diff mbox series

boning: use a read-write lock in bonding_show_bonds()

Message ID 20231108064641.65209-1-haifeng.xu@shopee.com
State New
Headers
Series boning: use a read-write lock in bonding_show_bonds() |

Commit Message

Haifeng Xu Nov. 8, 2023, 6:46 a.m. UTC 
  call stack:
......
PID: 210933  TASK: ffff92424e5ec080  CPU: 13  COMMAND: "kworker/u96:2"
[ffffa7a8e96bbac0] __schedule at ffffffffb0719898
[ffffa7a8e96bbb48] schedule at ffffffffb0719e9e
[ffffa7a8e96bbb68] rwsem_down_write_slowpath at ffffffffafb3167a
[ffffa7a8e96bbc00] down_write at ffffffffb071bfc1
[ffffa7a8e96bbc18] kernfs_remove_by_name_ns at ffffffffafe3593e
[ffffa7a8e96bbc48] sysfs_unmerge_group at ffffffffafe38922
[ffffa7a8e96bbc68] dpm_sysfs_remove at ffffffffb021c96a
[ffffa7a8e96bbc80] device_del at ffffffffb0209af8
[ffffa7a8e96bbcd0] netdev_unregister_kobject at ffffffffb04a6b0e
[ffffa7a8e96bbcf8] unregister_netdevice_many at ffffffffb046d3d9
[ffffa7a8e96bbd60] default_device_exit_batch at ffffffffb046d8d1
[ffffa7a8e96bbdd0] ops_exit_list at ffffffffb045e21d
[ffffa7a8e96bbe00] cleanup_net at ffffffffb045ea46
[ffffa7a8e96bbe60] process_one_work at ffffffffafad94bb
[ffffa7a8e96bbeb0] worker_thread at ffffffffafad96ad
[ffffa7a8e96bbf10] kthread at ffffffffafae132a
[ffffa7a8e96bbf50] ret_from_fork at ffffffffafa04b92

290858 PID: 278176  TASK: ffff925deb39a040  CPU: 32  COMMAND: "node-exporter"
[ffffa7a8d14dbb80] __schedule at ffffffffb0719898
[ffffa7a8d14dbc08] schedule at ffffffffb0719e9e
[ffffa7a8d14dbc28] schedule_preempt_disabled at ffffffffb071a24e
[ffffa7a8d14dbc38] __mutex_lock at ffffffffb071af28
[ffffa7a8d14dbcb8] __mutex_lock_slowpath at ffffffffb071b1a3
[ffffa7a8d14dbcc8] mutex_lock at ffffffffb071b1e2
[ffffa7a8d14dbce0] rtnl_lock at ffffffffb047f4b5
[ffffa7a8d14dbcf0] bonding_show_bonds at ffffffffc079b1a1 [bonding]
[ffffa7a8d14dbd20] class_attr_show at ffffffffb02117ce
[ffffa7a8d14dbd30] sysfs_kf_seq_show at ffffffffafe37ba1
[ffffa7a8d14dbd50] kernfs_seq_show at ffffffffafe35c07
[ffffa7a8d14dbd60] seq_read_iter at ffffffffafd9fce0
[ffffa7a8d14dbdc0] kernfs_fop_read_iter at ffffffffafe36a10
[ffffa7a8d14dbe00] new_sync_read at ffffffffafd6de23
[ffffa7a8d14dbe90] vfs_read at ffffffffafd6e64e
[ffffa7a8d14dbed0] ksys_read at ffffffffafd70977
[ffffa7a8d14dbf10] __x64_sys_read at ffffffffafd70a0a
[ffffa7a8d14dbf20] do_syscall_64 at ffffffffb070bf1c
[ffffa7a8d14dbf50] entry_SYSCALL_64_after_hwframe at ffffffffb080007c
......

Problem description:

Thread 210933 holds the rtnl_mutex and tries to acquire the kernfs_rwsem,
but there are many readers which hold the kernfs_rwsem, so it has to sleep
for a long time to wait the readers release the lock. Thread 278176 and any
other threads which call bonding_show_bonds() also need to wait because
they try to accuire the rtnl_mutex.

bonding_show_bonds() uses rtnl_mutex to protect the bond_list traversal.
However, the addition and deletion of bond_list are only performed in
bond_init()/bond_uninit(), so we can intoduce a separate read-write lock
to synchronize bond list mutation.

