aboutsummaryrefslogtreecommitdiffstats
path: root/target/linux/coldfire/patches/007-Add-eDMA-support-for-MCF5445x.patch
diff options
context:
space:
mode:
Diffstat (limited to 'target/linux/coldfire/patches/007-Add-eDMA-support-for-MCF5445x.patch')
-rw-r--r--target/linux/coldfire/patches/007-Add-eDMA-support-for-MCF5445x.patch1639
1 files changed, 1639 insertions, 0 deletions
diff --git a/target/linux/coldfire/patches/007-Add-eDMA-support-for-MCF5445x.patch b/target/linux/coldfire/patches/007-Add-eDMA-support-for-MCF5445x.patch
new file mode 100644
index 000000000..3e9d45cfb
--- /dev/null
+++ b/target/linux/coldfire/patches/007-Add-eDMA-support-for-MCF5445x.patch
@@ -0,0 +1,1639 @@
+From cd07e01d0c6c207a6138e9d5229dfa8c88ab3879 Mon Sep 17 00:00:00 2001
+From: Alison Wang <b18965@freescale.com>
+Date: Thu, 4 Aug 2011 09:59:40 +0800
+Subject: [PATCH 07/52] Add eDMA support for MCF5445x
+
+Add MCF5445x on-chip eDMA controller driver.
+
+Signed-off-by: Alison Wang <b18965@freescale.com>
+---
+ arch/m68k/include/asm/mcf_edma.h | 246 +++++++++
+ drivers/Makefile | 2 +
+ drivers/dma/Kconfig | 15 +
+ drivers/dma/Makefile | 4 +
+ drivers/dma/mcf_edma.c | 1029 ++++++++++++++++++++++++++++++++++++++
+ drivers/dma/mcf_edma_test.c | 276 ++++++++++
+ 6 files changed, 1572 insertions(+), 0 deletions(-)
+ create mode 100644 arch/m68k/include/asm/mcf_edma.h
+ create mode 100644 drivers/dma/mcf_edma.c
+ create mode 100644 drivers/dma/mcf_edma_test.c
+
+--- /dev/null
++++ b/arch/m68k/include/asm/mcf_edma.h
+@@ -0,0 +1,246 @@
++/*
++ * mcf_edma.h - Coldfire eDMA driver header file.
++ *
++ * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. All Rights Reserved.
++ *
++ * Add support for m5441x platform (Lanttor.Guo@freescale.com)
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License as published by the
++ * Free Software Foundation; either version 2 of the License, or (at your
++ * option) any later version.
++ */
++
++#ifndef _MCF_EDMA_H
++#define _MCF_EDMA_H
++
++#include <asm/mcfsim.h>
++#include <linux/interrupt.h>
++#if defined(CONFIG_M5445X)
++#include <asm/mcf5445x_edma.h>
++#elif defined(CONFIG_M5441X)
++#include <asm/mcf5441x_edma.h>
++#endif
++#include <linux/scatterlist.h>
++
++#define MCF_EDMA_INT0_CHANNEL_BASE (8)
++#define MCF_EDMA_INT0_CONTROLLER_BASE (64)
++#define MCF_EDMA_INT0_BASE (MCF_EDMA_INT0_CHANNEL_BASE + \
++ MCF_EDMA_INT0_CONTROLLER_BASE)
++#define MCF_EDMA_INT0_NUM (16)
++#define MCF_EDMA_INT0_END (MCF_EDMA_INT0_NUM)
++
++#if defined(CONFIG_M5441X)
++#define MCF_EDMA_INT1_CHANNEL_BASE (8)
++#define MCF_EDMA_INT1_CONTROLLER_BASE (128)
++#define MCF_EDMA_INT1_BASE (MCF_EDMA_INT1_CHANNEL_BASE + \
++ MCF_EDMA_INT1_CONTROLLER_BASE)
++#define MCF_EDMA_INT1_NUM (40)
++#define MCF_EDMA_INT1_END (MCF_EDMA_INT0_END + MCF_EDMA_INT1_NUM)
++
++#define MCF_EDMA_INT2_CHANNEL_BASE (0)
++#define MCF_EDMA_INT2_CONTROLLER_BASE (192)
++#define MCF_EDMA_INT2_BASE (MCF_EDMA_INT2_CHANNEL_BASE + \
++ MCF_EDMA_INT2_CONTROLLER_BASE)
++#define MCF_EDMA_INT2_NUM (8)
++#define MCF_EDMA_INT2_END (MCF_EDMA_INT1_END + MCF_EDMA_INT2_NUM)
++
++#endif
++
++#if defined(CONFIG_M5445X)
++#define MCF_EDMA_CHANNELS (16) /* 0-15 */
++#elif defined(CONFIG_M5441X)
++#define MCF_EDMA_CHANNELS (64) /* 0-63 */
++#endif
++
++#define MCF_EDMA_CHANNEL_ANY (0xFF)
++#define MCF_EDMA_INT_ERR (16) /* edma error interrupt */
++
++#define MCF_EDMA_TCD_PER_CHAN 256
++
++#ifdef CONFIG_M54455
++/* eDMA engine TCD memory description */
++
++struct TCD {
++ u32 saddr;
++ u16 attr;
++ u16 soff;
++ u32 nbytes;
++ u32 slast;
++ u32 daddr;
++ u16 citer;
++ u16 doff;
++ u32 dlast_sga;
++ u16 biter;
++ u16 csr;
++} __packed;
++
++struct fsl_edma_requestbuf {
++ dma_addr_t saddr;
++ dma_addr_t daddr;
++ u32 soff;
++ u32 doff;
++ u32 attr;
++ u32 minor_loop;
++ u32 len;
++};
++
++/*
++ * config the eDMA to use the TCD sg feature
++ *
++ * @channel: which channel. in fact this function is designed to satisfy
++ * the ATA driver TCD SG need, i.e. by now it is a special
++ * func, because it need prev alloc channel TCD physical memory
++ * first, we add the ATA's in the eDMA init only
++ * @buf: buffer array to fill the TCDs
++ * @nents: the size of the buf
++ */
++void mcf_edma_sg_config(int channel, struct fsl_edma_requestbuf *buf,
++ int nents);
++
++/*
++ * The zero-copy version of mcf_edma_sg_config()
++ */
++void mcf_edma_sglist_config(int channel, struct scatterlist *sgl, int n_elem,
++ int dma_dir, u32 addr, u32 attr,
++ u32 soff, u32 doff, u32 nbytes);
++#endif
++
++/* Setup transfer control descriptor (TCD)
++ * channel - descriptor number
++ * source - source address
++ * dest - destination address
++ * attr - attributes
++ * soff - source offset
++ * nbytes - number of bytes to be transfered in minor loop
++ * slast - last source address adjustment
++ * citer - major loop count
++ * biter - begining minor loop count
++ * doff - destination offset
++ * dlast_sga - last destination address adjustment
++ * major_int - generate interrupt after each major loop
++ * disable_req - disable DMA request after major loop
++ */
++void mcf_edma_set_tcd_params(int channel, u32 source, u32 dest,
++ u32 attr, u32 soff, u32 nbytes, u32 slast,
++ u32 citer, u32 biter, u32 doff, u32 dlast_sga,
++ int major_int, int disable_req);
++
++/* Setup transfer control descriptor (TCD) and enable halfway irq
++ * channel - descriptor number
++ * source - source address
++ * dest - destination address
++ * attr - attributes
++ * soff - source offset
++ * nbytes - number of bytes to be transfered in minor loop
++ * slast - last source address adjustment
++ * biter - major loop count
++ * doff - destination offset
++ * dlast_sga - last destination address adjustment
++ * disable_req - disable DMA request after major loop
++ */
++void mcf_edma_set_tcd_params_halfirq(int channel, u32 source, u32 dest,
++ u32 attr, u32 soff, u32 nbytes, u32 slast,
++ u32 biter, u32 doff, u32 dlast_sga,
++ int disable_req);
++
++/* check if dma is done
++ * channel - descriptor number
++ * return 1 if done
++ */
++int mcf_edma_check_done(int channel);
++
++/* Starts eDMA transfer on specified channel
++ * channel - eDMA TCD number
++ */
++static inline void
++mcf_edma_start_transfer(int channel)
++{
++ MCF_EDMA_SERQ = channel;
++ MCF_EDMA_SSRT = channel;
++}
++
++/* Restart eDMA transfer from halfirq
++ * channel - eDMA TCD number
++ */
++static inline void
++mcf_edma_confirm_halfirq(int channel)
++{
++ /*MCF_EDMA_TCD_CSR(channel) = 7;*/
++ MCF_EDMA_SSRT = channel;
++}
++
++/* Starts eDMA transfer on specified channel based on peripheral request
++ * channel - eDMA TCD number
++ */
++static inline void mcf_edma_enable_transfer(int channel)
++{
++ MCF_EDMA_SERQ = channel;
++}
++
++
++/* Stops eDMA transfer
++ * channel - eDMA TCD number
++ */
++static inline void
++mcf_edma_stop_transfer(int channel)
++{
++ MCF_EDMA_CINT = channel;
++ MCF_EDMA_CERQ = channel;
++}
++
++/* Confirm that interrupt has been handled
++ * channel - eDMA TCD number
++ */
++static inline void
++mcf_edma_confirm_interrupt_handled(int channel)
++{
++ MCF_EDMA_CINT = channel;
++}
++
++/**
++ * mcf_edma_request_channel - Request an eDMA channel
++ * @channel: channel number. In case it is equal to EDMA_CHANNEL_ANY
++ * it will be allocated a first free eDMA channel.
