diff mbox

[3/5] dw_spi: rework message processing

Message ID 1308158588-17249-4-git-send-email-dirk.brandewie@gmail.com
State Superseded, archived
Headers show

Commit Message

dirk.brandewie@gmail.com June 15, 2011, 5:23 p.m. UTC
From: Dirk Brandewie <dirk.brandewie@gmail.com>

NOTE: patch created git format-patch --break-rewrites=/50%

This patch reworks the message pump worker thread function to run
until all messages queued to the driver have been handled. The
function to handle individual spi_transfers is now a synchronus
function the tasklet to handle spi_transfers has been removed. Work
for the worker thread is only queued in host controller transfer
function.

Psuedo code for new thread function:
  message = get_message()
  while (message){
    for_each_transfer_in_msg(message){
      transfer_setup(transfer)
      do_transfer()
    }
    complete_message()
    message = get_message()
  }

Changes that fell out of the message thread changes:
Non-DMA transfers that are larger than the size of the controller FIFO
are handled as interrupt driven transfers.

Common FIFO handling functions shared PIO and interrupt transfers.

Simplified queue stop/start funcitons.

Cleanup fixes:
Changed exported all exported function names to have dw_spi_ prefix

Removed support for registering chip select control function. Setting
the slave chip select is handled by setting the SER (Slave enable
register)

Removed code that looked at the cs_change hint in the
spi_transfer. Software has no contorl over whether the slave chip
select is de-asserted at the end of the transfer.  Once the TX FIFO
goes empty the slave chip select is dropped.

Added dw_spi_{en,dis}able inline functions to replace spi_enable_chip()
Added dw_spi_{mask,umask}_intr inline functions

Signed-off-by: Dirk Brandewie <dirk.brandewie@gmail.com>
---
 drivers/spi/spi-dw-mid.c |   43 +-
 drivers/spi/spi-dw.c     | 1670 ++++++++++++++++++++--------------------------
 drivers/spi/spi-dw.h     |   92 ++--
 3 files changed, 809 insertions(+), 996 deletions(-)
 rewrite drivers/spi/spi-dw.c (55%)

Comments

Grant Likely June 16, 2011, 1:14 p.m. UTC | #1
On Wed, Jun 15, 2011 at 10:23:06AM -0700, dirk.brandewie@gmail.com wrote:
> From: Dirk Brandewie <dirk.brandewie@gmail.com>
> 
> NOTE: patch created git format-patch --break-rewrites=/50%
> 
> This patch reworks the message pump worker thread function to run
> until all messages queued to the driver have been handled. The
> function to handle individual spi_transfers is now a synchronus
> function the tasklet to handle spi_transfers has been removed. Work
> for the worker thread is only queued in host controller transfer
> function.
> 
> Psuedo code for new thread function:
>   message = get_message()
>   while (message){
>     for_each_transfer_in_msg(message){
>       transfer_setup(transfer)
>       do_transfer()
>     }
>     complete_message()
>     message = get_message()
>   }
> 
> Changes that fell out of the message thread changes:
> Non-DMA transfers that are larger than the size of the controller FIFO
> are handled as interrupt driven transfers.
> 
> Common FIFO handling functions shared PIO and interrupt transfers.
> 
> Simplified queue stop/start funcitons.
> 
> Cleanup fixes:
> Changed exported all exported function names to have dw_spi_ prefix
> 
> Removed support for registering chip select control function. Setting
> the slave chip select is handled by setting the SER (Slave enable
> register)

What about for implementations that use a GPIO for the SPI chip
select?  It is very common for board designs to use GPIOs for
multiplexing the SPI bus.

> 
> Removed code that looked at the cs_change hint in the
> spi_transfer. Software has no contorl over whether the slave chip
> select is de-asserted at the end of the transfer.  Once the TX FIFO
> goes empty the slave chip select is dropped.

This sounds wrong.  cs_change is *not* merely a hint.  It must be respected.
If the driver has no direct control over the CS line, then it is
incumbent on the driver to guaranteed that the cs deassert condition
does not occur.  This will probably mean chaining up all the transfers
in a message so that the TX FIFO remains full.  If cs_change is
requested, then the FIFO must be allowed to empty before kicking of
the next group of transfers.

> 
> Added dw_spi_{en,dis}able inline functions to replace spi_enable_chip()
> Added dw_spi_{mask,umask}_intr inline functions
> 
> Signed-off-by: Dirk Brandewie <dirk.brandewie@gmail.com>

As previously discussed both on-list and off, this is a major rework
and I'd like to hear from the folks who are affected by this driver
before I commit to merging it.  I don't like merging something so
large that it ends up basically being a driver rewrite, but if there
aren't any strong objections, I'll probably grudgingly accept it.

More comments below.

> ---
>  drivers/spi/spi-dw-mid.c |   43 +-
>  drivers/spi/spi-dw.c     | 1670 ++++++++++++++++++++--------------------------
>  drivers/spi/spi-dw.h     |   92 ++--
>  3 files changed, 809 insertions(+), 996 deletions(-)
>  rewrite drivers/spi/spi-dw.c (55%)
> 
> diff --git a/drivers/spi/spi-dw-mid.c b/drivers/spi/spi-dw-mid.c
> index 130e555..e44e37f 100644
> --- a/drivers/spi/spi-dw-mid.c
> +++ b/drivers/spi/spi-dw-mid.c
> @@ -38,7 +38,10 @@ static bool mid_spi_dma_chan_filter(struct dma_chan *chan, void *param)
>  {
>  	struct dw_spi *dws = param;
>  
> -	return dws->dmac && (&dws->dmac->dev == chan->device->dev);
> +	if (dws->dmac && &dws->dmac->dev == chan->device->dev)
> +		return true;
> +	else
> +		return false;

Why?  The old code is correct, and the new code is more verbose.

>  }
>  
>  static int mid_spi_dma_init(struct dw_spi *dws)
> @@ -103,10 +106,10 @@ static void dw_spi_dma_done(void *arg)
>  
>  	if (++dws->dma_chan_done != 2)
>  		return;
> -	dw_spi_xfer_done(dws);
> +	complete(&dws->xfer.complete);
>  }
>  
> -static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
> +static int mid_spi_dma_transfer(struct dw_spi *dws)
>  {
>  	struct dma_async_tx_descriptor *txdesc = NULL, *rxdesc = NULL;
>  	struct dma_chan *txchan, *rxchan;
> @@ -114,17 +117,17 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>  	u16 dma_ctrl = 0;
>  
>  	/* 1. setup DMA related registers */
> -	if (cs_change) {
> -		spi_enable_chip(dws, 0);
> -		dw_writew(dws, dmardlr, 0xf);
> -		dw_writew(dws, dmatdlr, 0x10);
> -		if (dws->tx_dma)
> -			dma_ctrl |= 0x2;
> -		if (dws->rx_dma)
> -			dma_ctrl |= 0x1;
> -		dw_writew(dws, dmacr, dma_ctrl);
> -		spi_enable_chip(dws, 1);
> -	}
> +
> +	dw_spi_disable(dws);
> +	dw_writew(dws, dmardlr, 0xf);
> +	dw_writew(dws, dmatdlr, 0x10);
> +	if (dws->xfer.tx_dma)
> +		dma_ctrl |= 0x2;
> +	if (dws->xfer.rx_dma)
> +		dma_ctrl |= 0x1;
> +	dw_writew(dws, dmacr, dma_ctrl);
> +	dw_spi_enable(dws);
> +
>  
>  	dws->dma_chan_done = 0;
>  	txchan = dws->txchan;
> @@ -141,8 +144,8 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>  				       (unsigned long) &txconf);
>  
>  	memset(&dws->tx_sgl, 0, sizeof(dws->tx_sgl));
> -	dws->tx_sgl.dma_address = dws->tx_dma;
> -	dws->tx_sgl.length = dws->len;
> +	dws->tx_sgl.dma_address = dws->xfer.tx_dma;
> +	dws->tx_sgl.length = dws->xfer.len;
>  
>  	txdesc = txchan->device->device_prep_slave_sg(txchan,
>  				&dws->tx_sgl,
> @@ -163,8 +166,8 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>  				       (unsigned long) &rxconf);
>  
>  	memset(&dws->rx_sgl, 0, sizeof(dws->rx_sgl));
> -	dws->rx_sgl.dma_address = dws->rx_dma;
> -	dws->rx_sgl.length = dws->len;
> +	dws->rx_sgl.dma_address = dws->xfer.rx_dma;
> +	dws->rx_sgl.length = dws->xfer.len;
>  
>  	rxdesc = rxchan->device->device_prep_slave_sg(rxchan,
>  				&dws->rx_sgl,
> @@ -188,7 +191,6 @@ static struct dw_spi_dma_ops mid_dma_ops = {
>  #endif
>  
>  /* Some specific info for SPI0 controller on Moorestown */
> -
>  /* HW info for MRST CLk Control Unit, one 32b reg */
>  #define MRST_SPI_CLK_BASE	100000000	/* 100m */
>  #define MRST_CLK_SPI0_REG	0xff11d86c
> @@ -202,12 +204,13 @@ int dw_spi_mid_init(struct dw_spi *dws)
>  {
>  	u32 *clk_reg, clk_cdiv;
>  
> +

nit: unrelated whitespace changes

>  	clk_reg = ioremap_nocache(MRST_CLK_SPI0_REG, 16);
>  	if (!clk_reg)
>  		return -ENOMEM;
>  
>  	/* get SPI controller operating freq info */
> -	clk_cdiv  = (readl(clk_reg) & CLK_SPI_CDIV_MASK) >> CLK_SPI_CDIV_OFFSET;
> +	clk_cdiv  = ((*clk_reg) & CLK_SPI_CDIV_MASK) >> CLK_SPI_CDIV_OFFSET;

Woah.  Changes correct usage of an io accessor to a direct dereference
of a register?  That's not cool.  Why is a direct dereference needed
here?  If you have to do something like this, then it must have a
comment explaining why.

>  	dws->max_freq = MRST_SPI_CLK_BASE / (clk_cdiv + 1);
>  	iounmap(clk_reg);
>  
> diff --git a/drivers/spi/spi-dw.c b/drivers/spi/spi-dw.c
> dissimilarity index 55%
> index ece5f69..2dacb8f 100644
> --- a/drivers/spi/spi-dw.c
> +++ b/drivers/spi/spi-dw.c
> @@ -1,936 +1,734 @@
> -/*
> - * Designware SPI core controller driver (refer pxa2xx_spi.c)
> - *
> - * Copyright (c) 2009, Intel Corporation.
> - *
> - * 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, write to the Free Software Foundation, Inc.,
> - * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
> - */
> -
> -#include <linux/dma-mapping.h>
> -#include <linux/interrupt.h>
> -#include <linux/highmem.h>
> -#include <linux/delay.h>
> -#include <linux/slab.h>
> -#include <linux/spi/spi.h>
> -
> -#include "spi-dw.h"
> -
> -#ifdef CONFIG_DEBUG_FS
> -#include <linux/debugfs.h>
> -#endif
> -
> -#define START_STATE	((void *)0)
> -#define RUNNING_STATE	((void *)1)
> -#define DONE_STATE	((void *)2)
> -#define ERROR_STATE	((void *)-1)
> -
> -#define QUEUE_RUNNING	0
> -#define QUEUE_STOPPED	1
> -
> -#define MRST_SPI_DEASSERT	0
> -#define MRST_SPI_ASSERT		1
> -
> -/* Slave spi_dev related */
> -struct chip_data {
> -	u16 cr0;
> -	u8 cs;			/* chip select pin */
> -	u8 n_bytes;		/* current is a 1/2/4 byte op */
> -	u8 tmode;		/* TR/TO/RO/EEPROM */
> -	u8 type;		/* SPI/SSP/MicroWire */
> -
> -	u8 poll_mode;		/* 1 means use poll mode */
> -
> -	u32 dma_width;
> -	u32 rx_threshold;
> -	u32 tx_threshold;
> -	u8 enable_dma;
> -	u8 bits_per_word;
> -	u16 clk_div;		/* baud rate divider */
> -	u32 speed_hz;		/* baud rate */
> -	void (*cs_control)(u32 command);
> -};
> -
> -#ifdef CONFIG_DEBUG_FS
> -static int spi_show_regs_open(struct inode *inode, struct file *file)
> -{
> -	file->private_data = inode->i_private;
> -	return 0;
> -}
> -
> -#define SPI_REGS_BUFSIZE	1024
> -static ssize_t  spi_show_regs(struct file *file, char __user *user_buf,
> -				size_t count, loff_t *ppos)
> -{
> -	struct dw_spi *dws;
> -	char *buf;
> -	u32 len = 0;
> -	ssize_t ret;
> -
> -	dws = file->private_data;
> -
> -	buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
> -	if (!buf)
> -		return 0;
> -
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"MRST SPI0 registers:\n");
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"=================================\n");
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"SER: \t\t0x%08x\n", dw_readl(dws, ser));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"SR: \t\t0x%08x\n", dw_readl(dws, sr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"IMR: \t\t0x%08x\n", dw_readl(dws, imr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"ISR: \t\t0x%08x\n", dw_readl(dws, isr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"DMATDLR: \t0x%08x\n", dw_readl(dws, dmatdlr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"DMARDLR: \t0x%08x\n", dw_readl(dws, dmardlr));
> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -			"=================================\n");
> -
> -	ret =  simple_read_from_buffer(user_buf, count, ppos, buf, len);
> -	kfree(buf);
> -	return ret;
> -}
> -
> -static const struct file_operations mrst_spi_regs_ops = {
> -	.owner		= THIS_MODULE,
> -	.open		= spi_show_regs_open,
> -	.read		= spi_show_regs,
> -	.llseek		= default_llseek,
> -};
> -
> -static int mrst_spi_debugfs_init(struct dw_spi *dws)
> -{
> -	dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
> -	if (!dws->debugfs)
> -		return -ENOMEM;
> -
> -	debugfs_create_file("registers", S_IFREG | S_IRUGO,
> -		dws->debugfs, (void *)dws, &mrst_spi_regs_ops);
> -	return 0;
> -}
> -
> -static void mrst_spi_debugfs_remove(struct dw_spi *dws)
> -{
> -	if (dws->debugfs)
> -		debugfs_remove_recursive(dws->debugfs);
> -}
> -
> -#else
> -static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
> -{
> -	return 0;
> -}
> -
> -static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
> -{
> -}
> -#endif /* CONFIG_DEBUG_FS */
> -
> -/* Return the max entries we can fill into tx fifo */
> -static inline u32 tx_max(struct dw_spi *dws)
> -{
> -	u32 tx_left, tx_room, rxtx_gap;
> -
> -	tx_left = (dws->tx_end - dws->tx) / dws->n_bytes;
> -	tx_room = dws->fifo_len - dw_readw(dws, txflr);
> -
> -	/*
> -	 * Another concern is about the tx/rx mismatch, we
> -	 * though to use (dws->fifo_len - rxflr - txflr) as
> -	 * one maximum value for tx, but it doesn't cover the
> -	 * data which is out of tx/rx fifo and inside the
> -	 * shift registers. So a control from sw point of
> -	 * view is taken.
> -	 */
> -	rxtx_gap =  ((dws->rx_end - dws->rx) - (dws->tx_end - dws->tx))
> -			/ dws->n_bytes;
> -
> -	return min3(tx_left, tx_room, (u32) (dws->fifo_len - rxtx_gap));
> -}
> -
> -/* Return the max entries we should read out of rx fifo */
> -static inline u32 rx_max(struct dw_spi *dws)
> -{
> -	u32 rx_left = (dws->rx_end - dws->rx) / dws->n_bytes;
> -
> -	return min(rx_left, (u32)dw_readw(dws, rxflr));
> -}
> -
> -static void dw_writer(struct dw_spi *dws)
> -{
> -	u32 max = tx_max(dws);
> -	u16 txw = 0;
> -
> -	while (max--) {
> -		/* Set the tx word if the transfer's original "tx" is not null */
> -		if (dws->tx_end - dws->len) {
> -			if (dws->n_bytes == 1)
> -				txw = *(u8 *)(dws->tx);
> -			else
> -				txw = *(u16 *)(dws->tx);
> -		}
> -		dw_writew(dws, dr, txw);
> -		dws->tx += dws->n_bytes;
> -	}
> -}
> -
> -static void dw_reader(struct dw_spi *dws)
> -{
> -	u32 max = rx_max(dws);
> -	u16 rxw;
> -
> -	while (max--) {
> -		rxw = dw_readw(dws, dr);
> -		/* Care rx only if the transfer's original "rx" is not null */
> -		if (dws->rx_end - dws->len) {
> -			if (dws->n_bytes == 1)
> -				*(u8 *)(dws->rx) = rxw;
> -			else
> -				*(u16 *)(dws->rx) = rxw;
> -		}
> -		dws->rx += dws->n_bytes;
> -	}
> -}
> -
> -static void *next_transfer(struct dw_spi *dws)
> -{
> -	struct spi_message *msg = dws->cur_msg;
> -	struct spi_transfer *trans = dws->cur_transfer;
> -
> -	/* Move to next transfer */
> -	if (trans->transfer_list.next != &msg->transfers) {
> -		dws->cur_transfer =
> -			list_entry(trans->transfer_list.next,
> -					struct spi_transfer,
> -					transfer_list);
> -		return RUNNING_STATE;
> -	} else
> -		return DONE_STATE;
> -}
> -
> -/*
> - * Note: first step is the protocol driver prepares
> - * a dma-capable memory, and this func just need translate
> - * the virt addr to physical
> - */
> -static int map_dma_buffers(struct dw_spi *dws)
> -{
> -	if (!dws->cur_msg->is_dma_mapped
> -		|| !dws->dma_inited
> -		|| !dws->cur_chip->enable_dma
> -		|| !dws->dma_ops)
> -		return 0;
> -
> -	if (dws->cur_transfer->tx_dma)
> -		dws->tx_dma = dws->cur_transfer->tx_dma;
> -
> -	if (dws->cur_transfer->rx_dma)
> -		dws->rx_dma = dws->cur_transfer->rx_dma;
> -
> -	return 1;
> -}
> -
> -/* Caller already set message->status; dma and pio irqs are blocked */
> -static void giveback(struct dw_spi *dws)
> -{
> -	struct spi_transfer *last_transfer;
> -	unsigned long flags;
> -	struct spi_message *msg;
> -
> -	spin_lock_irqsave(&dws->lock, flags);
> -	msg = dws->cur_msg;
> -	dws->cur_msg = NULL;
> -	dws->cur_transfer = NULL;
> -	dws->prev_chip = dws->cur_chip;
> -	dws->cur_chip = NULL;
> -	dws->dma_mapped = 0;
> -	queue_work(dws->workqueue, &dws->pump_messages);
> -	spin_unlock_irqrestore(&dws->lock, flags);
> -
> -	last_transfer = list_entry(msg->transfers.prev,
> -					struct spi_transfer,
> -					transfer_list);
> -
> -	if (!last_transfer->cs_change && dws->cs_control)
> -		dws->cs_control(MRST_SPI_DEASSERT);
> -
> -	msg->state = NULL;
> -	if (msg->complete)
> -		msg->complete(msg->context);
> -}
> -
> -static void int_error_stop(struct dw_spi *dws, const char *msg)
> -{
> -	/* Stop the hw */
> -	spi_enable_chip(dws, 0);
> -
> -	dev_err(&dws->master->dev, "%s\n", msg);
> -	dws->cur_msg->state = ERROR_STATE;
> -	tasklet_schedule(&dws->pump_transfers);
> -}
> -
> -void dw_spi_xfer_done(struct dw_spi *dws)
> -{
> -	/* Update total byte transferred return count actual bytes read */
> -	dws->cur_msg->actual_length += dws->len;
> -
> -	/* Move to next transfer */
> -	dws->cur_msg->state = next_transfer(dws);
> -
> -	/* Handle end of message */
> -	if (dws->cur_msg->state == DONE_STATE) {
> -		dws->cur_msg->status = 0;
> -		giveback(dws);
> -	} else
> -		tasklet_schedule(&dws->pump_transfers);
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_xfer_done);
> -
> -static irqreturn_t interrupt_transfer(struct dw_spi *dws)
> -{
> -	u16 irq_status = dw_readw(dws, isr);
> -
> -	/* Error handling */
> -	if (irq_status & (SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI)) {
> -		dw_readw(dws, txoicr);
> -		dw_readw(dws, rxoicr);
> -		dw_readw(dws, rxuicr);
> -		int_error_stop(dws, "interrupt_transfer: fifo overrun/underrun");
> -		return IRQ_HANDLED;
> -	}
> -
> -	dw_reader(dws);
> -	if (dws->rx_end == dws->rx) {
> -		spi_mask_intr(dws, SPI_INT_TXEI);
> -		dw_spi_xfer_done(dws);
> -		return IRQ_HANDLED;
> -	}
> -	if (irq_status & SPI_INT_TXEI) {
> -		spi_mask_intr(dws, SPI_INT_TXEI);
> -		dw_writer(dws);
> -		/* Enable TX irq always, it will be disabled when RX finished */
> -		spi_umask_intr(dws, SPI_INT_TXEI);
> -	}
> -
> -	return IRQ_HANDLED;
> -}
> -
> -static irqreturn_t dw_spi_irq(int irq, void *dev_id)
> -{
> -	struct dw_spi *dws = dev_id;
> -	u16 irq_status = dw_readw(dws, isr) & 0x3f;
> -
> -	if (!irq_status)
> -		return IRQ_NONE;
> -
> -	if (!dws->cur_msg) {
> -		spi_mask_intr(dws, SPI_INT_TXEI);
> -		return IRQ_HANDLED;
> -	}
> -
> -	return dws->transfer_handler(dws);
> -}
> -
> -/* Must be called inside pump_transfers() */
> -static void poll_transfer(struct dw_spi *dws)
> -{
> -	do {
> -		dw_writer(dws);
> -		dw_reader(dws);
> -		cpu_relax();
> -	} while (dws->rx_end > dws->rx);
> -
> -	dw_spi_xfer_done(dws);
> -}
> -
> -static void pump_transfers(unsigned long data)
> -{
> -	struct dw_spi *dws = (struct dw_spi *)data;
> -	struct spi_message *message = NULL;
> -	struct spi_transfer *transfer = NULL;
> -	struct spi_transfer *previous = NULL;
> -	struct spi_device *spi = NULL;
> -	struct chip_data *chip = NULL;
> -	u8 bits = 0;
> -	u8 imask = 0;
> -	u8 cs_change = 0;
> -	u16 txint_level = 0;
> -	u16 clk_div = 0;
> -	u32 speed = 0;
> -	u32 cr0 = 0;
> -
> -	/* Get current state information */
> -	message = dws->cur_msg;
> -	transfer = dws->cur_transfer;
> -	chip = dws->cur_chip;
> -	spi = message->spi;
> -
> -	if (unlikely(!chip->clk_div))
> -		chip->clk_div = dws->max_freq / chip->speed_hz;
> -
> -	if (message->state == ERROR_STATE) {
> -		message->status = -EIO;
> -		goto early_exit;
> -	}
> -
> -	/* Handle end of message */
> -	if (message->state == DONE_STATE) {
> -		message->status = 0;
> -		goto early_exit;
> -	}
> -
> -	/* Delay if requested at end of transfer*/
> -	if (message->state == RUNNING_STATE) {
> -		previous = list_entry(transfer->transfer_list.prev,
> -					struct spi_transfer,
> -					transfer_list);
> -		if (previous->delay_usecs)
> -			udelay(previous->delay_usecs);
> -	}
> -
> -	dws->n_bytes = chip->n_bytes;
> -	dws->dma_width = chip->dma_width;
> -	dws->cs_control = chip->cs_control;
> -
> -	dws->rx_dma = transfer->rx_dma;
> -	dws->tx_dma = transfer->tx_dma;
> -	dws->tx = (void *)transfer->tx_buf;
> -	dws->tx_end = dws->tx + transfer->len;
> -	dws->rx = transfer->rx_buf;
> -	dws->rx_end = dws->rx + transfer->len;
> -	dws->cs_change = transfer->cs_change;
> -	dws->len = dws->cur_transfer->len;
> -	if (chip != dws->prev_chip)
> -		cs_change = 1;
> -
> -	cr0 = chip->cr0;
> -
> -	/* Handle per transfer options for bpw and speed */
> -	if (transfer->speed_hz) {
> -		speed = chip->speed_hz;
> -
> -		if (transfer->speed_hz != speed) {
> -			speed = transfer->speed_hz;
> -			if (speed > dws->max_freq) {
> -				printk(KERN_ERR "MRST SPI0: unsupported"
> -					"freq: %dHz\n", speed);
> -				message->status = -EIO;
> -				goto early_exit;
> -			}
> -
> -			/* clk_div doesn't support odd number */
> -			clk_div = dws->max_freq / speed;
> -			clk_div = (clk_div + 1) & 0xfffe;
> -
> -			chip->speed_hz = speed;
> -			chip->clk_div = clk_div;
> -		}
> -	}
> -	if (transfer->bits_per_word) {
> -		bits = transfer->bits_per_word;
> -
> -		switch (bits) {
> -		case 8:
> -		case 16:
> -			dws->n_bytes = dws->dma_width = bits >> 3;
> -			break;
> -		default:
> -			printk(KERN_ERR "MRST SPI0: unsupported bits:"
> -				"%db\n", bits);
> -			message->status = -EIO;
> -			goto early_exit;
> -		}
> -
> -		cr0 = (bits - 1)
> -			| (chip->type << SPI_FRF_OFFSET)
> -			| (spi->mode << SPI_MODE_OFFSET)
> -			| (chip->tmode << SPI_TMOD_OFFSET);
> -	}
> -	message->state = RUNNING_STATE;
> -
> -	/*
> -	 * Adjust transfer mode if necessary. Requires platform dependent
> -	 * chipselect mechanism.
> -	 */
> -	if (dws->cs_control) {
> -		if (dws->rx && dws->tx)
> -			chip->tmode = SPI_TMOD_TR;
> -		else if (dws->rx)
> -			chip->tmode = SPI_TMOD_RO;
> -		else
> -			chip->tmode = SPI_TMOD_TO;
> -
> -		cr0 &= ~SPI_TMOD_MASK;
> -		cr0 |= (chip->tmode << SPI_TMOD_OFFSET);
> -	}
> -
> -	/* Check if current transfer is a DMA transaction */
> -	dws->dma_mapped = map_dma_buffers(dws);
> -
> -	/*
> -	 * Interrupt mode
> -	 * we only need set the TXEI IRQ, as TX/RX always happen syncronizely
> -	 */
> -	if (!dws->dma_mapped && !chip->poll_mode) {
> -		int templen = dws->len / dws->n_bytes;
> -		txint_level = dws->fifo_len / 2;
> -		txint_level = (templen > txint_level) ? txint_level : templen;
> -
> -		imask |= SPI_INT_TXEI | SPI_INT_TXOI | SPI_INT_RXUI | SPI_INT_RXOI;
> -		dws->transfer_handler = interrupt_transfer;
> -	}
> -
> -	/*
> -	 * Reprogram registers only if
> -	 *	1. chip select changes
> -	 *	2. clk_div is changed
> -	 *	3. control value changes
> -	 */
> -	if (dw_readw(dws, ctrl0) != cr0 || cs_change || clk_div || imask) {
> -		spi_enable_chip(dws, 0);
> -
> -		if (dw_readw(dws, ctrl0) != cr0)
> -			dw_writew(dws, ctrl0, cr0);
> -
> -		spi_set_clk(dws, clk_div ? clk_div : chip->clk_div);
> -		spi_chip_sel(dws, spi->chip_select);
> -
> -		/* Set the interrupt mask, for poll mode just disable all int */
> -		spi_mask_intr(dws, 0xff);
> -		if (imask)
> -			spi_umask_intr(dws, imask);
> -		if (txint_level)
> -			dw_writew(dws, txfltr, txint_level);
> -
> -		spi_enable_chip(dws, 1);
> -		if (cs_change)
> -			dws->prev_chip = chip;
> -	}
> -
> -	if (dws->dma_mapped)
> -		dws->dma_ops->dma_transfer(dws, cs_change);
> -
> -	if (chip->poll_mode)
> -		poll_transfer(dws);
> -
> -	return;
> -
> -early_exit:
> -	giveback(dws);
> -	return;
> -}
> -
> -static void pump_messages(struct work_struct *work)
> -{
> -	struct dw_spi *dws =
> -		container_of(work, struct dw_spi, pump_messages);
> -	unsigned long flags;
> -
> -	/* Lock queue and check for queue work */
> -	spin_lock_irqsave(&dws->lock, flags);
> -	if (list_empty(&dws->queue) || dws->run == QUEUE_STOPPED) {
> -		dws->busy = 0;
> -		spin_unlock_irqrestore(&dws->lock, flags);
> -		return;
> -	}
> -
> -	/* Make sure we are not already running a message */
> -	if (dws->cur_msg) {
> -		spin_unlock_irqrestore(&dws->lock, flags);
> -		return;
> -	}
> -
> -	/* Extract head of queue */
> -	dws->cur_msg = list_entry(dws->queue.next, struct spi_message, queue);
> -	list_del_init(&dws->cur_msg->queue);
> -
> -	/* Initial message state*/
> -	dws->cur_msg->state = START_STATE;
> -	dws->cur_transfer = list_entry(dws->cur_msg->transfers.next,
> -						struct spi_transfer,
> -						transfer_list);
> -	dws->cur_chip = spi_get_ctldata(dws->cur_msg->spi);
> -
> -	/* Mark as busy and launch transfers */
> -	tasklet_schedule(&dws->pump_transfers);
> -
> -	dws->busy = 1;
> -	spin_unlock_irqrestore(&dws->lock, flags);
> -}
> -
> -/* spi_device use this to queue in their spi_msg */
> -static int dw_spi_transfer(struct spi_device *spi, struct spi_message *msg)
> -{
> -	struct dw_spi *dws = spi_master_get_devdata(spi->master);
> -	unsigned long flags;
> -
> -	spin_lock_irqsave(&dws->lock, flags);
> -
> -	if (dws->run == QUEUE_STOPPED) {
> -		spin_unlock_irqrestore(&dws->lock, flags);
> -		return -ESHUTDOWN;
> -	}
> -
> -	msg->actual_length = 0;
> -	msg->status = -EINPROGRESS;
> -	msg->state = START_STATE;
> -
> -	list_add_tail(&msg->queue, &dws->queue);
> -
> -	if (dws->run == QUEUE_RUNNING && !dws->busy) {
> -
> -		if (dws->cur_transfer || dws->cur_msg)
> -			queue_work(dws->workqueue,
> -					&dws->pump_messages);
> -		else {
> -			/* If no other data transaction in air, just go */
> -			spin_unlock_irqrestore(&dws->lock, flags);
> -			pump_messages(&dws->pump_messages);
> -			return 0;
> -		}
> -	}
> -
> -	spin_unlock_irqrestore(&dws->lock, flags);
> -	return 0;
> -}
> -
> -/* This may be called twice for each spi dev */
> -static int dw_spi_setup(struct spi_device *spi)
> -{
> -	struct dw_spi_chip *chip_info = NULL;
> -	struct chip_data *chip;
> -
> -	if (spi->bits_per_word != 8 && spi->bits_per_word != 16)
> -		return -EINVAL;
> -
> -	/* Only alloc on first setup */
> -	chip = spi_get_ctldata(spi);
> -	if (!chip) {
> -		chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
> -		if (!chip)
> -			return -ENOMEM;
> -	}
> -
> -	/*
> -	 * Protocol drivers may change the chip settings, so...
> -	 * if chip_info exists, use it
> -	 */
> -	chip_info = spi->controller_data;
> -
> -	/* chip_info doesn't always exist */
> -	if (chip_info) {
> -		if (chip_info->cs_control)
> -			chip->cs_control = chip_info->cs_control;
> -
> -		chip->poll_mode = chip_info->poll_mode;
> -		chip->type = chip_info->type;
> -
> -		chip->rx_threshold = 0;
> -		chip->tx_threshold = 0;
> -
> -		chip->enable_dma = chip_info->enable_dma;
> -	}
> -
> -	if (spi->bits_per_word <= 8) {
> -		chip->n_bytes = 1;
> -		chip->dma_width = 1;
> -	} else if (spi->bits_per_word <= 16) {
> -		chip->n_bytes = 2;
> -		chip->dma_width = 2;
> -	} else {
> -		/* Never take >16b case for MRST SPIC */
> -		dev_err(&spi->dev, "invalid wordsize\n");
> -		return -EINVAL;
> -	}
> -	chip->bits_per_word = spi->bits_per_word;
> -
> -	if (!spi->max_speed_hz) {
> -		dev_err(&spi->dev, "No max speed HZ parameter\n");
> -		return -EINVAL;
> -	}
> -	chip->speed_hz = spi->max_speed_hz;
> -
> -	chip->tmode = 0; /* Tx & Rx */
> -	/* Default SPI mode is SCPOL = 0, SCPH = 0 */
> -	chip->cr0 = (chip->bits_per_word - 1)
> -			| (chip->type << SPI_FRF_OFFSET)
> -			| (spi->mode  << SPI_MODE_OFFSET)
> -			| (chip->tmode << SPI_TMOD_OFFSET);
> -
> -	spi_set_ctldata(spi, chip);
> -	return 0;
> -}
> -
> -static void dw_spi_cleanup(struct spi_device *spi)
> -{
> -	struct chip_data *chip = spi_get_ctldata(spi);
> -	kfree(chip);
> -}
> -
> -static int __devinit init_queue(struct dw_spi *dws)
> -{
> -	INIT_LIST_HEAD(&dws->queue);
> -	spin_lock_init(&dws->lock);
> -
> -	dws->run = QUEUE_STOPPED;
> -	dws->busy = 0;
> -
> -	tasklet_init(&dws->pump_transfers,
> -			pump_transfers,	(unsigned long)dws);
> -
> -	INIT_WORK(&dws->pump_messages, pump_messages);
> -	dws->workqueue = create_singlethread_workqueue(
> -					dev_name(dws->master->dev.parent));
> -	if (dws->workqueue == NULL)
> -		return -EBUSY;
> -
> -	return 0;
> -}
> -
> -static int start_queue(struct dw_spi *dws)
> -{
> -	unsigned long flags;
> -
> -	spin_lock_irqsave(&dws->lock, flags);
> -
> -	if (dws->run == QUEUE_RUNNING || dws->busy) {
> -		spin_unlock_irqrestore(&dws->lock, flags);
> -		return -EBUSY;
> -	}
> -
> -	dws->run = QUEUE_RUNNING;
> -	dws->cur_msg = NULL;
> -	dws->cur_transfer = NULL;
> -	dws->cur_chip = NULL;
> -	dws->prev_chip = NULL;
> -	spin_unlock_irqrestore(&dws->lock, flags);
> -
> -	queue_work(dws->workqueue, &dws->pump_messages);
> -
> -	return 0;
> -}
> -
> -static int stop_queue(struct dw_spi *dws)
> -{
> -	unsigned long flags;
> -	unsigned limit = 50;
> -	int status = 0;
> -
> -	spin_lock_irqsave(&dws->lock, flags);
> -	dws->run = QUEUE_STOPPED;
> -	while ((!list_empty(&dws->queue) || dws->busy) && limit--) {
> -		spin_unlock_irqrestore(&dws->lock, flags);
> -		msleep(10);
> -		spin_lock_irqsave(&dws->lock, flags);
> -	}
> -
> -	if (!list_empty(&dws->queue) || dws->busy)
> -		status = -EBUSY;
> -	spin_unlock_irqrestore(&dws->lock, flags);
> -
> -	return status;
> -}
> -
> -static int destroy_queue(struct dw_spi *dws)
> -{
> -	int status;
> -
> -	status = stop_queue(dws);
> -	if (status != 0)
> -		return status;
> -	destroy_workqueue(dws->workqueue);
> -	return 0;
> -}
> -
> -/* Restart the controller, disable all interrupts, clean rx fifo */
> -static void spi_hw_init(struct dw_spi *dws)
> -{
> -	spi_enable_chip(dws, 0);
> -	spi_mask_intr(dws, 0xff);
> -	spi_enable_chip(dws, 1);
> -
> -	/*
> -	 * Try to detect the FIFO depth if not set by interface driver,
> -	 * the depth could be from 2 to 256 from HW spec
> -	 */
> -	if (!dws->fifo_len) {
> -		u32 fifo;
> -		for (fifo = 2; fifo <= 257; fifo++) {
> -			dw_writew(dws, txfltr, fifo);
> -			if (fifo != dw_readw(dws, txfltr))
> -				break;
> -		}
> -
> -		dws->fifo_len = (fifo == 257) ? 0 : fifo;
> -		dw_writew(dws, txfltr, 0);
> -	}
> -}
> -
> -int __devinit dw_spi_add_host(struct dw_spi *dws)
> -{
> -	struct spi_master *master;
> -	int ret;
> -
> -	BUG_ON(dws == NULL);
> -
> -	master = spi_alloc_master(dws->parent_dev, 0);
> -	if (!master) {
> -		ret = -ENOMEM;
> -		goto exit;
> -	}
> -
> -	dws->master = master;
> -	dws->type = SSI_MOTO_SPI;
> -	dws->prev_chip = NULL;
> -	dws->dma_inited = 0;
> -	dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
> -
> -	ret = request_irq(dws->irq, dw_spi_irq, IRQF_SHARED,
> -			"dw_spi", dws);
> -	if (ret < 0) {
> -		dev_err(&master->dev, "can not get IRQ\n");
> -		goto err_free_master;
> -	}
> -
> -	master->mode_bits = SPI_CPOL | SPI_CPHA;
> -	master->bus_num = dws->bus_num;
> -	master->num_chipselect = dws->num_cs;
> -	master->cleanup = dw_spi_cleanup;
> -	master->setup = dw_spi_setup;
> -	master->transfer = dw_spi_transfer;
> -
> -	/* Basic HW init */
> -	spi_hw_init(dws);
> -
> -	if (dws->dma_ops && dws->dma_ops->dma_init) {
> -		ret = dws->dma_ops->dma_init(dws);
> -		if (ret) {
> -			dev_warn(&master->dev, "DMA init failed\n");
> -			dws->dma_inited = 0;
> -		}
> -	}
> -
> -	/* Initial and start queue */
> -	ret = init_queue(dws);
> -	if (ret) {
> -		dev_err(&master->dev, "problem initializing queue\n");
> -		goto err_diable_hw;
> -	}
> -	ret = start_queue(dws);
> -	if (ret) {
> -		dev_err(&master->dev, "problem starting queue\n");
> -		goto err_diable_hw;
> -	}
> -
> -	spi_master_set_devdata(master, dws);
> -	ret = spi_register_master(master);
> -	if (ret) {
> -		dev_err(&master->dev, "problem registering spi master\n");
> -		goto err_queue_alloc;
> -	}
> -
> -	mrst_spi_debugfs_init(dws);
> -	return 0;
> -
> -err_queue_alloc:
> -	destroy_queue(dws);
> -	if (dws->dma_ops && dws->dma_ops->dma_exit)
> -		dws->dma_ops->dma_exit(dws);
> -err_diable_hw:
> -	spi_enable_chip(dws, 0);
> -	free_irq(dws->irq, dws);
> -err_free_master:
> -	spi_master_put(master);
> -exit:
> -	return ret;
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_add_host);
> -
> -void __devexit dw_spi_remove_host(struct dw_spi *dws)
> -{
> -	int status = 0;
> -
> -	if (!dws)
> -		return;
> -	mrst_spi_debugfs_remove(dws);
> -
> -	/* Remove the queue */
> -	status = destroy_queue(dws);
> -	if (status != 0)
> -		dev_err(&dws->master->dev, "dw_spi_remove: workqueue will not "
> -			"complete, message memory not freed\n");
> -
> -	if (dws->dma_ops && dws->dma_ops->dma_exit)
> -		dws->dma_ops->dma_exit(dws);
> -	spi_enable_chip(dws, 0);
> -	/* Disable clk */
> -	spi_set_clk(dws, 0);
> -	free_irq(dws->irq, dws);
> -
> -	/* Disconnect from the SPI framework */
> -	spi_unregister_master(dws->master);
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_remove_host);
> -
> -int dw_spi_suspend_host(struct dw_spi *dws)
> -{
> -	int ret = 0;
> -
> -	ret = stop_queue(dws);
> -	if (ret)
> -		return ret;
> -	spi_enable_chip(dws, 0);
> -	spi_set_clk(dws, 0);
> -	return ret;
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
> -
> -int dw_spi_resume_host(struct dw_spi *dws)
> -{
> -	int ret;
> -
> -	spi_hw_init(dws);
> -	ret = start_queue(dws);
> -	if (ret)
> -		dev_err(&dws->master->dev, "fail to start queue (%d)\n", ret);
> -	return ret;
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_resume_host);
> -
> -MODULE_AUTHOR("Feng Tang <feng.tang@intel.com>");
> -MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
> -MODULE_LICENSE("GPL v2");
> +/*
> + * dw_spi.c - Designware SPI core controller driver

drop the filename.  It's incorrect, and not useful.  The useful bit is
the driver description.

