MIPS: Add Cavium OCTEON processor support files to arch/mips/cavium-octeon.

These are the rest of the new files needed to add OCTEON processor
support to the Linux kernel.  Other than Makefile and Kconfig which
should be obvious, we have:

csrc-octeon.c   -- Clock source driver for OCTEON.
dma-octeon.c    -- Helper functions for mapping DMA memory.
flash_setup.c   -- Register on-board flash with the MTD subsystem.
octeon-irq.c    -- OCTEON interrupt controller managment.
octeon-memcpy.S -- Optimized memcpy() implementation.
serial.c        -- Register 8250 platform driver and early console.
setup.c         -- Early architecture initialization.
smp.c           -- OCTEON SMP support.
octeon_switch.S -- Scheduler context switch for OCTEON.
c-octeon.c      -- OCTEON cache controller support.
cex-oct.S       -- OCTEON cache exception handler.

asm/mach-cavium-octeon/*.h -- Architecture include files.
Signed-off-by: Tomaso Paoletti <tpaoletti@caviumnetworks.com>
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

 create mode 100644 arch/mips/cavium-octeon/Kconfig
 create mode 100644 arch/mips/cavium-octeon/Makefile
 create mode 100644 arch/mips/cavium-octeon/csrc-octeon.c
 create mode 100644 arch/mips/cavium-octeon/dma-octeon.c
 create mode 100644 arch/mips/cavium-octeon/flash_setup.c
 create mode 100644 arch/mips/cavium-octeon/octeon-irq.c
 create mode 100644 arch/mips/cavium-octeon/octeon-memcpy.S
 create mode 100644 arch/mips/cavium-octeon/serial.c
 create mode 100644 arch/mips/cavium-octeon/setup.c
 create mode 100644 arch/mips/cavium-octeon/smp.c
 create mode 100644 arch/mips/include/asm/mach-cavium-octeon/cpu-feature-overrides.h
 create mode 100644 arch/mips/include/asm/mach-cavium-octeon/dma-coherence.h
 create mode 100644 arch/mips/include/asm/mach-cavium-octeon/irq.h
 create mode 100644 arch/mips/include/asm/mach-cavium-octeon/kernel-entry-init.h
 create mode 100644 arch/mips/include/asm/mach-cavium-octeon/war.h
 create mode 100644 arch/mips/include/asm/octeon/octeon.h
 create mode 100644 arch/mips/kernel/octeon_switch.S
 create mode 100644 arch/mips/mm/c-octeon.c
 create mode 100644 arch/mips/mm/cex-oct.S

arch/mips/cavium-octeon/Kconfig

+config CAVIUM_OCTEON_SPECIFIC_OPTIONS
+	bool "Enable Octeon specific options"
+	depends on CPU_CAVIUM_OCTEON
+	default "y"
+
+config CAVIUM_OCTEON_2ND_KERNEL
+	bool "Build the kernel to be used as a 2nd kernel on the same chip"
+	depends on CAVIUM_OCTEON_SPECIFIC_OPTIONS
+	default "n"
+	help
+	  This option configures this kernel to be linked at a different
+	  address and use the 2nd uart for output. This allows a kernel built
+	  with this option to be run at the same time as one built without this
+	  option.
+
+config CAVIUM_OCTEON_HW_FIX_UNALIGNED
+	bool "Enable hardware fixups of unaligned loads and stores"
+	depends on CAVIUM_OCTEON_SPECIFIC_OPTIONS
+	default "y"
+	help
+	  Configure the Octeon hardware to automatically fix unaligned loads
+	  and stores. Normally unaligned accesses are fixed using a kernel
+	  exception handler. This option enables the hardware automatic fixups,
+	  which requires only an extra 3 cycles. Disable this option if you
+	  are running code that relies on address exceptions on unaligned
+	  accesses.
+
+config CAVIUM_OCTEON_CVMSEG_SIZE
+	int "Number of L1 cache lines reserved for CVMSEG memory"
+	depends on CAVIUM_OCTEON_SPECIFIC_OPTIONS
+	range 0 54
+	default 1
+	help
+	  CVMSEG LM is a segment that accesses portions of the dcache as a
+	  local memory; the larger CVMSEG is, the smaller the cache is.
+	  This selects the size of CVMSEG LM, which is in cache blocks. The
+	  legally range is from zero to 54 cache blocks (i.e. CVMSEG LM is
+	  between zero and 6192 bytes).
+
+config CAVIUM_OCTEON_LOCK_L2
+	bool "Lock often used kernel code in the L2"
+	depends on CAVIUM_OCTEON_SPECIFIC_OPTIONS
+	default "y"
+	help
+	  Enable locking parts of the kernel into the L2 cache.
+
+config CAVIUM_OCTEON_LOCK_L2_TLB
+	bool "Lock the TLB handler in L2"
+	depends on CAVIUM_OCTEON_LOCK_L2
+	default "y"
+	help
+	  Lock the low level TLB fast path into L2.
+
+config CAVIUM_OCTEON_LOCK_L2_EXCEPTION
+	bool "Lock the exception handler in L2"
+	depends on CAVIUM_OCTEON_LOCK_L2
+	default "y"
+	help
+	  Lock the low level exception handler into L2.
+
+config CAVIUM_OCTEON_LOCK_L2_LOW_LEVEL_INTERRUPT
+	bool "Lock the interrupt handler in L2"
+	depends on CAVIUM_OCTEON_LOCK_L2
+	default "y"
+	help
+	  Lock the low level interrupt handler into L2.
+
+config CAVIUM_OCTEON_LOCK_L2_INTERRUPT
+	bool "Lock the 2nd level interrupt handler in L2"
+	depends on CAVIUM_OCTEON_LOCK_L2
+	default "y"
+	help
+	  Lock the 2nd level interrupt handler in L2.
+
+config CAVIUM_OCTEON_LOCK_L2_MEMCPY
+	bool "Lock memcpy() in L2"
+	depends on CAVIUM_OCTEON_LOCK_L2
+	default "y"
+	help
+	  Lock the kernel's implementation of memcpy() into L2.
+
+config ARCH_SPARSEMEM_ENABLE
+	def_bool y
+	select SPARSEMEM_STATIC
+	depends on CPU_CAVIUM_OCTEON

arch/mips/cavium-octeon/Makefile

+#
+# Makefile for the Cavium Octeon specific kernel interface routines
+# under Linux.
+#
+# This file is subject to the terms and conditions of the GNU General Public
+# License.  See the file "COPYING" in the main directory of this archive
+# for more details.
+#
+# Copyright (C) 2005-2008 Cavium Networks
+#
+
+obj-y := setup.o serial.o octeon-irq.o csrc-octeon.o
+obj-y += dma-octeon.o flash_setup.o
+obj-y += octeon-memcpy.o
+
+obj-$(CONFIG_SMP)                     += smp.o

arch/mips/cavium-octeon/csrc-octeon.c

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2007 by Ralf Baechle
+ */
+#include <linux/clocksource.h>
+#include <linux/init.h>
+
+#include <asm/time.h>
+
+#include <asm/octeon/octeon.h>
+#include <asm/octeon/cvmx-ipd-defs.h>
+
+/*
+ * Set the current core's cvmcount counter to the value of the
+ * IPD_CLK_COUNT.  We do this on all cores as they are brought
+ * on-line.  This allows for a read from a local cpu register to
+ * access a synchronized counter.
+ *
+ */
+void octeon_init_cvmcount(void)
+{
+	unsigned long flags;
+	unsigned loops = 2;
+
+	/* Clobber loops so GCC will not unroll the following while loop. */
+	asm("" : "+r" (loops));
+
+	local_irq_save(flags);
+	/*
+	 * Loop several times so we are executing from the cache,
+	 * which should give more deterministic timing.
+	 */
+	while (loops--)
+		write_c0_cvmcount(cvmx_read_csr(CVMX_IPD_CLK_COUNT));
+	local_irq_restore(flags);
+}
+
+static cycle_t octeon_cvmcount_read(void)
+{
+	return read_c0_cvmcount();
+}
+
+static struct clocksource clocksource_mips = {
+	.name		= "OCTEON_CVMCOUNT",
+	.read		= octeon_cvmcount_read,
+	.mask		= CLOCKSOURCE_MASK(64),
+	.flags		= CLOCK_SOURCE_IS_CONTINUOUS,
+};
+
+void __init plat_time_init(void)
+{
+	clocksource_mips.rating = 300;
+	clocksource_set_clock(&clocksource_mips, mips_hpt_frequency);
+	clocksource_register(&clocksource_mips);
+}