++ * @handler: dma handler
++ * @error_handler: dma error handler
++ * @irq_level: irq level for the dma handler
++ * @arg: argument to pass back
++ * @lock: optional spinlock to hold over interrupt
++ * @device_id: device id
++ *
++ * Returns allocatedd channel number if success or
++ * a negative value if failure.
++ */
++int mcf_edma_request_channel(int channel,
++ irqreturn_t(*handler) (int, void *),
++ void (*error_handler) (int, void *),
++ u8 irq_level,
++ void *arg,
++ spinlock_t *lock, const char *device_id);
++
++/**
++ * Update the channel callback/arg
++ * @channel: channel number
++ * @handler: dma handler
++ * @error_handler: dma error handler
++ * @arg: argument to pass back
++ *
++ * Returns 0 if success or a negative value if failure
++ */
++int mcf_edma_set_callback(int channel,
++ irqreturn_t(*handler) (int, void *),
++ void (*error_handler) (int, void *), void *arg);
++
++/**
++ * Free the edma channel
++ * @channel: channel number
++ * @arg: argument created with
++ *
++ * Returns 0 if success or a negative value if failure
++ */
++int mcf_edma_free_channel(int channel, void *arg);
++
++void mcf_edma_dump_channel(int channel);
++
++#endif /* _MCF_EDMA_H */
+--- a/drivers/Makefile
++++ b/drivers/Makefile
+@@ -39,6 +39,7 @@ obj-$(CONFIG_FB_I810) += video
+ obj-$(CONFIG_FB_INTEL) += video/intelfb/
+
+ obj-$(CONFIG_PARPORT) += parport/
++obj-$(CONFIG_COLDFIRE_EDMA) += dma/
+ obj-y += base/ block/ misc/ mfd/ nfc/
+ obj-$(CONFIG_NUBUS) += nubus/
+ obj-y += macintosh/
+@@ -114,6 +115,7 @@ obj-$(CONFIG_BCMA) += bcma/
+ obj-$(CONFIG_VHOST_NET) += vhost/
+ obj-$(CONFIG_VLYNQ) += vlynq/
+ obj-$(CONFIG_STAGING) += staging/
++obj-$(CONFIG_MCD_DMA) += dma/
+ obj-y += platform/
+ obj-y += ieee802154/
+ #common clk code
+--- a/drivers/dma/Kconfig
++++ b/drivers/dma/Kconfig
+@@ -114,6 +114,21 @@ config MPC512X_DMA
+ ---help---
+ Enable support for the Freescale MPC512x built-in DMA engine.
+
++config COLDFIRE_EDMA
++ tristate "Coldfire eDMA support"
++ default y
++ depends on COLDFIRE && (M5445X || M5441X)
++ help
++ Enable support for Coldfire eDMA controller. For example
++ used by Coldfire SSI Audio device driver.
++
++config COLDFIRE_EDMA_TEST
++ tristate "Coldfire eDMA simple test module"
++ default m
++ depends on COLDFIRE_EDMA
++ help
++ This is simple eDMA test module.
++
+ config MV_XOR
+ bool "Marvell XOR engine support"
+ depends on PLAT_ORION
+--- a/drivers/dma/Makefile
++++ b/drivers/dma/Makefile
+@@ -13,10 +13,14 @@ obj-$(CONFIG_INTEL_IOATDMA) += ioat/
+ obj-$(CONFIG_INTEL_IOP_ADMA) += iop-adma.o
+ obj-$(CONFIG_FSL_DMA) += fsldma.o
+ obj-$(CONFIG_MPC512X_DMA) += mpc512x_dma.o
++obj-$(CONFIG_COLDFIRE_EDMA) += mcf_edma.o
++obj-$(CONFIG_COLDFIRE_EDMA_TEST) += mcf_edma_test.o
+ obj-$(CONFIG_MV_XOR) += mv_xor.o
+ obj-$(CONFIG_DW_DMAC) += dw_dmac.o
+ obj-$(CONFIG_AT_HDMAC) += at_hdmac.o
+ obj-$(CONFIG_MX3_IPU) += ipu/
++obj-$(CONFIG_MCD_DMA) += mcddma.o
++mcddma-objs := MCD_dmaApi.o MCD_tasks.o MCD_tasksInit.o
+ obj-$(CONFIG_TXX9_DMAC) += txx9dmac.o
+ obj-$(CONFIG_SH_DMAE) += shdma.o
+ obj-$(CONFIG_COH901318) += coh901318.o coh901318_lli.o
+--- /dev/null
++++ b/drivers/dma/mcf_edma.c
+@@ -0,0 +1,1029 @@
++/*
++ * mcf_edma.c - eDMA driver for Coldfire.
++ *
++ * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. All Rights Reserved.
++ * Author: Andrey Butok
++ * Yaroslav Vinogradov
++ * Lanttor.Guo@freescale.com add m5441x platform support.
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License as published by the
++ * Free Software Foundation; either version 2 of the License, or (at your
++ * option) any later version.
++ *
++ * This program is distributed in the hope that 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ ***************************************************************************
++ * Changes:
++ * v0.002 29 February 2008 Andrey Butok, Freescale Semiconductor
++ * Added support of atomatic channel allocation from the
++ * channel pool.
++ * v0.001 12 February 2008 Andrey Butok
++ * Initial Release - developed on uClinux with 2.6.23 kernel.
++ * Based on coldfire_edma.c code
++ * of Yaroslav Vinogradov (Freescale Semiconductor, Inc.)
++ *
++ * NOTE: This driver was tested on MCF52277 platform.
++ * It should also work on other Coldfire platdorms with eDMA module.
++ *
++ * TBD: Try to make it more general.
++ * Try to integrate with current <asm/dma.h> <kernel/dma.c> API
++ * or use Intel DMA API
++ */
++
++#include <linux/dma-mapping.h>
++#include <asm/mcf_edma.h>
++#include <linux/init.h>
++#include <linux/module.h>
++#include <asm/coldfire.h>
++#include <linux/fs.h>
++#include <linux/cdev.h>
++#include <linux/seq_file.h>
++#include <linux/proc_fs.h>
++
++/* Please add here processors that were tested with this driver */
++#if !defined(CONFIG_M5227x) && !defined(CONFIG_M5445X) && \
++ !defined(CONFIG_M5441X)
++#error "The driver is not tested/designed for your processor!"
++#endif
++
++#define MCF_EDMA_DRIVER_VERSION "Revision: 0.003"
++#define MCF_EDMA_DRIVER_AUTHOR "Freescale Semiconductor Inc, Andrey Butok"
++#define MCF_EDMA_DRIVER_DESC "Coldfire EDMA driver."
++#define MCF_EDMA_DRIVER_INFO DRIVER_VERSION " " DRIVER_DESC
++#define MCF_EDMA_DRIVER_LICENSE "GPL"
++#define MCF_EDMA_DRIVER_NAME "mcf_edma"
++
++#define MCF_EDMA_DEV_MINOR (1)
++
++#undef MCF_EDMA_DEBUG
++
++#ifdef MCF_EDMA_DEBUG
++#define DBG(fmt, args...) printk(KERN_INFO "[%s] " fmt, \
++ __func__, ## args)
++#else
++#define DBG(fmt, args...) do {} while (0)
++#endif
++
++#define ERR(format, arg...) printk(KERN_ERR "%s:%s: " format "\n", \
++ __FILE__, __func__ , ## arg)
++#define INFO(stuff...) printk(KERN_INFO MCF_EDMA_DRIVER_NAME \
++ ": " stuff)
++
++/* DMA channel pool used for atomtic channel allocation.