> + *
> + * Copyright (c) 2009, Intel Corporation.
> + *
> + * 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, write to the Free Software Foundation, Inc.,
> + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
> + */
> +
> +#include <linux/dma-mapping.h>
> +#include <linux/interrupt.h>
> +#include <linux/highmem.h>
> +#include <linux/delay.h>
> +#include <linux/slab.h>
> +#include <linux/spi/spi.h>
> +
> +#include "spi-dw.h"
> +
> +#ifdef CONFIG_DEBUG_FS
> +#include <linux/debugfs.h>
> +#endif
> +
> +
> +#define QUEUE_RUNNING	0
> +#define QUEUE_STOPPED	1
> +
> +
> +/* Slave spi_dev related */
> +struct chip_data {
> +	struct spi_device *spi_dev;
> +	u32 cr0;
> +	u32 cs;			/* chip select pin */
> +	u32 n_bytes;		/* current is a 1/2/4 byte op */
> +	u32 type;		/* SPI/SSP/MicroWire */
> +
> +	u32 dma_width;
> +	u32 enable_dma;
> +	u32 bits_per_word;
> +	u32 clk_div;		/* baud rate divider */
> +	u32 speed_hz;		/* baud rate */
> +};
> +
> +#ifdef CONFIG_DEBUG_FS
> +static int spi_show_regs_open(struct inode *inode, struct file *file)
> +{
> +	file->private_data = inode->i_private;
> +	return 0;
> +}
> +
> +#define SPI_REGS_BUFSIZE	1024
> +static ssize_t  spi_show_regs(struct file *file, char __user *user_buf,
> +				size_t count, loff_t *ppos)
> +{
> +	struct dw_spi *dws;
> +	char *buf;
> +	u32 len = 0;
> +	ssize_t ret;
> +
> +	dws = file->private_data;
> +
> +	buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
> +	if (!buf)
> +		return 0;
> +
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"MRST SPI0 registers:\n");
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"=================================\n");
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"SER: \t\t0x%08x\n", dw_readl(dws, ser));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"SR: \t\t0x%08x\n", dw_readl(dws, sr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"IMR: \t\t0x%08x\n", dw_readl(dws, imr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"ISR: \t\t0x%08x\n", dw_readl(dws, isr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"DMATDLR: \t0x%08x\n", dw_readl(dws, dmatdlr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"DMARDLR: \t0x%08x\n", dw_readl(dws, dmardlr));
> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +			"=================================\n");
> +
> +	ret =  simple_read_from_buffer(user_buf, count, ppos, buf, len);
> +	kfree(buf);
> +	return ret;
> +}
> +
> +static const struct file_operations mrst_spi_regs_ops = {
> +	.owner		= THIS_MODULE,
> +	.open		= spi_show_regs_open,
> +	.read		= spi_show_regs,
> +	.llseek		= default_llseek,
> +};
> +
> +static int mrst_spi_debugfs_init(struct dw_spi *dws)
> +{
> +	dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
> +	if (!dws->debugfs)
> +		return -ENOMEM;
> +
> +	debugfs_create_file("registers", S_IFREG | S_IRUGO,
> +		dws->debugfs, (void *)dws, &mrst_spi_regs_ops);
> +	return 0;
> +}
> +
> +static void mrst_spi_debugfs_remove(struct dw_spi *dws)
> +{
> +	if (dws->debugfs)
> +		debugfs_remove_recursive(dws->debugfs);
> +}
> +
> +#else
> +static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
> +{
> +	return 0;
> +}
> +
> +static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
> +{
> +}
> +#endif /* CONFIG_DEBUG_FS */
> +
> +static irqreturn_t dw_spi_irq(int irq, void *dev_id)
> +{
> +	struct dw_spi *dws = dev_id;
> +	u16  irq_mask = 0x3f;
> +
> +	dws->xfer.irq_status = dw_readw(dws, isr) & irq_mask;
> +
> +	if (!dws->xfer.irq_status)
> +		return IRQ_NONE;
> +	if (dws->xfer.irq_status &
> +			(SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI)) {
> +		dw_readw(dws, txoicr);
> +		dw_readw(dws, rxoicr);
> +		dw_readw(dws, rxuicr);
> +		dws->xfer.err = -EIO;
> +		dw_spi_disable(dws);
> +		complete(&dws->xfer.complete);
> +		return IRQ_HANDLED;
> +	}
> +
> +	/* disable interrupts */
> +	dw_spi_mask_intr(dws, irq_mask);
> +	return IRQ_WAKE_THREAD;
> +}
> +struct spi_message *get_message(struct dw_spi *dws)
> +{
> +	struct spi_message *message = NULL;
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&dws->lock, flags);
> +	if (!list_empty(&dws->queue)) {
> +		message = list_entry(dws->queue.next,
> +				struct spi_message, queue);
> +		list_del_init(&message->queue);
> +	}
> +	spin_unlock_irqrestore(&dws->lock, flags);
> +	return message;
> +}
> +static inline u32 tx_max(struct dw_spi *dws)
> +{
> +	u32 tx_left, tx_room;
> +
> +	tx_left = (dws->xfer.len - dws->xfer.sent) / dws->xfer.n_bytes;
> +	tx_room = (dws->fifo_len - dw_readw(dws, txflr));
> +
> +	return min(tx_left, tx_room);
> +}
> +
> +/* Return the max entries we should read out of rx fifo */
> +static inline u32 rx_max(struct dw_spi *dws)
> +{
> +	u32 rx_left = (dws->xfer.len - dws->xfer.rcvd) / dws->xfer.n_bytes;
> +	return min(rx_left, (u32)dw_readw(dws, rxflr));
> +}
> +
> +static int transfer_setup(struct dw_spi *dws, struct chip_data *controller,
> +			struct spi_transfer *transfer, struct spi_message *msg)
> +{
> +	int err = 0;
> +	u32 cr0 = controller->cr0;
> +	u32 clk_div;
> +
> +	dws->xfer.tx_buf = transfer->tx_buf;
> +	dws->xfer.tx_dma = transfer->tx_dma;
> +	dws->xfer.rx_buf = transfer->rx_buf;
> +	dws->xfer.rx_dma = transfer->rx_dma;
> +
> +	dws->xfer.len = transfer->len;
> +	dws->xfer.n_bytes = controller->n_bytes;
> +	dws->xfer.sent = 0;
> +	dws->xfer.rcvd = 0;
> +	dws->xfer.msg = msg;
> +	dws->xfer.err = 0;
> +	dws->xfer.irq_status = 0;
> +	INIT_COMPLETION(dws->xfer.complete);
> +
> +	/* {tx, rx}_threshold should probably be a module param with
> +	 *  some reasonable default but these work for now.
> +	 */
> +	dws->xfer.tx_threshold = 10;
> +	dws->xfer.rx_threshold = dws->xfer.tx_threshold - 1;
> +
> +	/* we need to make the decsion about the type of transfer more
> +	 *  inteligently but this works for now
> +	 */
> +	if (transfer->len > dws->fifo_len)
> +		dws->xfer.type = INT_XFER;
> +	else
> +		dws->xfer.type = PIO_XFER;
> +
> +	if (controller->enable_dma &&
> +		msg->is_dma_mapped &&
> +		dws->dma_inited &&
> +		dws->dma_ops)
> +		dws->xfer.type = DMA_XFER;
> +
> +	/* Setup the controller based on parameters in the transfer
> +	 * each transfer can set the bit_per_word and the speed_hz to
> +	 * change these values in the controller the controller MUST
> +	 * be disabled
> +	 */
> +	if (unlikely(!controller->clk_div)) {
> +		controller->clk_div = dws->max_freq / controller->speed_hz;
> +		controller->clk_div = (controller->clk_div + 1) & 0xfffe;
> +		dw_spi_set_clk(dws, controller->clk_div);
> +		dev_err(&dws->master->dev, "setting default clk_div");
> +		err = 1;
> +	}
> +
> +	if (transfer->speed_hz) {
> +		if (transfer->speed_hz != controller->speed_hz) {
> +			if (transfer->speed_hz > dws->max_freq) {
> +				err = -EIO;
> +				goto out;
> +			}
> +
> +			clk_div = dws->max_freq / transfer->speed_hz;
> +			clk_div = (clk_div + 1) & 0xfffe;
> +			controller->clk_div = clk_div;
> +			controller->speed_hz = transfer->speed_hz;
> +			err = 1;
> +		}
> +	}
> +
> +	if (transfer->bits_per_word) {
> +		if (transfer->bits_per_word != 8 &&
> +			transfer->bits_per_word != 16) {
> +			err = -EIO;
> +			goto out;
> +		}
> +		cr0 &= ~SPI_DFS_MASK;
> +		cr0 |= transfer->bits_per_word - 1;
> +		err = 1;
> +	} else {
> +		cr0 &= ~SPI_DFS_MASK;
> +		cr0 |= controller->spi_dev->bits_per_word - 1;
> +	}
> +
> +	cr0 &= ~SPI_MODE_MASK;
> +	cr0 |= (controller->spi_dev->mode << SPI_MODE_OFFSET);
> +	controller->cr0 = cr0;
> +
> +	if (err || dw_readw(dws, ctrl0) != cr0) {
> +		dw_spi_disable(dws);
> +		dw_spi_chip_sel(dws, controller->spi_dev->chip_select);
> +		dw_writew(dws, ctrl0, cr0);
> +		dw_spi_set_clk(dws, controller->clk_div);
> +		dw_spi_enable(dws);
> +		err = 0;
> +	}
> +out:
> +	return err;
> +}
> +
> +static void tx_fifo_fill(struct dw_spi *dws)
> +{
> +	int room;
> +	u16 txw = 0;
> +	if (dws->xfer.sent < dws->xfer.len) {
> +		room = tx_max(dws);
> +		while (room--) {
> +			if (dws->xfer.tx_buf) {
> +				if (dws->xfer.n_bytes == 2)
> +					txw = *(u16 *)dws->xfer.tx_buf;
> +				else
> +					txw = *(u8 *)dws->xfer.tx_buf;
> +				dws->xfer.tx_buf += dws->xfer.n_bytes;
> +			}
> +			dw_writew(dws, dr, txw);
> +			dws->xfer.sent += dws->xfer.n_bytes;
> +		}
> +	}
> +}
> +
> +
> +static void rx_fifo_drain(struct dw_spi *dws)
> +{
> +	u16 rx_val;
> +	int avail;
> +
> +	if (dws->xfer.rcvd < dws->xfer.len) {
> +		avail = rx_max(dws);
> +		while (avail--) {
> +			rx_val = dw_readw(dws, dr);
> +			if (dws->xfer.rx_buf) {
> +				if (dws->xfer.n_bytes == 2)
> +					*(u16 *)(dws->xfer.rx_buf) =
> +						(u16)rx_val;
> +				else
> +					*dws->xfer.rx_buf = (u8)rx_val;
> +				dws->xfer.rx_buf += dws->xfer.n_bytes;
> +			}
> +			dws->xfer.rcvd += dws->xfer.n_bytes;
> +		}
> +	}
> +}
> +
> +static inline void do_pio(struct dw_spi *dws)
> +{
> +	while (dws->xfer.sent < dws->xfer.len ||
> +					dws->xfer.rcvd < dws->xfer.len) {
> +		tx_fifo_fill(dws);
> +		rx_fifo_drain(dws);
> +		cpu_relax();
> +	}
> +	complete(&dws->xfer.complete);
> +}
> +
> +static irqreturn_t dw_spi_irq_thread_handler(int irq, void *dev_id)
> +{
> +	struct dw_spi *dws = dev_id;
> +
> +	if (dws->xfer.irq_status & SPI_INT_TXEI) {
> +		rx_fifo_drain(dws);
> +		tx_fifo_fill(dws);
> +	}
> +
> +	if (dws->xfer.irq_status & SPI_INT_RXFI) {
> +		tx_fifo_fill(dws);
> +		rx_fifo_drain(dws);
> +	}
> +
> +	if (dws->xfer.len == dws->xfer.rcvd &&
> +		dws->xfer.len == dws->xfer.sent) {
> +		complete(&dws->xfer.complete);
> +		goto out;
> +	}
> +
> +	dw_spi_umask_intr(dws, SPI_INT_ALL);
> +out:
> +	return IRQ_HANDLED;
> +}
> +
> +static inline void do_int_xfer(struct dw_spi *dws)
> +{
> +	dw_spi_disable(dws);
> +	dw_writew(dws, txfltr, dws->xfer.tx_threshold);
> +	dw_writew(dws, rxfltr, dws->xfer.rx_threshold);
> +	dw_spi_enable(dws);
> +	dw_readw(dws, icr);
> +	tx_fifo_fill(dws);
> +	dw_writew(dws, imr, SPI_INT_ALL);
> +}
> +
> +static inline int do_transfer(struct dw_spi *dws)
> +{
> +	switch (dws->xfer.type) {
> +	case PIO_XFER:
> +		do_pio(dws);
> +		break;
> +	case INT_XFER:
> +		do_int_xfer(dws);
> +		break;
> +	case DMA_XFER:
> +		dws->dma_ops->dma_transfer(dws);
> +		break;
> +	default:
> +		BUG();
> +	}
> +
> +	wait_for_completion(&dws->xfer.complete);
> +
> +
> +	return dws->xfer.err;
> +}
> +
> +static void drain_message_queue(struct dw_spi *dws)
> +{
> +	struct spi_message *message;
> +
> +	message = get_message(dws);
> +	while (message) {
> +		message->status = -ESHUTDOWN;
> +		message->complete(message->context);
> +		message = get_message(dws);
> +	}
> +}
> +
> +static void pump_messages(struct work_struct *work)
> +{
> +	struct dw_spi *dws =
> +		container_of(work, struct dw_spi, pump_messages);
> +	struct spi_transfer *transfer;
> +	struct spi_message *message;
> +	struct chip_data *controller;
> +	int err = 0;
> +
> +
> +	message = get_message(dws);
> +
> +	while (message && dws->run != QUEUE_STOPPED) {
> +		controller = spi_get_ctldata(message->spi);
> +		list_for_each_entry(transfer, &message->transfers,
> +				transfer_list){
> +
> +			err = transfer_setup(dws, controller,
> +						transfer, message);
> +			if (err < 0) {
> +				dev_err(&dws->master->dev,
> +					"transfer_setup failed");
> +				dws->xfer.err = -EIO;
> +				break;
> +			}
> +
> +			err = do_transfer(dws);
> +			if (err < 0) {
> +				dev_err(&dws->master->dev,
> +					"do_transfer failed");
> +				break;
> +			}
> +			message->actual_length += dws->xfer.len;
> +
> +			if (transfer->delay_usecs)
> +				udelay(transfer->delay_usecs);
> +		}
> +
> +		message->status = dws->xfer.err;
> +		message->complete(message->context);
> +		message =  get_message(dws);
> +	}
> +	if (dws->run == QUEUE_STOPPED)
> +		drain_message_queue(dws);
> +}
> +
> +/* spi_device use this to queue in their spi_msg */
> +static int dw_spi_transfer(struct spi_device *spi, struct spi_message *msg)
> +{
> +	struct dw_spi *dws = spi_master_get_devdata(spi->master);
> +	unsigned long flags;
> +
> +
> +	spin_lock_irqsave(&dws->lock, flags);
> +
> +	msg->actual_length = 0;
> +	msg->status = -EINPROGRESS;
> +
> +	list_add_tail(&msg->queue, &dws->queue);
> +
> +	queue_work(dws->workqueue,
> +		&dws->pump_messages);
> +
> +	spin_unlock_irqrestore(&dws->lock, flags);
> +
> +	return 0;
> +}
> +
> +/* This may be called twice for each spi dev */
> +static int dw_spi_setup(struct spi_device *spi)
> +{
> +	struct dw_spi_chip *chip_info = NULL;
> +	struct chip_data *chip;
> +
> +	if (spi->bits_per_word != 8 && spi->bits_per_word != 16)
> +		return -EINVAL;
> +
> +	if (!spi->max_speed_hz) {
> +		dev_err(&spi->dev, "No max speed HZ parameter\n");
> +		return -EINVAL;
> +	}
> +
> +	/* Only alloc on first setup */
> +	chip = spi_get_ctldata(spi);
> +	if (!chip) {
> +		chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
> +		if (!chip)
> +			return -ENOMEM;
> +	}
> +	chip->spi_dev = spi;
> +
> +	/*
> +	 * Protocol drivers may change the chip settings, so...
> +	 * if chip_info exists, use it
> +	 */
> +	chip_info = spi->controller_data;
> +
> +	/* chip_info doesn't always exist */
> +	if (chip_info) {
> +		chip->type = chip_info->type;
> +		chip->enable_dma = chip_info->enable_dma;
> +	}
> +
> +	chip->bits_per_word = spi->bits_per_word;
> +	chip->n_bytes = chip->bits_per_word / 8;
> +	chip->dma_width = chip->bits_per_word / 8;
> +
> +	chip->speed_hz = spi->max_speed_hz;
> +
> +	/* Default SPI mode is SCPOL = 0, SCPH = 0 */
> +	chip->cr0 = (chip->bits_per_word - 1)
> +		| (chip->type << SPI_FRF_OFFSET)
> +		| (spi->mode  << SPI_MODE_OFFSET);
> +
> +	spi_set_ctldata(spi, chip);
> +	return 0;
> +}
> +
> +static void dw_spi_cleanup(struct spi_device *spi)
> +{
> +	struct chip_data *chip = spi_get_ctldata(spi);
> +	kfree(chip);
> +}
> +
> +static int __devinit init_queue(struct dw_spi *dws)
> +{
> +	INIT_LIST_HEAD(&dws->queue);
> +	spin_lock_init(&dws->lock);
> +
> +	dws->run = QUEUE_STOPPED;
> +
> +	INIT_WORK(&dws->pump_messages, pump_messages);
> +	dws->workqueue = create_singlethread_workqueue(
> +					dev_name(dws->master->dev.parent));
> +	if (dws->workqueue == NULL)
> +		return -EBUSY;
> +	dws->run = QUEUE_RUNNING;
> +	return 0;
> +}
> +
> +
> +int dw_spi_stop_queue(struct dw_spi *dws)
> +{
> +	unsigned long flags;
> +	int status = 0;
> +
> +	spin_lock_irqsave(&dws->lock, flags);
> +	dws->run = QUEUE_STOPPED;
> +
> +	if (!list_empty(&dws->queue))
> +		status = -EBUSY;
> +	spin_unlock_irqrestore(&dws->lock, flags);
> +
> +	return status;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_stop_queue);
> +
> +static int destroy_queue(struct dw_spi *dws)
> +{
> +	int status;
> +
> +	status = dw_spi_stop_queue(dws);
> +	if (status != 0)
> +		return status;
> +	destroy_workqueue(dws->workqueue);
> +	return 0;
> +}
> +
> +/* Restart the controller, disable all interrupts, clean rx fifo */
> +static void spi_hw_init(struct dw_spi *dws)
> +{
> +	dw_spi_disable(dws);
> +	dw_spi_mask_intr(dws, 0xff);
> +	dw_readw(dws, icr);
> +	dw_spi_enable(dws);
> +
> +	BUG_ON(!dws->fifo_len);
> +}
> +
> +int __devinit dw_spi_add_host(struct dw_spi *dws)
> +{
> +	struct spi_master *master;
> +	int ret;
> +
> +	BUG_ON(dws == NULL);
> +
> +	master = spi_alloc_master(dws->parent_dev, 0);
> +	if (!master) {
> +		ret = -ENOMEM;
> +		goto exit;
> +	}
> +
> +	dws->master = master;
> +	dws->type = SSI_MOTO_SPI;
> +
> +	init_completion(&dws->xfer.complete);
> +	dws->dma_inited = 0;
> +	/* Change to address of FIFO */
> +	dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
> +
> +	ret = request_threaded_irq(dws->irq, dw_spi_irq,
> +				dw_spi_irq_thread_handler,
> +				IRQF_SHARED, "dw_spi", dws);
> +	if (ret < 0) {
> +		dev_err(&master->dev, "can not get IRQ\n");
> +		goto err_free_master;
> +	}
> +
> +	master->mode_bits = SPI_CPOL | SPI_CPHA;
> +	master->bus_num = dws->bus_num;
> +	master->num_chipselect = dws->num_cs;
> +	master->cleanup = dw_spi_cleanup;
> +	master->setup = dw_spi_setup;
> +	master->transfer = dw_spi_transfer;
> +
> +	/* Basic HW init */
> +	spi_hw_init(dws);
> +
> +	if (dws->dma_ops && dws->dma_ops->dma_init) {
> +		ret = dws->dma_ops->dma_init(dws);
> +		if (ret) {
> +			dev_warn(&master->dev, "DMA init failed\n");
> +			dws->dma_inited = 0;
> +		}
> +	}
> +
> +	/* Initial and start queue */
> +	ret = init_queue(dws);
> +	if (ret) {
> +		dev_err(&master->dev, "problem initializing queue\n");
> +		goto err_diable_hw;
> +	}
> +
> +	spi_master_set_devdata(master, dws);
> +	ret = spi_register_master(master);
> +	if (ret) {
> +		dev_err(&master->dev, "problem registering spi master\n");
> +		goto err_queue_alloc;
> +	}
> +
> +	mrst_spi_debugfs_init(dws);
> +	return 0;
> +
> +err_queue_alloc:
> +	destroy_queue(dws);
> +	if (dws->dma_ops && dws->dma_ops->dma_exit)
> +		dws->dma_ops->dma_exit(dws);
> +err_diable_hw:
> +	dw_spi_disable(dws);
> +	free_irq(dws->irq, dws);
> +err_free_master:
> +	spi_master_put(master);
> +exit:
> +	return ret;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_add_host);
> +
> +void __devexit dw_spi_remove_host(struct dw_spi *dws)
> +{
> +	int status = 0;
> +
> +	if (!dws)
> +		return;
> +	mrst_spi_debugfs_remove(dws);
> +
> +	/* Remove the queue */
> +	status = destroy_queue(dws);
> +	if (status != 0)
> +		dev_err(&dws->master->dev, "dw_spi_remove: workqueue will not "
> +			"complete, message memory not freed\n");
> +
> +	if (dws->dma_ops && dws->dma_ops->dma_exit)
> +		dws->dma_ops->dma_exit(dws);
> +	dw_spi_disable(dws);
> +	dw_readw(dws, icr);
> +	free_irq(dws->irq, dws);
> +
> +	/* Disconnect from the SPI framework */
> +	spi_unregister_master(dws->master);
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_remove_host);
> +
> +int dw_spi_suspend_host(struct dw_spi *dws)
> +{
> +	int ret = 0;
> +
> +	ret = dw_spi_stop_queue(dws);
> +	if (ret)
> +		return ret;
> +	dw_spi_disable(dws);
> +	return ret;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
> +
> +int dw_spi_resume_host(struct dw_spi *dws)
> +{
> +	spi_hw_init(dws);
> +	dws->run = QUEUE_RUNNING;
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_resume_host);
> +
> +MODULE_AUTHOR("Feng Tang <feng.tang@intel.com>");
> +MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
> +MODULE_LICENSE("GPL v2");
> diff --git a/drivers/spi/spi-dw.h b/drivers/spi/spi-dw.h
> index 97baff6..b016b85 100644
> --- a/drivers/spi/spi-dw.h
> +++ b/drivers/spi/spi-dw.h
> @@ -7,6 +7,7 @@
>  
>  /* Bit fields in CTRLR0 */
>  #define SPI_DFS_OFFSET			0
> +#define SPI_DFS_MASK			0xf
>  
>  #define SPI_FRF_OFFSET			4
>  #define SPI_FRF_SPI			0x0
> @@ -17,6 +18,7 @@
>  #define SPI_MODE_OFFSET			6
>  #define SPI_SCPH_OFFSET			6
>  #define SPI_SCOL_OFFSET			7
> +#define SPI_MODE_MASK			(0x3 << SPI_MODE_OFFSET)
>  
>  #define SPI_TMOD_OFFSET			8
>  #define SPI_TMOD_MASK			(0x3 << SPI_TMOD_OFFSET)
> @@ -46,6 +48,7 @@
>  #define SPI_INT_RXOI			(1 << 3)
>  #define SPI_INT_RXFI			(1 << 4)
>  #define SPI_INT_MSTI			(1 << 5)
> +#define SPI_INT_ALL  0x3f
>  
>  /* TX RX interrupt level threshold, max can be 256 */
>  #define SPI_INT_THRESHOLD		32
> @@ -83,65 +86,67 @@ struct dw_spi;
>  struct dw_spi_dma_ops {
>  	int (*dma_init)(struct dw_spi *dws);
>  	void (*dma_exit)(struct dw_spi *dws);
> -	int (*dma_transfer)(struct dw_spi *dws, int cs_change);
> +	int (*dma_transfer)(struct dw_spi *dws);
> +};
> +
> +enum xfer_type {
> +	PIO_XFER,
> +	INT_XFER,
> +	DMA_XFER,
> +};
> +
> +	
> +struct xfer_state {
> +	const u8 *tx_buf;
> +	dma_addr_t tx_dma;
> +	u8 *rx_buf;
> +	dma_addr_t rx_dma;
> +	struct spi_message *msg;
> +	u32 n_bytes;
> +	u32 len;
> +	u32 sent;
> +	u32 rcvd;
> +	u32 err;
> +	u32 type;
> +	u32  tx_threshold;
> +	u32  rx_threshold;
> +	u32 irq_status;
> +	struct completion complete;
>  };
>  
>  struct dw_spi {
>  	struct spi_master	*master;
> -	struct spi_device	*cur_dev;
>  	struct device		*parent_dev;
>  	enum dw_ssi_type	type;
>  
>  	void __iomem		*regs;
>  	unsigned long		paddr;
>  	u32			iolen;
> -	int			irq;
> +	u32			irq;
>  	u32			fifo_len;	/* depth of the FIFO buffer */
>  	u32			max_freq;	/* max bus freq supported */
>  
>  	u16			bus_num;
>  	u16			num_cs;		/* supported slave numbers */
> -
> + 
>  	/* Driver message queue */
>  	struct workqueue_struct	*workqueue;
>  	struct work_struct	pump_messages;
>  	spinlock_t		lock;
>  	struct list_head	queue;
> -	int			busy;
> -	int			run;
> -
> -	/* Message Transfer pump */
> -	struct tasklet_struct	pump_transfers;
> +	u32			run;
>  
>  	/* Current message transfer state info */
> -	struct spi_message	*cur_msg;
> -	struct spi_transfer	*cur_transfer;
> -	struct chip_data	*cur_chip;
> -	struct chip_data	*prev_chip;
> -	size_t			len;
> -	void			*tx;
> -	void			*tx_end;
> -	void			*rx;
> -	void			*rx_end;
> -	int			dma_mapped;
> -	dma_addr_t		rx_dma;
> -	dma_addr_t		tx_dma;
> -	size_t			rx_map_len;
> -	size_t			tx_map_len;
> -	u8			n_bytes;	/* current is a 1/2 bytes op */
> -	u8			max_bits_per_word;	/* maxim is 16b */
> -	u32			dma_width;
> -	int			cs_change;
> -	irqreturn_t		(*transfer_handler)(struct dw_spi *dws);
> -	void			(*cs_control)(u32 command);
> +	struct xfer_state       xfer;
>  
>  	/* Dma info */
> -	int			dma_inited;
> +//	int 			dma_width;
> +	u32			dma_inited;
>  	struct dma_chan		*txchan;
>  	struct scatterlist	tx_sgl;
>  	struct dma_chan		*rxchan;
>  	struct scatterlist	rx_sgl;
> -	int			dma_chan_done;
> +	u32			dma_chan_done;
>  	struct device		*dma_dev;
>  	dma_addr_t		dma_addr; /* phy address of the Data register */
>  	struct dw_spi_dma_ops	*dma_ops;
> @@ -163,30 +168,36 @@ struct dw_spi {
>  	__raw_readw(&(((struct dw_spi_reg *)dw->regs)->name))
>  #define dw_writew(dw, name, val) \
>  	__raw_writew((val), &(((struct dw_spi_reg *)dw->regs)->name))
> +#define dw_readb(dw, name) \
> +	__raw_readb(&(((struct dw_spi_reg *)dw->regs)->name))
> +#define dw_writeb(dw, name, val) \
> +	__raw_writeb((val), &(((struct dw_spi_reg *)dw->regs)->name))
>  
> -static inline void spi_enable_chip(struct dw_spi *dws, int enable)
> +static inline void dw_spi_disable(struct dw_spi *dws)
>  {
> -	dw_writel(dws, ssienr, (enable ? 1 : 0));
> +	dw_writel(dws, ssienr, 0);
>  }
>  
> -static inline void spi_set_clk(struct dw_spi *dws, u16 div)
> +static inline void dw_spi_enable(struct dw_spi *dws)
> +{
> +	dw_writel(dws, ssienr, 1);
> +}
> +
> +static inline void dw_spi_set_clk(struct dw_spi *dws, u16 div)
>  {
>  	dw_writel(dws, baudr, div);
>  }
>  
> -static inline void spi_chip_sel(struct dw_spi *dws, u16 cs)
> +static inline void dw_spi_chip_sel(struct dw_spi *dws, u16 cs)
>  {
>  	if (cs > dws->num_cs)
>  		return;
>  
> -	if (dws->cs_control)
> -		dws->cs_control(1);
> -
>  	dw_writel(dws, ser, 1 << cs);
>  }
>  
>  /* Disable IRQ bits */
> -static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
> +static inline void dw_spi_mask_intr(struct dw_spi *dws, u32 mask)
>  {
>  	u32 new_mask;
>  
> @@ -195,7 +206,7 @@ static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
>  }
>  
>  /* Enable IRQ bits */
> -static inline void spi_umask_intr(struct dw_spi *dws, u32 mask)
> +static inline void dw_spi_umask_intr(struct dw_spi *dws, u32 mask)
>  {
>  	u32 new_mask;
>  
> @@ -208,6 +219,7 @@ extern void dw_spi_remove_host(struct dw_spi *dws);
>  extern int dw_spi_suspend_host(struct dw_spi *dws);
>  extern int dw_spi_resume_host(struct dw_spi *dws);
>  extern void dw_spi_xfer_done(struct dw_spi *dws);
> +extern int dw_spi_stop_queue(struct dw_spi *dws);
>  
>  /* platform related setup */
>  extern int dw_spi_mid_init(struct dw_spi *dws); /* Intel MID platforms */
> -- 
> 1.7.3.4
> 
> 
> ------------------------------------------------------------------------------
> EditLive Enterprise is the world's most technically advanced content
> authoring tool. Experience the power of Track Changes, Inline Image
> Editing and ensure content is compliant with Accessibility Checking.
> http://p.sf.net/sfu/ephox-dev2dev
> _______________________________________________
> spi-devel-general mailing list
> spi-devel-general@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/spi-devel-general

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
Feng Tang June 16, 2011, 2 p.m. UTC | #2
Hi Dirk,

On Thu, 16 Jun 2011 01:23:06 +0800
"dirk.brandewie@gmail.com" <dirk.brandewie@gmail.com> wrote:

> From: Dirk Brandewie <dirk.brandewie@gmail.com>
> 
> NOTE: patch created git format-patch --break-rewrites=/50%
> 
> This patch reworks the message pump worker thread function to run
> until all messages queued to the driver have been handled. The
> function to handle individual spi_transfers is now a synchronus
> function the tasklet to handle spi_transfers has been removed. Work
> for the worker thread is only queued in host controller transfer
> function.
> 
> Psuedo code for new thread function:
>   message = get_message()
>   while (message){
>     for_each_transfer_in_msg(message){
>       transfer_setup(transfer)
>       do_transfer()
>     }
>     complete_message()
>     message = get_message()
>   }
> 
> Changes that fell out of the message thread changes:
> Non-DMA transfers that are larger than the size of the controller FIFO
> are handled as interrupt driven transfers.
> 
> Common FIFO handling functions shared PIO and interrupt transfers.
> 
> Simplified queue stop/start funcitons.
> 
> Cleanup fixes:
> Changed exported all exported function names to have dw_spi_ prefix
> 
> Removed support for registering chip select control function. Setting
> the slave chip select is handled by setting the SER (Slave enable
> register)
> 
> Removed code that looked at the cs_change hint in the
> spi_transfer. Software has no contorl over whether the slave chip
> select is de-asserted at the end of the transfer.  Once the TX FIFO
> goes empty the slave chip select is dropped.
> 
> Added dw_spi_{en,dis}able inline functions to replace
> spi_enable_chip() Added dw_spi_{mask,umask}_intr inline functions


IMHO, the patch is too big, it contains too many changes to the original
drivers, and we can't see clearly what you've changed to each logical
code part or section, If possible, could you separate this patch to
several small ones in a logical way.

First, I have some questions, what devices have you tested with this patch?
high speed, low speed? Do you have any performance data to show the benefit
of this change? Current dw_spi driver has been tested with many devices, so
to not break them or cause obvious regression, we have to be cautious.

Here are some general comments according to the commit logs:
1. I think the threaded irq handling is a good idea. And let driver chose to
   use poll or interrupt is good, some other spi controller driver has used
   that way for a long time
2. Why you remove the cs_change code, in some case, the controller is only
   be used by one device, we don't need do the config for every single
   spi_transfer
3. Why do you remove the chip select control code, dw_spi controller hw has
   some problem in chip select controller by SER, and thus many devices has
   to use external GPIO has their chip select, this is real world usages!
4. I saw you enable both TX/RX interrupt when doing interrupt transfer, spi
   devices' TX/RX are born to be simutaneous, when one word is sent over
   TX line, a RX word will be received from RX line, so both the orignal
   interrupt transfer handling written by me and the later optimization
   from Alek Du only enable TX interrupt, which will only generate half of
   IRQs comparing to enble both TX/RX, this is huge when the data rate is
   several Mb per second
5. Why do you change the logic of filling TX FIFO, the logic comes from use
   case that the dw_spi driver is dealing with several high speed devices.