arch/mips/cavium-octeon/dma-octeon.c

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2000  Ani Joshi <ajoshi@unixbox.com>
+ * Copyright (C) 2000, 2001  Ralf Baechle <ralf@gnu.org>
+ * Copyright (C) 2005 Ilya A. Volynets-Evenbakh <ilya@total-knowledge.com>
+ * swiped from i386, and cloned for MIPS by Geert, polished by Ralf.
+ * IP32 changes by Ilya.
+ * Cavium Networks: Create new dma setup for Cavium Networks Octeon based on
+ * the kernels original.
+ */
+#include <linux/types.h>
+#include <linux/mm.h>
+
+#include <dma-coherence.h>
+
+dma_addr_t octeon_map_dma_mem(struct device *dev, void *ptr, size_t size)
+{
+	/* Without PCI/PCIe this function can be called for Octeon internal
+	   devices such as USB. These devices all support 64bit addressing */
+	mb();
+	return virt_to_phys(ptr);
+}
+
+void octeon_unmap_dma_mem(struct device *dev, dma_addr_t dma_addr)
+{
+	/* Without PCI/PCIe this function can be called for Octeon internal
+	 * devices such as USB. These devices all support 64bit addressing */
+	return;
+}

arch/mips/cavium-octeon/flash_setup.c

+/*
+ *   Octeon Bootbus flash setup
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2007, 2008 Cavium Networks
+ */
+#include <linux/kernel.h>
+#include <linux/mtd/mtd.h>
+#include <linux/mtd/map.h>
+#include <linux/mtd/partitions.h>
+
+#include <asm/octeon/octeon.h>
+
+static struct map_info flash_map;
+static struct mtd_info *mymtd;
+#ifdef CONFIG_MTD_PARTITIONS
+static int nr_parts;
+static struct mtd_partition *parts;
+static const char *part_probe_types[] = {
+	"cmdlinepart",
+#ifdef CONFIG_MTD_REDBOOT_PARTS
+	"RedBoot",
+#endif
+	NULL
+};
+#endif
+
+/**
+ * Module/ driver initialization.
+ *
+ * Returns Zero on success
+ */
+static int __init flash_init(void)
+{
+	/*
+	 * Read the bootbus region 0 setup to determine the base
+	 * address of the flash.
+	 */
+	union cvmx_mio_boot_reg_cfgx region_cfg;
+	region_cfg.u64 = cvmx_read_csr(CVMX_MIO_BOOT_REG_CFGX(0));
+	if (region_cfg.s.en) {
+		/*
+		 * The bootloader always takes the flash and sets its
+		 * address so the entire flash fits below
+		 * 0x1fc00000. This way the flash aliases to
+		 * 0x1fc00000 for booting. Software can access the
+		 * full flash at the true address, while core boot can
+		 * access 4MB.
+		 */
+		/* Use this name so old part lines work */
+		flash_map.name = "phys_mapped_flash";
+		flash_map.phys = region_cfg.s.base << 16;
+		flash_map.size = 0x1fc00000 - flash_map.phys;
+		flash_map.bankwidth = 1;
+		flash_map.virt = ioremap(flash_map.phys, flash_map.size);
+		pr_notice("Bootbus flash: Setting flash for %luMB flash at "
+			  "0x%08lx\n", flash_map.size >> 20, flash_map.phys);
+		simple_map_init(&flash_map);
+		mymtd = do_map_probe("cfi_probe", &flash_map);
+		if (mymtd) {
+			mymtd->owner = THIS_MODULE;
+
+#ifdef CONFIG_MTD_PARTITIONS
+			nr_parts = parse_mtd_partitions(mymtd,
+							part_probe_types,
+							&parts, 0);
+			if (nr_parts > 0)
+				add_mtd_partitions(mymtd, parts, nr_parts);
+			else
+				add_mtd_device(mymtd);
+#else
+			add_mtd_device(mymtd);
+#endif
+		} else {
+			pr_err("Failed to register MTD device for flash\n");
+		}
+	}
+	return 0;
+}
+
+late_initcall(flash_init);

arch/mips/cavium-octeon/serial.c

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2004-2007 Cavium Networks
+ */
+#include <linux/console.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+#include <linux/serial.h>
+#include <linux/serial_8250.h>
+#include <linux/serial_reg.h>
+#include <linux/tty.h>
+
+#include <asm/time.h>
+
+#include <asm/octeon/octeon.h>
+
+#ifdef CONFIG_GDB_CONSOLE
+#define DEBUG_UART 0
+#else
+#define DEBUG_UART 1
+#endif
+
+unsigned int octeon_serial_in(struct uart_port *up, int offset)
+{
+	int rv = cvmx_read_csr((uint64_t)(up->membase + (offset << 3)));
+	if (offset == UART_IIR && (rv & 0xf) == 7) {
+		/* Busy interrupt, read the USR (39) and try again. */
+		cvmx_read_csr((uint64_t)(up->membase + (39 << 3)));
+		rv = cvmx_read_csr((uint64_t)(up->membase + (offset << 3)));
+	}
+	return rv;
+}
+
+void octeon_serial_out(struct uart_port *up, int offset, int value)
+{
+	/*
+	 * If bits 6 or 7 of the OCTEON UART's LCR are set, it quits
+	 * working.
+	 */
+	if (offset == UART_LCR)
+		value &= 0x9f;
+	cvmx_write_csr((uint64_t)(up->membase + (offset << 3)), (u8)value);
+}
+
+/*
+ * Allocated in .bss, so it is all zeroed.
+ */
+#define OCTEON_MAX_UARTS 3
+static struct plat_serial8250_port octeon_uart8250_data[OCTEON_MAX_UARTS + 1];
+static struct platform_device octeon_uart8250_device = {
+	.name			= "serial8250",
+	.id			= PLAT8250_DEV_PLATFORM,
+	.dev			= {
+		.platform_data	= octeon_uart8250_data,
+	},
+};
+
+static void __init octeon_uart_set_common(struct plat_serial8250_port *p)
+{
+	p->flags = ASYNC_SKIP_TEST | UPF_SHARE_IRQ | UPF_FIXED_TYPE;
+	p->type = PORT_OCTEON;
+	p->iotype = UPIO_MEM;
+	p->regshift = 3;	/* I/O addresses are every 8 bytes */
+	p->uartclk = mips_hpt_frequency;
+	p->serial_in = octeon_serial_in;
+	p->serial_out = octeon_serial_out;
+}
+
+static int __init octeon_serial_init(void)
+{
+	int enable_uart0;
+	int enable_uart1;
+	int enable_uart2;
+	struct plat_serial8250_port *p;
+
+#ifdef CONFIG_CAVIUM_OCTEON_2ND_KERNEL
+	/*
+	 * If we are configured to run as the second of two kernels,
+	 * disable uart0 and enable uart1. Uart0 is owned by the first
+	 * kernel
+	 */
+	enable_uart0 = 0;
+	enable_uart1 = 1;
+#else
+	/*
+	 * We are configured for the first kernel. We'll enable uart0
+	 * if the bootloader told us to use 0, otherwise will enable
+	 * uart 1.
+	 */
+	enable_uart0 = (octeon_get_boot_uart() == 0);
+	enable_uart1 = (octeon_get_boot_uart() == 1);
+#ifdef CONFIG_KGDB
+	enable_uart1 = 1;
+#endif
+#endif
+
+	/* Right now CN52XX is the only chip with a third uart */
+	enable_uart2 = OCTEON_IS_MODEL(OCTEON_CN52XX);
+
+	p = octeon_uart8250_data;
+	if (enable_uart0) {
+		/* Add a ttyS device for hardware uart 0 */
+		octeon_uart_set_common(p);
+		p->membase = (void *) CVMX_MIO_UARTX_RBR(0);
+		p->mapbase = CVMX_MIO_UARTX_RBR(0) & ((1ull << 49) - 1);
+		p->irq = OCTEON_IRQ_UART0;
+		p++;
+	}
+
+	if (enable_uart1) {
+		/* Add a ttyS device for hardware uart 1 */
+		octeon_uart_set_common(p);
+		p->membase = (void *) CVMX_MIO_UARTX_RBR(1);
+		p->mapbase = CVMX_MIO_UARTX_RBR(1) & ((1ull << 49) - 1);
+		p->irq = OCTEON_IRQ_UART1;
+		p++;
+	}
+	if (enable_uart2) {
+		/* Add a ttyS device for hardware uart 2 */
+		octeon_uart_set_common(p);
+		p->membase = (void *) CVMX_MIO_UART2_RBR;
+		p->mapbase = CVMX_MIO_UART2_RBR & ((1ull << 49) - 1);
+		p->irq = OCTEON_IRQ_UART2;
+		p++;
+	}
+
+	BUG_ON(p > &octeon_uart8250_data[OCTEON_MAX_UARTS]);
+
+	return platform_device_register(&octeon_uart8250_device);
+}
+
+device_initcall(octeon_serial_init);