What's the benefits of this change?

1) All threads which call bonding_show_bonds() only wait when the
registration or unregistration of bond device happens.

2) There are many other users of rtnl_mutex, so bonding_show_bonds()
won't compete with them.

In a word, this change reduces the lock contention of rtnl_mutex.

Signed-off-by: Haifeng Xu <haifeng.xu@shopee.com>
---
 drivers/net/bonding/bond_main.c  | 4 ++++
 drivers/net/bonding/bond_sysfs.c | 6 ++++--
 include/net/bonding.h            | 3 +++
 3 files changed, 11 insertions(+), 2 deletions(-)

Comments

Eric Dumazet Nov. 8, 2023, 2:19 p.m. UTC | #1 
On Wed, Nov 8, 2023 at 7:47 AM Haifeng Xu <haifeng.xu@shopee.com> wrote:
>
> call stack:

These stacks should either be removed from the changelog, or moved
_after_ the description
of the problem. These are normal looking call stacks, you are not
fixing a crash or deadlock.

> ......
> PID: 210933  TASK: ffff92424e5ec080  CPU: 13  COMMAND: "kworker/u96:2"
> [ffffa7a8e96bbac0] __schedule at ffffffffb0719898
> [ffffa7a8e96bbb48] schedule at ffffffffb0719e9e
> [ffffa7a8e96bbb68] rwsem_down_write_slowpath at ffffffffafb3167a
> [ffffa7a8e96bbc00] down_write at ffffffffb071bfc1
> [ffffa7a8e96bbc18] kernfs_remove_by_name_ns at ffffffffafe3593e
> [ffffa7a8e96bbc48] sysfs_unmerge_group at ffffffffafe38922
> [ffffa7a8e96bbc68] dpm_sysfs_remove at ffffffffb021c96a
> [ffffa7a8e96bbc80] device_del at ffffffffb0209af8
> [ffffa7a8e96bbcd0] netdev_unregister_kobject at ffffffffb04a6b0e
> [ffffa7a8e96bbcf8] unregister_netdevice_many at ffffffffb046d3d9
> [ffffa7a8e96bbd60] default_device_exit_batch at ffffffffb046d8d1
> [ffffa7a8e96bbdd0] ops_exit_list at ffffffffb045e21d
> [ffffa7a8e96bbe00] cleanup_net at ffffffffb045ea46
> [ffffa7a8e96bbe60] process_one_work at ffffffffafad94bb
> [ffffa7a8e96bbeb0] worker_thread at ffffffffafad96ad
> [ffffa7a8e96bbf10] kthread at ffffffffafae132a
> [ffffa7a8e96bbf50] ret_from_fork at ffffffffafa04b92
>
> 290858 PID: 278176  TASK: ffff925deb39a040  CPU: 32  COMMAND: "node-exporter"
> [ffffa7a8d14dbb80] __schedule at ffffffffb0719898
> [ffffa7a8d14dbc08] schedule at ffffffffb0719e9e
> [ffffa7a8d14dbc28] schedule_preempt_disabled at ffffffffb071a24e
> [ffffa7a8d14dbc38] __mutex_lock at ffffffffb071af28
> [ffffa7a8d14dbcb8] __mutex_lock_slowpath at ffffffffb071b1a3
> [ffffa7a8d14dbcc8] mutex_lock at ffffffffb071b1e2
> [ffffa7a8d14dbce0] rtnl_lock at ffffffffb047f4b5
> [ffffa7a8d14dbcf0] bonding_show_bonds at ffffffffc079b1a1 [bonding]
> [ffffa7a8d14dbd20] class_attr_show at ffffffffb02117ce
> [ffffa7a8d14dbd30] sysfs_kf_seq_show at ffffffffafe37ba1
> [ffffa7a8d14dbd50] kernfs_seq_show at ffffffffafe35c07
> [ffffa7a8d14dbd60] seq_read_iter at ffffffffafd9fce0
> [ffffa7a8d14dbdc0] kernfs_fop_read_iter at ffffffffafe36a10
> [ffffa7a8d14dbe00] new_sync_read at ffffffffafd6de23
> [ffffa7a8d14dbe90] vfs_read at ffffffffafd6e64e
> [ffffa7a8d14dbed0] ksys_read at ffffffffafd70977
> [ffffa7a8d14dbf10] __x64_sys_read at ffffffffafd70a0a
> [ffffa7a8d14dbf20] do_syscall_64 at ffffffffb070bf1c
> [ffffa7a8d14dbf50] entry_SYSCALL_64_after_hwframe at ffffffffb080007c
> ......
>
> Problem description:
>
> Thread 210933 holds the rtnl_mutex and tries to acquire the kernfs_rwsem,
> but there are many readers which hold the kernfs_rwsem, so it has to sleep
> for a long time to wait the readers release the lock. Thread 278176 and any
> other threads which call bonding_show_bonds() also need to wait because
> they try to accuire the rtnl_mutex.