Thanks,
Feng




> 
> Signed-off-by: Dirk Brandewie <dirk.brandewie@gmail.com>
> ---
>  drivers/spi/spi-dw-mid.c |   43 +-
>  drivers/spi/spi-dw.c     | 1670
> ++++++++++++++++++++--------------------------
> drivers/spi/spi-dw.h     |   92 ++-- 3 files changed, 809
> insertions(+), 996 deletions(-) rewrite drivers/spi/spi-dw.c (55%)
> 
> diff --git a/drivers/spi/spi-dw-mid.c b/drivers/spi/spi-dw-mid.c
> index 130e555..e44e37f 100644
> --- a/drivers/spi/spi-dw-mid.c
> +++ b/drivers/spi/spi-dw-mid.c
> @@ -38,7 +38,10 @@ static bool mid_spi_dma_chan_filter(struct
> dma_chan *chan, void *param) {
>         struct dw_spi *dws = param;
> 
> -       return dws->dmac && (&dws->dmac->dev == chan->device->dev);
> +       if (dws->dmac && &dws->dmac->dev == chan->device->dev)
> +               return true;
> +       else
> +               return false;
>  }
> 
>  static int mid_spi_dma_init(struct dw_spi *dws)
> @@ -103,10 +106,10 @@ static void dw_spi_dma_done(void *arg)
> 
>         if (++dws->dma_chan_done != 2)
>                 return;
> -       dw_spi_xfer_done(dws);
> +       complete(&dws->xfer.complete);
>  }
> 
> -static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
> +static int mid_spi_dma_transfer(struct dw_spi *dws)
>  {
>         struct dma_async_tx_descriptor *txdesc = NULL, *rxdesc = NULL;
>         struct dma_chan *txchan, *rxchan;
> @@ -114,17 +117,17 @@ static int mid_spi_dma_transfer(struct dw_spi
> *dws, int cs_change) u16 dma_ctrl = 0;
> 
>         /* 1. setup DMA related registers */
> -       if (cs_change) {
> -               spi_enable_chip(dws, 0);
> -               dw_writew(dws, dmardlr, 0xf);
> -               dw_writew(dws, dmatdlr, 0x10);
> -               if (dws->tx_dma)
> -                       dma_ctrl |= 0x2;
> -               if (dws->rx_dma)
> -                       dma_ctrl |= 0x1;
> -               dw_writew(dws, dmacr, dma_ctrl);
> -               spi_enable_chip(dws, 1);
> -       }
> +
> +       dw_spi_disable(dws);
> +       dw_writew(dws, dmardlr, 0xf);
> +       dw_writew(dws, dmatdlr, 0x10);
> +       if (dws->xfer.tx_dma)
> +               dma_ctrl |= 0x2;
> +       if (dws->xfer.rx_dma)
> +               dma_ctrl |= 0x1;
> +       dw_writew(dws, dmacr, dma_ctrl);
> +       dw_spi_enable(dws);
> +
> 
>         dws->dma_chan_done = 0;
>         txchan = dws->txchan;
> @@ -141,8 +144,8 @@ static int mid_spi_dma_transfer(struct dw_spi
> *dws, int cs_change) (unsigned long) &txconf);
> 
>         memset(&dws->tx_sgl, 0, sizeof(dws->tx_sgl));
> -       dws->tx_sgl.dma_address = dws->tx_dma;
> -       dws->tx_sgl.length = dws->len;
> +       dws->tx_sgl.dma_address = dws->xfer.tx_dma;
> +       dws->tx_sgl.length = dws->xfer.len;
> 
>         txdesc = txchan->device->device_prep_slave_sg(txchan,
>                                 &dws->tx_sgl,
> @@ -163,8 +166,8 @@ static int mid_spi_dma_transfer(struct dw_spi
> *dws, int cs_change) (unsigned long) &rxconf);
> 
>         memset(&dws->rx_sgl, 0, sizeof(dws->rx_sgl));
> -       dws->rx_sgl.dma_address = dws->rx_dma;
> -       dws->rx_sgl.length = dws->len;
> +       dws->rx_sgl.dma_address = dws->xfer.rx_dma;
> +       dws->rx_sgl.length = dws->xfer.len;
> 
>         rxdesc = rxchan->device->device_prep_slave_sg(rxchan,
>                                 &dws->rx_sgl,
> @@ -188,7 +191,6 @@ static struct dw_spi_dma_ops mid_dma_ops = {
>  #endif
> 
>  /* Some specific info for SPI0 controller on Moorestown */
> -
>  /* HW info for MRST CLk Control Unit, one 32b reg */
>  #define MRST_SPI_CLK_BASE      100000000       /* 100m */
>  #define MRST_CLK_SPI0_REG      0xff11d86c
> @@ -202,12 +204,13 @@ int dw_spi_mid_init(struct dw_spi *dws)
>  {
>         u32 *clk_reg, clk_cdiv;
> 
> +
>         clk_reg = ioremap_nocache(MRST_CLK_SPI0_REG, 16);
>         if (!clk_reg)
>                 return -ENOMEM;
> 
>         /* get SPI controller operating freq info */
> -       clk_cdiv  = (readl(clk_reg) & CLK_SPI_CDIV_MASK) >>
> CLK_SPI_CDIV_OFFSET;
> +       clk_cdiv  = ((*clk_reg) & CLK_SPI_CDIV_MASK) >>
> CLK_SPI_CDIV_OFFSET; dws->max_freq = MRST_SPI_CLK_BASE / (clk_cdiv +
> 1); iounmap(clk_reg);
> 
> diff --git a/drivers/spi/spi-dw.c b/drivers/spi/spi-dw.c
> dissimilarity index 55%
> index ece5f69..2dacb8f 100644
> --- a/drivers/spi/spi-dw.c
> +++ b/drivers/spi/spi-dw.c
> @@ -1,936 +1,734 @@
> -/*
> - * Designware SPI core controller driver (refer pxa2xx_spi.c)
> - *
> - * Copyright (c) 2009, Intel Corporation.
> - *
> - * 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, write to the Free Software Foundation, Inc.,
> - * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
> - */
> -
> -#include <linux/dma-mapping.h>
> -#include <linux/interrupt.h>
> -#include <linux/highmem.h>
> -#include <linux/delay.h>
> -#include <linux/slab.h>
> -#include <linux/spi/spi.h>
> -
> -#include "spi-dw.h"
> -
> -#ifdef CONFIG_DEBUG_FS
> -#include <linux/debugfs.h>
> -#endif
> -
> -#define START_STATE    ((void *)0)
> -#define RUNNING_STATE  ((void *)1)
> -#define DONE_STATE     ((void *)2)
> -#define ERROR_STATE    ((void *)-1)
> -
> -#define QUEUE_RUNNING  0
> -#define QUEUE_STOPPED  1
> -
> -#define MRST_SPI_DEASSERT      0
> -#define MRST_SPI_ASSERT                1
> -
> -/* Slave spi_dev related */
> -struct chip_data {
> -       u16 cr0;
> -       u8 cs;                  /* chip select pin */
> -       u8 n_bytes;             /* current is a 1/2/4 byte op */
> -       u8 tmode;               /* TR/TO/RO/EEPROM */
> -       u8 type;                /* SPI/SSP/MicroWire */
> -
> -       u8 poll_mode;           /* 1 means use poll mode */
> -
> -       u32 dma_width;
> -       u32 rx_threshold;
> -       u32 tx_threshold;
> -       u8 enable_dma;
> -       u8 bits_per_word;
> -       u16 clk_div;            /* baud rate divider */
> -       u32 speed_hz;           /* baud rate */
> -       void (*cs_control)(u32 command);
> -};
> -
> -#ifdef CONFIG_DEBUG_FS
> -static int spi_show_regs_open(struct inode *inode, struct file *file)
> -{
> -       file->private_data = inode->i_private;
> -       return 0;
> -}
> -
> -#define SPI_REGS_BUFSIZE       1024
> -static ssize_t  spi_show_regs(struct file *file, char __user
> *user_buf,
> -                               size_t count, loff_t *ppos)
> -{
> -       struct dw_spi *dws;
> -       char *buf;
> -       u32 len = 0;
> -       ssize_t ret;
> -
> -       dws = file->private_data;
> -
> -       buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
> -       if (!buf)
> -               return 0;
> -
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "MRST SPI0 registers:\n");
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "=================================\n");
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "SER: \t\t0x%08x\n", dw_readl(dws, ser));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "SR: \t\t0x%08x\n", dw_readl(dws, sr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "IMR: \t\t0x%08x\n", dw_readl(dws, imr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "ISR: \t\t0x%08x\n", dw_readl(dws, isr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "DMATDLR: \t0x%08x\n", dw_readl(dws,
> dmatdlr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "DMARDLR: \t0x%08x\n", dw_readl(dws,
> dmardlr));
> -       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> -                       "=================================\n");
> -
> -       ret =  simple_read_from_buffer(user_buf, count, ppos, buf,
> len);
> -       kfree(buf);
> -       return ret;
> -}
> -
> -static const struct file_operations mrst_spi_regs_ops = {
> -       .owner          = THIS_MODULE,
> -       .open           = spi_show_regs_open,
> -       .read           = spi_show_regs,
> -       .llseek         = default_llseek,
> -};
> -
> -static int mrst_spi_debugfs_init(struct dw_spi *dws)
> -{
> -       dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
> -       if (!dws->debugfs)
> -               return -ENOMEM;
> -
> -       debugfs_create_file("registers", S_IFREG | S_IRUGO,
> -               dws->debugfs, (void *)dws, &mrst_spi_regs_ops);
> -       return 0;
> -}
> -
> -static void mrst_spi_debugfs_remove(struct dw_spi *dws)
> -{
> -       if (dws->debugfs)
> -               debugfs_remove_recursive(dws->debugfs);
> -}
> -
> -#else
> -static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
> -{
> -       return 0;
> -}
> -
> -static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
> -{
> -}
> -#endif /* CONFIG_DEBUG_FS */
> -
> -/* Return the max entries we can fill into tx fifo */
> -static inline u32 tx_max(struct dw_spi *dws)
> -{
> -       u32 tx_left, tx_room, rxtx_gap;
> -
> -       tx_left = (dws->tx_end - dws->tx) / dws->n_bytes;
> -       tx_room = dws->fifo_len - dw_readw(dws, txflr);
> -
> -       /*
> -        * Another concern is about the tx/rx mismatch, we
> -        * though to use (dws->fifo_len - rxflr - txflr) as
> -        * one maximum value for tx, but it doesn't cover the
> -        * data which is out of tx/rx fifo and inside the
> -        * shift registers. So a control from sw point of
> -        * view is taken.
> -        */
> -       rxtx_gap =  ((dws->rx_end - dws->rx) - (dws->tx_end -
> dws->tx))
> -                       / dws->n_bytes;
> -
> -       return min3(tx_left, tx_room, (u32) (dws->fifo_len -
> rxtx_gap)); -}
> -
> -/* Return the max entries we should read out of rx fifo */
> -static inline u32 rx_max(struct dw_spi *dws)
> -{
> -       u32 rx_left = (dws->rx_end - dws->rx) / dws->n_bytes;
> -
> -       return min(rx_left, (u32)dw_readw(dws, rxflr));
> -}
> -
> -static void dw_writer(struct dw_spi *dws)
> -{
> -       u32 max = tx_max(dws);
> -       u16 txw = 0;
> -
> -       while (max--) {
> -               /* Set the tx word if the transfer's original "tx" is
> not null */
> -               if (dws->tx_end - dws->len) {
> -                       if (dws->n_bytes == 1)
> -                               txw = *(u8 *)(dws->tx);
> -                       else
> -                               txw = *(u16 *)(dws->tx);
> -               }
> -               dw_writew(dws, dr, txw);
> -               dws->tx += dws->n_bytes;
> -       }
> -}
> -
> -static void dw_reader(struct dw_spi *dws)
> -{
> -       u32 max = rx_max(dws);
> -       u16 rxw;
> -
> -       while (max--) {
> -               rxw = dw_readw(dws, dr);
> -               /* Care rx only if the transfer's original "rx" is
> not null */
> -               if (dws->rx_end - dws->len) {
> -                       if (dws->n_bytes == 1)
> -                               *(u8 *)(dws->rx) = rxw;
> -                       else
> -                               *(u16 *)(dws->rx) = rxw;
> -               }
> -               dws->rx += dws->n_bytes;
> -       }
> -}
> -
> -static void *next_transfer(struct dw_spi *dws)
> -{
> -       struct spi_message *msg = dws->cur_msg;
> -       struct spi_transfer *trans = dws->cur_transfer;
> -
> -       /* Move to next transfer */
> -       if (trans->transfer_list.next != &msg->transfers) {
> -               dws->cur_transfer =
> -                       list_entry(trans->transfer_list.next,
> -                                       struct spi_transfer,
> -                                       transfer_list);
> -               return RUNNING_STATE;
> -       } else
> -               return DONE_STATE;
> -}
> -
> -/*
> - * Note: first step is the protocol driver prepares
> - * a dma-capable memory, and this func just need translate
> - * the virt addr to physical
> - */
> -static int map_dma_buffers(struct dw_spi *dws)
> -{
> -       if (!dws->cur_msg->is_dma_mapped
> -               || !dws->dma_inited
> -               || !dws->cur_chip->enable_dma
> -               || !dws->dma_ops)
> -               return 0;
> -
> -       if (dws->cur_transfer->tx_dma)
> -               dws->tx_dma = dws->cur_transfer->tx_dma;
> -
> -       if (dws->cur_transfer->rx_dma)
> -               dws->rx_dma = dws->cur_transfer->rx_dma;
> -
> -       return 1;
> -}
> -
> -/* Caller already set message->status; dma and pio irqs are blocked
> */ -static void giveback(struct dw_spi *dws)
> -{
> -       struct spi_transfer *last_transfer;
> -       unsigned long flags;
> -       struct spi_message *msg;
> -
> -       spin_lock_irqsave(&dws->lock, flags);
> -       msg = dws->cur_msg;
> -       dws->cur_msg = NULL;
> -       dws->cur_transfer = NULL;
> -       dws->prev_chip = dws->cur_chip;
> -       dws->cur_chip = NULL;
> -       dws->dma_mapped = 0;
> -       queue_work(dws->workqueue, &dws->pump_messages);
> -       spin_unlock_irqrestore(&dws->lock, flags);
> -
> -       last_transfer = list_entry(msg->transfers.prev,
> -                                       struct spi_transfer,
> -                                       transfer_list);
> -
> -       if (!last_transfer->cs_change && dws->cs_control)
> -               dws->cs_control(MRST_SPI_DEASSERT);
> -
> -       msg->state = NULL;
> -       if (msg->complete)
> -               msg->complete(msg->context);
> -}
> -
> -static void int_error_stop(struct dw_spi *dws, const char *msg)
> -{
> -       /* Stop the hw */
> -       spi_enable_chip(dws, 0);
> -
> -       dev_err(&dws->master->dev, "%s\n", msg);
> -       dws->cur_msg->state = ERROR_STATE;
> -       tasklet_schedule(&dws->pump_transfers);
> -}
> -
> -void dw_spi_xfer_done(struct dw_spi *dws)
> -{
> -       /* Update total byte transferred return count actual bytes
> read */
> -       dws->cur_msg->actual_length += dws->len;
> -
> -       /* Move to next transfer */
> -       dws->cur_msg->state = next_transfer(dws);
> -
> -       /* Handle end of message */
> -       if (dws->cur_msg->state == DONE_STATE) {
> -               dws->cur_msg->status = 0;
> -               giveback(dws);
> -       } else
> -               tasklet_schedule(&dws->pump_transfers);
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_xfer_done);
> -
> -static irqreturn_t interrupt_transfer(struct dw_spi *dws)
> -{
> -       u16 irq_status = dw_readw(dws, isr);
> -
> -       /* Error handling */
> -       if (irq_status & (SPI_INT_TXOI | SPI_INT_RXOI |
> SPI_INT_RXUI)) {
> -               dw_readw(dws, txoicr);
> -               dw_readw(dws, rxoicr);
> -               dw_readw(dws, rxuicr);
> -               int_error_stop(dws, "interrupt_transfer: fifo
> overrun/underrun");
> -               return IRQ_HANDLED;
> -       }
> -
> -       dw_reader(dws);
> -       if (dws->rx_end == dws->rx) {
> -               spi_mask_intr(dws, SPI_INT_TXEI);
> -               dw_spi_xfer_done(dws);
> -               return IRQ_HANDLED;
> -       }
> -       if (irq_status & SPI_INT_TXEI) {
> -               spi_mask_intr(dws, SPI_INT_TXEI);
> -               dw_writer(dws);
> -               /* Enable TX irq always, it will be disabled when RX
> finished */
> -               spi_umask_intr(dws, SPI_INT_TXEI);
> -       }
> -
> -       return IRQ_HANDLED;
> -}
> -
> -static irqreturn_t dw_spi_irq(int irq, void *dev_id)
> -{
> -       struct dw_spi *dws = dev_id;
> -       u16 irq_status = dw_readw(dws, isr) & 0x3f;
> -
> -       if (!irq_status)
> -               return IRQ_NONE;
> -
> -       if (!dws->cur_msg) {
> -               spi_mask_intr(dws, SPI_INT_TXEI);
> -               return IRQ_HANDLED;
> -       }
> -
> -       return dws->transfer_handler(dws);
> -}
> -
> -/* Must be called inside pump_transfers() */
> -static void poll_transfer(struct dw_spi *dws)
> -{
> -       do {
> -               dw_writer(dws);
> -               dw_reader(dws);
> -               cpu_relax();
> -       } while (dws->rx_end > dws->rx);
> -
> -       dw_spi_xfer_done(dws);
> -}
> -
> -static void pump_transfers(unsigned long data)
> -{
> -       struct dw_spi *dws = (struct dw_spi *)data;
> -       struct spi_message *message = NULL;
> -       struct spi_transfer *transfer = NULL;
> -       struct spi_transfer *previous = NULL;
> -       struct spi_device *spi = NULL;
> -       struct chip_data *chip = NULL;
> -       u8 bits = 0;
> -       u8 imask = 0;
> -       u8 cs_change = 0;
> -       u16 txint_level = 0;
> -       u16 clk_div = 0;
> -       u32 speed = 0;
> -       u32 cr0 = 0;
> -
> -       /* Get current state information */
> -       message = dws->cur_msg;
> -       transfer = dws->cur_transfer;
> -       chip = dws->cur_chip;
> -       spi = message->spi;
> -
> -       if (unlikely(!chip->clk_div))
> -               chip->clk_div = dws->max_freq / chip->speed_hz;
> -
> -       if (message->state == ERROR_STATE) {
> -               message->status = -EIO;
> -               goto early_exit;
> -       }
> -
> -       /* Handle end of message */
> -       if (message->state == DONE_STATE) {
> -               message->status = 0;
> -               goto early_exit;
> -       }
> -
> -       /* Delay if requested at end of transfer*/
> -       if (message->state == RUNNING_STATE) {
> -               previous = list_entry(transfer->transfer_list.prev,
> -                                       struct spi_transfer,
> -                                       transfer_list);
> -               if (previous->delay_usecs)
> -                       udelay(previous->delay_usecs);
> -       }
> -
> -       dws->n_bytes = chip->n_bytes;
> -       dws->dma_width = chip->dma_width;
> -       dws->cs_control = chip->cs_control;
> -
> -       dws->rx_dma = transfer->rx_dma;
> -       dws->tx_dma = transfer->tx_dma;
> -       dws->tx = (void *)transfer->tx_buf;
> -       dws->tx_end = dws->tx + transfer->len;
> -       dws->rx = transfer->rx_buf;
> -       dws->rx_end = dws->rx + transfer->len;
> -       dws->cs_change = transfer->cs_change;
> -       dws->len = dws->cur_transfer->len;
> -       if (chip != dws->prev_chip)
> -               cs_change = 1;
> -
> -       cr0 = chip->cr0;
> -
> -       /* Handle per transfer options for bpw and speed */
> -       if (transfer->speed_hz) {
> -               speed = chip->speed_hz;
> -
> -               if (transfer->speed_hz != speed) {
> -                       speed = transfer->speed_hz;
> -                       if (speed > dws->max_freq) {
> -                               printk(KERN_ERR "MRST SPI0:
> unsupported"
> -                                       "freq: %dHz\n", speed);
> -                               message->status = -EIO;
> -                               goto early_exit;
> -                       }
> -
> -                       /* clk_div doesn't support odd number */
> -                       clk_div = dws->max_freq / speed;
> -                       clk_div = (clk_div + 1) & 0xfffe;
> -
> -                       chip->speed_hz = speed;
> -                       chip->clk_div = clk_div;
> -               }
> -       }
> -       if (transfer->bits_per_word) {
> -               bits = transfer->bits_per_word;
> -
> -               switch (bits) {
> -               case 8:
> -               case 16:
> -                       dws->n_bytes = dws->dma_width = bits >> 3;
> -                       break;
> -               default:
> -                       printk(KERN_ERR "MRST SPI0: unsupported bits:"
> -                               "%db\n", bits);
> -                       message->status = -EIO;
> -                       goto early_exit;
> -               }
> -
> -               cr0 = (bits - 1)
> -                       | (chip->type << SPI_FRF_OFFSET)
> -                       | (spi->mode << SPI_MODE_OFFSET)
> -                       | (chip->tmode << SPI_TMOD_OFFSET);
> -       }
> -       message->state = RUNNING_STATE;
> -
> -       /*
> -        * Adjust transfer mode if necessary. Requires platform
> dependent
> -        * chipselect mechanism.
> -        */
> -       if (dws->cs_control) {
> -               if (dws->rx && dws->tx)
> -                       chip->tmode = SPI_TMOD_TR;
> -               else if (dws->rx)
> -                       chip->tmode = SPI_TMOD_RO;
> -               else
> -                       chip->tmode = SPI_TMOD_TO;
> -
> -               cr0 &= ~SPI_TMOD_MASK;
> -               cr0 |= (chip->tmode << SPI_TMOD_OFFSET);
> -       }
> -
> -       /* Check if current transfer is a DMA transaction */
> -       dws->dma_mapped = map_dma_buffers(dws);
> -
> -       /*
> -        * Interrupt mode
> -        * we only need set the TXEI IRQ, as TX/RX always happen
> syncronizely
> -        */
> -       if (!dws->dma_mapped && !chip->poll_mode) {
> -               int templen = dws->len / dws->n_bytes;
> -               txint_level = dws->fifo_len / 2;
> -               txint_level = (templen > txint_level) ? txint_level :
> templen; -
> -               imask |= SPI_INT_TXEI | SPI_INT_TXOI | SPI_INT_RXUI |
> SPI_INT_RXOI;
> -               dws->transfer_handler = interrupt_transfer;
> -       }
> -
> -       /*
> -        * Reprogram registers only if
> -        *      1. chip select changes
> -        *      2. clk_div is changed
> -        *      3. control value changes
> -        */
> -       if (dw_readw(dws, ctrl0) != cr0 || cs_change || clk_div ||
> imask) {
> -               spi_enable_chip(dws, 0);
> -
> -               if (dw_readw(dws, ctrl0) != cr0)
> -                       dw_writew(dws, ctrl0, cr0);
> -
> -               spi_set_clk(dws, clk_div ? clk_div : chip->clk_div);
> -               spi_chip_sel(dws, spi->chip_select);
> -
> -               /* Set the interrupt mask, for poll mode just disable
> all int */
> -               spi_mask_intr(dws, 0xff);
> -               if (imask)
> -                       spi_umask_intr(dws, imask);
> -               if (txint_level)
> -                       dw_writew(dws, txfltr, txint_level);
> -
> -               spi_enable_chip(dws, 1);
> -               if (cs_change)
> -                       dws->prev_chip = chip;
> -       }
> -
> -       if (dws->dma_mapped)
> -               dws->dma_ops->dma_transfer(dws, cs_change);
> -
> -       if (chip->poll_mode)
> -               poll_transfer(dws);
> -
> -       return;
> -
> -early_exit:
> -       giveback(dws);
> -       return;
> -}
> -
> -static void pump_messages(struct work_struct *work)
> -{
> -       struct dw_spi *dws =
> -               container_of(work, struct dw_spi, pump_messages);
> -       unsigned long flags;
> -
> -       /* Lock queue and check for queue work */
> -       spin_lock_irqsave(&dws->lock, flags);
> -       if (list_empty(&dws->queue) || dws->run == QUEUE_STOPPED) {
> -               dws->busy = 0;
> -               spin_unlock_irqrestore(&dws->lock, flags);
> -               return;
> -       }
> -
> -       /* Make sure we are not already running a message */
> -       if (dws->cur_msg) {
> -               spin_unlock_irqrestore(&dws->lock, flags);
> -               return;
> -       }
> -
> -       /* Extract head of queue */
> -       dws->cur_msg = list_entry(dws->queue.next, struct
> spi_message, queue);
> -       list_del_init(&dws->cur_msg->queue);
> -
> -       /* Initial message state*/
> -       dws->cur_msg->state = START_STATE;
> -       dws->cur_transfer = list_entry(dws->cur_msg->transfers.next,
> -                                               struct spi_transfer,
> -                                               transfer_list);
> -       dws->cur_chip = spi_get_ctldata(dws->cur_msg->spi);
> -
> -       /* Mark as busy and launch transfers */
> -       tasklet_schedule(&dws->pump_transfers);
> -
> -       dws->busy = 1;
> -       spin_unlock_irqrestore(&dws->lock, flags);
> -}
> -
> -/* spi_device use this to queue in their spi_msg */
> -static int dw_spi_transfer(struct spi_device *spi, struct
> spi_message *msg) -{
> -       struct dw_spi *dws = spi_master_get_devdata(spi->master);
> -       unsigned long flags;
> -
> -       spin_lock_irqsave(&dws->lock, flags);
> -
> -       if (dws->run == QUEUE_STOPPED) {
> -               spin_unlock_irqrestore(&dws->lock, flags);
> -               return -ESHUTDOWN;
> -       }
> -
> -       msg->actual_length = 0;
> -       msg->status = -EINPROGRESS;
> -       msg->state = START_STATE;
> -
> -       list_add_tail(&msg->queue, &dws->queue);
> -
> -       if (dws->run == QUEUE_RUNNING && !dws->busy) {
> -
> -               if (dws->cur_transfer || dws->cur_msg)
> -                       queue_work(dws->workqueue,
> -                                       &dws->pump_messages);
> -               else {
> -                       /* If no other data transaction in air, just
> go */
> -                       spin_unlock_irqrestore(&dws->lock, flags);
> -                       pump_messages(&dws->pump_messages);
> -                       return 0;
> -               }
> -       }
> -
> -       spin_unlock_irqrestore(&dws->lock, flags);
> -       return 0;
> -}
> -
> -/* This may be called twice for each spi dev */
> -static int dw_spi_setup(struct spi_device *spi)
> -{
> -       struct dw_spi_chip *chip_info = NULL;
> -       struct chip_data *chip;
> -
> -       if (spi->bits_per_word != 8 && spi->bits_per_word != 16)
> -               return -EINVAL;
> -
> -       /* Only alloc on first setup */
> -       chip = spi_get_ctldata(spi);
> -       if (!chip) {
> -               chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
> -               if (!chip)
> -                       return -ENOMEM;
> -       }
> -
> -       /*
> -        * Protocol drivers may change the chip settings, so...
> -        * if chip_info exists, use it
> -        */
> -       chip_info = spi->controller_data;
> -
> -       /* chip_info doesn't always exist */
> -       if (chip_info) {
> -               if (chip_info->cs_control)
> -                       chip->cs_control = chip_info->cs_control;
> -
> -               chip->poll_mode = chip_info->poll_mode;
> -               chip->type = chip_info->type;
> -
> -               chip->rx_threshold = 0;
> -               chip->tx_threshold = 0;
> -
> -               chip->enable_dma = chip_info->enable_dma;
> -       }
> -
> -       if (spi->bits_per_word <= 8) {
> -               chip->n_bytes = 1;
> -               chip->dma_width = 1;
> -       } else if (spi->bits_per_word <= 16) {
> -               chip->n_bytes = 2;
> -               chip->dma_width = 2;
> -       } else {
> -               /* Never take >16b case for MRST SPIC */
> -               dev_err(&spi->dev, "invalid wordsize\n");
> -               return -EINVAL;
> -       }
> -       chip->bits_per_word = spi->bits_per_word;
> -
> -       if (!spi->max_speed_hz) {
> -               dev_err(&spi->dev, "No max speed HZ parameter\n");
> -               return -EINVAL;
> -       }
> -       chip->speed_hz = spi->max_speed_hz;
> -
> -       chip->tmode = 0; /* Tx & Rx */
> -       /* Default SPI mode is SCPOL = 0, SCPH = 0 */
> -       chip->cr0 = (chip->bits_per_word - 1)
> -                       | (chip->type << SPI_FRF_OFFSET)
> -                       | (spi->mode  << SPI_MODE_OFFSET)
> -                       | (chip->tmode << SPI_TMOD_OFFSET);
> -
> -       spi_set_ctldata(spi, chip);
> -       return 0;
> -}
> -
> -static void dw_spi_cleanup(struct spi_device *spi)
> -{
> -       struct chip_data *chip = spi_get_ctldata(spi);
> -       kfree(chip);
> -}
> -
> -static int __devinit init_queue(struct dw_spi *dws)
> -{
> -       INIT_LIST_HEAD(&dws->queue);
> -       spin_lock_init(&dws->lock);
> -
> -       dws->run = QUEUE_STOPPED;
> -       dws->busy = 0;
> -
> -       tasklet_init(&dws->pump_transfers,
> -                       pump_transfers, (unsigned long)dws);
> -
> -       INIT_WORK(&dws->pump_messages, pump_messages);
> -       dws->workqueue = create_singlethread_workqueue(
> -
> dev_name(dws->master->dev.parent));
> -       if (dws->workqueue == NULL)
> -               return -EBUSY;
> -
> -       return 0;
> -}
> -
> -static int start_queue(struct dw_spi *dws)
> -{
> -       unsigned long flags;
> -
> -       spin_lock_irqsave(&dws->lock, flags);
> -
> -       if (dws->run == QUEUE_RUNNING || dws->busy) {
> -               spin_unlock_irqrestore(&dws->lock, flags);
> -               return -EBUSY;
> -       }
> -
> -       dws->run = QUEUE_RUNNING;
> -       dws->cur_msg = NULL;
> -       dws->cur_transfer = NULL;
> -       dws->cur_chip = NULL;
> -       dws->prev_chip = NULL;
> -       spin_unlock_irqrestore(&dws->lock, flags);
> -
> -       queue_work(dws->workqueue, &dws->pump_messages);
> -
> -       return 0;
> -}
> -
> -static int stop_queue(struct dw_spi *dws)
> -{
> -       unsigned long flags;
> -       unsigned limit = 50;
> -       int status = 0;
> -
> -       spin_lock_irqsave(&dws->lock, flags);
> -       dws->run = QUEUE_STOPPED;
> -       while ((!list_empty(&dws->queue) || dws->busy) && limit--) {
> -               spin_unlock_irqrestore(&dws->lock, flags);
> -               msleep(10);
> -               spin_lock_irqsave(&dws->lock, flags);
> -       }
> -
> -       if (!list_empty(&dws->queue) || dws->busy)
> -               status = -EBUSY;
> -       spin_unlock_irqrestore(&dws->lock, flags);
> -
> -       return status;
> -}
> -
> -static int destroy_queue(struct dw_spi *dws)
> -{
> -       int status;
> -
> -       status = stop_queue(dws);
> -       if (status != 0)
> -               return status;
> -       destroy_workqueue(dws->workqueue);
> -       return 0;
> -}
> -
> -/* Restart the controller, disable all interrupts, clean rx fifo */
> -static void spi_hw_init(struct dw_spi *dws)
> -{
> -       spi_enable_chip(dws, 0);
> -       spi_mask_intr(dws, 0xff);
> -       spi_enable_chip(dws, 1);
> -
> -       /*
> -        * Try to detect the FIFO depth if not set by interface
> driver,
> -        * the depth could be from 2 to 256 from HW spec
> -        */
> -       if (!dws->fifo_len) {
> -               u32 fifo;
> -               for (fifo = 2; fifo <= 257; fifo++) {
> -                       dw_writew(dws, txfltr, fifo);
> -                       if (fifo != dw_readw(dws, txfltr))
> -                               break;
> -               }
> -
> -               dws->fifo_len = (fifo == 257) ? 0 : fifo;
> -               dw_writew(dws, txfltr, 0);
> -       }
> -}
> -
> -int __devinit dw_spi_add_host(struct dw_spi *dws)
> -{
> -       struct spi_master *master;
> -       int ret;
> -
> -       BUG_ON(dws == NULL);
> -
> -       master = spi_alloc_master(dws->parent_dev, 0);
> -       if (!master) {
> -               ret = -ENOMEM;
> -               goto exit;
> -       }
> -
> -       dws->master = master;
> -       dws->type = SSI_MOTO_SPI;
> -       dws->prev_chip = NULL;
> -       dws->dma_inited = 0;
> -       dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
> -
> -       ret = request_irq(dws->irq, dw_spi_irq, IRQF_SHARED,
> -                       "dw_spi", dws);
> -       if (ret < 0) {
> -               dev_err(&master->dev, "can not get IRQ\n");
> -               goto err_free_master;
> -       }
> -
> -       master->mode_bits = SPI_CPOL | SPI_CPHA;
> -       master->bus_num = dws->bus_num;
> -       master->num_chipselect = dws->num_cs;
> -       master->cleanup = dw_spi_cleanup;
> -       master->setup = dw_spi_setup;
> -       master->transfer = dw_spi_transfer;
> -
> -       /* Basic HW init */
> -       spi_hw_init(dws);
> -
> -       if (dws->dma_ops && dws->dma_ops->dma_init) {
> -               ret = dws->dma_ops->dma_init(dws);
> -               if (ret) {
> -                       dev_warn(&master->dev, "DMA init failed\n");
> -                       dws->dma_inited = 0;
> -               }
> -       }
> -
> -       /* Initial and start queue */
> -       ret = init_queue(dws);
> -       if (ret) {
> -               dev_err(&master->dev, "problem initializing queue\n");
> -               goto err_diable_hw;
> -       }
> -       ret = start_queue(dws);
> -       if (ret) {
> -               dev_err(&master->dev, "problem starting queue\n");
> -               goto err_diable_hw;
> -       }
> -
> -       spi_master_set_devdata(master, dws);
> -       ret = spi_register_master(master);
> -       if (ret) {
> -               dev_err(&master->dev, "problem registering spi
> master\n");
> -               goto err_queue_alloc;
> -       }
> -
> -       mrst_spi_debugfs_init(dws);
> -       return 0;
> -
> -err_queue_alloc:
> -       destroy_queue(dws);
> -       if (dws->dma_ops && dws->dma_ops->dma_exit)
> -               dws->dma_ops->dma_exit(dws);
> -err_diable_hw:
> -       spi_enable_chip(dws, 0);
> -       free_irq(dws->irq, dws);
> -err_free_master:
> -       spi_master_put(master);
> -exit:
> -       return ret;
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_add_host);
> -
> -void __devexit dw_spi_remove_host(struct dw_spi *dws)
> -{
> -       int status = 0;
> -
> -       if (!dws)
> -               return;
> -       mrst_spi_debugfs_remove(dws);
> -
> -       /* Remove the queue */
> -       status = destroy_queue(dws);
> -       if (status != 0)
> -               dev_err(&dws->master->dev, "dw_spi_remove: workqueue
> will not "
> -                       "complete, message memory not freed\n");
> -
> -       if (dws->dma_ops && dws->dma_ops->dma_exit)
> -               dws->dma_ops->dma_exit(dws);
> -       spi_enable_chip(dws, 0);
> -       /* Disable clk */
> -       spi_set_clk(dws, 0);
> -       free_irq(dws->irq, dws);
> -
> -       /* Disconnect from the SPI framework */
> -       spi_unregister_master(dws->master);
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_remove_host);
> -
> -int dw_spi_suspend_host(struct dw_spi *dws)
> -{
> -       int ret = 0;
> -
> -       ret = stop_queue(dws);
> -       if (ret)
> -               return ret;
> -       spi_enable_chip(dws, 0);
> -       spi_set_clk(dws, 0);
> -       return ret;
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
> -
> -int dw_spi_resume_host(struct dw_spi *dws)
> -{
> -       int ret;
> -
> -       spi_hw_init(dws);
> -       ret = start_queue(dws);
> -       if (ret)
> -               dev_err(&dws->master->dev, "fail to start queue
> (%d)\n", ret);
> -       return ret;
> -}
> -EXPORT_SYMBOL_GPL(dw_spi_resume_host);
> -
> -MODULE_AUTHOR("Feng Tang <feng.tang@intel.com>");
> -MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
> -MODULE_LICENSE("GPL v2");
> +/*
> + * dw_spi.c - Designware SPI core controller driver
> + *
> + * Copyright (c) 2009, Intel Corporation.
> + *
> + * 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, write to the Free Software Foundation, Inc.,
> + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
> + */
> +
> +#include <linux/dma-mapping.h>
> +#include <linux/interrupt.h>
> +#include <linux/highmem.h>
> +#include <linux/delay.h>
> +#include <linux/slab.h>
> +#include <linux/spi/spi.h>
> +
> +#include "spi-dw.h"
> +
> +#ifdef CONFIG_DEBUG_FS
> +#include <linux/debugfs.h>
> +#endif
> +
> +
> +#define QUEUE_RUNNING  0
> +#define QUEUE_STOPPED  1
> +
> +
> +/* Slave spi_dev related */
> +struct chip_data {
> +       struct spi_device *spi_dev;
> +       u32 cr0;
> +       u32 cs;                 /* chip select pin */
> +       u32 n_bytes;            /* current is a 1/2/4 byte op */
> +       u32 type;               /* SPI/SSP/MicroWire */
> +
> +       u32 dma_width;
> +       u32 enable_dma;
> +       u32 bits_per_word;
> +       u32 clk_div;            /* baud rate divider */
> +       u32 speed_hz;           /* baud rate */
> +};
> +
> +#ifdef CONFIG_DEBUG_FS
> +static int spi_show_regs_open(struct inode *inode, struct file *file)
> +{
> +       file->private_data = inode->i_private;
> +       return 0;
> +}
> +
> +#define SPI_REGS_BUFSIZE       1024
> +static ssize_t  spi_show_regs(struct file *file, char __user
> *user_buf,
> +                               size_t count, loff_t *ppos)
> +{
> +       struct dw_spi *dws;
> +       char *buf;
> +       u32 len = 0;
> +       ssize_t ret;
> +
> +       dws = file->private_data;
> +
> +       buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
> +       if (!buf)
> +               return 0;
> +
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "MRST SPI0 registers:\n");
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "=================================\n");
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "SER: \t\t0x%08x\n", dw_readl(dws, ser));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "SR: \t\t0x%08x\n", dw_readl(dws, sr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "IMR: \t\t0x%08x\n", dw_readl(dws, imr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "ISR: \t\t0x%08x\n", dw_readl(dws, isr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "DMATDLR: \t0x%08x\n", dw_readl(dws,
> dmatdlr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "DMARDLR: \t0x%08x\n", dw_readl(dws,
> dmardlr));
> +       len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
> +                       "=================================\n");
> +
> +       ret =  simple_read_from_buffer(user_buf, count, ppos, buf,
> len);
> +       kfree(buf);
> +       return ret;
> +}
> +
> +static const struct file_operations mrst_spi_regs_ops = {
> +       .owner          = THIS_MODULE,
> +       .open           = spi_show_regs_open,
> +       .read           = spi_show_regs,
> +       .llseek         = default_llseek,
> +};
> +
> +static int mrst_spi_debugfs_init(struct dw_spi *dws)
> +{
> +       dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
> +       if (!dws->debugfs)
> +               return -ENOMEM;
> +
> +       debugfs_create_file("registers", S_IFREG | S_IRUGO,
> +               dws->debugfs, (void *)dws, &mrst_spi_regs_ops);
> +       return 0;
> +}
> +
> +static void mrst_spi_debugfs_remove(struct dw_spi *dws)
> +{
> +       if (dws->debugfs)
> +               debugfs_remove_recursive(dws->debugfs);
> +}
> +
> +#else
> +static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
> +{
> +       return 0;
> +}
> +
> +static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
> +{
> +}
> +#endif /* CONFIG_DEBUG_FS */
> +
> +static irqreturn_t dw_spi_irq(int irq, void *dev_id)
> +{
> +       struct dw_spi *dws = dev_id;
> +       u16  irq_mask = 0x3f;
> +
> +       dws->xfer.irq_status = dw_readw(dws, isr) & irq_mask;
> +
> +       if (!dws->xfer.irq_status)
> +               return IRQ_NONE;
> +       if (dws->xfer.irq_status &
> +                       (SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI))
> {
> +               dw_readw(dws, txoicr);
> +               dw_readw(dws, rxoicr);
> +               dw_readw(dws, rxuicr);
> +               dws->xfer.err = -EIO;
> +               dw_spi_disable(dws);
> +               complete(&dws->xfer.complete);
> +               return IRQ_HANDLED;
> +       }
> +
> +       /* disable interrupts */
> +       dw_spi_mask_intr(dws, irq_mask);
> +       return IRQ_WAKE_THREAD;
> +}
> +struct spi_message *get_message(struct dw_spi *dws)
> +{
> +       struct spi_message *message = NULL;
> +       unsigned long flags;
> +
> +       spin_lock_irqsave(&dws->lock, flags);
> +       if (!list_empty(&dws->queue)) {
> +               message = list_entry(dws->queue.next,
> +                               struct spi_message, queue);
> +               list_del_init(&message->queue);
> +       }
> +       spin_unlock_irqrestore(&dws->lock, flags);
> +       return message;
> +}
> +static inline u32 tx_max(struct dw_spi *dws)
> +{
> +       u32 tx_left, tx_room;
> +
> +       tx_left = (dws->xfer.len - dws->xfer.sent) /
> dws->xfer.n_bytes;
> +       tx_room = (dws->fifo_len - dw_readw(dws, txflr));
> +
> +       return min(tx_left, tx_room);
> +}
> +
> +/* Return the max entries we should read out of rx fifo */
> +static inline u32 rx_max(struct dw_spi *dws)
> +{
> +       u32 rx_left = (dws->xfer.len - dws->xfer.rcvd) /
> dws->xfer.n_bytes;
> +       return min(rx_left, (u32)dw_readw(dws, rxflr));
> +}
> +
> +static int transfer_setup(struct dw_spi *dws, struct chip_data
> *controller,
> +                       struct spi_transfer *transfer, struct
> spi_message *msg) +{
> +       int err = 0;
> +       u32 cr0 = controller->cr0;
> +       u32 clk_div;
> +
> +       dws->xfer.tx_buf = transfer->tx_buf;
> +       dws->xfer.tx_dma = transfer->tx_dma;
> +       dws->xfer.rx_buf = transfer->rx_buf;
> +       dws->xfer.rx_dma = transfer->rx_dma;
> +
> +       dws->xfer.len = transfer->len;
> +       dws->xfer.n_bytes = controller->n_bytes;
> +       dws->xfer.sent = 0;
> +       dws->xfer.rcvd = 0;
> +       dws->xfer.msg = msg;
> +       dws->xfer.err = 0;
> +       dws->xfer.irq_status = 0;
> +       INIT_COMPLETION(dws->xfer.complete);
> +
> +       /* {tx, rx}_threshold should probably be a module param with
> +        *  some reasonable default but these work for now.
> +        */
> +       dws->xfer.tx_threshold = 10;
> +       dws->xfer.rx_threshold = dws->xfer.tx_threshold - 1;
> +
> +       /* we need to make the decsion about the type of transfer more
> +        *  inteligently but this works for now
> +        */
> +       if (transfer->len > dws->fifo_len)
> +               dws->xfer.type = INT_XFER;
> +       else
> +               dws->xfer.type = PIO_XFER;
> +
> +       if (controller->enable_dma &&
> +               msg->is_dma_mapped &&
> +               dws->dma_inited &&
> +               dws->dma_ops)
> +               dws->xfer.type = DMA_XFER;
> +
> +       /* Setup the controller based on parameters in the transfer
> +        * each transfer can set the bit_per_word and the speed_hz to
> +        * change these values in the controller the controller MUST
> +        * be disabled
> +        */
> +       if (unlikely(!controller->clk_div)) {
> +               controller->clk_div = dws->max_freq /
> controller->speed_hz;
> +               controller->clk_div = (controller->clk_div + 1) &
> 0xfffe;
> +               dw_spi_set_clk(dws, controller->clk_div);
> +               dev_err(&dws->master->dev, "setting default clk_div");
> +               err = 1;
> +       }
> +
> +       if (transfer->speed_hz) {
> +               if (transfer->speed_hz != controller->speed_hz) {
> +                       if (transfer->speed_hz > dws->max_freq) {
> +                               err = -EIO;
> +                               goto out;
> +                       }
> +
> +                       clk_div = dws->max_freq / transfer->speed_hz;
> +                       clk_div = (clk_div + 1) & 0xfffe;
> +                       controller->clk_div = clk_div;
> +                       controller->speed_hz = transfer->speed_hz;
> +                       err = 1;
> +               }
> +       }
> +
> +       if (transfer->bits_per_word) {
> +               if (transfer->bits_per_word != 8 &&
> +                       transfer->bits_per_word != 16) {
> +                       err = -EIO;
> +                       goto out;
> +               }
> +               cr0 &= ~SPI_DFS_MASK;
> +               cr0 |= transfer->bits_per_word - 1;
> +               err = 1;
> +       } else {
> +               cr0 &= ~SPI_DFS_MASK;
> +               cr0 |= controller->spi_dev->bits_per_word - 1;
> +       }
> +
> +       cr0 &= ~SPI_MODE_MASK;
> +       cr0 |= (controller->spi_dev->mode << SPI_MODE_OFFSET);
> +       controller->cr0 = cr0;
> +
> +       if (err || dw_readw(dws, ctrl0) != cr0) {
> +               dw_spi_disable(dws);
> +               dw_spi_chip_sel(dws,
> controller->spi_dev->chip_select);
> +               dw_writew(dws, ctrl0, cr0);
> +               dw_spi_set_clk(dws, controller->clk_div);
> +               dw_spi_enable(dws);
> +               err = 0;
> +       }
> +out:
> +       return err;
> +}
> +
> +static void tx_fifo_fill(struct dw_spi *dws)
> +{
> +       int room;
> +       u16 txw = 0;
> +       if (dws->xfer.sent < dws->xfer.len) {
> +               room = tx_max(dws);
> +               while (room--) {
> +                       if (dws->xfer.tx_buf) {
> +                               if (dws->xfer.n_bytes == 2)
> +                                       txw = *(u16
> *)dws->xfer.tx_buf;
> +                               else
> +                                       txw = *(u8 *)dws->xfer.tx_buf;
> +                               dws->xfer.tx_buf += dws->xfer.n_bytes;
> +                       }
> +                       dw_writew(dws, dr, txw);
> +                       dws->xfer.sent += dws->xfer.n_bytes;
> +               }
> +       }
> +}
> +
> +
> +static void rx_fifo_drain(struct dw_spi *dws)
> +{
> +       u16 rx_val;
> +       int avail;
> +
> +       if (dws->xfer.rcvd < dws->xfer.len) {
> +               avail = rx_max(dws);
> +               while (avail--) {
> +                       rx_val = dw_readw(dws, dr);
> +                       if (dws->xfer.rx_buf) {
> +                               if (dws->xfer.n_bytes == 2)
> +                                       *(u16 *)(dws->xfer.rx_buf) =
> +                                               (u16)rx_val;
> +                               else
> +                                       *dws->xfer.rx_buf =
> (u8)rx_val;
> +                               dws->xfer.rx_buf += dws->xfer.n_bytes;
> +                       }
> +                       dws->xfer.rcvd += dws->xfer.n_bytes;
> +               }
> +       }
> +}
> +
> +static inline void do_pio(struct dw_spi *dws)
> +{
> +       while (dws->xfer.sent < dws->xfer.len ||
> +                                       dws->xfer.rcvd <
> dws->xfer.len) {
> +               tx_fifo_fill(dws);
> +               rx_fifo_drain(dws);
> +               cpu_relax();
> +       }
> +       complete(&dws->xfer.complete);
> +}
> +
> +static irqreturn_t dw_spi_irq_thread_handler(int irq, void *dev_id)
> +{
> +       struct dw_spi *dws = dev_id;
> +
> +       if (dws->xfer.irq_status & SPI_INT_TXEI) {
> +               rx_fifo_drain(dws);
> +               tx_fifo_fill(dws);
> +       }
> +
> +       if (dws->xfer.irq_status & SPI_INT_RXFI) {
> +               tx_fifo_fill(dws);
> +               rx_fifo_drain(dws);
> +       }
> +
> +       if (dws->xfer.len == dws->xfer.rcvd &&
> +               dws->xfer.len == dws->xfer.sent) {
> +               complete(&dws->xfer.complete);
> +               goto out;
> +       }
> +
> +       dw_spi_umask_intr(dws, SPI_INT_ALL);
> +out:
> +       return IRQ_HANDLED;
> +}
> +
> +static inline void do_int_xfer(struct dw_spi *dws)
> +{
> +       dw_spi_disable(dws);
> +       dw_writew(dws, txfltr, dws->xfer.tx_threshold);
> +       dw_writew(dws, rxfltr, dws->xfer.rx_threshold);
> +       dw_spi_enable(dws);
> +       dw_readw(dws, icr);
> +       tx_fifo_fill(dws);
> +       dw_writew(dws, imr, SPI_INT_ALL);
> +}
> +
> +static inline int do_transfer(struct dw_spi *dws)
> +{
> +       switch (dws->xfer.type) {
> +       case PIO_XFER:
> +               do_pio(dws);
> +               break;
> +       case INT_XFER:
> +               do_int_xfer(dws);
> +               break;
> +       case DMA_XFER:
> +               dws->dma_ops->dma_transfer(dws);
> +               break;
> +       default:
> +               BUG();
> +       }
> +
> +       wait_for_completion(&dws->xfer.complete);
> +
> +
> +       return dws->xfer.err;
> +}
> +
> +static void drain_message_queue(struct dw_spi *dws)
> +{
> +       struct spi_message *message;
> +
> +       message = get_message(dws);
> +       while (message) {
> +               message->status = -ESHUTDOWN;
> +               message->complete(message->context);
> +               message = get_message(dws);
> +       }
> +}
> +
> +static void pump_messages(struct work_struct *work)
> +{
> +       struct dw_spi *dws =
> +               container_of(work, struct dw_spi, pump_messages);
> +       struct spi_transfer *transfer;
> +       struct spi_message *message;
> +       struct chip_data *controller;
> +       int err = 0;
> +
> +
> +       message = get_message(dws);
> +
> +       while (message && dws->run != QUEUE_STOPPED) {
> +               controller = spi_get_ctldata(message->spi);
> +               list_for_each_entry(transfer, &message->transfers,
> +                               transfer_list){
> +
> +                       err = transfer_setup(dws, controller,
> +                                               transfer, message);
> +                       if (err < 0) {
> +                               dev_err(&dws->master->dev,
> +                                       "transfer_setup failed");
> +                               dws->xfer.err = -EIO;
> +                               break;
> +                       }
> +
> +                       err = do_transfer(dws);
> +                       if (err < 0) {
> +                               dev_err(&dws->master->dev,
> +                                       "do_transfer failed");
> +                               break;
> +                       }
> +                       message->actual_length += dws->xfer.len;
> +
> +                       if (transfer->delay_usecs)
> +                               udelay(transfer->delay_usecs);
> +               }
> +
> +               message->status = dws->xfer.err;
> +               message->complete(message->context);
> +               message =  get_message(dws);
> +       }
> +       if (dws->run == QUEUE_STOPPED)
> +               drain_message_queue(dws);
> +}
> +
> +/* spi_device use this to queue in their spi_msg */
> +static int dw_spi_transfer(struct spi_device *spi, struct
> spi_message *msg) +{
> +       struct dw_spi *dws = spi_master_get_devdata(spi->master);
> +       unsigned long flags;
> +
> +
> +       spin_lock_irqsave(&dws->lock, flags);
> +
> +       msg->actual_length = 0;
> +       msg->status = -EINPROGRESS;
> +
> +       list_add_tail(&msg->queue, &dws->queue);
> +
> +       queue_work(dws->workqueue,
> +               &dws->pump_messages);
> +
> +       spin_unlock_irqrestore(&dws->lock, flags);
> +
> +       return 0;
> +}
> +
> +/* This may be called twice for each spi dev */
> +static int dw_spi_setup(struct spi_device *spi)
> +{
> +       struct dw_spi_chip *chip_info = NULL;
> +       struct chip_data *chip;
> +
> +       if (spi->bits_per_word != 8 && spi->bits_per_word != 16)
> +               return -EINVAL;
> +
> +       if (!spi->max_speed_hz) {
> +               dev_err(&spi->dev, "No max speed HZ parameter\n");
> +               return -EINVAL;
> +       }
> +
> +       /* Only alloc on first setup */
> +       chip = spi_get_ctldata(spi);
> +       if (!chip) {
> +               chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
> +               if (!chip)
> +                       return -ENOMEM;
> +       }
> +       chip->spi_dev = spi;
> +
> +       /*
> +        * Protocol drivers may change the chip settings, so...
> +        * if chip_info exists, use it
> +        */
> +       chip_info = spi->controller_data;
> +
> +       /* chip_info doesn't always exist */
> +       if (chip_info) {
> +               chip->type = chip_info->type;
> +               chip->enable_dma = chip_info->enable_dma;
> +       }
> +
> +       chip->bits_per_word = spi->bits_per_word;
> +       chip->n_bytes = chip->bits_per_word / 8;
> +       chip->dma_width = chip->bits_per_word / 8;
> +
> +       chip->speed_hz = spi->max_speed_hz;
> +
> +       /* Default SPI mode is SCPOL = 0, SCPH = 0 */
> +       chip->cr0 = (chip->bits_per_word - 1)
> +               | (chip->type << SPI_FRF_OFFSET)
> +               | (spi->mode  << SPI_MODE_OFFSET);
> +
> +       spi_set_ctldata(spi, chip);
> +       return 0;
> +}
> +
> +static void dw_spi_cleanup(struct spi_device *spi)
> +{
> +       struct chip_data *chip = spi_get_ctldata(spi);
> +       kfree(chip);
> +}
> +
> +static int __devinit init_queue(struct dw_spi *dws)
> +{
> +       INIT_LIST_HEAD(&dws->queue);
> +       spin_lock_init(&dws->lock);
> +
> +       dws->run = QUEUE_STOPPED;
> +
> +       INIT_WORK(&dws->pump_messages, pump_messages);
> +       dws->workqueue = create_singlethread_workqueue(
> +
> dev_name(dws->master->dev.parent));
> +       if (dws->workqueue == NULL)
> +               return -EBUSY;
> +       dws->run = QUEUE_RUNNING;
> +       return 0;
> +}
> +
> +
> +int dw_spi_stop_queue(struct dw_spi *dws)
> +{
> +       unsigned long flags;
> +       int status = 0;
> +
> +       spin_lock_irqsave(&dws->lock, flags);
> +       dws->run = QUEUE_STOPPED;
> +
> +       if (!list_empty(&dws->queue))
> +               status = -EBUSY;
> +       spin_unlock_irqrestore(&dws->lock, flags);
> +
> +       return status;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_stop_queue);
> +
> +static int destroy_queue(struct dw_spi *dws)
> +{
> +       int status;
> +
> +       status = dw_spi_stop_queue(dws);
> +       if (status != 0)
> +               return status;
> +       destroy_workqueue(dws->workqueue);
> +       return 0;
> +}
> +
> +/* Restart the controller, disable all interrupts, clean rx fifo */
> +static void spi_hw_init(struct dw_spi *dws)
> +{
> +       dw_spi_disable(dws);
> +       dw_spi_mask_intr(dws, 0xff);
> +       dw_readw(dws, icr);
> +       dw_spi_enable(dws);
> +
> +       BUG_ON(!dws->fifo_len);
> +}
> +
> +int __devinit dw_spi_add_host(struct dw_spi *dws)
> +{
> +       struct spi_master *master;
> +       int ret;
> +
> +       BUG_ON(dws == NULL);
> +
> +       master = spi_alloc_master(dws->parent_dev, 0);
> +       if (!master) {
> +               ret = -ENOMEM;
> +               goto exit;
> +       }
> +
> +       dws->master = master;
> +       dws->type = SSI_MOTO_SPI;
> +
> +       init_completion(&dws->xfer.complete);
> +       dws->dma_inited = 0;
> +       /* Change to address of FIFO */
> +       dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
> +
> +       ret = request_threaded_irq(dws->irq, dw_spi_irq,
> +                               dw_spi_irq_thread_handler,
> +                               IRQF_SHARED, "dw_spi", dws);
> +       if (ret < 0) {
> +               dev_err(&master->dev, "can not get IRQ\n");
> +               goto err_free_master;
> +       }
> +
> +       master->mode_bits = SPI_CPOL | SPI_CPHA;
> +       master->bus_num = dws->bus_num;
> +       master->num_chipselect = dws->num_cs;
> +       master->cleanup = dw_spi_cleanup;
> +       master->setup = dw_spi_setup;
> +       master->transfer = dw_spi_transfer;
> +
> +       /* Basic HW init */
> +       spi_hw_init(dws);
> +
> +       if (dws->dma_ops && dws->dma_ops->dma_init) {
> +               ret = dws->dma_ops->dma_init(dws);
> +               if (ret) {
> +                       dev_warn(&master->dev, "DMA init failed\n");
> +                       dws->dma_inited = 0;
> +               }
> +       }
> +
> +       /* Initial and start queue */
> +       ret = init_queue(dws);
> +       if (ret) {
> +               dev_err(&master->dev, "problem initializing queue\n");
> +               goto err_diable_hw;
> +       }
> +
> +       spi_master_set_devdata(master, dws);
> +       ret = spi_register_master(master);
> +       if (ret) {
> +               dev_err(&master->dev, "problem registering spi
> master\n");
> +               goto err_queue_alloc;
> +       }
> +
> +       mrst_spi_debugfs_init(dws);
> +       return 0;
> +
> +err_queue_alloc:
> +       destroy_queue(dws);
> +       if (dws->dma_ops && dws->dma_ops->dma_exit)
> +               dws->dma_ops->dma_exit(dws);
> +err_diable_hw:
> +       dw_spi_disable(dws);
> +       free_irq(dws->irq, dws);
> +err_free_master:
> +       spi_master_put(master);
> +exit:
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_add_host);
> +
> +void __devexit dw_spi_remove_host(struct dw_spi *dws)
> +{
> +       int status = 0;
> +
> +       if (!dws)
> +               return;
> +       mrst_spi_debugfs_remove(dws);
> +
> +       /* Remove the queue */
> +       status = destroy_queue(dws);
> +       if (status != 0)
> +               dev_err(&dws->master->dev, "dw_spi_remove: workqueue
> will not "
> +                       "complete, message memory not freed\n");
> +
> +       if (dws->dma_ops && dws->dma_ops->dma_exit)
> +               dws->dma_ops->dma_exit(dws);
> +       dw_spi_disable(dws);
> +       dw_readw(dws, icr);
> +       free_irq(dws->irq, dws);
> +
> +       /* Disconnect from the SPI framework */
> +       spi_unregister_master(dws->master);
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_remove_host);
> +
> +int dw_spi_suspend_host(struct dw_spi *dws)
> +{
> +       int ret = 0;
> +
> +       ret = dw_spi_stop_queue(dws);
> +       if (ret)
> +               return ret;
> +       dw_spi_disable(dws);
> +       return ret;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
> +
> +int dw_spi_resume_host(struct dw_spi *dws)
> +{
> +       spi_hw_init(dws);
> +       dws->run = QUEUE_RUNNING;
> +       return 0;
> +}
> +EXPORT_SYMBOL_GPL(dw_spi_resume_host);
> +
> +MODULE_AUTHOR("Feng Tang <feng.tang@intel.com>");
> +MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
> +MODULE_LICENSE("GPL v2");
> diff --git a/drivers/spi/spi-dw.h b/drivers/spi/spi-dw.h
> index 97baff6..b016b85 100644
> --- a/drivers/spi/spi-dw.h
> +++ b/drivers/spi/spi-dw.h
> @@ -7,6 +7,7 @@
> 
>  /* Bit fields in CTRLR0 */
>  #define SPI_DFS_OFFSET                 0
> +#define SPI_DFS_MASK                   0xf
> 
>  #define SPI_FRF_OFFSET                 4
>  #define SPI_FRF_SPI                    0x0
> @@ -17,6 +18,7 @@
>  #define SPI_MODE_OFFSET                        6
>  #define SPI_SCPH_OFFSET                        6
>  #define SPI_SCOL_OFFSET                        7
> +#define SPI_MODE_MASK                  (0x3 << SPI_MODE_OFFSET)
> 
>  #define SPI_TMOD_OFFSET                        8
>  #define SPI_TMOD_MASK                  (0x3 << SPI_TMOD_OFFSET)
> @@ -46,6 +48,7 @@
>  #define SPI_INT_RXOI                   (1 << 3)
>  #define SPI_INT_RXFI                   (1 << 4)
>  #define SPI_INT_MSTI                   (1 << 5)
> +#define SPI_INT_ALL  0x3f
> 
>  /* TX RX interrupt level threshold, max can be 256 */
>  #define SPI_INT_THRESHOLD              32
> @@ -83,65 +86,67 @@ struct dw_spi;
>  struct dw_spi_dma_ops {
>         int (*dma_init)(struct dw_spi *dws);
>         void (*dma_exit)(struct dw_spi *dws);
> -       int (*dma_transfer)(struct dw_spi *dws, int cs_change);
> +       int (*dma_transfer)(struct dw_spi *dws);
> +};
> +
> +enum xfer_type {
> +       PIO_XFER,
> +       INT_XFER,
> +       DMA_XFER,
> +};
> +
> +
> +struct xfer_state {
> +       const u8 *tx_buf;
> +       dma_addr_t tx_dma;
> +       u8 *rx_buf;
> +       dma_addr_t rx_dma;
> +       struct spi_message *msg;
> +       u32 n_bytes;
> +       u32 len;
> +       u32 sent;
> +       u32 rcvd;
> +       u32 err;
> +       u32 type;
> +       u32  tx_threshold;
> +       u32  rx_threshold;
> +       u32 irq_status;
> +       struct completion complete;
>  };
> 
>  struct dw_spi {
>         struct spi_master       *master;
> -       struct spi_device       *cur_dev;
>         struct device           *parent_dev;
>         enum dw_ssi_type        type;
> 
>         void __iomem            *regs;
>         unsigned long           paddr;
>         u32                     iolen;
> -       int                     irq;
> +       u32                     irq;
>         u32                     fifo_len;       /* depth of the FIFO
> buffer */ u32                     max_freq;       /* max bus freq
> supported */
> 
>         u16                     bus_num;
>         u16                     num_cs;         /* supported slave
> numbers */ -
> +
>         /* Driver message queue */
>         struct workqueue_struct *workqueue;
>         struct work_struct      pump_messages;
>         spinlock_t              lock;
>         struct list_head        queue;
> -       int                     busy;
> -       int                     run;
> -
> -       /* Message Transfer pump */
> -       struct tasklet_struct   pump_transfers;
> +       u32                     run;
> 
>         /* Current message transfer state info */
> -       struct spi_message      *cur_msg;
> -       struct spi_transfer     *cur_transfer;
> -       struct chip_data        *cur_chip;
> -       struct chip_data        *prev_chip;
> -       size_t                  len;
> -       void                    *tx;
> -       void                    *tx_end;
> -       void                    *rx;
> -       void                    *rx_end;
> -       int                     dma_mapped;
> -       dma_addr_t              rx_dma;
> -       dma_addr_t              tx_dma;
> -       size_t                  rx_map_len;
> -       size_t                  tx_map_len;
> -       u8                      n_bytes;        /* current is a 1/2
> bytes op */
> -       u8                      max_bits_per_word;      /* maxim is
> 16b */
> -       u32                     dma_width;
> -       int                     cs_change;
> -       irqreturn_t             (*transfer_handler)(struct dw_spi
> *dws);
> -       void                    (*cs_control)(u32 command);
> +       struct xfer_state       xfer;
> 
>         /* Dma info */
> -       int                     dma_inited;
> +//     int                     dma_width;
> +       u32                     dma_inited;
>         struct dma_chan         *txchan;
>         struct scatterlist      tx_sgl;
>         struct dma_chan         *rxchan;
>         struct scatterlist      rx_sgl;
> -       int                     dma_chan_done;
> +       u32                     dma_chan_done;
>         struct device           *dma_dev;
>         dma_addr_t              dma_addr; /* phy address of the Data
> register */ struct dw_spi_dma_ops   *dma_ops;
> @@ -163,30 +168,36 @@ struct dw_spi {
>         __raw_readw(&(((struct dw_spi_reg *)dw->regs)->name))
>  #define dw_writew(dw, name, val) \
>         __raw_writew((val), &(((struct dw_spi_reg *)dw->regs)->name))
> +#define dw_readb(dw, name) \
> +       __raw_readb(&(((struct dw_spi_reg *)dw->regs)->name))
> +#define dw_writeb(dw, name, val) \
> +       __raw_writeb((val), &(((struct dw_spi_reg *)dw->regs)->name))
> 
> -static inline void spi_enable_chip(struct dw_spi *dws, int enable)
> +static inline void dw_spi_disable(struct dw_spi *dws)
>  {
> -       dw_writel(dws, ssienr, (enable ? 1 : 0));
> +       dw_writel(dws, ssienr, 0);
>  }
> 
> -static inline void spi_set_clk(struct dw_spi *dws, u16 div)
> +static inline void dw_spi_enable(struct dw_spi *dws)
> +{
> +       dw_writel(dws, ssienr, 1);
> +}
> +
> +static inline void dw_spi_set_clk(struct dw_spi *dws, u16 div)
>  {
>         dw_writel(dws, baudr, div);
>  }
> 
> -static inline void spi_chip_sel(struct dw_spi *dws, u16 cs)
> +static inline void dw_spi_chip_sel(struct dw_spi *dws, u16 cs)
>  {
>         if (cs > dws->num_cs)
>                 return;
> 
> -       if (dws->cs_control)
> -               dws->cs_control(1);
> -
>         dw_writel(dws, ser, 1 << cs);
>  }
> 
>  /* Disable IRQ bits */
> -static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
> +static inline void dw_spi_mask_intr(struct dw_spi *dws, u32 mask)
>  {
>         u32 new_mask;
> 
> @@ -195,7 +206,7 @@ static inline void spi_mask_intr(struct dw_spi
> *dws, u32 mask) }
> 
>  /* Enable IRQ bits */
> -static inline void spi_umask_intr(struct dw_spi *dws, u32 mask)
> +static inline void dw_spi_umask_intr(struct dw_spi *dws, u32 mask)
>  {
>         u32 new_mask;
> 
> @@ -208,6 +219,7 @@ extern void dw_spi_remove_host(struct dw_spi
> *dws); extern int dw_spi_suspend_host(struct dw_spi *dws);
>  extern int dw_spi_resume_host(struct dw_spi *dws);
>  extern void dw_spi_xfer_done(struct dw_spi *dws);
> +extern int dw_spi_stop_queue(struct dw_spi *dws);
> 
>  /* platform related setup */
>  extern int dw_spi_mid_init(struct dw_spi *dws); /* Intel MID
> platforms */ --
> 1.7.3.4
> 
> 
> ------------------------------------------------------------------------------
> EditLive Enterprise is the world's most technically advanced content
> authoring tool. Experience the power of Track Changes, Inline Image
> Editing and ensure content is compliant with Accessibility Checking.
> http://p.sf.net/sfu/ephox-dev2dev
> _______________________________________________
> spi-devel-general mailing list
> spi-devel-general@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/spi-devel-general