arch/mips/cavium-octeon/smp.c

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2004-2008 Cavium Networks
+ */
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/smp.h>
+#include <linux/interrupt.h>
+#include <linux/kernel_stat.h>
+#include <linux/sched.h>
+#include <linux/module.h>
+
+#include <asm/mmu_context.h>
+#include <asm/system.h>
+#include <asm/time.h>
+
+#include <asm/octeon/octeon.h>
+
+volatile unsigned long octeon_processor_boot = 0xff;
+volatile unsigned long octeon_processor_sp;
+volatile unsigned long octeon_processor_gp;
+
+static irqreturn_t mailbox_interrupt(int irq, void *dev_id)
+{
+	const int coreid = cvmx_get_core_num();
+	uint64_t action;
+
+	/* Load the mailbox register to figure out what we're supposed to do */
+	action = cvmx_read_csr(CVMX_CIU_MBOX_CLRX(coreid));
+
+	/* Clear the mailbox to clear the interrupt */
+	cvmx_write_csr(CVMX_CIU_MBOX_CLRX(coreid), action);
+
+	if (action & SMP_CALL_FUNCTION)
+		smp_call_function_interrupt();
+
+	/* Check if we've been told to flush the icache */
+	if (action & SMP_ICACHE_FLUSH)
+		asm volatile ("synci 0($0)\n");
+	return IRQ_HANDLED;
+}
+
+/**
+ * Cause the function described by call_data to be executed on the passed
+ * cpu.  When the function has finished, increment the finished field of
+ * call_data.
+ */
+void octeon_send_ipi_single(int cpu, unsigned int action)
+{
+	int coreid = cpu_logical_map(cpu);
+	/*
+	pr_info("SMP: Mailbox send cpu=%d, coreid=%d, action=%u\n", cpu,
+	       coreid, action);
+	*/
+	cvmx_write_csr(CVMX_CIU_MBOX_SETX(coreid), action);
+}
+
+static inline void octeon_send_ipi_mask(cpumask_t mask, unsigned int action)
+{
+	unsigned int i;
+
+	for_each_cpu_mask(i, mask)
+		octeon_send_ipi_single(i, action);
+}
+
+/**
+ * Detect available CPUs, populate phys_cpu_present_map
+ */
+static void octeon_smp_setup(void)
+{
+	const int coreid = cvmx_get_core_num();
+	int cpus;
+	int id;
+
+	int core_mask = octeon_get_boot_coremask();
+
+	cpus_clear(cpu_possible_map);
+	__cpu_number_map[coreid] = 0;
+	__cpu_logical_map[0] = coreid;
+	cpu_set(0, cpu_possible_map);
+
+	cpus = 1;
+	for (id = 0; id < 16; id++) {
+		if ((id != coreid) && (core_mask & (1 << id))) {
+			cpu_set(cpus, cpu_possible_map);
+			__cpu_number_map[id] = cpus;
+			__cpu_logical_map[cpus] = id;
+			cpus++;
+		}
+	}
+}
+
+/**
+ * Firmware CPU startup hook
+ *
+ */
+static void octeon_boot_secondary(int cpu, struct task_struct *idle)
+{
+	int count;
+
+	pr_info("SMP: Booting CPU%02d (CoreId %2d)...\n", cpu,
+		cpu_logical_map(cpu));
+
+	octeon_processor_sp = __KSTK_TOS(idle);
+	octeon_processor_gp = (unsigned long)(task_thread_info(idle));
+	octeon_processor_boot = cpu_logical_map(cpu);
+	mb();
+
+	count = 10000;
+	while (octeon_processor_sp && count) {
+		/* Waiting for processor to get the SP and GP */
+		udelay(1);
+		count--;
+	}
+	if (count == 0)
+		pr_err("Secondary boot timeout\n");
+}
+
+/**
+ * After we've done initial boot, this function is called to allow the
+ * board code to clean up state, if needed
+ */
+static void octeon_init_secondary(void)
+{
+	const int coreid = cvmx_get_core_num();
+	union cvmx_ciu_intx_sum0 interrupt_enable;
+
+	octeon_check_cpu_bist();
+	octeon_init_cvmcount();
+	/*
+	pr_info("SMP: CPU%d (CoreId %lu) started\n", cpu, coreid);
+	*/
+	/* Enable Mailbox interrupts to this core. These are the only
+	   interrupts allowed on line 3 */
+	cvmx_write_csr(CVMX_CIU_MBOX_CLRX(coreid), 0xffffffff);
+	interrupt_enable.u64 = 0;
+	interrupt_enable.s.mbox = 0x3;
+	cvmx_write_csr(CVMX_CIU_INTX_EN0((coreid * 2)), interrupt_enable.u64);
+	cvmx_write_csr(CVMX_CIU_INTX_EN0((coreid * 2 + 1)), 0);
+	cvmx_write_csr(CVMX_CIU_INTX_EN1((coreid * 2)), 0);
+	cvmx_write_csr(CVMX_CIU_INTX_EN1((coreid * 2 + 1)), 0);
+	/* Enable core interrupt processing for 2,3 and 7 */
+	set_c0_status(0x8c01);
+}
+
+/**
+ * Callout to firmware before smp_init
+ *
+ */
+void octeon_prepare_cpus(unsigned int max_cpus)
+{
+	cvmx_write_csr(CVMX_CIU_MBOX_CLRX(cvmx_get_core_num()), 0xffffffff);
+	if (request_irq(OCTEON_IRQ_MBOX0, mailbox_interrupt, IRQF_SHARED,
+			"mailbox0", mailbox_interrupt)) {
+		panic("Cannot request_irq(OCTEON_IRQ_MBOX0)\n");
+	}
+	if (request_irq(OCTEON_IRQ_MBOX1, mailbox_interrupt, IRQF_SHARED,
+			"mailbox1", mailbox_interrupt)) {
+		panic("Cannot request_irq(OCTEON_IRQ_MBOX1)\n");
+	}
+}
+
+/**
+ * Last chance for the board code to finish SMP initialization before
+ * the CPU is "online".
+ */
+static void octeon_smp_finish(void)
+{
+#ifdef CONFIG_CAVIUM_GDB
+	unsigned long tmp;
+	/* Pulse MCD0 signal on Ctrl-C to stop all the cores. Also set the MCD0
+	   to be not masked by this core so we know the signal is received by
+	   someone */
+	asm volatile ("dmfc0 %0, $22\n"
+		      "ori   %0, %0, 0x9100\n" "dmtc0 %0, $22\n" : "=r" (tmp));
+#endif
+
+	octeon_user_io_init();
+
+	/* to generate the first CPU timer interrupt */
+	write_c0_compare(read_c0_count() + mips_hpt_frequency / HZ);
+}
+
+/**
+ * Hook for after all CPUs are online
+ */
+static void octeon_cpus_done(void)
+{
+#ifdef CONFIG_CAVIUM_GDB
+	unsigned long tmp;
+	/* Pulse MCD0 signal on Ctrl-C to stop all the cores. Also set the MCD0
+	   to be not masked by this core so we know the signal is received by
+	   someone */
+	asm volatile ("dmfc0 %0, $22\n"
+		      "ori   %0, %0, 0x9100\n" "dmtc0 %0, $22\n" : "=r" (tmp));
+#endif
+}
+
+struct plat_smp_ops octeon_smp_ops = {
+	.send_ipi_single	= octeon_send_ipi_single,
+	.send_ipi_mask		= octeon_send_ipi_mask,
+	.init_secondary		= octeon_init_secondary,
+	.smp_finish		= octeon_smp_finish,
+	.cpus_done		= octeon_cpus_done,
+	.boot_secondary		= octeon_boot_secondary,
+	.smp_setup		= octeon_smp_setup,
+	.prepare_cpus		= octeon_prepare_cpus,
+};

arch/mips/include/asm/mach-cavium-octeon/cpu-feature-overrides.h

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2004 Cavium Networks
+ */
+#ifndef __ASM_MACH_CAVIUM_OCTEON_CPU_FEATURE_OVERRIDES_H
+#define __ASM_MACH_CAVIUM_OCTEON_CPU_FEATURE_OVERRIDES_H
+
+#include <linux/types.h>
+#include <asm/mipsregs.h>
+
+/*
+ * Cavium Octeons are MIPS64v2 processors
+ */
+#define cpu_dcache_line_size()	128
+#define cpu_icache_line_size()	128
+
+
+#define cpu_has_4kex		1
+#define cpu_has_3k_cache	0
+#define cpu_has_4k_cache	0
+#define cpu_has_tx39_cache	0
+#define cpu_has_fpu		0
+#define cpu_has_counter		1
+#define cpu_has_watch		1
+#define cpu_has_divec		1
+#define cpu_has_vce		0
+#define cpu_has_cache_cdex_p	0
+#define cpu_has_cache_cdex_s	0
+#define cpu_has_prefetch	1
+
+/*
+ * We should disable LL/SC on non SMP systems as it is faster to
+ * disable interrupts for atomic access than a LL/SC.  Unfortunatly we
+ * cannot as this breaks asm/futex.h
+ */
+#define cpu_has_llsc		1
+#define cpu_has_vtag_icache	1
+#define cpu_has_dc_aliases	0
+#define cpu_has_ic_fills_f_dc	0
+#define cpu_has_64bits		1
+#define cpu_has_octeon_cache	1
+#define cpu_has_saa		octeon_has_saa()
+#define cpu_has_mips32r1	0
+#define cpu_has_mips32r2	0
+#define cpu_has_mips64r1	0
+#define cpu_has_mips64r2	1
+#define cpu_has_dsp		0
+#define cpu_has_mipsmt		0
+#define cpu_has_userlocal	0
+#define cpu_has_vint		0
+#define cpu_has_veic		0
+#define ARCH_HAS_READ_CURRENT_TIMER 1
+#define ARCH_HAS_IRQ_PER_CPU	1
+#define ARCH_HAS_SPINLOCK_PREFETCH 1
+#define spin_lock_prefetch(x) prefetch(x)
+#define PREFETCH_STRIDE 128
+
+static inline int read_current_timer(unsigned long *result)
+{
+	asm volatile ("rdhwr %0,$31\n"
+#ifndef CONFIG_64BIT
+		      "\tsll %0, 0"
+#endif
+		      : "=r" (*result));
+	return 0;
+}
+
+static inline int octeon_has_saa(void)
+{
+	int id;
+	asm volatile ("mfc0 %0, $15,0" : "=r" (id));
+	return id >= 0x000d0300;
+}
+
+#endif

