Re: Apparent regression in blktests since 5.18-rc1+

[Yanjun Zhu <yanjun.zhu@linux.dev>]

在 2022/5/7 9:55, Yanjun Zhu 写道:
>
> 在 2022/5/7 9:29, Jason Gunthorpe 写道:
>> On Sat, May 07, 2022 at 08:29:31AM +0800, Yanjun Zhu wrote:
>>
>>>> If I try to run the SRP test 002 with the soft-RoCE driver, the
>>>> following appears:
>>>>
>>>> [  749.901966] ================================
>>>> [  749.903638] WARNING: inconsistent lock state
>>>> [  749.905376] 5.18.0-rc5-dbg+ #1 Not tainted
>>>> [  749.907039] --------------------------------
>>>> [  749.908699] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
>>>> [  749.910646] ksoftirqd/5/40 [HC0[0]:SC1[1]:HE0:SE0] takes:
>>>> [  749.912499] ffff88818244d350 (&xa->xa_lock#14){+.?.}-{2:2}, at:
>>>> rxe_pool_get_index+0x73/0x170 [rdma_rxe]
>>>> [  749.914691] {SOFTIRQ-ON-W} state was registered at:
>>>> [  749.916648]   __lock_acquire+0x45b/0xce0
>>>> [  749.918599]   lock_acquire+0x18a/0x450
>>>> [  749.920480]   _raw_spin_lock+0x34/0x50
>>>> [  749.922580]   __rxe_add_to_pool+0xcc/0x140 [rdma_rxe]
>>>> [  749.924583]   rxe_alloc_pd+0x2d/0x40 [rdma_rxe]
>>>> [  749.926394]   __ib_alloc_pd+0xa3/0x270 [ib_core]
>>>> [  749.928579]   ib_mad_port_open+0x44a/0x790 [ib_core]
>>>> [  749.930640]   ib_mad_init_device+0x8e/0x110 [ib_core]
>>>> [  749.932495]   add_client_context+0x26a/0x330 [ib_core]
>>>> [  749.934302]   enable_device_and_get+0x169/0x2b0 [ib_core]
>>>> [  749.936217]   ib_register_device+0x26f/0x330 [ib_core]
>>>> [  749.938020]   rxe_register_device+0x1b4/0x1d0 [rdma_rxe]
>>>> [  749.939794]   rxe_add+0x8c/0xc0 [rdma_rxe]
>>>> [  749.941552]   rxe_net_add+0x5b/0x90 [rdma_rxe]
>>>> [  749.943356]   rxe_newlink+0x71/0x80 [rdma_rxe]
>>>> [  749.945182]   nldev_newlink+0x21e/0x370 [ib_core]
>>>> [  749.946917]   rdma_nl_rcv_msg+0x200/0x410 [ib_core]
>>>> [  749.948657]   rdma_nl_rcv+0x140/0x220 [ib_core]
>>>> [  749.950373]   netlink_unicast+0x307/0x460
>>>> [  749.952063]   netlink_sendmsg+0x422/0x750
>>>> [  749.953672]   __sys_sendto+0x1c2/0x250
>>>> [  749.955281]   __x64_sys_sendto+0x7f/0x90
>>>> [  749.956849]   do_syscall_64+0x35/0x80
>>>> [  749.958353]   entry_SYSCALL_64_after_hwframe+0x44/0xae
>>>> [  749.959942] irq event stamp: 1411849
>>>> [  749.961517] hardirqs last  enabled at (1411848): 
>>>> [<ffffffff810cdb28>]
>>>> __local_bh_enable_ip+0x88/0xf0
>>>> [  749.963338] hardirqs last disabled at (1411849): 
>>>> [<ffffffff81ebf24d>]
>>>> _raw_spin_lock_irqsave+0x5d/0x60
>>>> [  749.965214] softirqs last  enabled at (1411838): 
>>>> [<ffffffff82200467>]
>>>> __do_softirq+0x467/0x6e1
>>>> [  749.967027] softirqs last disabled at (1411843): 
>>>> [<ffffffff810cd947>]
>>>> run_ksoftirqd+0x37/0x60
>>> To this, Please use this patch series
>>> news://nntp.lore.kernel.org:119/20220422194416.983549-1-yanjun.zhu@linux.dev 
>>>
>> No, that is the wrong fix for this. This is mismatched lock modes with
>> the lookup path in the BH, the fix is to consistently use BH locking
>> with the xarray everwhere or to use RCU. I'm expecting to go with
>> Bob's RCU patch.
>
> Bob's RCU patch causes some atomic problems. Not sure these problems 
> can be fixed properly.

I delved into Bob's rcu patch series, in this 
https://patchwork.kernel.org/project/linux-rdma/patch/20220421014042.26985-9-rpearsonhpe@gmail.com/,

Sometimes __rxe_cleanup is called between spin_lock_irq and spin_unlock_irq.

With Bob's rcu patch, this will cause oop.

Best Regards,

Zhu Yanjun

>
> Zhu Yanjun
>
>>
>> We still need a proper patch for the AH problem.
>>
>> Jason