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
dirk.brandewie@gmail.com June 16, 2011, 4:03 p.m. UTC | #3
On 06/16/2011 06:14 AM, Grant Likely wrote:
> On Wed, Jun 15, 2011 at 10:23:06AM -0700, dirk.brandewie@gmail.com wrote:
>> From: Dirk Brandewie<dirk.brandewie@gmail.com>
>>
>> NOTE: patch created git format-patch --break-rewrites=/50%
>>
>> This patch reworks the message pump worker thread function to run
>> until all messages queued to the driver have been handled. The
>> function to handle individual spi_transfers is now a synchronus
>> function the tasklet to handle spi_transfers has been removed. Work
>> for the worker thread is only queued in host controller transfer
>> function.
>>
>> Psuedo code for new thread function:
>>    message = get_message()
>>    while (message){
>>      for_each_transfer_in_msg(message){
>>        transfer_setup(transfer)
>>        do_transfer()
>>      }
>>      complete_message()
>>      message = get_message()
>>    }
>>
>> Changes that fell out of the message thread changes:
>> Non-DMA transfers that are larger than the size of the controller FIFO
>> are handled as interrupt driven transfers.
>>
>> Common FIFO handling functions shared PIO and interrupt transfers.
>>
>> Simplified queue stop/start funcitons.
>>
>> Cleanup fixes:
>> Changed exported all exported function names to have dw_spi_ prefix
>>
>> Removed support for registering chip select control function. Setting
>> the slave chip select is handled by setting the SER (Slave enable
>> register)
>
> What about for implementations that use a GPIO for the SPI chip
> select?  It is very common for board designs to use GPIOs for
> multiplexing the SPI bus.

OK I can put that back. I took it out for the following reason.

The Slave Enable Register (SER) must be set for the IP block to start the 
transfer.  The bit in SER directly control a separate chip select pin while 
there is an active transfer.  This gives four chip selects in the direct mapping 
case and up to fifteen if the slave chip selects are muxed together.

>
>>
>> Removed code that looked at the cs_change hint in the
>> spi_transfer. Software has no contorl over whether the slave chip
>> select is de-asserted at the end of the transfer.  Once the TX FIFO
>> goes empty the slave chip select is dropped.
>
> This sounds wrong.  cs_change is *not* merely a hint.  It must be respected.
> If the driver has no direct control over the CS line, then it is
> incumbent on the driver to guaranteed that the cs deassert condition
> does not occur.  This will probably mean chaining up all the transfers
> in a message so that the TX FIFO remains full.  If cs_change is
> requested, then the FIFO must be allowed to empty before kicking of
> the next group of transfers.
>

I agree it is wrong that the driver can not directly control the chip select :-)

I could probably manage to get the PIO and interrupt case to honor this but 
would add a fair bit of complexity, I have no idea how to make the DMA case work 
I am open to suggestions :-).  The current driver has this issue I just made it 
explicit maybe we should add a warning message in the driver to let client 
driver writers aware of the Silicon behaviour.

ATM I believe that my driver exhibits the same behaviour as the current driver. 
  I verified this with a logic analyser for the PIO and interrupt case.

>>
>> Added dw_spi_{en,dis}able inline functions to replace spi_enable_chip()
>> Added dw_spi_{mask,umask}_intr inline functions
>>
>> Signed-off-by: Dirk Brandewie<dirk.brandewie@gmail.com>
>
> As previously discussed both on-list and off, this is a major rework
> and I'd like to hear from the folks who are affected by this driver
> before I commit to merging it.  I don't like merging something so
> large that it ends up basically being a driver rewrite, but if there
> aren't any strong objections, I'll probably grudgingly accept it.
>

I agree, I have no desire to break anyone :-)  This driver is in use in all 
internal Meego builds for Moorsetown and Medfield based on v2.6.37+ that I am 
aware of.  It has been tested on the Moorsetown (Sooke harbor) and Meedfield 
(iCDK) development along with five different OEM designs.

It is also in use on multiple android based platforms.

> More comments below.
>
>> ---
>>   drivers/spi/spi-dw-mid.c |   43 +-
>>   drivers/spi/spi-dw.c     | 1670 ++++++++++++++++++++--------------------------
>>   drivers/spi/spi-dw.h     |   92 ++--
>>   3 files changed, 809 insertions(+), 996 deletions(-)
>>   rewrite drivers/spi/spi-dw.c (55%)
>>
>> diff --git a/drivers/spi/spi-dw-mid.c b/drivers/spi/spi-dw-mid.c
>> index 130e555..e44e37f 100644
>> --- a/drivers/spi/spi-dw-mid.c
>> +++ b/drivers/spi/spi-dw-mid.c
>> @@ -38,7 +38,10 @@ static bool mid_spi_dma_chan_filter(struct dma_chan *chan, void *param)
>>   {
>>   	struct dw_spi *dws = param;
>>
>> -	return dws->dmac&&  (&dws->dmac->dev == chan->device->dev);
>> +	if (dws->dmac&&  &dws->dmac->dev == chan->device->dev)
>> +		return true;
>> +	else
>> +		return false;
>
> Why?  The old code is correct, and the new code is more verbose.
>

I will back this change out
>>   }
>>
>>   static int mid_spi_dma_init(struct dw_spi *dws)
>> @@ -103,10 +106,10 @@ static void dw_spi_dma_done(void *arg)
>>
>>   	if (++dws->dma_chan_done != 2)
>>   		return;
>> -	dw_spi_xfer_done(dws);
>> +	complete(&dws->xfer.complete);
>>   }
>>
>> -static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>> +static int mid_spi_dma_transfer(struct dw_spi *dws)
>>   {
>>   	struct dma_async_tx_descriptor *txdesc = NULL, *rxdesc = NULL;
>>   	struct dma_chan *txchan, *rxchan;
>> @@ -114,17 +117,17 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>>   	u16 dma_ctrl = 0;
>>
>>   	/* 1. setup DMA related registers */
>> -	if (cs_change) {
>> -		spi_enable_chip(dws, 0);
>> -		dw_writew(dws, dmardlr, 0xf);
>> -		dw_writew(dws, dmatdlr, 0x10);
>> -		if (dws->tx_dma)
>> -			dma_ctrl |= 0x2;
>> -		if (dws->rx_dma)
>> -			dma_ctrl |= 0x1;
>> -		dw_writew(dws, dmacr, dma_ctrl);
>> -		spi_enable_chip(dws, 1);
>> -	}
>> +
>> +	dw_spi_disable(dws);
>> +	dw_writew(dws, dmardlr, 0xf);
>> +	dw_writew(dws, dmatdlr, 0x10);
>> +	if (dws->xfer.tx_dma)
>> +		dma_ctrl |= 0x2;
>> +	if (dws->xfer.rx_dma)
>> +		dma_ctrl |= 0x1;
>> +	dw_writew(dws, dmacr, dma_ctrl);
>> +	dw_spi_enable(dws);
>> +
>>
>>   	dws->dma_chan_done = 0;
>>   	txchan = dws->txchan;
>> @@ -141,8 +144,8 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>>   				       (unsigned long)&txconf);
>>
>>   	memset(&dws->tx_sgl, 0, sizeof(dws->tx_sgl));
>> -	dws->tx_sgl.dma_address = dws->tx_dma;
>> -	dws->tx_sgl.length = dws->len;
>> +	dws->tx_sgl.dma_address = dws->xfer.tx_dma;
>> +	dws->tx_sgl.length = dws->xfer.len;
>>
>>   	txdesc = txchan->device->device_prep_slave_sg(txchan,
>>   				&dws->tx_sgl,
>> @@ -163,8 +166,8 @@ static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
>>   				       (unsigned long)&rxconf);
>>
>>   	memset(&dws->rx_sgl, 0, sizeof(dws->rx_sgl));
>> -	dws->rx_sgl.dma_address = dws->rx_dma;
>> -	dws->rx_sgl.length = dws->len;
>> +	dws->rx_sgl.dma_address = dws->xfer.rx_dma;
>> +	dws->rx_sgl.length = dws->xfer.len;
>>
>>   	rxdesc = rxchan->device->device_prep_slave_sg(rxchan,
>>   				&dws->rx_sgl,
>> @@ -188,7 +191,6 @@ static struct dw_spi_dma_ops mid_dma_ops = {
>>   #endif
>>
>>   /* Some specific info for SPI0 controller on Moorestown */
>> -
>>   /* HW info for MRST CLk Control Unit, one 32b reg */
>>   #define MRST_SPI_CLK_BASE	100000000	/* 100m */
>>   #define MRST_CLK_SPI0_REG	0xff11d86c
>> @@ -202,12 +204,13 @@ int dw_spi_mid_init(struct dw_spi *dws)
>>   {
>>   	u32 *clk_reg, clk_cdiv;
>>
>> +
>
> nit: unrelated whitespace changes

ack

>
>>   	clk_reg = ioremap_nocache(MRST_CLK_SPI0_REG, 16);
>>   	if (!clk_reg)
>>   		return -ENOMEM;
>>
>>   	/* get SPI controller operating freq info */
>> -	clk_cdiv  = (readl(clk_reg)&  CLK_SPI_CDIV_MASK)>>  CLK_SPI_CDIV_OFFSET;
>> +	clk_cdiv  = ((*clk_reg)&  CLK_SPI_CDIV_MASK)>>  CLK_SPI_CDIV_OFFSET;
>
> Woah.  Changes correct usage of an io accessor to a direct dereference
> of a register?  That's not cool.  Why is a direct dereference needed
> here?  If you have to do something like this, then it must have a
> comment explaining why.
>

I don't remember exactly why I made this change I will back it out and test and 
add the appropriate comments if it is still needed.