++ * You can edit this list. First candidates are "Not used/Reserved" channels */
++u8 mcf_edma_channel_pool[] = { 1, /* Not used */
++ 0, /* External DMA request */
++ 5, /* UART1 Receive */
++ 6, /* UART1 Transmit */
++ 7, /* UART2 Receive */
++ 8, /* UART2 Transmit */
++#if defined(CONFIG_M5441X)
++ 16,
++ 55,
++ 56,
++ 63,
++#endif
++};
++
++/*
++ * Callback handler data for each TCD
++ */
++struct mcf_edma_isr_record {
++ irqreturn_t(*irq_handler) (int, void *); /* interrupt handler */
++ void (*error_handler) (int, void *); /* error interrupt handler */
++ void *arg; /* argument to pass back */
++ int allocated; /* busy flag */
++ spinlock_t *lock; /* spin lock (optional) */
++ const char *device_id; /* dev id string, used in procfs */
++};
++
++/*
++ * Device structure
++ */
++struct mcf_edma_dev {
++ struct cdev cdev; /* character device */
++ struct mcf_edma_isr_record dma_interrupt_handlers[MCF_EDMA_CHANNELS];
++};
++
++/* allocated major device number */
++static int mcf_edma_major;
++
++/* device driver structure */
++static struct mcf_edma_dev *mcf_edma_devp;
++
++#ifdef CONFIG_M54455
++/* PATA controller structure */
++static struct {
++ struct TCD *pata_tcd_va;
++ dma_addr_t pata_tcd_pa;
++} fsl_pata_dma_tcd;
++#endif
++
++/* device driver file operations */
++const struct file_operations mcf_edma_fops = {
++ .owner = THIS_MODULE,
++};
++
++/**
++ * mcf_edma_isr - eDMA channel interrupt handler
++ * @irq: interrupt number
++ * @dev_id: argument
++ */
++static irqreturn_t
++mcf_edma_isr(int irq, void *dev_id)
++{
++ int channel = -1;
++ int result = IRQ_HANDLED;
++
++#if defined(CONFIG_M5445X)
++ channel = irq - MCF_EDMA_INT0_BASE;
++#elif defined(CONFIG_M5441X)
++ if (irq >= MCF_EDMA_INT0_BASE &&
++ irq < MCF_EDMA_INT0_BASE + MCF_EDMA_INT0_NUM)
++ channel = irq - MCF_EDMA_INT0_BASE;
++ else if (irq >= MCF_EDMA_INT1_BASE &&
++ irq < MCF_EDMA_INT1_BASE + MCF_EDMA_INT1_NUM)
++ channel = irq - MCF_EDMA_INT1_BASE + MCF_EDMA_INT0_END;
++ else if (irq == MCF_EDMA_INT2_BASE &&
++ irq < MCF_EDMA_INT2_BASE + MCF_EDMA_INT2_NUM) {
++ int i;
++ for (i = 0; i < MCF_EDMA_INT2_NUM; i++) {
++ if ((MCF_EDMA_INTH >> 24) & (0x1 << i)) {
++ channel = irq - MCF_EDMA_INT2_BASE +
++ MCF_EDMA_INT1_END + i;
++ break;
++ }
++ }
++ } else {
++ ERR("Bad irq number at isr!\n");
++ return result;
++ }
++#endif
++
++ DBG("\n");
++
++ if ((mcf_edma_devp != NULL) &&
++ (mcf_edma_devp->dma_interrupt_handlers[channel].irq_handler)) {
++ /* call user irq handler */
++ if (mcf_edma_devp->dma_interrupt_handlers[channel].lock)
++ spin_lock(mcf_edma_devp->
++ dma_interrupt_handlers[channel].lock);
++
++ result =
++ mcf_edma_devp->dma_interrupt_handlers[channel].
++ irq_handler(channel,
++ mcf_edma_devp->dma_interrupt_handlers[channel].
++ arg);
++
++ if (mcf_edma_devp->dma_interrupt_handlers[channel].lock)
++ spin_unlock(mcf_edma_devp->
++ dma_interrupt_handlers[channel].lock);
++ } else {
++ /* no irq handler so just ack it */
++ mcf_edma_confirm_interrupt_handled(channel);
++ ERR(" No handler for DMA channel (%d)\n", channel);
++ }
++
++ return result;
++}
++
++/**
++ * mcf_edma_error_isr - eDMA error interrupt handler
++ * @irq: interrupt number
++ * @dev_id: argument
++ */
++static irqreturn_t
++mcf_edma_error_isr(int irq, void *dev_id)
++{
++ int i;
++
++#if defined(CONFIG_M5445X)
++ u16 err;
++
++ err = MCF_EDMA_ERR;
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ if (err & (1 << i)) {
++ if (mcf_edma_devp != NULL &&
++ mcf_edma_devp->dma_interrupt_handlers[i].
++ error_handler)
++ mcf_edma_devp->dma_interrupt_handlers[i].
++ error_handler(i,
++ mcf_edma_devp->
++ dma_interrupt_handlers[i].
++ arg);
++ else
++ ERR(" DMA error on channel (%d)\n", i);
++ }
++ }
++#elif defined(CONFIG_M5441X)
++ u32 errl, errh;
++
++ errl = MCF_EDMA_ERRL;
++ errh = MCF_EDMA_ERRH;
++
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ if ((errl & (1 << i)) || (errh & (1 << (i - 32)))) {
++ if (mcf_edma_devp != NULL &&
++ mcf_edma_devp->dma_interrupt_handlers[i].
++ error_handler)
++ mcf_edma_devp->dma_interrupt_handlers[i].
