static void __init
mpc85xx_setup_pci1(struct pci_controller *hose)
{
volatile struct ccsr_pci *pci;
volatile struct ccsr_guts *guts;
unsigned short temps;
bd_t *binfo = (bd_t *) __res;
pci = ioremap(binfo->bi_immr_base + MPC85xx_PCI1_OFFSET,
MPC85xx_PCI1_SIZE);
guts = ioremap(binfo->bi_immr_base + MPC85xx_GUTS_OFFSET,
MPC85xx_GUTS_SIZE);
early_read_config_word(hose, 0, 0, PCI_COMMAND, &temps);
temps |= PCI_COMMAND_SERR | PCI_COMMAND_MASTER | PCI_COMMAND_MEMORY;
early_write_config_word(hose, 0, 0, PCI_COMMAND, temps);
#define PORDEVSR_PCI (0x00800000) /* PCI Mode */
if (guts->pordevsr & PORDEVSR_PCI) {
early_write_config_byte(hose, 0, 0, PCI_LATENCY_TIMER, 0x80);
} else {
/* PCI-X init */
temps = PCI_X_CMD_MAX_SPLIT | PCI_X_CMD_MAX_READ
| PCI_X_CMD_ERO | PCI_X_CMD_DPERR_E;
early_write_config_word(hose, 0, 0, PCIX_COMMAND, temps);
}
/* Disable all windows (except powar0 since its ignored) */
pci->powar1 = 0;
pci->powar2 = 0;
pci->powar3 = 0;
pci->powar4 = 0;
pci->piwar1 = 0;
pci->piwar2 = 0;
pci->piwar3 = 0;
/* Setup Phys:PCI 1:1 outbound mem window @ MPC85XX_PCI1_LOWER_MEM */
pci->potar1 = (MPC85XX_PCI1_LOWER_MEM >> 12) & 0x000fffff;
pci->potear1 = 0x00000000;
pci->powbar1 = (MPC85XX_PCI1_LOWER_MEM >> 12) & 0x000fffff;
/* Enable, Mem R/W */
pci->powar1 = 0x80044000 |
(__ilog2(MPC85XX_PCI1_UPPER_MEM - MPC85XX_PCI1_LOWER_MEM + 1) - 1);
/* Setup outbound IO windows @ MPC85XX_PCI1_IO_BASE */
pci->potar2 = (MPC85XX_PCI1_LOWER_IO >> 12) & 0x000fffff;
pci->potear2 = 0x00000000;
pci->powbar2 = (MPC85XX_PCI1_IO_BASE >> 12) & 0x000fffff;
/* Enable, IO R/W */
pci->powar2 = 0x80088000 | (__ilog2(MPC85XX_PCI1_IO_SIZE) - 1);
/* Setup 2G inbound Memory Window @ 0 */
pci->pitar1 = 0x00000000;
pci->piwbar1 = 0x00000000;
pci->piwar1 = 0xa0f5501e; /* Enable, Prefetch, Local
Mem, Snoop R/W, 2G */
}
/* mv64360_get_irq()
*
* This function returns the lowest interrupt number of all interrupts that
* are currently asserted.
*
* Output Variable(s):
* None.
*
* Returns:
* int <interrupt number> or -2 (bogus interrupt)
*
*/
int
mv64360_get_irq(void)
{
int irq;
int irq_gpp;
#ifdef CONFIG_SMP
/*
* Second CPU gets only doorbell (message) interrupts.
* The doorbell interrupt is BIT28 in the main interrupt low cause reg.
*/
int cpu_nr = smp_processor_id();
if (cpu_nr == 1) {
if (!(mv64x60_read(&bh, MV64360_IC_MAIN_CAUSE_LO) &
(1 << MV64x60_IRQ_DOORBELL)))
return -1;
return mv64360_irq_base + MV64x60_IRQ_DOORBELL;
}
#endif
irq = mv64x60_read(&bh, MV64360_IC_MAIN_CAUSE_LO);
irq = __ilog2((irq & 0x3dfffffe) & ppc_cached_irq_mask[0]);
if (irq == -1) {
irq = mv64x60_read(&bh, MV64360_IC_MAIN_CAUSE_HI);
irq = __ilog2((irq & 0x1f0003f7) & ppc_cached_irq_mask[1]);
if (irq == -1)
irq = -2; /* bogus interrupt, should never happen */
else {
if ((irq >= 24) && (irq < MV64x60_IRQ_DOORBELL)) {
irq_gpp = mv64x60_read(&bh,
MV64x60_GPP_INTR_CAUSE);
irq_gpp = __ilog2(irq_gpp &
ppc_cached_irq_mask[2]);
if (irq_gpp == -1)
irq = -2;
else {
irq = irq_gpp + 64;
mv64x60_write(&bh,
MV64x60_GPP_INTR_CAUSE,
~(1 << (irq - 64)));
}
}
else
irq += 32;
}
}
(void)mv64x60_read(&bh, MV64x60_GPP_INTR_CAUSE);
if (irq < 0)
return (irq);
else
return (mv64360_irq_base + irq);
}
void __init
adjust_total_lowmem(void)
{
unsigned long max_low_mem = MAX_LOW_MEM;
#ifdef HAVE_BATS
unsigned long bat_max = 0x10000000;
unsigned long align;
unsigned long ram = total_lowmem;
int is601 = 0;
/* 601s have smaller BATs */
if (PVR_VER(mfspr(PVR)) == 1) {
bat_max = 0x00800000;
is601 = 1;
}
/* Make sure we don't map a block larger than the
smallest alignment of the physical address. */
/* alignment of ram_phys_base */
align = ~(ram_phys_base-1) & ram_phys_base;
/* set BAT block size to MIN(max_size, align) */
if (align && align < bat_max)
bat_max = align;
/* Calculate BAT values */
__bat2 = 1UL << __ilog2(ram);
if (__bat2 > bat_max)
__bat2 = bat_max;
ram -= __bat2;
if (ram) {
__bat3 = 1UL << __ilog2(ram);
if (__bat3 > bat_max)
__bat3 = bat_max;
ram -= __bat3;
}
printk(KERN_INFO "Memory BAT mapping: BAT2=%ldMb, BAT3=%ldMb, residual: %ldMb\n",
__bat2 >> 20, __bat3 >> 20, ram >> 20);
/* On SMP, we limit the lowmem to the area mapped with BATs.
