[Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Pavel Cheblakov]

This is a general driver for cards based on PLX90xx PCI-bridges.
It supports following cards:
 - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
 - Adlink PCI-7841/cPCI-7841 SE card
 - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
 - esd CAN-PCI/PMC/266
 - esd CAN-PCIe/2000
 - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
 - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)

This driver has been ported from socketcan of mainline Linux kernel 2.6.34.
Support of esd cards has been added by Matthias Fuchs in linux-next

Signed-off-by: Pavel Cheblakov <P.B.Cheblakov@domain.hid>
---
 ksrc/drivers/can/sja1000/Kconfig         |   15 +
 ksrc/drivers/can/sja1000/Makefile        |    7 +
 ksrc/drivers/can/sja1000/rtcan_plx_pci.c |  608 ++++++++++++++++++++++++++++++
 3 files changed, 630 insertions(+), 0 deletions(-)
 create mode 100644 ksrc/drivers/can/sja1000/rtcan_plx_pci.c

diff --git a/ksrc/drivers/can/sja1000/Kconfig b/ksrc/drivers/can/sja1000/Kconfig
index 56f4691..39c18e4 100644
--- a/ksrc/drivers/can/sja1000/Kconfig
+++ b/ksrc/drivers/can/sja1000/Kconfig
@@ -41,6 +41,21 @@ config XENO_DRIVERS_CAN_SJA1000_IXXAT_PCI
 	the second channel working, Xenomai's shared interrupt support
 	must be enabled.
 
+config XENO_DRIVERS_CAN_SJA1000_PLX_PCI
+	depends on XENO_DRIVERS_CAN_SJA1000 && PCI
+	tristate "PLX-based PCI Cards"
+	help
+
+	This driver is for CAN interface cards based on
+	the PLX90xx PCI bridge.
+	Driver supports now:
+	 - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
+	 - Adlink PCI-7841/cPCI-7841 SE card
+	 - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
+	 - esd CAN-PCI/PMC/266
+	 - esd CAN-PCIe/2000
+	 - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
+	 - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
 
 config XENO_DRIVERS_CAN_SJA1000_EMS_PCI
 	depends on XENO_DRIVERS_CAN_SJA1000 && PCI
diff --git a/ksrc/drivers/can/sja1000/Makefile b/ksrc/drivers/can/sja1000/Makefile
index 09a03ab..389ca04 100644
--- a/ksrc/drivers/can/sja1000/Makefile
+++ b/ksrc/drivers/can/sja1000/Makefile
@@ -7,6 +7,7 @@ EXTRA_CFLAGS += -D__IN_XENOMAI__ -Iinclude/xenomai -Idrivers/xenomai/can -Idrive
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000) += xeno_can_sja1000.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_PEAK_PCI) += xeno_can_peak_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_PEAK_DNG) += xeno_can_peak_dng.o
+obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_PLX_PCI) += xeno_can_plx_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_IXXAT_PCI) += xeno_can_ixxat_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_EMS_PCI) += xeno_can_ems_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_ESD_PCI) += xeno_can_esd_pci.o
@@ -16,6 +17,7 @@ obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_MEM) += xeno_can_mem.o
 xeno_can_sja1000-y := rtcan_sja1000.o rtcan_sja1000_proc.o
 xeno_can_peak_pci-y := rtcan_peak_pci.o
 xeno_can_peak_dng-y := rtcan_peak_dng.o
+xeno_can_plx_pci-y := rtcan_plx_pci.o
 xeno_can_ixxat_pci-y := rtcan_ixxat_pci.o
 xeno_can_ems_pci-y := rtcan_ems_pci.o
 xeno_can_esd_pci-y := rtcan_esd_pci.o
@@ -31,6 +33,7 @@ O_TARGET := built-in.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000) += xeno_can_sja1000.o 
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_PEAK_PCI) += xeno_can_peak_pci.o 
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_PEAK_DNG) += xeno_can_peak_dng.o 
+obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_PLX_PCI) += xeno_can_plx_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_IXXAT_PCI) += xeno_can_ixxat_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_EMS_PCI) += xeno_can_ems_pci.o
 obj-$(CONFIG_XENO_DRIVERS_CAN_SJA1000_ESD_PCI) += xeno_can_esd_pci.o
@@ -42,6 +45,7 @@ list-multi := xeno_can_sja1000.o
 xeno_can_sja1000-objs := rtcan_sja1000.o rtcan_sja1000_proc.o
 xeno_can_peak_pci-objs := rtcan_peak_pci.o
 xeno_can_peak_dng-objs := rtcan_peak_dng.o
+xeno_can_plx_pci-objs := rtcan_plx_pci.o
 xeno_can_ixxat_pci-objs := rtcan_ixxat_pci.o
 xeno_can_ems_pci-objs := rtcan_ems_pci.o
 xeno_can_esd_pci-objs := rtcan_esd_pci.o
@@ -63,6 +67,9 @@ xeno_can_peak_pci.o: $(xeno_can_peak_pci-objs)
 xeno_can_peak_dng.o: $(xeno_can_peak_dng-objs)
 	$(LD) -r -o $@ $(xeno_can_peak_dng-objs)
 