>>   	dws->max_freq = MRST_SPI_CLK_BASE / (clk_cdiv + 1);
>>   	iounmap(clk_reg);
>>
>> diff --git a/drivers/spi/spi-dw.c b/drivers/spi/spi-dw.c
>> dissimilarity index 55%
>> index ece5f69..2dacb8f 100644
>> --- a/drivers/spi/spi-dw.c
>> +++ b/drivers/spi/spi-dw.c
>> @@ -1,936 +1,734 @@
>> -/*
>> - * Designware SPI core controller driver (refer pxa2xx_spi.c)
>> - *
>> - * Copyright (c) 2009, Intel Corporation.
>> - *
>> - * 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, write to the Free Software Foundation, Inc.,
>> - * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
>> - */
>> -
>> -#include<linux/dma-mapping.h>
>> -#include<linux/interrupt.h>
>> -#include<linux/highmem.h>
>> -#include<linux/delay.h>
>> -#include<linux/slab.h>
>> -#include<linux/spi/spi.h>
>> -
>> -#include "spi-dw.h"
>> -
>> -#ifdef CONFIG_DEBUG_FS
>> -#include<linux/debugfs.h>
>> -#endif
>> -
>> -#define START_STATE	((void *)0)
>> -#define RUNNING_STATE	((void *)1)
>> -#define DONE_STATE	((void *)2)
>> -#define ERROR_STATE	((void *)-1)
>> -
>> -#define QUEUE_RUNNING	0
>> -#define QUEUE_STOPPED	1
>> -
>> -#define MRST_SPI_DEASSERT	0
>> -#define MRST_SPI_ASSERT		1
>> -
>> -/* Slave spi_dev related */
>> -struct chip_data {
>> -	u16 cr0;
>> -	u8 cs;			/* chip select pin */
>> -	u8 n_bytes;		/* current is a 1/2/4 byte op */
>> -	u8 tmode;		/* TR/TO/RO/EEPROM */
>> -	u8 type;		/* SPI/SSP/MicroWire */
>> -
>> -	u8 poll_mode;		/* 1 means use poll mode */
>> -
>> -	u32 dma_width;
>> -	u32 rx_threshold;
>> -	u32 tx_threshold;
>> -	u8 enable_dma;
>> -	u8 bits_per_word;
>> -	u16 clk_div;		/* baud rate divider */
>> -	u32 speed_hz;		/* baud rate */
>> -	void (*cs_control)(u32 command);
>> -};
>> -
>> -#ifdef CONFIG_DEBUG_FS
>> -static int spi_show_regs_open(struct inode *inode, struct file *file)
>> -{
>> -	file->private_data = inode->i_private;
>> -	return 0;
>> -}
>> -
>> -#define SPI_REGS_BUFSIZE	1024
>> -static ssize_t  spi_show_regs(struct file *file, char __user *user_buf,
>> -				size_t count, loff_t *ppos)
>> -{
>> -	struct dw_spi *dws;
>> -	char *buf;
>> -	u32 len = 0;
>> -	ssize_t ret;
>> -
>> -	dws = file->private_data;
>> -
>> -	buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
>> -	if (!buf)
>> -		return 0;
>> -
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"MRST SPI0 registers:\n");
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"=================================\n");
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"SER: \t\t0x%08x\n", dw_readl(dws, ser));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"SR: \t\t0x%08x\n", dw_readl(dws, sr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"IMR: \t\t0x%08x\n", dw_readl(dws, imr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"ISR: \t\t0x%08x\n", dw_readl(dws, isr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"DMATDLR: \t0x%08x\n", dw_readl(dws, dmatdlr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"DMARDLR: \t0x%08x\n", dw_readl(dws, dmardlr));
>> -	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> -			"=================================\n");
>> -
>> -	ret =  simple_read_from_buffer(user_buf, count, ppos, buf, len);
>> -	kfree(buf);
>> -	return ret;
>> -}
>> -
>> -static const struct file_operations mrst_spi_regs_ops = {
>> -	.owner		= THIS_MODULE,
>> -	.open		= spi_show_regs_open,
>> -	.read		= spi_show_regs,
>> -	.llseek		= default_llseek,
>> -};
>> -
>> -static int mrst_spi_debugfs_init(struct dw_spi *dws)
>> -{
>> -	dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
>> -	if (!dws->debugfs)
>> -		return -ENOMEM;
>> -
>> -	debugfs_create_file("registers", S_IFREG | S_IRUGO,
>> -		dws->debugfs, (void *)dws,&mrst_spi_regs_ops);
>> -	return 0;
>> -}
>> -
>> -static void mrst_spi_debugfs_remove(struct dw_spi *dws)
>> -{
>> -	if (dws->debugfs)
>> -		debugfs_remove_recursive(dws->debugfs);
>> -}
>> -
>> -#else
>> -static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
>> -{
>> -	return 0;
>> -}
>> -
>> -static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
>> -{
>> -}
>> -#endif /* CONFIG_DEBUG_FS */
>> -
>> -/* Return the max entries we can fill into tx fifo */
>> -static inline u32 tx_max(struct dw_spi *dws)
>> -{
>> -	u32 tx_left, tx_room, rxtx_gap;
>> -
>> -	tx_left = (dws->tx_end - dws->tx) / dws->n_bytes;
>> -	tx_room = dws->fifo_len - dw_readw(dws, txflr);
>> -
>> -	/*
>> -	 * Another concern is about the tx/rx mismatch, we
>> -	 * though to use (dws->fifo_len - rxflr - txflr) as
>> -	 * one maximum value for tx, but it doesn't cover the
>> -	 * data which is out of tx/rx fifo and inside the
>> -	 * shift registers. So a control from sw point of
>> -	 * view is taken.
>> -	 */
>> -	rxtx_gap =  ((dws->rx_end - dws->rx) - (dws->tx_end - dws->tx))
>> -			/ dws->n_bytes;
>> -
>> -	return min3(tx_left, tx_room, (u32) (dws->fifo_len - rxtx_gap));
>> -}
>> -
>> -/* Return the max entries we should read out of rx fifo */
>> -static inline u32 rx_max(struct dw_spi *dws)
>> -{
>> -	u32 rx_left = (dws->rx_end - dws->rx) / dws->n_bytes;
>> -
>> -	return min(rx_left, (u32)dw_readw(dws, rxflr));
>> -}
>> -
>> -static void dw_writer(struct dw_spi *dws)
>> -{
>> -	u32 max = tx_max(dws);
>> -	u16 txw = 0;
>> -
>> -	while (max--) {
>> -		/* Set the tx word if the transfer's original "tx" is not null */
>> -		if (dws->tx_end - dws->len) {
>> -			if (dws->n_bytes == 1)
>> -				txw = *(u8 *)(dws->tx);
>> -			else
>> -				txw = *(u16 *)(dws->tx);
>> -		}
>> -		dw_writew(dws, dr, txw);
>> -		dws->tx += dws->n_bytes;
>> -	}
>> -}
>> -
>> -static void dw_reader(struct dw_spi *dws)
>> -{
>> -	u32 max = rx_max(dws);
>> -	u16 rxw;
>> -
>> -	while (max--) {
>> -		rxw = dw_readw(dws, dr);
>> -		/* Care rx only if the transfer's original "rx" is not null */
>> -		if (dws->rx_end - dws->len) {
>> -			if (dws->n_bytes == 1)
>> -				*(u8 *)(dws->rx) = rxw;
>> -			else
>> -				*(u16 *)(dws->rx) = rxw;
>> -		}
>> -		dws->rx += dws->n_bytes;
>> -	}
>> -}
>> -
>> -static void *next_transfer(struct dw_spi *dws)
>> -{
>> -	struct spi_message *msg = dws->cur_msg;
>> -	struct spi_transfer *trans = dws->cur_transfer;
>> -
>> -	/* Move to next transfer */
>> -	if (trans->transfer_list.next !=&msg->transfers) {
>> -		dws->cur_transfer =
>> -			list_entry(trans->transfer_list.next,
>> -					struct spi_transfer,
>> -					transfer_list);
>> -		return RUNNING_STATE;
>> -	} else
>> -		return DONE_STATE;
>> -}
>> -
>> -/*
>> - * Note: first step is the protocol driver prepares
>> - * a dma-capable memory, and this func just need translate
>> - * the virt addr to physical
>> - */
>> -static int map_dma_buffers(struct dw_spi *dws)
>> -{
>> -	if (!dws->cur_msg->is_dma_mapped
>> -		|| !dws->dma_inited
>> -		|| !dws->cur_chip->enable_dma
>> -		|| !dws->dma_ops)
>> -		return 0;
>> -
>> -	if (dws->cur_transfer->tx_dma)
>> -		dws->tx_dma = dws->cur_transfer->tx_dma;
>> -
>> -	if (dws->cur_transfer->rx_dma)
>> -		dws->rx_dma = dws->cur_transfer->rx_dma;
>> -
>> -	return 1;
>> -}
>> -
>> -/* Caller already set message->status; dma and pio irqs are blocked */
>> -static void giveback(struct dw_spi *dws)
>> -{
>> -	struct spi_transfer *last_transfer;
>> -	unsigned long flags;
>> -	struct spi_message *msg;
>> -
>> -	spin_lock_irqsave(&dws->lock, flags);
>> -	msg = dws->cur_msg;
>> -	dws->cur_msg = NULL;
>> -	dws->cur_transfer = NULL;
>> -	dws->prev_chip = dws->cur_chip;
>> -	dws->cur_chip = NULL;
>> -	dws->dma_mapped = 0;
>> -	queue_work(dws->workqueue,&dws->pump_messages);
>> -	spin_unlock_irqrestore(&dws->lock, flags);
>> -
>> -	last_transfer = list_entry(msg->transfers.prev,
>> -					struct spi_transfer,
>> -					transfer_list);
>> -
>> -	if (!last_transfer->cs_change&&  dws->cs_control)
>> -		dws->cs_control(MRST_SPI_DEASSERT);
>> -
>> -	msg->state = NULL;
>> -	if (msg->complete)
>> -		msg->complete(msg->context);
>> -}
>> -
>> -static void int_error_stop(struct dw_spi *dws, const char *msg)
>> -{
>> -	/* Stop the hw */
>> -	spi_enable_chip(dws, 0);
>> -
>> -	dev_err(&dws->master->dev, "%s\n", msg);
>> -	dws->cur_msg->state = ERROR_STATE;
>> -	tasklet_schedule(&dws->pump_transfers);
>> -}
>> -
>> -void dw_spi_xfer_done(struct dw_spi *dws)
>> -{
>> -	/* Update total byte transferred return count actual bytes read */
>> -	dws->cur_msg->actual_length += dws->len;
>> -
>> -	/* Move to next transfer */
>> -	dws->cur_msg->state = next_transfer(dws);
>> -
>> -	/* Handle end of message */
>> -	if (dws->cur_msg->state == DONE_STATE) {
>> -		dws->cur_msg->status = 0;
>> -		giveback(dws);
>> -	} else
>> -		tasklet_schedule(&dws->pump_transfers);
>> -}
>> -EXPORT_SYMBOL_GPL(dw_spi_xfer_done);
>> -
>> -static irqreturn_t interrupt_transfer(struct dw_spi *dws)
>> -{
>> -	u16 irq_status = dw_readw(dws, isr);
>> -
>> -	/* Error handling */
>> -	if (irq_status&  (SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI)) {
>> -		dw_readw(dws, txoicr);
>> -		dw_readw(dws, rxoicr);
>> -		dw_readw(dws, rxuicr);
>> -		int_error_stop(dws, "interrupt_transfer: fifo overrun/underrun");
>> -		return IRQ_HANDLED;
>> -	}
>> -
>> -	dw_reader(dws);
>> -	if (dws->rx_end == dws->rx) {
>> -		spi_mask_intr(dws, SPI_INT_TXEI);
>> -		dw_spi_xfer_done(dws);
>> -		return IRQ_HANDLED;
>> -	}
>> -	if (irq_status&  SPI_INT_TXEI) {
>> -		spi_mask_intr(dws, SPI_INT_TXEI);
>> -		dw_writer(dws);
>> -		/* Enable TX irq always, it will be disabled when RX finished */
>> -		spi_umask_intr(dws, SPI_INT_TXEI);
>> -	}
>> -
>> -	return IRQ_HANDLED;
>> -}
>> -
>> -static irqreturn_t dw_spi_irq(int irq, void *dev_id)
>> -{
>> -	struct dw_spi *dws = dev_id;
>> -	u16 irq_status = dw_readw(dws, isr)&  0x3f;
>> -
>> -	if (!irq_status)
>> -		return IRQ_NONE;
>> -
>> -	if (!dws->cur_msg) {
>> -		spi_mask_intr(dws, SPI_INT_TXEI);
>> -		return IRQ_HANDLED;
>> -	}
>> -
>> -	return dws->transfer_handler(dws);
>> -}
>> -
>> -/* Must be called inside pump_transfers() */
>> -static void poll_transfer(struct dw_spi *dws)
>> -{
>> -	do {
>> -		dw_writer(dws);
>> -		dw_reader(dws);
>> -		cpu_relax();
>> -	} while (dws->rx_end>  dws->rx);
>> -
>> -	dw_spi_xfer_done(dws);
>> -}
>> -
>> -static void pump_transfers(unsigned long data)
>> -{
>> -	struct dw_spi *dws = (struct dw_spi *)data;
>> -	struct spi_message *message = NULL;
>> -	struct spi_transfer *transfer = NULL;
>> -	struct spi_transfer *previous = NULL;
>> -	struct spi_device *spi = NULL;
>> -	struct chip_data *chip = NULL;
>> -	u8 bits = 0;
>> -	u8 imask = 0;
>> -	u8 cs_change = 0;
>> -	u16 txint_level = 0;
>> -	u16 clk_div = 0;
>> -	u32 speed = 0;
>> -	u32 cr0 = 0;
>> -
>> -	/* Get current state information */
>> -	message = dws->cur_msg;
>> -	transfer = dws->cur_transfer;
>> -	chip = dws->cur_chip;
>> -	spi = message->spi;
>> -
>> -	if (unlikely(!chip->clk_div))
>> -		chip->clk_div = dws->max_freq / chip->speed_hz;
>> -
>> -	if (message->state == ERROR_STATE) {
>> -		message->status = -EIO;
>> -		goto early_exit;
>> -	}
>> -
>> -	/* Handle end of message */
>> -	if (message->state == DONE_STATE) {
>> -		message->status = 0;
>> -		goto early_exit;
>> -	}
>> -
>> -	/* Delay if requested at end of transfer*/
>> -	if (message->state == RUNNING_STATE) {
>> -		previous = list_entry(transfer->transfer_list.prev,
>> -					struct spi_transfer,
>> -					transfer_list);
>> -		if (previous->delay_usecs)
>> -			udelay(previous->delay_usecs);
>> -	}
>> -
>> -	dws->n_bytes = chip->n_bytes;
>> -	dws->dma_width = chip->dma_width;
>> -	dws->cs_control = chip->cs_control;
>> -
>> -	dws->rx_dma = transfer->rx_dma;
>> -	dws->tx_dma = transfer->tx_dma;
>> -	dws->tx = (void *)transfer->tx_buf;
>> -	dws->tx_end = dws->tx + transfer->len;
>> -	dws->rx = transfer->rx_buf;
>> -	dws->rx_end = dws->rx + transfer->len;
>> -	dws->cs_change = transfer->cs_change;
>> -	dws->len = dws->cur_transfer->len;
>> -	if (chip != dws->prev_chip)
>> -		cs_change = 1;
>> -
>> -	cr0 = chip->cr0;
>> -
>> -	/* Handle per transfer options for bpw and speed */
>> -	if (transfer->speed_hz) {
>> -		speed = chip->speed_hz;
>> -
>> -		if (transfer->speed_hz != speed) {
>> -			speed = transfer->speed_hz;
>> -			if (speed>  dws->max_freq) {
>> -				printk(KERN_ERR "MRST SPI0: unsupported"
>> -					"freq: %dHz\n", speed);
>> -				message->status = -EIO;
>> -				goto early_exit;
>> -			}
>> -
>> -			/* clk_div doesn't support odd number */
>> -			clk_div = dws->max_freq / speed;
>> -			clk_div = (clk_div + 1)&  0xfffe;
>> -
>> -			chip->speed_hz = speed;
>> -			chip->clk_div = clk_div;
>> -		}
>> -	}
>> -	if (transfer->bits_per_word) {
>> -		bits = transfer->bits_per_word;
>> -
>> -		switch (bits) {
>> -		case 8:
>> -		case 16:
>> -			dws->n_bytes = dws->dma_width = bits>>  3;
>> -			break;
>> -		default:
>> -			printk(KERN_ERR "MRST SPI0: unsupported bits:"
>> -				"%db\n", bits);
>> -			message->status = -EIO;
>> -			goto early_exit;
>> -		}
>> -
>> -		cr0 = (bits - 1)
>> -			| (chip->type<<  SPI_FRF_OFFSET)
>> -			| (spi->mode<<  SPI_MODE_OFFSET)
>> -			| (chip->tmode<<  SPI_TMOD_OFFSET);
>> -	}
>> -	message->state = RUNNING_STATE;
>> -
>> -	/*
>> -	 * Adjust transfer mode if necessary. Requires platform dependent
>> -	 * chipselect mechanism.
>> -	 */
>> -	if (dws->cs_control) {
>> -		if (dws->rx&&  dws->tx)
>> -			chip->tmode = SPI_TMOD_TR;
>> -		else if (dws->rx)
>> -			chip->tmode = SPI_TMOD_RO;
>> -		else
>> -			chip->tmode = SPI_TMOD_TO;
>> -
>> -		cr0&= ~SPI_TMOD_MASK;
>> -		cr0 |= (chip->tmode<<  SPI_TMOD_OFFSET);
>> -	}
>> -
>> -	/* Check if current transfer is a DMA transaction */
>> -	dws->dma_mapped = map_dma_buffers(dws);
>> -
>> -	/*
>> -	 * Interrupt mode
>> -	 * we only need set the TXEI IRQ, as TX/RX always happen syncronizely
>> -	 */
>> -	if (!dws->dma_mapped&&  !chip->poll_mode) {
>> -		int templen = dws->len / dws->n_bytes;
>> -		txint_level = dws->fifo_len / 2;
>> -		txint_level = (templen>  txint_level) ? txint_level : templen;
>> -
>> -		imask |= SPI_INT_TXEI | SPI_INT_TXOI | SPI_INT_RXUI | SPI_INT_RXOI;
>> -		dws->transfer_handler = interrupt_transfer;
>> -	}
>> -
>> -	/*
>> -	 * Reprogram registers only if
>> -	 *	1. chip select changes
>> -	 *	2. clk_div is changed
>> -	 *	3. control value changes
>> -	 */
>> -	if (dw_readw(dws, ctrl0) != cr0 || cs_change || clk_div || imask) {
>> -		spi_enable_chip(dws, 0);
>> -
>> -		if (dw_readw(dws, ctrl0) != cr0)
>> -			dw_writew(dws, ctrl0, cr0);
>> -
>> -		spi_set_clk(dws, clk_div ? clk_div : chip->clk_div);
>> -		spi_chip_sel(dws, spi->chip_select);
>> -
>> -		/* Set the interrupt mask, for poll mode just disable all int */
>> -		spi_mask_intr(dws, 0xff);
>> -		if (imask)
>> -			spi_umask_intr(dws, imask);
>> -		if (txint_level)
>> -			dw_writew(dws, txfltr, txint_level);
>> -
>> -		spi_enable_chip(dws, 1);
>> -		if (cs_change)
>> -			dws->prev_chip = chip;
>> -	}
>> -
>> -	if (dws->dma_mapped)
>> -		dws->dma_ops->dma_transfer(dws, cs_change);
>> -
>> -	if (chip->poll_mode)
>> -		poll_transfer(dws);
>> -
>> -	return;
>> -
>> -early_exit:
>> -	giveback(dws);
>> -	return;
>> -}
>> -
>> -static void pump_messages(struct work_struct *work)
>> -{
>> -	struct dw_spi *dws =
>> -		container_of(work, struct dw_spi, pump_messages);
>> -	unsigned long flags;
>> -
>> -	/* Lock queue and check for queue work */
>> -	spin_lock_irqsave(&dws->lock, flags);
>> -	if (list_empty(&dws->queue) || dws->run == QUEUE_STOPPED) {
>> -		dws->busy = 0;
>> -		spin_unlock_irqrestore(&dws->lock, flags);
>> -		return;
>> -	}
>> -
>> -	/* Make sure we are not already running a message */
>> -	if (dws->cur_msg) {
>> -		spin_unlock_irqrestore(&dws->lock, flags);
>> -		return;
>> -	}
>> -
>> -	/* Extract head of queue */
>> -	dws->cur_msg = list_entry(dws->queue.next, struct spi_message, queue);
>> -	list_del_init(&dws->cur_msg->queue);
>> -
>> -	/* Initial message state*/
>> -	dws->cur_msg->state = START_STATE;
>> -	dws->cur_transfer = list_entry(dws->cur_msg->transfers.next,
>> -						struct spi_transfer,
>> -						transfer_list);
>> -	dws->cur_chip = spi_get_ctldata(dws->cur_msg->spi);
>> -
>> -	/* Mark as busy and launch transfers */
>> -	tasklet_schedule(&dws->pump_transfers);
>> -
>> -	dws->busy = 1;
>> -	spin_unlock_irqrestore(&dws->lock, flags);
>> -}
>> -
>> -/* spi_device use this to queue in their spi_msg */
>> -static int dw_spi_transfer(struct spi_device *spi, struct spi_message *msg)
>> -{
>> -	struct dw_spi *dws = spi_master_get_devdata(spi->master);
>> -	unsigned long flags;
>> -
>> -	spin_lock_irqsave(&dws->lock, flags);
>> -
>> -	if (dws->run == QUEUE_STOPPED) {
>> -		spin_unlock_irqrestore(&dws->lock, flags);
>> -		return -ESHUTDOWN;
>> -	}
>> -
>> -	msg->actual_length = 0;
>> -	msg->status = -EINPROGRESS;
>> -	msg->state = START_STATE;
>> -
>> -	list_add_tail(&msg->queue,&dws->queue);
>> -
>> -	if (dws->run == QUEUE_RUNNING&&  !dws->busy) {
>> -
>> -		if (dws->cur_transfer || dws->cur_msg)
>> -			queue_work(dws->workqueue,
>> -					&dws->pump_messages);
>> -		else {
>> -			/* If no other data transaction in air, just go */
>> -			spin_unlock_irqrestore(&dws->lock, flags);
>> -			pump_messages(&dws->pump_messages);
>> -			return 0;
>> -		}
>> -	}
>> -
>> -	spin_unlock_irqrestore(&dws->lock, flags);
>> -	return 0;
>> -}
>> -
>> -/* This may be called twice for each spi dev */
>> -static int dw_spi_setup(struct spi_device *spi)
>> -{
>> -	struct dw_spi_chip *chip_info = NULL;
>> -	struct chip_data *chip;
>> -
>> -	if (spi->bits_per_word != 8&&  spi->bits_per_word != 16)
>> -		return -EINVAL;
>> -
>> -	/* Only alloc on first setup */
>> -	chip = spi_get_ctldata(spi);
>> -	if (!chip) {
>> -		chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
>> -		if (!chip)
>> -			return -ENOMEM;
>> -	}
>> -
>> -	/*
>> -	 * Protocol drivers may change the chip settings, so...
>> -	 * if chip_info exists, use it
>> -	 */
>> -	chip_info = spi->controller_data;
>> -
>> -	/* chip_info doesn't always exist */
>> -	if (chip_info) {
>> -		if (chip_info->cs_control)
>> -			chip->cs_control = chip_info->cs_control;
>> -
>> -		chip->poll_mode = chip_info->poll_mode;
>> -		chip->type = chip_info->type;
>> -
>> -		chip->rx_threshold = 0;
>> -		chip->tx_threshold = 0;
>> -
>> -		chip->enable_dma = chip_info->enable_dma;
>> -	}
>> -
>> -	if (spi->bits_per_word<= 8) {
>> -		chip->n_bytes = 1;
>> -		chip->dma_width = 1;
>> -	} else if (spi->bits_per_word<= 16) {
>> -		chip->n_bytes = 2;
>> -		chip->dma_width = 2;
>> -	} else {
>> -		/* Never take>16b case for MRST SPIC */
>> -		dev_err(&spi->dev, "invalid wordsize\n");
>> -		return -EINVAL;
>> -	}
>> -	chip->bits_per_word = spi->bits_per_word;
>> -
>> -	if (!spi->max_speed_hz) {
>> -		dev_err(&spi->dev, "No max speed HZ parameter\n");
>> -		return -EINVAL;
>> -	}
>> -	chip->speed_hz = spi->max_speed_hz;
>> -
>> -	chip->tmode = 0; /* Tx&  Rx */
>> -	/* Default SPI mode is SCPOL = 0, SCPH = 0 */
>> -	chip->cr0 = (chip->bits_per_word - 1)
>> -			| (chip->type<<  SPI_FRF_OFFSET)
>> -			| (spi->mode<<  SPI_MODE_OFFSET)
>> -			| (chip->tmode<<  SPI_TMOD_OFFSET);
>> -
>> -	spi_set_ctldata(spi, chip);
>> -	return 0;
>> -}
>> -
>> -static void dw_spi_cleanup(struct spi_device *spi)
>> -{
>> -	struct chip_data *chip = spi_get_ctldata(spi);
>> -	kfree(chip);
>> -}
>> -
>> -static int __devinit init_queue(struct dw_spi *dws)
>> -{
>> -	INIT_LIST_HEAD(&dws->queue);
>> -	spin_lock_init(&dws->lock);
>> -
>> -	dws->run = QUEUE_STOPPED;
>> -	dws->busy = 0;
>> -
>> -	tasklet_init(&dws->pump_transfers,
>> -			pump_transfers,	(unsigned long)dws);
>> -
>> -	INIT_WORK(&dws->pump_messages, pump_messages);
>> -	dws->workqueue = create_singlethread_workqueue(
>> -					dev_name(dws->master->dev.parent));
>> -	if (dws->workqueue == NULL)
>> -		return -EBUSY;
>> -
>> -	return 0;
>> -}
>> -
>> -static int start_queue(struct dw_spi *dws)
>> -{
>> -	unsigned long flags;
>> -
>> -	spin_lock_irqsave(&dws->lock, flags);
>> -
>> -	if (dws->run == QUEUE_RUNNING || dws->busy) {
>> -		spin_unlock_irqrestore(&dws->lock, flags);
>> -		return -EBUSY;
>> -	}
>> -
>> -	dws->run = QUEUE_RUNNING;
>> -	dws->cur_msg = NULL;
>> -	dws->cur_transfer = NULL;
>> -	dws->cur_chip = NULL;
>> -	dws->prev_chip = NULL;
>> -	spin_unlock_irqrestore(&dws->lock, flags);
>> -
>> -	queue_work(dws->workqueue,&dws->pump_messages);
>> -
>> -	return 0;
>> -}
>> -
>> -static int stop_queue(struct dw_spi *dws)
>> -{
>> -	unsigned long flags;
>> -	unsigned limit = 50;
>> -	int status = 0;
>> -
>> -	spin_lock_irqsave(&dws->lock, flags);
>> -	dws->run = QUEUE_STOPPED;
>> -	while ((!list_empty(&dws->queue) || dws->busy)&&  limit--) {
>> -		spin_unlock_irqrestore(&dws->lock, flags);
>> -		msleep(10);
>> -		spin_lock_irqsave(&dws->lock, flags);
>> -	}
>> -
>> -	if (!list_empty(&dws->queue) || dws->busy)
>> -		status = -EBUSY;
>> -	spin_unlock_irqrestore(&dws->lock, flags);
>> -
>> -	return status;
>> -}
>> -
>> -static int destroy_queue(struct dw_spi *dws)
>> -{
>> -	int status;
>> -
>> -	status = stop_queue(dws);
>> -	if (status != 0)
>> -		return status;
>> -	destroy_workqueue(dws->workqueue);
>> -	return 0;
>> -}
>> -
>> -/* Restart the controller, disable all interrupts, clean rx fifo */
>> -static void spi_hw_init(struct dw_spi *dws)
>> -{
>> -	spi_enable_chip(dws, 0);
>> -	spi_mask_intr(dws, 0xff);
>> -	spi_enable_chip(dws, 1);
>> -
>> -	/*
>> -	 * Try to detect the FIFO depth if not set by interface driver,
>> -	 * the depth could be from 2 to 256 from HW spec
>> -	 */
>> -	if (!dws->fifo_len) {
>> -		u32 fifo;
>> -		for (fifo = 2; fifo<= 257; fifo++) {
>> -			dw_writew(dws, txfltr, fifo);
>> -			if (fifo != dw_readw(dws, txfltr))
>> -				break;
>> -		}
>> -
>> -		dws->fifo_len = (fifo == 257) ? 0 : fifo;
>> -		dw_writew(dws, txfltr, 0);
>> -	}
>> -}
>> -
>> -int __devinit dw_spi_add_host(struct dw_spi *dws)
>> -{
>> -	struct spi_master *master;
>> -	int ret;
>> -
>> -	BUG_ON(dws == NULL);
>> -
>> -	master = spi_alloc_master(dws->parent_dev, 0);
>> -	if (!master) {
>> -		ret = -ENOMEM;
>> -		goto exit;
>> -	}
>> -
>> -	dws->master = master;
>> -	dws->type = SSI_MOTO_SPI;
>> -	dws->prev_chip = NULL;
>> -	dws->dma_inited = 0;
>> -	dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
>> -
>> -	ret = request_irq(dws->irq, dw_spi_irq, IRQF_SHARED,
>> -			"dw_spi", dws);
>> -	if (ret<  0) {
>> -		dev_err(&master->dev, "can not get IRQ\n");
>> -		goto err_free_master;
>> -	}
>> -
>> -	master->mode_bits = SPI_CPOL | SPI_CPHA;
>> -	master->bus_num = dws->bus_num;
>> -	master->num_chipselect = dws->num_cs;
>> -	master->cleanup = dw_spi_cleanup;
>> -	master->setup = dw_spi_setup;
>> -	master->transfer = dw_spi_transfer;
>> -
>> -	/* Basic HW init */
>> -	spi_hw_init(dws);
>> -
>> -	if (dws->dma_ops&&  dws->dma_ops->dma_init) {
>> -		ret = dws->dma_ops->dma_init(dws);
>> -		if (ret) {
>> -			dev_warn(&master->dev, "DMA init failed\n");
>> -			dws->dma_inited = 0;
>> -		}
>> -	}
>> -
>> -	/* Initial and start queue */
>> -	ret = init_queue(dws);
>> -	if (ret) {
>> -		dev_err(&master->dev, "problem initializing queue\n");
>> -		goto err_diable_hw;
>> -	}
>> -	ret = start_queue(dws);
>> -	if (ret) {
>> -		dev_err(&master->dev, "problem starting queue\n");
>> -		goto err_diable_hw;
>> -	}
>> -
>> -	spi_master_set_devdata(master, dws);
>> -	ret = spi_register_master(master);
>> -	if (ret) {
>> -		dev_err(&master->dev, "problem registering spi master\n");
>> -		goto err_queue_alloc;
>> -	}
>> -
>> -	mrst_spi_debugfs_init(dws);
>> -	return 0;
>> -
>> -err_queue_alloc:
>> -	destroy_queue(dws);
>> -	if (dws->dma_ops&&  dws->dma_ops->dma_exit)
>> -		dws->dma_ops->dma_exit(dws);
>> -err_diable_hw:
>> -	spi_enable_chip(dws, 0);
>> -	free_irq(dws->irq, dws);
>> -err_free_master:
>> -	spi_master_put(master);
>> -exit:
>> -	return ret;
>> -}
>> -EXPORT_SYMBOL_GPL(dw_spi_add_host);
>> -
>> -void __devexit dw_spi_remove_host(struct dw_spi *dws)
>> -{
>> -	int status = 0;
>> -
>> -	if (!dws)
>> -		return;
>> -	mrst_spi_debugfs_remove(dws);
>> -
>> -	/* Remove the queue */
>> -	status = destroy_queue(dws);
>> -	if (status != 0)
>> -		dev_err(&dws->master->dev, "dw_spi_remove: workqueue will not "
>> -			"complete, message memory not freed\n");
>> -
>> -	if (dws->dma_ops&&  dws->dma_ops->dma_exit)
>> -		dws->dma_ops->dma_exit(dws);
>> -	spi_enable_chip(dws, 0);
>> -	/* Disable clk */
>> -	spi_set_clk(dws, 0);
>> -	free_irq(dws->irq, dws);
>> -
>> -	/* Disconnect from the SPI framework */
>> -	spi_unregister_master(dws->master);
>> -}
>> -EXPORT_SYMBOL_GPL(dw_spi_remove_host);
>> -
>> -int dw_spi_suspend_host(struct dw_spi *dws)
>> -{
>> -	int ret = 0;
>> -
>> -	ret = stop_queue(dws);
>> -	if (ret)
>> -		return ret;
>> -	spi_enable_chip(dws, 0);
>> -	spi_set_clk(dws, 0);
>> -	return ret;
>> -}
>> -EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
>> -
>> -int dw_spi_resume_host(struct dw_spi *dws)
>> -{
>> -	int ret;
>> -
>> -	spi_hw_init(dws);
>> -	ret = start_queue(dws);
>> -	if (ret)
>> -		dev_err(&dws->master->dev, "fail to start queue (%d)\n", ret);
>> -	return ret;
>> -}
>> -EXPORT_SYMBOL_GPL(dw_spi_resume_host);
>> -
>> -MODULE_AUTHOR("Feng Tang<feng.tang@intel.com>");
>> -MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
>> -MODULE_LICENSE("GPL v2");
>> +/*
>> + * dw_spi.c - Designware SPI core controller driver
>
> drop the filename.  It's incorrect, and not useful.  The useful bit is
> the driver description.
>

