diff mbox

[21/32] mmc: bcm2835-sdhost: Add new driver for the internal SD controller.

Message ID 1464817421-8519-22-git-send-email-kraxel@redhat.com (mailing list archive)
State New, archived
Headers show

Commit Message

Gerd Hoffmann June 1, 2016, 9:43 p.m. UTC
From: Eric Anholt <eric@anholt.net>

The 2835 has two SD controllers: The Arasan SDHCI controller that we
currently use, and a custom SD controller.  The custom one runs faster

The code was originally written by Phil Elwell in the downstream
Rasbperry Pi tree, and I did a major cleanup on it (+319, -707 lines
out of the original 2055) for inclusion.

Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/mmc/host/Kconfig          |    9 +
 drivers/mmc/host/Makefile         |    1 +
 drivers/mmc/host/bcm2835-sdhost.c | 1666 +++++++++++++++++++++++++++++++++++++
 3 files changed, 1676 insertions(+)
 create mode 100644 drivers/mmc/host/bcm2835-sdhost.c

Comments

Arnd Bergmann June 1, 2016, 10:18 p.m. UTC | #1
On Wednesday, June 1, 2016 11:43:30 PM CEST Gerd Hoffmann wrote:

> +static void bcm2835_sdhost_reset_internal(struct bcm2835_host *host)
> +{
> +	u32 temp;
> +
> +	bcm2835_sdhost_set_power(host, false);
> +
> +	bcm2835_sdhost_write(host, 0, SDCMD);
> +	bcm2835_sdhost_write(host, 0, SDARG);
> +	bcm2835_sdhost_write(host, 0xf00000, SDTOUT);
> +	bcm2835_sdhost_write(host, 0, SDCDIV);
> +	bcm2835_sdhost_write(host, 0x7f8, SDHSTS); /* Write 1s to clear */
> +	bcm2835_sdhost_write(host, 0, SDHCFG);
> +	bcm2835_sdhost_write(host, 0, SDHBCT);
> +	bcm2835_sdhost_write(host, 0, SDHBLC);
> +
> +	/* Limit fifo usage due to silicon bug */
> +	temp = bcm2835_sdhost_read(host, SDEDM);
> +	temp &= ~((SDEDM_THRESHOLD_MASK << SDEDM_READ_THRESHOLD_SHIFT) |
> +		  (SDEDM_THRESHOLD_MASK << SDEDM_WRITE_THRESHOLD_SHIFT));
> +	temp |= (FIFO_READ_THRESHOLD << SDEDM_READ_THRESHOLD_SHIFT) |
> +		(FIFO_WRITE_THRESHOLD << SDEDM_WRITE_THRESHOLD_SHIFT);
> +	bcm2835_sdhost_write(host, temp, SDEDM);
> +	mdelay(10);
> +	bcm2835_sdhost_set_power(host, true);
> +	mdelay(10);

Are you sure you cannot use msleep() here?

> +	host->clock = 0;
> +	bcm2835_sdhost_write(host, host->hcfg, SDHCFG);
> +	bcm2835_sdhost_write(host, host->cdiv, SDCDIV);
> +	mmiowb();
> +}

What is the barrier for? Same question for all the other instances

> +
> +static void bcm2835_sdhost_reset(struct mmc_host *mmc)
> +{
> +	struct bcm2835_host *host = mmc_priv(mmc);
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&host->lock, flags);
> +	bcm2835_sdhost_reset_internal(host);
> +	spin_unlock_irqrestore(&host->lock, flags);
> +}

The spinlock seems to only protect the reset from happening concurrently
with bcm2835_sdhost_dma_complete() here. So if you ensure that this
function is no longer pending, you can probably use msleep() above.

> +static void bcm2835_sdhost_set_ios(struct mmc_host *mmc, struct mmc_ios *ios);

Maybe rearrange the order of the functions to avoid forward declarations?

> +	spin_lock_init(&host->lock);
> +
> +	if (IS_ERR_OR_NULL(host->dma_chan_tx) ||
> +	    IS_ERR_OR_NULL(host->dma_chan_rx)) {
> +		pr_err("%s: unable to initialise DMA channels. Falling back to PIO\n",
> +		       mmc_hostname(mmc));

When are these ever NULL?

> +	/* Parse OF address directly to get the physical address for
> +	 * DMA to our registers.
> +	 */
> +	host->phys_addr = be32_to_cpup(of_get_address(pdev->dev.of_node, 0,
> +						      NULL, NULL));

This looks broken on 64-bit systems when #address-cells=<2>, or if there
is an intermediate bus with a non-empty ranges property.

What's wrong with platform_get_resource(pdev, IORESOURCE_MEM, 0)?

	Arnd
Eric Anholt June 2, 2016, 6:12 p.m. UTC | #2
Arnd Bergmann <arnd@arndb.de> writes:

> On Wednesday, June 1, 2016 11:43:30 PM CEST Gerd Hoffmann wrote:
>> +	/* Parse OF address directly to get the physical address for
>> +	 * DMA to our registers.
>> +	 */
>> +	host->phys_addr = be32_to_cpup(of_get_address(pdev->dev.of_node, 0,
>> +						      NULL, NULL));
>
> This looks broken on 64-bit systems when #address-cells=<2>, or if there
> is an intermediate bus with a non-empty ranges property.
>
> What's wrong with platform_get_resource(pdev, IORESOURCE_MEM, 0)?

I'll get to the rest of the review later, but this is a weird pattern
that we have in several bcm2835-related drivers.

We need the address as seen by the DMA controller, not the address from
the ARM's perspective (which is offset by the dma-ranges DT property).
This is the way that people have figured out to get that address.
Arnd Bergmann June 3, 2016, 8:42 a.m. UTC | #3
On Thursday, June 2, 2016 11:12:38 AM CEST Eric Anholt wrote:
> Arnd Bergmann <arnd@arndb.de> writes:
> 
> > On Wednesday, June 1, 2016 11:43:30 PM CEST Gerd Hoffmann wrote:
> >> +    /* Parse OF address directly to get the physical address for
> >> +     * DMA to our registers.
> >> +     */
> >> +    host->phys_addr = be32_to_cpup(of_get_address(pdev->dev.of_node, 0,
> >> +                                                  NULL, NULL));
> >
> > This looks broken on 64-bit systems when #address-cells=<2>, or if there
> > is an intermediate bus with a non-empty ranges property.
> >
> > What's wrong with platform_get_resource(pdev, IORESOURCE_MEM, 0)?
> 
> I'll get to the rest of the review later, but this is a weird pattern
> that we have in several bcm2835-related drivers.
> 
> We need the address as seen by the DMA controller, not the address from
> the ARM's perspective (which is offset by the dma-ranges DT property).
> This is the way that people have figured out to get that address.

I see. This is indeed a bit tricky, as there is no easy way to know
how the dmaengine is wired up to the slave. A correct solution is
probably to follow the 'dma-ranges' up from the master to the common
parent and then follow the 'ranges' back to the slave, but that
requires coming up with a new exported function.

It's perhaps good enough in the meantime to read out the address from
the slave directly, but you should not assume that the first
cell has the complete information and instead should at least
evaluate #address-cells.

	Arnd
diff mbox

Patch

diff --git a/drivers/mmc/host/Kconfig b/drivers/mmc/host/Kconfig
index 0aa484c..67efe28 100644
--- a/drivers/mmc/host/Kconfig
+++ b/drivers/mmc/host/Kconfig
@@ -779,6 +779,15 @@  config MMC_TOSHIBA_PCI
 	depends on PCI
 	help
 