+xeno_can_plx_pci.o: $(xeno_can_plx_pci-objs)
+	$(LD) -r -o $@ $(xeno_can_plx_pci-objs)
+
 xeno_can_ixxat_pci.o: $(xeno_can_ixxat_pci-objs)
 	$(LD) -r -o $@ $(xeno_can_ixxat_pci-objs)
 
diff --git a/ksrc/drivers/can/sja1000/rtcan_plx_pci.c b/ksrc/drivers/can/sja1000/rtcan_plx_pci.c
new file mode 100644
index 0000000..6a5366b
--- /dev/null
+++ b/ksrc/drivers/can/sja1000/rtcan_plx_pci.c
@@ -0,0 +1,608 @@
+/*
+ * Copyright (C) 2008-2010 Pavel Cheblakov <P.B.Cheblakov@domain.hid>
+ *
+ * Derived from the ems_pci.c driver:
+ *	Copyright (C) 2007 Wolfgang Grandegger <wg@domain.hid>
+ *	Copyright (C) 2008 Markus Plessing <plessing@domain.hid>
+ *	Copyright (C) 2008 Sebastian Haas <haas@domain.hid>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the version 2 of the GNU General Public License
+ * as published by the Free Software Foundation
+ *
+ * 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.
+ */
+
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/pci.h>
+#include <linux/io.h>
+
+#include <rtdm/rtdm_driver.h>
+
+/* CAN device profile */
+#include <rtdm/rtcan.h>
+#include <rtcan_dev.h>
+#include <rtcan_raw.h>
+#include <rtcan_internal.h>
+#include <rtcan_sja1000.h>
+#include <rtcan_sja1000_regs.h>
+
+#define RTCAN_DRV_NAME "rt_sja1000_plx_pci"
+#define RTCAN_DEV_NAME "rtcan%d"
+
+MODULE_AUTHOR("Pavel Cheblakov <P.B.Cheblakov@domain.hid>");
+MODULE_DESCRIPTION("RTCAN driver for PLX90xx PCI-bridge cards with "
+		   "the SJA1000 chips");
+MODULE_SUPPORTED_DEVICE("Adlink PCI-7841/cPCI-7841, "
+			"Adlink PCI-7841/cPCI-7841 SE, "
+			"Marathon CAN-bus-PCI, "
+			"TEWS TECHNOLOGIES TPMC810, "
+			"esd CAN-PCI/CPCI/PCI104/200, "
+			"esd CAN-PCI/PMC/266, "
+			"esd CAN-PCIe/2000")
+MODULE_LICENSE("GPL v2");
+
+#define PLX_PCI_MAX_CHAN 2
+
+struct plx_pci_card {
+	int channels;			/* detected channels count */
+	struct rtcan_device *rtcan_dev[PLX_PCI_MAX_CHAN];
+	void __iomem *conf_addr;
+
+	/* Pointer to device-dependent reset function */
+	void (*reset_func)(struct pci_dev *pdev);
+};
+
+#define PLX_PCI_CAN_CLOCK (16000000 / 2)
+
+/* PLX9030/9050/9052 registers */
+#define PLX_INTCSR	0x4c		/* Interrupt Control/Status */
+#define PLX_CNTRL	0x50		/* User I/O, Direct Slave Response,
+					 * Serial EEPROM, and Initialization
+					 * Control register
+					 */
+
+#define PLX_LINT1_EN	0x1		/* Local interrupt 1 enable */
+#define PLX_LINT2_EN	(1 << 3)	/* Local interrupt 2 enable */
+#define PLX_PCI_INT_EN	(1 << 6)	/* PCI Interrupt Enable */
+#define PLX_PCI_RESET	(1 << 30)	/* PCI Adapter Software Reset */
+
+/* PLX9056 registers */
+#define PLX9056_INTCSR	0x68		/* Interrupt Control/Status */
+#define PLX9056_CNTRL	0x6c		/* Control / Software Reset */
+
+#define PLX9056_LINTI	(1 << 11)
+#define PLX9056_PCI_INT_EN (1 << 8)
+#define PLX9056_PCI_RCR	(1 << 29)	/* Read Configuration Registers */
+
+/*
+ * The board configuration is probably following:
+ * RX1 is connected to ground.
+ * TX1 is not connected.
+ * CLKO is not connected.
+ * Setting the OCR register to 0xDA is a good idea.
+ * This means normal output mode, push-pull and the correct polarity.
+ */
+#define PLX_PCI_OCR	(SJA_OCR_MODE_NORMAL | SJA_OCR_TX0_PUSHPULL | SJA_OCR_TX1_PUSHPULL)
+
+/*
+ * In the CDR register, you should set CBP to 1.
+ * You will probably also want to set the clock divider value to 7
+ * (meaning direct oscillator output) because the second SJA1000 chip
+ * is driven by the first one CLKOUT output.
+ */
+#define PLX_PCI_CDR			(SJA_CDR_CBP | SJA_CDR_CAN_MODE)
+
+/* SJA1000 Control Register in the BasicCAN Mode */
+#define SJA_CR				0x00
+
+/* States of some SJA1000 registers after hardware reset in the BasicCAN mode*/
+#define REG_CR_BASICCAN_INITIAL		0x21
+#define REG_CR_BASICCAN_INITIAL_MASK	0xa1
+#define REG_SR_BASICCAN_INITIAL		0x0c
+#define REG_IR_BASICCAN_INITIAL		0xe0
+
+/* States of some SJA1000 registers after hardware reset in the PeliCAN mode*/
+#define REG_MOD_PELICAN_INITIAL		0x01
+#define REG_SR_PELICAN_INITIAL		0x3c
+#define REG_IR_PELICAN_INITIAL		0x00
+
+#define ADLINK_PCI_VENDOR_ID		0x144A
+#define ADLINK_PCI_DEVICE_ID		0x7841
+
+#define ESD_PCI_SUB_SYS_ID_PCI200	0x0004
+#define ESD_PCI_SUB_SYS_ID_PCI266	0x0009
+#define ESD_PCI_SUB_SYS_ID_PMC266	0x000e
+#define ESD_PCI_SUB_SYS_ID_CPCI200	0x010b
+#define ESD_PCI_SUB_SYS_ID_PCIE2000	0x0200
+#define ESD_PCI_SUB_SYS_ID_PCI104200	0x0501
+
+#define MARATHON_PCI_DEVICE_ID		0x2715
+
