/*
 * Copyright (c) 2007 Michal Kebrt
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 *
 * - Redistributions of source code must retain the above copyright
 *   notice, this list of conditions and the following disclaimer.
 * - Redistributions in binary form must reproduce the above copyright
 *   notice, this list of conditions and the following disclaimer in the
 *   documentation and/or other materials provided with the distribution.
 * - The name of the author may not be used to endorse or promote products
 *   derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

/** @addtogroup arm32
 * @{
 */
/** @file
 *  @brief CPU identification.
 */

#include <arch/cache.h>
#include <arch/cpu.h>
#include <arch/cp15.h>
#include <cpu.h>
#include <arch.h>
#include <print.h>

static inline unsigned log2(unsigned val)
{
	unsigned log = 0;
	--val;
	while (val) {
		++log;
		val >>= 1;
	}
	return log;
}

static unsigned dcache_ways(unsigned level);
static unsigned dcache_sets(unsigned level);
static unsigned dcache_linesize_log(unsigned level);


/** Implementers (vendor) names */
static const char * implementer(unsigned id)
{
	switch (id)
	{
	case 0x41: return "ARM Limited";
	case 0x44: return "Digital Equipment Corporation";
	case 0x4d: return "Motorola, Freescale Semiconductor Inc.";
	case 0x51: return "Qualcomm Inc.";
	case 0x56: return "Marvell Semiconductor Inc.";
	case 0x69: return "Intel Corporation";
	}
	return "Unknown implementer";
}

/** Architecture names */
static const char * architecture_string(cpu_arch_t *arch)
{
	static const char *arch_data[] = {
		"ARM",       /* 0x0 */
		"ARMv4",       /* 0x1 */
		"ARMv4T",      /* 0x2 */
		"ARMv5",       /* 0x3 */
		"ARMv5T",      /* 0x4 */
		"ARMv5TE",     /* 0x5 */
		"ARMv5TEJ",    /* 0x6 */
		"ARMv6"        /* 0x7 */
	};
	if (arch->arch_num < (sizeof(arch_data) / sizeof(arch_data[0])))
		return arch_data[arch->arch_num];
	else
		return arch_data[0];
}


/** Retrieves processor identification from CP15 register 0.
 *
 * @param cpu Structure for storing CPU identification.
 * See page B4-1630 of ARM Architecture Reference Manual.
 */
static void arch_cpu_identify(cpu_arch_t *cpu)
{
	const uint32_t ident = MIDR_read();

	cpu->imp_num = (ident >> MIDR_IMPLEMENTER_SHIFT) & MIDR_IMPLEMENTER_MASK;
	cpu->variant_num = (ident >> MIDR_VARIANT_SHIFT) & MIDR_VARIANT_MASK;
	cpu->arch_num = (ident >> MIDR_ARCHITECTURE_SHIFT) & MIDR_ARCHITECTURE_MASK;
	cpu->prim_part_num = (ident >> MIDR_PART_NUMBER_SHIFT) & MIDR_PART_NUMBER_MASK;
	cpu->rev_num = (ident >> MIDR_REVISION_SHIFT) & MIDR_REVISION_MASK;

	// TODO CPUs with arch_num == 0xf use CPUID scheme for identification
	cpu->dcache_levels = dcache_levels();

	for (unsigned i = 0; i < cpu->dcache_levels; ++i) {
		cpu->dcache[i].ways = dcache_ways(i);
		cpu->dcache[i].sets = dcache_sets(i);
		cpu->dcache[i].way_shift = 31 - log2(cpu->dcache[i].ways);
		cpu->dcache[i].set_shift = dcache_linesize_log(i);
		cpu->dcache[i].line_size = 1 << dcache_linesize_log(i);
		printf("Found DCache L%u: %u-way, %u sets, %u byte lines "
		    "(shifts: w%u, s%u)\n", i + 1, cpu->dcache[i].ways,
		    cpu->dcache[i].sets, cpu->dcache[i].line_size,
		    cpu->dcache[i].way_shift, cpu->dcache[i].set_shift);
	}
}