+config MMC_BCM2835_SDHOST
+	tristate "platform support for the BCM2835 SDHOST MMC Controller"
+	depends on ARCH_BCM2835 || COMPILE_TEST
+	help
+	  This selects the BCM2835 SDHOST MMC controller. If you have a BCM2835
+	  platform with SD or MMC devices, say Y or M here.
+
+	  If unsure, say N.
+
 config MMC_MTK
 	tristate "MediaTek SD/MMC Card Interface support"
 	depends on HAS_DMA
diff --git a/drivers/mmc/host/Makefile b/drivers/mmc/host/Makefile
index af918d2..1e1b42b 100644
--- a/drivers/mmc/host/Makefile
+++ b/drivers/mmc/host/Makefile
@@ -57,6 +57,7 @@  obj-$(CONFIG_MMC_MOXART)	+= moxart-mmc.o
 obj-$(CONFIG_MMC_SUNXI)		+= sunxi-mmc.o
 obj-$(CONFIG_MMC_USDHI6ROL0)	+= usdhi6rol0.o
 obj-$(CONFIG_MMC_TOSHIBA_PCI)	+= toshsd.o
+obj-$(CONFIG_MMC_BCM2835_SDHOST)	+= bcm2835-sdhost.o
 
 obj-$(CONFIG_MMC_REALTEK_PCI)	+= rtsx_pci_sdmmc.o
 obj-$(CONFIG_MMC_REALTEK_USB)	+= rtsx_usb_sdmmc.o
diff --git a/drivers/mmc/host/bcm2835-sdhost.c b/drivers/mmc/host/bcm2835-sdhost.c
new file mode 100644
index 0000000..cafe6ad
--- /dev/null
+++ b/drivers/mmc/host/bcm2835-sdhost.c
@@ -0,0 +1,1666 @@ 
+/*
+ * BCM2835 SD host driver.
+ *
+ * Author:      Phil Elwell <phil@raspberrypi.org>
+ *              Copyright (C) 2015-2016 Raspberry Pi (Trading) Ltd.
+ *
+ * Based on
+ *  mmc-bcm2835.c by Gellert Weisz
+ * which is, in turn, based on
+ *  sdhci-bcm2708.c by Broadcom
+ *  sdhci-bcm2835.c by Stephen Warren and Oleksandr Tymoshenko
+ *  sdhci.c and sdhci-pci.c by Pierre Ossman
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/delay.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/mmc/mmc.h>
+#include <linux/mmc/host.h>
+#include <linux/mmc/sd.h>
+#include <linux/scatterlist.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/clk.h>
+#include <linux/platform_device.h>
+#include <linux/err.h>
+#include <linux/blkdev.h>
+#include <linux/dmaengine.h>
+#include <linux/dma-mapping.h>
+#include <linux/of_dma.h>
+#include <linux/time.h>
+#include <linux/workqueue.h>
+
+#define SDCMD  0x00 /* Command to SD card              - 16 R/W */
+#define SDARG  0x04 /* Argument to SD card             - 32 R/W */
+#define SDTOUT 0x08 /* Start value for timeout counter - 32 R/W */
+#define SDCDIV 0x0c /* Start value for clock divider   - 11 R/W */
+#define SDRSP0 0x10 /* SD card response (31:0)         - 32 R   */
+#define SDRSP1 0x14 /* SD card response (63:32)        - 32 R   */
+#define SDRSP2 0x18 /* SD card response (95:64)        - 32 R   */
+#define SDRSP3 0x1c /* SD card response (127:96)       - 32 R   */
+#define SDHSTS 0x20 /* SD host status                  - 11 R   */
+#define SDVDD  0x30 /* SD card power control           -  1 R/W */
+#define SDEDM  0x34 /* Emergency Debug Mode            - 13 R/W */
+#define SDHCFG 0x38 /* Host configuration              -  2 R/W */
+#define SDHBCT 0x3c /* Host byte count (debug)         - 32 R/W */
+#define SDDATA 0x40 /* Data to/from SD card            - 32 R/W */
+#define SDHBLC 0x50 /* Host block count (SDIO/SDHC)    -  9 R/W */
+
+#define SDCMD_NEW_FLAG                  0x8000
+#define SDCMD_FAIL_FLAG                 0x4000
+#define SDCMD_BUSYWAIT                  0x800
+#define SDCMD_NO_RESPONSE               0x400
+#define SDCMD_LONG_RESPONSE             0x200
+#define SDCMD_WRITE_CMD                 0x80
+#define SDCMD_READ_CMD                  0x40
+#define SDCMD_CMD_MASK                  0x3f
+
+#define SDCDIV_MAX_CDIV                 0x7ff
+
+#define SDHSTS_BUSY_IRPT                0x400
+#define SDHSTS_BLOCK_IRPT               0x200
+#define SDHSTS_SDIO_IRPT                0x100
+#define SDHSTS_REW_TIME_OUT             0x80
+#define SDHSTS_CMD_TIME_OUT             0x40
+#define SDHSTS_CRC16_ERROR              0x20
+#define SDHSTS_CRC7_ERROR               0x10
+#define SDHSTS_FIFO_ERROR               0x08
+/* Reserved */
+/* Reserved */
+#define SDHSTS_DATA_FLAG                0x01
+
+#define SDHSTS_TRANSFER_ERROR_MASK      (SDHSTS_CRC7_ERROR | \
+					 SDHSTS_CRC16_ERROR | \
+					 SDHSTS_REW_TIME_OUT | \
+					 SDHSTS_FIFO_ERROR)
+
+#define SDHSTS_ERROR_MASK               (SDHSTS_CMD_TIME_OUT | \
+					 SDHSTS_TRANSFER_ERROR_MASK)
+
+#define SDHCFG_BUSY_IRPT_EN     BIT(10)
+#define SDHCFG_BLOCK_IRPT_EN    BIT(8)
+#define SDHCFG_SDIO_IRPT_EN     BIT(5)
+#define SDHCFG_DATA_IRPT_EN     BIT(4)
+#define SDHCFG_SLOW_CARD        BIT(3)
+#define SDHCFG_WIDE_EXT_BUS     BIT(2)
+#define SDHCFG_WIDE_INT_BUS     BIT(1)
+#define SDHCFG_REL_CMD_LINE     BIT(0)
+
+#define SDEDM_FORCE_DATA_MODE   BIT(19)
+#define SDEDM_CLOCK_PULSE       BIT(20)
+#define SDEDM_BYPASS            BIT(21)
+
+#define SDEDM_WRITE_THRESHOLD_SHIFT 9
+#define SDEDM_READ_THRESHOLD_SHIFT 14
+#define SDEDM_THRESHOLD_MASK     0x1f
+
+#define SDEDM_FSM_MASK           0xf
+#define SDEDM_FSM_IDENTMODE      0x0
+#define SDEDM_FSM_DATAMODE       0x1
+#define SDEDM_FSM_READDATA       0x2
+#define SDEDM_FSM_WRITEDATA      0x3
+#define SDEDM_FSM_READWAIT       0x4
+#define SDEDM_FSM_READCRC        0x5
+#define SDEDM_FSM_WRITECRC       0x6
+#define SDEDM_FSM_WRITEWAIT1     0x7
+#define SDEDM_FSM_POWERDOWN      0x8
+#define SDEDM_FSM_POWERUP        0x9
+#define SDEDM_FSM_WRITESTART1    0xa
+#define SDEDM_FSM_WRITESTART2    0xb
+#define SDEDM_FSM_GENPULSES      0xc
+#define SDEDM_FSM_WRITEWAIT2     0xd
+#define SDEDM_FSM_STARTPOWDOWN   0xf
+
+#define SDDATA_FIFO_WORDS        16
+
+#define FIFO_READ_THRESHOLD     4
+#define FIFO_WRITE_THRESHOLD    4
+#define SDDATA_FIFO_PIO_BURST   8
+#define CMD_DALLY_US            1
+
+struct bcm2835_host {
+	spinlock_t		lock;
+
+	void __iomem		*ioaddr;
+	u32			phys_addr;
+
+	struct mmc_host		*mmc;
+
+	u32			pio_timeout;	/* In jiffies */
+
+	int			clock;		/* Current clock speed */
+
+	unsigned int		max_clk;	/* Max possible freq */
+
+	struct tasklet_struct	finish_tasklet;	/* Tasklet structures */
+
+	struct work_struct	cmd_wait_wq;	/* Workqueue function */
+
+	struct timer_list	timer;		/* Timer for timeouts */
+
+	struct sg_mapping_iter	sg_miter;	/* SG state for PIO */
+	unsigned int		blocks;		/* remaining PIO blocks */
+
+	int			irq;		/* Device IRQ */
+
+	u32			cmd_quick_poll_retries;
+	u32			ns_per_fifo_word;
+
+	/* cached registers */
+	u32			hcfg;
+	u32			cdiv;
+
+	/* Current request */
+	struct mmc_request		*mrq;
+	/* Current command */
+	struct mmc_command		*cmd;
+	/* Current data request */
+	struct mmc_data			*data;
+	/* Data finished before cmd */
+	bool				data_complete:1;