+#define TEWS_PCI_VENDOR_ID		0x1498
+#define TEWS_PCI_DEVICE_ID_TMPC810	0x032A
+
+static void plx_pci_reset_common(struct pci_dev *pdev);
+static void plx_pci_reset_marathon(struct pci_dev *pdev);
+static void plx9056_pci_reset_common(struct pci_dev *pdev);
+
+struct plx_pci_channel_map {
+	u32 bar;
+	u32 offset;
+	u32 size;		/* 0x00 - auto, e.g. length of entire bar */
+};
+
+struct plx_pci_card_info {
+	const char *name;
+	int channel_count;
+	u32 can_clock;
+	u8 ocr;			/* output control register */
+	u8 cdr;			/* clock divider register */
+
+	/* Parameters for mapping local configuration space */
+	struct plx_pci_channel_map conf_map;
+
+	/* Parameters for mapping the SJA1000 chips */
+	struct plx_pci_channel_map chan_map_tbl[PLX_PCI_MAX_CHAN];
+
+	/* Pointer to device-dependent reset function */
+	void (*reset_func)(struct pci_dev *pdev);
+};
+
+static struct plx_pci_card_info plx_pci_card_info_adlink __devinitdata = {
+	"Adlink PCI-7841/cPCI-7841", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{1, 0x00, 0x00}, { {2, 0x00, 0x80}, {2, 0x80, 0x80} },
+	&plx_pci_reset_common
+	/* based on PLX9052 */
+};
+
+static struct plx_pci_card_info plx_pci_card_info_adlink_se __devinitdata = {
+	"Adlink PCI-7841/cPCI-7841 SE", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{0, 0x00, 0x00}, { {2, 0x00, 0x80}, {2, 0x80, 0x80} },
+	&plx_pci_reset_common
+	/* based on PLX9052 */
+};
+
+static struct plx_pci_card_info plx_pci_card_info_esd200 __devinitdata = {
+	"esd CAN-PCI/CPCI/PCI104/200", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{0, 0x00, 0x00}, { {2, 0x00, 0x80}, {2, 0x100, 0x80} },
+	&plx_pci_reset_common
+	/* based on PLX9030/9050 */
+};
+
+static struct plx_pci_card_info plx_pci_card_info_esd266 __devinitdata = {
+	"esd CAN-PCI/PMC/266", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{0, 0x00, 0x00}, { {2, 0x00, 0x80}, {2, 0x100, 0x80} },
+	&plx9056_pci_reset_common
+	/* based on PLX9056 */
+};
+
+static struct plx_pci_card_info plx_pci_card_info_esd2000 __devinitdata = {
+	"esd CAN-PCIe/2000", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{0, 0x00, 0x00}, { {2, 0x00, 0x80}, {2, 0x100, 0x80} },
+	&plx9056_pci_reset_common
+	/* based on PEX8311 */
+};
+
+static struct plx_pci_card_info plx_pci_card_info_marathon __devinitdata = {
+	"Marathon CAN-bus-PCI", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{0, 0x00, 0x00}, { {2, 0x00, 0x00}, {4, 0x00, 0x00} },
+	&plx_pci_reset_marathon
+	/* based on PLX9052 */
+};
+
+static struct plx_pci_card_info plx_pci_card_info_tews __devinitdata = {
+	"TEWS TECHNOLOGIES TPMC810", 2,
+	PLX_PCI_CAN_CLOCK, PLX_PCI_OCR, PLX_PCI_CDR,
+	{0, 0x00, 0x00}, { {2, 0x000, 0x80}, {2, 0x100, 0x80} },
+	&plx_pci_reset_common
+	/* based on PLX9030 */
+};
+
+static DEFINE_PCI_DEVICE_TABLE(plx_pci_tbl) = {
+	{
+		/* Adlink PCI-7841/cPCI-7841 */
+		ADLINK_PCI_VENDOR_ID, ADLINK_PCI_DEVICE_ID,
+		PCI_ANY_ID, PCI_ANY_ID,
+		PCI_CLASS_NETWORK_OTHER << 8, ~0,
+		(kernel_ulong_t)&plx_pci_card_info_adlink
+	},
+	{
+		/* Adlink PCI-7841/cPCI-7841 SE */
+		ADLINK_PCI_VENDOR_ID, ADLINK_PCI_DEVICE_ID,
+		PCI_ANY_ID, PCI_ANY_ID,
+		PCI_CLASS_COMMUNICATION_OTHER << 8, ~0,
+		(kernel_ulong_t)&plx_pci_card_info_adlink_se
+	},
+	{
+		/* esd CAN-PCI/200 */
+		PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_9050,
+		PCI_VENDOR_ID_ESDGMBH, ESD_PCI_SUB_SYS_ID_PCI200,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_esd200
+	},
+	{
+		/* esd CAN-CPCI/200 */
+		PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_9030,
+		PCI_VENDOR_ID_ESDGMBH, ESD_PCI_SUB_SYS_ID_CPCI200,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_esd200
+	},
+	{
+		/* esd CAN-PCI104/200 */
+		PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_9030,
+		PCI_VENDOR_ID_ESDGMBH, ESD_PCI_SUB_SYS_ID_PCI104200,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_esd200
+	},
+	{
+		/* esd CAN-PCI/266 */
+		PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_9056,
+		PCI_VENDOR_ID_ESDGMBH, ESD_PCI_SUB_SYS_ID_PCI266,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_esd266
+	},
+	{
+		/* esd CAN-PMC/266 */
+		PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_9056,
+		PCI_VENDOR_ID_ESDGMBH, ESD_PCI_SUB_SYS_ID_PMC266,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_esd266
+	},
+	{
+		/* esd CAN-PCIE/2000 */
+		PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_9056,