acquire

>
> bonding_show_bonds() uses rtnl_mutex to protect the bond_list traversal.
> However, the addition and deletion of bond_list are only performed in
> bond_init()/bond_uninit(), so we can intoduce a separate read-write lock

introduce

> to synchronize bond list mutation.
>
> What's the benefits of this change?
>
> 1) All threads which call bonding_show_bonds() only wait when the
> registration or unregistration of bond device happens.
>
> 2) There are many other users of rtnl_mutex, so bonding_show_bonds()
> won't compete with them.
>
> In a word, this change reduces the lock contention of rtnl_mutex.
>

This looks good to me, but please note:

1) This is net-next material, please resend next week, because
net-next is currently closed during the merge window.

2) Using a spell checker would point few typos (including in the title
"boning" -> "bonding")

Thanks.
Haifeng Xu Nov. 9, 2023, 2:43 a.m. UTC | #2 
On 2023/11/8 22:19, Eric Dumazet wrote:
> On Wed, Nov 8, 2023 at 7:47 AM Haifeng Xu <haifeng.xu@shopee.com> wrote:
>>
>> call stack:
> 
> These stacks should either be removed from the changelog, or moved
> _after_ the description
> of the problem. These are normal looking call stacks, you are not
> fixing a crash or deadlock.
> 
>> ......
>> PID: 210933  TASK: ffff92424e5ec080  CPU: 13  COMMAND: "kworker/u96:2"
>> [ffffa7a8e96bbac0] __schedule at ffffffffb0719898
>> [ffffa7a8e96bbb48] schedule at ffffffffb0719e9e
>> [ffffa7a8e96bbb68] rwsem_down_write_slowpath at ffffffffafb3167a
>> [ffffa7a8e96bbc00] down_write at ffffffffb071bfc1
>> [ffffa7a8e96bbc18] kernfs_remove_by_name_ns at ffffffffafe3593e
>> [ffffa7a8e96bbc48] sysfs_unmerge_group at ffffffffafe38922
>> [ffffa7a8e96bbc68] dpm_sysfs_remove at ffffffffb021c96a
>> [ffffa7a8e96bbc80] device_del at ffffffffb0209af8
>> [ffffa7a8e96bbcd0] netdev_unregister_kobject at ffffffffb04a6b0e
>> [ffffa7a8e96bbcf8] unregister_netdevice_many at ffffffffb046d3d9
>> [ffffa7a8e96bbd60] default_device_exit_batch at ffffffffb046d8d1
>> [ffffa7a8e96bbdd0] ops_exit_list at ffffffffb045e21d
>> [ffffa7a8e96bbe00] cleanup_net at ffffffffb045ea46
>> [ffffa7a8e96bbe60] process_one_work at ffffffffafad94bb
>> [ffffa7a8e96bbeb0] worker_thread at ffffffffafad96ad
>> [ffffa7a8e96bbf10] kthread at ffffffffafae132a
>> [ffffa7a8e96bbf50] ret_from_fork at ffffffffafa04b92
>>
>> 290858 PID: 278176  TASK: ffff925deb39a040  CPU: 32  COMMAND: "node-exporter"