arch/mips/include/asm/mach-cavium-octeon/dma-coherence.h

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2006  Ralf Baechle <ralf@linux-mips.org>
+ *
+ *
+ * Similar to mach-generic/dma-coherence.h except
+ * plat_device_is_coherent hard coded to return 1.
+ *
+ */
+#ifndef __ASM_MACH_CAVIUM_OCTEON_DMA_COHERENCE_H
+#define __ASM_MACH_CAVIUM_OCTEON_DMA_COHERENCE_H
+
+struct device;
+
+dma_addr_t octeon_map_dma_mem(struct device *, void *, size_t);
+void octeon_unmap_dma_mem(struct device *, dma_addr_t);
+
+static inline dma_addr_t plat_map_dma_mem(struct device *dev, void *addr,
+	size_t size)
+{
+	return octeon_map_dma_mem(dev, addr, size);
+}
+
+static inline dma_addr_t plat_map_dma_mem_page(struct device *dev,
+	struct page *page)
+{
+	return octeon_map_dma_mem(dev, page_address(page), PAGE_SIZE);
+}
+
+static inline unsigned long plat_dma_addr_to_phys(dma_addr_t dma_addr)
+{
+	return dma_addr;
+}
+
+static inline void plat_unmap_dma_mem(struct device *dev, dma_addr_t dma_addr)
+{
+	octeon_unmap_dma_mem(dev, dma_addr);
+}
+
+static inline int plat_dma_supported(struct device *dev, u64 mask)
+{
+	return 1;
+}
+
+static inline void plat_extra_sync_for_device(struct device *dev)
+{
+	mb();
+}
+
+static inline int plat_device_is_coherent(struct device *dev)
+{
+	return 1;
+}
+
+static inline int plat_dma_mapping_error(struct device *dev,
+					 dma_addr_t dma_addr)
+{
+	return dma_addr == -1;
+}
+
+#endif /* __ASM_MACH_CAVIUM_OCTEON_DMA_COHERENCE_H */