+		PCI_VENDOR_ID_ESDGMBH, ESD_PCI_SUB_SYS_ID_PCIE2000,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_esd2000
+	},
+	{
+		/* Marathon CAN-bus-PCI card */
+		PCI_VENDOR_ID_PLX, MARATHON_PCI_DEVICE_ID,
+		PCI_ANY_ID, PCI_ANY_ID,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_marathon
+	},
+	{
+		/* TEWS TECHNOLOGIES TPMC810 card */
+		TEWS_PCI_VENDOR_ID, TEWS_PCI_DEVICE_ID_TMPC810,
+		PCI_ANY_ID, PCI_ANY_ID,
+		0, 0,
+		(kernel_ulong_t)&plx_pci_card_info_tews
+	},
+	{ 0,}
+};
+MODULE_DEVICE_TABLE(pci, plx_pci_tbl);
+
+static u8 plx_pci_read_reg(const struct rtcan_device *dev, int port)
+{
+	return ioread8((void* __iomem)dev->base_addr + port);
+}
+
+static void plx_pci_write_reg(const struct rtcan_device *dev, int port, u8 val)
+{
+	iowrite8(val, (void* __iomem)dev->base_addr + port);
+}
+
+/*
+ * Check if a CAN controller is present at the specified location
+ * by trying to switch 'em from the Basic mode into the PeliCAN mode.
+ * Also check states of some registers in reset mode.
+ */
+static inline int plx_pci_check_sja1000(const struct rtcan_device *dev)
+{
+	int flag = 0;
+
+	struct rtcan_sja1000 *chip = (struct rtcan_sja1000 *)dev->priv;
+
+	/*
+	 * Check registers after hardware reset (the Basic mode)
+	 * See states on p. 10 of the Datasheet.
+	 */
+	if ((chip->read_reg(dev, SJA_CR) & REG_CR_BASICCAN_INITIAL_MASK) ==
+	    REG_CR_BASICCAN_INITIAL &&
+	    (chip->read_reg(dev, SJA_SR) == REG_SR_BASICCAN_INITIAL) &&
+	    (chip->read_reg(dev, SJA_IR) == REG_IR_BASICCAN_INITIAL))
+		flag = 1;
+
+	/* Bring the SJA1000 into the PeliCAN mode*/
+	chip->write_reg(dev, SJA_CDR, SJA_CDR_CAN_MODE);
+
+	/*
+	 * Check registers after reset in the PeliCAN mode.
+	 * See states on p. 23 of the Datasheet.
+	 */
+	if (chip->read_reg(dev, SJA_MOD) == REG_MOD_PELICAN_INITIAL &&
+	    chip->read_reg(dev, SJA_SR) == REG_SR_PELICAN_INITIAL &&
+	    chip->read_reg(dev, SJA_IR) == REG_IR_PELICAN_INITIAL)
+		return flag;
+
+	return 0;
+}
+
+/*
+ * PLX9030/50/52 software reset
+ * Also LRESET# asserts and brings to reset device on the Local Bus (if wired).
+ * For most cards it's enough for reset the SJA1000 chips.
+ */
+static void plx_pci_reset_common(struct pci_dev *pdev)
+{
+	struct plx_pci_card *card = pci_get_drvdata(pdev);
+	u32 cntrl;
+
+	cntrl = ioread32(card->conf_addr + PLX_CNTRL);
+	cntrl |= PLX_PCI_RESET;
+	iowrite32(cntrl, card->conf_addr + PLX_CNTRL);
+	udelay(100);
+	cntrl ^= PLX_PCI_RESET;
+	iowrite32(cntrl, card->conf_addr + PLX_CNTRL);
+};
+
+/*
+ * PLX9056 software reset
+ * Assert LRESET# and reset device(s) on the Local Bus (if wired).
+ */
+static void plx9056_pci_reset_common(struct pci_dev *pdev)
+{
+	struct plx_pci_card *card = pci_get_drvdata(pdev);
+	u32 cntrl;
+
+	/* issue a local bus reset */
+	cntrl = ioread32(card->conf_addr + PLX9056_CNTRL);
+	cntrl |= PLX_PCI_RESET;
+	iowrite32(cntrl, card->conf_addr + PLX9056_CNTRL);
+	udelay(100);
+	cntrl ^= PLX_PCI_RESET;
+	iowrite32(cntrl, card->conf_addr + PLX9056_CNTRL);
+
+	/* reload local configuration from EEPROM */
+	cntrl |= PLX9056_PCI_RCR;
+	iowrite32(cntrl, card->conf_addr + PLX9056_CNTRL);
+
+	/*
+	 * There is no safe way to poll for the end
+	 * of reconfiguration process. Waiting for 10ms
+	 * is safe.
+	 */
+	mdelay(10);
+
+	cntrl ^= PLX9056_PCI_RCR;
+	iowrite32(cntrl, card->conf_addr + PLX9056_CNTRL);
+};
+
+/* Special reset function for Marathon card */
+static void plx_pci_reset_marathon(struct pci_dev *pdev)
+{
+	void __iomem *reset_addr;
+	int i;
+	int reset_bar[2] = {3, 5};
+
+	plx_pci_reset_common(pdev);
+
+	for (i = 0; i < 2; i++) {
+		reset_addr = pci_iomap(pdev, reset_bar[i], 0);
+		if (!reset_addr) {
+			dev_err(&pdev->dev, "Failed to remap reset "
+				"space %d (BAR%d)\n", i, reset_bar[i]);
+		} else {
+			/* reset the SJA1000 chip */
+			iowrite8(0x1, reset_addr);
+			udelay(100);
+			pci_iounmap(pdev, reset_addr);
+		}
+	}
+}
+
+static void plx_pci_del_card(struct pci_dev *pdev)
+{
+	struct plx_pci_card *card = pci_get_drvdata(pdev);
+	struct rtcan_device *dev;
+	int i = 0;
+
+	for (i = 0; i < card->channels; i++) {
+		dev = card->rtcan_dev[i];
+		if (!dev)
+			continue;
+
+		dev_info(&pdev->dev, "Removing %s\n", dev->name);
+		rtcan_sja1000_unregister(dev);
+		if (dev->base_addr)
+			pci_iounmap(pdev, (void* __iomem)dev->base_addr);
+		rtcan_dev_free(dev);
+	}
+
+	card->reset_func(pdev);
+
+	/*
+	 * Disable interrupts from PCI-card and disable local
+	 * interrupts
+	 */
+	if (pdev->device != PCI_DEVICE_ID_PLX_9056)
