Message-ID: <1551504025-3541-4-git-send-email-skomatineni@nvidia.com>
Date:   Fri, 1 Mar 2019 21:20:18 -0800
From:   Sowjanya Komatineni <skomatineni@...dia.com>
To:     <adrian.hunter@...el.com>, <ulf.hansson@...aro.org>,
        <robh+dt@...nel.org>, <mark.rutland@....com>,
        <riteshh@...eaurora.org>
CC:     <thierry.reding@...il.com>, <jonathanh@...dia.com>,
        <anrao@...dia.com>, <linux-tegra@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>, <linux-mmc@...r.kernel.org>,
        <devicetree@...r.kernel.org>,
        Sowjanya Komatineni <skomatineni@...dia.com>
Subject: [PATCH V1 04/11] mmc: tegra: update hw tuning process

This patch includes below HW tuning related fixes.
- configures tuning parameters as per Tegra TRM
- WAR fix for manual tap change
- HW auto-tuning post process

As per Tegra TRM, SDR50 mode tuning execution takes upto maximum
of 256 tuning iterations and SDR104/HS200/HS400 modes tuning
execution takes upto maximum of 128 tuning iterations.

This patch programs tuning control register with maximum tuning
iterations needed based on the timing along with the start tap,
multiplier, and step size used by the HW tuning.

Tegra210 has a known issue of glitch on trimmer output when the
tap value is changed with the trimmer input clock running and the
WAR is to disable card clock before sending tuning command and
after sending tuning command wait for 1usec and issue SW reset
followed by enabling card clock.

This WAR is applicable when changing tap value manually as well.
Tegra SDHCI driver has this implemented correctly for manual tap
change but missing SW reset before enabling card clock during
sending tuning command.

Issuing SW reset during tuning command as a part of WAR and is
applicable in cases where tuning is performed with single step size
for more iterations. This patch includes this fix.

HW auto-tuning finds the best largest passing window and sets the
tap at the middle of the window. With some devices like sandisk
eMMC driving fast edges and due to high tap to tap delay in the
Tegra chipset, auto-tuning does not detect falling tap between the
valid windows resulting in a parital window or a merged window and
the best tap is set at the signal transition which is actually the
worst tap location.

Recommended SW solution is to detect if the best passing window
picked by the HW tuning is a partial or a merged window based on
min and max tap delays found from chip characterization across
PVT and perform tuning correction to pick the best tap.

This patch has this implemention of post tuning and uses tuned tap
delay.

Signed-off-by: Sowjanya Komatineni <skomatineni@...dia.com>
---
 drivers/mmc/host/sdhci-tegra.c | 218 ++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 217 insertions(+), 1 deletion(-)

diff --git a/drivers/mmc/host/sdhci-tegra.c b/drivers/mmc/host/sdhci-tegra.c
index 46086dd43bfb..2086e0eced88 100644
--- a/drivers/mmc/host/sdhci-tegra.c
+++ b/drivers/mmc/host/sdhci-tegra.c
@@ -66,6 +66,23 @@
 
 #define SDHCI_VNDR_TUN_CTRL0_0				0x1c0
 #define SDHCI_VNDR_TUN_CTRL0_TUN_HW_TAP			0x20000
+#define SDHCI_VNDR_TUN_CTRL0_START_TAP_VAL_MASK		0x03fc0000
+#define SDHCI_VNDR_TUN_CTRL0_START_TAP_VAL_SHIFT	18
+#define SDHCI_VNDR_TUN_CTRL0_MUL_M_MASK			0x00001fc0
+#define SDHCI_VNDR_TUN_CTRL0_MUL_M_SHIFT		6
+#define SDHCI_VNDR_TUN_CTRL0_TUN_ITER_MASK		0x000e000
+#define SDHCI_VNDR_TUN_CTRL0_TUN_ITER_SHIFT		13
+#define TRIES_40					0
+#define TRIES_128					2
+#define TRIES_256					4
+#define SDHCI_VNDR_TUN_CTRL0_TUN_WORD_SEL_MASK		0x7
+
+#define SDHCI_TEGRA_VNDR_TUN_CTRL1_0			0x1c4
+#define SDHCI_TEGRA_VNDR_TUN_STATUS0			0x1C8
+#define SDHCI_TEGRA_VNDR_TUN_STATUS1			0x1CC
+#define SDHCI_TEGRA_VNDR_TUN_STATUS1_TAP_MASK		0xFF
+#define SDHCI_TEGRA_VNDR_TUN_STATUS1_END_TAP_SHIFT	0x8
+#define TUNING_WORD_BIT_SIZE				32
 
