.h, which moves the details of
how to break a build and emit an error message for a specific compiler
to the headers where these details should be. Following in the
tradition of the POSIX assert macro, compiletime_assert creates a
build-time error when the supplied condition is *false*.
Next, we add BUILD_BUG_ON_MSG to bug.h which simply wraps
compiletime_assert, inverting the logic, so that it fails when the
condition is *true*, consistent with the language "build bug on." This
macro allows you to specify the error message you want emitted when the
supplied condition is true.
Finally, we remove all other code from bug.h that mucks with these
details (BUILD_BUG & BUILD_BUG_ON), and have them all call
BUILD_BUG_ON_MSG. This not only reduces source code bloat, but also
prevents the possibility of code being changed for one macro and not for
the other (which was previously the case for BUILD_BUG and
BUILD_BUG_ON).
Since __compiletime_error_fallback is now only used in compiler.h, I'm
considering it a private macro and removing the double negation that's
now extraneous.
[akpm@linux-foundation.org: checkpatch fixes]
Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-22 07:41:55 +07:00
|
|
|
do { ((void)sizeof(char[1 - 2 * condition])); } while (0)
|
2014-06-05 06:11:16 +07:00
|
|
|
# endif
|
|
|
|
#endif
|
|
|
|
#ifndef __compiletime_error_fallback
|
compiler.h, bug.h: prevent double error messages with BUILD_BUG{,_ON}
Prior to the introduction of __attribute__((error("msg"))) in gcc 4.3,
creating compile-time errors required a little trickery.
BUILD_BUG{,_ON} uses this attribute when available to generate
compile-time errors, but also uses the negative-sized array trick for
older compilers, resulting in two error messages in some cases. The
reason it's "some" cases is that as of gcc 4.4, the negative-sized array
will not create an error in some situations, like inline functions.
This patch replaces the negative-sized array code with the new
__compiletime_error_fallback() macro which expands to the same thing
unless the the error attribute is available, in which case it expands to
do{}while(0), resulting in exactly one compile-time error on all
versions of gcc.
Note that we are not changing the negative-sized array code for the
unoptimized version of BUILD_BUG_ON, since it has the potential to catch
problems that would be disabled in later versions of gcc were
__compiletime_error_fallback used. The reason is that that an
unoptimized build can't always remove calls to an error-attributed
function call (like we are using) that should effectively become dead
code if it were optimized. However, using a negative-sized array with a
similar value will not result in an false-positive (error). The only
caveat being that it will also fail to catch valid conditions, which we
should be expecting in an unoptimized build anyway.
Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-22 07:41:54 +07:00
|
|
|
# define __compiletime_error_fallback(condition) do { } while (0)
|
2009-10-02 21:50:50 +07:00
|
|
|
#endif
|
compiler.h, bug.h: prevent double error messages with BUILD_BUG{,_ON}
Prior to the introduction of __attribute__((error("msg"))) in gcc 4.3,
creating compile-time errors required a little trickery.
BUILD_BUG{,_ON} uses this attribute when available to generate
compile-time errors, but also uses the negative-sized array trick for
older compilers, resulting in two error messages in some cases. The
reason it's "some" cases is that as of gcc 4.4, the negative-sized array
will not create an error in some situations, like inline functions.
This patch replaces the negative-sized array code with the new
__compiletime_error_fallback() macro which expands to the same thing
unless the the error attribute is available, in which case it expands to
do{}while(0), resulting in exactly one compile-time error on all
versions of gcc.
Note that we are not changing the negative-sized array code for the
unoptimized version of BUILD_BUG_ON, since it has the potential to catch
problems that would be disabled in later versions of gcc were
__compiletime_error_fallback used. The reason is that that an
unoptimized build can't always remove calls to an error-attributed
function call (like we are using) that should effectively become dead
code if it were optimized. However, using a negative-sized array with a
similar value will not result in an false-positive (error). The only
caveat being that it will also fail to catch valid conditions, which we
should be expecting in an unoptimized build anyway.
Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-22 07:41:54 +07:00
|
|
|
|
bug.h, compiler.h: introduce compiletime_assert & BUILD_BUG_ON_MSG
Introduce compiletime_assert to compiler.h, which moves the details of
how to break a build and emit an error message for a specific compiler
to the headers where these details should be. Following in the
tradition of the POSIX assert macro, compiletime_assert creates a
build-time error when the supplied condition is *false*.
Next, we add BUILD_BUG_ON_MSG to bug.h which simply wraps
compiletime_assert, inverting the logic, so that it fails when the
condition is *true*, consistent with the language "build bug on." This
macro allows you to specify the error message you want emitted when the
supplied condition is true.
Finally, we remove all other code from bug.h that mucks with these
details (BUILD_BUG & BUILD_BUG_ON), and have them all call
BUILD_BUG_ON_MSG. This not only reduces source code bloat, but also
prevents the possibility of code being changed for one macro and not for
the other (which was previously the case for BUILD_BUG and
BUILD_BUG_ON).
Since __compiletime_error_fallback is now only used in compiler.h, I'm
considering it a private macro and removing the double negation that's
now extraneous.