+		iowrite32(0x0, card->conf_addr + PLX_INTCSR);
+	else
+		iowrite32(0x0, card->conf_addr + PLX9056_INTCSR);
+
+	if (card->conf_addr)
+		pci_iounmap(pdev, card->conf_addr);
+
+	kfree(card);
+
+	pci_disable_device(pdev);
+	pci_set_drvdata(pdev, NULL);
+}
+
+/*
+ * Probe PLX90xx based device for the SJA1000 chips and register each
+ * available CAN channel to SJA1000 Socket-CAN subsystem.
+ */
+static int __devinit plx_pci_add_card(struct pci_dev *pdev,
+				      const struct pci_device_id *ent)
+{
+	struct rtcan_sja1000 *chip;
+	struct rtcan_device *dev;
+	struct plx_pci_card *card;
+	struct plx_pci_card_info *ci;
+	int err, i;
+	u32 val;
+	void __iomem *addr;
+
+	ci = (struct plx_pci_card_info *)ent->driver_data;
+
+	if (pci_enable_device(pdev) < 0) {
+		dev_err(&pdev->dev, "Failed to enable PCI device\n");
+		return -ENODEV;
+	}
+
+	dev_info(&pdev->dev, "Detected \"%s\" card at slot #%i\n",
+		 ci->name, PCI_SLOT(pdev->devfn));
+
+	/* Allocate card structures to hold addresses, ... */
+	card = kzalloc(sizeof(*card), GFP_KERNEL);
+	if (!card) {
+		dev_err(&pdev->dev, "Unable to allocate memory\n");
+		pci_disable_device(pdev);
+		return -ENOMEM;
+	}
+
+	pci_set_drvdata(pdev, card);
+
+	card->channels = 0;
+
+	/* Remap PLX90xx configuration space */
+	addr = pci_iomap(pdev, ci->conf_map.bar, ci->conf_map.size);
+	if (!addr) {
+		err = -ENOMEM;
+		dev_err(&pdev->dev, "Failed to remap configuration space "
+			"(BAR%d)\n", ci->conf_map.bar);
+		goto failure_cleanup;
+	}
+	card->conf_addr = addr + ci->conf_map.offset;
+
+	ci->reset_func(pdev);
+	card->reset_func = ci->reset_func;
+
+	/* Detect available channels */
+	for (i = 0; i < ci->channel_count; i++) {
+		struct plx_pci_channel_map *cm = &ci->chan_map_tbl[i];
+
+		dev = rtcan_dev_alloc(sizeof(struct rtcan_sja1000),
+				      sizeof(struct plx_pci_card));
+		if (!dev) {
+			err = -ENOMEM;
+			goto failure_cleanup;
+		}
+
+		strncpy(dev->name, RTCAN_DEV_NAME, IFNAMSIZ);
+		dev->board_name = ci->name;
+
+		card->rtcan_dev[i] = dev;
+		chip = card->rtcan_dev[i]->priv;
+		chip->irq_flags = RTDM_IRQTYPE_SHARED;
+		chip->irq_num = pdev->irq;
+
+		/*
+		 * Remap IO space of the SJA1000 chips
+		 * This is device-dependent mapping
+		 */
+		addr = pci_iomap(pdev, cm->bar, cm->size);
+		if (!addr) {
+			err = -ENOMEM;
+			dev_err(&pdev->dev, "Failed to remap BAR%d\n", cm->bar);
+			goto failure_cleanup;
+		}
+
+		dev->base_addr = (unsigned long)(addr + cm->offset);
+		chip->read_reg = plx_pci_read_reg;
+		chip->write_reg = plx_pci_write_reg;
+
+		/* Check if channel is present */
+		if (plx_pci_check_sja1000(dev)) {
+			dev->can_sys_clock = ci->can_clock;
+			chip->ocr = ci->ocr;
+			chip->cdr = ci->cdr;
+
+			/* Register SJA1000 device */
+			err = rtcan_sja1000_register(dev);
+			if (err) {
+				dev_err(&pdev->dev, "Registering device failed "
+					"(err=%d)\n", err);
+				rtcan_dev_free(dev);
+				goto failure_cleanup;
+			}
+
+			card->channels++;
+
+			dev_info(&pdev->dev, "Channel #%d at 0x%p, irq %d "
+				 "registered as %s\n", i + 1,
+				 (void* __iomem)dev->base_addr, chip->irq_num,
+				 dev->name);
+		} else {
+			dev_err(&pdev->dev, "Channel #%d not detected\n",
+				i + 1);
+			rtcan_dev_free(dev);
+		}
+	}
+
+	if (!card->channels) {
+		err = -ENODEV;
+		goto failure_cleanup;
+	}
+
+	/*
+	 * Enable interrupts from PCI-card (PLX90xx) and enable Local_1,
+	 * Local_2 interrupts from the SJA1000 chips
+	 */
+	if (pdev->device != PCI_DEVICE_ID_PLX_9056) {
+		val = ioread32(card->conf_addr + PLX_INTCSR);
+		if (pdev->subsystem_vendor == PCI_VENDOR_ID_ESDGMBH)
+			val |= PLX_LINT1_EN | PLX_PCI_INT_EN;
+		else
+			val |= PLX_LINT1_EN | PLX_LINT2_EN | PLX_PCI_INT_EN;
+		iowrite32(val, card->conf_addr + PLX_INTCSR);
+	} else {
+		iowrite32(PLX9056_LINTI | PLX9056_PCI_INT_EN,
+			  card->conf_addr + PLX9056_INTCSR);
+	}
+	return 0;
+
+failure_cleanup:
+	dev_err(&pdev->dev, "Error: %d. Cleaning Up.\n", err);
+
+	plx_pci_del_card(pdev);
+
+	return err;
+}
+
+static struct pci_driver plx_pci_driver = {
+	.name = RTCAN_DRV_NAME,
+	.id_table = plx_pci_tbl,
+	.probe = plx_pci_add_card,
+	.remove = plx_pci_del_card,
+};
+
+static int __init plx_pci_init(void)
+{
+	return pci_register_driver(&plx_pci_driver);
+}
+
+static void __exit plx_pci_exit(void)
+{
+	pci_unregister_driver(&plx_pci_driver);
+}
+
+module_init(plx_pci_init);
+module_exit(plx_pci_exit);
-- 
1.6.0.6