>> [ffffa7a8d14dbb80] __schedule at ffffffffb0719898
>> [ffffa7a8d14dbc08] schedule at ffffffffb0719e9e
>> [ffffa7a8d14dbc28] schedule_preempt_disabled at ffffffffb071a24e
>> [ffffa7a8d14dbc38] __mutex_lock at ffffffffb071af28
>> [ffffa7a8d14dbcb8] __mutex_lock_slowpath at ffffffffb071b1a3
>> [ffffa7a8d14dbcc8] mutex_lock at ffffffffb071b1e2
>> [ffffa7a8d14dbce0] rtnl_lock at ffffffffb047f4b5
>> [ffffa7a8d14dbcf0] bonding_show_bonds at ffffffffc079b1a1 [bonding]
>> [ffffa7a8d14dbd20] class_attr_show at ffffffffb02117ce
>> [ffffa7a8d14dbd30] sysfs_kf_seq_show at ffffffffafe37ba1
>> [ffffa7a8d14dbd50] kernfs_seq_show at ffffffffafe35c07
>> [ffffa7a8d14dbd60] seq_read_iter at ffffffffafd9fce0
>> [ffffa7a8d14dbdc0] kernfs_fop_read_iter at ffffffffafe36a10
>> [ffffa7a8d14dbe00] new_sync_read at ffffffffafd6de23
>> [ffffa7a8d14dbe90] vfs_read at ffffffffafd6e64e
>> [ffffa7a8d14dbed0] ksys_read at ffffffffafd70977
>> [ffffa7a8d14dbf10] __x64_sys_read at ffffffffafd70a0a
>> [ffffa7a8d14dbf20] do_syscall_64 at ffffffffb070bf1c
>> [ffffa7a8d14dbf50] entry_SYSCALL_64_after_hwframe at ffffffffb080007c
>> ......
>>
>> Problem description:
>>
>> Thread 210933 holds the rtnl_mutex and tries to acquire the kernfs_rwsem,
>> but there are many readers which hold the kernfs_rwsem, so it has to sleep
>> for a long time to wait the readers release the lock. Thread 278176 and any
>> other threads which call bonding_show_bonds() also need to wait because
>> they try to accuire the rtnl_mutex.
> 
> acquire
> 
>>
>> bonding_show_bonds() uses rtnl_mutex to protect the bond_list traversal.
>> However, the addition and deletion of bond_list are only performed in
>> bond_init()/bond_uninit(), so we can intoduce a separate read-write lock
> 
> introduce
> 
>> to synchronize bond list mutation.
>>
>> What's the benefits of this change?
>>
>> 1) All threads which call bonding_show_bonds() only wait when the
>> registration or unregistration of bond device happens.
>>
>> 2) There are many other users of rtnl_mutex, so bonding_show_bonds()
>> won't compete with them.
>>
>> In a word, this change reduces the lock contention of rtnl_mutex.
>>
> 
> This looks good to me, but please note:
> 
> 1) This is net-next material, please resend next week, because
> net-next is currently closed during the merge window.
> 
> 2) Using a spell checker would point few typos (including in the title
> "boning" -> "bonding")
> 
> Thanks.

