From 7df1d3b44368209517fc7dc0b35ee2621a94722e Mon Sep 17 00:00:00 2001
From: Jonathan Bell <jonathan@raspberrypi.com>
Date: Thu, 18 Jan 2024 14:32:16 +0000
Subject: [PATCH] mmc: bcm2835-sdhost: use Host Software Queueing mechanism

See commit 511ce378e16f ("mmc: Add MMC host software queue support")

Introduced in 5.8, this feature lets the block layer issue up to 2
pending requests to the MMC layer which in certain cases can improve
throughput, but in the case of this driver can significantly reduce
context switching when performing random reads.

On bcm2837 with a performant class A1 card, context switches under FIO
random 4k reads go from ~8800 per second to ~5800, with a reduction in
hardIRQs per second from ~5800 to ~4000. There is no appreciable
difference in throughput.

For bcm2835, and for workloads other than random read, HSQ is a wash in
terms of throughput and CPU load.

So, use it by default.

Signed-off-by: Jonathan Bell <jonathan@raspberrypi.com>
---
 drivers/mmc/host/Kconfig          |  1 +
 drivers/mmc/host/bcm2835-sdhost.c | 27 ++++++++++++++++++++++-----
 2 files changed, 23 insertions(+), 5 deletions(-)

diff --git a/drivers/mmc/host/Kconfig b/drivers/mmc/host/Kconfig
index bb63a9908d8d3d..bc220f028191af 100644
--- a/drivers/mmc/host/Kconfig
+++ b/drivers/mmc/host/Kconfig
@@ -37,6 +37,7 @@ config MMC_BCM2835_PIO_DMA_BARRIER
 config MMC_BCM2835_SDHOST
 	tristate "Support for the SDHost controller on BCM2708/9"
 	depends on ARCH_BCM2835
+	select MMC_HSQ
 	help
 	  This selects the SDHost controller on BCM2835/6.
 
diff --git a/drivers/mmc/host/bcm2835-sdhost.c b/drivers/mmc/host/bcm2835-sdhost.c
index 40eb557e68fd35..8e95a9df2f8812 100644
--- a/drivers/mmc/host/bcm2835-sdhost.c
+++ b/drivers/mmc/host/bcm2835-sdhost.c
@@ -57,6 +57,7 @@
 
 /* For mmc_card_blockaddr */
 #include "../core/card.h"
+#include "mmc_hsq.h"
 
 #define DRIVER_NAME "sdhost-bcm2835"
 
@@ -1891,13 +1892,16 @@ static void bcm2835_sdhost_tasklet_finish(unsigned long param)
 				mmc_hostname(host->mmc));
 	}
 
-	mmc_request_done(host->mmc, mrq);
+	if (!mmc_hsq_finalize_request(host->mmc, mrq))
+		mmc_request_done(host->mmc, mrq);
 	log_event("TSK>", mrq, 0);
 }
 
-int bcm2835_sdhost_add_host(struct bcm2835_host *host)
+int bcm2835_sdhost_add_host(struct platform_device *pdev)
 {
+	struct bcm2835_host *host = platform_get_drvdata(pdev);
 	struct mmc_host *mmc;
+	struct mmc_hsq *hsq;
 	struct dma_slave_config cfg;
 	char pio_limit_string[20];
 	int ret;
@@ -1992,6 +1996,16 @@ int bcm2835_sdhost_add_host(struct bcm2835_host *host)
 		goto untasklet;
 	}
 
+	hsq = devm_kzalloc(&pdev->dev, sizeof(*hsq), GFP_KERNEL);
+	if (!hsq) {
+		ret = -ENOMEM;
+		goto free_irq;
+	}
+
+	ret = mmc_hsq_init(hsq, host->mmc);
+	if (ret)
+		goto free_irq;
+
 	mmc_add_host(mmc);
 
 	pio_limit_string[0] = '\0';
@@ -2004,6 +2018,9 @@ int bcm2835_sdhost_add_host(struct bcm2835_host *host)
 
 	return 0;
 
+free_irq:
+	free_irq(host->irq, host);
+
 untasklet:
 	tasklet_kill(&host->finish_tasklet);
 
@@ -2134,12 +2151,12 @@ static int bcm2835_sdhost_probe(struct platform_device *pdev)
 
 	host->firmware_sets_cdiv = (msg[1] != ~0);
 
-	ret = bcm2835_sdhost_add_host(host);
+	platform_set_drvdata(pdev, host);
+
+	ret = bcm2835_sdhost_add_host(pdev);
 	if (ret)
 		goto err;
 
-	platform_set_drvdata(pdev, host);
-
 	pr_debug("bcm2835_sdhost_probe -> OK\n");
 
 	return 0;