++ error_handler(i, mcf_edma_devp->
++ dma_interrupt_handlers[i].arg);
++ else
++ ERR(" DMA error on channel (%d)\n", i);
++ }
++ }
++#endif
++ MCF_EDMA_CERR = MCF_EDMA_CERR_CAER;
++ return IRQ_HANDLED;
++}
++
++/**
++ * mcf_edma_check_done - Check if channel is finished or not
++ * @channel: channel number
++ * return: 0 if not done yet
++ */
++int
++mcf_edma_check_done(int channel)
++{
++ if (channel < 0 || channel > MCF_EDMA_CHANNELS)
++ return 1;
++
++ return MCF_EDMA_TCD_CSR(channel) & MCF_EDMA_TCD_CSR_DONE;
++}
++EXPORT_SYMBOL(mcf_edma_check_done);
++
++/**
++ * mcf_edma_set_tcd_params - Set transfer control descriptor (TCD)
++ * @channel: channel number
++ * @source: source address
++ * @dest: destination address
++ * @attr: attributes
++ * @soff: source offset
++ * @nbytes: number of bytes to be transfered in minor loop
++ * @slast: last source address adjustment
++ * @citer: major loop count
++ * @biter: beginning major loop count
++ * @doff: destination offset
++ * @dlast_sga: last destination address adjustment
++ * @major_int: generate interrupt after each major loop
++ * @disable_req: disable DMA request after major loop
++ */
++void
++mcf_edma_set_tcd_params(int channel, u32 source, u32 dest,
++ u32 attr, u32 soff, u32 nbytes, u32 slast,
++ u32 citer, u32 biter, u32 doff, u32 dlast_sga,
++ int major_int, int disable_req)
++{
++ DBG("(%d)\n", channel);
++
++ if (channel < 0 || channel > MCF_EDMA_CHANNELS)
++ return;
++
++ MCF_EDMA_TCD_SADDR(channel) = source;
++ MCF_EDMA_TCD_DADDR(channel) = dest;
++ MCF_EDMA_TCD_ATTR(channel) = attr;
++ MCF_EDMA_TCD_SOFF(channel) = MCF_EDMA_TCD_SOFF_SOFF(soff);
++ MCF_EDMA_TCD_NBYTES(channel) = MCF_EDMA_TCD_NBYTES_NBYTES(nbytes);
++ MCF_EDMA_TCD_SLAST(channel) = MCF_EDMA_TCD_SLAST_SLAST(slast);
++ MCF_EDMA_TCD_CITER(channel) = MCF_EDMA_TCD_CITER_CITER(citer);
++ MCF_EDMA_TCD_BITER(channel) = MCF_EDMA_TCD_BITER_BITER(biter);
++ MCF_EDMA_TCD_DOFF(channel) = MCF_EDMA_TCD_DOFF_DOFF(doff);
++ MCF_EDMA_TCD_DLAST_SGA(channel) =
++ MCF_EDMA_TCD_DLAST_SGA_DLAST_SGA(dlast_sga);
++ MCF_EDMA_TCD_CSR(channel) = 0x0000;
++
++ /* interrupt at the end of major loop */
++ if (major_int)
++ MCF_EDMA_TCD_CSR(channel) |= MCF_EDMA_TCD_CSR_INT_MAJOR;
++ else
++ MCF_EDMA_TCD_CSR(channel) &= ~MCF_EDMA_TCD_CSR_INT_MAJOR;
++
++ /* disable request at the end of major loop of transfer or not */
++ if (disable_req)
++ MCF_EDMA_TCD_CSR(channel) |= MCF_EDMA_TCD_CSR_D_REQ;
++ else
++ MCF_EDMA_TCD_CSR(channel) &= ~MCF_EDMA_TCD_CSR_D_REQ;
++
++ /* enable error interrupt */
++ MCF_EDMA_SEEI = MCF_EDMA_SEEI_SEEI(channel);
++}
++EXPORT_SYMBOL(mcf_edma_set_tcd_params);
++#ifdef CONFIG_M54455
++/**
++ * mcf_edma_sg_config - config an eDMA channel to use the S/G tcd feature
++ * @channel: channel number
++ * @buf: the array of tcd sg
++ * @nents: number of tcd sg array, the max is 256 set but can modify
++ *
++ * limitation:
++ * currently this function is only for PATA RX/TX on MCF54455,
++ * so eDMA init does not allocate TCD memory for other memory
++ *
++ * TODO:
++ * any one who need this feature shoule add his own TCD memory init
++ */
++void mcf_edma_sg_config(int channel, struct fsl_edma_requestbuf *buf,
++ int nents)
++{
++ struct TCD *vtcd = (struct TCD *)fsl_pata_dma_tcd.pata_tcd_va;
++ u32 ptcd = fsl_pata_dma_tcd.pata_tcd_pa;
++ struct fsl_edma_requestbuf *pb = buf;
++ int i;
++
++ if (channel < MCF_EDMA_CHAN_ATA_RX || channel > MCF_EDMA_CHAN_ATA_TX) {
++ printk(KERN_ERR "mcf edma sg config err, not support\n");
++ return;
++ }
++ if (nents > MCF_EDMA_TCD_PER_CHAN) {
++ printk(KERN_ERR "Too many SGs, please confirm.%d > %d\n",
++ nents, MCF_EDMA_TCD_PER_CHAN);
++ return;
++ }
++
++ /* build our tcd sg array */
++ for (i = 0; i < nents; i++) {
++ memset(vtcd, 0 , sizeof(struct TCD));
++ vtcd->saddr = pb->saddr;
++ vtcd->daddr = pb->daddr;
++ vtcd->attr = pb->attr;
++ vtcd->soff = pb->soff;
++ vtcd->doff = pb->doff;
++ vtcd->nbytes = pb->minor_loop;
++ vtcd->citer = vtcd->biter = pb->len/pb->minor_loop;
++
++ if (i != nents - 1) {
++ vtcd->csr |= MCF_EDMA_TCD_CSR_E_SG;/* we are tcd sg */
++ vtcd->dlast_sga =
++ (u32)(ptcd + (i + 1)*sizeof(struct TCD));
++ } else {
++ /*this is the last sg, so enable the major int*/
++ vtcd->csr |= MCF_EDMA_TCD_CSR_INT_MAJOR
++ |MCF_EDMA_TCD_CSR_D_REQ;
++ }
++ pb++;
++ vtcd++;
++ }
++
++ /* Now setup the firset TCD for this sg to the edma enginee */
++ vtcd = fsl_pata_dma_tcd.pata_tcd_va;
++
++ MCF_EDMA_TCD_CSR(channel) = 0x0000;
++ MCF_EDMA_TCD_SADDR(channel) = vtcd->saddr;
++ MCF_EDMA_TCD_DADDR(channel) = vtcd->daddr;
++ MCF_EDMA_TCD_ATTR(channel) = vtcd->attr;
++ MCF_EDMA_TCD_SOFF(channel) = MCF_EDMA_TCD_SOFF_SOFF(vtcd->soff);
++ MCF_EDMA_TCD_NBYTES(channel) = MCF_EDMA_TCD_NBYTES_NBYTES(vtcd->nbytes);
++ MCF_EDMA_TCD_SLAST(channel) = MCF_EDMA_TCD_SLAST_SLAST(vtcd->slast);
++ MCF_EDMA_TCD_CITER(channel) = MCF_EDMA_TCD_CITER_CITER(vtcd->citer);
++ MCF_EDMA_TCD_BITER(channel) = MCF_EDMA_TCD_BITER_BITER(vtcd->biter);
++ MCF_EDMA_TCD_DOFF(channel) = MCF_EDMA_TCD_DOFF_DOFF(vtcd->doff);
++ MCF_EDMA_TCD_DLAST_SGA(channel) =
++ MCF_EDMA_TCD_DLAST_SGA_DLAST_SGA(vtcd->dlast_sga);
++ MCF_EDMA_TCD_CSR(channel) |= vtcd->csr;
++}
++EXPORT_SYMBOL(mcf_edma_sg_config);
++
++/**
++ * The zero-copy version of mcf_edma_sg_config
++ * dma_dir : indicate teh addr direction
++ */
++void mcf_edma_sglist_config(int channel, struct scatterlist *sgl, int n_elem,
++ int dma_dir, u32 addr, u32 attr,
++ u32 soff, u32 doff, u32 nbytes)
++{
++ struct TCD *vtcd = (struct TCD *)fsl_pata_dma_tcd.pata_tcd_va;
++ u32 ptcd = fsl_pata_dma_tcd.pata_tcd_pa;
++ struct scatterlist *sg;
++ u32 si;
++
++ if (channel < MCF_EDMA_CHAN_ATA_RX || channel > MCF_EDMA_CHAN_ATA_TX) {
++ printk(KERN_ERR "mcf edma sg config err, not support\n");
++ return;
++ }
++ if (n_elem > MCF_EDMA_TCD_PER_CHAN) {
++ printk(KERN_ERR "Too many SGs, please confirm.%d > %d\n",
++ n_elem, MCF_EDMA_TCD_PER_CHAN);
++ return;
++ }
++
++ /* build our tcd sg array */
++ if (dma_dir == DMA_TO_DEVICE) { /* write */
++ for_each_sg(sgl, sg, n_elem, si) {
++ memset(vtcd, 0 , sizeof(struct TCD));
++ vtcd->saddr = sg_dma_address(sg);
++ vtcd->daddr = addr;
++ vtcd->attr = attr;
++ vtcd->soff = soff;
++ vtcd->doff = doff;
++ vtcd->nbytes = nbytes;
++ vtcd->citer = vtcd->biter = sg_dma_len(sg)/nbytes;
++
++ if (si != n_elem - 1) {
++ /* we are tcd sg */
++ vtcd->csr |= MCF_EDMA_TCD_CSR_E_SG;
++ vtcd->dlast_sga = (u32)(ptcd + (si + 1) * \
++ sizeof(struct TCD));
++ } else {
++ /*this is the last sg, so enable the major int*/
++ vtcd->csr |= MCF_EDMA_TCD_CSR_INT_MAJOR
++ |MCF_EDMA_TCD_CSR_D_REQ;
++ }
++ vtcd++;
++ }
++ } else {
++ for_each_sg(sgl, sg, n_elem, si) {
++ memset(vtcd, 0 , sizeof(struct TCD));