[akpm@linux-foundation.org: checkpatch fixes]
Signed-off-by: Daniel Santos <daniel.santos@pobox.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Joe Perches <joe@perches.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-22 07:41:55 +07:00
|
|
|
#define __compiletime_assert(condition, msg, prefix, suffix) \
|
|
|
|
do { \
|
|
|
|
bool __cond = !(condition); \
|
|
|
|
extern void prefix ## suffix(void) __compiletime_error(msg); \
|
|
|
|
if (__cond) \
|
|
|
|
prefix ## suffix(); \
|
|
|
|
__compiletime_error_fallback(__cond); \
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
#define _compiletime_assert(condition, msg, prefix, suffix) \
|
|
|
|
__compiletime_assert(condition, msg, prefix, suffix)
|
|
|
|
|
|
|
|
/**
|
|
|
|
* compiletime_assert - break build and emit msg if condition is false
|
|
|
|
* @condition: a compile-time constant condition to check
|
|
|
|
* @msg: a message to emit if condition is false
|
|
|
|
*
|
|
|
|
* In tradition of POSIX assert, this macro will break the build if the
|
|
|
|
* supplied condition is *false*, emitting the supplied error message if the
|
|
|
|
* compiler has support to do so.
|
|
|
|
*/
|
|
|
|
#define compiletime_assert(condition, msg) \
|
|
|
|
_compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
|
|
|
|
|
2013-11-06 20:57:36 +07:00
|
|
|
#define compiletime_assert_atomic_type(t) \
|
|
|
|
compiletime_assert(__native_word(t), \
|
|
|
|
"Need native word sized stores/loads for atomicity.")
|
|
|
|
|
2008-05-11 09:51:16 +07:00
|
|
|
/*
|
|
|
|
* Prevent the compiler from merging or refetching accesses. The compiler
|
|
|
|
* is also forbidden from reordering successive instances of ACCESS_ONCE(),
|
|
|
|
* but only when the compiler is aware of some particular ordering. One way
|
|
|
|
* to make the compiler aware of ordering is to put the two invocations of
|
|
|
|
* ACCESS_ONCE() in different C statements.
|
|
|
|
*
|
2014-11-25 16:16:39 +07:00
|
|
|
* ACCESS_ONCE will only work on scalar types. For union types, ACCESS_ONCE
|
|
|
|
* on a union member will work as long as the size of the member matches the
|
|
|
|
* size of the union and the size is smaller than word size.
|
|
|
|
*
|
|
|
|
* The major use cases of ACCESS_ONCE used to be (1) Mediating communication
|
|
|
|
* between process-level code and irq/NMI handlers, all running on the same CPU,
|
|
|
|
* and (2) Ensuring that the compiler does not fold, spindle, or otherwise
|
|
|
|
* mutilate accesses that either do not require ordering or that interact
|
|
|
|
* with an explicit memory barrier or atomic instruction that provides the
|
|
|
|
* required ordering.
|
|
|
|
*
|
2015-04-30 18:57:21 +07:00
|
|
|
* If possible use READ_ONCE()/WRITE_ONCE() instead.
|
2008-05-11 09:51:16 +07:00
|
|
|
*/
|
2014-11-25 16:16:39 +07:00
|
|
|
#define __ACCESS_ONCE(x) ({ \
|
2015-01-12 18:13:39 +07:00
|
|
|
__maybe_unused typeof(x) __var = (__force typeof(x)) 0; \
|
2014-11-25 16:16:39 +07:00
|
|
|
(volatile typeof(x) *)&(x); })
|
|
|
|
#define ACCESS_ONCE(x) (*__ACCESS_ONCE(x))
|
2008-05-11 09:51:16 +07:00
|
|
|
|
2015-05-27 08:39:36 +07:00
|
|
|
/**
|
|
|
|
* lockless_dereference() - safely load a pointer for later dereference
|
|
|
|
* @p: The pointer to load
|
|
|
|
*
|
|
|
|
* Similar to rcu_dereference(), but for situations where the pointed-to
|
|
|
|
* object's lifetime is managed by something other than RCU. That
|
|
|
|
* "something other" might be reference counting or simple immortality.
|
2016-05-22 17:48:27 +07:00
|
|
|
*
|
2016-08-26 13:16:00 +07:00
|
|
|
* The seemingly unused variable ___typecheck_p validates that @p is
|
|
|
|
* indeed a pointer type by using a pointer to typeof(*p) as the type.
|
|
|
|
* Taking a pointer to typeof(*p) again is needed in case p is void *.
|
2015-05-27 08:39:36 +07:00
|
|
|
*/
|
|
|
|
#define lockless_dereference(p) \
|
|
|
|
({ \
|
2015-05-28 14:20:58 +07:00
|
|
|
typeof(p) _________p1 = READ_ONCE(p); \
|
2016-08-26 13:16:00 +07:00
|
|
|
typeof(*(p)) *___typecheck_p __maybe_unused; \
|
2015-05-27 08:39:36 +07:00
|
|
|
smp_read_barrier_depends(); /* Dependency order vs. p above. */ \
|
|
|
|
(_________p1); \
|
|
|
|
})
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
#endif /* __LINUX_COMPILER_H */
|