arch/mips/include/asm/mach-cavium-octeon/irq.h

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2004-2008 Cavium Networks
+ */
+#ifndef __OCTEON_IRQ_H__
+#define __OCTEON_IRQ_H__
+
+#define NR_IRQS OCTEON_IRQ_LAST
+#define MIPS_CPU_IRQ_BASE OCTEON_IRQ_SW0
+
+/* 0 - 7 represent the i8259 master */
+#define OCTEON_IRQ_I8259M0	0
+#define OCTEON_IRQ_I8259M1	1
+#define OCTEON_IRQ_I8259M2	2
+#define OCTEON_IRQ_I8259M3	3
+#define OCTEON_IRQ_I8259M4	4
+#define OCTEON_IRQ_I8259M5	5
+#define OCTEON_IRQ_I8259M6	6
+#define OCTEON_IRQ_I8259M7	7
+/* 8 - 15 represent the i8259 slave */
+#define OCTEON_IRQ_I8259S0	8
+#define OCTEON_IRQ_I8259S1	9
+#define OCTEON_IRQ_I8259S2	10
+#define OCTEON_IRQ_I8259S3	11
+#define OCTEON_IRQ_I8259S4	12
+#define OCTEON_IRQ_I8259S5	13
+#define OCTEON_IRQ_I8259S6	14
+#define OCTEON_IRQ_I8259S7	15
+/* 16 - 23 represent the 8 MIPS standard interrupt sources */
+#define OCTEON_IRQ_SW0		16
+#define OCTEON_IRQ_SW1		17
+#define OCTEON_IRQ_CIU0		18
+#define OCTEON_IRQ_CIU1		19
+#define OCTEON_IRQ_CIU4		20
+#define OCTEON_IRQ_5		21
+#define OCTEON_IRQ_PERF		22
+#define OCTEON_IRQ_TIMER	23
+/* 24 - 87 represent the sources in CIU_INTX_EN0 */
+#define OCTEON_IRQ_WORKQ0	24
+#define OCTEON_IRQ_WORKQ1	25
+#define OCTEON_IRQ_WORKQ2	26
+#define OCTEON_IRQ_WORKQ3	27
+#define OCTEON_IRQ_WORKQ4	28
+#define OCTEON_IRQ_WORKQ5	29
+#define OCTEON_IRQ_WORKQ6	30
+#define OCTEON_IRQ_WORKQ7	31
+#define OCTEON_IRQ_WORKQ8	32
+#define OCTEON_IRQ_WORKQ9	33
+#define OCTEON_IRQ_WORKQ10	34
+#define OCTEON_IRQ_WORKQ11	35
+#define OCTEON_IRQ_WORKQ12	36
+#define OCTEON_IRQ_WORKQ13	37
+#define OCTEON_IRQ_WORKQ14	38
+#define OCTEON_IRQ_WORKQ15	39
+#define OCTEON_IRQ_GPIO0	40
+#define OCTEON_IRQ_GPIO1	41
+#define OCTEON_IRQ_GPIO2	42
+#define OCTEON_IRQ_GPIO3	43
+#define OCTEON_IRQ_GPIO4	44
+#define OCTEON_IRQ_GPIO5	45
+#define OCTEON_IRQ_GPIO6	46
+#define OCTEON_IRQ_GPIO7	47
+#define OCTEON_IRQ_GPIO8	48
+#define OCTEON_IRQ_GPIO9	49
+#define OCTEON_IRQ_GPIO10	50
+#define OCTEON_IRQ_GPIO11	51
+#define OCTEON_IRQ_GPIO12	52
+#define OCTEON_IRQ_GPIO13	53
+#define OCTEON_IRQ_GPIO14	54
+#define OCTEON_IRQ_GPIO15	55
+#define OCTEON_IRQ_MBOX0	56
+#define OCTEON_IRQ_MBOX1	57
+#define OCTEON_IRQ_UART0	58
+#define OCTEON_IRQ_UART1	59
+#define OCTEON_IRQ_PCI_INT0	60
+#define OCTEON_IRQ_PCI_INT1	61
+#define OCTEON_IRQ_PCI_INT2	62
+#define OCTEON_IRQ_PCI_INT3	63
+#define OCTEON_IRQ_PCI_MSI0	64
+#define OCTEON_IRQ_PCI_MSI1	65
+#define OCTEON_IRQ_PCI_MSI2	66
+#define OCTEON_IRQ_PCI_MSI3	67
+#define OCTEON_IRQ_RESERVED68	68	/* Summary of CIU_INT_SUM1 */
+#define OCTEON_IRQ_TWSI		69
+#define OCTEON_IRQ_RML		70
+#define OCTEON_IRQ_TRACE	71
+#define OCTEON_IRQ_GMX_DRP0	72
+#define OCTEON_IRQ_GMX_DRP1	73
+#define OCTEON_IRQ_IPD_DRP	74
+#define OCTEON_IRQ_KEY_ZERO	75
+#define OCTEON_IRQ_TIMER0	76
+#define OCTEON_IRQ_TIMER1	77
+#define OCTEON_IRQ_TIMER2	78
+#define OCTEON_IRQ_TIMER3	79
+#define OCTEON_IRQ_USB0		80
+#define OCTEON_IRQ_PCM		81
+#define OCTEON_IRQ_MPI		82
+#define OCTEON_IRQ_TWSI2	83
+#define OCTEON_IRQ_POWIQ	84
+#define OCTEON_IRQ_IPDPPTHR	85
+#define OCTEON_IRQ_MII0		86
+#define OCTEON_IRQ_BOOTDMA	87
+/* 88 - 151 represent the sources in CIU_INTX_EN1 */
+#define OCTEON_IRQ_WDOG0	88
+#define OCTEON_IRQ_WDOG1	89
+#define OCTEON_IRQ_WDOG2	90
+#define OCTEON_IRQ_WDOG3	91
+#define OCTEON_IRQ_WDOG4	92
+#define OCTEON_IRQ_WDOG5	93
+#define OCTEON_IRQ_WDOG6	94
+#define OCTEON_IRQ_WDOG7	95
+#define OCTEON_IRQ_WDOG8	96
+#define OCTEON_IRQ_WDOG9	97
+#define OCTEON_IRQ_WDOG10	98
+#define OCTEON_IRQ_WDOG11	99
+#define OCTEON_IRQ_WDOG12	100
+#define OCTEON_IRQ_WDOG13	101
+#define OCTEON_IRQ_WDOG14	102
+#define OCTEON_IRQ_WDOG15	103
+#define OCTEON_IRQ_UART2	104
+#define OCTEON_IRQ_USB1		105
+#define OCTEON_IRQ_MII1		106
+#define OCTEON_IRQ_RESERVED107	107
+#define OCTEON_IRQ_RESERVED108	108
+#define OCTEON_IRQ_RESERVED109	109
+#define OCTEON_IRQ_RESERVED110	110
+#define OCTEON_IRQ_RESERVED111	111
+#define OCTEON_IRQ_RESERVED112	112
+#define OCTEON_IRQ_RESERVED113	113
+#define OCTEON_IRQ_RESERVED114	114
+#define OCTEON_IRQ_RESERVED115	115
+#define OCTEON_IRQ_RESERVED116	116
+#define OCTEON_IRQ_RESERVED117	117
+#define OCTEON_IRQ_RESERVED118	118
+#define OCTEON_IRQ_RESERVED119	119
+#define OCTEON_IRQ_RESERVED120	120
+#define OCTEON_IRQ_RESERVED121	121
+#define OCTEON_IRQ_RESERVED122	122
+#define OCTEON_IRQ_RESERVED123	123
+#define OCTEON_IRQ_RESERVED124	124
+#define OCTEON_IRQ_RESERVED125	125
+#define OCTEON_IRQ_RESERVED126	126
+#define OCTEON_IRQ_RESERVED127	127
+#define OCTEON_IRQ_RESERVED128	128
+#define OCTEON_IRQ_RESERVED129	129
+#define OCTEON_IRQ_RESERVED130	130
+#define OCTEON_IRQ_RESERVED131	131
+#define OCTEON_IRQ_RESERVED132	132
+#define OCTEON_IRQ_RESERVED133	133
+#define OCTEON_IRQ_RESERVED134	134
+#define OCTEON_IRQ_RESERVED135	135
+#define OCTEON_IRQ_RESERVED136	136
+#define OCTEON_IRQ_RESERVED137	137
+#define OCTEON_IRQ_RESERVED138	138
+#define OCTEON_IRQ_RESERVED139	139
+#define OCTEON_IRQ_RESERVED140	140
+#define OCTEON_IRQ_RESERVED141	141
+#define OCTEON_IRQ_RESERVED142	142
+#define OCTEON_IRQ_RESERVED143	143
+#define OCTEON_IRQ_RESERVED144	144
+#define OCTEON_IRQ_RESERVED145	145
+#define OCTEON_IRQ_RESERVED146	146
+#define OCTEON_IRQ_RESERVED147	147
+#define OCTEON_IRQ_RESERVED148	148
+#define OCTEON_IRQ_RESERVED149	149
+#define OCTEON_IRQ_RESERVED150	150
+#define OCTEON_IRQ_RESERVED151	151
+
+#ifdef CONFIG_PCI_MSI
+/* 152 - 215 represent the MSI interrupts 0-63 */
+#define OCTEON_IRQ_MSI_BIT0	152
+#define OCTEON_IRQ_MSI_BIT1	153
+#define OCTEON_IRQ_MSI_BIT2	154
+#define OCTEON_IRQ_MSI_BIT3	155
+#define OCTEON_IRQ_MSI_BIT4	156
+#define OCTEON_IRQ_MSI_BIT5	157
+#define OCTEON_IRQ_MSI_BIT6	158
+#define OCTEON_IRQ_MSI_BIT7	159
+#define OCTEON_IRQ_MSI_BIT8	160
+#define OCTEON_IRQ_MSI_BIT9	161
+#define OCTEON_IRQ_MSI_BIT10	162
+#define OCTEON_IRQ_MSI_BIT11	163
+#define OCTEON_IRQ_MSI_BIT12	164
+#define OCTEON_IRQ_MSI_BIT13	165
+#define OCTEON_IRQ_MSI_BIT14	166
+#define OCTEON_IRQ_MSI_BIT15	167
+#define OCTEON_IRQ_MSI_BIT16	168
+#define OCTEON_IRQ_MSI_BIT17	169
+#define OCTEON_IRQ_MSI_BIT18	170
+#define OCTEON_IRQ_MSI_BIT19	171
+#define OCTEON_IRQ_MSI_BIT20	172
+#define OCTEON_IRQ_MSI_BIT21	173
+#define OCTEON_IRQ_MSI_BIT22	174
+#define OCTEON_IRQ_MSI_BIT23	175
+#define OCTEON_IRQ_MSI_BIT24	176
+#define OCTEON_IRQ_MSI_BIT25	177
+#define OCTEON_IRQ_MSI_BIT26	178
+#define OCTEON_IRQ_MSI_BIT27	179
+#define OCTEON_IRQ_MSI_BIT28	180
+#define OCTEON_IRQ_MSI_BIT29	181
+#define OCTEON_IRQ_MSI_BIT30	182
+#define OCTEON_IRQ_MSI_BIT31	183
+#define OCTEON_IRQ_MSI_BIT32	184
+#define OCTEON_IRQ_MSI_BIT33	185
+#define OCTEON_IRQ_MSI_BIT34	186
+#define OCTEON_IRQ_MSI_BIT35	187
+#define OCTEON_IRQ_MSI_BIT36	188
+#define OCTEON_IRQ_MSI_BIT37	189
+#define OCTEON_IRQ_MSI_BIT38	190
+#define OCTEON_IRQ_MSI_BIT39	191
+#define OCTEON_IRQ_MSI_BIT40	192
+#define OCTEON_IRQ_MSI_BIT41	193
+#define OCTEON_IRQ_MSI_BIT42	194
+#define OCTEON_IRQ_MSI_BIT43	195
+#define OCTEON_IRQ_MSI_BIT44	196
+#define OCTEON_IRQ_MSI_BIT45	197
+#define OCTEON_IRQ_MSI_BIT46	198
+#define OCTEON_IRQ_MSI_BIT47	199
+#define OCTEON_IRQ_MSI_BIT48	200
+#define OCTEON_IRQ_MSI_BIT49	201
+#define OCTEON_IRQ_MSI_BIT50	202
+#define OCTEON_IRQ_MSI_BIT51	203
+#define OCTEON_IRQ_MSI_BIT52	204
+#define OCTEON_IRQ_MSI_BIT53	205
+#define OCTEON_IRQ_MSI_BIT54	206
+#define OCTEON_IRQ_MSI_BIT55	207
+#define OCTEON_IRQ_MSI_BIT56	208
+#define OCTEON_IRQ_MSI_BIT57	209
+#define OCTEON_IRQ_MSI_BIT58	210
+#define OCTEON_IRQ_MSI_BIT59	211
+#define OCTEON_IRQ_MSI_BIT60	212
+#define OCTEON_IRQ_MSI_BIT61	213
+#define OCTEON_IRQ_MSI_BIT62	214
+#define OCTEON_IRQ_MSI_BIT63	215
+
+#define OCTEON_IRQ_LAST         216
+#else
+#define OCTEON_IRQ_LAST         152
+#endif
+
+#endif

