On Mon, 2015-11-02 at 00:24 +0000, Al Viro wrote:
> This ought to be a bit cleaner. Eric, could you test the variant below on your
> setup?
Sure !
5 runs of :
lpaa24:~# taskset ff0ff ./opensock -t 16 -n 10000000 -l 10
total = 4386311
total = 4560402
total = 4437309
total = 4516227
total = 4478778
With 48 threads :
./opensock -t 48 -n 10000000 -l 10
total = 4940245
total = 4848513
total = 4813153
total = 4813946
total = 5127804
Perf output taken on the 16 threads run :
74.71% opensock opensock [.] memset
|
--- memset
5.64% opensock [kernel.kallsyms] [k] queued_spin_lock_slowpath
|
--- queued_spin_lock_slowpath
|
|--99.89%-- _raw_spin_lock
| |
| |--52.74%-- __close_fd
| | sys_close
| | entry_SYSCALL_64_fastpath
| | __libc_close
| | |
| | --100.00%-- 0x0
| |
| |--46.97%-- __alloc_fd
| | get_unused_fd_flags
| | sock_map_fd
| | sys_socket
| | entry_SYSCALL_64_fastpath
| | __socket
| | |
| | --100.00%-- 0x0
| --0.30%-- [...]
--0.11%-- [...]
1.69% opensock [kernel.kallsyms] [k] _raw_spin_lock
|
--- _raw_spin_lock
|
|--27.37%-- get_unused_fd_flags
| sock_map_fd
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--22.40%-- sys_close
| entry_SYSCALL_64_fastpath
| __libc_close
|
|--15.79%-- cache_alloc_refill
| |
| |--99.27%-- kmem_cache_alloc
| | |
| | |--81.25%-- sk_prot_alloc
| | | sk_alloc
| | | inet_create
| | | __sock_create
| | | sock_create
| | | sys_socket
| | | entry_SYSCALL_64_fastpath
| | | __socket
| | |
| | |--9.08%-- sock_alloc_inode
| | | alloc_inode
| | | new_inode_pseudo
| | | sock_alloc
| | | __sock_create
| | | sock_create
| | | sys_socket
| | | entry_SYSCALL_64_fastpath
| | | __socket
| | |
| | |--4.98%-- __d_alloc
| | | d_alloc_pseudo
| | | sock_alloc_file
| | | sock_map_fd
| | | sys_socket
| | | entry_SYSCALL_64_fastpath
| | | __socket
| | |
| | --4.69%-- get_empty_filp
| | alloc_file
| | sock_alloc_file
| | sock_map_fd
| | sys_socket
| | entry_SYSCALL_64_fastpath
| | __socket
| |
| --0.73%-- kmem_cache_alloc_trace
| sock_alloc_inode
| alloc_inode
| new_inode_pseudo
| sock_alloc
| __sock_create
| sock_create
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--10.80%-- sock_alloc_file
| sock_map_fd
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--7.47%-- sock_alloc
| __sock_create
| sock_create
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--6.96%-- kmem_cache_alloc
| |
| |--72.94%-- sk_prot_alloc
| | sk_alloc
| | inet_create
| | __sock_create
| | sock_create
| | sys_socket
| | entry_SYSCALL_64_fastpath
| | __socket
| |
| |--15.51%-- sock_alloc_inode
| | alloc_inode
| | new_inode_pseudo
| | sock_alloc
| | __sock_create
| | sock_create
| | sys_socket
| | entry_SYSCALL_64_fastpath
| | __socket
| |
| |--7.59%-- get_empty_filp
| | alloc_file
| | sock_alloc_file
| | sock_map_fd
| | sys_socket
| | entry_SYSCALL_64_fastpath
| | __socket
| |
| --3.96%-- __d_alloc
| d_alloc_pseudo
| sock_alloc_file
| sock_map_fd
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--3.74%-- d_instantiate
| sock_alloc_file
| sock_map_fd
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--2.03%-- iput
| __dentry_kill
| dput
| __fput
| ____fput
| task_work_run
| prepare_exit_to_usermode
| syscall_return_slowpath
| int_ret_from_sys_call
| __libc_close
|
|--0.60%-- __fsnotify_inode_delete
| __destroy_inode
| destroy_inode
| evict
| iput
| __dentry_kill
| dput
| __fput
| ____fput
| task_work_run
| prepare_exit_to_usermode
| syscall_return_slowpath
| int_ret_from_sys_call
| __libc_close
|
|--0.55%-- evict
| iput
| __dentry_kill
| dput
| __fput
| ____fput
| task_work_run
| prepare_exit_to_usermode
| syscall_return_slowpath
| int_ret_from_sys_call
| __libc_close
|
|--0.53%-- inet_release
| sock_release
| sock_close
| __fput
| ____fput
| task_work_run
| prepare_exit_to_usermode
| syscall_return_slowpath
| int_ret_from_sys_call
| __libc_close
|
|--0.51%-- __fput
| ____fput
| task_work_run
| prepare_exit_to_usermode
| syscall_return_slowpath
Perf taken on the 48 threads run :
48.34% opensock [kernel.kallsyms] [k] queued_spin_lock_slowpath
|
--- queued_spin_lock_slowpath
|
|--99.93%-- _raw_spin_lock
| |
| |--50.11%-- __close_fd
| | sys_close
| | entry_SYSCALL_64_fastpath
| | __libc_close
| | |
| | --100.00%-- 0x0
| |
| |--49.85%-- __alloc_fd
| | get_unused_fd_flags
| | sock_map_fd
| | sys_socket
| | entry_SYSCALL_64_fastpath
| | __socket
| | |
| | --100.00%-- 0x0
| --0.03%-- [...]
--0.07%-- [...]
41.03% opensock opensock [.] memset
|
--- memset
0.69% opensock [kernel.kallsyms] [k] _raw_spin_lock
|
--- _raw_spin_lock
|
|--30.22%-- sys_close
| entry_SYSCALL_64_fastpath
| __libc_close
| |
| --100.00%-- 0x0
|
|--30.15%-- get_unused_fd_flags
| sock_map_fd
| sys_socket
| entry_SYSCALL_64_fastpath
| __socket
|
|--14.61%-- cache_alloc_refill
| |
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