+	/* Drain the fifo when finishing */
+	bool				flush_fifo:1;
+	/* Wait for busy interrupt */
+	bool				use_busy:1;
+	/* Send CMD23 */
+	bool				use_sbc:1;
+
+	/*DMA part*/
+	struct dma_chan			*dma_chan_rx;
+	struct dma_chan			*dma_chan_tx;
+	/* Channel in use */
+	struct dma_chan			*dma_chan;
+	struct dma_async_tx_descriptor	*dma_desc;
+	u32				dma_dir;
+	u32				drain_words;
+	struct page			*drain_page;
+	u32				drain_offset;
+
+	bool				use_dma;
+	/*end of DMA part*/
+	/* maximum length of time spent waiting */
+	int				max_delay;
+	/* Maximum block count for PIO (0 = always DMA) */
+	u32				pio_limit;
+};
+
+static inline void bcm2835_sdhost_write(struct bcm2835_host *host,
+					u32 val, int reg)
+{
+	writel(val, host->ioaddr + reg);
+}
+
+static inline u32 bcm2835_sdhost_read(struct bcm2835_host *host, int reg)
+{
+	return readl(host->ioaddr + reg);
+}
+
+static inline u32 bcm2835_sdhost_read_relaxed(struct bcm2835_host *host,
+					      int reg)
+{
+	return readl_relaxed(host->ioaddr + reg);
+}
+
+static void bcm2835_sdhost_dumpcmd(struct bcm2835_host *host,
+				   struct mmc_command *cmd,
+				   const char *label)
+{
+	if (!cmd)
+		return;
+
+	pr_err("%s:%c%s op %d arg 0x%x flags 0x%x - resp %08x %08x %08x %08x, err %d\n",
+	       mmc_hostname(host->mmc),
+	       (cmd == host->cmd) ? '>' : ' ',
+	       label, cmd->opcode, cmd->arg, cmd->flags,
+	       cmd->resp[0], cmd->resp[1], cmd->resp[2], cmd->resp[3],
+	       cmd->error);
+}
+
+static void bcm2835_sdhost_dumpregs(struct bcm2835_host *host)
+{
+	if (host->mrq) {
+		bcm2835_sdhost_dumpcmd(host, host->mrq->sbc, "sbc");
+		bcm2835_sdhost_dumpcmd(host, host->mrq->cmd, "cmd");
+		if (host->mrq->data) {
+			pr_err("%s: data blocks %x blksz %x - err %d\n",
+			       mmc_hostname(host->mmc),
+			       host->mrq->data->blocks,
+			       host->mrq->data->blksz,
+			       host->mrq->data->error);
+		}
+		bcm2835_sdhost_dumpcmd(host, host->mrq->stop, "stop");
+	}
+
+	pr_err("%s: =========== REGISTER DUMP ===========\n",
+	       mmc_hostname(host->mmc));
+
+	pr_err("%s: SDCMD  0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDCMD));
+	pr_err("%s: SDARG  0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDARG));
+	pr_err("%s: SDTOUT 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDTOUT));
+	pr_err("%s: SDCDIV 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDCDIV));
+	pr_err("%s: SDRSP0 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDRSP0));
+	pr_err("%s: SDRSP1 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDRSP1));
+	pr_err("%s: SDRSP2 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDRSP2));
+	pr_err("%s: SDRSP3 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDRSP3));
+	pr_err("%s: SDHSTS 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDHSTS));
+	pr_err("%s: SDVDD  0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDVDD));
+	pr_err("%s: SDEDM  0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDEDM));
+	pr_err("%s: SDHCFG 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDHCFG));
+	pr_err("%s: SDHBCT 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDHBCT));
+	pr_err("%s: SDHBLC 0x%08x\n",
+	       mmc_hostname(host->mmc),
+	       bcm2835_sdhost_read(host, SDHBLC));
+
+	pr_err("%s: ===========================================\n",
+	       mmc_hostname(host->mmc));
+}
+
+static void bcm2835_sdhost_set_power(struct bcm2835_host *host, bool on)
+{
+	bcm2835_sdhost_write(host, on ? 1 : 0, SDVDD);
+}
+
+static void bcm2835_sdhost_reset_internal(struct bcm2835_host *host)
+{
+	u32 temp;
+
+	bcm2835_sdhost_set_power(host, false);
+
+	bcm2835_sdhost_write(host, 0, SDCMD);
+	bcm2835_sdhost_write(host, 0, SDARG);
+	bcm2835_sdhost_write(host, 0xf00000, SDTOUT);
+	bcm2835_sdhost_write(host, 0, SDCDIV);
+	bcm2835_sdhost_write(host, 0x7f8, SDHSTS); /* Write 1s to clear */
+	bcm2835_sdhost_write(host, 0, SDHCFG);
+	bcm2835_sdhost_write(host, 0, SDHBCT);
+	bcm2835_sdhost_write(host, 0, SDHBLC);
+
+	/* Limit fifo usage due to silicon bug */
+	temp = bcm2835_sdhost_read(host, SDEDM);
+	temp &= ~((SDEDM_THRESHOLD_MASK << SDEDM_READ_THRESHOLD_SHIFT) |
+		  (SDEDM_THRESHOLD_MASK << SDEDM_WRITE_THRESHOLD_SHIFT));
+	temp |= (FIFO_READ_THRESHOLD << SDEDM_READ_THRESHOLD_SHIFT) |
+		(FIFO_WRITE_THRESHOLD << SDEDM_WRITE_THRESHOLD_SHIFT);
+	bcm2835_sdhost_write(host, temp, SDEDM);
+	mdelay(10);
+	bcm2835_sdhost_set_power(host, true);
+	mdelay(10);
+	host->clock = 0;
+	bcm2835_sdhost_write(host, host->hcfg, SDHCFG);
+	bcm2835_sdhost_write(host, host->cdiv, SDCDIV);
+	mmiowb();
+}
+
+static void bcm2835_sdhost_reset(struct mmc_host *mmc)
+{
+	struct bcm2835_host *host = mmc_priv(mmc);
+	unsigned long flags;
+
+	spin_lock_irqsave(&host->lock, flags);
+	bcm2835_sdhost_reset_internal(host);
+	spin_unlock_irqrestore(&host->lock, flags);
+}
+
+static void bcm2835_sdhost_set_ios(struct mmc_host *mmc, struct mmc_ios *ios);
+
+static void bcm2835_sdhost_init(struct bcm2835_host *host, int soft)
+{
+	pr_debug("bcm2835_sdhost_init(%d)\n", soft);
+
+	/* Set interrupt enables */
+	host->hcfg = SDHCFG_BUSY_IRPT_EN;
+
+	bcm2835_sdhost_reset_internal(host);
+
+	if (soft) {
+		/* force clock reconfiguration */
+		host->clock = 0;
+		bcm2835_sdhost_set_ios(host->mmc, &host->mmc->ios);
+	}
+}
+
+static void bcm2835_sdhost_wait_transfer_complete(struct bcm2835_host *host)
+{
+	int timediff;
+	u32 alternate_idle;
+	u32 edm;
+
+	alternate_idle = (host->mrq->data->flags & MMC_DATA_READ) ?
+		SDEDM_FSM_READWAIT : SDEDM_FSM_WRITESTART1;
+
+	edm = bcm2835_sdhost_read(host, SDEDM);
+
+	timediff = 0;
+
+	while (1) {
+		u32 fsm = edm & SDEDM_FSM_MASK;
+
+		if ((fsm == SDEDM_FSM_IDENTMODE) ||
+		    (fsm == SDEDM_FSM_DATAMODE))
+			break;
+		if (fsm == alternate_idle) {
+			bcm2835_sdhost_write(host,
+					     edm | SDEDM_FORCE_DATA_MODE,
+					     SDEDM);
+			break;
+		}
+
+		timediff++;
+		if (timediff == 100000) {
+			pr_err("%s: wait_transfer_complete - still waiting after %d retries\n",
+			       mmc_hostname(host->mmc),
+			       timediff);
+			bcm2835_sdhost_dumpregs(host);
+			host->mrq->data->error = -ETIMEDOUT;
+			return;
+		}
+		cpu_relax();
+		edm = bcm2835_sdhost_read(host, SDEDM);
+	}
+}
+
+static void bcm2835_sdhost_finish_data(struct bcm2835_host *host);
+
+static void bcm2835_sdhost_dma_complete(void *param)
+{
+	struct bcm2835_host *host = param;
+	struct mmc_data *data = host->data;
+	unsigned long flags;