arch/mips/include/asm/mach-cavium-octeon/kernel-entry-init.h

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2005-2008 Cavium Networks, Inc
+ */
+#ifndef __ASM_MACH_CAVIUM_OCTEON_KERNEL_ENTRY_H
+#define __ASM_MACH_CAVIUM_OCTEON_KERNEL_ENTRY_H
+
+
+#define CP0_CYCLE_COUNTER $9, 6
+#define CP0_CVMCTL_REG $9, 7
+#define CP0_CVMMEMCTL_REG $11,7
+#define CP0_PRID_REG $15, 0
+#define CP0_PRID_OCTEON_PASS1 0x000d0000
+#define CP0_PRID_OCTEON_CN30XX 0x000d0200
+
+.macro  kernel_entry_setup
+	# Registers set by bootloader:
+	# (only 32 bits set by bootloader, all addresses are physical
+	# addresses, and need to have the appropriate memory region set
+	# by the kernel
+	# a0 = argc
+	# a1 = argv (kseg0 compat addr)
+	# a2 = 1 if init core, zero otherwise
+	# a3 = address of boot descriptor block
+	.set push
+	.set arch=octeon
+	# Read the cavium mem control register
+	dmfc0   v0, CP0_CVMMEMCTL_REG
+	# Clear the lower 6 bits, the CVMSEG size
+	dins    v0, $0, 0, 6
+	ori     v0, CONFIG_CAVIUM_OCTEON_CVMSEG_SIZE
+	dmtc0   v0, CP0_CVMMEMCTL_REG	# Write the cavium mem control register
+	dmfc0   v0, CP0_CVMCTL_REG	# Read the cavium control register
+#ifdef CONFIG_CAVIUM_OCTEON_HW_FIX_UNALIGNED
+	# Disable unaligned load/store support but leave HW fixup enabled
+	or  v0, v0, 0x5001
+	xor v0, v0, 0x1001
+#else
+	# Disable unaligned load/store and HW fixup support
+	or  v0, v0, 0x5001
+	xor v0, v0, 0x5001
+#endif
+	# Read the processor ID register
+	mfc0 v1, CP0_PRID_REG
+	# Disable instruction prefetching (Octeon Pass1 errata)
+	or  v0, v0, 0x2000
+	# Skip reenable of prefetching for Octeon Pass1
+	beq v1, CP0_PRID_OCTEON_PASS1, skip
+	nop
+	# Reenable instruction prefetching, not on Pass1
+	xor v0, v0, 0x2000
+	# Strip off pass number off of processor id
+	srl v1, 8
+	sll v1, 8
+	# CN30XX needs some extra stuff turned off for better performance
+	bne v1, CP0_PRID_OCTEON_CN30XX, skip
+	nop
+	# CN30XX Use random Icache replacement
+	or  v0, v0, 0x400
+	# CN30XX Disable instruction prefetching
+	or  v0, v0, 0x2000
+skip:
+	# Write the cavium control register
+	dmtc0   v0, CP0_CVMCTL_REG
+	sync
+	# Flush dcache after config change
+	cache   9, 0($0)
+	# Get my core id
+	rdhwr   v0, $0
+	# Jump the master to kernel_entry
+	bne     a2, zero, octeon_main_processor
+	nop
+
+#ifdef CONFIG_SMP
+
+	#
+	# All cores other than the master need to wait here for SMP bootstrap
+	# to begin
+	#
+
+	# This is the variable where the next core to boot os stored
+	PTR_LA  t0, octeon_processor_boot
+octeon_spin_wait_boot:
+	# Get the core id of the next to be booted
+	LONG_L  t1, (t0)
+	# Keep looping if it isn't me
+	bne t1, v0, octeon_spin_wait_boot
+	nop
+	# Get my GP from the global variable
+	PTR_LA  t0, octeon_processor_gp
+	LONG_L  gp, (t0)
+	# Get my SP from the global variable
+	PTR_LA  t0, octeon_processor_sp
+	LONG_L  sp, (t0)
+	# Set the SP global variable to zero so the master knows we've started
+	LONG_S  zero, (t0)
+#ifdef __OCTEON__
+	syncw
+	syncw
+#else
+	sync
+#endif
+	# Jump to the normal Linux SMP entry point
+	j   smp_bootstrap
+	nop
+#else /* CONFIG_SMP */
+
+	#
+	# Someone tried to boot SMP with a non SMP kernel. All extra cores
+	# will halt here.
+	#
+octeon_wait_forever:
+	wait
+	b   octeon_wait_forever
+	nop
+
+#endif /* CONFIG_SMP */
+octeon_main_processor:
+	.set pop
+.endm
+
+/*
+ * Do SMP slave processor setup necessary before we can savely execute C code.
+ */
+	.macro  smp_slave_setup
+	.endm
+
+#endif /* __ASM_MACH_CAVIUM_OCTEON_KERNEL_ENTRY_H */

arch/mips/include/asm/mach-cavium-octeon/war.h

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2002, 2004, 2007 by Ralf Baechle <ralf@linux-mips.org>
+ * Copyright (C) 2008 Cavium Networks <support@caviumnetworks.com>
+ */
+#ifndef __ASM_MIPS_MACH_CAVIUM_OCTEON_WAR_H
+#define __ASM_MIPS_MACH_CAVIUM_OCTEON_WAR_H
+
+#define R4600_V1_INDEX_ICACHEOP_WAR	0
+#define R4600_V1_HIT_CACHEOP_WAR	0
+#define R4600_V2_HIT_CACHEOP_WAR	0
+#define R5432_CP0_INTERRUPT_WAR		0
+#define BCM1250_M3_WAR			0
+#define SIBYTE_1956_WAR			0
+#define MIPS4K_ICACHE_REFILL_WAR	0
+#define MIPS_CACHE_SYNC_WAR		0
+#define TX49XX_ICACHE_INDEX_INV_WAR	0
+#define RM9000_CDEX_SMP_WAR		0
+#define ICACHE_REFILLS_WORKAROUND_WAR	0
+#define R10000_LLSC_WAR			0
+#define MIPS34K_MISSED_ITLB_WAR		0
+
+#endif /* __ASM_MIPS_MACH_CAVIUM_OCTEON_WAR_H */