 #define SDHCI_TEGRA_AUTO_CAL_CONFIG			0x1e4
 #define SDHCI_AUTO_CAL_START				BIT(31)
@@ -97,6 +114,8 @@
 struct sdhci_tegra_soc_data {
 	const struct sdhci_pltfm_data *pdata;
 	u32 nvquirks;
+	u8 min_tap_delay;
+	u8 max_tap_delay;
 };
 
 /* Magic pull up and pull down pad calibration offsets */
@@ -136,6 +155,8 @@ struct sdhci_tegra {
 	u32 default_trim;
 	u32 dqs_trim;
 	bool enable_hwcq;
+	unsigned long curr_clk_rate;
+	u8 tuned_tap_delay;
 };
 
 static u16 tegra_sdhci_readw(struct sdhci_host *host, int reg)
@@ -241,6 +262,7 @@ static void tegra210_sdhci_writew(struct sdhci_host *host, u16 val, int reg)
 
 	if (is_tuning_cmd) {
 		udelay(1);
+		sdhci_reset(host, SDHCI_RESET_CMD | SDHCI_RESET_DATA);
 		tegra_sdhci_configure_card_clk(host, clk_enabled);
 	}
 }
@@ -722,6 +744,7 @@ static void tegra_sdhci_set_clock(struct sdhci_host *host, unsigned int clock)
 	 */
 	host_clk = tegra_host->ddr_signaling ? clock * 2 : clock;
 	clk_set_rate(pltfm_host->clk, host_clk);
+	tegra_host->curr_clk_rate = host_clk;
 	if (tegra_host->ddr_signaling)
 		host->max_clk = host_clk;
 	else
@@ -770,6 +793,162 @@ static void tegra_sdhci_hs400_dll_cal(struct sdhci_host *host)
 			"HS400 delay line calibration timed out\n");
 }
 