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	if (host->dma_chan) {
+		dma_unmap_sg(host->dma_chan->device->dev,
+			     data->sg, data->sg_len,
+			     host->dma_dir);
+
+		host->dma_chan = NULL;
+	}
+
+	if (host->drain_words) {
+		void *page;
+		u32 *buf;
+
+		page = kmap_atomic(host->drain_page);
+		buf = page + host->drain_offset;
+
+		while (host->drain_words) {
+			u32 edm = bcm2835_sdhost_read(host, SDEDM);
+
+			if ((edm >> 4) & 0x1f)
+				*(buf++) = bcm2835_sdhost_read(host,
+							       SDDATA);
+			host->drain_words--;
+		}
+
+		kunmap_atomic(page);
+	}
+
+	bcm2835_sdhost_finish_data(host);
+
+	spin_unlock_irqrestore(&host->lock, flags);
+}
+
+static void bcm2835_sdhost_transfer_block_pio(struct bcm2835_host *host,
+					      bool is_read)
+{
+	unsigned long flags;
+	size_t blksize, len;
+	u32 *buf;
+	unsigned long wait_max;
+
+	blksize = host->data->blksz;
+
+	wait_max = jiffies + msecs_to_jiffies(host->pio_timeout);
+
+	local_irq_save(flags);
+
+	while (blksize) {
+		int copy_words;
+		u32 hsts = 0;
+
+		if (!sg_miter_next(&host->sg_miter)) {
+			host->data->error = -EINVAL;
+			break;
+		}
+
+		len = min(host->sg_miter.length, blksize);
+		if (len % 4) {
+			host->data->error = -EINVAL;
+			break;
+		}
+
+		blksize -= len;
+		host->sg_miter.consumed = len;
+
+		buf = (u32 *)host->sg_miter.addr;
+
+		copy_words = len / 4;
+
+		while (copy_words) {
+			int burst_words, words;
+			u32 edm;
+
+			burst_words = SDDATA_FIFO_PIO_BURST;
+			if (burst_words > copy_words)
+				burst_words = copy_words;
+			edm = bcm2835_sdhost_read(host, SDEDM);
+			if (is_read)
+				words = ((edm >> 4) & 0x1f);
+			else
+				words = SDDATA_FIFO_WORDS - ((edm >> 4) & 0x1f);
+
+			if (words < burst_words) {
+				int fsm_state = (edm & SDEDM_FSM_MASK);
+
+				if ((is_read &&
+				     (fsm_state != SDEDM_FSM_READDATA &&
+				      fsm_state != SDEDM_FSM_READWAIT &&
+				      fsm_state != SDEDM_FSM_READCRC)) ||
+				    (!is_read &&
+				     (fsm_state != SDEDM_FSM_WRITEDATA &&
+				      fsm_state != SDEDM_FSM_WRITESTART1 &&
+				      fsm_state != SDEDM_FSM_WRITESTART2))) {
+					hsts = bcm2835_sdhost_read(host,
+								   SDHSTS);
+					pr_err("%s: fsm %x, hsts %x\n",
+					       mmc_hostname(host->mmc),
+					       fsm_state, hsts);
+					if (hsts & SDHSTS_ERROR_MASK)
+						break;
+				}
+
+				if (time_after(jiffies, wait_max)) {
+					pr_err("%s: PIO %s timeout - EDM %x\n",
+					       is_read ? "read" : "write",
+					       mmc_hostname(host->mmc),
+					       edm);
+					hsts = SDHSTS_REW_TIME_OUT;
+					break;
+				}
+				ndelay((burst_words - words) *
+				       host->ns_per_fifo_word);
+				continue;
+			} else if (words > copy_words) {
+				words = copy_words;
+			}
+
+			copy_words -= words;
+
+			while (words) {
+				if (is_read) {
+					*(buf++) = bcm2835_sdhost_read(host,
+								       SDDATA);
+				} else {
+					bcm2835_sdhost_write(host, *(buf++),
+							     SDDATA);
+				}
+				words--;
+			}
+		}
+
+		if (hsts & SDHSTS_ERROR_MASK)
+			break;
+	}
+
+	sg_miter_stop(&host->sg_miter);
+
+	local_irq_restore(flags);
+}
+
+static void bcm2835_sdhost_transfer_pio(struct bcm2835_host *host)
+{
+	u32 sdhsts;
+	bool is_read;
+
+	is_read = (host->data->flags & MMC_DATA_READ) != 0;
+	bcm2835_sdhost_transfer_block_pio(host, is_read);
+
+	sdhsts = bcm2835_sdhost_read(host, SDHSTS);
+	if (sdhsts & (SDHSTS_CRC16_ERROR |
+		      SDHSTS_CRC7_ERROR |
+		      SDHSTS_FIFO_ERROR)) {
+		pr_err("%s: %s transfer error - HSTS %x\n",
+		       mmc_hostname(host->mmc),
+		       is_read ? "read" : "write",
+		       sdhsts);
+		host->data->error = -EILSEQ;
+	} else if ((sdhsts & (SDHSTS_CMD_TIME_OUT |
+			      SDHSTS_REW_TIME_OUT))) {
+		pr_err("%s: %s timeout error - HSTS %x\n",
+		       mmc_hostname(host->mmc),
+		       is_read ? "read" : "write",
+		       sdhsts);
+		host->data->error = -ETIMEDOUT;
+	}
+}
+
+static void bcm2835_sdhost_prepare_dma(struct bcm2835_host *host,
+				       struct mmc_data *data)
+{
+	int len, dir_data, dir_slave;
+	struct dma_async_tx_descriptor *desc = NULL;
+	struct dma_chan *dma_chan;
+
+	pr_debug("bcm2835_sdhost_prepare_dma()\n");
+
+	if (data->flags & MMC_DATA_READ) {
+		dma_chan = host->dma_chan_rx;
+		dir_data = DMA_FROM_DEVICE;
+		dir_slave = DMA_DEV_TO_MEM;
+	} else {
+		dma_chan = host->dma_chan_tx;
+		dir_data = DMA_TO_DEVICE;
+		dir_slave = DMA_MEM_TO_DEV;
+	}
+
+	/* The block doesn't manage the FIFO DREQs properly for
+	 * multi-block transfers, so don't attempt to DMA the final
+	 * few words.  Unfortunately this requires the final sg entry
+	 * to be trimmed.  N.B. This code demands that the overspill
+	 * is contained in a single sg entry.
+	 */
+
+	host->drain_words = 0;
+	if ((data->blocks > 1) && (dir_data == DMA_FROM_DEVICE)) {
+		struct scatterlist *sg;
+		u32 len;
+		int i;
+
+		len = min((u32)(FIFO_READ_THRESHOLD - 1) * 4,
+			  (u32)data->blocks * data->blksz);
+
+		for_each_sg(data->sg, sg, data->sg_len, i) {
+			if (sg_is_last(sg)) {
+				WARN_ON(sg->length < len);
+				sg->length -= len;
+				host->drain_page = (struct page *)sg->page_link;
+				host->drain_offset = sg->offset + sg->length;
+			}
+		}
+		host->drain_words = len / 4;
+	}
+
+	len = dma_map_sg(dma_chan->device->dev, data->sg, data->sg_len,
+			 dir_data);
+
+	if (len > 0) {
+		desc = dmaengine_prep_slave_sg(dma_chan, data->sg,
+					       len, dir_slave,
+					       DMA_PREP_INTERRUPT |
+					       DMA_CTRL_ACK);
+	}
+
+	if (desc) {
+		desc->callback = bcm2835_sdhost_dma_complete;
+		desc->callback_param = host;
+		host->dma_desc = desc;
+		host->dma_chan = dma_chan;
+		host->dma_dir = dir_data;
+	}
+}
+
+static void bcm2835_sdhost_start_dma(struct bcm2835_host *host)
+{
+	dmaengine_submit(host->dma_desc);
+	dma_async_issue_pending(host->dma_chan);
+}
+
+static void bcm2835_sdhost_set_transfer_irqs(struct bcm2835_host *host)
+{
+	u32 all_irqs = SDHCFG_DATA_IRPT_EN | SDHCFG_BLOCK_IRPT_EN |
+		SDHCFG_BUSY_IRPT_EN;
+
+	if (host->dma_desc) {
+		host->hcfg = (host->hcfg & ~all_irqs) |
+			SDHCFG_BUSY_IRPT_EN;
+	} else {
+		host->hcfg = (host->hcfg & ~all_irqs) |
+			SDHCFG_DATA_IRPT_EN |
+			SDHCFG_BUSY_IRPT_EN;
+	}
+
+	bcm2835_sdhost_write(host, host->hcfg, SDHCFG);
+}
+
+static void bcm2835_sdhost_prepare_data(struct bcm2835_host *host,
+					struct mmc_command *cmd)
+{
+	struct mmc_data *data = cmd->data;
+
+	WARN_ON(host->data);
+
+	host->data = data;
+	if (!data)
+		return;
+
+	/* Sanity checks */
