lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Mon, 18 Mar 2019 15:00:39 +0800
From:   "jianchao.wang" <jianchao.w.wang@...cle.com>
To:     James Smart <jsmart2021@...il.com>, axboe@...nel.dk
Cc:     hch@....de, jthumshirn@...e.de, hare@...e.de, josef@...icpanda.com,
        bvanassche@....org, sagi@...mberg.me, keith.busch@...el.com,
        linux-block@...r.kernel.org, linux-nvme@...ts.infradead.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 7/8] nvme: use blk_mq_queue_tag_busy_iter

Hi James

On 3/16/19 12:33 AM, James Smart wrote:
> On 3/15/2019 1:57 AM, Jianchao Wang wrote:
>> blk_mq_tagset_busy_iter is not safe that it could get stale request
>> in tags->rqs[]. Use blk_mq_queue_tag_busy_iter here. A new helper
>> interface nvme_iterate_inflight_rqs is introduced to iterate
>> all of the ns under a ctrl.
>>
>> Signed-off-by: Jianchao Wang <jianchao.w.wang@...cle.com>
>> ---
>>   drivers/nvme/host/core.c   | 12 ++++++++++++
>>   drivers/nvme/host/fc.c     | 12 ++++++------
>>   drivers/nvme/host/nvme.h   |  2 ++
>>   drivers/nvme/host/pci.c    |  5 +++--
>>   drivers/nvme/host/rdma.c   |  6 +++---
>>   drivers/nvme/host/tcp.c    |  5 +++--
>>   drivers/nvme/target/loop.c |  6 +++---
>>   7 files changed, 32 insertions(+), 16 deletions(-)
>>
>> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
>> index 6a9dd68..5760fad 100644
>> --- a/drivers/nvme/host/core.c
>> +++ b/drivers/nvme/host/core.c
>> @@ -3816,6 +3816,18 @@ void nvme_start_queues(struct nvme_ctrl *ctrl)
>>   }
>>   EXPORT_SYMBOL_GPL(nvme_start_queues);
>>   +void nvme_iterate_inflight_rqs(struct nvme_ctrl *ctrl,
>> +        busy_iter_fn *fn, void *data)
>> +{
>> +    struct nvme_ns *ns;
>> +
>> +    down_read(&ctrl->namespaces_rwsem);
>> +    list_for_each_entry(ns, &ctrl->namespaces, list)
>> +        blk_mq_queue_tag_busy_iter(ns->queue, fn, data, true);
>> +    up_read(&ctrl->namespaces_rwsem);
>> +}
>> +EXPORT_SYMBOL_GPL(nvme_iterate_inflight_rqs);
> 
> Hmm... so this aborts ios for namespaces. How about ios that aren't for a namespace but rather for the controller itself.  For example, anything on the admin queue ?   Rather critical for those ios be terminated as well.
> 

The blk_mq_tagset_busy_iter which is currently used by nvme is iterating the in-flight ios for the controller.
But due to blk_mq_tagset_busy_iter is not safe which is talked in the patch 0/8,
so change it to blk_mq_queue_busy_tag_iter which is for a namespace/request_queue.

Thanks
Jianchao

Powered by blists - more mailing lists