+static int tegra_sdhci_get_max_tuning_loop(struct sdhci_host *host)
+{
+	u32 val;
+
+	val = sdhci_readl(host, SDHCI_VNDR_TUN_CTRL0_0);
+	val &= SDHCI_VNDR_TUN_CTRL0_TUN_ITER_MASK;
+	val = val >> SDHCI_VNDR_TUN_CTRL0_TUN_ITER_SHIFT;
+	if (val == TRIES_128)
+		return 128;
+	else if (val == TRIES_256)
+		return 256;
+	else
+		return 40;
+}
+
+static void tegra_sdhci_tap_correction(struct sdhci_host *host, u8 num_iter,
+				       u8 thd_up, u8 thd_low, u8 fixed_tap)
+{
+	struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host);
+	struct sdhci_tegra *tegra_host = sdhci_pltfm_priv(pltfm_host);
+	u32 val, tun_status;
+	u8 word, bit, edge1, tap, window;
+	bool tap_result;
+	bool start_fail = false;
+	bool start_pass = false;
+	bool end_pass = false;
+	bool first_fail = false;
+	bool first_pass = false;
+	u8 start_pass_tap = 0;
+	u8 end_pass_tap = 0;
+	u8 first_fail_tap = 0;
+	u8 first_pass_tap = 0;
+	u8 total_tuning_words = num_iter / TUNING_WORD_BIT_SIZE;
+
+	/*
+	 * Read auto-tuned results and extract good valid passing window by
+	 * filtering out un-wanted bubble/partial/merged windows.
+	 */
+	for (word = 0; word < total_tuning_words; word++) {
+		val = sdhci_readl(host, SDHCI_VNDR_TUN_CTRL0_0);
+		val &= ~SDHCI_VNDR_TUN_CTRL0_TUN_WORD_SEL_MASK;
+		val |= word;
+		sdhci_writel(host, val, SDHCI_VNDR_TUN_CTRL0_0);
+		tun_status = sdhci_readl(host, SDHCI_TEGRA_VNDR_TUN_STATUS0);
+		bit = 0;
+		while (bit < TUNING_WORD_BIT_SIZE) {
+			tap = word * TUNING_WORD_BIT_SIZE + bit;
+			tap_result = tun_status & (1 << bit);
+			if (!tap_result && !start_fail) {
+				start_fail = true;
+				if (!first_fail) {
+					first_fail_tap = tap;
+					first_fail = true;
+				}
+
+			} else if (tap_result && start_fail && !start_pass) {
+				start_pass_tap = tap;
+				start_pass = true;
+				if (!first_pass) {
+					first_pass_tap = tap;
+					first_pass = true;
+				}
+
+			} else if (!tap_result && start_fail && start_pass &&
+				   !end_pass) {
+				end_pass_tap = tap - 1;
+				end_pass = true;
+			} else if (tap_result && start_pass && start_fail &&
+				   end_pass) {
+				window = end_pass_tap - start_pass_tap;
+				/* discard merged window and bubble window */
+				if (window >= thd_up || window < thd_low) {
+					start_pass_tap = tap;
+					end_pass = false;
+				} else {
+					/* set tap at middle of valid window */
+					tap = start_pass_tap + window / 2;
+					tegra_host->tuned_tap_delay = tap;
+					return;
+				}
+			}
+
+			bit++;
+		}
+	}
+
+	if (!first_fail) {
+		WARN_ON("no edge detected, continue with hw tuned delay.\n");
+	} else if (first_pass) {
+		/* set tap location at fixed tap relative to the first edge */
+		edge1 = first_fail_tap + (first_pass_tap - first_fail_tap) / 2;
+		if (edge1 - 1 > fixed_tap)
+			tegra_host->tuned_tap_delay = edge1 - fixed_tap;
+		else
+			tegra_host->tuned_tap_delay = edge1 + fixed_tap;
+	}
+}
+
+static void tegra_sdhci_post_tuning(struct sdhci_host *host)
+{
+	struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host);
+	struct sdhci_tegra *tegra_host = sdhci_pltfm_priv(pltfm_host);
+	const struct sdhci_tegra_soc_data *soc_data = tegra_host->soc_data;
+	u32 avg_tap_dly, val, min_tap_dly, max_tap_dly;
+	u8 fixed_tap, start_tap, end_tap, window_width;
+	u8 thdupper, thdlower;
+	u8 num_iter;
+	u32 clk_rate_mhz, period_ps, bestcase, worstcase;
+
+	/* retain HW tuned tap to use incase if no correction is needed */
+	val = sdhci_readl(host, SDHCI_TEGRA_VENDOR_CLOCK_CTRL);
+	tegra_host->tuned_tap_delay = (val & SDHCI_CLOCK_CTRL_TAP_MASK) >>
+				      SDHCI_CLOCK_CTRL_TAP_SHIFT;
+	if (soc_data->min_tap_delay && soc_data->max_tap_delay) {
+		min_tap_dly = soc_data->min_tap_delay;
+		max_tap_dly = soc_data->max_tap_delay;
+		clk_rate_mhz = tegra_host->curr_clk_rate / USEC_PER_SEC;
+		period_ps = USEC_PER_SEC / clk_rate_mhz;
+		bestcase = period_ps / min_tap_dly;
+		worstcase = period_ps / max_tap_dly;
+		/*
+		 * Upper and Lower bound thresholds used to detect merged and
+		 * bubble windows
+		 */
+		thdupper = (2 * worstcase + bestcase) / 2;
+		thdlower = worstcase / 4;
+		/*
+		 * fixed tap is used when HW tuning result contains single edge
+		 * and tap is set at fixed tap delay relative to the first edge
+		 */
+		avg_tap_dly = (period_ps * 2) / (min_tap_dly + max_tap_dly);
+		fixed_tap = avg_tap_dly / 2;
+
+		val = sdhci_readl(host, SDHCI_TEGRA_VNDR_TUN_STATUS1);
+		start_tap = val & SDHCI_TEGRA_VNDR_TUN_STATUS1_TAP_MASK;
+		end_tap = (val >> SDHCI_TEGRA_VNDR_TUN_STATUS1_END_TAP_SHIFT) &
+			  SDHCI_TEGRA_VNDR_TUN_STATUS1_TAP_MASK;
+		window_width = end_tap - start_tap;
+		num_iter = tegra_sdhci_get_max_tuning_loop(host);
+		/*
+		 * partial window includes edges of the tuning range.
+		 * merged window includes more taps so window width is higher
+		 * than upper threshold.
+		 */
+		if (start_tap == 0 || (end_tap == (num_iter - 1)) ||
+		    (end_tap == num_iter - 2) || window_width >= thdupper) {
+			pr_debug("%s: Apply tuning correction\n",
+				 mmc_hostname(host->mmc));
+			tegra_sdhci_tap_correction(host, num_iter, thdupper,
+						   thdlower, fixed_tap);
+		}
+	}
+
+	tegra_sdhci_set_tap(host, tegra_host->tuned_tap_delay);
+}
+
 static void tegra_sdhci_set_uhs_signaling(struct sdhci_host *host,
 					  unsigned timing)
 {
@@ -778,17 +957,22 @@ static void tegra_sdhci_set_uhs_signaling(struct sdhci_host *host,
 	bool set_default_tap = false;
 	bool set_dqs_trim = false;
 	bool do_hs400_dll_cal = false;
+	u8 iter = TRIES_256;
+	u32 val;
 
 	tegra_host->ddr_signaling = false;
 	switch (timing) {
 	case MMC_TIMING_UHS_SDR50:
+		break;
 	case MMC_TIMING_UHS_SDR104:
 	case MMC_TIMING_MMC_HS200:
 		/* Don't set default tap on tunable modes. */
+		iter = TRIES_128;
 		break;
 	case MMC_TIMING_MMC_HS400:
 		set_dqs_trim = true;
 		do_hs400_dll_cal = true;
+		iter = TRIES_128;
 		break;
 	case MMC_TIMING_MMC_DDR52:
 	case MMC_TIMING_UHS_DDR50:
@@ -800,11 +984,23 @@ static void tegra_sdhci_set_uhs_signaling(struct sdhci_host *host,
 		break;
 	}
 