* We also assume nobody will do SMP with 601s
*/
#ifdef CONFIG_SMP
if (!is601)
max_low_mem = __bat2 + __bat3;
#endif /* CONFIG_SMP */
#endif /* HAVE_BATS */
if (total_lowmem > max_low_mem) {
total_lowmem = max_low_mem;
#ifndef CONFIG_HIGHMEM
printk(KERN_INFO "Warning, memory limited to %ld Mb, use CONFIG_HIGHMEM"
" to reach %ld Mb\n", max_low_mem >> 20, total_lowmem >> 20);
total_memory = total_lowmem;
#endif /* CONFIG_HIGHMEM */
}
static void __init
mpc86xx_setup_pcie(struct pci_controller *hose, u32 pcie_offset, u32 lower_mem,
u32 upper_mem, u32 io_base)
{
volatile struct ccsr_pcie *pcie;
u16 cmd;
bd_t *binfo = (bd_t *) __res;
pcie = ioremap(binfo->bi_immr_base + pcie_offset,
MPC86xx_PCIE_SIZE);
early_read_config_word(hose, 0, 0, PCI_COMMAND, &cmd);
cmd |= PCI_COMMAND_SERR | PCI_COMMAND_MASTER |
PCI_COMMAND_MEMORY | PCI_COMMAND_IO;
early_write_config_word(hose, 0, 0, PCI_COMMAND, cmd);
early_write_config_byte(hose, 0, 0, PCI_LATENCY_TIMER, 0x80);
/* Disable all windows (except pcieowar0 since its ignored) */
pcie->pcieowar1 = 0;
pcie->pcieowar2 = 0;
pcie->pcieowar3 = 0;
pcie->pcieowar4 = 0;
pcie->pcieiwar1 = 0;
pcie->pcieiwar1 = 0;
pcie->pcieiwar2 = 0;
pcie->pcieiwar3 = 0;
/* Setup Phys:PCIE 1:1 outbound mem window @ MPC86XX_PCIEn_LOWER_MEM */
pcie->pcieotar1 = (lower_mem >> 12) & 0x000fffff;
pcie->pcieotear1 = 0x00000000;
pcie->pcieowbar1 = (lower_mem >> 12) & 0x000fffff;
/* Enable, Mem R/W */
pcie->pcieowar1 = 0x80044000 |
(__ilog2(upper_mem - lower_mem + 1) - 1);
/* Setup outbound IO windows @ MPC86XX_PCIEn_IO_BASE */
pcie->pcieotar2 = (MPC86XX_PCIE_LOWER_IO >> 12) & 0x000fffff;
pcie->pcieotear2 = 0x00000000;
pcie->pcieowbar2 = (io_base >> 12) & 0x000fffff;
/* Enable, IO R/W */
pcie->pcieowar2 = 0x80088000 | (__ilog2(MPC86XX_PCIE_IO_SIZE) - 1);
/* Setup 2G inbound Memory Window @ 0 */
pcie->pcieitar1 = 0x00000000;
pcie->pcieiwbar1 = 0x00000000;
/* Enable, Prefetch, Local Mem, Snoop R/W, 2G */
pcie->pcieiwar1 = 0xa0f5501e;
}
unsigned int mv64x60_get_irq(void)
{
u32 cause;
int level1;
irq_hw_number_t hwirq;
int virq = 0;
cause = in_le32(mv64x60_irq_reg_base + MV64X60_IC_CPU0_SELECT_CAUSE);
if (cause & MV64X60_SELECT_CAUSE_HIGH) {
cause &= mv64x60_cached_high_mask;
level1 = MV64x60_LEVEL1_HIGH;
if (cause & MV64X60_HIGH_GPP_GROUPS) {
cause = in_le32(mv64x60_gpp_reg_base +
MV64x60_GPP_INTR_CAUSE);
cause &= mv64x60_cached_gpp_mask;
level1 = MV64x60_LEVEL1_GPP;
}
} else {
cause &= mv64x60_cached_low_mask;
level1 = MV64x60_LEVEL1_LOW;
}
if (cause) {
hwirq = (level1 << MV64x60_LEVEL1_OFFSET) | __ilog2(cause);
virq = irq_linear_revmap(mv64x60_irq_host, hwirq);
}
return virq;
}
/* atmu setup for fsl pci/pcie controller */
void __init setup_pci_atmu(struct pci_controller *hose, struct resource *rsrc)
{
struct ccsr_pci __iomem *pci;
int i;
pr_debug("PCI memory map start 0x%016llx, size 0x%016llx\n",
(u64)rsrc->start, (u64)rsrc->end - (u64)rsrc->start + 1);
pci = ioremap(rsrc->start, rsrc->end - rsrc->start + 1);
/* Disable all windows (except powar0 since its ignored) */
for(i = 1; i < 5; i++)
out_be32(&pci->pow[i].powar, 0);
for(i = 0; i < 3; i++)
out_be32(&pci->piw[i].piwar, 0);
/* Setup outbound MEM window */
for(i = 0; i < 3; i++)
if (hose->mem_resources[i].flags & IORESOURCE_MEM){
resource_size_t pci_addr_start =
hose->mem_resources[i].start -
hose->pci_mem_offset;
pr_debug("PCI MEM resource start 0x%016llx, size 0x%016llx.\n",
(u64)hose->mem_resources[i].start,
(u64)hose->mem_resources[i].end
- (u64)hose->mem_resources[i].start + 1);
out_be32(&pci->pow[i+1].potar, (pci_addr_start >> 12));
out_be32(&pci->pow[i+1].potear, 0);
out_be32(&pci->pow[i+1].powbar,
(hose->mem_resources[i].start >> 12));
/* Enable, Mem R/W */
out_be32(&pci->pow[i+1].powar, 0x80044000
| (__ilog2(hose->mem_resources[i].end
- hose->mem_resources[i].start + 1) - 1));
}
/*************************************************************************
* fixed sdram init -- doesn't use serial presence detect.
************************************************************************/
int fixed_sdram(void)
{
immap_t *im = (immap_t *) CFG_IMMR;
u32 msize = CFG_DDR_SIZE * 1024 * 1024;
u32 msize_log2 = __ilog2(msize);
im->sysconf.ddrlaw[0].bar = CFG_DDR_SDRAM_BASE >> 12;
im->sysconf.ddrlaw[0].ar = LBLAWAR_EN | (msize_log2 - 1);
im->sysconf.ddrcdr = CFG_DDRCDR_VALUE;
udelay(50000);
im->ddr.sdram_clk_cntl = CFG_DDR_SDRAM_CLK_CNTL;
udelay(1000);
im->ddr.csbnds[0].csbnds = CFG_DDR_CS0_BNDS;
im->ddr.cs_config[0] = CFG_DDR_CS0_CONFIG;
udelay(1000);
im->ddr.timing_cfg_0 = CFG_DDR_TIMING_0;
im->ddr.timing_cfg_1 = CFG_DDR_TIMING_1;
im->ddr.timing_cfg_2 = CFG_DDR_TIMING_2;
im->ddr.timing_cfg_3 = CFG_DDR_TIMING_3;
im->ddr.sdram_cfg = CFG_DDR_SDRAM_CFG;
im->ddr.sdram_cfg2 = CFG_DDR_SDRAM_CFG2;
im->ddr.sdram_mode = CFG_DDR_MODE;
im->ddr.sdram_mode2 = CFG_DDR_MODE2;
im->ddr.sdram_interval = CFG_DDR_INTERVAL;
sync();
udelay(1000);
im->ddr.sdram_cfg |= SDRAM_CFG_MEM_EN;
udelay(2000);
return CFG_DDR_SIZE;
}
static int sun4i_spi_set_speed(struct udevice *dev, uint speed)
{
struct sun4i_spi_platdata *plat = dev_get_platdata(dev);
struct sun4i_spi_priv *priv = dev_get_priv(dev);
unsigned int div;
u32 reg;
if (speed > plat->max_hz)
speed = plat->max_hz;
if (speed < SUN4I_SPI_MIN_RATE)
speed = SUN4I_SPI_MIN_RATE;
/*
* Setup clock divider.
