This looks like the ib_device ref count bug we are trying to resolve. I am in a tussle with Leon about style. Should get fixed very soon.
bob
-----Original Message-----
From: Bart Van Assche <bvanassche@xxxxxxx>
Sent: Sunday, February 28, 2021 4:29 PM
To: linux-rdma@xxxxxxxxxxxxxxx
Cc: Bob Pearson <rpearsonhpe@xxxxxxxxx>; jgg@xxxxxxxxxx; Yi Zhang <yi.zhang@xxxxxxxxxx>
Subject: Regression in rdma_rxe driver?
Hi,
If I run the following command:
(cd ~bart/software/blktests && ./check -q srp/001)
against the for-next branch of the RDMA git repository (commit
7fad751c2062 ("RDMA/srp: Apply the __packed attribute to members instead of structures") then the following appears in the kernel log:
Feb 28 14:24:04 ubuntu-vm kernel: WARNING: CPU: 5 PID: 84 at
drivers/infiniband/sw/rxe/rxe_comp.c:761 rxe_completer+0xdc5/0x10d0 [rdma_rxe] Feb 28 14:24:04 ubuntu-vm kernel: Call Trace: [ ... ] Feb 28 14:24:05 ubuntu-vm kernel: WARNING: CPU: 5 PID: 39 at
lib/refcount.c:28 refcount_warn_saturate+0x154/0x160
Feb 28 14:24:05 ubuntu-vm kernel: Call Trace: [ ... ] Feb 28 14:24:11 ubuntu-vm kernel: WARNING: CPU: 5 PID: 1471 at
lib/refcount.c:19 refcount_warn_saturate+0xa8/0x160 Feb 28 14:24:11 ubuntu-vm kernel: Call Trace: [ ... ] Feb 28 14:24:17 ubuntu-vm kernel: WARNING: CPU: 6 PID: 1501 at
drivers/infiniband/core/device.c:671 ib_dealloc_device+0x104/0x110 [ib_core] Feb 28 14:24:17 ubuntu-vm kernel: Call Trace: [ ... ] Feb 28 14:24:18 ubuntu-vm kernel: WARNING: CPU: 4 PID: 170 at
drivers/infiniband/core/device.c:493 ib_device_release+0xd3/0xe0 [ib_core] Feb 28 14:24:18 ubuntu-vm kernel: Call Trace: [ ... ]
Is this a known issue?
Thanks,
Bart.