+	WARN_ON(data->blksz * data->blocks > 524288);
+	WARN_ON(data->blksz > host->mmc->max_blk_size);
+	WARN_ON(data->blocks > 65535);
+
+	host->data_complete = false;
+	host->flush_fifo = false;
+	host->data->bytes_xfered = 0;
+
+	if (!host->dma_desc) {
+		/* Use PIO */
+		int flags = SG_MITER_ATOMIC;
+
+		if (data->flags & MMC_DATA_READ)
+			flags |= SG_MITER_TO_SG;
+		else
+			flags |= SG_MITER_FROM_SG;
+		sg_miter_start(&host->sg_miter, data->sg, data->sg_len, flags);
+		host->blocks = data->blocks;
+	}
+
+	bcm2835_sdhost_set_transfer_irqs(host);
+
+	bcm2835_sdhost_write(host, data->blksz, SDHBCT);
+	bcm2835_sdhost_write(host, data->blocks, SDHBLC);
+}
+
+bool bcm2835_sdhost_send_command(struct bcm2835_host *host,
+				 struct mmc_command *cmd)
+{
+	u32 sdcmd, sdhsts;
+	unsigned long timeout;
+	int delay;
+
+	WARN_ON(host->cmd);
+
+	if (cmd->data) {
+		pr_debug("%s: send_command %d 0x%x (flags 0x%x) - %s %d*%d\n",
+			 mmc_hostname(host->mmc),
+			 cmd->opcode, cmd->arg, cmd->flags,
+			 (cmd->data->flags & MMC_DATA_READ) ?
+			 "read" : "write", cmd->data->blocks,
+			 cmd->data->blksz);
+	} else {
+		pr_debug("%s: send_command %d 0x%x (flags 0x%x)\n",
+			 mmc_hostname(host->mmc),
+			 cmd->opcode, cmd->arg, cmd->flags);
+	}
+
+	/* Wait max 100 ms */
+	timeout = 10000;
+
+	while (bcm2835_sdhost_read(host, SDCMD) & SDCMD_NEW_FLAG) {
+		if (timeout == 0) {
+			pr_err("%s: previous command never completed.\n",
+			       mmc_hostname(host->mmc));
+			bcm2835_sdhost_dumpregs(host);
+			cmd->error = -EILSEQ;
+			tasklet_schedule(&host->finish_tasklet);
+			return false;
+		}
+		timeout--;
+		udelay(10);
+	}
+
+	delay = (10000 - timeout) / 100;
+	if (delay > host->max_delay) {
+		host->max_delay = delay;
+		pr_warn("%s: controller hung for %d ms\n",
+			mmc_hostname(host->mmc),
+			host->max_delay);
+	}
+
+	timeout = jiffies;
+	if (!cmd->data && cmd->busy_timeout > 9000)
+		timeout += DIV_ROUND_UP(cmd->busy_timeout, 1000) * HZ + HZ;
+	else
+		timeout += 10 * HZ;
+	mod_timer(&host->timer, timeout);
+
+	host->cmd = cmd;
+
+	/* Clear any error flags */
+	sdhsts = bcm2835_sdhost_read(host, SDHSTS);
+	if (sdhsts & SDHSTS_ERROR_MASK)
+		bcm2835_sdhost_write(host, sdhsts, SDHSTS);
+
+	if ((cmd->flags & MMC_RSP_136) && (cmd->flags & MMC_RSP_BUSY)) {
+		pr_err("%s: unsupported response type!\n",
+		       mmc_hostname(host->mmc));
+		cmd->error = -EINVAL;
+		tasklet_schedule(&host->finish_tasklet);
+		return false;
+	}
+
+	bcm2835_sdhost_prepare_data(host, cmd);
+
+	bcm2835_sdhost_write(host, cmd->arg, SDARG);
+
+	sdcmd = cmd->opcode & SDCMD_CMD_MASK;
+
+	host->use_busy = false;
+	if (!(cmd->flags & MMC_RSP_PRESENT)) {
+		sdcmd |= SDCMD_NO_RESPONSE;
+	} else {
+		if (cmd->flags & MMC_RSP_136)
+			sdcmd |= SDCMD_LONG_RESPONSE;
+		if (cmd->flags & MMC_RSP_BUSY) {
+			sdcmd |= SDCMD_BUSYWAIT;
+			host->use_busy = true;
+		}
+	}
+
+	if (cmd->data) {
+		if (cmd->data->flags & MMC_DATA_WRITE)
+			sdcmd |= SDCMD_WRITE_CMD;
+		if (cmd->data->flags & MMC_DATA_READ)
+			sdcmd |= SDCMD_READ_CMD;
+	}
+
+	bcm2835_sdhost_write(host, sdcmd | SDCMD_NEW_FLAG, SDCMD);
+
+	return true;
+}
+
+static void bcm2835_sdhost_finish_command(struct bcm2835_host *host,
+					  unsigned long *irq_flags);
+
+static void bcm2835_sdhost_transfer_complete(struct bcm2835_host *host)
+{
+	struct mmc_data *data;
+
+	WARN_ON(!host->data_complete);
+
+	data = host->data;
+	host->data = NULL;
+
+	pr_debug("transfer_complete(error %d, stop %d)\n",
+		 data->error, data->stop ? 1 : 0);
+
+	/* Need to send CMD12 if -
+	 * a) open-ended multiblock transfer (no CMD23)
+	 * b) error in multiblock transfer
+	 */
+	if (host->mrq->stop && (data->error || !host->use_sbc)) {
+		if (bcm2835_sdhost_send_command(host, host->mrq->stop)) {
+			/* No busy, so poll for completion */
+			if (!host->use_busy)
+				bcm2835_sdhost_finish_command(host, NULL);
+		}
+	} else {
+		bcm2835_sdhost_wait_transfer_complete(host);
+		tasklet_schedule(&host->finish_tasklet);
+	}
+}
+
+static void bcm2835_sdhost_finish_data(struct bcm2835_host *host)
+{
+	struct mmc_data *data;
+
+	data = host->data;
+
+	pr_debug("finish_data(error %d, stop %d, sbc %d)\n",
+		 data->error, data->stop ? 1 : 0,
+		 host->mrq->sbc ? 1 : 0);
+
+	host->hcfg &= ~(SDHCFG_DATA_IRPT_EN | SDHCFG_BLOCK_IRPT_EN);
+	bcm2835_sdhost_write(host, host->hcfg, SDHCFG);
+
+	data->bytes_xfered = data->error ? 0 : (data->blksz * data->blocks);
+
+	host->data_complete = true;
+
+	if (host->cmd) {
+		/* Data managed to finish before the
+		 * command completed. Make sure we do
+		 * things in the proper order.
+		 */
+		pr_debug("Finished early - HSTS %x\n",
+			 bcm2835_sdhost_read(host, SDHSTS));
+	} else {
+		bcm2835_sdhost_transfer_complete(host);
+	}
+}
+
+/* If irq_flags is valid, the caller is in a thread context and is
+ * allowed to sleep
+ */
+static void bcm2835_sdhost_finish_command(struct bcm2835_host *host,
+					  unsigned long *irq_flags)
+{
+	u32 sdcmd;
+	u32 retries;
+
+	pr_debug("finish_command(%x)\n", bcm2835_sdhost_read(host, SDCMD));
+
+	/* Poll quickly at first */
+
+	retries = host->cmd_quick_poll_retries;
+	if (!retries) {
+		/* Work out how many polls take 1us by timing 10us */
+		struct timeval start, now;
+		int us_diff;
+
+		retries = 1;
+		do {
+			int i;
+
+			retries *= 2;
+
+			do_gettimeofday(&start);
+
+			for (i = 0; i < retries; i++) {
+				cpu_relax();
+				sdcmd = bcm2835_sdhost_read(host, SDCMD);
+			}
+
+			do_gettimeofday(&now);
+			us_diff = (now.tv_sec - start.tv_sec) * 1000000 +
+				(now.tv_usec - start.tv_usec);
+		} while (us_diff < 10);
+
+		host->cmd_quick_poll_retries =
+			((retries * us_diff + 9) * CMD_DALLY_US) / 10 + 1;
+		retries = 1; /* We've already waited long enough this time */
+	}
+
+	retries = host->cmd_quick_poll_retries;
+	for (sdcmd = bcm2835_sdhost_read(host, SDCMD);
+	     (sdcmd & SDCMD_NEW_FLAG) && !(sdcmd & SDCMD_FAIL_FLAG) && retries;
+	     retries--) {
+		cpu_relax();
+		sdcmd = bcm2835_sdhost_read(host, SDCMD);
+	}
+
+	if (!retries) {
+		unsigned long wait_max;
+
+		if (!irq_flags) {
+			/* Schedule the work */
+			schedule_work(&host->cmd_wait_wq);
+			return;
+		}
+
+		/* Wait max 100 ms */
+		wait_max = jiffies + msecs_to_jiffies(100);
+		while (time_before(jiffies, wait_max)) {
+			spin_unlock_irqrestore(&host->lock, *irq_flags);
+			usleep_range(1, 10);
+			spin_lock_irqsave(&host->lock, *irq_flags);
+			sdcmd = bcm2835_sdhost_read(host, SDCMD);
