Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit e9495c70 authored by Subhash Jadavani's avatar Subhash Jadavani Committed by Gerrit - the friendly Code Review server
Browse files

mmc: block: serialize the requests if we are scaled down 



When in SVS2 on low load scenario and there are lots of requests
queued for CMDQ we need to wait till the queue is empty to scale
back up to Nominal even if there is a sudden increase in load.
This impacts performance where lots of IO get executed in SVS2
frequency since the queue is full. As SVS2 is a low load use case
we can serialize the requests and not queue them in parallel
without impacting other use cases. This makes sure the queue gets
empty faster and we will be able to scale up to Nominal frequency
when needed.

Change-Id: Idbe7e939e01327061dfa5de93c0eaed59b910592
Signed-off-by: default avatarSubhash Jadavani <subhashj@codeaurora.org>
Signed-off-by: default avatarVenkat Gopalakrishnan <venkatg@codeaurora.org>
parent 83ff89ab

drivers/mmc/card/block.c

+15 −0
Original line number Diff line number Diff line
@@ -2807,6 +2807,7 @@ static int mmc_blk_cmdq_issue_rw_rq(struct mmc_queue *mq, struct request *req) 
	struct mmc_queue_req *active_mqrq;

	struct mmc_card *card = mq->card;

	struct mmc_host *host = card->host;

	struct mmc_cmdq_context_info *ctx = &host->cmdq_ctx;

	struct mmc_cmdq_req *mc_rq;

	int ret = 0;
@@ -2831,6 +2832,20 @@ static int mmc_blk_cmdq_issue_rw_rq(struct mmc_queue *mq, struct request *req) 
		    && (rq_data_dir(req) == READ))

			host->cmdq_ctx.active_small_sector_read_reqs++;

	}

	/*

	 * When in SVS2 on low load scenario and there are lots of requests

	 * queued for CMDQ we need to wait till the queue is empty to scale

	 * back up to Nominal even if there is a sudden increase in load.

	 * This impacts performance where lots of IO get executed in SVS2

	 * frequency since the queue is full. As SVS2 is a low load use case

	 * we can serialize the requests and not queue them in parallel

	 * without impacting other use cases. This makes sure the queue gets

	 * empty faster and we will be able to scale up to Nominal frequency

	 * when needed.

	 */

	if (!ret && (host->clk_scaling.state == MMC_LOAD_LOW))

		wait_event_interruptible(ctx->queue_empty_wq,

					(!ctx->active_reqs));



	return ret;

}