From mboxrd@z Thu Jan  1 00:00:00 1970
From: Mike Christie <michaelc@cs.wisc.edu>
Subject: Re: [PATCH] IB/srp: Process REQ_PREEMPT requests correctly
Date: Thu, 29 Jan 2015 20:06:20 -0600
Message-ID: <54CAE71C.6020002@cs.wisc.edu>
References: <54CA2F48.605@sandisk.com> <54CA8C4C.1080508@cs.wisc.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Return-path: <linux-scsi-owner@vger.kernel.org>
In-Reply-To: <54CA8C4C.1080508@cs.wisc.edu>
Sender: linux-scsi-owner@vger.kernel.org
To: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: Roland Dreier <roland@purestorage.com>, Max Gurtuvoy <maxg@mellanox.com>, Sagi Grimberg <sagig@mellanox.com>, linux-rdma <linux-rdma@vger.kernel.org>, "linux-scsi@vger.kernel.org" <linux-scsi@vger.kernel.org>
List-Id: linux-rdma@vger.kernel.org

On 1/29/15, 1:38 PM, Mike Christie wrote:
> On 1/29/15, 7:02 AM, Bart Van Assche wrote:
>> Delay REQ_PREEMPT requests submitted against a blocked device
>> until the device is unblocked by returning SCSI_MLQUEUE_HOST_BUSY
>> to the SCSI mid-layer. This avoids that a rescan shortly after a
>> cable pull sporadically triggers the following kernel oops:
>>
>> BUG: unable to handle kernel paging request at ffffc9001a6bc084
>> IP: [<ffffffffa04e08f2>] mlx4_ib_post_send+0xd2/0xb30 [mlx4_ib]
>> Process rescan-scsi-bus (pid: 9241, threadinfo ffff88053484a000, task
>> ffff880534aae100)
>> Call Trace:
>>   [<ffffffffa0718135>] srp_post_send+0x65/0x70 [ib_srp]
>>   [<ffffffffa071b9df>] srp_queuecommand+0x1cf/0x3e0 [ib_srp]
>>   [<ffffffffa0001ff1>] scsi_dispatch_cmd+0x101/0x280 [scsi_mod]
>>   [<ffffffffa0009ad1>] scsi_request_fn+0x411/0x4d0 [scsi_mod]
>>   [<ffffffff81223b37>] __blk_run_queue+0x27/0x30
>>   [<ffffffff8122a8d2>] blk_execute_rq_nowait+0x82/0x110
>>   [<ffffffff8122a9c2>] blk_execute_rq+0x62/0xf0
>>   [<ffffffffa000b0e8>] scsi_execute+0xe8/0x190 [scsi_mod]
>>   [<ffffffffa000b2f3>] scsi_execute_req+0xa3/0x130 [scsi_mod]
>>   [<ffffffffa000c1aa>] scsi_probe_lun+0x17a/0x450 [scsi_mod]
>>   [<ffffffffa000ce86>] scsi_probe_and_add_lun+0x156/0x480 [scsi_mod]
>>   [<ffffffffa000dc2f>] __scsi_scan_target+0xdf/0x1f0 [scsi_mod]
>>   [<ffffffffa000dfa3>] scsi_scan_host_selected+0x183/0x1c0 [scsi_mod]
>>   [<ffffffffa000edfb>] scsi_scan+0xdb/0xe0 [scsi_mod]
>>   [<ffffffffa000ee13>] store_scan+0x13/0x20 [scsi_mod]
>>   [<ffffffff811c8d9b>] sysfs_write_file+0xcb/0x160
>>   [<ffffffff811589de>] vfs_write+0xce/0x140
>>   [<ffffffff81158b53>] sys_write+0x53/0xa0
>>   [<ffffffff81464592>] system_call_fastpath+0x16/0x1b
>>   [<00007f611c9d9300>] 0x7f611c9d92ff
>>
>> Reported-by: Max Gurtuvoy <maxg@mellanox.com>
>> Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
>> Cc: Sagi Grimberg <sagig@mellanox.com>
>> Cc: <stable@vger.kernel.org>
>> ---
>>   drivers/infiniband/ulp/srp/ib_srp.c | 7 ++++++-
>>   1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/infiniband/ulp/srp/ib_srp.c
>> b/drivers/infiniband/ulp/srp/ib_srp.c
>> index 0747c05..77a7a2f 100644
>> --- a/drivers/infiniband/ulp/srp/ib_srp.c
>> +++ b/drivers/infiniband/ulp/srp/ib_srp.c
>> @@ -2003,8 +2003,13 @@ static int srp_queuecommand(struct Scsi_Host
>> *shost, struct scsi_cmnd *scmnd)
>>       if (in_scsi_eh)
>>           mutex_lock(&rport->mutex);
>>
>> +    /*
>> +     * The "blocked" state of SCSI devices is ignored by the SCSI
>> core for
>> +     * REQ_PREEMPT requests. Hence the explicit check below for the SCSI
>> +     * device state.
>> +     */
>>       scmnd->result = srp_chkready(target->rport);
>> -    if (unlikely(scmnd->result))
>> +    if (unlikely(scmnd->result != 0 ||
>> scsi_device_blocked(scmnd->device)))
>>           goto err;
>>
>>       WARN_ON_ONCE(scmnd->request->tag < 0);
>>
>
> What is the case where a driver blocks the device and can handle or
> wants commands? iSCSI and FC also do not want commands, even PREEMPT
> ones, at this time. It looks like they have been hitting internal checks
> to prevent hitting similar issues.

I think I figured this out. I think we want to change the 
scsi_prep_state_check check instead of each driver/class.

It looks like for the SDEV_QUIESCE state we want to allow REQ_PREEMPT 
commands. James would know best, but I think SPI needs that ability.

For SDEV_BLOCK/SDEV_CREATED_BLOCK, it looks like drivers/classes that 
use that state do not want any commands to be queued at that time, 
because the transport is normally down.