+			if (!(sdcmd & SDCMD_NEW_FLAG) ||
+			    (sdcmd & SDCMD_FAIL_FLAG))
+				break;
+		}
+	}
+
+	/* Check for errors */
+	if (sdcmd & SDCMD_NEW_FLAG) {
+		pr_err("%s: command never completed.\n",
+		       mmc_hostname(host->mmc));
+		bcm2835_sdhost_dumpregs(host);
+		host->cmd->error = -EIO;
+		tasklet_schedule(&host->finish_tasklet);
+		return;
+	} else if (sdcmd & SDCMD_FAIL_FLAG) {
+		u32 sdhsts = bcm2835_sdhost_read(host, SDHSTS);
+
+		/* Clear the errors */
+		bcm2835_sdhost_write(host, SDHSTS_ERROR_MASK, SDHSTS);
+
+		if (!(sdhsts & SDHSTS_CRC7_ERROR) ||
+		    (host->cmd->opcode != 1)) {
+			if (sdhsts & SDHSTS_CMD_TIME_OUT) {
+				host->cmd->error = -ETIMEDOUT;
+			} else {
+				pr_err("%s: unexpected command %d error\n",
+				       mmc_hostname(host->mmc),
+				       host->cmd->opcode);
+				bcm2835_sdhost_dumpregs(host);
+				host->cmd->error = -EILSEQ;
+			}
+			tasklet_schedule(&host->finish_tasklet);
+			return;
+		}
+	}
+
+	if (host->cmd->flags & MMC_RSP_PRESENT) {
+		if (host->cmd->flags & MMC_RSP_136) {
+			int i;
+
+			for (i = 0; i < 4; i++) {
+				host->cmd->resp[3 - i] =
+					bcm2835_sdhost_read(host,
+							    SDRSP0 + i * 4);
+			}
+
+			pr_debug("%s: finish_command %08x %08x %08x %08x\n",
+				 mmc_hostname(host->mmc),
+				 host->cmd->resp[0], host->cmd->resp[1],
+				 host->cmd->resp[2], host->cmd->resp[3]);
+		} else {
+			host->cmd->resp[0] = bcm2835_sdhost_read(host, SDRSP0);
+			pr_debug("%s: finish_command %08x\n",
+				 mmc_hostname(host->mmc),
+				 host->cmd->resp[0]);
+		}
+	}
+
+	if (host->cmd == host->mrq->sbc) {
+		/* Finished CMD23, now send actual command. */
+		host->cmd = NULL;
+		if (bcm2835_sdhost_send_command(host, host->mrq->cmd)) {
+			if (host->data && host->dma_desc)
+				/* DMA transfer starts now, PIO starts
+				 * after irq
+				 */
+				bcm2835_sdhost_start_dma(host);
+
+			if (!host->use_busy)
+				bcm2835_sdhost_finish_command(host, NULL);
+		}
+	} else if (host->cmd == host->mrq->stop) {
+		/* Finished CMD12 */
+		tasklet_schedule(&host->finish_tasklet);
+	} else {
+		/* Processed actual command. */
+		host->cmd = NULL;
+		if (!host->data)
+			tasklet_schedule(&host->finish_tasklet);
+		else if (host->data_complete)
+			bcm2835_sdhost_transfer_complete(host);
+	}
+}
+
+static void bcm2835_sdhost_timeout(unsigned long data)
+{
+	struct bcm2835_host *host;
+	unsigned long flags;
+
+	host = (struct bcm2835_host *)data;
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	if (host->mrq) {
+		pr_err("%s: timeout waiting for hardware interrupt.\n",
+		       mmc_hostname(host->mmc));
+		bcm2835_sdhost_dumpregs(host);
+
+		if (host->data) {
+			host->data->error = -ETIMEDOUT;
+			bcm2835_sdhost_finish_data(host);
+		} else {
+			if (host->cmd)
+				host->cmd->error = -ETIMEDOUT;
+			else
+				host->mrq->cmd->error = -ETIMEDOUT;
+
+			pr_debug("timeout_timer tasklet_schedule\n");
+			tasklet_schedule(&host->finish_tasklet);
+		}
+	}
+
+	mmiowb();
+	spin_unlock_irqrestore(&host->lock, flags);
+}
+
+static void bcm2835_sdhost_busy_irq(struct bcm2835_host *host, u32 intmask)
+{
+	if (!host->cmd) {
+		pr_err("%s: got command busy interrupt 0x%08x even though no command operation was in progress.\n",
+		       mmc_hostname(host->mmc), (unsigned)intmask);
+		bcm2835_sdhost_dumpregs(host);
+		return;
+	}
+
+	if (!host->use_busy) {
+		pr_err("%s: got command busy interrupt 0x%08x even though not expecting one.\n",
+		       mmc_hostname(host->mmc), (unsigned)intmask);
+		bcm2835_sdhost_dumpregs(host);
+		return;
+	}
+	host->use_busy = false;
+
+	if (intmask & SDHSTS_ERROR_MASK) {
+		pr_err("sdhost_busy_irq: intmask %x, data %p\n",
+		       intmask, host->mrq->data);
+		if (intmask & SDHSTS_CRC7_ERROR) {
+			host->cmd->error = -EILSEQ;
+		} else if (intmask & (SDHSTS_CRC16_ERROR |
+				    SDHSTS_FIFO_ERROR)) {
+			if (host->mrq->data)
+				host->mrq->data->error = -EILSEQ;
+			else
+				host->cmd->error = -EILSEQ;
+		} else if (intmask & SDHSTS_REW_TIME_OUT) {
+			if (host->mrq->data)
+				host->mrq->data->error = -ETIMEDOUT;
+			else
+				host->cmd->error = -ETIMEDOUT;
+		} else if (intmask & SDHSTS_CMD_TIME_OUT) {
+			host->cmd->error = -ETIMEDOUT;
+		}
+
+		bcm2835_sdhost_dumpregs(host);
+	} else {
+		bcm2835_sdhost_finish_command(host, NULL);
+	}
+}
+
+static void bcm2835_sdhost_data_irq(struct bcm2835_host *host, u32 intmask)
+{
+	/* There are no dedicated data/space available interrupt
+	 * status bits, so it is necessary to use the single shared
+	 * data/space available FIFO status bits. It is therefore not
+	 * an error to get here when there is no data transfer in
+	 * progress.
+	 */
+	if (!host->data)
+		return;
+
+	if (intmask & (SDHSTS_CRC16_ERROR |
+		       SDHSTS_FIFO_ERROR |
+		       SDHSTS_REW_TIME_OUT)) {
+		if (intmask & (SDHSTS_CRC16_ERROR |
+			       SDHSTS_FIFO_ERROR))
+			host->data->error = -EILSEQ;
+		else
+			host->data->error = -ETIMEDOUT;
+	}
+
+	if (host->data->error) {
+		bcm2835_sdhost_finish_data(host);
+	} else if (host->data->flags & MMC_DATA_WRITE) {
+		/* Use the block interrupt for writes after the first block */
+		host->hcfg &= ~(SDHCFG_DATA_IRPT_EN);
+		host->hcfg |= SDHCFG_BLOCK_IRPT_EN;
+		bcm2835_sdhost_write(host, host->hcfg, SDHCFG);
+		bcm2835_sdhost_transfer_pio(host);
+	} else {
+		bcm2835_sdhost_transfer_pio(host);
+		host->blocks--;
+		if ((host->blocks == 0) || host->data->error)
+			bcm2835_sdhost_finish_data(host);
+	}
+}
+
+static void bcm2835_sdhost_block_irq(struct bcm2835_host *host, u32 intmask)
+{
+	if (!host->data) {
+		pr_err("%s: got block interrupt 0x%08x even though no data operation was in progress.\n",
+		       mmc_hostname(host->mmc), (unsigned)intmask);
+		bcm2835_sdhost_dumpregs(host);
+		return;
+	}
+
+	if (intmask & (SDHSTS_CRC16_ERROR |
+		       SDHSTS_FIFO_ERROR |
+		       SDHSTS_REW_TIME_OUT)) {
+		if (intmask & (SDHSTS_CRC16_ERROR |
+			       SDHSTS_FIFO_ERROR))
+			host->data->error = -EILSEQ;
+		else
+			host->data->error = -ETIMEDOUT;
+	}
+
+	if (!host->dma_desc) {
+		WARN_ON(!host->blocks);
+		if (host->data->error || (--host->blocks == 0))
+			bcm2835_sdhost_finish_data(host);
+		else
+			bcm2835_sdhost_transfer_pio(host);
+	} else if (host->data->flags & MMC_DATA_WRITE) {
+		bcm2835_sdhost_finish_data(host);
+	}
+}
+
+static irqreturn_t bcm2835_sdhost_irq(int irq, void *dev_id)
+{
+	irqreturn_t result = IRQ_NONE;
+	struct bcm2835_host *host = dev_id;
+	u32 intmask;