*
* We have two choices there. Either we can use the clock
* divide rate 1, which is calculated thanks to this formula:
* SPI_CLK = MOD_CLK / (2 ^ (cdr + 1))
* Or we can use CDR2, which is calculated with the formula:
* SPI_CLK = MOD_CLK / (2 * (cdr + 1))
* Whether we use the former or the latter is set through the
* DRS bit.
*
* First try CDR2, and if we can't reach the expected
* frequency, fall back to CDR1.
*/
div = SUN4I_SPI_MAX_RATE / (2 * speed);
reg = readl(&priv->regs->cctl);
if (div <= (SUN4I_CLK_CTL_CDR2_MASK + 1)) {
if (div > 0)
div--;
reg &= ~(SUN4I_CLK_CTL_CDR2_MASK | SUN4I_CLK_CTL_DRS);
reg |= SUN4I_CLK_CTL_CDR2(div) | SUN4I_CLK_CTL_DRS;
} else {
div = __ilog2(SUN4I_SPI_MAX_RATE) - __ilog2(speed);
reg &= ~((SUN4I_CLK_CTL_CDR1_MASK << 8) | SUN4I_CLK_CTL_DRS);
reg |= SUN4I_CLK_CTL_CDR1(div);
}
priv->freq = speed;
writel(reg, &priv->regs->cctl);
return 0;
}
/* Chip Select Configuration (CSn_CONFIG) */
static void set_csn_config(int i, fsl_ddr_cfg_regs_t *ddr,
const memctl_options_t *popts,
const dimm_params_t *dimm_params)
{
unsigned int cs_n_en = 0; /* Chip Select enable */
unsigned int intlv_en = 0; /* Memory controller interleave enable */
unsigned int intlv_ctl = 0; /* Interleaving control */
unsigned int ap_n_en = 0; /* Chip select n auto-precharge enable */
unsigned int odt_rd_cfg = 0; /* ODT for reads configuration */
unsigned int odt_wr_cfg = 0; /* ODT for writes configuration */
unsigned int ba_bits_cs_n = 0; /* Num of bank bits for SDRAM on CSn */
unsigned int row_bits_cs_n = 0; /* Num of row bits for SDRAM on CSn */
unsigned int col_bits_cs_n = 0; /* Num of ocl bits for SDRAM on CSn */
/* Compute CS_CONFIG only for existing ranks of each DIMM. */
if ((((i&1) == 0)
&& (dimm_params[i/2].n_ranks == 1))
|| (dimm_params[i/2].n_ranks == 2)) {
unsigned int n_banks_per_sdram_device;
cs_n_en = 1;
if (i == 0) {
/* These fields only available in CS0_CONFIG */
intlv_en = popts->memctl_interleaving;
intlv_ctl = popts->memctl_interleaving_mode;
}
ap_n_en = popts->cs_local_opts[i].auto_precharge;
odt_rd_cfg = popts->cs_local_opts[i].odt_rd_cfg;
odt_wr_cfg = popts->cs_local_opts[i].odt_wr_cfg;
n_banks_per_sdram_device
= dimm_params[i/2].n_banks_per_sdram_device;
ba_bits_cs_n = __ilog2(n_banks_per_sdram_device) - 2;
row_bits_cs_n = dimm_params[i/2].n_row_addr - 12;
col_bits_cs_n = dimm_params[i/2].n_col_addr - 8;
}
/* FIXME: intlv_en, intlv_ctl only on CS0_CONFIG */
if (i != 0) {
intlv_en = 0;
intlv_ctl = 0;
}
ddr->cs[i].config = (0
| ((cs_n_en & 0x1) << 31)
| ((intlv_en & 0x3) << 29)
| ((intlv_en & 0xf) << 24)
| ((ap_n_en & 0x1) << 23)
/* XXX: some implementation only have 1 bit starting at left */
| ((odt_rd_cfg & 0x7) << 20)
/* XXX: Some implementation only have 1 bit starting at left */
| ((odt_wr_cfg & 0x7) << 16)
| ((ba_bits_cs_n & 0x3) << 14)
| ((row_bits_cs_n & 0x7) << 8)
| ((col_bits_cs_n & 0x7) << 0)
);
}
static void __init
mpc85xx_setup_pci2(struct pci_controller *hose)
{
volatile struct ccsr_pci *pci;
unsigned short temps;
bd_t *binfo = (bd_t *) __res;
pci = ioremap(binfo->bi_immr_base + MPC85xx_PCI2_OFFSET,
MPC85xx_PCI2_SIZE);
early_read_config_word(hose, hose->bus_offset, 0, PCI_COMMAND, &temps);
temps |= PCI_COMMAND_SERR | PCI_COMMAND_MASTER | PCI_COMMAND_MEMORY;
early_write_config_word(hose, hose->bus_offset, 0, PCI_COMMAND, temps);
early_write_config_byte(hose, hose->bus_offset, 0, PCI_LATENCY_TIMER, 0x80);
/* Disable all windows (except powar0 since its ignored) */
pci->powar1 = 0;
pci->powar2 = 0;
pci->powar3 = 0;
pci->powar4 = 0;
pci->piwar1 = 0;
pci->piwar2 = 0;
pci->piwar3 = 0;
/* Setup Phys:PCI 1:1 outbound mem window @ MPC85XX_PCI2_LOWER_MEM */
pci->potar1 = (MPC85XX_PCI2_LOWER_MEM >> 12) & 0x000fffff;
pci->potear1 = 0x00000000;
pci->powbar1 = (MPC85XX_PCI2_LOWER_MEM >> 12) & 0x000fffff;
/* Enable, Mem R/W */
pci->powar1 = 0x80044000 |
(__ilog2(MPC85XX_PCI2_UPPER_MEM - MPC85XX_PCI2_LOWER_MEM + 1) - 1);
/* Setup outbound IO windows @ MPC85XX_PCI2_IO_BASE */
pci->potar2 = (MPC85XX_PCI2_LOWER_IO >> 12) & 0x000fffff;
pci->potear2 = 0x00000000;
pci->powbar2 = (MPC85XX_PCI2_IO_BASE >> 12) & 0x000fffff;
/* Enable, IO R/W */
pci->powar2 = 0x80088000 | (__ilog2(MPC85XX_PCI2_IO_SIZE) - 1);
/* Setup 2G inbound Memory Window @ 0 */
pci->pitar1 = 0x00000000;
pci->piwbar1 = 0x00000000;
pci->piwar1 = 0xa0f5501e; /* Enable, Prefetch, Local
Mem, Snoop R/W, 2G */
}
const char * step_to_string(unsigned int step) {
unsigned int s = __ilog2(step);
if ((1 << s) != step)
return step_string_tbl[7];
return step_string_tbl[s];
}
/*
* gt64260_get_irq()
*
* This function returns the lowest interrupt number of all interrupts that
* are currently asserted.
*
* Input Variable(s):
* struct pt_regs* not used
*
* Output Variable(s):
* None.