Thanks for your review, I 'll send a new patch next week.
Jiri Pirko Nov. 9, 2023, 3:47 p.m. UTC | #3 
s/boning/bonding/
?
Stephen Hemminger Nov. 9, 2023, 5:55 p.m. UTC | #4 
On Wed,  8 Nov 2023 06:46:41 +0000
Haifeng Xu <haifeng.xu@shopee.com> wrote:

> call stack:
> ......
> PID: 210933  TASK: ffff92424e5ec080  CPU: 13  COMMAND: "kworker/u96:2"
> [ffffa7a8e96bbac0] __schedule at ffffffffb0719898
> [ffffa7a8e96bbb48] schedule at ffffffffb0719e9e
> [ffffa7a8e96bbb68] rwsem_down_write_slowpath at ffffffffafb3167a
> [ffffa7a8e96bbc00] down_write at ffffffffb071bfc1
> [ffffa7a8e96bbc18] kernfs_remove_by_name_ns at ffffffffafe3593e
> [ffffa7a8e96bbc48] sysfs_unmerge_group at ffffffffafe38922
> [ffffa7a8e96bbc68] dpm_sysfs_remove at ffffffffb021c96a
> [ffffa7a8e96bbc80] device_del at ffffffffb0209af8
> [ffffa7a8e96bbcd0] netdev_unregister_kobject at ffffffffb04a6b0e
> [ffffa7a8e96bbcf8] unregister_netdevice_many at ffffffffb046d3d9
> [ffffa7a8e96bbd60] default_device_exit_batch at ffffffffb046d8d1
> [ffffa7a8e96bbdd0] ops_exit_list at ffffffffb045e21d
> [ffffa7a8e96bbe00] cleanup_net at ffffffffb045ea46
> [ffffa7a8e96bbe60] process_one_work at ffffffffafad94bb
> [ffffa7a8e96bbeb0] worker_thread at ffffffffafad96ad
> [ffffa7a8e96bbf10] kthread at ffffffffafae132a
> [ffffa7a8e96bbf50] ret_from_fork at ffffffffafa04b92
> 
> 290858 PID: 278176  TASK: ffff925deb39a040  CPU: 32  COMMAND: "node-exporter"
> [ffffa7a8d14dbb80] __schedule at ffffffffb0719898
> [ffffa7a8d14dbc08] schedule at ffffffffb0719e9e
> [ffffa7a8d14dbc28] schedule_preempt_disabled at ffffffffb071a24e
> [ffffa7a8d14dbc38] __mutex_lock at ffffffffb071af28
> [ffffa7a8d14dbcb8] __mutex_lock_slowpath at ffffffffb071b1a3
> [ffffa7a8d14dbcc8] mutex_lock at ffffffffb071b1e2
> [ffffa7a8d14dbce0] rtnl_lock at ffffffffb047f4b5
> [ffffa7a8d14dbcf0] bonding_show_bonds at ffffffffc079b1a1 [bonding]
> [ffffa7a8d14dbd20] class_attr_show at ffffffffb02117ce
> [ffffa7a8d14dbd30] sysfs_kf_seq_show at ffffffffafe37ba1
> [ffffa7a8d14dbd50] kernfs_seq_show at ffffffffafe35c07
> [ffffa7a8d14dbd60] seq_read_iter at ffffffffafd9fce0
> [ffffa7a8d14dbdc0] kernfs_fop_read_iter at ffffffffafe36a10
> [ffffa7a8d14dbe00] new_sync_read at ffffffffafd6de23
> [ffffa7a8d14dbe90] vfs_read at ffffffffafd6e64e
> [ffffa7a8d14dbed0] ksys_read at ffffffffafd70977
> [ffffa7a8d14dbf10] __x64_sys_read at ffffffffafd70a0a
> [ffffa7a8d14dbf20] do_syscall_64 at ffffffffb070bf1c
> [ffffa7a8d14dbf50] entry_SYSCALL_64_after_hwframe at ffffffffb080007c
> ......
> 
> Problem description:
> 
> Thread 210933 holds the rtnl_mutex and tries to acquire the kernfs_rwsem,
> but there are many readers which hold the kernfs_rwsem, so it has to sleep
> for a long time to wait the readers release the lock. Thread 278176 and any
> other threads which call bonding_show_bonds() also need to wait because
> they try to accuire the rtnl_mutex.
> 
> bonding_show_bonds() uses rtnl_mutex to protect the bond_list traversal.
> However, the addition and deletion of bond_list are only performed in
> bond_init()/bond_uninit(), so we can intoduce a separate read-write lock
> to synchronize bond list mutation.
> 
> What's the benefits of this change?
> 
> 1) All threads which call bonding_show_bonds() only wait when the
> registration or unregistration of bond device happens.
> 
> 2) There are many other users of rtnl_mutex, so bonding_show_bonds()
> won't compete with them.
> 
> In a word, this change reduces the lock contention of rtnl_mutex.
> 
> Signed-off-by: Haifeng Xu <haifeng.xu@shopee.com>
> ---
>  drivers/net/bonding/bond_main.c  | 4 ++++
>  drivers/net/bonding/bond_sysfs.c | 6 ++++--
>  include/net/bonding.h            | 3 +++
>  3 files changed, 11 insertions(+), 2 deletions(-)

Reader-writer locks are slower than spin locks and should be discouraged.
Would it be possible to use RCU here instead?
Eric Dumazet Nov. 9, 2023, 6:03 p.m. UTC | #5 
On Thu, Nov 9, 2023 at 6:55 PM Stephen Hemminger
<stephen@networkplumber.org> wrote:

> Reader-writer locks are slower than spin locks and should be discouraged.
> Would it be possible to use RCU here instead?

I doubt there is a case for repeatedly reading this file ?