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Sebastian Smolorz]

Pavel Cheblakov wrote:
> This is a general driver for cards based on PLX90xx PCI-bridges.
> It supports following cards:
>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
>  - Adlink PCI-7841/cPCI-7841 SE card
>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
>  - esd CAN-PCI/PMC/266
>  - esd CAN-PCIe/2000
>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)

The esd cards mentioned above are supported by the RTCAN driver 
xeno_can_esd_pci. Why do you propose a new driver instead of extending the 
existing one?

-- 
Sebastian

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Wolfgang Grandegger]

Hi Sebastian,

On 05/18/2010 01:42 PM, Sebastian Smolorz wrote:
> Pavel Cheblakov wrote:
>> This is a general driver for cards based on PLX90xx PCI-bridges.
>> It supports following cards:
>>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
>>  - Adlink PCI-7841/cPCI-7841 SE card
>>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
>>  - esd CAN-PCI/PMC/266
>>  - esd CAN-PCIe/2000
>>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
>>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
> 
> The esd cards mentioned above are supported by the RTCAN driver 
> xeno_can_esd_pci. Why do you propose a new driver instead of extending the 
> existing one?

For Socket-CAN, this driver is supposed to support all PLX PCI based
boards including the esd CAN PCI cards and also the IXXAT PCI board (not
yet done, though). The RTCAN driver xeno_can_esd_pci is a *dedicated*
driver for that card without generic support for the PLX PCI chips.
Extending it makes little sense. The question is if we want to drop
xeno_can_esd_pci and xeno_can_ixxat_pci.

Wolfgang.

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[P.B.Cheblakov]

Wolfgang Grandegger wrote:
> Hi Sebastian,
>
> On 05/18/2010 01:42 PM, Sebastian Smolorz wrote:
>   
>> Pavel Cheblakov wrote:
>>     
>>> This is a general driver for cards based on PLX90xx PCI-bridges.
>>> It supports following cards:
>>>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
>>>  - Adlink PCI-7841/cPCI-7841 SE card
>>>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
>>>  - esd CAN-PCI/PMC/266
>>>  - esd CAN-PCIe/2000
>>>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
>>>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
>>>       
>> The esd cards mentioned above are supported by the RTCAN driver 
>> xeno_can_esd_pci. Why do you propose a new driver instead of extending the 
>> existing one?
>>     
>
> For Socket-CAN, this driver is supposed to support all PLX PCI based
> boards including the esd CAN PCI cards and also the IXXAT PCI board (not
> yet done, though). The RTCAN driver xeno_can_esd_pci is a *dedicated*
> driver for that card without generic support for the PLX PCI chips.
> Extending it makes little sense. The question is if we want to drop
> xeno_can_esd_pci and xeno_can_ixxat_pci.
>
> Wolfgang.
>   
I subscribe to Wolfgang.
Moreover, during an adaptation of Socket-CAN plx_pci driver to Xenomai, 
I remarked that these drivers are too similar. In general, the 
differences consist only of slightly changed API (early version of 
Socket-CAN internal API) and different prefixes for some names.

