Re: Regarding the recent new blk-mq timeout handling

2018-06-12 Thread jianchao.wang
On 06/12/2018 09:01 PM, jianchao.wang wrote: > Hi ming > > Thanks for your kindly response. > > On 06/12/2018 06:17 PM, Ming Lei wrote: >> On Tue, Jun 12, 2018 at 6:04 PM, jianchao.wang >> wrote: >>> Hi Jens and Christoph >>> >>> In the recent commit of new blk-mq timeout handling, we don't ha

Re: Regarding the recent new blk-mq timeout handling

2018-06-12 Thread jianchao.wang
Hi ming Thanks for your kindly response. On 06/12/2018 06:17 PM, Ming Lei wrote: > On Tue, Jun 12, 2018 at 6:04 PM, jianchao.wang > wrote: >> Hi Jens and Christoph >> >> In the recent commit of new blk-mq timeout handling, we don't have any >> protection >> on timed out request against the comp

Re: Regarding the recent new blk-mq timeout handling

2018-06-12 Thread Ming Lei
On Tue, Jun 12, 2018 at 6:04 PM, jianchao.wang wrote: > Hi Jens and Christoph > > In the recent commit of new blk-mq timeout handling, we don't have any > protection > on timed out request against the completion path. We just hold a request->ref > count, > it just could avoid the request tag to

Regarding the recent new blk-mq timeout handling

2018-06-12 Thread jianchao.wang
Hi Jens and Christoph In the recent commit of new blk-mq timeout handling, we don't have any protection on timed out request against the completion path. We just hold a request->ref count, it just could avoid the request tag to be released and life-recycle, but not completion. For the scsi mid