OK

>> + *
>> + * Copyright (c) 2009, Intel Corporation.
>> + *
>> + * 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, write to the Free Software Foundation, Inc.,
>> + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
>> + */
>> +
>> +#include<linux/dma-mapping.h>
>> +#include<linux/interrupt.h>
>> +#include<linux/highmem.h>
>> +#include<linux/delay.h>
>> +#include<linux/slab.h>
>> +#include<linux/spi/spi.h>
>> +
>> +#include "spi-dw.h"
>> +
>> +#ifdef CONFIG_DEBUG_FS
>> +#include<linux/debugfs.h>
>> +#endif
>> +
>> +
>> +#define QUEUE_RUNNING	0
>> +#define QUEUE_STOPPED	1
>> +
>> +
>> +/* Slave spi_dev related */
>> +struct chip_data {
>> +	struct spi_device *spi_dev;
>> +	u32 cr0;
>> +	u32 cs;			/* chip select pin */
>> +	u32 n_bytes;		/* current is a 1/2/4 byte op */
>> +	u32 type;		/* SPI/SSP/MicroWire */
>> +
>> +	u32 dma_width;
>> +	u32 enable_dma;
>> +	u32 bits_per_word;
>> +	u32 clk_div;		/* baud rate divider */
>> +	u32 speed_hz;		/* baud rate */
>> +};
>> +
>> +#ifdef CONFIG_DEBUG_FS
>> +static int spi_show_regs_open(struct inode *inode, struct file *file)
>> +{
>> +	file->private_data = inode->i_private;
>> +	return 0;
>> +}
>> +
>> +#define SPI_REGS_BUFSIZE	1024
>> +static ssize_t  spi_show_regs(struct file *file, char __user *user_buf,
>> +				size_t count, loff_t *ppos)
>> +{
>> +	struct dw_spi *dws;
>> +	char *buf;
>> +	u32 len = 0;
>> +	ssize_t ret;
>> +
>> +	dws = file->private_data;
>> +
>> +	buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
>> +	if (!buf)
>> +		return 0;
>> +
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"MRST SPI0 registers:\n");
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"=================================\n");
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"SER: \t\t0x%08x\n", dw_readl(dws, ser));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"SR: \t\t0x%08x\n", dw_readl(dws, sr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"IMR: \t\t0x%08x\n", dw_readl(dws, imr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"ISR: \t\t0x%08x\n", dw_readl(dws, isr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"DMATDLR: \t0x%08x\n", dw_readl(dws, dmatdlr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"DMARDLR: \t0x%08x\n", dw_readl(dws, dmardlr));
>> +	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
>> +			"=================================\n");
>> +
>> +	ret =  simple_read_from_buffer(user_buf, count, ppos, buf, len);
>> +	kfree(buf);
>> +	return ret;
>> +}
>> +
>> +static const struct file_operations mrst_spi_regs_ops = {
>> +	.owner		= THIS_MODULE,
>> +	.open		= spi_show_regs_open,
>> +	.read		= spi_show_regs,
>> +	.llseek		= default_llseek,
>> +};
>> +
>> +static int mrst_spi_debugfs_init(struct dw_spi *dws)
>> +{
>> +	dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
>> +	if (!dws->debugfs)
>> +		return -ENOMEM;
>> +
>> +	debugfs_create_file("registers", S_IFREG | S_IRUGO,
>> +		dws->debugfs, (void *)dws,&mrst_spi_regs_ops);
>> +	return 0;
>> +}
>> +
>> +static void mrst_spi_debugfs_remove(struct dw_spi *dws)
>> +{
>> +	if (dws->debugfs)
>> +		debugfs_remove_recursive(dws->debugfs);
>> +}
>> +
>> +#else
>> +static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
>> +{
>> +	return 0;
>> +}
>> +
>> +static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
>> +{
>> +}
>> +#endif /* CONFIG_DEBUG_FS */
>> +
>> +static irqreturn_t dw_spi_irq(int irq, void *dev_id)
>> +{
>> +	struct dw_spi *dws = dev_id;
>> +	u16  irq_mask = 0x3f;
>> +
>> +	dws->xfer.irq_status = dw_readw(dws, isr)&  irq_mask;
>> +
>> +	if (!dws->xfer.irq_status)
>> +		return IRQ_NONE;
>> +	if (dws->xfer.irq_status&
>> +			(SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI)) {
>> +		dw_readw(dws, txoicr);
>> +		dw_readw(dws, rxoicr);
>> +		dw_readw(dws, rxuicr);
>> +		dws->xfer.err = -EIO;
>> +		dw_spi_disable(dws);
>> +		complete(&dws->xfer.complete);
>> +		return IRQ_HANDLED;
>> +	}
>> +
>> +	/* disable interrupts */
>> +	dw_spi_mask_intr(dws, irq_mask);
>> +	return IRQ_WAKE_THREAD;
>> +}
>> +struct spi_message *get_message(struct dw_spi *dws)
>> +{
>> +	struct spi_message *message = NULL;
>> +	unsigned long flags;
>> +
>> +	spin_lock_irqsave(&dws->lock, flags);
>> +	if (!list_empty(&dws->queue)) {
>> +		message = list_entry(dws->queue.next,
>> +				struct spi_message, queue);
>> +		list_del_init(&message->queue);
>> +	}
>> +	spin_unlock_irqrestore(&dws->lock, flags);
>> +	return message;
>> +}
>> +static inline u32 tx_max(struct dw_spi *dws)
>> +{
>> +	u32 tx_left, tx_room;
>> +
>> +	tx_left = (dws->xfer.len - dws->xfer.sent) / dws->xfer.n_bytes;
>> +	tx_room = (dws->fifo_len - dw_readw(dws, txflr));
>> +
>> +	return min(tx_left, tx_room);
>> +}
>> +
>> +/* Return the max entries we should read out of rx fifo */
>> +static inline u32 rx_max(struct dw_spi *dws)
>> +{
>> +	u32 rx_left = (dws->xfer.len - dws->xfer.rcvd) / dws->xfer.n_bytes;
>> +	return min(rx_left, (u32)dw_readw(dws, rxflr));
>> +}
>> +
>> +static int transfer_setup(struct dw_spi *dws, struct chip_data *controller,
>> +			struct spi_transfer *transfer, struct spi_message *msg)
>> +{
>> +	int err = 0;
>> +	u32 cr0 = controller->cr0;
>> +	u32 clk_div;
>> +
>> +	dws->xfer.tx_buf = transfer->tx_buf;
>> +	dws->xfer.tx_dma = transfer->tx_dma;
>> +	dws->xfer.rx_buf = transfer->rx_buf;
>> +	dws->xfer.rx_dma = transfer->rx_dma;
>> +
>> +	dws->xfer.len = transfer->len;
>> +	dws->xfer.n_bytes = controller->n_bytes;
>> +	dws->xfer.sent = 0;
>> +	dws->xfer.rcvd = 0;
>> +	dws->xfer.msg = msg;
>> +	dws->xfer.err = 0;
>> +	dws->xfer.irq_status = 0;
>> +	INIT_COMPLETION(dws->xfer.complete);
>> +
>> +	/* {tx, rx}_threshold should probably be a module param with
>> +	 *  some reasonable default but these work for now.
>> +	 */
>> +	dws->xfer.tx_threshold = 10;
>> +	dws->xfer.rx_threshold = dws->xfer.tx_threshold - 1;
>> +
>> +	/* we need to make the decsion about the type of transfer more
>> +	 *  inteligently but this works for now
>> +	 */
>> +	if (transfer->len>  dws->fifo_len)
>> +		dws->xfer.type = INT_XFER;
>> +	else
>> +		dws->xfer.type = PIO_XFER;
>> +
>> +	if (controller->enable_dma&&
>> +		msg->is_dma_mapped&&
>> +		dws->dma_inited&&
>> +		dws->dma_ops)
>> +		dws->xfer.type = DMA_XFER;
>> +
>> +	/* Setup the controller based on parameters in the transfer
>> +	 * each transfer can set the bit_per_word and the speed_hz to
>> +	 * change these values in the controller the controller MUST
>> +	 * be disabled
>> +	 */
>> +	if (unlikely(!controller->clk_div)) {
>> +		controller->clk_div = dws->max_freq / controller->speed_hz;
>> +		controller->clk_div = (controller->clk_div + 1)&  0xfffe;
>> +		dw_spi_set_clk(dws, controller->clk_div);
>> +		dev_err(&dws->master->dev, "setting default clk_div");
>> +		err = 1;
>> +	}
>> +
>> +	if (transfer->speed_hz) {
>> +		if (transfer->speed_hz != controller->speed_hz) {
>> +			if (transfer->speed_hz>  dws->max_freq) {
>> +				err = -EIO;
>> +				goto out;
>> +			}
>> +
>> +			clk_div = dws->max_freq / transfer->speed_hz;
>> +			clk_div = (clk_div + 1)&  0xfffe;
>> +			controller->clk_div = clk_div;
>> +			controller->speed_hz = transfer->speed_hz;
>> +			err = 1;
>> +		}
>> +	}
>> +
>> +	if (transfer->bits_per_word) {
>> +		if (transfer->bits_per_word != 8&&
>> +			transfer->bits_per_word != 16) {
>> +			err = -EIO;
>> +			goto out;
>> +		}
>> +		cr0&= ~SPI_DFS_MASK;
>> +		cr0 |= transfer->bits_per_word - 1;
>> +		err = 1;
>> +	} else {
>> +		cr0&= ~SPI_DFS_MASK;
>> +		cr0 |= controller->spi_dev->bits_per_word - 1;
>> +	}
>> +
>> +	cr0&= ~SPI_MODE_MASK;
>> +	cr0 |= (controller->spi_dev->mode<<  SPI_MODE_OFFSET);
>> +	controller->cr0 = cr0;
>> +
>> +	if (err || dw_readw(dws, ctrl0) != cr0) {
>> +		dw_spi_disable(dws);
>> +		dw_spi_chip_sel(dws, controller->spi_dev->chip_select);
>> +		dw_writew(dws, ctrl0, cr0);
>> +		dw_spi_set_clk(dws, controller->clk_div);
>> +		dw_spi_enable(dws);
>> +		err = 0;
>> +	}
>> +out:
>> +	return err;
>> +}
>> +
>> +static void tx_fifo_fill(struct dw_spi *dws)
>> +{
>> +	int room;
>> +	u16 txw = 0;
>> +	if (dws->xfer.sent<  dws->xfer.len) {
>> +		room = tx_max(dws);
>> +		while (room--) {
>> +			if (dws->xfer.tx_buf) {
>> +				if (dws->xfer.n_bytes == 2)
>> +					txw = *(u16 *)dws->xfer.tx_buf;
>> +				else
>> +					txw = *(u8 *)dws->xfer.tx_buf;
>> +				dws->xfer.tx_buf += dws->xfer.n_bytes;
>> +			}
>> +			dw_writew(dws, dr, txw);
>> +			dws->xfer.sent += dws->xfer.n_bytes;
>> +		}
>> +	}
>> +}
>> +
>> +
>> +static void rx_fifo_drain(struct dw_spi *dws)
>> +{
>> +	u16 rx_val;
>> +	int avail;
>> +
>> +	if (dws->xfer.rcvd<  dws->xfer.len) {
>> +		avail = rx_max(dws);
>> +		while (avail--) {
>> +			rx_val = dw_readw(dws, dr);
>> +			if (dws->xfer.rx_buf) {
>> +				if (dws->xfer.n_bytes == 2)
>> +					*(u16 *)(dws->xfer.rx_buf) =
>> +						(u16)rx_val;
>> +				else
>> +					*dws->xfer.rx_buf = (u8)rx_val;
>> +				dws->xfer.rx_buf += dws->xfer.n_bytes;
>> +			}
>> +			dws->xfer.rcvd += dws->xfer.n_bytes;
>> +		}
>> +	}
>> +}
>> +
>> +static inline void do_pio(struct dw_spi *dws)
>> +{
>> +	while (dws->xfer.sent<  dws->xfer.len ||
>> +					dws->xfer.rcvd<  dws->xfer.len) {
>> +		tx_fifo_fill(dws);
>> +		rx_fifo_drain(dws);
>> +		cpu_relax();
>> +	}
>> +	complete(&dws->xfer.complete);
>> +}
>> +
>> +static irqreturn_t dw_spi_irq_thread_handler(int irq, void *dev_id)
>> +{
>> +	struct dw_spi *dws = dev_id;
>> +
>> +	if (dws->xfer.irq_status&  SPI_INT_TXEI) {
>> +		rx_fifo_drain(dws);
>> +		tx_fifo_fill(dws);
>> +	}
>> +
>> +	if (dws->xfer.irq_status&  SPI_INT_RXFI) {
>> +		tx_fifo_fill(dws);
>> +		rx_fifo_drain(dws);
>> +	}
>> +
>> +	if (dws->xfer.len == dws->xfer.rcvd&&
>> +		dws->xfer.len == dws->xfer.sent) {
>> +		complete(&dws->xfer.complete);
>> +		goto out;
>> +	}
>> +
>> +	dw_spi_umask_intr(dws, SPI_INT_ALL);
>> +out:
>> +	return IRQ_HANDLED;
>> +}
>> +
>> +static inline void do_int_xfer(struct dw_spi *dws)
>> +{
>> +	dw_spi_disable(dws);
>> +	dw_writew(dws, txfltr, dws->xfer.tx_threshold);
>> +	dw_writew(dws, rxfltr, dws->xfer.rx_threshold);
>> +	dw_spi_enable(dws);
>> +	dw_readw(dws, icr);
>> +	tx_fifo_fill(dws);
>> +	dw_writew(dws, imr, SPI_INT_ALL);
>> +}
>> +
>> +static inline int do_transfer(struct dw_spi *dws)
>> +{
>> +	switch (dws->xfer.type) {
>> +	case PIO_XFER:
>> +		do_pio(dws);
>> +		break;
>> +	case INT_XFER:
>> +		do_int_xfer(dws);
>> +		break;
>> +	case DMA_XFER:
>> +		dws->dma_ops->dma_transfer(dws);
>> +		break;
>> +	default:
>> +		BUG();
>> +	}
>> +
>> +	wait_for_completion(&dws->xfer.complete);
>> +
>> +
>> +	return dws->xfer.err;
>> +}
>> +
>> +static void drain_message_queue(struct dw_spi *dws)
>> +{
>> +	struct spi_message *message;
>> +
>> +	message = get_message(dws);
>> +	while (message) {
>> +		message->status = -ESHUTDOWN;
>> +		message->complete(message->context);
>> +		message = get_message(dws);
>> +	}
>> +}
>> +
>> +static void pump_messages(struct work_struct *work)
>> +{
>> +	struct dw_spi *dws =
>> +		container_of(work, struct dw_spi, pump_messages);
>> +	struct spi_transfer *transfer;
>> +	struct spi_message *message;
>> +	struct chip_data *controller;
>> +	int err = 0;
>> +
>> +
>> +	message = get_message(dws);
>> +
>> +	while (message&&  dws->run != QUEUE_STOPPED) {
>> +		controller = spi_get_ctldata(message->spi);
>> +		list_for_each_entry(transfer,&message->transfers,
>> +				transfer_list){
>> +
>> +			err = transfer_setup(dws, controller,
>> +						transfer, message);
>> +			if (err<  0) {
>> +				dev_err(&dws->master->dev,
>> +					"transfer_setup failed");
>> +				dws->xfer.err = -EIO;
>> +				break;
>> +			}
>> +
>> +			err = do_transfer(dws);
>> +			if (err<  0) {
>> +				dev_err(&dws->master->dev,
>> +					"do_transfer failed");
>> +				break;
>> +			}
>> +			message->actual_length += dws->xfer.len;
>> +
>> +			if (transfer->delay_usecs)
>> +				udelay(transfer->delay_usecs);
>> +		}
>> +
>> +		message->status = dws->xfer.err;
>> +		message->complete(message->context);
>> +		message =  get_message(dws);
>> +	}
>> +	if (dws->run == QUEUE_STOPPED)
>> +		drain_message_queue(dws);
>> +}
>> +
>> +/* spi_device use this to queue in their spi_msg */
>> +static int dw_spi_transfer(struct spi_device *spi, struct spi_message *msg)
>> +{
>> +	struct dw_spi *dws = spi_master_get_devdata(spi->master);
>> +	unsigned long flags;
>> +
>> +
>> +	spin_lock_irqsave(&dws->lock, flags);
>> +
>> +	msg->actual_length = 0;
>> +	msg->status = -EINPROGRESS;
>> +
>> +	list_add_tail(&msg->queue,&dws->queue);
>> +
>> +	queue_work(dws->workqueue,
>> +		&dws->pump_messages);
>> +
>> +	spin_unlock_irqrestore(&dws->lock, flags);
>> +
>> +	return 0;
>> +}
>> +
>> +/* This may be called twice for each spi dev */
>> +static int dw_spi_setup(struct spi_device *spi)
>> +{
>> +	struct dw_spi_chip *chip_info = NULL;
>> +	struct chip_data *chip;
>> +
>> +	if (spi->bits_per_word != 8&&  spi->bits_per_word != 16)
>> +		return -EINVAL;
>> +
>> +	if (!spi->max_speed_hz) {
>> +		dev_err(&spi->dev, "No max speed HZ parameter\n");
>> +		return -EINVAL;
>> +	}
>> +
>> +	/* Only alloc on first setup */
>> +	chip = spi_get_ctldata(spi);
>> +	if (!chip) {
>> +		chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
>> +		if (!chip)
>> +			return -ENOMEM;
>> +	}
>> +	chip->spi_dev = spi;
>> +
>> +	/*
>> +	 * Protocol drivers may change the chip settings, so...
>> +	 * if chip_info exists, use it
>> +	 */
>> +	chip_info = spi->controller_data;
>> +
>> +	/* chip_info doesn't always exist */
>> +	if (chip_info) {
>> +		chip->type = chip_info->type;
>> +		chip->enable_dma = chip_info->enable_dma;
>> +	}
>> +
>> +	chip->bits_per_word = spi->bits_per_word;
>> +	chip->n_bytes = chip->bits_per_word / 8;
>> +	chip->dma_width = chip->bits_per_word / 8;
>> +
>> +	chip->speed_hz = spi->max_speed_hz;
>> +
>> +	/* Default SPI mode is SCPOL = 0, SCPH = 0 */
>> +	chip->cr0 = (chip->bits_per_word - 1)
>> +		| (chip->type<<  SPI_FRF_OFFSET)
>> +		| (spi->mode<<  SPI_MODE_OFFSET);
>> +
>> +	spi_set_ctldata(spi, chip);
>> +	return 0;
>> +}
>> +
>> +static void dw_spi_cleanup(struct spi_device *spi)
>> +{
>> +	struct chip_data *chip = spi_get_ctldata(spi);
>> +	kfree(chip);
>> +}
>> +
>> +static int __devinit init_queue(struct dw_spi *dws)
>> +{
>> +	INIT_LIST_HEAD(&dws->queue);
>> +	spin_lock_init(&dws->lock);
>> +
>> +	dws->run = QUEUE_STOPPED;
>> +
>> +	INIT_WORK(&dws->pump_messages, pump_messages);
>> +	dws->workqueue = create_singlethread_workqueue(
>> +					dev_name(dws->master->dev.parent));
>> +	if (dws->workqueue == NULL)
>> +		return -EBUSY;
>> +	dws->run = QUEUE_RUNNING;
>> +	return 0;
>> +}
>> +
>> +
>> +int dw_spi_stop_queue(struct dw_spi *dws)
>> +{
>> +	unsigned long flags;
>> +	int status = 0;
>> +
>> +	spin_lock_irqsave(&dws->lock, flags);
>> +	dws->run = QUEUE_STOPPED;
>> +
>> +	if (!list_empty(&dws->queue))
>> +		status = -EBUSY;
>> +	spin_unlock_irqrestore(&dws->lock, flags);
>> +
>> +	return status;
>> +}
>> +EXPORT_SYMBOL_GPL(dw_spi_stop_queue);
>> +
>> +static int destroy_queue(struct dw_spi *dws)
>> +{
>> +	int status;
>> +
>> +	status = dw_spi_stop_queue(dws);
>> +	if (status != 0)
>> +		return status;
>> +	destroy_workqueue(dws->workqueue);
>> +	return 0;
>> +}
>> +
>> +/* Restart the controller, disable all interrupts, clean rx fifo */
>> +static void spi_hw_init(struct dw_spi *dws)
>> +{
>> +	dw_spi_disable(dws);
>> +	dw_spi_mask_intr(dws, 0xff);
>> +	dw_readw(dws, icr);
>> +	dw_spi_enable(dws);
>> +
>> +	BUG_ON(!dws->fifo_len);
>> +}
>> +
>> +int __devinit dw_spi_add_host(struct dw_spi *dws)
>> +{
>> +	struct spi_master *master;
>> +	int ret;
>> +
>> +	BUG_ON(dws == NULL);
>> +
>> +	master = spi_alloc_master(dws->parent_dev, 0);
>> +	if (!master) {
>> +		ret = -ENOMEM;
>> +		goto exit;
>> +	}
>> +
>> +	dws->master = master;
>> +	dws->type = SSI_MOTO_SPI;
>> +
>> +	init_completion(&dws->xfer.complete);
>> +	dws->dma_inited = 0;
>> +	/* Change to address of FIFO */
>> +	dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
>> +
>> +	ret = request_threaded_irq(dws->irq, dw_spi_irq,
>> +				dw_spi_irq_thread_handler,
>> +				IRQF_SHARED, "dw_spi", dws);
>> +	if (ret<  0) {
>> +		dev_err(&master->dev, "can not get IRQ\n");
>> +		goto err_free_master;
>> +	}
>> +
>> +	master->mode_bits = SPI_CPOL | SPI_CPHA;
>> +	master->bus_num = dws->bus_num;
>> +	master->num_chipselect = dws->num_cs;
>> +	master->cleanup = dw_spi_cleanup;
>> +	master->setup = dw_spi_setup;
>> +	master->transfer = dw_spi_transfer;
>> +
>> +	/* Basic HW init */
>> +	spi_hw_init(dws);
>> +
>> +	if (dws->dma_ops&&  dws->dma_ops->dma_init) {
>> +		ret = dws->dma_ops->dma_init(dws);
>> +		if (ret) {
>> +			dev_warn(&master->dev, "DMA init failed\n");
>> +			dws->dma_inited = 0;
>> +		}
>> +	}
>> +
>> +	/* Initial and start queue */
>> +	ret = init_queue(dws);
>> +	if (ret) {
>> +		dev_err(&master->dev, "problem initializing queue\n");
>> +		goto err_diable_hw;
>> +	}
>> +
>> +	spi_master_set_devdata(master, dws);
>> +	ret = spi_register_master(master);
>> +	if (ret) {
>> +		dev_err(&master->dev, "problem registering spi master\n");
>> +		goto err_queue_alloc;
>> +	}
>> +
>> +	mrst_spi_debugfs_init(dws);
>> +	return 0;
>> +
>> +err_queue_alloc:
>> +	destroy_queue(dws);
>> +	if (dws->dma_ops&&  dws->dma_ops->dma_exit)
>> +		dws->dma_ops->dma_exit(dws);
>> +err_diable_hw:
>> +	dw_spi_disable(dws);
>> +	free_irq(dws->irq, dws);
>> +err_free_master:
>> +	spi_master_put(master);
>> +exit:
>> +	return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(dw_spi_add_host);
>> +
>> +void __devexit dw_spi_remove_host(struct dw_spi *dws)
>> +{
>> +	int status = 0;
>> +
>> +	if (!dws)
>> +		return;
>> +	mrst_spi_debugfs_remove(dws);
>> +
>> +	/* Remove the queue */
>> +	status = destroy_queue(dws);
>> +	if (status != 0)
>> +		dev_err(&dws->master->dev, "dw_spi_remove: workqueue will not "
>> +			"complete, message memory not freed\n");
>> +
>> +	if (dws->dma_ops&&  dws->dma_ops->dma_exit)
>> +		dws->dma_ops->dma_exit(dws);
>> +	dw_spi_disable(dws);
>> +	dw_readw(dws, icr);
>> +	free_irq(dws->irq, dws);
>> +
>> +	/* Disconnect from the SPI framework */
>> +	spi_unregister_master(dws->master);
>> +}
>> +EXPORT_SYMBOL_GPL(dw_spi_remove_host);
>> +
>> +int dw_spi_suspend_host(struct dw_spi *dws)
>> +{
>> +	int ret = 0;
>> +
>> +	ret = dw_spi_stop_queue(dws);
>> +	if (ret)
>> +		return ret;
>> +	dw_spi_disable(dws);
>> +	return ret;
>> +}
>> +EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
>> +
>> +int dw_spi_resume_host(struct dw_spi *dws)
>> +{
>> +	spi_hw_init(dws);
>> +	dws->run = QUEUE_RUNNING;
>> +	return 0;
>> +}
>> +EXPORT_SYMBOL_GPL(dw_spi_resume_host);
>> +
>> +MODULE_AUTHOR("Feng Tang<feng.tang@intel.com>");
>> +MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
>> +MODULE_LICENSE("GPL v2");
>> diff --git a/drivers/spi/spi-dw.h b/drivers/spi/spi-dw.h
>> index 97baff6..b016b85 100644
>> --- a/drivers/spi/spi-dw.h
>> +++ b/drivers/spi/spi-dw.h
>> @@ -7,6 +7,7 @@
>>
>>   /* Bit fields in CTRLR0 */
>>   #define SPI_DFS_OFFSET			0
>> +#define SPI_DFS_MASK			0xf
>>
>>   #define SPI_FRF_OFFSET			4
>>   #define SPI_FRF_SPI			0x0
>> @@ -17,6 +18,7 @@
>>   #define SPI_MODE_OFFSET			6
>>   #define SPI_SCPH_OFFSET			6
>>   #define SPI_SCOL_OFFSET			7
>> +#define SPI_MODE_MASK			(0x3<<  SPI_MODE_OFFSET)
>>
>>   #define SPI_TMOD_OFFSET			8
>>   #define SPI_TMOD_MASK			(0x3<<  SPI_TMOD_OFFSET)
>> @@ -46,6 +48,7 @@
>>   #define SPI_INT_RXOI			(1<<  3)
>>   #define SPI_INT_RXFI			(1<<  4)
>>   #define SPI_INT_MSTI			(1<<  5)
>> +#define SPI_INT_ALL  0x3f
>>
>>   /* TX RX interrupt level threshold, max can be 256 */
>>   #define SPI_INT_THRESHOLD		32
>> @@ -83,65 +86,67 @@ struct dw_spi;
>>   struct dw_spi_dma_ops {
>>   	int (*dma_init)(struct dw_spi *dws);
>>   	void (*dma_exit)(struct dw_spi *dws);
>> -	int (*dma_transfer)(struct dw_spi *dws, int cs_change);
>> +	int (*dma_transfer)(struct dw_spi *dws);
>> +};
>> +
>> +enum xfer_type {
>> +	PIO_XFER,
>> +	INT_XFER,
>> +	DMA_XFER,
>> +};
>> +
>> +	
>> +struct xfer_state {
>> +	const u8 *tx_buf;
>> +	dma_addr_t tx_dma;
>> +	u8 *rx_buf;
>> +	dma_addr_t rx_dma;
>> +	struct spi_message *msg;
>> +	u32 n_bytes;
>> +	u32 len;
>> +	u32 sent;
>> +	u32 rcvd;
>> +	u32 err;
>> +	u32 type;
>> +	u32  tx_threshold;
>> +	u32  rx_threshold;
>> +	u32 irq_status;
>> +	struct completion complete;
>>   };
>>
>>   struct dw_spi {
>>   	struct spi_master	*master;
>> -	struct spi_device	*cur_dev;
>>   	struct device		*parent_dev;
>>   	enum dw_ssi_type	type;
>>
>>   	void __iomem		*regs;
>>   	unsigned long		paddr;
>>   	u32			iolen;
>> -	int			irq;
>> +	u32			irq;
>>   	u32			fifo_len;	/* depth of the FIFO buffer */
>>   	u32			max_freq;	/* max bus freq supported */
>>
>>   	u16			bus_num;
>>   	u16			num_cs;		/* supported slave numbers */
>> -
>> +
>>   	/* Driver message queue */
>>   	struct workqueue_struct	*workqueue;
>>   	struct work_struct	pump_messages;
>>   	spinlock_t		lock;
>>   	struct list_head	queue;
>> -	int			busy;
>> -	int			run;
>> -
>> -	/* Message Transfer pump */
>> -	struct tasklet_struct	pump_transfers;
>> +	u32			run;
>>
>>   	/* Current message transfer state info */
>> -	struct spi_message	*cur_msg;
>> -	struct spi_transfer	*cur_transfer;
>> -	struct chip_data	*cur_chip;
>> -	struct chip_data	*prev_chip;
>> -	size_t			len;
>> -	void			*tx;
>> -	void			*tx_end;
>> -	void			*rx;
>> -	void			*rx_end;
>> -	int			dma_mapped;
>> -	dma_addr_t		rx_dma;
>> -	dma_addr_t		tx_dma;
>> -	size_t			rx_map_len;
>> -	size_t			tx_map_len;
>> -	u8			n_bytes;	/* current is a 1/2 bytes op */
>> -	u8			max_bits_per_word;	/* maxim is 16b */
>> -	u32			dma_width;
>> -	int			cs_change;
>> -	irqreturn_t		(*transfer_handler)(struct dw_spi *dws);
>> -	void			(*cs_control)(u32 command);
>> +	struct xfer_state       xfer;
>>
>>   	/* Dma info */
>> -	int			dma_inited;
>> +//	int 			dma_width;
>> +	u32			dma_inited;
>>   	struct dma_chan		*txchan;
>>   	struct scatterlist	tx_sgl;
>>   	struct dma_chan		*rxchan;
>>   	struct scatterlist	rx_sgl;
>> -	int			dma_chan_done;
>> +	u32			dma_chan_done;
>>   	struct device		*dma_dev;
>>   	dma_addr_t		dma_addr; /* phy address of the Data register */
>>   	struct dw_spi_dma_ops	*dma_ops;
>> @@ -163,30 +168,36 @@ struct dw_spi {
>>   	__raw_readw(&(((struct dw_spi_reg *)dw->regs)->name))
>>   #define dw_writew(dw, name, val) \
>>   	__raw_writew((val),&(((struct dw_spi_reg *)dw->regs)->name))
>> +#define dw_readb(dw, name) \
>> +	__raw_readb(&(((struct dw_spi_reg *)dw->regs)->name))
>> +#define dw_writeb(dw, name, val) \
>> +	__raw_writeb((val),&(((struct dw_spi_reg *)dw->regs)->name))
>>
>> -static inline void spi_enable_chip(struct dw_spi *dws, int enable)
>> +static inline void dw_spi_disable(struct dw_spi *dws)
>>   {
>> -	dw_writel(dws, ssienr, (enable ? 1 : 0));
>> +	dw_writel(dws, ssienr, 0);
>>   }
>>
>> -static inline void spi_set_clk(struct dw_spi *dws, u16 div)
>> +static inline void dw_spi_enable(struct dw_spi *dws)
>> +{
>> +	dw_writel(dws, ssienr, 1);
>> +}
>> +
>> +static inline void dw_spi_set_clk(struct dw_spi *dws, u16 div)
>>   {
>>   	dw_writel(dws, baudr, div);
>>   }
>>
>> -static inline void spi_chip_sel(struct dw_spi *dws, u16 cs)
>> +static inline void dw_spi_chip_sel(struct dw_spi *dws, u16 cs)
>>   {
>>   	if (cs>  dws->num_cs)
>>   		return;
>>
>> -	if (dws->cs_control)
>> -		dws->cs_control(1);
>> -
>>   	dw_writel(dws, ser, 1<<  cs);
>>   }
>>
>>   /* Disable IRQ bits */
>> -static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
>> +static inline void dw_spi_mask_intr(struct dw_spi *dws, u32 mask)
>>   {
>>   	u32 new_mask;
>>
>> @@ -195,7 +206,7 @@ static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
>>   }
>>
>>   /* Enable IRQ bits */
>> -static inline void spi_umask_intr(struct dw_spi *dws, u32 mask)
>> +static inline void dw_spi_umask_intr(struct dw_spi *dws, u32 mask)
>>   {
>>   	u32 new_mask;
>>
>> @@ -208,6 +219,7 @@ extern void dw_spi_remove_host(struct dw_spi *dws);
>>   extern int dw_spi_suspend_host(struct dw_spi *dws);
>>   extern int dw_spi_resume_host(struct dw_spi *dws);
>>   extern void dw_spi_xfer_done(struct dw_spi *dws);
>> +extern int dw_spi_stop_queue(struct dw_spi *dws);
>>
>>   /* platform related setup */
>>   extern int dw_spi_mid_init(struct dw_spi *dws); /* Intel MID platforms */
>> --
>> 1.7.3.4
>>

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
Grant Likely June 16, 2011, 4:40 p.m. UTC | #4
On Thu, Jun 16, 2011 at 09:03:49AM -0700, Dirk Brandewie wrote:
> On 06/16/2011 06:14 AM, Grant Likely wrote:
> >On Wed, Jun 15, 2011 at 10:23:06AM -0700, dirk.brandewie@gmail.com wrote:
> >>From: Dirk Brandewie<dirk.brandewie@gmail.com>
> >>
> >>NOTE: patch created git format-patch --break-rewrites=/50%
> >>
> >>This patch reworks the message pump worker thread function to run
> >>until all messages queued to the driver have been handled. The
> >>function to handle individual spi_transfers is now a synchronus
> >>function the tasklet to handle spi_transfers has been removed. Work
> >>for the worker thread is only queued in host controller transfer
> >>function.
> >>
> >>Psuedo code for new thread function:
> >>   message = get_message()
> >>   while (message){
> >>     for_each_transfer_in_msg(message){
> >>       transfer_setup(transfer)
> >>       do_transfer()
> >>     }
> >>     complete_message()
> >>     message = get_message()
> >>   }
> >>
> >>Changes that fell out of the message thread changes:
> >>Non-DMA transfers that are larger than the size of the controller FIFO
> >>are handled as interrupt driven transfers.
> >>
> >>Common FIFO handling functions shared PIO and interrupt transfers.
> >>
> >>Simplified queue stop/start funcitons.
> >>
> >>Cleanup fixes:
> >>Changed exported all exported function names to have dw_spi_ prefix
> >>
> >>Removed support for registering chip select control function. Setting
> >>the slave chip select is handled by setting the SER (Slave enable
> >>register)
> >
> >What about for implementations that use a GPIO for the SPI chip
> >select?  It is very common for board designs to use GPIOs for
> >multiplexing the SPI bus.
> 
> OK I can put that back. I took it out for the following reason.
> 
> The Slave Enable Register (SER) must be set for the IP block to
> start the transfer.  The bit in SER directly control a separate chip
> select pin while there is an active transfer.  This gives four chip
> selects in the direct mapping case and up to fifteen if the slave
> chip selects are muxed together.
> 
> >
> >>
> >>Removed code that looked at the cs_change hint in the
> >>spi_transfer. Software has no contorl over whether the slave chip
> >>select is de-asserted at the end of the transfer.  Once the TX FIFO
> >>goes empty the slave chip select is dropped.
> >
> >This sounds wrong.  cs_change is *not* merely a hint.  It must be respected.
> >If the driver has no direct control over the CS line, then it is
> >incumbent on the driver to guaranteed that the cs deassert condition
> >does not occur.  This will probably mean chaining up all the transfers
> >in a message so that the TX FIFO remains full.  If cs_change is
> >requested, then the FIFO must be allowed to empty before kicking of
> >the next group of transfers.
> >
> 
> I agree it is wrong that the driver can not directly control the chip select :-)
> 
> I could probably manage to get the PIO and interrupt case to honor
> this but would add a fair bit of complexity, I have no idea how to
> make the DMA case work I am open to suggestions :-).  The current
> driver has this issue I just made it explicit maybe we should add a
> warning message in the driver to let client driver writers aware of
> the Silicon behaviour.

Use scatter/gather DMA or a bounce buffer to chain up contiguous
transfers into a single DMA operation.  "Best effort" is not okay in
this situation, the behaviour is required even if it does add
complexity.

g.

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
dirk.brandewie@gmail.com June 16, 2011, 5:28 p.m. UTC | #5
On 06/16/2011 07:00 AM, Feng Tang wrote:
> Hi Dirk,
>
> IMHO, the patch is too big, it contains too many changes to the original
> drivers, and we can't see clearly what you've changed to each logical
> code part or section, If possible, could you separate this patch to
> several small ones in a logical way.
>
> First, I have some questions, what devices have you tested with this patch?
> high speed, low speed? Do you have any performance data to show the benefit
> of this change? Current dw_spi driver has been tested with many devices, so
> to not break them or cause obvious regression, we have to be cautious.

See Thread with grant for list of environments where it has been tested.
The boot time of the platforms it is being used on decreased 2-5 seconds with no 
regressions reported.  It has been in use/under test for ~3 months in various 
Meego and Android builds.  It clears all the bugs reported against the driver 
that I am aware of.  If you can give me pointers to teams/projects that are 
using v2.6.37+ kernels I would be more that happy to provide them with patches 
for their kernel to ensure we get the most comprehensive test possible.

>
> Here are some general comments according to the commit logs:
> 1. I think the threaded irq handling is a good idea. And let driver chose to
>     use poll or interrupt is good, some other spi controller driver has used
>     that way for a long time
> 2. Why you remove the cs_change code, in some case, the controller is only
>     be used by one device, we don't need do the config for every single
>     spi_transfer

There is no guarantee that all the transfers in a given spi_message have the 
same values for  speed_hz, bits_per_word, cs_change and delay_usecs atleast 
nothing I could find put that restriction in place. Since we need to deal with 
possible changes (although unlikely) it gets rid some state we need to maintain 
and makes the code path common for all transfers.

> 3. Why do you remove the chip select control code, dw_spi controller hw has
>     some problem in chip select controller by SER, and thus many devices has
>     to use external GPIO has their chip select, this is real world usages!

Which devices/platforms are you referring to?  I was unable to find any 
platforms or client drivers using this functionality. If they are not public 
please respond internally.  In any case it is mute since I already agreed to put 
is back in in my response to Grant.

> 4. I saw you enable both TX/RX interrupt when doing interrupt transfer, spi
>     devices' TX/RX are born to be simutaneous, when one word is sent over
>     TX line, a RX word will be received from RX line, so both the orignal
>     interrupt transfer handling written by me and the later optimization
>     from Alek Du only enable TX interrupt, which will only generate half of
>     IRQs comparing to enble both TX/RX, this is huge when the data rate is
>     several Mb per second

I the current driver the txfltr register is set to 0 (FIFO empty) in the 
interrupt transfer case which will drop chip select every FIFO length bytes.

In my transfer setup the RX FIFO interrupt is set to a value lower than the TX 
threshold which will keep both interrupts from firing at the same time.

The TX interrupt will drive the transfer until there are less than tx_threshold 
bytes left to transfer then by the RX interrupt to drive the remainder of the 
transfer without dropping chip select.

> 5. Why do you change the logic of filling TX FIFO, the logic comes from use
>     case that the dw_spi driver is dealing with several high speed devices.
>

The version of the driver that I started with did not have the current fifo 
handling code.  I finished my changes before they showed up in the AC tree or 
the upstream kernel.  I picked up most to the fifo handling logic from the 
current driver with the exception of the rxtx_gap calculation which was not 
needed with the way I am maintaining the state and the addition of a check to 
avoid the register read in tx_max/rx_max if all bytes have been sent/received.
This avoids some overhead in the interrupt case.