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Sebastian Smolorz]

Hi Wolfgang,

Wolfgang Grandegger wrote:
> On 05/18/2010 01:42 PM, Sebastian Smolorz wrote:
> > Pavel Cheblakov wrote:
> >> This is a general driver for cards based on PLX90xx PCI-bridges.
> >> It supports following cards:
> >>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
> >>  - Adlink PCI-7841/cPCI-7841 SE card
> >>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
> >>  - esd CAN-PCI/PMC/266
> >>  - esd CAN-PCIe/2000
> >>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
> >>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
> >
> > The esd cards mentioned above are supported by the RTCAN driver
> > xeno_can_esd_pci. Why do you propose a new driver instead of extending
> > the existing one?
> 
> For Socket-CAN, this driver is supposed to support all PLX PCI based
> boards including the esd CAN PCI cards and also the IXXAT PCI board (not
> yet done, though). The RTCAN driver xeno_can_esd_pci is a *dedicated*
> driver for that card without generic support for the PLX PCI chips.
> Extending it makes little sense. The question is if we want to drop
> xeno_can_esd_pci and xeno_can_ixxat_pci.

With extending the esd_pci driver I meant to take this driver as a basis to 
add support for more cards. Of course this would mean to rename the driver 
in order to reflect that. I'm not against a unification of esd_pci and 
ixxat_pci but it's unneccessary work to write a new RTCAN driver if there 
exists another one which is tried and tested in the field. So my question 
was why  Pavel did not take the esd_pci driver as the starting position.

So talking about potential issues in Pavel's code (e.g. the function 
plx_pci_check_sja1000()) could be unnecessary if he derived the new driver 
from esd_pci .

-- 
Sebastian

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Wolfgang Grandegger]

On 05/18/2010 02:29 PM, Sebastian Smolorz wrote:
> Hi Wolfgang,
> 
> Wolfgang Grandegger wrote:
>> On 05/18/2010 01:42 PM, Sebastian Smolorz wrote:
>>> Pavel Cheblakov wrote:
>>>> This is a general driver for cards based on PLX90xx PCI-bridges.
>>>> It supports following cards:
>>>>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
>>>>  - Adlink PCI-7841/cPCI-7841 SE card
>>>>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
>>>>  - esd CAN-PCI/PMC/266
>>>>  - esd CAN-PCIe/2000
>>>>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
>>>>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
>>>
>>> The esd cards mentioned above are supported by the RTCAN driver
>>> xeno_can_esd_pci. Why do you propose a new driver instead of extending
>>> the existing one?
>>
>> For Socket-CAN, this driver is supposed to support all PLX PCI based
>> boards including the esd CAN PCI cards and also the IXXAT PCI board (not
>> yet done, though). The RTCAN driver xeno_can_esd_pci is a *dedicated*
>> driver for that card without generic support for the PLX PCI chips.
>> Extending it makes little sense. The question is if we want to drop
>> xeno_can_esd_pci and xeno_can_ixxat_pci.
> 
> With extending the esd_pci driver I meant to take this driver as a basis to 
> add support for more cards. Of course this would mean to rename the driver 
> in order to reflect that. I'm not against a unification of esd_pci and 
> ixxat_pci but it's unneccessary work to write a new RTCAN driver if there 
> exists another one which is tried and tested in the field. So my question 
> was why  Pavel did not take the esd_pci driver as the starting position.

He did *not* write a new RTCAN driver. He ported plx_pci.c from the
mainline kernel to rtcan, which is much less work and even less error prune.

> So talking about potential issues in Pavel's code (e.g. the function 
> plx_pci_check_sja1000()) could be unnecessary if he derived the new driver 
> from esd_pci .

Does it harm? It is necessary to find out the number of channels on some
cards.

Wolfgang.

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Sebastian Smolorz]

Wolfgang Grandegger wrote:
> On 05/18/2010 02:29 PM, Sebastian Smolorz wrote:
> > Hi Wolfgang,
> >
> > Wolfgang Grandegger wrote:
> >> On 05/18/2010 01:42 PM, Sebastian Smolorz wrote:
> >>> Pavel Cheblakov wrote:
> >>>> This is a general driver for cards based on PLX90xx PCI-bridges.
> >>>> It supports following cards:
> >>>>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
> >>>>  - Adlink PCI-7841/cPCI-7841 SE card
> >>>>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
> >>>>  - esd CAN-PCI/PMC/266
> >>>>  - esd CAN-PCIe/2000
> >>>>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
> >>>>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
> >>>
> >>> The esd cards mentioned above are supported by the RTCAN driver
> >>> xeno_can_esd_pci. Why do you propose a new driver instead of
> >>> extending the existing one?
> >>
> >> For Socket-CAN, this driver is supposed to support all PLX PCI based
> >> boards including the esd CAN PCI cards and also the IXXAT PCI board
> >> (not yet done, though). The RTCAN driver xeno_can_esd_pci is a
> >> *dedicated* driver for that card without generic support for the PLX
> >> PCI chips. Extending it makes little sense. The question is if we want
> >> to drop xeno_can_esd_pci and xeno_can_ixxat_pci.
> >
> > With extending the esd_pci driver I meant to take this driver as a
> > basis to add support for more cards. Of course this would mean to
> > rename the driver in order to reflect that. I'm not against a
> > unification of esd_pci and ixxat_pci but it's unneccessary work to
> > write a new RTCAN driver if there exists another one which is tried and
> > tested in the field. So my question was why  Pavel did not take the
> > esd_pci driver as the starting position.
> 
> He did *not* write a new RTCAN driver. He ported plx_pci.c from the
> mainline kernel to rtcan,