/** Enables unaligned access and caching for armv6+ */
void cpu_arch_init(void)
{
	uint32_t control_reg = SCTLR_read();
	
	/* Turn off tex remap, RAZ/WI prior to armv7 */
	control_reg &= ~SCTLR_TEX_REMAP_EN_FLAG;
	/* Turn off accessed flag, RAZ/WI prior to armv7 */
	control_reg &= ~(SCTLR_ACCESS_FLAG_EN_FLAG | SCTLR_HW_ACCESS_FLAG_EN_FLAG);

	/* Unaligned access is supported on armv6+ */
#if defined(PROCESSOR_ARCH_armv7_a) | defined(PROCESSOR_ARCH_armv6)
	/* Enable unaligned access, RAZ/WI prior to armv6
	 * switchable on armv6, RAO/WI writes on armv7,
	 * see ARM Architecture Reference Manual ARMv7-A and ARMv7-R edition
	 * L.3.1 (p. 2456) */
	control_reg |= SCTLR_UNALIGNED_EN_FLAG;
	/* Disable alignment checks, this turns unaligned access to undefined,
	 * unless U bit is set. */
	control_reg &= ~SCTLR_ALIGN_CHECK_EN_FLAG;
	/* Enable caching, On arm prior to armv7 there is only one level
	 * of caches. Data cache is coherent.
	 * "This means that the behavior of accesses from the same observer to
	 * different VAs, that are translated to the same PA
	 * with the same memory attributes, is fully coherent."
	 *    ARM Architecture Reference Manual ARMv7-A and ARMv7-R Edition
	 *    B3.11.1 (p. 1383)
	 * We are safe to turn this on. For arm v6 see ch L.6.2 (p. 2469)
	 * L2 Cache for armv7 is enabled by default (i.e. controlled by
	 * this flag).
	 */
	control_reg |= SCTLR_CACHE_EN_FLAG;
#endif
#ifdef PROCESSOR_ARCH_armv7_a
	 /* ICache coherency is elaborate on in barrier.h.
	  * VIPT and PIPT caches need maintenance only on code modify,
	  * so it should be safe for general use.
	  * Enable branch predictors too as they follow the same rules
	  * as ICache and they can be flushed together
	  */
	if ((CTR_read() & CTR_L1I_POLICY_MASK) != CTR_L1I_POLICY_AIVIVT) {
		control_reg |=
		    SCTLR_INST_CACHE_EN_FLAG | SCTLR_BRANCH_PREDICT_EN_FLAG;
	}
#endif
	SCTLR_write(control_reg);

#ifdef CONFIG_FPU
	fpu_setup();
#endif

#ifdef PROCESSOR_ARCH_armv7_a
	if ((ID_PFR1_read() & ID_PFR1_GEN_TIMER_EXT_MASK) !=
	    ID_PFR1_GEN_TIMER_EXT) {
		PMCR_write(PMCR_read() | PMCR_E_FLAG | PMCR_D_FLAG);
		PMCNTENSET_write(PMCNTENSET_CYCLE_COUNTER_EN_FLAG);
	}
#endif
}

/** Retrieves processor identification and stores it to #CPU.arch */
void cpu_identify(void)
{
	arch_cpu_identify(&CPU->arch);
}

/** Prints CPU identification. */
void cpu_print_report(cpu_t *m)
{
	printf("cpu%d: vendor=%s, architecture=%s, part number=%x, "
	    "variant=%x, revision=%x\n",
	    m->id, implementer(m->arch.imp_num),
	    architecture_string(&m->arch), m->arch.prim_part_num,
	    m->arch.variant_num, m->arch.rev_num);
}

/** See chapter B4.1.19 of ARM Architecture Reference Manual */
static unsigned dcache_linesize_log(unsigned level)
{
#ifdef PROCESSOR_ARCH_armv7_a
	CSSELR_write((level & CCSELR_LEVEL_MASK) << CCSELR_LEVEL_SHIFT);
	const unsigned ls_log = 2 +
	    ((CCSIDR_read() >> CCSIDR_LINESIZE_SHIFT) & CCSIDR_LINESIZE_MASK);
	return ls_log + 2; //return log2(bytes)
#endif
	return 0;