+	val = sdhci_readl(host, SDHCI_VNDR_TUN_CTRL0_0);
+	val &= ~(SDHCI_VNDR_TUN_CTRL0_TUN_ITER_MASK |
+		 SDHCI_VNDR_TUN_CTRL0_START_TAP_VAL_MASK |
+		 SDHCI_VNDR_TUN_CTRL0_MUL_M_MASK);
+	val |= (iter << SDHCI_VNDR_TUN_CTRL0_TUN_ITER_SHIFT |
+		0 << SDHCI_VNDR_TUN_CTRL0_START_TAP_VAL_SHIFT |
+		1 << SDHCI_VNDR_TUN_CTRL0_MUL_M_SHIFT);
+	sdhci_writel(host, val, SDHCI_VNDR_TUN_CTRL0_0);
+	sdhci_writel(host, 0, SDHCI_TEGRA_VNDR_TUN_CTRL1_0);
+
 	sdhci_set_uhs_signaling(host, timing);
 
 	tegra_sdhci_pad_autocalib(host);
 
-	if (set_default_tap)
+	if (tegra_host->tuned_tap_delay && !set_default_tap)
+		tegra_sdhci_set_tap(host, tegra_host->tuned_tap_delay);
+	else
 		tegra_sdhci_set_tap(host, tegra_host->default_tap);
 
 	if (set_dqs_trim)
@@ -1090,6 +1286,8 @@ static const struct sdhci_ops tegra210_sdhci_ops = {
 	.set_uhs_signaling = tegra_sdhci_set_uhs_signaling,
 	.voltage_switch = tegra_sdhci_voltage_switch,
 	.get_max_clock = tegra_sdhci_get_max_clock,
+	.get_max_tuning_loop_count = tegra_sdhci_get_max_tuning_loop,
+	.post_tuning = tegra_sdhci_post_tuning,
 };
 
 static const struct sdhci_pltfm_data sdhci_tegra210_pdata = {
@@ -1110,6 +1308,8 @@ static const struct sdhci_tegra_soc_data soc_data_tegra210 = {
 		    NVQUIRK_DIS_CARD_CLK_CONFIG_TAP |
 		    NVQUIRK_ENABLE_SDR50 |
 		    NVQUIRK_ENABLE_SDR104,
+	.min_tap_delay = 106,
+	.max_tap_delay = 185,
 };
 
 static const struct sdhci_ops tegra186_sdhci_ops = {
@@ -1121,6 +1321,8 @@ static const struct sdhci_ops tegra186_sdhci_ops = {
 	.set_uhs_signaling = tegra_sdhci_set_uhs_signaling,
 	.voltage_switch = tegra_sdhci_voltage_switch,
 	.get_max_clock = tegra_sdhci_get_max_clock,
+	.get_max_tuning_loop_count = tegra_sdhci_get_max_tuning_loop,
+	.post_tuning = tegra_sdhci_post_tuning,
 	.irq = sdhci_tegra_cqhci_irq,
 };
 
@@ -1150,9 +1352,23 @@ static const struct sdhci_tegra_soc_data soc_data_tegra186 = {
 		    NVQUIRK_DIS_CARD_CLK_CONFIG_TAP |
 		    NVQUIRK_ENABLE_SDR50 |
 		    NVQUIRK_ENABLE_SDR104,
+	.min_tap_delay = 84,
+	.max_tap_delay = 136,
+};
+
+static const struct sdhci_tegra_soc_data soc_data_tegra194 = {
+	.pdata = &sdhci_tegra186_pdata,
+	.nvquirks = NVQUIRK_NEEDS_PAD_CONTROL |
+		    NVQUIRK_HAS_PADCALIB |
+		    NVQUIRK_DIS_CARD_CLK_CONFIG_TAP |
+		    NVQUIRK_ENABLE_SDR50 |
+		    NVQUIRK_ENABLE_SDR104,
+	.min_tap_delay = 96,
+	.max_tap_delay = 139,
 };
 
 static const struct of_device_id sdhci_tegra_dt_match[] = {
+	{ .compatible = "nvidia,tegra194-sdhci", .data = &soc_data_tegra194 },
 	{ .compatible = "nvidia,tegra186-sdhci", .data = &soc_data_tegra186 },
 	{ .compatible = "nvidia,tegra210-sdhci", .data = &soc_data_tegra210 },
 	{ .compatible = "nvidia,tegra124-sdhci", .data = &soc_data_tegra124 },
-- 
2.7.4

Powered by blists - more mailing lists