So it is a new RTCAN driver - never mind, just nitpicking. ;-)

As I said, I'm in favour of a unified driver which supports more cards than 
esd_pci and ixxat_pci. The question is how we proceed best when we include 
rt_sja1000_plx_pci. I suggest not to remove esd_pci immediately but rather 
mark it deprecated with a reference to the new driver, e.g. in the Kconfig 
help text. That way we would have a fallback driver in case of unforeseen 
issues with the new driver.

> which is much less work and even less error
>  prune.
> 
> > So talking about potential issues in Pavel's code (e.g. the function
> > plx_pci_check_sja1000()) could be unnecessary if he derived the new
> > driver from esd_pci .
> 
> Does it harm? It is necessary to find out the number of channels on some
> cards.

Yes, you are right.

-- 
Sebastian

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Wolfgang Grandegger]

On 05/18/2010 03:29 PM, Sebastian Smolorz wrote:
> Wolfgang Grandegger wrote:
>> On 05/18/2010 02:29 PM, Sebastian Smolorz wrote:
>>> Hi Wolfgang,
>>>
>>> Wolfgang Grandegger wrote:
>>>> On 05/18/2010 01:42 PM, Sebastian Smolorz wrote:
>>>>> Pavel Cheblakov wrote:
>>>>>> This is a general driver for cards based on PLX90xx PCI-bridges.
>>>>>> It supports following cards:
>>>>>>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
>>>>>>  - Adlink PCI-7841/cPCI-7841 SE card
>>>>>>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
>>>>>>  - esd CAN-PCI/PMC/266
>>>>>>  - esd CAN-PCIe/2000
>>>>>>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
>>>>>>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
>>>>>
>>>>> The esd cards mentioned above are supported by the RTCAN driver
>>>>> xeno_can_esd_pci. Why do you propose a new driver instead of
>>>>> extending the existing one?
>>>>
>>>> For Socket-CAN, this driver is supposed to support all PLX PCI based
>>>> boards including the esd CAN PCI cards and also the IXXAT PCI board
>>>> (not yet done, though). The RTCAN driver xeno_can_esd_pci is a
>>>> *dedicated* driver for that card without generic support for the PLX
>>>> PCI chips. Extending it makes little sense. The question is if we want
>>>> to drop xeno_can_esd_pci and xeno_can_ixxat_pci.
>>>
>>> With extending the esd_pci driver I meant to take this driver as a
>>> basis to add support for more cards. Of course this would mean to
>>> rename the driver in order to reflect that. I'm not against a
>>> unification of esd_pci and ixxat_pci but it's unneccessary work to
>>> write a new RTCAN driver if there exists another one which is tried and
>>> tested in the field. So my question was why  Pavel did not take the
>>> esd_pci driver as the starting position.
>>
>> He did *not* write a new RTCAN driver. He ported plx_pci.c from the
>> mainline kernel to rtcan,
> 
> So it is a new RTCAN driver - never mind, just nitpicking. ;-)
> 
> As I said, I'm in favour of a unified driver which supports more cards than 
> esd_pci and ixxat_pci. The question is how we proceed best when we include 
> rt_sja1000_plx_pci. I suggest not to remove esd_pci immediately but rather 
> mark it deprecated with a reference to the new driver, e.g. in the Kconfig 
> help text. That way we would have a fallback driver in case of unforeseen 
> issues with the new driver.

Fine with me. Marking it as deprecated is a good thing, though. Note
that the esd part of the plx_pci driver was contributed by Matthias
Fuchs, including support for the *new* esd pci cards.

Wolfgang.

Re: [Xenomai-core] [PATCH] rtcan: add rtcan_plx_pci driver

[Wolfgang Grandegger]

Hi Pavel,

On 05/18/2010 01:28 PM, Pavel Cheblakov wrote:
> This is a general driver for cards based on PLX90xx PCI-bridges.
> It supports following cards:
>  - Adlink PCI-7841/cPCI-7841 card (http://www.adlinktech.com/)
>  - Adlink PCI-7841/cPCI-7841 SE card
>  - esd CAN-PCI/CPCI/PCI104/200 (http://www.esd.eu/)
>  - esd CAN-PCI/PMC/266
>  - esd CAN-PCIe/2000
>  - Marathon CAN-bus-PCI card (http://www.marathon.ru/)
>  - TEWS TECHNOLOGIES TPMC810 card (http://www.tews.com/)
> 
> This driver has been ported from socketcan of mainline Linux kernel 2.6.34.
> Support of esd cards has been added by Matthias Fuchs in linux-next
> 
> Signed-off-by: Pavel Cheblakov <P.B.Cheblakov@domain.hid>

As discussed, could you please label the Kconfig entry for the old esd
PCI driver as deprecated. Apart from that the patch looks good and you
can add my "Acked-by: Wolfgang Grandegger <wg@domain.hid>".

Thanks,

Wolfgang.