++ vtcd->daddr = sg_dma_address(sg);
++ vtcd->saddr = addr;
++ vtcd->attr = attr;
++ vtcd->soff = soff;
++ vtcd->doff = doff;
++ vtcd->nbytes = nbytes;
++ vtcd->citer = vtcd->biter = sg_dma_len(sg)/nbytes;
++
++ if (si != n_elem - 1) {
++ /* we are tcd sg */
++ vtcd->csr |= MCF_EDMA_TCD_CSR_E_SG;
++ vtcd->dlast_sga = (u32)(ptcd + (si + 1) * \
++ sizeof(struct TCD));
++ } else {
++ /*this is the last sg, so enable the major int*/
++ vtcd->csr |= MCF_EDMA_TCD_CSR_INT_MAJOR
++ |MCF_EDMA_TCD_CSR_D_REQ;
++ }
++ vtcd++;
++ }
++ }
++
++ /* Now setup the firset TCD for this sg to the edma enginee */
++ vtcd = fsl_pata_dma_tcd.pata_tcd_va;
++
++ MCF_EDMA_TCD_CSR(channel) = 0x0000;
++ MCF_EDMA_TCD_SADDR(channel) = vtcd->saddr;
++ MCF_EDMA_TCD_DADDR(channel) = vtcd->daddr;
++ MCF_EDMA_TCD_ATTR(channel) = vtcd->attr;
++ MCF_EDMA_TCD_SOFF(channel) = MCF_EDMA_TCD_SOFF_SOFF(vtcd->soff);
++ MCF_EDMA_TCD_NBYTES(channel) = MCF_EDMA_TCD_NBYTES_NBYTES(vtcd->nbytes);
++ MCF_EDMA_TCD_SLAST(channel) = MCF_EDMA_TCD_SLAST_SLAST(vtcd->slast);
++ MCF_EDMA_TCD_CITER(channel) = MCF_EDMA_TCD_CITER_CITER(vtcd->citer);
++ MCF_EDMA_TCD_BITER(channel) = MCF_EDMA_TCD_BITER_BITER(vtcd->biter);
++ MCF_EDMA_TCD_DOFF(channel) = MCF_EDMA_TCD_DOFF_DOFF(vtcd->doff);
++ MCF_EDMA_TCD_DLAST_SGA(channel) =
++ MCF_EDMA_TCD_DLAST_SGA_DLAST_SGA(vtcd->dlast_sga);
++
++ MCF_EDMA_TCD_CSR(channel) |= vtcd->csr;
++}
++EXPORT_SYMBOL(mcf_edma_sglist_config);
++#endif
++/**
++ * mcf_edma_set_tcd_params_halfirq - Set TCD AND enable half irq
++ * @channel: channel number
++ * @source: source address
++ * @dest: destination address
++ * @attr: attributes
++ * @soff: source offset
++ * @nbytes: number of bytes to be transfered in minor loop
++ * @slast: last source address adjustment
++ * @biter: beginning major loop count
++ * @doff: destination offset
++ * @dlast_sga: last destination address adjustment
++ * @disable_req: disable DMA request after major loop
++ */
++void
++mcf_edma_set_tcd_params_halfirq(int channel, u32 source, u32 dest,
++ u32 attr, u32 soff, u32 nbytes, u32 slast,
++ u32 biter, u32 doff, u32 dlast_sga,
++ int disable_req)
++{
++ DBG("(%d)\n", channel);
++
++ if (channel < 0 || channel > MCF_EDMA_CHANNELS)
++ return;
++
++ mcf_edma_set_tcd_params(channel, source, dest,
++ attr, soff, nbytes, slast,
++ biter, biter, doff, dlast_sga,
++ 1/*0*/, disable_req);
++
++ if (biter < 2)
++ printk(KERN_ERR "MCF_EDMA: Request for halfway irq denied\n");
++
++ /* interrupt midway through major loop */
++ MCF_EDMA_TCD_CSR(channel) |= MCF_EDMA_TCD_CSR_INT_HALF;
++}
++EXPORT_SYMBOL(mcf_edma_set_tcd_params_halfirq);
++
++/**
++ * mcf_edma_request_channel - Request an eDMA channel
++ * @channel: channel number. In case it is equal to EDMA_CHANNEL_ANY
++ * it will be allocated a first free eDMA channel.
++ * @handler: dma handler
++ * @error_handler: dma error handler
++ * @irq_level: irq level for the dma handler
++ * @arg: argument to pass back
++ * @lock: optional spinlock to hold over interrupt
++ * @device_id: device id
++ *
++ * Returns allocatedd channel number if success or
++ * a negative value if failure.
++ */
++int
++mcf_edma_request_channel(int channel,
++ irqreturn_t(*handler) (int, void *),
++ void (*error_handler) (int, void *),
++ u8 irq_level,
++ void *arg, spinlock_t *lock, const char *device_id)
++{
++ DBG("\n channel=%d\n", channel);
++
++ if (mcf_edma_devp != NULL
++ && ((channel >= 0 && channel <= MCF_EDMA_CHANNELS)
++ || (channel == MCF_EDMA_CHANNEL_ANY))) {
++ if (channel == MCF_EDMA_CHANNEL_ANY) {
++ int i;
++ for (i = 0; i < sizeof(mcf_edma_channel_pool); i++) {
++ if (mcf_edma_devp->dma_interrupt_handlers
++ [mcf_edma_channel_pool[i]].allocated ==
++ 0) {
++ channel = mcf_edma_channel_pool[i];
++ break;
++ }
++ };
++ if (channel == MCF_EDMA_CHANNEL_ANY)
++ return -EBUSY;
++ } else {
++ if (mcf_edma_devp->dma_interrupt_handlers[channel].
++ allocated)
++ return -EBUSY;
++ }
++
++ mcf_edma_devp->dma_interrupt_handlers[channel].allocated = 1;
++ mcf_edma_devp->dma_interrupt_handlers[channel].irq_handler =
++ handler;
++ mcf_edma_devp->dma_interrupt_handlers[channel].error_handler =
++ error_handler;
++ mcf_edma_devp->dma_interrupt_handlers[channel].arg = arg;
++ mcf_edma_devp->dma_interrupt_handlers[channel].lock = lock;
++ mcf_edma_devp->dma_interrupt_handlers[channel].device_id =
++ device_id;
++
++ /* Initalize interrupt controller to allow eDMA interrupts */
++#if defined(CONFIG_M5445X)
++ MCF_INTC0_ICR(MCF_EDMA_INT0_CHANNEL_BASE + channel) = irq_level;
++ MCF_INTC0_CIMR = MCF_EDMA_INT0_CHANNEL_BASE + channel;
++#elif defined(CONFIG_M5441X)
++ if (channel >= 0 && channel < MCF_EDMA_INT0_END) {
++ MCF_INTC0_ICR(MCF_EDMA_INT0_CHANNEL_BASE + channel) =
++ irq_level;
++ MCF_INTC0_CIMR = MCF_EDMA_INT0_CHANNEL_BASE + channel;
++ } else if (channel >= MCF_EDMA_INT0_END &&
++ channel < MCF_EDMA_INT1_END) {
++ MCF_INTC1_ICR(MCF_EDMA_INT1_CHANNEL_BASE +
++ (channel - MCF_EDMA_INT0_END)) = irq_level;
++ MCF_INTC1_CIMR = MCF_EDMA_INT1_CHANNEL_BASE +
++ (channel - MCF_EDMA_INT0_END);
++ } else if (channel >= MCF_EDMA_INT1_END &&
++ channel < MCF_EDMA_INT2_END) {
++ MCF_INTC2_ICR(MCF_EDMA_INT2_CHANNEL_BASE) = irq_level;
++ MCF_INTC2_CIMR = MCF_EDMA_INT2_CHANNEL_BASE;
++ } else
++ ERR("Bad channel number!\n");
++#endif
++ return channel;
++ }
++ return -EINVAL;
++}
++EXPORT_SYMBOL(mcf_edma_request_channel);
++
++/**
++ * mcf_edma_set_callback - Update the channel callback/arg
++ * @channel: channel number
++ * @handler: dma handler
++ * @error_handler: dma error handler
++ * @arg: argument to pass back
++ *
++ * Returns 0 if success or a negative value if failure
++ */
++int
++mcf_edma_set_callback(int channel,
++ irqreturn_t(*handler) (int, void *),
++ void (*error_handler) (int, void *), void *arg)
++{
++ DBG("\n");
++
++ if (mcf_edma_devp != NULL && channel >= 0
++ && channel <= MCF_EDMA_CHANNELS
++ && mcf_edma_devp->dma_interrupt_handlers[channel].allocated) {
++ mcf_edma_devp->dma_interrupt_handlers[channel].irq_handler =
++ handler;
++ mcf_edma_devp->dma_interrupt_handlers[channel].error_handler =
++ error_handler;
++ mcf_edma_devp->dma_interrupt_handlers[channel].arg = arg;
++ return 0;
++ }
++ return -EINVAL;
++}
++EXPORT_SYMBOL(mcf_edma_set_callback);
++
++/**
++ * mcf_edma_free_channel - Free the edma channel
++ * @channel: channel number
++ * @arg: argument created with
++ *
++ * Returns 0 if success or a negative value if failure
++ */
++int
++mcf_edma_free_channel(int channel, void *arg)
++{
++ DBG("\n");
++
++ if (mcf_edma_devp != NULL && channel >= 0
++ && channel <= MCF_EDMA_CHANNELS) {
++ if (mcf_edma_devp->dma_interrupt_handlers[channel].allocated) {
++#if 1
++ if (mcf_edma_devp->dma_interrupt_handlers[channel].