+
+	spin_lock(&host->lock);
+
+	intmask = bcm2835_sdhost_read(host, SDHSTS);
+
+	bcm2835_sdhost_write(host,
+			     SDHSTS_BUSY_IRPT |
+			     SDHSTS_BLOCK_IRPT |
+			     SDHSTS_SDIO_IRPT |
+			     SDHSTS_DATA_FLAG,
+			     SDHSTS);
+
+	if (intmask & SDHSTS_BLOCK_IRPT) {
+		bcm2835_sdhost_block_irq(host, intmask);
+		result = IRQ_HANDLED;
+	}
+
+	if (intmask & SDHSTS_BUSY_IRPT) {
+		bcm2835_sdhost_busy_irq(host, intmask);
+		result = IRQ_HANDLED;
+	}
+
+	/* There is no true data interrupt status bit, so it is
+	 * necessary to qualify the data flag with the interrupt
+	 * enable bit.
+	 */
+	if ((intmask & SDHSTS_DATA_FLAG) &&
+	    (host->hcfg & SDHCFG_DATA_IRPT_EN)) {
+		bcm2835_sdhost_data_irq(host, intmask);
+		result = IRQ_HANDLED;
+	}
+
+	mmiowb();
+
+	spin_unlock(&host->lock);
+
+	return result;
+}
+
+void bcm2835_sdhost_set_clock(struct bcm2835_host *host, unsigned int clock)
+{
+	int div = 0; /* Initialized for compiler warning */
+
+	/* The SDCDIV register has 11 bits, and holds (div - 2).  But
+	 * in data mode the max is 50MHz wihout a minimum, and only
+	 * the bottom 3 bits are used. Since the switch over is
+	 * automatic (unless we have marked the card as slow...),
+	 * chosen values have to make sense in both modes.  Ident mode
+	 * must be 100-400KHz, so can range check the requested
+	 * clock. CMD15 must be used to return to data mode, so this
+	 * can be monitored.
+	 *
+	 * clock 250MHz -> 0->125MHz, 1->83.3MHz, 2->62.5MHz, 3->50.0MHz
+	 *                 4->41.7MHz, 5->35.7MHz, 6->31.3MHz, 7->27.8MHz
+	 *
+	 *		 623->400KHz/27.8MHz
+	 *		 reset value (507)->491159/50MHz
+	 *
+	 * BUT, the 3-bit clock divisor in data mode is too small if
+	 * the core clock is higher than 250MHz, so instead use the
+	 * SLOW_CARD configuration bit to force the use of the ident
+	 * clock divisor at all times.
+	 */
+
+	host->mmc->actual_clock = 0;
+
+	if (clock < 100000) {
+		/* Can't stop the clock, but make it as slow as possible
+		 * to show willing
+		 */
+		host->cdiv = SDCDIV_MAX_CDIV;
+		bcm2835_sdhost_write(host, host->cdiv, SDCDIV);
+		return;
+	}
+
+	div = host->max_clk / clock;
+	if (div < 2)
+		div = 2;
+	if ((host->max_clk / div) > clock)
+		div++;
+	div -= 2;
+
+	if (div > SDCDIV_MAX_CDIV)
+		div = SDCDIV_MAX_CDIV;
+
+	clock = host->max_clk / (div + 2);
+	host->mmc->actual_clock = clock;
+
+	/* Calibrate some delays */
+
+	host->ns_per_fifo_word = (1000000000 / clock) *
+		((host->mmc->caps & MMC_CAP_4_BIT_DATA) ? 8 : 32);
+
+	host->cdiv = div;
+	bcm2835_sdhost_write(host, host->cdiv, SDCDIV);
+
+	/* Set the timeout to 500ms */
+	bcm2835_sdhost_write(host, host->mmc->actual_clock / 2, SDTOUT);
+}
+
+static void bcm2835_sdhost_request(struct mmc_host *mmc,
+				   struct mmc_request *mrq)
+{
+	struct bcm2835_host *host;
+	unsigned long flags;
+	u32 edm, fsm;
+
+	host = mmc_priv(mmc);
+
+	/* Reset the error statuses in case this is a retry */
+	if (mrq->sbc)
+		mrq->sbc->error = 0;
+	if (mrq->cmd)
+		mrq->cmd->error = 0;
+	if (mrq->data)
+		mrq->data->error = 0;
+	if (mrq->stop)
+		mrq->stop->error = 0;
+
+	if (mrq->data && !is_power_of_2(mrq->data->blksz)) {
+		pr_err("%s: unsupported block size (%d bytes)\n",
+		       mmc_hostname(mmc), mrq->data->blksz);
+		mrq->cmd->error = -EINVAL;
+		mmc_request_done(mmc, mrq);
+		return;
+	}
+
+	if (host->use_dma && mrq->data &&
+	    (mrq->data->blocks > host->pio_limit))
+		bcm2835_sdhost_prepare_dma(host, mrq->data);
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	WARN_ON(host->mrq);
+	host->mrq = mrq;
+
+	edm = bcm2835_sdhost_read(host, SDEDM);
+	fsm = edm & SDEDM_FSM_MASK;
+
+	if ((fsm != SDEDM_FSM_IDENTMODE) &&
+	    (fsm != SDEDM_FSM_DATAMODE)) {
+		pr_err("%s: previous command (%d) not complete (EDM %x)\n",
+		       mmc_hostname(host->mmc),
+		       bcm2835_sdhost_read(host, SDCMD) & SDCMD_CMD_MASK,
+		       edm);
+		bcm2835_sdhost_dumpregs(host);
+		mrq->cmd->error = -EILSEQ;
+		tasklet_schedule(&host->finish_tasklet);
+		mmiowb();
+		spin_unlock_irqrestore(&host->lock, flags);
+		return;
+	}
+
+	host->use_sbc = !!mrq->sbc && (host->mrq->data->flags & MMC_DATA_READ);
+	if (host->use_sbc) {
+		if (bcm2835_sdhost_send_command(host, mrq->sbc)) {
+			if (!host->use_busy)
+				bcm2835_sdhost_finish_command(host, &flags);
+		}
+	} else if (bcm2835_sdhost_send_command(host, mrq->cmd)) {
+		if (host->data && host->dma_desc) {
+			/* DMA transfer starts now, PIO starts after irq */
+			bcm2835_sdhost_start_dma(host);
+		}
+
+		if (!host->use_busy)
+			bcm2835_sdhost_finish_command(host, &flags);
+	}
+
+	mmiowb();
+
+	spin_unlock_irqrestore(&host->lock, flags);
+}
+
+static void bcm2835_sdhost_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
+{
+	struct bcm2835_host *host = mmc_priv(mmc);
+	unsigned long flags;
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	if (!ios->clock || ios->clock != host->clock) {
+		bcm2835_sdhost_set_clock(host, ios->clock);
+		host->clock = ios->clock;
+	}
+
+	/* set bus width */
+	host->hcfg &= ~SDHCFG_WIDE_EXT_BUS;
+	if (ios->bus_width == MMC_BUS_WIDTH_4)
+		host->hcfg |= SDHCFG_WIDE_EXT_BUS;
+
+	host->hcfg |= SDHCFG_WIDE_INT_BUS;
+
+	/* Disable clever clock switching, to cope with fast core clocks */
+	host->hcfg |= SDHCFG_SLOW_CARD;
+
+	bcm2835_sdhost_write(host, host->hcfg, SDHCFG);
+
+	mmiowb();
+
+	spin_unlock_irqrestore(&host->lock, flags);
+}
+
+static struct mmc_host_ops bcm2835_sdhost_ops = {
+	.request = bcm2835_sdhost_request,
+	.set_ios = bcm2835_sdhost_set_ios,
+	.hw_reset = bcm2835_sdhost_reset,
+};
+
+static void bcm2835_sdhost_cmd_wait_work(struct work_struct *work)
+{
+	struct bcm2835_host *host;
+	unsigned long flags;
+
+	host = container_of(work, struct bcm2835_host, cmd_wait_wq);
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	/* If this tasklet gets rescheduled while running, it will
+	 * be run again afterwards but without any active request.
+	 */
+	if (!host->mrq) {
+		spin_unlock_irqrestore(&host->lock, flags);
+		return;
+	}
+
+	bcm2835_sdhost_finish_command(host, &flags);
+
+	mmiowb();
+
+	spin_unlock_irqrestore(&host->lock, flags);
+}
+
+static void bcm2835_sdhost_tasklet_finish(unsigned long param)
+{
+	struct bcm2835_host *host;
+	unsigned long flags;
+	struct mmc_request *mrq;
+	struct dma_chan *terminate_chan = NULL;
+
+	host = (struct bcm2835_host *)param;
+
+	spin_lock_irqsave(&host->lock, flags);
+
+	/* If this tasklet gets rescheduled while running, it will
