log2: make is_power_of_2() integer constant expression when possible
PATCH 1/1 log2: make is_power_of_2() integer constant expression when possible
While is_power_of_2() is an inline function and likely gets optimized for compile time constant arguments, it still doesn't produce an integer constant expression that could be used in, say, static data initialization or case labels.
Make is_power_of_2() an integer constant expression when possible, otherwise using the inline function to avoid multiple evaluation of the parameter.
---
The alternative would be to define both a function and a macro version of is_power_of_2(), and let the callers decide what to use.
--- include/linux/log2.h | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-)
include/linux/log2.h +10 -5
@@ -37,6 +37,14 @@ int __ilog2_u64(u64 n) }#endif +#define __IS_POWER_OF_2 (__n) ((__n) != 0 && (((__n) & ((__n) - 1)) == 0)) + +static inline __attribute__((const)) +bool __is_power_of_2 (unsigned long n) +{ +return __IS_POWER_OF_2 (n); +} +/** * is_power_of_2() - check if a value is a power of two * @n: the value to check
@@ -45,11 +53,8 @@ int __ilog2_u64(u64 n)* *not* considered a power of two. * Return: true if @n is a power of 2, otherwise false. */ -static inline __attribute__((const)) -bool is_power_of_2 (unsigned long n) -{ -return (n !=0 && ((n & (n -1 )) ==0 )); -} +#define is_power_of_2 (n) \ + __builtin_choose_expr(__builtin_constant_p(n), __IS_POWER_OF_2(n), __is_power_of_2(n))/** * __roundup_pow_of_two() - round up to nearest power of two
Signature · 1 line
2.20.1
Quoting Jani Nikula (2019-03-01 12:52:07)
Quoted 11 lines
While
is_power_of_2()is an inline function and likely gets optimized for compile time constant arguments, it still doesn't produce an integer constant expression that could be used in, say, static data initialization or case labels.Make
is_power_of_2()an integer constant expression when possible, otherwise using the inline function to avoid multiple evaluation of the parameter.Cc Chris Wilson <[email protected]>Signed-off-by Jani Nikula <[email protected]>
It does what it says on the tin and allows for
static const is_pot = is_power_of_two(32);
or in the actual case of interest,
BUILD_BUG_ON(is_power_of_two(x) ? test1(x) : test2(x)).
The only question remains can build_bug.h include ilog2.h and use this macro instead of having its own copy? Or that may be overkill for a single commonplace macro.
-Chris
Quoted 4 lines · On Fri, 1 Mar 2019 14:52:07 +0200 Jani Nikula <[email protected]> wrote:
While
is_power_of_2()is an inline function and likely gets optimized for compile time constant arguments, it still doesn't produce an integer constant expression that could be used in, say, static data initialization or case labels.
hm, what code wants to do these things?
Quoted 3 lines
Make
is_power_of_2()an integer constant expression when possible, otherwise using the inline function to avoid multiple evaluation of the parameter.
Spose so. But I fear that some gcc versions will get it right and others will mess it up. While this patch is under test I think it would be best to also have at least one or two callsites which actually use the compile-time evaluation feature. Possible?
Quoted 9 lines · Andrew Morton
Quoted 4 lines · On Fri, 1 Mar 2019 14:52:07 +0200 Jani Nikula <[email protected]> wrote:
While
is_power_of_2()is an inline function and likely gets optimized for compile time constant arguments, it still doesn't produce an integer constant expression that could be used in, say, static data initialization or case labels.hm, what code wants to do these things?
Quoted 8 lines
Quoted 3 lines
Make
is_power_of_2()an integer constant expression when possible, otherwise using the inline function to avoid multiple evaluation of the parameter.Spose so. But I fear that some gcc versions will get it right and others will mess it up. While this patch is under test I think it would be best to also have at least one or two callsites which actually use the compile-time evaluation feature. Possible?
I have some drm/i915 patches that have a local compile-time version of IS_POWER_OF_2(). It'll take a while before they hit Linus' master. This can wait.
BR,
Jani.
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
A message that is not in the archive (sent off-list or not yet received): 201903041956.eLAFVhUK%[email protected]
Quoted 17 lines · kbuild test robot
Hi Jani,
I love your patch! Perhaps something to improve:
[auto build test WARNING on linus/master]
[also build test WARNING on v5.0 next-20190301]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]url: https://github.com/0day-ci/linux/commits/Jani-Nikula/log2-make-is_power_of_2-integer-constant-expression-when-possible/20190304-153529
config: microblaze-mmu_defconfig (attached as .config)
compiler: microblaze-linux-gcc (GCC) 8.2.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=8.2.0 make.cross ARCH=microblaze
Quoted 4 lines
All warnings (new ones prefixed by >>):
Quoted 1 line
Quoted 1 line
arch/microblaze/mm/pgtable.c:185: warning: "is_power_of_2" redefined#define is_power_of_2(x) ((x) != 0 && (((x) & ((x) - 1)) == 0))
Oh wow, a copy of is_power_of_2() introduced 10 years ago, apparently unused since its introduction.
BR,
Jani.
Quoted 23 lines
In file included from
include/linux/kernel.h:12,
fromarch/microblaze/mm/pgtable.c:30:
include/linux/log2.h:56: note: this is the location of the previous definition#define is_power_of_2(n) \vim +/is_power_of_2 +185
arch/microblaze/mm/pgtable.c
15902bf6Michal Simek 2009-05-26 18315902bf6Michal Simek 2009-05-26 184 /* is x a power of 2? */15902bf6Michal Simek 2009-05-26 @185 #defineis_power_of_2(x)((x) != 0 && (((x) & ((x) - 1)) == 0))15902bf6Michal Simek 2009-05-26 186:::::: The code at line 185 was first introduced by commit
::::::15902bf63c8332946e5a1f48a72e3ae22874b11bmicroblaze_mmu_v2: Page table - ioremap - pgtable.c/h, section update:::::: TO: Michal Simek <[email protected]>
:::::: CC: Michal Simek <[email protected]>---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center