++ arg != arg)
++ return -EBUSY;
++#endif
++
++ mcf_edma_devp->dma_interrupt_handlers[channel].
++ allocated = 0;
++ mcf_edma_devp->dma_interrupt_handlers[channel].arg =
++ NULL;
++ mcf_edma_devp->dma_interrupt_handlers[channel].
++ irq_handler = NULL;
++ mcf_edma_devp->dma_interrupt_handlers[channel].
++ error_handler = NULL;
++ mcf_edma_devp->dma_interrupt_handlers[channel].lock =
++ NULL;
++ }
++
++ /* make sure error interrupt is disabled */
++ MCF_EDMA_CEEI = MCF_EDMA_CEEI_CEEI(channel);
++
++ return 0;
++ }
++ return -EINVAL;
++}
++EXPORT_SYMBOL(mcf_edma_free_channel);
++
++/**
++ * mcf_edma_cleanup - cleanup driver allocated resources
++ */
++static void
++mcf_edma_cleanup(void)
++{
++ dev_t devno;
++ int i;
++
++ DBG("\n");
++
++ /* disable all error ints */
++ MCF_EDMA_CEEI = MCF_EDMA_CEEI_CAEE;
++
++ /* free interrupts/memory */
++ if (mcf_edma_devp) {
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ #if defined(CONFIG_M5445X)
++ free_irq(MCF_EDMA_INT0_BASE + i, mcf_edma_devp);
++ #elif defined(CONFIG_M5441X)
++ if (i >= 0 && i < MCF_EDMA_INT0_END)
++ free_irq(MCF_EDMA_INT0_BASE + i, mcf_edma_devp);
++ else if (i >= MCF_EDMA_INT0_END &&
++ i <= MCF_EDMA_INT1_END)
++ free_irq(MCF_EDMA_INT1_BASE +
++ (i - MCF_EDMA_INT0_END), mcf_edma_devp);
++ else if (i >= MCF_EDMA_INT1_END &&
++ i < MCF_EDMA_INT2_END) {
++ free_irq(MCF_EDMA_INT2_BASE, mcf_edma_devp);
++ break;
++ } else {
++ ERR("Bad irq number!\n");
++ return;
++ }
++ #endif
++ }
++
++ free_irq(MCF_EDMA_INT0_BASE + MCF_EDMA_INT_ERR, mcf_edma_devp);
++ cdev_del(&mcf_edma_devp->cdev);
++ kfree(mcf_edma_devp);
++ }
++
++ /* unregister character device */
++ devno = MKDEV(mcf_edma_major, 0);
++ unregister_chrdev_region(devno, 1);
++}
++
++/**
++ * mcf_edma_dump_channel - dump a channel information
++ */
++void
++mcf_edma_dump_channel(int channel)
++{
++ printk(KERN_DEBUG "EDMA Channel %d\n", channel);
++ printk(KERN_DEBUG " TCD Base = 0x%x\n",
++ (int)&MCF_EDMA_TCD_SADDR(channel));
++ printk(KERN_DEBUG " SRCADDR = 0x%lx\n",
++ MCF_EDMA_TCD_SADDR(channel));
++ printk(KERN_DEBUG " SRCOFF = 0x%x\n",
++ MCF_EDMA_TCD_SOFF(channel));
++ printk(KERN_DEBUG " XFR ATTRIB = 0x%x\n",
++ MCF_EDMA_TCD_ATTR(channel));
++ printk(KERN_DEBUG " SRCLAST = 0x%lx\n",
++ MCF_EDMA_TCD_SLAST(channel));
++ printk(KERN_DEBUG " DSTADDR = 0x%lx\n",
++ MCF_EDMA_TCD_DADDR(channel));
++ printk(KERN_DEBUG " MINOR BCNT = 0x%lx\n",
++ MCF_EDMA_TCD_NBYTES(channel));
++ printk(KERN_DEBUG " CUR_LOOP_CNT = 0x%x\n",
++ MCF_EDMA_TCD_CITER(channel)&0x1ff);
++ printk(KERN_DEBUG " BEG_LOOP_CNT = 0x%x\n",
++ MCF_EDMA_TCD_BITER(channel)&0x1ff);
++ printk(KERN_DEBUG " STATUS = 0x%x\n",
++ MCF_EDMA_TCD_CSR(channel));
++
++}
++EXPORT_SYMBOL(mcf_edma_dump_channel);
++
++#ifdef CONFIG_PROC_FS
++/*
++ * proc file system support
++ */
++
++#define FREE_CHANNEL "free"
++#define DEVICE_UNKNOWN "device unknown"
++
++/**
++ * mcf_edma_proc_show - print out proc info
++ * @m: seq_file
++ * @v:
++ */
++static int
++mcf_edma_proc_show(struct seq_file *m, void *v)
++{
++ int i;
++
++ if (mcf_edma_devp == NULL)
++ return 0;
++
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ if (mcf_edma_devp->dma_interrupt_handlers[i].allocated) {
++ if (mcf_edma_devp->dma_interrupt_handlers[i].device_id)
++ seq_printf(m, "%2d: %s\n", i,
++ mcf_edma_devp->
++ dma_interrupt_handlers[i].