> Thanks,
> Feng
>

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
Grant Likely June 16, 2011, 5:38 p.m. UTC | #6
On Thu, Jun 16, 2011 at 11:28 AM, Dirk Brandewie
<dirk.brandewie@gmail.com> wrote:
> On 06/16/2011 07:00 AM, Feng Tang wrote:
>>
>> Hi Dirk,
>>
>> IMHO, the patch is too big, it contains too many changes to the original
>> drivers, and we can't see clearly what you've changed to each logical
>> code part or section, If possible, could you separate this patch to
>> several small ones in a logical way.
>>
>> First, I have some questions, what devices have you tested with this
>> patch?
>> high speed, low speed? Do you have any performance data to show the
>> benefit
>> of this change? Current dw_spi driver has been tested with many devices,
>> so
>> to not break them or cause obvious regression, we have to be cautious.
>
> See Thread with grant for list of environments where it has been tested.
> The boot time of the platforms it is being used on decreased 2-5 seconds
> with no regressions reported.  It has been in use/under test for ~3 months
> in various Meego and Android builds.  It clears all the bugs reported
> against the driver that I am aware of.  If you can give me pointers to
> teams/projects that are using v2.6.37+ kernels I would be more that happy to
> provide them with patches for their kernel to ensure we get the most
> comprehensive test possible.
>
>>
>> Here are some general comments according to the commit logs:
>> 1. I think the threaded irq handling is a good idea. And let driver chose
>> to
>>    use poll or interrupt is good, some other spi controller driver has
>> used
>>    that way for a long time
>> 2. Why you remove the cs_change code, in some case, the controller is only
>>    be used by one device, we don't need do the config for every single
>>    spi_transfer
>
> There is no guarantee that all the transfers in a given spi_message have the
> same values for  speed_hz, bits_per_word, cs_change and delay_usecs atleast
> nothing I could find put that restriction in place. Since we need to deal
> with possible changes (although unlikely) it gets rid some state we need to
> maintain and makes the code path common for all transfers.
>
>> 3. Why do you remove the chip select control code, dw_spi controller hw
>> has
>>    some problem in chip select controller by SER, and thus many devices
>> has
>>    to use external GPIO has their chip select, this is real world usages!
>
> Which devices/platforms are you referring to?  I was unable to find any
> platforms or client drivers using this functionality. If they are not public
> please respond internally.  In any case it is mute since I already agreed to
> put is back in in my response to Grant.
>
>> 4. I saw you enable both TX/RX interrupt when doing interrupt transfer,
>> spi
>>    devices' TX/RX are born to be simutaneous, when one word is sent over
>>    TX line, a RX word will be received from RX line, so both the orignal
>>    interrupt transfer handling written by me and the later optimization
>>    from Alek Du only enable TX interrupt, which will only generate half of
>>    IRQs comparing to enble both TX/RX, this is huge when the data rate is
>>    several Mb per second
>
> I the current driver the txfltr register is set to 0 (FIFO empty) in the
> interrupt transfer case which will drop chip select every FIFO length bytes.
>
> In my transfer setup the RX FIFO interrupt is set to a value lower than the
> TX threshold which will keep both interrupts from firing at the same time.
>
> The TX interrupt will drive the transfer until there are less than
> tx_threshold bytes left to transfer then by the RX interrupt to drive the
> remainder of the transfer without dropping chip select.

Be careful here.  Can you guarantee that the kernel will process the
IRQ before the FIFO drains?  If not, then you'll need something more
reliable.

g.

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
dirk.brandewie@gmail.com June 16, 2011, 7:52 p.m. UTC | #7
On 06/16/2011 10:38 AM, Grant Likely wrote:
> On Thu, Jun 16, 2011 at 11:28 AM, Dirk Brandewie
> <dirk.brandewie@gmail.com>  wrote:
>> On 06/16/2011 07:00 AM, Feng Tang wrote:
>>>
>>> 4. I saw you enable both TX/RX interrupt when doing interrupt transfer,
>>> spi
>>>     devices' TX/RX are born to be simutaneous, when one word is sent over
>>>     TX line, a RX word will be received from RX line, so both the orignal
>>>     interrupt transfer handling written by me and the later optimization
>>>     from Alek Du only enable TX interrupt, which will only generate half of
>>>     IRQs comparing to enble both TX/RX, this is huge when the data rate is
>>>     several Mb per second
>>
>> I the current driver the txfltr register is set to 0 (FIFO empty) in the
>> interrupt transfer case which will drop chip select every FIFO length bytes.
>>
>> In my transfer setup the RX FIFO interrupt is set to a value lower than the
>> TX threshold which will keep both interrupts from firing at the same time.
>>
>> The TX interrupt will drive the transfer until there are less than
>> tx_threshold bytes left to transfer then by the RX interrupt to drive the
>> remainder of the transfer without dropping chip select.
>
> Be careful here.  Can you guarantee that the kernel will process the
> IRQ before the FIFO drains?  If not, then you'll need something more
> reliable.

I can't guarantee when IRQ are processed :-)

Without actual control over when chip is de-asserted I don't think we guarantee 
that chip select will be not dropped, chip select is dropped when the TX fifo 
goes empty.

With a sufficiently high bitrate client, heavily loaded system and fixed size 
fifo we are domed at some point.  As tx_threshold approaches fifo depth you give 
yourself a bigger buffer in time but increase the number of interrupts you need 
to deal with increasing the system load.

For high bitrate devices the client driver should be using DMA IMHO to work 
around this limitation in the designware IP.  Of course this doesn't solve the 
problem of chip select dropping between transfers in the same message.

I took a SWAG at tx_threshold in my changes to try to limit the number of 
interrupts it will need tuning for a reasonable default and likely the addition 
of a module parameter or platform data to set it for any given SOC/system.

--Dirk





------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
Feng Tang June 17, 2011, 1:22 a.m. UTC | #8
On Fri, 17 Jun 2011 01:28:16 +0800
Dirk Brandewie <dirk.brandewie@gmail.com> wrote:

> On 06/16/2011 07:00 AM, Feng Tang wrote:
> > Hi Dirk,
> >
> > IMHO, the patch is too big, it contains too many changes to the
> > original drivers, and we can't see clearly what you've changed to
> > each logical code part or section, If possible, could you separate
> > this patch to several small ones in a logical way.
> >
> > First, I have some questions, what devices have you tested with
> > this patch? high speed, low speed? Do you have any performance data
> > to show the benefit of this change? Current dw_spi driver has been
> > tested with many devices, so to not break them or cause obvious
> > regression, we have to be cautious.
> 
> See Thread with grant for list of environments where it has been
> tested. The boot time of the platforms it is being used on decreased
> 2-5 seconds with no regressions reported.  It has been in use/under
> test for ~3 months in various Meego and Android builds.  It clears
> all the bugs reported against the driver that I am aware of.  If you
> can give me pointers to teams/projects that are using v2.6.37+
> kernels I would be more that happy to provide them with patches for
> their kernel to ensure we get the most comprehensive test possible.

I was not asking about the environments, but the actual devices connect.
to and work with the dw_spi controller, here is a quote from one of
my previous email of devices listing:

----------------------------------------------------------------------------
I don't know all the devices and users, but here is what I know: I've tested
Max3110 spi-uart (in-tree), Option GTM501L high-speed 3G modem (out of tree),
ektf1236 spi touch screen (out of tree). Alek Du (Cced) should have tested
current dw_spi driver with some spi bluetooth device and modem device. Also
the original author fordw_spi_mmio.c Jean-Hugues Deschense should have some
experience too. 
-----------------------------------------------------------------------------
Meego or Android doesn't mean much for a device driver, as all these OSes are
using the same Linux kernel.

Is "2-5 seconds boot time cuts" the only performance data you got? The current
driver has 2 phases, first is the original one from me, the second is the 
optimization by Alek Du which introduce the batch operation for TX/RX FIFO, 
Alek mentioned the boot time cut too, so I guess the boot time gain is tested
against the original driver, not the current mainstream driver, right?

For the kernel version you mentioned, my thought is the kernel version diff
between 2.6.35 and 2.6.37 doesn't mean much to a simple dw_spi device driver's
performance? some users of dw_spi driver are still using 2.6.35 kernel with
all new optimizations back ported, I have no power to force them upgrade the
kernel, but if the dw_spi driver is the same, then the performance data is
mostly trustable.

> 
> >
> > Here are some general comments according to the commit logs:
> > 1. I think the threaded irq handling is a good idea. And let driver
> > chose to use poll or interrupt is good, some other spi controller
> > driver has used that way for a long time
> > 2. Why you remove the cs_change code, in some case, the controller
> > is only be used by one device, we don't need do the config for
> > every single spi_transfer
> 
> There is no guarantee that all the transfers in a given spi_message
> have the same values for  speed_hz, bits_per_word, cs_change and
> delay_usecs atleast nothing I could find put that restriction in
> place. Since we need to deal with possible changes (although
> unlikely) it gets rid some state we need to maintain and makes the
> code path common for all transfers.

right, if you look at the driver, cs_change is only judge condition
for judge, speed_hz, bits_per_word's change are counted in too. Otherwise
the driver should fail years ago.

> 
> > 3. Why do you remove the chip select control code, dw_spi
> > controller hw has some problem in chip select controller by SER,
> > and thus many devices has to use external GPIO has their chip
> > select, this is real world usages!
> 
> Which devices/platforms are you referring to?  I was unable to find
> any platforms or client drivers using this functionality. If they are
> not public please respond internally.  In any case it is mute since I
> already agreed to put is back in in my response to Grant.

Some intel platforms used/are using that, I got requests to help them
on it. Also some community guys are using that (though I don't know the
detail),  as this piece of code is introduced by Shore George once made
some change according to their specific change, see commit 052dc7c4

> 
> > 4. I saw you enable both TX/RX interrupt when doing interrupt
> > transfer, spi devices' TX/RX are born to be simutaneous, when one
> > word is sent over TX line, a RX word will be received from RX line,
> > so both the orignal interrupt transfer handling written by me and
> > the later optimization from Alek Du only enable TX interrupt, which
> > will only generate half of IRQs comparing to enble both TX/RX, this
> > is huge when the data rate is several Mb per second
> 
> I the current driver the txfltr register is set to 0 (FIFO empty) in
> the interrupt transfer case which will drop chip select every FIFO
> length bytes.
> 
> In my transfer setup the RX FIFO interrupt is set to a value lower
> than the TX threshold which will keep both interrupts from firing at
> the same time.
> 
> The TX interrupt will drive the transfer until there are less than
> tx_threshold bytes left to transfer then by the RX interrupt to drive
> the remainder of the transfer without dropping chip select.

What I set is half-empty for TX interrupt, so the FIFO will get filled
long before it's drained. I don't have any HW protocol analyzer or
logical analyzer, so I can't totally guarantee it won't happen

> 
> > 5. Why do you change the logic of filling TX FIFO, the logic comes
> > from use case that the dw_spi driver is dealing with several high
> > speed devices.
> >
> 
> The version of the driver that I started with did not have the
> current fifo handling code.  I finished my changes before they showed
> up in the AC tree or the upstream kernel.  I picked up most to the
> fifo handling logic from the current driver with the exception of the
> rxtx_gap calculation which was not needed with the way I am
> maintaining the state and the addition of a check to avoid the
> register read in tx_max/rx_max if all bytes have been sent/received.
> This avoids some overhead in the interrupt case.

If there is no performance again, I would hold on any change.

> 
> > Thanks,
> > Feng
> >

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
Du, Alek June 17, 2011, 1:34 a.m. UTC | #9
On Thu, 16 Jun 2011 22:00:03 +0800
"Tang, Feng" <feng.tang@intel.com> wrote:

> Hi Dirk,
> 
> On Thu, 16 Jun 2011 01:23:06 +0800
> "dirk.brandewie@gmail.com" <dirk.brandewie@gmail.com> wrote:
> 
> > From: Dirk Brandewie <dirk.brandewie@gmail.com>
> >
> > NOTE: patch created git format-patch --break-rewrites=/50%
> >
> > This patch reworks the message pump worker thread function to run
> > until all messages queued to the driver have been handled. The
> > function to handle individual spi_transfers is now a synchronus
> > function the tasklet to handle spi_transfers has been removed. Work
> > for the worker thread is only queued in host controller transfer
> > function.
> >
> > Psuedo code for new thread function:
> >   message = get_message()
> >   while (message){
> >     for_each_transfer_in_msg(message){
> >       transfer_setup(transfer)
> >       do_transfer()
> >     }
> >     complete_message()
> >     message = get_message()
> >   }
> >
> > Changes that fell out of the message thread changes:
> > Non-DMA transfers that are larger than the size of the controller FIFO
> > are handled as interrupt driven transfers.
> >
> > Common FIFO handling functions shared PIO and interrupt transfers.
> >
> > Simplified queue stop/start funcitons.
> >
> > Cleanup fixes:
> > Changed exported all exported function names to have dw_spi_ prefix
> >
> > Removed support for registering chip select control function. Setting
> > the slave chip select is handled by setting the SER (Slave enable
> > register)
> >

Removing this, you already break the TI spi BT module which requires that it
controls the chip select instead of controller.

When TX FIFO empty (means TX draining fast) the CS pin will de-assert.
In order to support those spi slaves that cannot stand this, the chip
select control function callback was added to this driver. Please don't
break previous contribution please.

And again, it is meaningless to enable both TX and RX fifo threshold interrupts,
since they are connected in hardware.

Thanks,
Alek

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
diff mbox

Patch

diff --git a/drivers/spi/spi-dw-mid.c b/drivers/spi/spi-dw-mid.c
index 130e555..e44e37f 100644
--- a/drivers/spi/spi-dw-mid.c
+++ b/drivers/spi/spi-dw-mid.c
@@ -38,7 +38,10 @@  static bool mid_spi_dma_chan_filter(struct dma_chan *chan, void *param)
 {
 	struct dw_spi *dws = param;
 
-	return dws->dmac && (&dws->dmac->dev == chan->device->dev);
+	if (dws->dmac && &dws->dmac->dev == chan->device->dev)
+		return true;
+	else
+		return false;
 }
 
 static int mid_spi_dma_init(struct dw_spi *dws)
@@ -103,10 +106,10 @@  static void dw_spi_dma_done(void *arg)
 
 	if (++dws->dma_chan_done != 2)
 		return;
-	dw_spi_xfer_done(dws);
+	complete(&dws->xfer.complete);
 }
 
-static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
+static int mid_spi_dma_transfer(struct dw_spi *dws)
 {
 	struct dma_async_tx_descriptor *txdesc = NULL, *rxdesc = NULL;
 	struct dma_chan *txchan, *rxchan;
@@ -114,17 +117,17 @@  static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
 	u16 dma_ctrl = 0;
 
 	/* 1. setup DMA related registers */
-	if (cs_change) {
-		spi_enable_chip(dws, 0);
-		dw_writew(dws, dmardlr, 0xf);
-		dw_writew(dws, dmatdlr, 0x10);
-		if (dws->tx_dma)
-			dma_ctrl |= 0x2;
-		if (dws->rx_dma)
-			dma_ctrl |= 0x1;
-		dw_writew(dws, dmacr, dma_ctrl);
-		spi_enable_chip(dws, 1);
-	}
+
+	dw_spi_disable(dws);
+	dw_writew(dws, dmardlr, 0xf);
+	dw_writew(dws, dmatdlr, 0x10);
+	if (dws->xfer.tx_dma)
+		dma_ctrl |= 0x2;
+	if (dws->xfer.rx_dma)
+		dma_ctrl |= 0x1;
+	dw_writew(dws, dmacr, dma_ctrl);
+	dw_spi_enable(dws);
+
 
 	dws->dma_chan_done = 0;
 	txchan = dws->txchan;
@@ -141,8 +144,8 @@  static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
 				       (unsigned long) &txconf);
 
 	memset(&dws->tx_sgl, 0, sizeof(dws->tx_sgl));
-	dws->tx_sgl.dma_address = dws->tx_dma;
-	dws->tx_sgl.length = dws->len;
+	dws->tx_sgl.dma_address = dws->xfer.tx_dma;
+	dws->tx_sgl.length = dws->xfer.len;
 
 	txdesc = txchan->device->device_prep_slave_sg(txchan,
 				&dws->tx_sgl,
@@ -163,8 +166,8 @@  static int mid_spi_dma_transfer(struct dw_spi *dws, int cs_change)
 				       (unsigned long) &rxconf);
 
 	memset(&dws->rx_sgl, 0, sizeof(dws->rx_sgl));
-	dws->rx_sgl.dma_address = dws->rx_dma;
-	dws->rx_sgl.length = dws->len;
+	dws->rx_sgl.dma_address = dws->xfer.rx_dma;
+	dws->rx_sgl.length = dws->xfer.len;
 
 	rxdesc = rxchan->device->device_prep_slave_sg(rxchan,
 				&dws->rx_sgl,
@@ -188,7 +191,6 @@  static struct dw_spi_dma_ops mid_dma_ops = {
 #endif
 
 /* Some specific info for SPI0 controller on Moorestown */
-
 /* HW info for MRST CLk Control Unit, one 32b reg */
 #define MRST_SPI_CLK_BASE	100000000	/* 100m */
 #define MRST_CLK_SPI0_REG	0xff11d86c
@@ -202,12 +204,13 @@  int dw_spi_mid_init(struct dw_spi *dws)
 {
 	u32 *clk_reg, clk_cdiv;
 
+
 	clk_reg = ioremap_nocache(MRST_CLK_SPI0_REG, 16);
 	if (!clk_reg)
 		return -ENOMEM;
 
 	/* get SPI controller operating freq info */
-	clk_cdiv  = (readl(clk_reg) & CLK_SPI_CDIV_MASK) >> CLK_SPI_CDIV_OFFSET;
+	clk_cdiv  = ((*clk_reg) & CLK_SPI_CDIV_MASK) >> CLK_SPI_CDIV_OFFSET;
 	dws->max_freq = MRST_SPI_CLK_BASE / (clk_cdiv + 1);
 	iounmap(clk_reg);
 