+	 * be run again afterwards but without any active request.
+	 */
+	if (!host->mrq) {
+		spin_unlock_irqrestore(&host->lock, flags);
+		return;
+	}
+
+	del_timer(&host->timer);
+
+	mrq = host->mrq;
+
+	host->mrq = NULL;
+	host->cmd = NULL;
+	host->data = NULL;
+
+	mmiowb();
+
+	host->dma_desc = NULL;
+	terminate_chan = host->dma_chan;
+	host->dma_chan = NULL;
+
+	spin_unlock_irqrestore(&host->lock, flags);
+
+	if (terminate_chan) {
+		int err = dmaengine_terminate_all(terminate_chan);
+
+		if (err)
+			pr_err("%s: failed to terminate DMA (%d)\n",
+			       mmc_hostname(host->mmc), err);
+	}
+
+	mmc_request_done(host->mmc, mrq);
+}
+
+int bcm2835_sdhost_add_host(struct bcm2835_host *host)
+{
+	struct mmc_host *mmc;
+	struct dma_slave_config cfg;
+	char pio_limit_string[20];
+	int ret;
+
+	mmc = host->mmc;
+
+	bcm2835_sdhost_reset_internal(host);
+
+	mmc->f_max = host->max_clk;
+	mmc->f_min = host->max_clk / SDCDIV_MAX_CDIV;
+
+	mmc->max_busy_timeout = ~0 / (mmc->f_max / 1000);
+
+	pr_debug("f_max %d, f_min %d, max_busy_timeout %d\n",
+		 mmc->f_max, mmc->f_min, mmc->max_busy_timeout);
+
+	/* host controller capabilities */
+	mmc->caps |=
+		MMC_CAP_SD_HIGHSPEED | MMC_CAP_MMC_HIGHSPEED |
+		MMC_CAP_NEEDS_POLL | MMC_CAP_HW_RESET | MMC_CAP_ERASE |
+		MMC_CAP_CMD23;
+
+	spin_lock_init(&host->lock);
+
+	if (IS_ERR_OR_NULL(host->dma_chan_tx) ||
+	    IS_ERR_OR_NULL(host->dma_chan_rx)) {
+		pr_err("%s: unable to initialise DMA channels. Falling back to PIO\n",
+		       mmc_hostname(mmc));
+		host->use_dma = false;
+	} else {
+		host->use_dma = true;
+
+		cfg.src_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
+		cfg.dst_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
+		cfg.slave_id = 13;		/* DREQ channel */
+
+		cfg.direction = DMA_MEM_TO_DEV;
+		cfg.src_addr = 0;
+		cfg.dst_addr = host->phys_addr + SDDATA;
+		ret = dmaengine_slave_config(host->dma_chan_tx, &cfg);
+		if (ret)
+			host->use_dma = false;
+
+		cfg.direction = DMA_DEV_TO_MEM;
+		cfg.src_addr = host->phys_addr + SDDATA;
+		cfg.dst_addr = 0;
+		ret = dmaengine_slave_config(host->dma_chan_rx, &cfg);
+		if (ret)
+			host->use_dma = false;
+	}
+
+	mmc->max_segs = 128;
+	mmc->max_req_size = 524288;
+	mmc->max_seg_size = mmc->max_req_size;
+	mmc->max_blk_size = 512;
+	mmc->max_blk_count =  65535;
+
+	/* report supported voltage ranges */
+	mmc->ocr_avail = MMC_VDD_32_33 | MMC_VDD_33_34;
+
+	tasklet_init(&host->finish_tasklet,
+		     bcm2835_sdhost_tasklet_finish, (unsigned long)host);
+
+	INIT_WORK(&host->cmd_wait_wq, bcm2835_sdhost_cmd_wait_work);
+
+	setup_timer(&host->timer, bcm2835_sdhost_timeout,
+		    (unsigned long)host);
+
+	bcm2835_sdhost_init(host, 0);
+
+	ret = request_irq(host->irq, bcm2835_sdhost_irq, 0 /*IRQF_SHARED*/,
+			  mmc_hostname(mmc), host);
+	if (ret) {
+		pr_err("%s: failed to request IRQ %d: %d\n",
+		       mmc_hostname(mmc), host->irq, ret);
+		goto untasklet;
+	}
+
+	mmiowb();
+	mmc_add_host(mmc);
+
+	pio_limit_string[0] = '\0';
+	if (host->use_dma && (host->pio_limit > 0))
+		sprintf(pio_limit_string, " (>%d)", host->pio_limit);
+	pr_info("%s: loaded - DMA %s%s\n",
+		mmc_hostname(mmc),
+		host->use_dma ? "enabled" : "disabled",
+		pio_limit_string);
+
+	return 0;
+
+untasklet:
+	tasklet_kill(&host->finish_tasklet);
+
+	return ret;
+}
+
+static int bcm2835_sdhost_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct clk *clk;
+	struct resource *iomem;
+	struct bcm2835_host *host;
+	struct mmc_host *mmc;
+	int ret;
+
+	pr_debug("bcm2835_sdhost_probe\n");
+	mmc = mmc_alloc_host(sizeof(*host), dev);
+	if (!mmc)
+		return -ENOMEM;
+
+	mmc->ops = &bcm2835_sdhost_ops;
+	host = mmc_priv(mmc);
+	host->mmc = mmc;
+	host->cmd_quick_poll_retries = 0;
+	host->pio_timeout = msecs_to_jiffies(500);
+	host->pio_limit = 1;
+	host->max_delay = 1; /* Warn if over 1ms */
+	spin_lock_init(&host->lock);
+
+	iomem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	host->ioaddr = devm_ioremap_resource(dev, iomem);
+	if (IS_ERR(host->ioaddr)) {
+		ret = PTR_ERR(host->ioaddr);
+		goto err;
+	}
+
+	/* Parse OF address directly to get the physical address for
+	 * DMA to our registers.
+	 */
+	host->phys_addr = be32_to_cpup(of_get_address(pdev->dev.of_node, 0,
+						      NULL, NULL));
+
+	pr_debug(" - ioaddr %lx, iomem->start %lx, phys_addr %lx\n",
+		 (unsigned long)host->ioaddr,
+		 (unsigned long)iomem->start,
+		 (unsigned long)host->phys_addr);
+
+	host->dma_chan = NULL;
+	host->dma_desc = NULL;
+
+	host->dma_chan_tx = dma_request_slave_channel(dev, "tx");
+	host->dma_chan_rx = dma_request_slave_channel(dev, "rx");
+
+	clk = devm_clk_get(dev, NULL);
+	if (IS_ERR(clk)) {
+		ret = PTR_ERR(clk);
+		if (ret != -EPROBE_DEFER)
+			dev_err(dev, "could not get clk: %d\n", ret);
+		goto err;
+	}
+
+	host->max_clk = clk_get_rate(clk);
+
+	host->irq = platform_get_irq(pdev, 0);
+	if (host->irq <= 0) {
+		dev_err(dev, "get IRQ failed\n");
+		ret = -EINVAL;
+		goto err;
+	}
+
+	pr_debug(" - max_clk %lx, irq %d\n",
+		 (unsigned long)host->max_clk,
+		 (int)host->irq);
+
+	mmc_of_parse(mmc);
+
+	ret = bcm2835_sdhost_add_host(host);
+	if (ret)
+		goto err;
+
+	platform_set_drvdata(pdev, host);
+
+	pr_debug("bcm2835_sdhost_probe -> OK\n");
+
+	return 0;
+
+err:
+	pr_debug("bcm2835_sdhost_probe -> err %d\n", ret);
+	mmc_free_host(mmc);
+
+	return ret;
+}
+
+static int bcm2835_sdhost_remove(struct platform_device *pdev)
+{
+	struct bcm2835_host *host = platform_get_drvdata(pdev);
+
+	pr_debug("bcm2835_sdhost_remove\n");
+
+	mmc_remove_host(host->mmc);
+
+	bcm2835_sdhost_set_power(host, false);
+
+	free_irq(host->irq, host);
+
+	del_timer_sync(&host->timer);
+
+	tasklet_kill(&host->finish_tasklet);
+
+	mmc_free_host(host->mmc);
+	platform_set_drvdata(pdev, NULL);
+
+	pr_debug("bcm2835_sdhost_remove - OK\n");
+	return 0;
+}
+
+static const struct of_device_id bcm2835_sdhost_match[] = {
+	{ .compatible = "brcm,bcm2835-sdhost" },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, bcm2835_sdhost_match);
+
+static struct platform_driver bcm2835_sdhost_driver = {
+	.probe      = bcm2835_sdhost_probe,
+	.remove     = bcm2835_sdhost_remove,
+	.driver     = {
+		.name		= "sdhost-bcm2835",
+		.of_match_table	= bcm2835_sdhost_match,
+	},
+};
+module_platform_driver(bcm2835_sdhost_driver);
+
+MODULE_ALIAS("platform:sdhost-bcm2835");
+MODULE_DESCRIPTION("BCM2835 SDHost driver");
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Phil Elwell");