*
* Returns:
* int <interrupt number> or -2 (bogus interrupt)
*/
int
gt64260_get_irq(struct pt_regs *regs)
{
int irq;
int irq_gpp;
irq = mv64x60_read(&bh, GT64260_IC_MAIN_CAUSE_LO);
irq = __ilog2((irq & 0x3dfffffe) & ppc_cached_irq_mask[0]);
if (irq == -1) {
irq = mv64x60_read(&bh, GT64260_IC_MAIN_CAUSE_HI);
irq = __ilog2((irq & 0x0f000db7) & ppc_cached_irq_mask[1]);
if (irq == -1)
irq = -2; /* bogus interrupt, should never happen */
else {
if (irq >= 24) {
irq_gpp = mv64x60_read(&bh,
MV64x60_GPP_INTR_CAUSE);
irq_gpp = __ilog2(irq_gpp &
ppc_cached_irq_mask[2]);
if (irq_gpp == -1)
irq = -2;
else {
irq = irq_gpp + 64;
mv64x60_write(&bh,
MV64x60_GPP_INTR_CAUSE,
~(1 << (irq - 64)));
}
} else
irq += 32;
}
}
(void)mv64x60_read(&bh, MV64x60_GPP_INTR_CAUSE);
if (irq < 0)
return (irq);
else
return (gt64260_irq_base + irq);
}
void __init ps3_hpte_init(unsigned long htab_size)
{
ppc_md.hpte_invalidate = ps3_hpte_invalidate;
ppc_md.hpte_updatepp = ps3_hpte_updatepp;
ppc_md.hpte_updateboltedpp = ps3_hpte_updateboltedpp;
ppc_md.hpte_insert = ps3_hpte_insert;
ppc_md.hpte_remove = ps3_hpte_remove;
ppc_md.hpte_clear_all = ps3_hpte_clear;
ppc64_pft_size = __ilog2(htab_size);
}
phys_size_t initdram(int board_type)
{
u32 msize;
volatile immap_t *immr = (immap_t *)CONFIG_SYS_IMMR;
volatile clk83xx_t *clk = (clk83xx_t *)&immr->clk;
/* Enable PCI_CLK[0:1] */
clk->occr |= 0xc0000000;
udelay(2000);
#if defined(CONFIG_SPD_EEPROM)
msize = spd_sdram();
#else
immap_t *im = (immap_t *) CONFIG_SYS_IMMR;
u32 msize_log2;
msize = CONFIG_SYS_DDR_SIZE;
msize_log2 = __ilog2(msize);
im->sysconf.ddrlaw[0].bar = CONFIG_SYS_DDR_SDRAM_BASE & 0xfffff000;
im->sysconf.ddrlaw[0].ar = LBLAWAR_EN | (msize_log2 - 1);
im->sysconf.ddrcdr = CONFIG_SYS_DDRCDR_VALUE;
udelay(50000);
im->ddr.sdram_clk_cntl = CONFIG_SYS_DDR_SDRAM_CLK_CNTL;
udelay(1000);
im->ddr.csbnds[0].csbnds = CONFIG_SYS_DDR_CS0_BNDS;
im->ddr.cs_config[0] = CONFIG_SYS_DDR_CS0_CONFIG;
udelay(1000);
im->ddr.timing_cfg_0 = CONFIG_SYS_DDR_TIMING_0;
im->ddr.timing_cfg_1 = CONFIG_SYS_DDR_TIMING_1;
im->ddr.timing_cfg_2 = CONFIG_SYS_DDR_TIMING_2;
im->ddr.timing_cfg_3 = CONFIG_SYS_DDR_TIMING_3;
im->ddr.sdram_cfg = CONFIG_SYS_DDR_SDRAM_CFG;
im->ddr.sdram_cfg2 = CONFIG_SYS_DDR_SDRAM_CFG2;
im->ddr.sdram_mode = CONFIG_SYS_DDR_MODE;
im->ddr.sdram_mode2 = CONFIG_SYS_DDR_MODE2;
im->ddr.sdram_interval = CONFIG_SYS_DDR_INTERVAL;
__asm__ __volatile__("sync");
udelay(1000);
im->ddr.sdram_cfg |= SDRAM_CFG_MEM_EN;
udelay(2000);
#endif
setup_serdes();
return msize << 20;
}
unsigned long calc_cam_sz(unsigned long ram, unsigned long virt,
phys_addr_t phys)
{
unsigned int camsize = __ilog2(ram);
unsigned int align = __ffs(virt | phys);
unsigned long max_cam;
if ((mfspr(SPRN_MMUCFG) & MMUCFG_MAVN) == MMUCFG_MAVN_V1) {
/* Convert (4^max) kB to (2^max) bytes */
max_cam = ((mfspr(SPRN_TLB1CFG) >> 16) & 0xf) * 2 + 10;
camsize &= ~1U;
align &= ~1U;
} else {
const char * step_to_string(unsigned int step) {
unsigned int s = __ilog2(step);
if ((1 << s) != step)
return step_string_tbl[7];
if (s >= ARRAY_SIZE(step_string_tbl)) {
printf("Error for the step in %s\n", __func__);
s = 0;
}
return step_string_tbl[s];
}
/* Fixed sdram init -- doesn't use serial presence detect.
*
* This is useful for faster booting in configs where the RAM is unlikely
* to be changed, or for things like NAND booting where space is tight.
*/
static long fixed_sdram(void)
{
volatile immap_t *im = (volatile immap_t *)CFG_IMMR;
u32 msize = CFG_DDR_SIZE * 1024 * 1024;
u32 msize_log2 = __ilog2(msize);
im->sysconf.ddrlaw[0].bar = CFG_DDR_SDRAM_BASE >> 12;
im->sysconf.ddrlaw[0].ar = LBLAWAR_EN | (msize_log2 - 1);
im->sysconf.ddrcdr = CFG_DDRCDR_VALUE;
/*
* Erratum DDR3 requires a 50ms delay after clearing DDRCDR[DDR_cfg],
* or the DDR2 controller may fail to initialize correctly.
*/
udelay(50000);
im->ddr.csbnds[0].csbnds = (msize - 1) >> 24;
im->ddr.cs_config[0] = CFG_DDR_CONFIG;
/* Currently we use only one CS, so disable the other bank. */
im->ddr.cs_config[1] = 0;
im->ddr.sdram_clk_cntl = CFG_DDR_CLK_CNTL;
im->ddr.timing_cfg_3 = CFG_DDR_TIMING_3;
im->ddr.timing_cfg_1 = CFG_DDR_TIMING_1;
im->ddr.timing_cfg_2 = CFG_DDR_TIMING_2;
im->ddr.timing_cfg_0 = CFG_DDR_TIMING_0;
#ifndef CFG_8313ERDB_BROKEN_PMC
if (im->pmc.pmccr1 & PMCCR1_POWER_OFF)
im->ddr.sdram_cfg = CFG_SDRAM_CFG | SDRAM_CFG_BI;
else
#endif
im->ddr.sdram_cfg = CFG_SDRAM_CFG;
im->ddr.sdram_cfg2 = CFG_SDRAM_CFG2;
im->ddr.sdram_mode = CFG_DDR_MODE;
im->ddr.sdram_mode2 = CFG_DDR_MODE_2;
im->ddr.sdram_interval = CFG_DDR_INTERVAL;
sync();
/* enable DDR controller */
im->ddr.sdram_cfg |= SDRAM_CFG_MEM_EN;
return msize;
}
/* Depending on whether this is called from iSeries or pSeries setup
* code, the location of the msChunks struct may or may not have
* to be reloc'd, so we force the caller to do that for us by passing
* in a pointer to the structure.