diff --git a/drivers/spi/spi-dw.c b/drivers/spi/spi-dw.c
dissimilarity index 55%
index ece5f69..2dacb8f 100644
--- a/drivers/spi/spi-dw.c
+++ b/drivers/spi/spi-dw.c
@@ -1,936 +1,734 @@ 
-/*
- * Designware SPI core controller driver (refer pxa2xx_spi.c)
- *
- * Copyright (c) 2009, Intel Corporation.
- *
- * 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, write to the Free Software Foundation, Inc.,
- * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
- */
-
-#include <linux/dma-mapping.h>
-#include <linux/interrupt.h>
-#include <linux/highmem.h>
-#include <linux/delay.h>
-#include <linux/slab.h>
-#include <linux/spi/spi.h>
-
-#include "spi-dw.h"
-
-#ifdef CONFIG_DEBUG_FS
-#include <linux/debugfs.h>
-#endif
-
-#define START_STATE	((void *)0)
-#define RUNNING_STATE	((void *)1)
-#define DONE_STATE	((void *)2)
-#define ERROR_STATE	((void *)-1)
-
-#define QUEUE_RUNNING	0
-#define QUEUE_STOPPED	1
-
-#define MRST_SPI_DEASSERT	0
-#define MRST_SPI_ASSERT		1
-
-/* Slave spi_dev related */
-struct chip_data {
-	u16 cr0;
-	u8 cs;			/* chip select pin */
-	u8 n_bytes;		/* current is a 1/2/4 byte op */
-	u8 tmode;		/* TR/TO/RO/EEPROM */
-	u8 type;		/* SPI/SSP/MicroWire */
-
-	u8 poll_mode;		/* 1 means use poll mode */
-
-	u32 dma_width;
-	u32 rx_threshold;
-	u32 tx_threshold;
-	u8 enable_dma;
-	u8 bits_per_word;
-	u16 clk_div;		/* baud rate divider */
-	u32 speed_hz;		/* baud rate */
-	void (*cs_control)(u32 command);
-};
-
-#ifdef CONFIG_DEBUG_FS
-static int spi_show_regs_open(struct inode *inode, struct file *file)
-{
-	file->private_data = inode->i_private;
-	return 0;
-}
-
-#define SPI_REGS_BUFSIZE	1024
-static ssize_t  spi_show_regs(struct file *file, char __user *user_buf,
-				size_t count, loff_t *ppos)
-{
-	struct dw_spi *dws;
-	char *buf;
-	u32 len = 0;
-	ssize_t ret;
-
-	dws = file->private_data;
-
-	buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
-	if (!buf)
-		return 0;
-
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"MRST SPI0 registers:\n");
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"=================================\n");
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"SER: \t\t0x%08x\n", dw_readl(dws, ser));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"SR: \t\t0x%08x\n", dw_readl(dws, sr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"IMR: \t\t0x%08x\n", dw_readl(dws, imr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"ISR: \t\t0x%08x\n", dw_readl(dws, isr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"DMATDLR: \t0x%08x\n", dw_readl(dws, dmatdlr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"DMARDLR: \t0x%08x\n", dw_readl(dws, dmardlr));
-	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
-			"=================================\n");
-
-	ret =  simple_read_from_buffer(user_buf, count, ppos, buf, len);
-	kfree(buf);
-	return ret;
-}
-
-static const struct file_operations mrst_spi_regs_ops = {
-	.owner		= THIS_MODULE,
-	.open		= spi_show_regs_open,
-	.read		= spi_show_regs,
-	.llseek		= default_llseek,
-};
-
-static int mrst_spi_debugfs_init(struct dw_spi *dws)
-{
-	dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
-	if (!dws->debugfs)
-		return -ENOMEM;
-
-	debugfs_create_file("registers", S_IFREG | S_IRUGO,
-		dws->debugfs, (void *)dws, &mrst_spi_regs_ops);
-	return 0;
-}
-
-static void mrst_spi_debugfs_remove(struct dw_spi *dws)
-{
-	if (dws->debugfs)
-		debugfs_remove_recursive(dws->debugfs);
-}
-
-#else
-static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
-{
-	return 0;
-}
-
-static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
-{
-}
-#endif /* CONFIG_DEBUG_FS */
-
-/* Return the max entries we can fill into tx fifo */
-static inline u32 tx_max(struct dw_spi *dws)
-{
-	u32 tx_left, tx_room, rxtx_gap;
-
-	tx_left = (dws->tx_end - dws->tx) / dws->n_bytes;
-	tx_room = dws->fifo_len - dw_readw(dws, txflr);
-
-	/*
-	 * Another concern is about the tx/rx mismatch, we
-	 * though to use (dws->fifo_len - rxflr - txflr) as
-	 * one maximum value for tx, but it doesn't cover the
-	 * data which is out of tx/rx fifo and inside the
-	 * shift registers. So a control from sw point of
-	 * view is taken.
-	 */
-	rxtx_gap =  ((dws->rx_end - dws->rx) - (dws->tx_end - dws->tx))
-			/ dws->n_bytes;
-
-	return min3(tx_left, tx_room, (u32) (dws->fifo_len - rxtx_gap));
-}
-
-/* Return the max entries we should read out of rx fifo */
-static inline u32 rx_max(struct dw_spi *dws)
-{
-	u32 rx_left = (dws->rx_end - dws->rx) / dws->n_bytes;
-
-	return min(rx_left, (u32)dw_readw(dws, rxflr));
-}
-
-static void dw_writer(struct dw_spi *dws)
-{
-	u32 max = tx_max(dws);
-	u16 txw = 0;
-
-	while (max--) {
-		/* Set the tx word if the transfer's original "tx" is not null */
-		if (dws->tx_end - dws->len) {
-			if (dws->n_bytes == 1)
-				txw = *(u8 *)(dws->tx);
-			else
-				txw = *(u16 *)(dws->tx);
-		}
-		dw_writew(dws, dr, txw);
-		dws->tx += dws->n_bytes;
-	}
-}
-
-static void dw_reader(struct dw_spi *dws)
-{
-	u32 max = rx_max(dws);
-	u16 rxw;
-
-	while (max--) {
-		rxw = dw_readw(dws, dr);
-		/* Care rx only if the transfer's original "rx" is not null */
-		if (dws->rx_end - dws->len) {
-			if (dws->n_bytes == 1)
-				*(u8 *)(dws->rx) = rxw;
-			else
-				*(u16 *)(dws->rx) = rxw;
-		}
-		dws->rx += dws->n_bytes;
-	}
-}
-
-static void *next_transfer(struct dw_spi *dws)
-{
-	struct spi_message *msg = dws->cur_msg;
-	struct spi_transfer *trans = dws->cur_transfer;
-
-	/* Move to next transfer */
-	if (trans->transfer_list.next != &msg->transfers) {
-		dws->cur_transfer =
-			list_entry(trans->transfer_list.next,
-					struct spi_transfer,
-					transfer_list);
-		return RUNNING_STATE;
-	} else
-		return DONE_STATE;
-}
-
-/*
- * Note: first step is the protocol driver prepares
- * a dma-capable memory, and this func just need translate
- * the virt addr to physical
- */
-static int map_dma_buffers(struct dw_spi *dws)
-{
-	if (!dws->cur_msg->is_dma_mapped
-		|| !dws->dma_inited
-		|| !dws->cur_chip->enable_dma
-		|| !dws->dma_ops)
-		return 0;
-
-	if (dws->cur_transfer->tx_dma)
-		dws->tx_dma = dws->cur_transfer->tx_dma;
-
-	if (dws->cur_transfer->rx_dma)
-		dws->rx_dma = dws->cur_transfer->rx_dma;
-
-	return 1;
-}
-
-/* Caller already set message->status; dma and pio irqs are blocked */
-static void giveback(struct dw_spi *dws)
-{
-	struct spi_transfer *last_transfer;
-	unsigned long flags;
-	struct spi_message *msg;
-
-	spin_lock_irqsave(&dws->lock, flags);
-	msg = dws->cur_msg;
-	dws->cur_msg = NULL;
-	dws->cur_transfer = NULL;
-	dws->prev_chip = dws->cur_chip;
-	dws->cur_chip = NULL;
-	dws->dma_mapped = 0;
-	queue_work(dws->workqueue, &dws->pump_messages);
-	spin_unlock_irqrestore(&dws->lock, flags);
-
-	last_transfer = list_entry(msg->transfers.prev,
-					struct spi_transfer,
-					transfer_list);
-
-	if (!last_transfer->cs_change && dws->cs_control)
-		dws->cs_control(MRST_SPI_DEASSERT);
-
-	msg->state = NULL;
-	if (msg->complete)
-		msg->complete(msg->context);
-}
-
-static void int_error_stop(struct dw_spi *dws, const char *msg)
-{
-	/* Stop the hw */
-	spi_enable_chip(dws, 0);
-
-	dev_err(&dws->master->dev, "%s\n", msg);
-	dws->cur_msg->state = ERROR_STATE;
-	tasklet_schedule(&dws->pump_transfers);
-}
-
-void dw_spi_xfer_done(struct dw_spi *dws)
-{
-	/* Update total byte transferred return count actual bytes read */
-	dws->cur_msg->actual_length += dws->len;
-
-	/* Move to next transfer */
-	dws->cur_msg->state = next_transfer(dws);
-
-	/* Handle end of message */
-	if (dws->cur_msg->state == DONE_STATE) {
-		dws->cur_msg->status = 0;
-		giveback(dws);
-	} else
-		tasklet_schedule(&dws->pump_transfers);
-}
-EXPORT_SYMBOL_GPL(dw_spi_xfer_done);
-
-static irqreturn_t interrupt_transfer(struct dw_spi *dws)
-{
-	u16 irq_status = dw_readw(dws, isr);
-
-	/* Error handling */
-	if (irq_status & (SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI)) {
-		dw_readw(dws, txoicr);
-		dw_readw(dws, rxoicr);
-		dw_readw(dws, rxuicr);
-		int_error_stop(dws, "interrupt_transfer: fifo overrun/underrun");
-		return IRQ_HANDLED;
-	}
-
-	dw_reader(dws);
-	if (dws->rx_end == dws->rx) {
-		spi_mask_intr(dws, SPI_INT_TXEI);
-		dw_spi_xfer_done(dws);
-		return IRQ_HANDLED;
-	}
-	if (irq_status & SPI_INT_TXEI) {
-		spi_mask_intr(dws, SPI_INT_TXEI);
-		dw_writer(dws);
-		/* Enable TX irq always, it will be disabled when RX finished */
-		spi_umask_intr(dws, SPI_INT_TXEI);
-	}
-
-	return IRQ_HANDLED;
-}
-
-static irqreturn_t dw_spi_irq(int irq, void *dev_id)
-{
-	struct dw_spi *dws = dev_id;
-	u16 irq_status = dw_readw(dws, isr) & 0x3f;
-
-	if (!irq_status)
-		return IRQ_NONE;
-
-	if (!dws->cur_msg) {
-		spi_mask_intr(dws, SPI_INT_TXEI);
-		return IRQ_HANDLED;
-	}
-
-	return dws->transfer_handler(dws);
-}
-
-/* Must be called inside pump_transfers() */
-static void poll_transfer(struct dw_spi *dws)
-{
-	do {
-		dw_writer(dws);
-		dw_reader(dws);
-		cpu_relax();
-	} while (dws->rx_end > dws->rx);
-
-	dw_spi_xfer_done(dws);
-}
-
-static void pump_transfers(unsigned long data)
-{
-	struct dw_spi *dws = (struct dw_spi *)data;
-	struct spi_message *message = NULL;
-	struct spi_transfer *transfer = NULL;
-	struct spi_transfer *previous = NULL;
-	struct spi_device *spi = NULL;
-	struct chip_data *chip = NULL;
-	u8 bits = 0;
-	u8 imask = 0;
-	u8 cs_change = 0;
-	u16 txint_level = 0;
-	u16 clk_div = 0;
-	u32 speed = 0;
-	u32 cr0 = 0;
-
-	/* Get current state information */
-	message = dws->cur_msg;
-	transfer = dws->cur_transfer;
-	chip = dws->cur_chip;
-	spi = message->spi;
-
-	if (unlikely(!chip->clk_div))
-		chip->clk_div = dws->max_freq / chip->speed_hz;
-
-	if (message->state == ERROR_STATE) {
-		message->status = -EIO;
-		goto early_exit;
-	}
-
-	/* Handle end of message */
-	if (message->state == DONE_STATE) {
-		message->status = 0;
-		goto early_exit;
-	}
-
-	/* Delay if requested at end of transfer*/
-	if (message->state == RUNNING_STATE) {
-		previous = list_entry(transfer->transfer_list.prev,
-					struct spi_transfer,
-					transfer_list);
-		if (previous->delay_usecs)
-			udelay(previous->delay_usecs);
-	}
-
-	dws->n_bytes = chip->n_bytes;
-	dws->dma_width = chip->dma_width;
-	dws->cs_control = chip->cs_control;
-
-	dws->rx_dma = transfer->rx_dma;
-	dws->tx_dma = transfer->tx_dma;
-	dws->tx = (void *)transfer->tx_buf;
-	dws->tx_end = dws->tx + transfer->len;
-	dws->rx = transfer->rx_buf;
-	dws->rx_end = dws->rx + transfer->len;
-	dws->cs_change = transfer->cs_change;
-	dws->len = dws->cur_transfer->len;
-	if (chip != dws->prev_chip)
-		cs_change = 1;
-
-	cr0 = chip->cr0;
-
-	/* Handle per transfer options for bpw and speed */
-	if (transfer->speed_hz) {
-		speed = chip->speed_hz;
-
-		if (transfer->speed_hz != speed) {
-			speed = transfer->speed_hz;
-			if (speed > dws->max_freq) {
-				printk(KERN_ERR "MRST SPI0: unsupported"
-					"freq: %dHz\n", speed);
-				message->status = -EIO;
-				goto early_exit;
-			}
-
-			/* clk_div doesn't support odd number */
-			clk_div = dws->max_freq / speed;
-			clk_div = (clk_div + 1) & 0xfffe;
-
-			chip->speed_hz = speed;
-			chip->clk_div = clk_div;
-		}
-	}
-	if (transfer->bits_per_word) {
-		bits = transfer->bits_per_word;
-
-		switch (bits) {
-		case 8:
-		case 16:
-			dws->n_bytes = dws->dma_width = bits >> 3;
-			break;
-		default:
-			printk(KERN_ERR "MRST SPI0: unsupported bits:"
-				"%db\n", bits);
-			message->status = -EIO;
-			goto early_exit;
-		}
-
-		cr0 = (bits - 1)
-			| (chip->type << SPI_FRF_OFFSET)
-			| (spi->mode << SPI_MODE_OFFSET)
-			| (chip->tmode << SPI_TMOD_OFFSET);
-	}
-	message->state = RUNNING_STATE;
-
-	/*
-	 * Adjust transfer mode if necessary. Requires platform dependent
-	 * chipselect mechanism.
-	 */
-	if (dws->cs_control) {
-		if (dws->rx && dws->tx)
-			chip->tmode = SPI_TMOD_TR;
-		else if (dws->rx)
-			chip->tmode = SPI_TMOD_RO;
-		else
-			chip->tmode = SPI_TMOD_TO;
-
-		cr0 &= ~SPI_TMOD_MASK;
-		cr0 |= (chip->tmode << SPI_TMOD_OFFSET);
-	}
-
-	/* Check if current transfer is a DMA transaction */
-	dws->dma_mapped = map_dma_buffers(dws);
-
-	/*
-	 * Interrupt mode
-	 * we only need set the TXEI IRQ, as TX/RX always happen syncronizely
-	 */
-	if (!dws->dma_mapped && !chip->poll_mode) {
-		int templen = dws->len / dws->n_bytes;
-		txint_level = dws->fifo_len / 2;
-		txint_level = (templen > txint_level) ? txint_level : templen;
-
-		imask |= SPI_INT_TXEI | SPI_INT_TXOI | SPI_INT_RXUI | SPI_INT_RXOI;
-		dws->transfer_handler = interrupt_transfer;
-	}
-
-	/*
-	 * Reprogram registers only if
-	 *	1. chip select changes
-	 *	2. clk_div is changed
-	 *	3. control value changes
-	 */
-	if (dw_readw(dws, ctrl0) != cr0 || cs_change || clk_div || imask) {
-		spi_enable_chip(dws, 0);
-
-		if (dw_readw(dws, ctrl0) != cr0)
-			dw_writew(dws, ctrl0, cr0);
-
-		spi_set_clk(dws, clk_div ? clk_div : chip->clk_div);
-		spi_chip_sel(dws, spi->chip_select);
-
-		/* Set the interrupt mask, for poll mode just disable all int */
-		spi_mask_intr(dws, 0xff);
-		if (imask)
-			spi_umask_intr(dws, imask);
-		if (txint_level)
-			dw_writew(dws, txfltr, txint_level);
-
-		spi_enable_chip(dws, 1);
-		if (cs_change)
-			dws->prev_chip = chip;
-	}
-
-	if (dws->dma_mapped)
-		dws->dma_ops->dma_transfer(dws, cs_change);
-
-	if (chip->poll_mode)
-		poll_transfer(dws);
-
-	return;
-
-early_exit:
-	giveback(dws);
-	return;
-}
-
-static void pump_messages(struct work_struct *work)
-{
-	struct dw_spi *dws =
-		container_of(work, struct dw_spi, pump_messages);
-	unsigned long flags;
-
-	/* Lock queue and check for queue work */
-	spin_lock_irqsave(&dws->lock, flags);
-	if (list_empty(&dws->queue) || dws->run == QUEUE_STOPPED) {
-		dws->busy = 0;
-		spin_unlock_irqrestore(&dws->lock, flags);
-		return;
-	}
-
-	/* Make sure we are not already running a message */
-	if (dws->cur_msg) {
-		spin_unlock_irqrestore(&dws->lock, flags);
-		return;
-	}
-
-	/* Extract head of queue */
-	dws->cur_msg = list_entry(dws->queue.next, struct spi_message, queue);
-	list_del_init(&dws->cur_msg->queue);
-
-	/* Initial message state*/
-	dws->cur_msg->state = START_STATE;
-	dws->cur_transfer = list_entry(dws->cur_msg->transfers.next,
-						struct spi_transfer,
-						transfer_list);
-	dws->cur_chip = spi_get_ctldata(dws->cur_msg->spi);
-
-	/* Mark as busy and launch transfers */
-	tasklet_schedule(&dws->pump_transfers);
-
-	dws->busy = 1;
-	spin_unlock_irqrestore(&dws->lock, flags);
-}
-
-/* spi_device use this to queue in their spi_msg */
-static int dw_spi_transfer(struct spi_device *spi, struct spi_message *msg)
-{
-	struct dw_spi *dws = spi_master_get_devdata(spi->master);
-	unsigned long flags;
-
-	spin_lock_irqsave(&dws->lock, flags);
-
-	if (dws->run == QUEUE_STOPPED) {
-		spin_unlock_irqrestore(&dws->lock, flags);
-		return -ESHUTDOWN;
-	}
-
-	msg->actual_length = 0;
-	msg->status = -EINPROGRESS;
-	msg->state = START_STATE;
-
-	list_add_tail(&msg->queue, &dws->queue);
-
-	if (dws->run == QUEUE_RUNNING && !dws->busy) {
-
-		if (dws->cur_transfer || dws->cur_msg)
-			queue_work(dws->workqueue,
-					&dws->pump_messages);
-		else {
-			/* If no other data transaction in air, just go */
-			spin_unlock_irqrestore(&dws->lock, flags);
-			pump_messages(&dws->pump_messages);
-			return 0;
-		}
-	}
-
-	spin_unlock_irqrestore(&dws->lock, flags);
-	return 0;
-}
-
-/* This may be called twice for each spi dev */
-static int dw_spi_setup(struct spi_device *spi)
-{
-	struct dw_spi_chip *chip_info = NULL;
-	struct chip_data *chip;
-
-	if (spi->bits_per_word != 8 && spi->bits_per_word != 16)
-		return -EINVAL;
-
-	/* Only alloc on first setup */
-	chip = spi_get_ctldata(spi);
-	if (!chip) {
-		chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
-		if (!chip)
-			return -ENOMEM;
-	}
-
-	/*
-	 * Protocol drivers may change the chip settings, so...
-	 * if chip_info exists, use it
-	 */
-	chip_info = spi->controller_data;
-
-	/* chip_info doesn't always exist */
-	if (chip_info) {
-		if (chip_info->cs_control)
-			chip->cs_control = chip_info->cs_control;
-
-		chip->poll_mode = chip_info->poll_mode;
-		chip->type = chip_info->type;
-
-		chip->rx_threshold = 0;
-		chip->tx_threshold = 0;
-
-		chip->enable_dma = chip_info->enable_dma;
-	}
-
-	if (spi->bits_per_word <= 8) {
-		chip->n_bytes = 1;
-		chip->dma_width = 1;
-	} else if (spi->bits_per_word <= 16) {
-		chip->n_bytes = 2;
-		chip->dma_width = 2;
-	} else {
-		/* Never take >16b case for MRST SPIC */
-		dev_err(&spi->dev, "invalid wordsize\n");
-		return -EINVAL;
-	}
-	chip->bits_per_word = spi->bits_per_word;
-
-	if (!spi->max_speed_hz) {
-		dev_err(&spi->dev, "No max speed HZ parameter\n");
-		return -EINVAL;
-	}
-	chip->speed_hz = spi->max_speed_hz;
-
-	chip->tmode = 0; /* Tx & Rx */
-	/* Default SPI mode is SCPOL = 0, SCPH = 0 */
-	chip->cr0 = (chip->bits_per_word - 1)
-			| (chip->type << SPI_FRF_OFFSET)
-			| (spi->mode  << SPI_MODE_OFFSET)
-			| (chip->tmode << SPI_TMOD_OFFSET);
-
-	spi_set_ctldata(spi, chip);
-	return 0;
-}
-
-static void dw_spi_cleanup(struct spi_device *spi)
-{
-	struct chip_data *chip = spi_get_ctldata(spi);
-	kfree(chip);
-}
-
-static int __devinit init_queue(struct dw_spi *dws)
-{
-	INIT_LIST_HEAD(&dws->queue);
-	spin_lock_init(&dws->lock);
-
-	dws->run = QUEUE_STOPPED;
-	dws->busy = 0;
-
-	tasklet_init(&dws->pump_transfers,
-			pump_transfers,	(unsigned long)dws);
-
-	INIT_WORK(&dws->pump_messages, pump_messages);
-	dws->workqueue = create_singlethread_workqueue(
-					dev_name(dws->master->dev.parent));
-	if (dws->workqueue == NULL)
-		return -EBUSY;
-
-	return 0;
-}
-
-static int start_queue(struct dw_spi *dws)
-{
-	unsigned long flags;
-
-	spin_lock_irqsave(&dws->lock, flags);
-
-	if (dws->run == QUEUE_RUNNING || dws->busy) {
-		spin_unlock_irqrestore(&dws->lock, flags);
-		return -EBUSY;
-	}
-
-	dws->run = QUEUE_RUNNING;
-	dws->cur_msg = NULL;
-	dws->cur_transfer = NULL;
-	dws->cur_chip = NULL;
-	dws->prev_chip = NULL;
-	spin_unlock_irqrestore(&dws->lock, flags);
-
-	queue_work(dws->workqueue, &dws->pump_messages);
-
-	return 0;
-}
-
-static int stop_queue(struct dw_spi *dws)
-{
-	unsigned long flags;
-	unsigned limit = 50;
-	int status = 0;
-
-	spin_lock_irqsave(&dws->lock, flags);
-	dws->run = QUEUE_STOPPED;
-	while ((!list_empty(&dws->queue) || dws->busy) && limit--) {
-		spin_unlock_irqrestore(&dws->lock, flags);
-		msleep(10);
-		spin_lock_irqsave(&dws->lock, flags);
-	}
-
-	if (!list_empty(&dws->queue) || dws->busy)
-		status = -EBUSY;
-	spin_unlock_irqrestore(&dws->lock, flags);
-
-	return status;
-}
-
-static int destroy_queue(struct dw_spi *dws)
-{
-	int status;
-
-	status = stop_queue(dws);
-	if (status != 0)
-		return status;
-	destroy_workqueue(dws->workqueue);
-	return 0;
-}
-
-/* Restart the controller, disable all interrupts, clean rx fifo */
-static void spi_hw_init(struct dw_spi *dws)
-{
-	spi_enable_chip(dws, 0);
-	spi_mask_intr(dws, 0xff);
-	spi_enable_chip(dws, 1);
-
-	/*
-	 * Try to detect the FIFO depth if not set by interface driver,
-	 * the depth could be from 2 to 256 from HW spec
-	 */
-	if (!dws->fifo_len) {
-		u32 fifo;
-		for (fifo = 2; fifo <= 257; fifo++) {
-			dw_writew(dws, txfltr, fifo);
-			if (fifo != dw_readw(dws, txfltr))
-				break;
-		}
-
-		dws->fifo_len = (fifo == 257) ? 0 : fifo;
-		dw_writew(dws, txfltr, 0);
-	}
-}
-
-int __devinit dw_spi_add_host(struct dw_spi *dws)
-{
-	struct spi_master *master;
-	int ret;
-
-	BUG_ON(dws == NULL);
-
-	master = spi_alloc_master(dws->parent_dev, 0);
-	if (!master) {
-		ret = -ENOMEM;
-		goto exit;
-	}
-
-	dws->master = master;
-	dws->type = SSI_MOTO_SPI;
-	dws->prev_chip = NULL;
-	dws->dma_inited = 0;
-	dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
-
-	ret = request_irq(dws->irq, dw_spi_irq, IRQF_SHARED,
-			"dw_spi", dws);
-	if (ret < 0) {
-		dev_err(&master->dev, "can not get IRQ\n");
-		goto err_free_master;
-	}
-
-	master->mode_bits = SPI_CPOL | SPI_CPHA;
-	master->bus_num = dws->bus_num;
-	master->num_chipselect = dws->num_cs;
-	master->cleanup = dw_spi_cleanup;
-	master->setup = dw_spi_setup;
-	master->transfer = dw_spi_transfer;
-
-	/* Basic HW init */
-	spi_hw_init(dws);
-
-	if (dws->dma_ops && dws->dma_ops->dma_init) {
-		ret = dws->dma_ops->dma_init(dws);
-		if (ret) {
-			dev_warn(&master->dev, "DMA init failed\n");
-			dws->dma_inited = 0;
-		}
-	}
-
-	/* Initial and start queue */
-	ret = init_queue(dws);
-	if (ret) {
-		dev_err(&master->dev, "problem initializing queue\n");
-		goto err_diable_hw;
-	}
-	ret = start_queue(dws);
-	if (ret) {
-		dev_err(&master->dev, "problem starting queue\n");
-		goto err_diable_hw;
-	}
-
-	spi_master_set_devdata(master, dws);
-	ret = spi_register_master(master);
-	if (ret) {
-		dev_err(&master->dev, "problem registering spi master\n");
-		goto err_queue_alloc;
-	}
-
-	mrst_spi_debugfs_init(dws);
-	return 0;
-
-err_queue_alloc:
-	destroy_queue(dws);
-	if (dws->dma_ops && dws->dma_ops->dma_exit)
-		dws->dma_ops->dma_exit(dws);
-err_diable_hw:
-	spi_enable_chip(dws, 0);
-	free_irq(dws->irq, dws);
-err_free_master:
-	spi_master_put(master);
-exit:
-	return ret;
-}
-EXPORT_SYMBOL_GPL(dw_spi_add_host);
-
-void __devexit dw_spi_remove_host(struct dw_spi *dws)
-{
-	int status = 0;
-
-	if (!dws)
-		return;
-	mrst_spi_debugfs_remove(dws);
-
-	/* Remove the queue */
-	status = destroy_queue(dws);
-	if (status != 0)
-		dev_err(&dws->master->dev, "dw_spi_remove: workqueue will not "
-			"complete, message memory not freed\n");
-
-	if (dws->dma_ops && dws->dma_ops->dma_exit)
-		dws->dma_ops->dma_exit(dws);
-	spi_enable_chip(dws, 0);
-	/* Disable clk */
-	spi_set_clk(dws, 0);
-	free_irq(dws->irq, dws);
-
-	/* Disconnect from the SPI framework */
-	spi_unregister_master(dws->master);
-}
-EXPORT_SYMBOL_GPL(dw_spi_remove_host);
-
-int dw_spi_suspend_host(struct dw_spi *dws)
-{
-	int ret = 0;
-
-	ret = stop_queue(dws);
-	if (ret)
-		return ret;
-	spi_enable_chip(dws, 0);
-	spi_set_clk(dws, 0);
-	return ret;
-}
-EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
-
-int dw_spi_resume_host(struct dw_spi *dws)
-{
-	int ret;
-
-	spi_hw_init(dws);
-	ret = start_queue(dws);
-	if (ret)
-		dev_err(&dws->master->dev, "fail to start queue (%d)\n", ret);
-	return ret;
-}
-EXPORT_SYMBOL_GPL(dw_spi_resume_host);
-
-MODULE_AUTHOR("Feng Tang <feng.tang@intel.com>");
-MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
-MODULE_LICENSE("GPL v2");
+/*
+ * dw_spi.c - Designware SPI core controller driver
+ *
+ * Copyright (c) 2009, Intel Corporation.
+ *
+ * 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, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/interrupt.h>
+#include <linux/highmem.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+#include <linux/spi/spi.h>
+
+#include "spi-dw.h"
+
+#ifdef CONFIG_DEBUG_FS
+#include <linux/debugfs.h>
+#endif
+
+
+#define QUEUE_RUNNING	0
+#define QUEUE_STOPPED	1
+
+
+/* Slave spi_dev related */
+struct chip_data {
+	struct spi_device *spi_dev;
+	u32 cr0;
+	u32 cs;			/* chip select pin */
+	u32 n_bytes;		/* current is a 1/2/4 byte op */
+	u32 type;		/* SPI/SSP/MicroWire */
+
+	u32 dma_width;
+	u32 enable_dma;
+	u32 bits_per_word;
+	u32 clk_div;		/* baud rate divider */
+	u32 speed_hz;		/* baud rate */
+};
+
+#ifdef CONFIG_DEBUG_FS
+static int spi_show_regs_open(struct inode *inode, struct file *file)
+{
+	file->private_data = inode->i_private;
+	return 0;
+}
+
+#define SPI_REGS_BUFSIZE	1024
+static ssize_t  spi_show_regs(struct file *file, char __user *user_buf,
+				size_t count, loff_t *ppos)
+{
+	struct dw_spi *dws;
+	char *buf;
+	u32 len = 0;
+	ssize_t ret;
+
+	dws = file->private_data;
+
+	buf = kzalloc(SPI_REGS_BUFSIZE, GFP_KERNEL);
+	if (!buf)
+		return 0;
+
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"MRST SPI0 registers:\n");
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"=================================\n");
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"CTRL0: \t\t0x%08x\n", dw_readl(dws, ctrl0));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"CTRL1: \t\t0x%08x\n", dw_readl(dws, ctrl1));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"SSIENR: \t0x%08x\n", dw_readl(dws, ssienr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"SER: \t\t0x%08x\n", dw_readl(dws, ser));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"BAUDR: \t\t0x%08x\n", dw_readl(dws, baudr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"TXFTLR: \t0x%08x\n", dw_readl(dws, txfltr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"RXFTLR: \t0x%08x\n", dw_readl(dws, rxfltr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"TXFLR: \t\t0x%08x\n", dw_readl(dws, txflr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"RXFLR: \t\t0x%08x\n", dw_readl(dws, rxflr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"SR: \t\t0x%08x\n", dw_readl(dws, sr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"IMR: \t\t0x%08x\n", dw_readl(dws, imr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"ISR: \t\t0x%08x\n", dw_readl(dws, isr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"DMACR: \t\t0x%08x\n", dw_readl(dws, dmacr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"DMATDLR: \t0x%08x\n", dw_readl(dws, dmatdlr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"DMARDLR: \t0x%08x\n", dw_readl(dws, dmardlr));
+	len += snprintf(buf + len, SPI_REGS_BUFSIZE - len,
+			"=================================\n");
+
+	ret =  simple_read_from_buffer(user_buf, count, ppos, buf, len);
+	kfree(buf);
+	return ret;
+}
+
+static const struct file_operations mrst_spi_regs_ops = {
+	.owner		= THIS_MODULE,
+	.open		= spi_show_regs_open,
+	.read		= spi_show_regs,
+	.llseek		= default_llseek,
+};
+
+static int mrst_spi_debugfs_init(struct dw_spi *dws)
+{
+	dws->debugfs = debugfs_create_dir("mrst_spi", NULL);
+	if (!dws->debugfs)
+		return -ENOMEM;
+
+	debugfs_create_file("registers", S_IFREG | S_IRUGO,
+		dws->debugfs, (void *)dws, &mrst_spi_regs_ops);
+	return 0;
+}
+
+static void mrst_spi_debugfs_remove(struct dw_spi *dws)
+{
+	if (dws->debugfs)
+		debugfs_remove_recursive(dws->debugfs);
+}
+
+#else
+static inline int mrst_spi_debugfs_init(struct dw_spi *dws)
+{
+	return 0;
+}
+
+static inline void mrst_spi_debugfs_remove(struct dw_spi *dws)
+{
+}
+#endif /* CONFIG_DEBUG_FS */
+
+static irqreturn_t dw_spi_irq(int irq, void *dev_id)
+{
+	struct dw_spi *dws = dev_id;
+	u16  irq_mask = 0x3f;
+
+	dws->xfer.irq_status = dw_readw(dws, isr) & irq_mask;
+
+	if (!dws->xfer.irq_status)
+		return IRQ_NONE;
+	if (dws->xfer.irq_status &
+			(SPI_INT_TXOI | SPI_INT_RXOI | SPI_INT_RXUI)) {
+		dw_readw(dws, txoicr);
+		dw_readw(dws, rxoicr);
+		dw_readw(dws, rxuicr);
+		dws->xfer.err = -EIO;
+		dw_spi_disable(dws);
+		complete(&dws->xfer.complete);
+		return IRQ_HANDLED;
+	}
+
+	/* disable interrupts */
+	dw_spi_mask_intr(dws, irq_mask);
+	return IRQ_WAKE_THREAD;
+}
+struct spi_message *get_message(struct dw_spi *dws)
+{
+	struct spi_message *message = NULL;
+	unsigned long flags;
+
+	spin_lock_irqsave(&dws->lock, flags);
+	if (!list_empty(&dws->queue)) {
+		message = list_entry(dws->queue.next,
+				struct spi_message, queue);
+		list_del_init(&message->queue);
+	}
+	spin_unlock_irqrestore(&dws->lock, flags);
+	return message;
+}
+static inline u32 tx_max(struct dw_spi *dws)
+{
+	u32 tx_left, tx_room;
+
+	tx_left = (dws->xfer.len - dws->xfer.sent) / dws->xfer.n_bytes;
+	tx_room = (dws->fifo_len - dw_readw(dws, txflr));
+
+	return min(tx_left, tx_room);
+}
+
+/* Return the max entries we should read out of rx fifo */
+static inline u32 rx_max(struct dw_spi *dws)
+{
+	u32 rx_left = (dws->xfer.len - dws->xfer.rcvd) / dws->xfer.n_bytes;
+	return min(rx_left, (u32)dw_readw(dws, rxflr));
+}
+
+static int transfer_setup(struct dw_spi *dws, struct chip_data *controller,
+			struct spi_transfer *transfer, struct spi_message *msg)
+{
+	int err = 0;
+	u32 cr0 = controller->cr0;
+	u32 clk_div;
+
+	dws->xfer.tx_buf = transfer->tx_buf;
+	dws->xfer.tx_dma = transfer->tx_dma;
+	dws->xfer.rx_buf = transfer->rx_buf;
+	dws->xfer.rx_dma = transfer->rx_dma;
+
+	dws->xfer.len = transfer->len;
+	dws->xfer.n_bytes = controller->n_bytes;
+	dws->xfer.sent = 0;
+	dws->xfer.rcvd = 0;
+	dws->xfer.msg = msg;
+	dws->xfer.err = 0;
+	dws->xfer.irq_status = 0;
+	INIT_COMPLETION(dws->xfer.complete);
+
+	/* {tx, rx}_threshold should probably be a module param with
+	 *  some reasonable default but these work for now.
+	 */
+	dws->xfer.tx_threshold = 10;
+	dws->xfer.rx_threshold = dws->xfer.tx_threshold - 1;
+
+	/* we need to make the decsion about the type of transfer more
+	 *  inteligently but this works for now
+	 */
+	if (transfer->len > dws->fifo_len)
+		dws->xfer.type = INT_XFER;
+	else
+		dws->xfer.type = PIO_XFER;
+
+	if (controller->enable_dma &&
+		msg->is_dma_mapped &&
+		dws->dma_inited &&
+		dws->dma_ops)
+		dws->xfer.type = DMA_XFER;
+
+	/* Setup the controller based on parameters in the transfer
+	 * each transfer can set the bit_per_word and the speed_hz to
+	 * change these values in the controller the controller MUST
+	 * be disabled
+	 */
+	if (unlikely(!controller->clk_div)) {
+		controller->clk_div = dws->max_freq / controller->speed_hz;
+		controller->clk_div = (controller->clk_div + 1) & 0xfffe;
+		dw_spi_set_clk(dws, controller->clk_div);
+		dev_err(&dws->master->dev, "setting default clk_div");
+		err = 1;
+	}
+
+	if (transfer->speed_hz) {
+		if (transfer->speed_hz != controller->speed_hz) {
+			if (transfer->speed_hz > dws->max_freq) {
+				err = -EIO;
+				goto out;
+			}
+
+			clk_div = dws->max_freq / transfer->speed_hz;
+			clk_div = (clk_div + 1) & 0xfffe;
+			controller->clk_div = clk_div;
+			controller->speed_hz = transfer->speed_hz;
+			err = 1;
+		}
+	}
+
+	if (transfer->bits_per_word) {
+		if (transfer->bits_per_word != 8 &&
+			transfer->bits_per_word != 16) {
+			err = -EIO;
+			goto out;
+		}
+		cr0 &= ~SPI_DFS_MASK;
+		cr0 |= transfer->bits_per_word - 1;
+		err = 1;
+	} else {
+		cr0 &= ~SPI_DFS_MASK;
+		cr0 |= controller->spi_dev->bits_per_word - 1;
+	}
+
+	cr0 &= ~SPI_MODE_MASK;
+	cr0 |= (controller->spi_dev->mode << SPI_MODE_OFFSET);
+	controller->cr0 = cr0;
+
+	if (err || dw_readw(dws, ctrl0) != cr0) {
+		dw_spi_disable(dws);
+		dw_spi_chip_sel(dws, controller->spi_dev->chip_select);
+		dw_writew(dws, ctrl0, cr0);
+		dw_spi_set_clk(dws, controller->clk_div);
+		dw_spi_enable(dws);
+		err = 0;
+	}
+out:
+	return err;
+}
+
+static void tx_fifo_fill(struct dw_spi *dws)
+{
+	int room;
+	u16 txw = 0;
+	if (dws->xfer.sent < dws->xfer.len) {
+		room = tx_max(dws);
+		while (room--) {
+			if (dws->xfer.tx_buf) {
+				if (dws->xfer.n_bytes == 2)
+					txw = *(u16 *)dws->xfer.tx_buf;
+				else
+					txw = *(u8 *)dws->xfer.tx_buf;
+				dws->xfer.tx_buf += dws->xfer.n_bytes;
+			}
+			dw_writew(dws, dr, txw);
+			dws->xfer.sent += dws->xfer.n_bytes;
+		}
+	}
+}
+
+
+static void rx_fifo_drain(struct dw_spi *dws)
+{
+	u16 rx_val;
+	int avail;
+
+	if (dws->xfer.rcvd < dws->xfer.len) {
+		avail = rx_max(dws);
+		while (avail--) {
+			rx_val = dw_readw(dws, dr);
+			if (dws->xfer.rx_buf) {
+				if (dws->xfer.n_bytes == 2)
+					*(u16 *)(dws->xfer.rx_buf) =
+						(u16)rx_val;
+				else
+					*dws->xfer.rx_buf = (u8)rx_val;
+				dws->xfer.rx_buf += dws->xfer.n_bytes;
+			}
+			dws->xfer.rcvd += dws->xfer.n_bytes;
+		}
+	}
+}
+
+static inline void do_pio(struct dw_spi *dws)
+{
+	while (dws->xfer.sent < dws->xfer.len ||
+					dws->xfer.rcvd < dws->xfer.len) {
+		tx_fifo_fill(dws);
+		rx_fifo_drain(dws);
+		cpu_relax();
+	}
+	complete(&dws->xfer.complete);
+}
+
+static irqreturn_t dw_spi_irq_thread_handler(int irq, void *dev_id)
+{
+	struct dw_spi *dws = dev_id;
+
+	if (dws->xfer.irq_status & SPI_INT_TXEI) {
+		rx_fifo_drain(dws);
+		tx_fifo_fill(dws);
+	}
+
+	if (dws->xfer.irq_status & SPI_INT_RXFI) {
+		tx_fifo_fill(dws);
+		rx_fifo_drain(dws);
+	}
+
+	if (dws->xfer.len == dws->xfer.rcvd &&
+		dws->xfer.len == dws->xfer.sent) {
+		complete(&dws->xfer.complete);
+		goto out;
+	}
+
+	dw_spi_umask_intr(dws, SPI_INT_ALL);
+out:
+	return IRQ_HANDLED;
+}
+
+static inline void do_int_xfer(struct dw_spi *dws)
+{
+	dw_spi_disable(dws);
+	dw_writew(dws, txfltr, dws->xfer.tx_threshold);
+	dw_writew(dws, rxfltr, dws->xfer.rx_threshold);
+	dw_spi_enable(dws);
+	dw_readw(dws, icr);
+	tx_fifo_fill(dws);
+	dw_writew(dws, imr, SPI_INT_ALL);
+}
+
+static inline int do_transfer(struct dw_spi *dws)
+{
+	switch (dws->xfer.type) {
+	case PIO_XFER:
+		do_pio(dws);
+		break;
+	case INT_XFER:
+		do_int_xfer(dws);
+		break;
+	case DMA_XFER:
+		dws->dma_ops->dma_transfer(dws);
+		break;
+	default:
+		BUG();
+	}
+
+	wait_for_completion(&dws->xfer.complete);
+
+
+	return dws->xfer.err;
+}
+
+static void drain_message_queue(struct dw_spi *dws)
+{
+	struct spi_message *message;
+
+	message = get_message(dws);
+	while (message) {
+		message->status = -ESHUTDOWN;
+		message->complete(message->context);
+		message = get_message(dws);
+	}
+}
+
+static void pump_messages(struct work_struct *work)
+{
+	struct dw_spi *dws =
+		container_of(work, struct dw_spi, pump_messages);
+	struct spi_transfer *transfer;
+	struct spi_message *message;
+	struct chip_data *controller;
+	int err = 0;
+
+
+	message = get_message(dws);
+
+	while (message && dws->run != QUEUE_STOPPED) {
+		controller = spi_get_ctldata(message->spi);
+		list_for_each_entry(transfer, &message->transfers,
+				transfer_list){
+
+			err = transfer_setup(dws, controller,
+						transfer, message);
+			if (err < 0) {
+				dev_err(&dws->master->dev,
+					"transfer_setup failed");
+				dws->xfer.err = -EIO;
+				break;
+			}
+
+			err = do_transfer(dws);
+			if (err < 0) {
+				dev_err(&dws->master->dev,
+					"do_transfer failed");
+				break;
+			}
+			message->actual_length += dws->xfer.len;
+
+			if (transfer->delay_usecs)
+				udelay(transfer->delay_usecs);
+		}
+
+		message->status = dws->xfer.err;
+		message->complete(message->context);
+		message =  get_message(dws);
+	}
+	if (dws->run == QUEUE_STOPPED)
+		drain_message_queue(dws);
+}
+
+/* spi_device use this to queue in their spi_msg */
+static int dw_spi_transfer(struct spi_device *spi, struct spi_message *msg)
+{
+	struct dw_spi *dws = spi_master_get_devdata(spi->master);
+	unsigned long flags;
+
+
+	spin_lock_irqsave(&dws->lock, flags);
+
+	msg->actual_length = 0;
+	msg->status = -EINPROGRESS;
+
+	list_add_tail(&msg->queue, &dws->queue);
+
+	queue_work(dws->workqueue,
+		&dws->pump_messages);
+
+	spin_unlock_irqrestore(&dws->lock, flags);
+
+	return 0;
+}
+
+/* This may be called twice for each spi dev */
+static int dw_spi_setup(struct spi_device *spi)
+{
+	struct dw_spi_chip *chip_info = NULL;
+	struct chip_data *chip;
+
+	if (spi->bits_per_word != 8 && spi->bits_per_word != 16)
+		return -EINVAL;
+
+	if (!spi->max_speed_hz) {
+		dev_err(&spi->dev, "No max speed HZ parameter\n");
+		return -EINVAL;
+	}
+
+	/* Only alloc on first setup */
+	chip = spi_get_ctldata(spi);
+	if (!chip) {
+		chip = kzalloc(sizeof(struct chip_data), GFP_KERNEL);
+		if (!chip)
+			return -ENOMEM;
+	}
+	chip->spi_dev = spi;
+
+	/*
+	 * Protocol drivers may change the chip settings, so...
+	 * if chip_info exists, use it
+	 */
+	chip_info = spi->controller_data;
+
+	/* chip_info doesn't always exist */
+	if (chip_info) {
+		chip->type = chip_info->type;
+		chip->enable_dma = chip_info->enable_dma;
+	}
+
+	chip->bits_per_word = spi->bits_per_word;
+	chip->n_bytes = chip->bits_per_word / 8;
+	chip->dma_width = chip->bits_per_word / 8;
+
+	chip->speed_hz = spi->max_speed_hz;
+
+	/* Default SPI mode is SCPOL = 0, SCPH = 0 */
+	chip->cr0 = (chip->bits_per_word - 1)
+		| (chip->type << SPI_FRF_OFFSET)
+		| (spi->mode  << SPI_MODE_OFFSET);
+
+	spi_set_ctldata(spi, chip);
+	return 0;
+}
+
+static void dw_spi_cleanup(struct spi_device *spi)
+{
+	struct chip_data *chip = spi_get_ctldata(spi);
+	kfree(chip);
+}
+
+static int __devinit init_queue(struct dw_spi *dws)
+{
+	INIT_LIST_HEAD(&dws->queue);
+	spin_lock_init(&dws->lock);
+
+	dws->run = QUEUE_STOPPED;
+
+	INIT_WORK(&dws->pump_messages, pump_messages);
+	dws->workqueue = create_singlethread_workqueue(
+					dev_name(dws->master->dev.parent));
+	if (dws->workqueue == NULL)
+		return -EBUSY;
+	dws->run = QUEUE_RUNNING;
+	return 0;
+}
+
+
+int dw_spi_stop_queue(struct dw_spi *dws)
+{
+	unsigned long flags;
+	int status = 0;
+
+	spin_lock_irqsave(&dws->lock, flags);
+	dws->run = QUEUE_STOPPED;
+
+	if (!list_empty(&dws->queue))
+		status = -EBUSY;
+	spin_unlock_irqrestore(&dws->lock, flags);
+
+	return status;
+}
+EXPORT_SYMBOL_GPL(dw_spi_stop_queue);
+
+static int destroy_queue(struct dw_spi *dws)
+{
+	int status;
+
+	status = dw_spi_stop_queue(dws);
+	if (status != 0)
+		return status;
+	destroy_workqueue(dws->workqueue);
+	return 0;
+}
+
+/* Restart the controller, disable all interrupts, clean rx fifo */
+static void spi_hw_init(struct dw_spi *dws)
+{
+	dw_spi_disable(dws);
+	dw_spi_mask_intr(dws, 0xff);
+	dw_readw(dws, icr);
+	dw_spi_enable(dws);
+
+	BUG_ON(!dws->fifo_len);
+}
+
+int __devinit dw_spi_add_host(struct dw_spi *dws)
+{
+	struct spi_master *master;
+	int ret;
+
+	BUG_ON(dws == NULL);
+
+	master = spi_alloc_master(dws->parent_dev, 0);
+	if (!master) {
+		ret = -ENOMEM;
+		goto exit;
+	}
+
+	dws->master = master;
+	dws->type = SSI_MOTO_SPI;
+
+	init_completion(&dws->xfer.complete);
+	dws->dma_inited = 0;
+	/* Change to address of FIFO */
+	dws->dma_addr = (dma_addr_t)(dws->paddr + 0x60);
+
+	ret = request_threaded_irq(dws->irq, dw_spi_irq,
+				dw_spi_irq_thread_handler,
+				IRQF_SHARED, "dw_spi", dws);
+	if (ret < 0) {
+		dev_err(&master->dev, "can not get IRQ\n");
+		goto err_free_master;
+	}
+
+	master->mode_bits = SPI_CPOL | SPI_CPHA;
+	master->bus_num = dws->bus_num;
+	master->num_chipselect = dws->num_cs;
+	master->cleanup = dw_spi_cleanup;
+	master->setup = dw_spi_setup;
+	master->transfer = dw_spi_transfer;
+
+	/* Basic HW init */
+	spi_hw_init(dws);
+
+	if (dws->dma_ops && dws->dma_ops->dma_init) {
+		ret = dws->dma_ops->dma_init(dws);
+		if (ret) {
+			dev_warn(&master->dev, "DMA init failed\n");
+			dws->dma_inited = 0;
+		}
+	}
+
+	/* Initial and start queue */
+	ret = init_queue(dws);
+	if (ret) {
+		dev_err(&master->dev, "problem initializing queue\n");
+		goto err_diable_hw;
+	}
+
+	spi_master_set_devdata(master, dws);
+	ret = spi_register_master(master);
+	if (ret) {
+		dev_err(&master->dev, "problem registering spi master\n");
+		goto err_queue_alloc;
+	}
+
+	mrst_spi_debugfs_init(dws);
+	return 0;
+
+err_queue_alloc:
+	destroy_queue(dws);
+	if (dws->dma_ops && dws->dma_ops->dma_exit)
+		dws->dma_ops->dma_exit(dws);
+err_diable_hw:
+	dw_spi_disable(dws);
+	free_irq(dws->irq, dws);
+err_free_master:
+	spi_master_put(master);
+exit:
+	return ret;
+}
+EXPORT_SYMBOL_GPL(dw_spi_add_host);
+
+void __devexit dw_spi_remove_host(struct dw_spi *dws)
+{
+	int status = 0;
+
+	if (!dws)
+		return;
+	mrst_spi_debugfs_remove(dws);
+
+	/* Remove the queue */
+	status = destroy_queue(dws);
+	if (status != 0)
+		dev_err(&dws->master->dev, "dw_spi_remove: workqueue will not "
+			"complete, message memory not freed\n");
+
+	if (dws->dma_ops && dws->dma_ops->dma_exit)
+		dws->dma_ops->dma_exit(dws);
+	dw_spi_disable(dws);
+	dw_readw(dws, icr);
+	free_irq(dws->irq, dws);
+
+	/* Disconnect from the SPI framework */
+	spi_unregister_master(dws->master);
+}
+EXPORT_SYMBOL_GPL(dw_spi_remove_host);
+
+int dw_spi_suspend_host(struct dw_spi *dws)
+{
+	int ret = 0;
+
+	ret = dw_spi_stop_queue(dws);
+	if (ret)
+		return ret;
+	dw_spi_disable(dws);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(dw_spi_suspend_host);
+
+int dw_spi_resume_host(struct dw_spi *dws)
+{
+	spi_hw_init(dws);
+	dws->run = QUEUE_RUNNING;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(dw_spi_resume_host);
+
+MODULE_AUTHOR("Feng Tang <feng.tang@intel.com>");
+MODULE_DESCRIPTION("Driver for DesignWare SPI controller core");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/spi/spi-dw.h b/drivers/spi/spi-dw.h
index 97baff6..b016b85 100644
--- a/drivers/spi/spi-dw.h
+++ b/drivers/spi/spi-dw.h
@@ -7,6 +7,7 @@ 
 
 /* Bit fields in CTRLR0 */
 #define SPI_DFS_OFFSET			0
+#define SPI_DFS_MASK			0xf
 
 #define SPI_FRF_OFFSET			4
 #define SPI_FRF_SPI			0x0
@@ -17,6 +18,7 @@ 
 #define SPI_MODE_OFFSET			6
 #define SPI_SCPH_OFFSET			6
 #define SPI_SCOL_OFFSET			7
+#define SPI_MODE_MASK			(0x3 << SPI_MODE_OFFSET)
 
 #define SPI_TMOD_OFFSET			8
 #define SPI_TMOD_MASK			(0x3 << SPI_TMOD_OFFSET)
@@ -46,6 +48,7 @@ 
 #define SPI_INT_RXOI			(1 << 3)
 #define SPI_INT_RXFI			(1 << 4)
 #define SPI_INT_MSTI			(1 << 5)
+#define SPI_INT_ALL  0x3f
 
 /* TX RX interrupt level threshold, max can be 256 */
 #define SPI_INT_THRESHOLD		32
@@ -83,65 +86,67 @@  struct dw_spi;
 struct dw_spi_dma_ops {
 	int (*dma_init)(struct dw_spi *dws);
 	void (*dma_exit)(struct dw_spi *dws);
-	int (*dma_transfer)(struct dw_spi *dws, int cs_change);
+	int (*dma_transfer)(struct dw_spi *dws);
+};
+
+enum xfer_type {
+	PIO_XFER,
+	INT_XFER,
+	DMA_XFER,
+};
+
+	
+struct xfer_state {
+	const u8 *tx_buf;
+	dma_addr_t tx_dma;
+	u8 *rx_buf;
+	dma_addr_t rx_dma;
+	struct spi_message *msg;
+	u32 n_bytes;
+	u32 len;
+	u32 sent;
+	u32 rcvd;
+	u32 err;
+	u32 type;
+	u32  tx_threshold;
+	u32  rx_threshold;
+	u32 irq_status;
+	struct completion complete;
 };
 
 struct dw_spi {
 	struct spi_master	*master;
-	struct spi_device	*cur_dev;
 	struct device		*parent_dev;
 	enum dw_ssi_type	type;
 
 	void __iomem		*regs;
 	unsigned long		paddr;
 	u32			iolen;
-	int			irq;
+	u32			irq;
 	u32			fifo_len;	/* depth of the FIFO buffer */
 	u32			max_freq;	/* max bus freq supported */
 
 	u16			bus_num;
 	u16			num_cs;		/* supported slave numbers */
-
+ 
 	/* Driver message queue */
 	struct workqueue_struct	*workqueue;
 	struct work_struct	pump_messages;
 	spinlock_t		lock;
 	struct list_head	queue;
-	int			busy;
-	int			run;
-
-	/* Message Transfer pump */
-	struct tasklet_struct	pump_transfers;
+	u32			run;
 
 	/* Current message transfer state info */
-	struct spi_message	*cur_msg;
-	struct spi_transfer	*cur_transfer;
-	struct chip_data	*cur_chip;
-	struct chip_data	*prev_chip;
-	size_t			len;
-	void			*tx;
-	void			*tx_end;
-	void			*rx;
-	void			*rx_end;
-	int			dma_mapped;
-	dma_addr_t		rx_dma;
-	dma_addr_t		tx_dma;
-	size_t			rx_map_len;
-	size_t			tx_map_len;
-	u8			n_bytes;	/* current is a 1/2 bytes op */
-	u8			max_bits_per_word;	/* maxim is 16b */
-	u32			dma_width;
-	int			cs_change;
-	irqreturn_t		(*transfer_handler)(struct dw_spi *dws);
-	void			(*cs_control)(u32 command);
+	struct xfer_state       xfer;
 
 	/* Dma info */
-	int			dma_inited;
+//	int 			dma_width;
+	u32			dma_inited;
 	struct dma_chan		*txchan;
 	struct scatterlist	tx_sgl;
 	struct dma_chan		*rxchan;
 	struct scatterlist	rx_sgl;
-	int			dma_chan_done;
+	u32			dma_chan_done;
 	struct device		*dma_dev;
 	dma_addr_t		dma_addr; /* phy address of the Data register */
 	struct dw_spi_dma_ops	*dma_ops;
@@ -163,30 +168,36 @@  struct dw_spi {
 	__raw_readw(&(((struct dw_spi_reg *)dw->regs)->name))
 #define dw_writew(dw, name, val) \
 	__raw_writew((val), &(((struct dw_spi_reg *)dw->regs)->name))
+#define dw_readb(dw, name) \
+	__raw_readb(&(((struct dw_spi_reg *)dw->regs)->name))
+#define dw_writeb(dw, name, val) \
+	__raw_writeb((val), &(((struct dw_spi_reg *)dw->regs)->name))
 
-static inline void spi_enable_chip(struct dw_spi *dws, int enable)
+static inline void dw_spi_disable(struct dw_spi *dws)
 {
-	dw_writel(dws, ssienr, (enable ? 1 : 0));
+	dw_writel(dws, ssienr, 0);
 }
 
-static inline void spi_set_clk(struct dw_spi *dws, u16 div)
+static inline void dw_spi_enable(struct dw_spi *dws)
+{
+	dw_writel(dws, ssienr, 1);
+}
+
+static inline void dw_spi_set_clk(struct dw_spi *dws, u16 div)
 {
 	dw_writel(dws, baudr, div);
 }
 
-static inline void spi_chip_sel(struct dw_spi *dws, u16 cs)
+static inline void dw_spi_chip_sel(struct dw_spi *dws, u16 cs)
 {
 	if (cs > dws->num_cs)
 		return;
 
-	if (dws->cs_control)
-		dws->cs_control(1);
-
 	dw_writel(dws, ser, 1 << cs);
 }
 
 /* Disable IRQ bits */
-static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
+static inline void dw_spi_mask_intr(struct dw_spi *dws, u32 mask)
 {
 	u32 new_mask;
 
@@ -195,7 +206,7 @@  static inline void spi_mask_intr(struct dw_spi *dws, u32 mask)
 }
 
 /* Enable IRQ bits */
-static inline void spi_umask_intr(struct dw_spi *dws, u32 mask)
+static inline void dw_spi_umask_intr(struct dw_spi *dws, u32 mask)
 {
 	u32 new_mask;
 
@@ -208,6 +219,7 @@  extern void dw_spi_remove_host(struct dw_spi *dws);
 extern int dw_spi_suspend_host(struct dw_spi *dws);
 extern int dw_spi_resume_host(struct dw_spi *dws);
 extern void dw_spi_xfer_done(struct dw_spi *dws);
+extern int dw_spi_stop_queue(struct dw_spi *dws);
 
 /* platform related setup */
 extern int dw_spi_mid_init(struct dw_spi *dws); /* Intel MID platforms */