This 'lock' is only for slow paths, it doesn't really matter what it is.
Haifeng Xu Nov. 10, 2023, 1:59 a.m. UTC | #6 
On 2023/11/9 23:47, Jiri Pirko wrote:
> 
> s/boning/bonding/
> ?

Yes, Eric has pointed out the problem in last email.
Thanks.
Haifeng Xu Nov. 10, 2023, 2:35 a.m. UTC | #7 
On 2023/11/10 01:55, Stephen Hemminger wrote:
> On Wed,  8 Nov 2023 06:46:41 +0000
> Haifeng Xu <haifeng.xu@shopee.com> wrote:
> 
>> call stack:
>> ......
>> PID: 210933  TASK: ffff92424e5ec080  CPU: 13  COMMAND: "kworker/u96:2"
>> [ffffa7a8e96bbac0] __schedule at ffffffffb0719898
>> [ffffa7a8e96bbb48] schedule at ffffffffb0719e9e
>> [ffffa7a8e96bbb68] rwsem_down_write_slowpath at ffffffffafb3167a
>> [ffffa7a8e96bbc00] down_write at ffffffffb071bfc1
>> [ffffa7a8e96bbc18] kernfs_remove_by_name_ns at ffffffffafe3593e
>> [ffffa7a8e96bbc48] sysfs_unmerge_group at ffffffffafe38922
>> [ffffa7a8e96bbc68] dpm_sysfs_remove at ffffffffb021c96a
>> [ffffa7a8e96bbc80] device_del at ffffffffb0209af8
>> [ffffa7a8e96bbcd0] netdev_unregister_kobject at ffffffffb04a6b0e
>> [ffffa7a8e96bbcf8] unregister_netdevice_many at ffffffffb046d3d9
>> [ffffa7a8e96bbd60] default_device_exit_batch at ffffffffb046d8d1
>> [ffffa7a8e96bbdd0] ops_exit_list at ffffffffb045e21d
>> [ffffa7a8e96bbe00] cleanup_net at ffffffffb045ea46
>> [ffffa7a8e96bbe60] process_one_work at ffffffffafad94bb
>> [ffffa7a8e96bbeb0] worker_thread at ffffffffafad96ad
>> [ffffa7a8e96bbf10] kthread at ffffffffafae132a
>> [ffffa7a8e96bbf50] ret_from_fork at ffffffffafa04b92
>>
>> 290858 PID: 278176  TASK: ffff925deb39a040  CPU: 32  COMMAND: "node-exporter"
>> [ffffa7a8d14dbb80] __schedule at ffffffffb0719898
>> [ffffa7a8d14dbc08] schedule at ffffffffb0719e9e
>> [ffffa7a8d14dbc28] schedule_preempt_disabled at ffffffffb071a24e
>> [ffffa7a8d14dbc38] __mutex_lock at ffffffffb071af28
>> [ffffa7a8d14dbcb8] __mutex_lock_slowpath at ffffffffb071b1a3
>> [ffffa7a8d14dbcc8] mutex_lock at ffffffffb071b1e2
>> [ffffa7a8d14dbce0] rtnl_lock at ffffffffb047f4b5
>> [ffffa7a8d14dbcf0] bonding_show_bonds at ffffffffc079b1a1 [bonding]
>> [ffffa7a8d14dbd20] class_attr_show at ffffffffb02117ce
>> [ffffa7a8d14dbd30] sysfs_kf_seq_show at ffffffffafe37ba1
>> [ffffa7a8d14dbd50] kernfs_seq_show at ffffffffafe35c07
>> [ffffa7a8d14dbd60] seq_read_iter at ffffffffafd9fce0
>> [ffffa7a8d14dbdc0] kernfs_fop_read_iter at ffffffffafe36a10
>> [ffffa7a8d14dbe00] new_sync_read at ffffffffafd6de23
>> [ffffa7a8d14dbe90] vfs_read at ffffffffafd6e64e
>> [ffffa7a8d14dbed0] ksys_read at ffffffffafd70977
>> [ffffa7a8d14dbf10] __x64_sys_read at ffffffffafd70a0a
>> [ffffa7a8d14dbf20] do_syscall_64 at ffffffffb070bf1c
>> [ffffa7a8d14dbf50] entry_SYSCALL_64_after_hwframe at ffffffffb080007c
>> ......
>>
>> Problem description:
>>
>> Thread 210933 holds the rtnl_mutex and tries to acquire the kernfs_rwsem,
>> but there are many readers which hold the kernfs_rwsem, so it has to sleep
>> for a long time to wait the readers release the lock. Thread 278176 and any
>> other threads which call bonding_show_bonds() also need to wait because
>> they try to accuire the rtnl_mutex.
>>
>> bonding_show_bonds() uses rtnl_mutex to protect the bond_list traversal.
>> However, the addition and deletion of bond_list are only performed in
>> bond_init()/bond_uninit(), so we can intoduce a separate read-write lock
>> to synchronize bond list mutation.
>>
>> What's the benefits of this change?
>>
>> 1) All threads which call bonding_show_bonds() only wait when the
>> registration or unregistration of bond device happens.
>>
>> 2) There are many other users of rtnl_mutex, so bonding_show_bonds()
>> won't compete with them.
>>
>> In a word, this change reduces the lock contention of rtnl_mutex.
>>
>> Signed-off-by: Haifeng Xu <haifeng.xu@shopee.com>
>> ---
>>  drivers/net/bonding/bond_main.c  | 4 ++++
>>  drivers/net/bonding/bond_sysfs.c | 6 ++++--
>>  include/net/bonding.h            | 3 +++
>>  3 files changed, 11 insertions(+), 2 deletions(-)
> 
> Reader-writer locks are slower than spin locks and should be discouraged> Would it be possible to use RCU here instead?
 