*/
unsigned long
msChunks_alloc(unsigned long mem, unsigned long num_chunks, unsigned long chunk_size)
{
unsigned long offset = reloc_offset();
struct msChunks *_msChunks = PTRRELOC(&msChunks);
_msChunks->num_chunks = num_chunks;
_msChunks->chunk_size = chunk_size;
_msChunks->chunk_shift = __ilog2(chunk_size);
_msChunks->chunk_mask = (1UL<<_msChunks->chunk_shift)-1;
mem = _ALIGN(mem, sizeof(msChunks_entry));
_msChunks->abs = (msChunks_entry *)(mem + offset);
mem += num_chunks * sizeof(msChunks_entry);
return mem;
}
static void iommu_buses_init(void)
{
struct pci_controller* phb;
struct device_node *dn, *first_dn;
int num_slots, num_slots_ilog2;
int first_phb = 1;
unsigned long tcetable_ilog2;
/*
* We default to a TCE table that maps 2GB (4MB table, 22 bits),
* however some machines have a 3GB IO hole and for these we
* create a table that maps 1GB (2MB table, 21 bits)
*/
if (io_hole_start < 0x80000000UL)
tcetable_ilog2 = 21;
else
tcetable_ilog2 = 22;
/* XXX Should we be using pci_root_buses instead? -ojn
*/
for (phb=hose_head; phb; phb=phb->next) {
first_dn = ((struct device_node *)phb->arch_data)->child;
/* Carve 2GB into the largest dma_window_size possible */
for (dn = first_dn, num_slots = 0; dn != NULL; dn = dn->sibling)
num_slots++;
num_slots_ilog2 = __ilog2(num_slots);
if ((1<<num_slots_ilog2) != num_slots)
num_slots_ilog2++;
phb->dma_window_size = 1 << (tcetable_ilog2 - num_slots_ilog2);
/* Reserve 16MB of DMA space on the first PHB.
* We should probably be more careful and use firmware props.
* In reality this space is remapped, not lost. But we don't
* want to get that smart to handle it -- too much work.
*/
phb->dma_window_base_cur = first_phb ? (1 << 12) : 0;
first_phb = 0;
for (dn = first_dn; dn != NULL; dn = dn->sibling)
iommu_devnode_init(dn);
}
}
unsigned long calc_cam_sz(unsigned long ram, unsigned long virt,
phys_addr_t phys)
{
unsigned int camsize = __ilog2(ram) & ~1U;
unsigned int align = __ffs(virt | phys) & ~1U;
unsigned long max_cam = (mfspr(SPRN_TLB1CFG) >> 16) & 0xf;
/* Convert (4^max) kB to (2^max) bytes */
max_cam = max_cam * 2 + 10;
if (camsize > align)
camsize = align;
if (camsize > max_cam)
camsize = max_cam;
return 1UL << camsize;
}
int set_last_law(phys_addr_t addr, enum law_size sz, enum law_trgt_if id)
{
u32 idx;
/* we have no LAWs free */
if (gd->used_laws == -1)
return -1;
/* grab the last free law */
idx = __ilog2(~(gd->used_laws));
if (idx >= FSL_HW_NUM_LAWS)
return -1;
set_law(idx, addr, sz, id);
return idx;
}
static void init_cache_info(struct ppc_cache_info *info, u32 size, u32 lsize,
u32 bsize, u32 sets)
{
info->size = size;
info->sets = sets;
info->line_size = lsize;
info->block_size = bsize;
info->log_block_size = __ilog2(bsize);
if (bsize)
info->blocks_per_page = PAGE_SIZE / bsize;
else
info->blocks_per_page = 0;
if (sets == 0)
info->assoc = 0xffff;
else
info->assoc = size / (sets * lsize);
}
static void stm32_sdmmc2_start_data(struct stm32_sdmmc2_priv *priv,
struct mmc_data *data,
struct stm32_sdmmc2_ctx *ctx)
{
u32 data_ctrl, idmabase0;
/* Configure the SDMMC DPSM (Data Path State Machine) */
data_ctrl = (__ilog2(data->blocksize) <<
SDMMC_DCTRL_DBLOCKSIZE_SHIFT) &
SDMMC_DCTRL_DBLOCKSIZE;
if (data->flags & MMC_DATA_READ) {
data_ctrl |= SDMMC_DCTRL_DTDIR;
idmabase0 = (u32)data->dest;
} else {
idmabase0 = (u32)data->src;
}
/* Set the SDMMC Data TimeOut value */
writel(SDMMC_CMD_TIMEOUT, priv->base + SDMMC_DTIMER);
/* Set the SDMMC DataLength value */
writel(ctx->data_length, priv->base + SDMMC_DLEN);
/* Write to SDMMC DCTRL */
writel(data_ctrl, priv->base + SDMMC_DCTRL);
/* Cache align */
ctx->cache_start = rounddown(idmabase0, ARCH_DMA_MINALIGN);
ctx->cache_end = roundup(idmabase0 + ctx->data_length,
ARCH_DMA_MINALIGN);
/*
* Flush data cache before DMA start (clean and invalidate)
* Clean also needed for read
* Avoid issue on buffer not cached-aligned
*/
flush_dcache_range(ctx->cache_start, ctx->cache_end);
/* Enable internal DMA */
writel(idmabase0, priv->base + SDMMC_IDMABASE0);
writel(SDMMC_IDMACTRL_IDMAEN, priv->base + SDMMC_IDMACTRL);
}
/*
* Set up a variable-size TLB entry (tlbcam). The parameters are not checked;
* in particular size must be a power of 4 between 4k and the max supported by
* an implementation; max may further be limited by what can be represented in
* an unsigned long (for example, 32-bit implementations cannot support a 4GB
* size).
