64 bits in open_fds are mapped to a common bit in full_fds_bits. It is very
likely that a bit in full_fds_bits has been cleared before in
__clear_open_fds()'s operation. Check the clear bit in full_fds_bits before
clearing to avoid unnecessary write and cache bouncing. See commit fc90888d07b8
("vfs: conditionally clear close-on-exec flag") for a similar optimization.
Together with patch 1, they improves pts/blogbench-1.1.0 read for 27%, and write
for 14% on Intel ICX 160 cores configuration with v6.10-rc4.

Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Yu Ma <yu.ma@intel.com>
---
 fs/file.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/fs/file.c b/fs/file.c
index 50e900a47107..b4d25f6d4c19 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -268,7 +268,9 @@ static inline void __set_open_fd(unsigned int fd, struct fdtable *fdt)
 static inline void __clear_open_fd(unsigned int fd, struct fdtable *fdt)
 {
 	__clear_bit(fd, fdt->open_fds);
-	__clear_bit(fd / BITS_PER_LONG, fdt->full_fds_bits);
+	fd /= BITS_PER_LONG;
+	if (test_bit(fd, fdt->full_fds_bits))
+	    __clear_bit(fd, fdt->full_fds_bits);
 }
 
 static inline bool fd_is_open(unsigned int fd, const struct fdtable *fdt)
-- 
2.43.0

On Sat 22-06-24 11:49:03, Yu Ma wrote:
> 64 bits in open_fds are mapped to a common bit in full_fds_bits. It is very
> likely that a bit in full_fds_bits has been cleared before in
> __clear_open_fds()'s operation. Check the clear bit in full_fds_bits before
> clearing to avoid unnecessary write and cache bouncing. See commit fc90888d07b8
> ("vfs: conditionally clear close-on-exec flag") for a similar optimization.
> Together with patch 1, they improves pts/blogbench-1.1.0 read for 27%, and write
> for 14% on Intel ICX 160 cores configuration with v6.10-rc4.
> 
> Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
> Signed-off-by: Yu Ma <yu.ma@intel.com>

Nice. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
>  fs/file.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/file.c b/fs/file.c
> index 50e900a47107..b4d25f6d4c19 100644
> --- a/fs/file.c
> +++ b/fs/file.c
> @@ -268,7 +268,9 @@ static inline void __set_open_fd(unsigned int fd, struct fdtable *fdt)
>  static inline void __clear_open_fd(unsigned int fd, struct fdtable *fdt)
>  {
>  	__clear_bit(fd, fdt->open_fds);
> -	__clear_bit(fd / BITS_PER_LONG, fdt->full_fds_bits);
> +	fd /= BITS_PER_LONG;
> +	if (test_bit(fd, fdt->full_fds_bits))
> +	    __clear_bit(fd, fdt->full_fds_bits);
>  }
>  
>  static inline bool fd_is_open(unsigned int fd, const struct fdtable *fdt)
> -- 
> 2.43.0
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

On 6/25/2024 7:54 PM, Jan Kara wrote:
> On Sat 22-06-24 11:49:03, Yu Ma wrote:
>> 64 bits in open_fds are mapped to a common bit in full_fds_bits. It is very
>> likely that a bit in full_fds_bits has been cleared before in
>> __clear_open_fds()'s operation. Check the clear bit in full_fds_bits before
>> clearing to avoid unnecessary write and cache bouncing. See commit fc90888d07b8
>> ("vfs: conditionally clear close-on-exec flag") for a similar optimization.
>> Together with patch 1, they improves pts/blogbench-1.1.0 read for 27%, and write
>> for 14% on Intel ICX 160 cores configuration with v6.10-rc4.
>>
>> Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
>> Signed-off-by: Yu Ma <yu.ma@intel.com>
> Nice. Feel free to add:
>
> Reviewed-by: Jan Kara <jack@suse.cz>
>
> 								Honza

Copy that, thanks Honza :)


>> ---
>>   fs/file.c | 4 +++-
>>   1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/fs/file.c b/fs/file.c
>> index 50e900a47107..b4d25f6d4c19 100644
>> --- a/fs/file.c
>> +++ b/fs/file.c
>> @@ -268,7 +268,9 @@ static inline void __set_open_fd(unsigned int fd, struct fdtable *fdt)
>>   static inline void __clear_open_fd(unsigned int fd, struct fdtable *fdt)
>>   {
>>   	__clear_bit(fd, fdt->open_fds);
>> -	__clear_bit(fd / BITS_PER_LONG, fdt->full_fds_bits);
>> +	fd /= BITS_PER_LONG;
>> +	if (test_bit(fd, fdt->full_fds_bits))
>> +	    __clear_bit(fd, fdt->full_fds_bits);
>>   }
>>   
>>   static inline bool fd_is_open(unsigned int fd, const struct fdtable *fdt)
>> -- 
>> 2.43.0
>>