++ device_id);
++ else
++ seq_printf(m, "%2d: %s\n", i, DEVICE_UNKNOWN);
++ } else
++ seq_printf(m, "%2d: %s\n", i, FREE_CHANNEL);
++ }
++ return 0;
++}
++
++/**
++ * mcf_edma_proc_open - open the proc file
++ * @inode: inode ptr
++ * @file: file ptr
++ */
++static int
++mcf_edma_proc_open(struct inode *inode, struct file *file)
++{
++ return single_open(file, mcf_edma_proc_show, NULL);
++}
++
++static const struct file_operations mcf_edma_proc_operations = {
++ .open = mcf_edma_proc_open,
++ .read = seq_read,
++ .llseek = seq_lseek,
++ .release = single_release,
++};
++
++/**
++ * mcf_edma_proc_init - initialize proc filesystem
++ */
++static int __init
++mcf_edma_proc_init(void)
++{
++ struct proc_dir_entry *e;
++
++ e = create_proc_entry("edma", 0, NULL);
++ if (e)
++ e->proc_fops = &mcf_edma_proc_operations;
++
++ return 0;
++}
++
++#endif
++
++/**
++ * mcf_edma_init - eDMA module init
++ */
++static int __init
++mcf_edma_init(void)
++{
++ dev_t dev;
++ int result;
++ int i;
++#ifdef CONFIG_M54455
++ u32 offset;
++#endif
++
++#if defined(CONFIG_M5441X)
++ /* edma group priority, default grp0 > grp1 > grp2 > grp3 */
++ u32 grp0_pri = MCF_EDMA_CR_GRP0PRI(0x00);
++ u32 grp1_pri = MCF_EDMA_CR_GRP1PRI(0x01);
++ u32 grp2_pri = MCF_EDMA_CR_GRP2PRI(0x02);
++ u32 grp3_pri = MCF_EDMA_CR_GRP3PRI(0x03);
++#endif
++
++ DBG("Entry\n");
++
++ /* allocate free major number */
++ result =
++ alloc_chrdev_region(&dev, MCF_EDMA_DEV_MINOR, 1,
++ MCF_EDMA_DRIVER_NAME);
++ if (result < 0) {
++ ERR("Error %d can't get major number.\n", result);
++ return result;
++ }
++ mcf_edma_major = MAJOR(dev);
++
++ /* allocate device driver structure */
++ mcf_edma_devp = kmalloc(sizeof(struct mcf_edma_dev), GFP_KERNEL);
++ if (!mcf_edma_devp) {
++ result = -ENOMEM;
++ goto fail;
++ }
++
++ /* init handlers (no handlers for beginning) */
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ mcf_edma_devp->dma_interrupt_handlers[i].irq_handler = NULL;
++ mcf_edma_devp->dma_interrupt_handlers[i].error_handler = NULL;
++ mcf_edma_devp->dma_interrupt_handlers[i].arg = NULL;
++ mcf_edma_devp->dma_interrupt_handlers[i].allocated = 0;
++ mcf_edma_devp->dma_interrupt_handlers[i].lock = NULL;
++ mcf_edma_devp->dma_interrupt_handlers[i].device_id = NULL;
++ MCF_EDMA_TCD_CSR(i) = 0x0000;
++ }
++
++ /* register char device */
++ cdev_init(&mcf_edma_devp->cdev, &mcf_edma_fops);
++ mcf_edma_devp->cdev.owner = THIS_MODULE;
++ mcf_edma_devp->cdev.ops = &mcf_edma_fops;
++ result = cdev_add(&mcf_edma_devp->cdev, dev, 1);
++ if (result) {
++ ERR("Error %d adding coldfire-dma device.\n", result);
++ result = -ENODEV;
++ goto fail;
++ }
++
++ /* request/enable irq for each eDMA channel */
++#if defined(CONFIG_M5445X)
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ result = request_irq(MCF_EDMA_INT0_BASE + i,
++ mcf_edma_isr, IRQF_DISABLED,
++ MCF_EDMA_DRIVER_NAME, mcf_edma_devp);
++ if (result) {
++ ERR("Cannot request irq %d\n",
++ (MCF_EDMA_INT0_BASE + i));
++ result = -EBUSY;
++ goto fail;
++ }
++ }
++#elif defined(CONFIG_M5441X)
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ if (i >= 0 && i < MCF_EDMA_INT0_END) {
++ result = request_irq(MCF_EDMA_INT0_BASE + i,
++ mcf_edma_isr, IRQF_DISABLED,
++ MCF_EDMA_DRIVER_NAME,
++ mcf_edma_devp);
++
++ if (result) {
++ ERR("Cannot request irq %d\n",
++ (MCF_EDMA_INT0_BASE + i));
++ result = -EBUSY;
++ goto fail;
++ }
++ } else if (i >= MCF_EDMA_INT0_END && i < MCF_EDMA_INT1_END) {
++ result = request_irq(MCF_EDMA_INT1_BASE +
++ (i - MCF_EDMA_INT0_END),
++ mcf_edma_isr, IRQF_DISABLED,
++ MCF_EDMA_DRIVER_NAME,
++ mcf_edma_devp);
++
++ if (result) {
++ ERR("Cannot request irq %d\n",
++ (MCF_EDMA_INT1_BASE +
++ (i - MCF_EDMA_INT0_END)));
++ result = -EBUSY;
++ goto fail;
++ }
++ } else if (i >= MCF_EDMA_INT1_END && MCF_EDMA_INT2_END) {
++ result = request_irq(MCF_EDMA_INT2_BASE,
++ mcf_edma_isr, IRQF_DISABLED,
++ MCF_EDMA_DRIVER_NAME,
++ mcf_edma_devp);
++ if (result) {
++ ERR("Cannot request irq %d\n",
++ MCF_EDMA_INT2_BASE);
++ result = -EBUSY;
++ goto fail;
++ }
++ break;
++ } else {
++ ERR(" Cannot request irq because of wrong number!\n");
++ result = -EBUSY;
++ goto fail;
++ }
++ }
++#endif
++
++ /* request error interrupt */
++ result = request_irq(MCF_EDMA_INT0_BASE + MCF_EDMA_INT_ERR,
++ mcf_edma_error_isr, IRQF_DISABLED,
++ MCF_EDMA_DRIVER_NAME, mcf_edma_devp);
++ if (result) {
++ ERR("Cannot request irq %d\n",
++ (MCF_EDMA_INT0_BASE + MCF_EDMA_INT_ERR));
++ result = -EBUSY;
++ goto fail;
++ }
++
++#if defined(CONFIG_M5445X)
++ MCF_EDMA_CR = 0;
++#elif defined(CONFIG_M5441X)
++ MCF_EDMA_CR = (0 | grp0_pri | grp1_pri | grp2_pri | grp3_pri);
++ DBG("MCF_EDMA_CR = %lx\n", MCF_EDMA_CR);
++#endif
++
++#ifdef CONFIG_M54455
++ fsl_pata_dma_tcd.pata_tcd_va = (struct TCD *) dma_alloc_coherent(NULL,
++ MCF_EDMA_TCD_PER_CHAN + 1,
++ &fsl_pata_dma_tcd.pata_tcd_pa,
++ GFP_KERNEL);
++
++ if (!fsl_pata_dma_tcd.pata_tcd_va) {
++ printk(KERN_INFO "MCF eDMA alllocate tcd memeory failed\n");
++ goto fail;
++ }
++
++
++ offset = (fsl_pata_dma_tcd.pata_tcd_pa & (sizeof(struct TCD)-1)) ;
++ if (offset) {
++ /*
++ * up align the addr to 32B to match the eDMA enginee require,
++ * ie. sizeof tcd boundary
++ * */
++ printk(KERN_INFO "pata tcd original:pa-%x[%x]\n",
++ fsl_pata_dma_tcd.pata_tcd_pa,
++ (u32)fsl_pata_dma_tcd.pata_tcd_va);
++
++ fsl_pata_dma_tcd.pata_tcd_pa += sizeof(struct TCD) - offset;
++ fsl_pata_dma_tcd.pata_tcd_va += sizeof(struct TCD) - offset;
++
++ printk(KERN_INFO "pata tcd realigned:pa-%x[%x]\n",
++ fsl_pata_dma_tcd.pata_tcd_pa,
++ (u32)fsl_pata_dma_tcd.pata_tcd_va);
++ }
++#endif
++#ifdef CONFIG_PROC_FS
++ mcf_edma_proc_init();
++#endif
++
++ INFO("Initialized successfully\n");
++ return 0;
++fail:
++ mcf_edma_cleanup();
++ return result;
++}
++
++/**
++ * mcf_edma_exit - eDMA module exit
++ */
++static void __exit
++mcf_edma_exit(void)
++{
++ mcf_edma_cleanup();
++}
++
++#ifdef CONFIG_COLDFIRE_EDMA_MODULE
++module_init(mcf_edma_init);
++module_exit(mcf_edma_exit);
++#else
++/* get us in early */
++postcore_initcall(mcf_edma_init);
++#endif
++
++MODULE_DESCRIPTION(MCF_EDMA_DRIVER_INFO);
++MODULE_AUTHOR(MCF_EDMA_DRIVER_AUTHOR);
++MODULE_LICENSE(MCF_EDMA_DRIVER_LICENSE);
+--- /dev/null
++++ b/drivers/dma/mcf_edma_test.c
+@@ -0,0 +1,276 @@
++/*
++ * mcf_edma_test.c - simple test/example module for Coldfire eDMA.
++ *
++ * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. All Rights Reserved.
++ * Author: Andrey Butok
++ *
++ * This program is free software; you can redistribute it and/or modify it
++ * under the terms of the GNU General Public License as published by the
++ * Free Software Foundation; either version 2 of the License, or (at your
++ * option) any later version.
++ *
++ * This program is distributed in the hope that 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
++ *
++ ***************************************************************************
++ * Changes:
++ * v0.001 29 February 2008 Andrey Butok
++ * Initial Release
++ *
++ * NOTE: This module tests eDMA driver performing
++ * a simple memory to memory transfer with a 32 bit
++ * source and destination transfer size that generates
++ * an interrupt when the transfer is complete.