*/
void settlbcam(int index, unsigned long virt, phys_addr_t phys,
unsigned long size, unsigned long flags, unsigned int pid)
{
unsigned int tsize;
tsize = __ilog2(size) - 10;
#ifdef CONFIG_SMP
if ((flags & _PAGE_NO_CACHE) == 0)
flags |= _PAGE_COHERENT;
#endif
TLBCAM[index].MAS0 = MAS0_TLBSEL(1) | MAS0_ESEL(index) | MAS0_NV(index+1);
TLBCAM[index].MAS1 = MAS1_VALID | MAS1_IPROT | MAS1_TSIZE(tsize) | MAS1_TID(pid);
TLBCAM[index].MAS2 = virt & PAGE_MASK;
TLBCAM[index].MAS2 |= (flags & _PAGE_WRITETHRU) ? MAS2_W : 0;
TLBCAM[index].MAS2 |= (flags & _PAGE_NO_CACHE) ? MAS2_I : 0;
TLBCAM[index].MAS2 |= (flags & _PAGE_COHERENT) ? MAS2_M : 0;
TLBCAM[index].MAS2 |= (flags & _PAGE_GUARDED) ? MAS2_G : 0;
TLBCAM[index].MAS2 |= (flags & _PAGE_ENDIAN) ? MAS2_E : 0;
TLBCAM[index].MAS3 = (phys & MAS3_RPN) | MAS3_SX | MAS3_SR;
TLBCAM[index].MAS3 |= ((flags & _PAGE_RW) ? MAS3_SW : 0);
if (mmu_has_feature(MMU_FTR_BIG_PHYS))
TLBCAM[index].MAS7 = (u64)phys >> 32;
/* Below is unlikely -- only for large user pages or similar */
if (pte_user(flags)) {
TLBCAM[index].MAS3 |= MAS3_UX | MAS3_UR;
TLBCAM[index].MAS3 |= ((flags & _PAGE_RW) ? MAS3_UW : 0);
}
tlbcam_addrs[index].start = virt;
tlbcam_addrs[index].limit = virt + size - 1;
tlbcam_addrs[index].phys = phys;
loadcam_entry(index);
}
/*************************************************************************
* fixed sdram init -- doesn't use serial presence detect.
************************************************************************/
int fixed_sdram(void)
{
volatile immap_t *im = (immap_t *) CONFIG_SYS_IMMR;
u32 msize = CONFIG_SYS_DDR_SIZE * 1024 * 1024;
u32 msize_log2 = __ilog2(msize);
im->sysconf.ddrlaw[0].bar = CONFIG_SYS_DDR_SDRAM_BASE & 0xfffff000;
im->sysconf.ddrlaw[0].ar = LBLAWAR_EN | (msize_log2 - 1);
#if (CONFIG_SYS_DDR_SIZE != 512)
#warning Currenly any ddr size other than 512 is not supported
#endif
im->sysconf.ddrcdr = CONFIG_SYS_DDRCDR_VALUE;
udelay(50000);
im->ddr.sdram_clk_cntl = CONFIG_SYS_DDR_SDRAM_CLK_CNTL;
udelay(1000);
im->ddr.csbnds[0].csbnds = CONFIG_SYS_DDR_CS0_BNDS;
im->ddr.cs_config[0] = CONFIG_SYS_DDR_CS0_CONFIG;
udelay(1000);
im->ddr.timing_cfg_0 = CONFIG_SYS_DDR_TIMING_0;
im->ddr.timing_cfg_1 = CONFIG_SYS_DDR_TIMING_1;
im->ddr.timing_cfg_2 = CONFIG_SYS_DDR_TIMING_2;
im->ddr.timing_cfg_3 = CONFIG_SYS_DDR_TIMING_3;
im->ddr.sdram_cfg = CONFIG_SYS_DDR_SDRAM_CFG;
im->ddr.sdram_cfg2 = CONFIG_SYS_DDR_SDRAM_CFG2;
im->ddr.sdram_mode = CONFIG_SYS_DDR_MODE;
im->ddr.sdram_mode2 = CONFIG_SYS_DDR_MODE2;
im->ddr.sdram_interval = CONFIG_SYS_DDR_INTERVAL;
__asm__ __volatile__("sync");
udelay(1000);
im->ddr.sdram_cfg |= SDRAM_CFG_MEM_EN;
udelay(2000);
return CONFIG_SYS_DDR_SIZE;
}
static unsigned long __init htab_get_table_size(void)
{
unsigned long mem_size, rnd_mem_size, pteg_count;
/* If hash size isn't already provided by the platform, we try to
* retrieve it from the device-tree. If it's not there neither, we
* calculate it now based on the total RAM size
*/
if (ppc64_pft_size == 0)
of_scan_flat_dt(htab_dt_scan_pftsize, NULL);
if (ppc64_pft_size)
return 1UL << ppc64_pft_size;
/* round mem_size up to next power of 2 */
mem_size = lmb_phys_mem_size();
rnd_mem_size = 1UL << __ilog2(mem_size);
if (rnd_mem_size < mem_size)
rnd_mem_size <<= 1;
/* # pages / 2 */
pteg_count = max(rnd_mem_size >> (12 + 1), 1UL << 11);
return pteg_count << 7;
}
/*
* Initialize some remaining members of the ppc64_caches and systemcfg
* structures
* (at least until we get rid of them completely). This is mostly some
* cache informations about the CPU that will be used by cache flush
* routines and/or provided to userland
*/
static void __init initialize_cache_info(void)
{
struct device_node *np;
unsigned long num_cpus = 0;
DBG(" -> initialize_cache_info()\n");
for (np = NULL; (np = of_find_node_by_type(np, "cpu"));) {
num_cpus += 1;
/* We're assuming *all* of the CPUs have the same
* d-cache and i-cache sizes... -Peter
*/
if ( num_cpus == 1 ) {
const u32 *sizep, *lsizep;
u32 size, lsize;
size = 0;
lsize = cur_cpu_spec->dcache_bsize;
sizep = of_get_property(np, "d-cache-size", NULL);
if (sizep != NULL)
size = *sizep;
lsizep = of_get_property(np, "d-cache-block-size", NULL);
/* fallback if block size missing */
if (lsizep == NULL)
lsizep = of_get_property(np, "d-cache-line-size", NULL);
if (lsizep != NULL)
lsize = *lsizep;
if (sizep == 0 || lsizep == 0)
DBG("Argh, can't find dcache properties ! "
"sizep: %p, lsizep: %p\n", sizep, lsizep);
ppc64_caches.dsize = size;
ppc64_caches.dline_size = lsize;
ppc64_caches.log_dline_size = __ilog2(lsize);
ppc64_caches.dlines_per_page = PAGE_SIZE / lsize;
size = 0;
lsize = cur_cpu_spec->icache_bsize;
sizep = of_get_property(np, "i-cache-size", NULL);
if (sizep != NULL)
size = *sizep;
lsizep = of_get_property(np, "i-cache-block-size", NULL);
if (lsizep == NULL)
lsizep = of_get_property(np, "i-cache-line-size", NULL);
if (lsizep != NULL)
lsize = *lsizep;
if (sizep == 0 || lsizep == 0)
DBG("Argh, can't find icache properties ! "
"sizep: %p, lsizep: %p\n", sizep, lsizep);
ppc64_caches.isize = size;
ppc64_caches.iline_size = lsize;
ppc64_caches.log_iline_size = __ilog2(lsize);
ppc64_caches.ilines_per_page = PAGE_SIZE / lsize;
}
}
DBG(" <- initialize_cache_info()\n");
}
static void pci_init_bus(int bus, struct pci_region *reg)
{
volatile immap_t *immr = (volatile immap_t *)CONFIG_SYS_IMMR;
volatile pot83xx_t *pot = immr->ios.pot;
volatile pcictrl83xx_t *pci_ctrl = &immr->pci_ctrl[bus];
struct pci_controller *hose = &pci_hose[bus];
u32 dev;
u16 reg16;
int i;
if (bus == 1)
pot += 3;
/* Setup outbound translation windows */
for (i = 0; i < 3; i++, reg++, pot++) {
if (reg->size == 0)
break;
hose->regions[i] = *reg;
hose->region_count++;
pot->potar = reg->bus_start >> 12;
pot->pobar = reg->phys_start >> 12;
pot->pocmr = ~(reg->size - 1) >> 12;
if (reg->flags & PCI_REGION_IO)
pot->pocmr |= POCMR_IO;
#ifdef CONFIG_83XX_PCI_STREAMING
else if (reg->flags & PCI_REGION_PREFETCH)
pot->pocmr |= POCMR_SE;
#endif
if (bus == 1)
pot->pocmr |= POCMR_DST;
pot->pocmr |= POCMR_EN;
}
/* Point inbound translation at RAM */
pci_ctrl->pitar1 = 0;
pci_ctrl->pibar1 = 0;
pci_ctrl->piebar1 = 0;
pci_ctrl->piwar1 = PIWAR_EN | PIWAR_PF | PIWAR_RTT_SNOOP |
PIWAR_WTT_SNOOP | (__ilog2(gd->ram_size - 1));
i = hose->region_count++;
hose->regions[i].bus_start = 0;
hose->regions[i].phys_start = 0;
hose->regions[i].size = gd->ram_size;
hose->regions[i].flags = PCI_REGION_MEM | PCI_REGION_SYS_MEMORY;
hose->first_busno = pci_last_busno() + 1;
hose->last_busno = 0xff;
pci_setup_indirect(hose, CONFIG_SYS_IMMR + 0x8300 + bus * 0x80,
CONFIG_SYS_IMMR + 0x8304 + bus * 0x80);
pci_register_hose(hose);
/*
* Write to Command register
*/
reg16 = 0xff;
dev = PCI_BDF(hose->first_busno, 0, 0);
pci_hose_read_config_word(hose, dev, PCI_COMMAND, ®16);
reg16 |= PCI_COMMAND_SERR | PCI_COMMAND_MASTER | PCI_COMMAND_MEMORY;
pci_hose_write_config_word(hose, dev, PCI_COMMAND, reg16);
/*
* Clear non-reserved bits in status register.