arch/mips/include/asm/octeon/octeon.h

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2004-2008 Cavium Networks
+ */
+#ifndef __ASM_OCTEON_OCTEON_H
+#define __ASM_OCTEON_OCTEON_H
+
+#include "cvmx.h"
+
+extern uint64_t octeon_bootmem_alloc_range_phys(uint64_t size,
+						uint64_t alignment,
+						uint64_t min_addr,
+						uint64_t max_addr,
+						int do_locking);
+extern void *octeon_bootmem_alloc(uint64_t size, uint64_t alignment,
+				  int do_locking);
+extern void *octeon_bootmem_alloc_range(uint64_t size, uint64_t alignment,
+					uint64_t min_addr, uint64_t max_addr,
+					int do_locking);
+extern void *octeon_bootmem_alloc_named(uint64_t size, uint64_t alignment,
+					char *name);
+extern void *octeon_bootmem_alloc_named_range(uint64_t size, uint64_t min_addr,
+					      uint64_t max_addr, uint64_t align,
+					      char *name);
+extern void *octeon_bootmem_alloc_named_address(uint64_t size, uint64_t address,
+						char *name);
+extern int octeon_bootmem_free_named(char *name);
+extern void octeon_bootmem_lock(void);
+extern void octeon_bootmem_unlock(void);
+
+extern int octeon_is_simulation(void);
+extern int octeon_is_pci_host(void);
+extern int octeon_usb_is_ref_clk(void);
+extern uint64_t octeon_get_clock_rate(void);
+extern const char *octeon_board_type_string(void);
+extern const char *octeon_get_pci_interrupts(void);
+extern int octeon_get_southbridge_interrupt(void);
+extern int octeon_get_boot_coremask(void);
+extern int octeon_get_boot_num_arguments(void);
+extern const char *octeon_get_boot_argument(int arg);
+extern void octeon_hal_setup_reserved32(void);
+extern void octeon_user_io_init(void);
+struct octeon_cop2_state;
+extern unsigned long octeon_crypto_enable(struct octeon_cop2_state *state);
+extern void octeon_crypto_disable(struct octeon_cop2_state *state,
+				  unsigned long flags);
+
+extern void octeon_init_cvmcount(void);
+
+#define OCTEON_ARGV_MAX_ARGS	64
+#define OCTOEN_SERIAL_LEN	20
+
+struct octeon_boot_descriptor {
+	/* Start of block referenced by assembly code - do not change! */
+	uint32_t desc_version;
+	uint32_t desc_size;
+	uint64_t stack_top;
+	uint64_t heap_base;
+	uint64_t heap_end;
+	/* Only used by bootloader */
+	uint64_t entry_point;
+	uint64_t desc_vaddr;
+	/* End of This block referenced by assembly code - do not change! */
+	uint32_t exception_base_addr;
+	uint32_t stack_size;
+	uint32_t heap_size;
+	/* Argc count for application. */
+	uint32_t argc;
+	uint32_t argv[OCTEON_ARGV_MAX_ARGS];
+
+#define  BOOT_FLAG_INIT_CORE		(1 << 0)
+#define  OCTEON_BL_FLAG_DEBUG		(1 << 1)
+#define  OCTEON_BL_FLAG_NO_MAGIC	(1 << 2)
+	/* If set, use uart1 for console */
+#define  OCTEON_BL_FLAG_CONSOLE_UART1	(1 << 3)
+	/* If set, use PCI console */
+#define  OCTEON_BL_FLAG_CONSOLE_PCI	(1 << 4)
+	/* Call exit on break on serial port */
+#define  OCTEON_BL_FLAG_BREAK		(1 << 5)
+
+	uint32_t flags;
+	uint32_t core_mask;
+	/* DRAM size in megabyes. */
+	uint32_t dram_size;
+	/* physical address of free memory descriptor block. */
+	uint32_t phy_mem_desc_addr;
+	/* used to pass flags from app to debugger. */
+	uint32_t debugger_flags_base_addr;
+	/* CPU clock speed, in hz. */
+	uint32_t eclock_hz;
+	/* DRAM clock speed, in hz. */
+	uint32_t dclock_hz;
+	/* SPI4 clock in hz. */
+	uint32_t spi_clock_hz;
+	uint16_t board_type;
+	uint8_t board_rev_major;
+	uint8_t board_rev_minor;
+	uint16_t chip_type;
+	uint8_t chip_rev_major;
+	uint8_t chip_rev_minor;
+	char board_serial_number[OCTOEN_SERIAL_LEN];
+	uint8_t mac_addr_base[6];
+	uint8_t mac_addr_count;
+	uint64_t cvmx_desc_vaddr;
+};
+
+union octeon_cvmemctl {
+	uint64_t u64;
+	struct {
+		/* RO 1 = BIST fail, 0 = BIST pass */
+		uint64_t tlbbist:1;
+		/* RO 1 = BIST fail, 0 = BIST pass */
+		uint64_t l1cbist:1;
+		/* RO 1 = BIST fail, 0 = BIST pass */
+		uint64_t l1dbist:1;
+		/* RO 1 = BIST fail, 0 = BIST pass */
+		uint64_t dcmbist:1;
+		/* RO 1 = BIST fail, 0 = BIST pass */
+		uint64_t ptgbist:1;
+		/* RO 1 = BIST fail, 0 = BIST pass */
+		uint64_t wbfbist:1;
+		/* Reserved */
+		uint64_t reserved:22;
+		/* R/W If set, marked write-buffer entries time out
+		 * the same as as other entries; if clear, marked
+		 * write-buffer entries use the maximum timeout. */
+		uint64_t dismarkwblongto:1;
+		/* R/W If set, a merged store does not clear the
+		 * write-buffer entry timeout state. */
+		uint64_t dismrgclrwbto:1;
+		/* R/W Two bits that are the MSBs of the resultant
+		 * CVMSEG LM word location for an IOBDMA. The other 8
+		 * bits come from the SCRADDR field of the IOBDMA. */
+		uint64_t iobdmascrmsb:2;
+		/* R/W If set, SYNCWS and SYNCS only order marked
+		 * stores; if clear, SYNCWS and SYNCS only order
+		 * unmarked stores. SYNCWSMARKED has no effect when
+		 * DISSYNCWS is set. */
+		uint64_t syncwsmarked:1;
+		/* R/W If set, SYNCWS acts as SYNCW and SYNCS acts as
+		 * SYNC. */
+		uint64_t dissyncws:1;
+		/* R/W If set, no stall happens on write buffer
+		 * full. */
+		uint64_t diswbfst:1;
+		/* R/W If set (and SX set), supervisor-level
+		 * loads/stores can use XKPHYS addresses with
+		 * VA<48>==0 */
+		uint64_t xkmemenas:1;
+		/* R/W If set (and UX set), user-level loads/stores
+		 * can use XKPHYS addresses with VA<48>==0 */
+		uint64_t xkmemenau:1;
+		/* R/W If set (and SX set), supervisor-level
+		 * loads/stores can use XKPHYS addresses with
+		 * VA<48>==1 */
+		uint64_t xkioenas:1;
+		/* R/W If set (and UX set), user-level loads/stores
+		 * can use XKPHYS addresses with VA<48>==1 */
+		uint64_t xkioenau:1;
+		/* R/W If set, all stores act as SYNCW (NOMERGE must
+		 * be set when this is set) RW, reset to 0. */
+		uint64_t allsyncw:1;
+		/* R/W If set, no stores merge, and all stores reach
+		 * the coherent bus in order. */
+		uint64_t nomerge:1;
+		/* R/W Selects the bit in the counter used for DID
+		 * time-outs 0 = 231, 1 = 230, 2 = 229, 3 =
+		 * 214. Actual time-out is between 1x and 2x this
+		 * interval. For example, with DIDTTO=3, expiration
+		 * interval is between 16K and 32K. */
+		uint64_t didtto:2;
+		/* R/W If set, the (mem) CSR clock never turns off. */
+		uint64_t csrckalwys:1;
+		/* R/W If set, mclk never turns off. */
+		uint64_t mclkalwys:1;
+		/* R/W Selects the bit in the counter used for write
+		 * buffer flush time-outs (WBFLT+11) is the bit
+		 * position in an internal counter used to determine
+		 * expiration. The write buffer expires between 1x and
+		 * 2x this interval. For example, with WBFLT = 0, a
+		 * write buffer expires between 2K and 4K cycles after
+		 * the write buffer entry is allocated. */
+		uint64_t wbfltime:3;
+		/* R/W If set, do not put Istream in the L2 cache. */
+		uint64_t istrnol2:1;
+		/* R/W The write buffer threshold. */
+		uint64_t wbthresh:4;
+		/* Reserved */
+		uint64_t reserved2:2;
+		/* R/W If set, CVMSEG is available for loads/stores in
+		 * kernel/debug mode. */
+		uint64_t cvmsegenak:1;
+		/* R/W If set, CVMSEG is available for loads/stores in
+		 * supervisor mode. */
+		uint64_t cvmsegenas:1;
+		/* R/W If set, CVMSEG is available for loads/stores in
+		 * user mode. */
+		uint64_t cvmsegenau:1;
+		/* R/W Size of local memory in cache blocks, 54 (6912
+		 * bytes) is max legal value. */
+		uint64_t lmemsz:6;
+	} s;
+};
+
+struct octeon_cf_data {
+	unsigned long	base_region_bias;
+	unsigned int	base_region;	/* The chip select region used by CF */
+	int		is16bit;	/* 0 - 8bit, !0 - 16bit */
+	int		dma_engine;	/* -1 for no DMA */
+};
+
+extern void octeon_write_lcd(const char *s);
+extern void octeon_check_cpu_bist(void);
+extern int octeon_get_boot_debug_flag(void);
+extern int octeon_get_boot_uart(void);
+
+struct uart_port;
+extern unsigned int octeon_serial_in(struct uart_port *, int);
+extern void octeon_serial_out(struct uart_port *, int, int);
+
+/**
+ * Write a 32bit value to the Octeon NPI register space
+ *
+ * @address: Address to write to
+ * @val:     Value to write
+ */
+static inline void octeon_npi_write32(uint64_t address, uint32_t val)
+{
+	cvmx_write64_uint32(address ^ 4, val);
+	cvmx_read64_uint32(address ^ 4);
+}
+
+
+/**
+ * Read a 32bit value from the Octeon NPI register space
+ *
+ * @address: Address to read
+ * Returns The result
+ */
+static inline uint32_t octeon_npi_read32(uint64_t address)
+{
+	return cvmx_read64_uint32(address ^ 4);
+}
+
+#endif /* __ASM_OCTEON_OCTEON_H */

