[PATCH v3] mmc: mtk-sd: reduce CIT for better performance

[Wenbin Mei]

CQHCI_SSC1 indicates to CQE the polling period to use when using periodic
SEND_QUEUE_STATUS(CMD13) polling.
Since MSDC CQE uses msdc_hclk as ITCFVAL, so driver should use hclk
frequency to get the actual time.
The default value 0x1000 that corresponds to 150us for MediaTek SoCs, let's
decrease it to 0x40 that corresponds to 2.35us, which can improve the
performance of some eMMC devices.

Signed-off-by: Wenbin Mei <wenbin.mei@xxxxxxxxxxxx>
---
 drivers/mmc/host/cqhci.h  |  1 +
 drivers/mmc/host/mtk-sd.c | 47 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 48 insertions(+)

diff --git a/drivers/mmc/host/cqhci.h b/drivers/mmc/host/cqhci.h
index ba9387ed90eb..292b89ebd978 100644
--- a/drivers/mmc/host/cqhci.h
+++ b/drivers/mmc/host/cqhci.h
@@ -23,6 +23,7 @@
 /* capabilities */
 #define CQHCI_CAP			0x04
 #define CQHCI_CAP_CS			0x10000000 /* Crypto Support */
+#define CQHCI_CAP_ITCFMUL(x)		(((x) & GENMASK(15, 12)) >> 12)
 
 /* configuration */
 #define CQHCI_CFG			0x08
diff --git a/drivers/mmc/host/mtk-sd.c b/drivers/mmc/host/mtk-sd.c
index edade0e54a0c..c221ef8a6992 100644
--- a/drivers/mmc/host/mtk-sd.c
+++ b/drivers/mmc/host/mtk-sd.c
@@ -473,6 +473,7 @@ struct msdc_host {
 	struct msdc_tune_para def_tune_para; /* default tune setting */
 	struct msdc_tune_para saved_tune_para; /* tune result of CMD21/CMD19 */
 	struct cqhci_host *cq_host;
+	u32 cq_ssc1_time;
 };
 
 static const struct mtk_mmc_compatible mt2701_compat = {
@@ -2450,9 +2451,50 @@ static void msdc_hs400_enhanced_strobe(struct mmc_host *mmc,
 	}
 }
 
+static void msdc_cqe_cit_cal(struct msdc_host *host, u64 timer_ns)
+{
+	struct mmc_host *mmc = mmc_from_priv(host);
+	struct cqhci_host *cq_host = mmc->cqe_private;
+	u8 itcfmul;
+	u32 hclk_freq;
+	u64 value;
+
+	/* Since MSDC CQE uses msdc_hclk as ITCFVAL, so driver should use hclk
+	 * frequency to get the actual time for CIT.
+	 */
+	if (host->h_clk) {
+		hclk_freq = clk_get_rate(host->h_clk);
+		itcfmul = CQHCI_CAP_ITCFMUL(cqhci_readl(cq_host, CQHCI_CAP));
+		switch (itcfmul) {
+		case 0x0:
+			do_div(hclk_freq, 1000);
+			break;
+		case 0x1:
+			do_div(hclk_freq, 100);
+			break;
+		case 0x2:
+			do_div(hclk_freq, 10);
+			break;
+		case 0x3:
+			break;
+		case 0x4:
+			hclk_freq = hclk_freq * 10;
+			break;
+		default:
+			host->cq_ssc1_time = 0x40;
+			return;
+		value = hclk_freq * timer_ns;
+		do_div(value, 1000000000ULL);
+		host->cq_ssc1_time = value;
+	} else {
+		host->cq_ssc1_time = 0x40;
+	}
+}
+
 static void msdc_cqe_enable(struct mmc_host *mmc)
 {
 	struct msdc_host *host = mmc_priv(mmc);
+	struct cqhci_host *cq_host = mmc->cqe_private;
 
 	/* enable cmdq irq */
 	writel(MSDC_INT_CMDQ, host->base + MSDC_INTEN);
@@ -2462,6 +2504,9 @@ static void msdc_cqe_enable(struct mmc_host *mmc)
 	msdc_set_busy_timeout(host, 20 * 1000000000ULL, 0);
 	/* default read data timeout 1s */
 	msdc_set_timeout(host, 1000000000ULL, 0);
+
+	/* Set the send status command idle timer */
+	cqhci_writel(cq_host, host->cq_ssc1_time, CQHCI_SSC1);
 }
 
 static void msdc_cqe_disable(struct mmc_host *mmc, bool recovery)
@@ -2803,6 +2848,8 @@ static int msdc_drv_probe(struct platform_device *pdev)
 		/* cqhci 16bit length */
 		/* 0 size, means 65536 so we don't have to -1 here */
 		mmc->max_seg_size = 64 * 1024;
+		/* Reduce CIT to 0x40 that corresponds to 2.35us */
+		msdc_cqe_cit_cal(host, 2350);
 	}
 
 	ret = devm_request_irq(&pdev->dev, host->irq, msdc_irq,
-- 
2.25.1