*/
pci_hose_write_config_word(hose, dev, PCI_STATUS, 0xffff);
pci_hose_write_config_byte(hose, dev, PCI_LATENCY_TIMER, 0x80);
pci_hose_write_config_byte(hose, dev, PCI_CACHE_LINE_SIZE, 0x08);
#ifdef CONFIG_PCI_SCAN_SHOW
printf("PCI: Bus Dev VenId DevId Class Int\n");
#endif
#ifndef CONFIG_PCISLAVE
/*
* Hose scan.
*/
hose->last_busno = pci_hose_scan(hose);
#endif
}
/*
* fixed sdram init:
* The board doesn't use memory modules that have serial presence
* detect or similar mechanism for discovery of the DRAM settings
*/
long int fixed_sdram(ddr512x_config_t *mddrc_config,
u32 *dram_init_seq, int seq_sz)
{
volatile immap_t *im = (immap_t *)CONFIG_SYS_IMMR;
u32 msize = CONFIG_SYS_MAX_RAM_SIZE;
u32 msize_log2 = __ilog2(msize);
u32 i;
/* take default settings and init sequence if necessary */
if (mddrc_config == NULL)
mddrc_config = &default_mddrc_config;
if (dram_init_seq == NULL) {
dram_init_seq = default_init_seq;
seq_sz = sizeof(default_init_seq)/sizeof(u32);
}
/* Initialize IO Control */
out_be32(&im->io_ctrl.io_control_mem, CONFIG_SYS_IOCTRL_MUX_DDR);
/* Initialize DDR Local Window */
out_be32(&im->sysconf.ddrlaw.bar, CONFIG_SYS_DDR_BASE & 0xFFFFF000);
out_be32(&im->sysconf.ddrlaw.ar, msize_log2 - 1);
sync_law(&im->sysconf.ddrlaw.ar);
/* DDR Enable */
out_be32(&im->mddrc.ddr_sys_config, MDDRC_SYS_CFG_EN);
/* Initialize DDR Priority Manager */
out_be32(&im->mddrc.prioman_config1, CONFIG_SYS_MDDRCGRP_PM_CFG1);
out_be32(&im->mddrc.prioman_config2, CONFIG_SYS_MDDRCGRP_PM_CFG2);
out_be32(&im->mddrc.hiprio_config, CONFIG_SYS_MDDRCGRP_HIPRIO_CFG);
out_be32(&im->mddrc.lut_table0_main_upper, CONFIG_SYS_MDDRCGRP_LUT0_MU);
out_be32(&im->mddrc.lut_table0_main_lower, CONFIG_SYS_MDDRCGRP_LUT0_ML);
out_be32(&im->mddrc.lut_table1_main_upper, CONFIG_SYS_MDDRCGRP_LUT1_MU);
out_be32(&im->mddrc.lut_table1_main_lower, CONFIG_SYS_MDDRCGRP_LUT1_ML);
out_be32(&im->mddrc.lut_table2_main_upper, CONFIG_SYS_MDDRCGRP_LUT2_MU);
out_be32(&im->mddrc.lut_table2_main_lower, CONFIG_SYS_MDDRCGRP_LUT2_ML);
out_be32(&im->mddrc.lut_table3_main_upper, CONFIG_SYS_MDDRCGRP_LUT3_MU);
out_be32(&im->mddrc.lut_table3_main_lower, CONFIG_SYS_MDDRCGRP_LUT3_ML);
out_be32(&im->mddrc.lut_table4_main_upper, CONFIG_SYS_MDDRCGRP_LUT4_MU);
out_be32(&im->mddrc.lut_table4_main_lower, CONFIG_SYS_MDDRCGRP_LUT4_ML);
out_be32(&im->mddrc.lut_table0_alternate_upper, CONFIG_SYS_MDDRCGRP_LUT0_AU);
out_be32(&im->mddrc.lut_table0_alternate_lower, CONFIG_SYS_MDDRCGRP_LUT0_AL);
out_be32(&im->mddrc.lut_table1_alternate_upper, CONFIG_SYS_MDDRCGRP_LUT1_AU);
out_be32(&im->mddrc.lut_table1_alternate_lower, CONFIG_SYS_MDDRCGRP_LUT1_AL);
out_be32(&im->mddrc.lut_table2_alternate_upper, CONFIG_SYS_MDDRCGRP_LUT2_AU);
out_be32(&im->mddrc.lut_table2_alternate_lower, CONFIG_SYS_MDDRCGRP_LUT2_AL);
out_be32(&im->mddrc.lut_table3_alternate_upper, CONFIG_SYS_MDDRCGRP_LUT3_AU);
out_be32(&im->mddrc.lut_table3_alternate_lower, CONFIG_SYS_MDDRCGRP_LUT3_AL);
out_be32(&im->mddrc.lut_table4_alternate_upper, CONFIG_SYS_MDDRCGRP_LUT4_AU);
out_be32(&im->mddrc.lut_table4_alternate_lower, CONFIG_SYS_MDDRCGRP_LUT4_AL);
/*
* Initialize MDDRC
* put MDDRC in CMD mode and
* set the max time between refreshes to 0 during init process
*/
out_be32(&im->mddrc.ddr_sys_config,
mddrc_config->ddr_sys_config | MDDRC_SYS_CFG_CMD_MASK);
out_be32(&im->mddrc.ddr_time_config0,
mddrc_config->ddr_time_config0 & MDDRC_REFRESH_ZERO_MASK);
out_be32(&im->mddrc.ddr_time_config1,
mddrc_config->ddr_time_config1);
out_be32(&im->mddrc.ddr_time_config2,
mddrc_config->ddr_time_config2);
/* Initialize DDR with either default or supplied init sequence */
for (i = 0; i < seq_sz; i++)
out_be32(&im->mddrc.ddr_command, dram_init_seq[i]);
/* Start MDDRC */
out_be32(&im->mddrc.ddr_time_config0, mddrc_config->ddr_time_config0);
out_be32(&im->mddrc.ddr_sys_config, mddrc_config->ddr_sys_config);
msize = get_ram_size(CONFIG_SYS_DDR_BASE, CONFIG_SYS_MAX_RAM_SIZE);
/* Fix DDR Local Window for new size */
out_be32(&im->sysconf.ddrlaw.ar, __ilog2(msize) - 1);
sync_law(&im->sysconf.ddrlaw.ar);
return msize;
}
/*
* compute_lowest_common_dimm_parameters()
*
* Determine the worst-case DIMM timing parameters from the set of DIMMs
* whose parameters have been computed into the array pointed to
* by dimm_params.