++ */
++
++#include <linux/dma-mapping.h>
++#include <linux/init.h>
++#include <linux/fs.h>
++#include <linux/cdev.h>
++#include <linux/seq_file.h>
++#include <linux/proc_fs.h>
++#include <asm/mcfsim.h>
++#include <asm/coldfire.h>
++#include <asm/mcf_edma.h>
++#include <asm/cacheflush.h>
++
++#define MCF_EDMA_TEST_DRIVER_VERSION "Revision: 0.001"
++#define MCF_EDMA_TEST_DRIVER_AUTHOR \
++ "Freescale Semiconductor Inc, Andrey Butok"
++#define MCF_EDMA_TEST_DRIVER_DESC \
++ "Simple testing module for Coldfire eDMA "
++#define MCF_EDMA_TEST_DRIVER_INFO \
++ MCF_EDMA_TEST_DRIVER_VERSION " " MCF_EDMA_TEST_DRIVER_DESC
++#define MCF_EDMA_TEST_DRIVER_LICENSE "GPL"
++#define MCF_EDMA_TEST_DRIVER_NAME "mcf_edma_test"
++
++#ifndef TRUE
++#define TRUE 1
++#define FALSE 0
++#endif
++
++#define TEST_HALFIRQ
++
++/* Global variable used to signal main process when interrupt is recognized */
++static volatile int mcf_edma_test_interrupt;
++volatile int *mcf_edma_test_interrupt_p =
++ (volatile int *) &mcf_edma_test_interrupt;
++
++/********************************************************************/
++static irqreturn_t
++mcf_edma_test_handler(int channel, void *dev_id)
++{
++ int done = mcf_edma_check_done(channel);
++
++ /* Clear interrupt flag */
++ mcf_edma_confirm_interrupt_handled(channel);
++
++ if (done) {
++ printk(KERN_INFO "DMA Finished\n");
++
++ /* Set interrupt status flag to TRUE */
++ mcf_edma_test_interrupt = TRUE;
++ } else {
++ printk(KERN_INFO "DMA Halfway Done\n");
++
++ /* restart DMA. */
++ mcf_edma_confirm_halfirq(channel);
++ }
++
++ return IRQ_HANDLED;
++}
++
++static void
++mcf_edma_test_error_handler(int channel, void *dev_id)
++{
++ printk(KERN_INFO "DMA ERROR: Channel = %d\n", channel);
++ printk(KERN_INFO " EDMA_ES = 0x%lx\n", (MCF_EDMA_ES));
++ mcf_edma_dump_channel(channel);
++}
++
++/********************************************************************/
++
++int
++mcf_edma_test_block_compare(u8 *block1, u8 *block2, u32 size)
++{
++ u32 i;
++
++ for (i = 0; i < (size); i++) {
++ if ((*(u8 *) (block1 + i)) != (*(u8 *) (block2 + i))) {
++ printk(KERN_INFO "Data Mismatch index=0x%x len=0x%x "
++ "block1=0x%p block2=0x%p\n",
++ i, size, block1, block2);
++ return FALSE;
++ }
++ }
++
++ return TRUE;
++}
++
++/********************************************************************/
++
++void
++mcf_edma_test_run(void)
++{
++ u16 byte_count;
++ u32 i, j;
++ u8 *start_address;
++ u8 *dest_address;
++ u32 test_data;
++ int channel;
++ u32 allocated_channels_low = 0;
++ u32 allocated_channels_high = 0;
++
++ printk(KERN_INFO "\n===============================================\n");
++ printk(KERN_INFO "\nStarting eDMA transfer test!\n");
++
++ /* Initialize test variables */
++ byte_count = 0x2000;
++ test_data = 0xA5A5A5A5;
++
++ /* DMA buffer must be from GFP_DMA zone, so it will not be cached */
++ start_address = kmalloc(byte_count, GFP_DMA);
++ if (start_address == NULL) {
++ printk(KERN_INFO MCF_EDMA_TEST_DRIVER_NAME
++ ": failed to allocate DMA[%d] buffer\n", byte_count);
++ goto err_out;
++ }
++ dest_address = kmalloc(byte_count, /*GFP_KERNEL*/GFP_DMA);
++ if (dest_address == NULL) {
++ printk(KERN_INFO MCF_EDMA_TEST_DRIVER_NAME
++ ": failed to allocate DMA[%d] buffer\n", byte_count);
++ goto err_free_mem;
++ }
++
++ /* Test all automatically allocated DMA channels. The test data is
++ * complemented at the end of the loop, so that the testData value
++ * isn't the same twice in a row */
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ /* request eDMA channel */
++ channel = mcf_edma_request_channel(MCF_EDMA_CHANNEL_ANY,
++ mcf_edma_test_handler,
++ mcf_edma_test_error_handler,
++ 0x6,
++ NULL,
++ NULL,
++ MCF_EDMA_TEST_DRIVER_NAME);
++ if (channel < 0)
++ goto test_end;
++
++
++ if (channel >= 0 && channel < 32)
++ allocated_channels_low |= (1 << channel);
++ else if (channel >= 32 && channel < 64)
++ allocated_channels_high |= (1 << (channel - 32));
++
++ /* Initialize data for DMA to move */
++ for (j = 0; j < byte_count; j = j + 4) {
++ *((u32 *) (start_address + j)) = test_data;
++ *((u32 *) (dest_address + j)) = ~test_data;
++ }
++
++ /* Clear interrupt status indicator */
++ mcf_edma_test_interrupt = FALSE;
++
++ /* Configure DMA Channel TCD */
++#ifndef TEST_HALFIRQ
++ /* regular irq on completion */
++ mcf_edma_set_tcd_params(channel,
++ (u32)virt_to_phys(start_address),
++ (u32)virt_to_phys(dest_address),
++ (0 | MCF_EDMA_TCD_ATTR_SSIZE_32BIT |
++ MCF_EDMA_TCD_ATTR_DSIZE_32BIT), 0x04,
++ byte_count, 0x0, 1, 1, 0x04, 0x0, 0x1,
++ 0x0);
++#else
++ /* half completion irq */
++ mcf_edma_set_tcd_params_halfirq(channel,
++ (u32)virt_to_phys(start_address),
++ (u32)virt_to_phys(dest_address),
++ (MCF_EDMA_TCD_ATTR_SSIZE_32BIT |
++ MCF_EDMA_TCD_ATTR_DSIZE_32BIT),
++ 0x04, /* soff */
++ byte_count/2, /* bytes/loop */
++ 0x0, /* slast */
++ 2, /* loop count */
++ 0x04, /* doff */
++ 0x0, /* dlast_sga */
++ 0x0); /* req dis */
++#endif
++
++ printk(KERN_INFO "DMA Channel %d Bytes = 0x%x\n",
++ channel, byte_count);
++ /* Start DMA. */
++ mcf_edma_start_transfer(channel);
++
++ printk(KERN_INFO "DMA channel %d started.\n", channel);
++
++ /* Wait for DMA to complete */
++ while (!*mcf_edma_test_interrupt_p)
++ ;
++
++ /* Test data */
++ if (mcf_edma_test_block_compare
++ (start_address, dest_address, byte_count))
++ printk(KERN_INFO "Data are moved correctly.\n");
++ else
++ printk(KERN_INFO "ERROR!!! Data error!\n");
++
++ printk(KERN_INFO "DMA channel %d test complete.\n", channel);
++ printk(KERN_INFO "-------------------------------\n");
++
++ /* Complement test data so next channel test does not
++ * use same values */
++ test_data = ~test_data;
++ }
++
++test_end:
++ printk(KERN_INFO "All tests are complete\n\n");
++ printk(KERN_INFO
++ "It has been automatically allocated %d eDMA channels:\n", i);
++ for (i = 0; i < MCF_EDMA_CHANNELS; i++) {
++ if ((allocated_channels_low & (1 << i)) ||
++ (allocated_channels_high & (1 << (i - 32)))) {
++ printk(KERN_INFO "%d,\n", i);
++ mcf_edma_free_channel(i, NULL);
++ }
++ }
++ printk(KERN_INFO "===============================================\n\n");
++
++ kfree(dest_address);
++err_free_mem:
++ kfree(start_address);
++err_out:
++ return;
++}
++
++/********************************************************************/
++
++static int __init
++mcf_edma_test_init(void)
++{
++ mcf_edma_test_run();
++
++ /* We intentionaly return -EAGAIN to prevent keeping
++ * the module. It does all its work from init()
++ * and doesn't offer any runtime functionality */
++ return -EAGAIN;
++}
++
++static void __exit
++mcf_edma_test_exit(void)
++{
++}
++
++module_init(mcf_edma_test_init);
++module_exit(mcf_edma_test_exit);
++
++MODULE_DESCRIPTION(MCF_EDMA_TEST_DRIVER_INFO);
++MODULE_AUTHOR(MCF_EDMA_TEST_DRIVER_AUTHOR);
++MODULE_LICENSE(MCF_EDMA_TEST_DRIVER_LICENSE);