In most cases，there are many threads which want to iterate over the bond_list,
the registration or unregistration of bond device rarely happens.
diff mbox series

Patch

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 51d47eda1c87..ac4773d19beb 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -5951,7 +5951,9 @@  static void bond_uninit(struct net_device *bond_dev)
 
 	bond_set_slave_arr(bond, NULL, NULL);
 
+	write_lock(&bonding_dev_lock);
 	list_del(&bond->bond_list);
+	write_unlock(&bonding_dev_lock);
 
 	bond_debug_unregister(bond);
 }
@@ -6364,7 +6366,9 @@  static int bond_init(struct net_device *bond_dev)
 	spin_lock_init(&bond->stats_lock);
 	netdev_lockdep_set_classes(bond_dev);
 
+	write_lock(&bonding_dev_lock);
 	list_add_tail(&bond->bond_list, &bn->dev_list);
+	write_unlock(&bonding_dev_lock);
 
 	bond_prepare_sysfs_group(bond);
 
diff --git a/drivers/net/bonding/bond_sysfs.c b/drivers/net/bonding/bond_sysfs.c
index 2805135a7205..e107c1d7a6bf 100644
--- a/drivers/net/bonding/bond_sysfs.c
+++ b/drivers/net/bonding/bond_sysfs.c
@@ -28,6 +28,8 @@ 
 
 #define to_bond(cd)	((struct bonding *)(netdev_priv(to_net_dev(cd))))
 
+DEFINE_RWLOCK(bonding_dev_lock);
+
 /* "show" function for the bond_masters attribute.
  * The class parameter is ignored.
  */
@@ -40,7 +42,7 @@  static ssize_t bonding_show_bonds(const struct class *cls,
 	int res = 0;
 	struct bonding *bond;
 
-	rtnl_lock();
+	read_lock(&bonding_dev_lock);
 
 	list_for_each_entry(bond, &bn->dev_list, bond_list) {
 		if (res > (PAGE_SIZE - IFNAMSIZ)) {
@@ -55,7 +57,7 @@  static ssize_t bonding_show_bonds(const struct class *cls,
 	if (res)
 		buf[res-1] = '\n'; /* eat the leftover space */
 
-	rtnl_unlock();
+	read_unlock(&bonding_dev_lock);
 	return res;
 }
 
diff --git a/include/net/bonding.h b/include/net/bonding.h
index 5b8b1b644a2d..584ba4b5b8df 100644
--- a/include/net/bonding.h
+++ b/include/net/bonding.h
@@ -777,6 +777,9 @@  extern struct rtnl_link_ops bond_link_ops;
 /* exported from bond_sysfs_slave.c */
 extern const struct sysfs_ops slave_sysfs_ops;
 
+/* exported from bond_sysfs.c */
+extern rwlock_t bonding_dev_lock;
+
 /* exported from bond_3ad.c */
 extern const u8 lacpdu_mcast_addr[];