btrfs: make ASSERT no-op in release builds
PATCH 1/1 btrfs: make ASSERT no-op in release builds
The current definition of ASSERT(cond) as (void)(cond) is redundant, because all checks are without side effects and don't affect code logic.
However, some checks has READ_ONCE in them or other 'compiler-unfriendly' behaviour. For example, ASSERT(list_empty) in btrfs_add_dealloc_inode was compiled to redundant mov because of this.
This patch replaces ASSERT with BUILD_BUG_ON_INVALID for !CONFIG_BTRFS_ASSERT builds.
---
.o size reductions are not that big, for example on defconfig + btrfs fs/btrfs/*.o size went from 3280528 to 3277936, so compiler was pretty efficient on his own
--- fs/btrfs/messages.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/btrfs/messages.h +1 -1
@@ -168,7 +168,7 @@ do { \#endif #else -#define ASSERT(cond, args...) (void)(cond) +#define ASSERT(cond, args...) BUILD_BUG_ON_INVALID(cond)#endif #ifdef CONFIG_BTRFS_DEBUG
base-commit: e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6
Signature · 1 line
2.51.1.dirty
Quoted 25 lines · Gladyshev Ilya
The current definition of
ASSERT(cond)as(void)(cond)is redundant, because all checks are without side effects and don't affect code logic.However, some checks has
READ_ONCEin them or other 'compiler-unfriendly' behaviour. For example,ASSERT(list_empty)inbtrfs_add_dealloc_inodewas compiled to redundant mov because of this.This patch replaces ASSERT with
BUILD_BUG_ON_INVALIDfor !CONFIG_BTRFS_ASSERT builds.Signed-off-by Gladyshev Ilya <[email protected]>---
.o size reductions are not that big, for example on defconfig + btrfsfs/btrfs/*.osize went from 3280528 to 3277936, so compiler was pretty efficient on his own--- fs/btrfs/messages.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)fs/btrfs/messages.h +0
-0
fs/btrfs/messages.h do { \ · line 171
Quoted 5 lines
fs/btrfs/messages.h +1
-1@@ -168,7 +168,7 @@ do { \#endif #else -#define ASSERT(cond, args...) (void)(cond) +#define ASSERT(cond, args...) BUILD_BUG_ON_INVALID(cond)
And I do not think it's a good idea to use BUILD_BUG_ON_INVALID() here, most ASSERT()s are checking runtime conditions, I understand you want to avoid real code generation, but in that case there are more straightforward solutions, like "do {} while (0)" as no-op.
Thanks,
Qu
Quoted 5 lines
fs/btrfs/messages.h +0
-0#endif #ifdef CONFIG_BTRFS_DEBUG base-commit:
e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6
Quoted 28 lines · Qu Wenruo
Quoted 27 lines · Gladyshev Ilya
The current definition of
ASSERT(cond)as(void)(cond)is redundant, because all checks are without side effects and don't affect code logic.However, some checks has
READ_ONCEin them or other 'compiler-unfriendly' behaviour. For example,ASSERT(list_empty)inbtrfs_add_dealloc_inodewas compiled to redundant mov because of this.This patch replaces ASSERT with
BUILD_BUG_ON_INVALIDfor !CONFIG_BTRFS_ASSERT builds.Signed-off-by Gladyshev Ilya <[email protected]>---
.o size reductions are not that big, for example on defconfig + btrfsfs/btrfs/*.osize went from 3280528 to 3277936, so compiler was pretty efficient on his own--- fs/btrfs/messages.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)fs/btrfs/messages.h +0
-0@@ -168,7 +168,7 @@ do { \#endif
Quoted 8 lines
fs/btrfs/messages.h do { \ · line 171
Quoted 3 lines
fs/btrfs/messages.h +1
-1#else -#define ASSERT(cond, args...) (void)(cond) +#define ASSERT(cond, args...) BUILD_BUG_ON_INVALID(cond)And I do not think it's a good idea to use
BUILD_BUG_ON_INVALID()here, most ASSERT()s are checking runtime conditions, I understand you want to avoid real code generation, but in that case there are more straightforward solutions, like "do {} while (0)" as no-op.
It's supposed to be no-op but also compile checked, so the do/while(0) will not do that. What BUILD_BUG_ON_INVALID is basically a sizeof(cond) so it's the right thing but the naming is confusing, we can possibly open code it.
kernel test robot
build botkernel test robot noticed the following build warnings:
Hi Gladyshev,
kernel test robot noticed the following build warnings:
[auto build test WARNING on e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6]
url: https://github.com/intel-lab-lkp/linux/commits/Gladyshev-Ilya/btrfs-make-ASSERT-no-op-in-release-builds/20251031-024059 base: e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6 patch link: https://lore.kernel.org/r/20251030182322.4085697-1-foxido%40foxido.dev patch subject: [PATCH] btrfs: make ASSERT no-op in release builds config: x86_64-kexec (https://download.01.org/0day-ci/archive/20251031/[email protected]/config) compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251031/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All warnings (new ones prefixed by >>):
>> fs/btrfs/raid56.c:302:13: warning: function 'full_page_sectors_uptodate' is not needed and will not be emitted [-Wunneeded-internal-declaration]
302 | static bool full_page_sectors_uptodate(struct btrfs_raid_bio *rbio,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
vim +/full_page_sectors_uptodate +302 fs/btrfs/raid56.c
53b381b3abeb86 David Woodhouse 2013-01-29 301
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 @302 static bool full_page_sectors_uptodate(struct btrfs_raid_bio *rbio,
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 303 unsigned int page_nr)
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 304 {
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 305 const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 306 const u32 sectors_per_page = PAGE_SIZE / sectorsize;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 307 int i;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 308
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 309 ASSERT(page_nr < rbio->nr_pages);
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 310
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 311 for (i = sectors_per_page * page_nr;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 312 i < sectors_per_page * page_nr + sectors_per_page;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 313 i++) {
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 314 if (!rbio->stripe_sectors[i].uptodate)
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 315 return false;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 316 }
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 317 return true;
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 318 }
d4e28d9b5f04d8 Qu Wenruo 2022-04-01 319 Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
Quoted 19 lines · kernel test robot
Hi Gladyshev,
kernel test robot noticed the following build warnings:
[auto build test WARNING on
e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6]url: https://github.com/intel-lab-lkp/linux/commits/Gladyshev-Ilya/btrfs-make-ASSERT-no-op-in-release-builds/20251031-024059 base: e53642b87a4f4b03a8d7e5f8507fc3cd0c595ea6 patch link: https://lore.kernel.org/r/20251030182322.4085697-1-foxido%40foxido.dev patch subject: [PATCH] btrfs: make ASSERT no-op in release builds config: x86_64-kexec (https://download.01.org/0day-ci/archive/20251031/[email protected]/config) compiler: clang version 20.1.8 (https://github.com/llvm/llvm-project 87f0227cb60147a26a1eeb4fb06e3b505e9c7261) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20251031/[email protected]/reproduce)If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Quoted 6 lines
All warnings (new ones prefixed by >>):
Quoted 1 line
Quoted 1 line
fs/btrfs/raid56.c:302:13: warning: function 'full_page_sectors_uptodate' is not needed and will not be emitted [-Wunneeded-internal-declaration]302 | static bool full_page_sectors_uptodate(struct
btrfs_raid_bio*rbio,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
Just in case it is not obvious: full_page_sectors_uptodate() is only called within an ASSERT() macro, so after this change, it is only referenced within sizeof(), so it won't be emitted in .text (which may be a bug). Presumably that is expected in this case, so I would recommend marking this as __maybe_unused to avoid the warning.
Cheers,
Nathan