}

/** See chapter B4.1.19 of ARM Architecture Reference Manual */
static unsigned dcache_ways(unsigned level)
{
#ifdef PROCESSOR_ARCH_armv7_a
	CSSELR_write((level & CCSELR_LEVEL_MASK) << CCSELR_LEVEL_SHIFT);
	const unsigned ways = 1 +
	    ((CCSIDR_read() >> CCSIDR_ASSOC_SHIFT) & CCSIDR_ASSOC_MASK);
	return ways;
#endif
	return 0;
}

/** See chapter B4.1.19 of ARM Architecture Reference Manual */
static unsigned dcache_sets(unsigned level)
{
#ifdef PROCESSOR_ARCH_armv7_a
	CSSELR_write((level & CCSELR_LEVEL_MASK) << CCSELR_LEVEL_SHIFT);
	const unsigned sets = 1 +
	    ((CCSIDR_read() >> CCSIDR_NUMSETS_SHIFT) & CCSIDR_NUMSETS_MASK);
	return sets;
#endif
	return 0;
}

unsigned dcache_levels(void)
{
	unsigned levels = 0;
#ifdef PROCESSOR_ARCH_armv7_a
	const uint32_t val = CLIDR_read();
	for (unsigned i = 1; i <= 7; ++i) {
		const unsigned ctype = CLIDR_CACHE(i, val);
		switch (ctype) {
		case CLIDR_DCACHE_ONLY:
		case CLIDR_SEP_CACHE:
		case CLIDR_UNI_CACHE:
			++levels;
		default:
			(void)0;
		}
	}
#endif
	return levels;
}

static void dcache_clean_manual(unsigned level, bool invalidate,
    unsigned ways, unsigned sets, unsigned way_shift, unsigned set_shift)
{

	for (unsigned i = 0; i < ways; ++i) {
		for (unsigned j = 0; j < sets; ++j) {
			const uint32_t val =
			    ((level & 0x7) << 1) |
			    (j << set_shift) | (i << way_shift);
			if (invalidate)
				DCCISW_write(val);
			else
				DCCSW_write(val);
		}
	}
}

void dcache_flush(void)
{
	/* See ARM Architecture Reference Manual ch. B4.2.1 p. B4-1724 */
	const unsigned levels = dcache_levels();
	for (unsigned i = 0; i < levels; ++i) {
		const unsigned ways = dcache_ways(i);
		const unsigned sets = dcache_sets(i);
		const unsigned way_shift =  31 - log2(ways);
		const unsigned set_shift = dcache_linesize_log(i);
		dcache_clean_manual(i, false, ways, sets, way_shift, set_shift);
	}
}

void dcache_flush_invalidate(void)
{
	/* See ARM Architecture Reference Manual ch. B4.2.1 p. B4-1724 */
	const unsigned levels = dcache_levels();
	for (unsigned i = 0; i < levels; ++i) {
		const unsigned ways = dcache_ways(i);
		const unsigned sets = dcache_sets(i);
		const unsigned way_shift =  31 - log2(ways);
		const unsigned set_shift = dcache_linesize_log(i);
		dcache_clean_manual(i, true, ways, sets, way_shift, set_shift);
	}
}


void cpu_dcache_flush(void)
{
	for (unsigned i = 0; i < CPU->arch.dcache_levels; ++i)
		dcache_clean_manual(i, false,
		    CPU->arch.dcache[i].ways, CPU->arch.dcache[i].sets,
		    CPU->arch.dcache[i].way_shift, CPU->arch.dcache[i].set_shift);
}

void cpu_dcache_flush_invalidate(void)
{
	const unsigned levels =  dcache_levels();
	for (unsigned i = 0; i < levels; ++i)
		dcache_clean_manual(i, true,
		    CPU->arch.dcache[i].ways, CPU->arch.dcache[i].sets,
		    CPU->arch.dcache[i].way_shift, CPU->arch.dcache[i].set_shift);
}

void icache_invalidate(void)
{
	ICIALLU_write(0);
}

/** @}
 */