public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* Re: Poll CQ syncing problem
       [not found] ` <3ba1baab-e2ac-358d-3b3b-ff4a27405c93@mellanox.com>
@ 2017-03-01 14:51   ` Christoph Hellwig
  2017-03-01 15:28     ` Noa Osherovich
  0 siblings, 1 reply; 2+ messages in thread
From: Christoph Hellwig @ 2017-03-01 14:51 UTC (permalink / raw)
  To: Noa Osherovich; +Cc: hch, sagi, linux-rdma, Majd Dibbiny, tj, linux-kernel

On Wed, Mar 01, 2017 at 04:30:26PM +0200, Noa Osherovich wrote:
> Analysis:
> Since ib_comp_wq isn't single threaded, two works can run in parallel for the same CQ,
> executing __ib_process_cq.

They shouldn't.  Each CQ has a single work_struct, and any given work_struct
should only be executing at once:

"Note that the flag ``WQ_NON_REENTRANT`` no longer exists as all
workqueues are now non-reentrant - any work item is guaranteed to be
executed by at most one worker system-wide at any given time."

> Since this function isn't thread safe and the wc array is shared, it causes a data corruption
> which eventually crashes in the MAD layer due to a double list_del of the same element.

This should not be the case.  What kernel version are you testing and does
it contain any patches touching core kernel code?

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: Poll CQ syncing problem
  2017-03-01 14:51   ` Poll CQ syncing problem Christoph Hellwig
@ 2017-03-01 15:28     ` Noa Osherovich
  0 siblings, 0 replies; 2+ messages in thread
From: Noa Osherovich @ 2017-03-01 15:28 UTC (permalink / raw)
  To: Christoph Hellwig; +Cc: sagi, linux-rdma, Majd Dibbiny, tj, linux-kernel

On 3/1/2017 4:51 PM, Christoph Hellwig wrote:

> On Wed, Mar 01, 2017 at 04:30:26PM +0200, Noa Osherovich wrote:
>> Analysis:
>> Since ib_comp_wq isn't single threaded, two works can run in parallel for the same CQ,
>> executing __ib_process_cq.
> They shouldn't.  Each CQ has a single work_struct, and any given work_struct
> should only be executing at once:
>
> "Note that the flag ``WQ_NON_REENTRANT`` no longer exists as all
> workqueues are now non-reentrant - any work item is guaranteed to be
> executed by at most one worker system-wide at any given time."
>
>> Since this function isn't thread safe and the wc array is shared, it causes a data corruption
>> which eventually crashes in the MAD layer due to a double list_del of the same element.
> This should not be the case.  What kernel version are you testing and does
> it contain any patches touching core kernel code?

Thanks Christoph for the quick response.

Currently we see this only in old kernels. I'll investigate this more and update.

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2017-03-01 15:46 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <b4355d22-fc79-c860-de8a-5a4d468c884d@mellanox.com>
     [not found] ` <3ba1baab-e2ac-358d-3b3b-ff4a27405c93@mellanox.com>
2017-03-01 14:51   ` Poll CQ syncing problem Christoph Hellwig
2017-03-01 15:28     ` Noa Osherovich

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox