RE: [PATCH RFC] rds: ib: Reduce the contention caused by the asynchronous workers to flush the mr pool

[Praveen Kannoju]

-----Original Message-----
From: Jason Gunthorpe [mailto:jgg@xxxxxxxx] 
Sent: 19 January 2022 06:47 PM
To: Praveen Kannoju <praveen.kannoju@xxxxxxxxxx>
Cc: Leon Romanovsky <leon@xxxxxxxxxx>; Santosh Shilimkar <santosh.shilimkar@xxxxxxxxxx>; David S . Miller <davem@xxxxxxxxxxxxx>; kuba@xxxxxxxxxx; netdev@xxxxxxxxxxxxxxx; linux-rdma@xxxxxxxxxxxxxxx; rds-devel@xxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Rama Nichanamatlu <rama.nichanamatlu@xxxxxxxxxx>; Rajesh Sivaramasubramaniom <rajesh.sivaramasubramaniom@xxxxxxxxxx>
Subject: Re: [PATCH RFC] rds: ib: Reduce the contention caused by the asynchronous workers to flush the mr pool

On Wed, Jan 19, 2022 at 01:12:29PM +0000, Praveen Kannoju wrote:

> Yes, we are using the barriers. I was justifying the usage of
> smp_rmb() and smp_wmb() over smp_load_acquire() and
> smp_store_release() in the patch.

You failed to justify it.

Jason

Apologies, if my earlier point is not clear, Jason.
Let me reframe:

1. The introduced bool variable "flush_ongoing", is being accessed only in the function "rds_ib_free_mr" while spawning asynchronous workers.

2. The ordering guaranteed by smp_rmb() and smp_wmb() would be sufficient for such simple usage and hence we did not use smp_load_acquire() and smp_store_release().

3. In case the function "rds_ib_free_mr", misses to spawn the flush function, the same will be requested by the allocation path  "rds_ib_alloc_frmr" which in-turn calls "rds_ib_try_reuse_ibmr", which finally calls the flush function "rds_ib_flush_mr_pool" to obtain mr, during mr allocation requests.

4. If you still insist, we can modify the patch to use  smp_load_acquire()  and smp_store_release().

Regards,
Praveen.