arch/mips/mm/c-octeon.c

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2005-2007 Cavium Networks
+ */
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/bitops.h>
+#include <linux/cpu.h>
+#include <linux/io.h>
+
+#include <asm/bcache.h>
+#include <asm/bootinfo.h>
+#include <asm/cacheops.h>
+#include <asm/cpu-features.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/r4kcache.h>
+#include <asm/system.h>
+#include <asm/mmu_context.h>
+#include <asm/war.h>
+
+#include <asm/octeon/octeon.h>
+
+unsigned long long cache_err_dcache[NR_CPUS];
+
+/**
+ * Octeon automatically flushes the dcache on tlb changes, so
+ * from Linux's viewpoint it acts much like a physically
+ * tagged cache. No flushing is needed
+ *
+ */
+static void octeon_flush_data_cache_page(unsigned long addr)
+{
+    /* Nothing to do */
+}
+
+static inline void octeon_local_flush_icache(void)
+{
+	asm volatile ("synci 0($0)");
+}
+
+/*
+ * Flush local I-cache for the specified range.
+ */
+static void local_octeon_flush_icache_range(unsigned long start,
+					    unsigned long end)
+{
+	octeon_local_flush_icache();
+}
+
+/**
+ * Flush caches as necessary for all cores affected by a
+ * vma. If no vma is supplied, all cores are flushed.
+ *
+ * @vma:    VMA to flush or NULL to flush all icaches.
+ */
+static void octeon_flush_icache_all_cores(struct vm_area_struct *vma)
+{
+	extern void octeon_send_ipi_single(int cpu, unsigned int action);
+#ifdef CONFIG_SMP
+	int cpu;
+	cpumask_t mask;
+#endif
+
+	mb();
+	octeon_local_flush_icache();
+#ifdef CONFIG_SMP
+	preempt_disable();
+	cpu = smp_processor_id();
+
+	/*
+	 * If we have a vma structure, we only need to worry about
+	 * cores it has been used on
+	 */
+	if (vma)
+		mask = vma->vm_mm->cpu_vm_mask;
+	else
+		mask = cpu_online_map;
+	cpu_clear(cpu, mask);
+	for_each_cpu_mask(cpu, mask)
+		octeon_send_ipi_single(cpu, SMP_ICACHE_FLUSH);
+
+	preempt_enable();
+#endif
+}
+
+
+/**
+ * Called to flush the icache on all cores
+ */
+static void octeon_flush_icache_all(void)
+{
+	octeon_flush_icache_all_cores(NULL);
+}
+
+
+/**
+ * Called to flush all memory associated with a memory
+ * context.
+ *
+ * @mm:     Memory context to flush
+ */
+static void octeon_flush_cache_mm(struct mm_struct *mm)
+{
+	/*
+	 * According to the R4K version of this file, CPUs without
+	 * dcache aliases don't need to do anything here
+	 */
+}
+
+
+/**
+ * Flush a range of kernel addresses out of the icache
+ *
+ */
+static void octeon_flush_icache_range(unsigned long start, unsigned long end)
+{
+	octeon_flush_icache_all_cores(NULL);
+}
+
+
+/**
+ * Flush the icache for a trampoline. These are used for interrupt
+ * and exception hooking.
+ *
+ * @addr:   Address to flush
+ */
+static void octeon_flush_cache_sigtramp(unsigned long addr)
+{
+	struct vm_area_struct *vma;
+
+	vma = find_vma(current->mm, addr);
+	octeon_flush_icache_all_cores(vma);
+}
+
+
+/**
+ * Flush a range out of a vma
+ *
+ * @vma:    VMA to flush
+ * @start:
+ * @end:
+ */
+static void octeon_flush_cache_range(struct vm_area_struct *vma,
+				     unsigned long start, unsigned long end)
+{
+	if (vma->vm_flags & VM_EXEC)
+		octeon_flush_icache_all_cores(vma);
+}
+
+
+/**
+ * Flush a specific page of a vma
+ *
+ * @vma:    VMA to flush page for
+ * @page:   Page to flush
+ * @pfn:
+ */
+static void octeon_flush_cache_page(struct vm_area_struct *vma,
+				    unsigned long page, unsigned long pfn)
+{
+	if (vma->vm_flags & VM_EXEC)
+		octeon_flush_icache_all_cores(vma);
+}
+
+
+/**
+ * Probe Octeon's caches
+ *
+ */
+static void __devinit probe_octeon(void)
+{
+	unsigned long icache_size;
+	unsigned long dcache_size;
+	unsigned int config1;
+	struct cpuinfo_mips *c = &current_cpu_data;
+
+	switch (c->cputype) {
+	case CPU_CAVIUM_OCTEON:
+		config1 = read_c0_config1();
+		c->icache.linesz = 2 << ((config1 >> 19) & 7);
+		c->icache.sets = 64 << ((config1 >> 22) & 7);
+		c->icache.ways = 1 + ((config1 >> 16) & 7);
+		c->icache.flags |= MIPS_CACHE_VTAG;
+		icache_size =
+			c->icache.sets * c->icache.ways * c->icache.linesz;
+		c->icache.waybit = ffs(icache_size / c->icache.ways) - 1;
+		c->dcache.linesz = 128;
+		if (OCTEON_IS_MODEL(OCTEON_CN3XXX))
+			c->dcache.sets = 1; /* CN3XXX has one Dcache set */
+		else
+			c->dcache.sets = 2; /* CN5XXX has two Dcache sets */
+		c->dcache.ways = 64;
+		dcache_size =
+			c->dcache.sets * c->dcache.ways * c->dcache.linesz;
+		c->dcache.waybit = ffs(dcache_size / c->dcache.ways) - 1;
+		c->options |= MIPS_CPU_PREFETCH;
+		break;
+
+	default:
+		panic("Unsupported Cavium Networks CPU type\n");
+		break;
+	}
+
+	/* compute a couple of other cache variables */
+	c->icache.waysize = icache_size / c->icache.ways;
+	c->dcache.waysize = dcache_size / c->dcache.ways;
+
+	c->icache.sets = icache_size / (c->icache.linesz * c->icache.ways);
+	c->dcache.sets = dcache_size / (c->dcache.linesz * c->dcache.ways);
+
+	if (smp_processor_id() == 0) {
+		pr_notice("Primary instruction cache %ldkB, %s, %d way, "
+			  "%d sets, linesize %d bytes.\n",
+			  icache_size >> 10,
+			  cpu_has_vtag_icache ?
+				"virtually tagged" : "physically tagged",
+			  c->icache.ways, c->icache.sets, c->icache.linesz);
+
+		pr_notice("Primary data cache %ldkB, %d-way, %d sets, "
+			  "linesize %d bytes.\n",
+			  dcache_size >> 10, c->dcache.ways,
+			  c->dcache.sets, c->dcache.linesz);
+	}
+}
+
+
+/**
+ * Setup the Octeon cache flush routines
+ *
+ */
+void __devinit octeon_cache_init(void)
+{
+	extern unsigned long ebase;
+	extern char except_vec2_octeon;
+
+	memcpy((void *)(ebase + 0x100), &except_vec2_octeon, 0x80);
+	octeon_flush_cache_sigtramp(ebase + 0x100);
+
+	probe_octeon();
+
+	shm_align_mask = PAGE_SIZE - 1;
+
+	flush_cache_all			= octeon_flush_icache_all;
+	__flush_cache_all		= octeon_flush_icache_all;
+	flush_cache_mm			= octeon_flush_cache_mm;
+	flush_cache_page		= octeon_flush_cache_page;
+	flush_cache_range		= octeon_flush_cache_range;
+	flush_cache_sigtramp		= octeon_flush_cache_sigtramp;
+	flush_icache_all		= octeon_flush_icache_all;
+	flush_data_cache_page		= octeon_flush_data_cache_page;
+	flush_icache_range		= octeon_flush_icache_range;
+	local_flush_icache_range	= local_octeon_flush_icache_range;
+
+	build_clear_page();
+	build_copy_page();
+}
+
+/**
+ * Handle a cache error exception
+ */
+
+static void  cache_parity_error_octeon(int non_recoverable)
+{
+	unsigned long coreid = cvmx_get_core_num();
+	uint64_t icache_err = read_octeon_c0_icacheerr();
+
+	pr_err("Cache error exception:\n");
+	pr_err("cp0_errorepc == %lx\n", read_c0_errorepc());
+	if (icache_err & 1) {
+		pr_err("CacheErr (Icache) == %llx\n",
+		       (unsigned long long)icache_err);
+		write_octeon_c0_icacheerr(0);
+	}
+	if (cache_err_dcache[coreid] & 1) {
+		pr_err("CacheErr (Dcache) == %llx\n",
+		       (unsigned long long)cache_err_dcache[coreid]);
+		cache_err_dcache[coreid] = 0;
+	}
+
+	if (non_recoverable)
+		panic("Can't handle cache error: nested exception");
+}
+
+/**
+ * Called when the the exception is not recoverable
+ */
+
+asmlinkage void cache_parity_error_octeon_recoverable(void)
+{
+	cache_parity_error_octeon(0);
+}
+
+/**
+ * Called when the the exception is recoverable
+ */
+
+asmlinkage void cache_parity_error_octeon_non_recoverable(void)
+{
+	cache_parity_error_octeon(1);
+}
+

arch/mips/mm/cex-oct.S

+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2006 Cavium Networks
+ * Cache error handler
+ */
+
+#include <asm/asm.h>
+#include <asm/regdef.h>
+#include <asm/mipsregs.h>
+#include <asm/stackframe.h>
+
+/*
+ * Handle cache error. Indicate to the second level handler whether
+ * the exception is recoverable.
+ */
+	LEAF(except_vec2_octeon)
+
+	.set    push
+	.set	mips64r2
+	.set	noreorder
+	.set	noat
+
+
+	/* due to an errata we need to read the COP0 CacheErr (Dcache)
+	 * before any cache/DRAM access	 */
+
+	rdhwr   k0, $0        /* get core_id */
+	PTR_LA  k1, cache_err_dcache
+	sll     k0, k0, 3
+	PTR_ADDU k1, k0, k1    /* k1 = &cache_err_dcache[core_id] */
+
+	dmfc0   k0, CP0_CACHEERR, 1
+	sd      k0, (k1)
+	dmtc0   $0, CP0_CACHEERR, 1
+
+        /* check whether this is a nested exception */
+	mfc0    k1, CP0_STATUS
+	andi    k1, k1, ST0_EXL
+	beqz    k1, 1f
+	 nop
+	j	cache_parity_error_octeon_non_recoverable
+	 nop
+
+	/* exception is recoverable */
+1:	j	handle_cache_err
+	 nop
+
+	.set    pop
+	END(except_vec2_octeon)
+
+ /* We need to jump to handle_cache_err so that the previous handler
+  * can fit within 0x80 bytes. We also move from 0xFFFFFFFFAXXXXXXX
+  * space (uncached) to the 0xFFFFFFFF8XXXXXXX space (cached).  */
+	LEAF(handle_cache_err)
+	.set    push
+        .set    noreorder
+        .set    noat
+
+	SAVE_ALL
+	KMODE
+	jal     cache_parity_error_octeon_recoverable
+	nop
+	j       ret_from_exception
+	nop
+
+	.set pop
+	END(handle_cache_err)