*/
unsigned int
compute_lowest_common_dimm_parameters(const dimm_params_t *dimm_params,
common_timing_params_t *outpdimm,
unsigned int number_of_dimms)
{
unsigned int i, j;
unsigned int tCKmin_X_ps = 0;
unsigned int tCKmax_ps = 0xFFFFFFFF;
unsigned int tCKmax_max_ps = 0;
unsigned int tRCD_ps = 0;
unsigned int tRP_ps = 0;
unsigned int tRAS_ps = 0;
unsigned int tWR_ps = 0;
unsigned int tWTR_ps = 0;
unsigned int tRFC_ps = 0;
unsigned int tRRD_ps = 0;
unsigned int tRC_ps = 0;
unsigned int refresh_rate_ps = 0;
unsigned int tIS_ps = 0;
unsigned int tIH_ps = 0;
unsigned int tDS_ps = 0;
unsigned int tDH_ps = 0;
unsigned int tRTP_ps = 0;
unsigned int tDQSQ_max_ps = 0;
unsigned int tQHS_ps = 0;
unsigned int temp1, temp2;
unsigned int additive_latency = 0;
#if !defined(CONFIG_FSL_DDR3)
const unsigned int mclk_ps = get_memory_clk_period_ps();
unsigned int lowest_good_caslat;
unsigned int not_ok;
debug("using mclk_ps = %u\n", mclk_ps);
#endif
temp1 = 0;
for (i = 0; i < number_of_dimms; i++) {
/*
* If there are no ranks on this DIMM,
* it probably doesn't exist, so skip it.
*/
if (dimm_params[i].n_ranks == 0) {
temp1++;
continue;
}
if (dimm_params[i].n_ranks == 4 && i != 0) {
printf("Found Quad-rank DIMM in wrong bank, ignored."
" Software may not run as expected.\n");
temp1++;
continue;
}
if (dimm_params[i].n_ranks == 4 && \
CONFIG_CHIP_SELECTS_PER_CTRL/CONFIG_DIMM_SLOTS_PER_CTLR < 4) {
printf("Found Quad-rank DIMM, not able to support.");
temp1++;
continue;
}
/*
* Find minimum tCKmax_ps to find fastest slow speed,
* i.e., this is the slowest the whole system can go.
*/
tCKmax_ps = min(tCKmax_ps, dimm_params[i].tCKmax_ps);
/* Either find maximum value to determine slowest
* speed, delay, time, period, etc */
tCKmin_X_ps = max(tCKmin_X_ps, dimm_params[i].tCKmin_X_ps);
tCKmax_max_ps = max(tCKmax_max_ps, dimm_params[i].tCKmax_ps);
tRCD_ps = max(tRCD_ps, dimm_params[i].tRCD_ps);
tRP_ps = max(tRP_ps, dimm_params[i].tRP_ps);
tRAS_ps = max(tRAS_ps, dimm_params[i].tRAS_ps);
tWR_ps = max(tWR_ps, dimm_params[i].tWR_ps);
tWTR_ps = max(tWTR_ps, dimm_params[i].tWTR_ps);
tRFC_ps = max(tRFC_ps, dimm_params[i].tRFC_ps);
tRRD_ps = max(tRRD_ps, dimm_params[i].tRRD_ps);
tRC_ps = max(tRC_ps, dimm_params[i].tRC_ps);
tIS_ps = max(tIS_ps, dimm_params[i].tIS_ps);
tIH_ps = max(tIH_ps, dimm_params[i].tIH_ps);
tDS_ps = max(tDS_ps, dimm_params[i].tDS_ps);
tDH_ps = max(tDH_ps, dimm_params[i].tDH_ps);
tRTP_ps = max(tRTP_ps, dimm_params[i].tRTP_ps);
tQHS_ps = max(tQHS_ps, dimm_params[i].tQHS_ps);
refresh_rate_ps = max(refresh_rate_ps,
dimm_params[i].refresh_rate_ps);
/*
* Find maximum tDQSQ_max_ps to find slowest.
*
* FIXME: is finding the slowest value the correct
* strategy for this parameter?
*/
tDQSQ_max_ps = max(tDQSQ_max_ps, dimm_params[i].tDQSQ_max_ps);
}
outpdimm->ndimms_present = number_of_dimms - temp1;
if (temp1 == number_of_dimms) {
debug("no dimms this memory controller\n");
return 0;
}
outpdimm->tCKmin_X_ps = tCKmin_X_ps;
outpdimm->tCKmax_ps = tCKmax_ps;
outpdimm->tCKmax_max_ps = tCKmax_max_ps;
outpdimm->tRCD_ps = tRCD_ps;
outpdimm->tRP_ps = tRP_ps;
outpdimm->tRAS_ps = tRAS_ps;
outpdimm->tWR_ps = tWR_ps;
outpdimm->tWTR_ps = tWTR_ps;
outpdimm->tRFC_ps = tRFC_ps;
outpdimm->tRRD_ps = tRRD_ps;
outpdimm->tRC_ps = tRC_ps;
outpdimm->refresh_rate_ps = refresh_rate_ps;
outpdimm->tIS_ps = tIS_ps;
outpdimm->tIH_ps = tIH_ps;
outpdimm->tDS_ps = tDS_ps;
outpdimm->tDH_ps = tDH_ps;
outpdimm->tRTP_ps = tRTP_ps;
outpdimm->tDQSQ_max_ps = tDQSQ_max_ps;
outpdimm->tQHS_ps = tQHS_ps;
/* Determine common burst length for all DIMMs. */
temp1 = 0xff;
for (i = 0; i < number_of_dimms; i++) {
if (dimm_params[i].n_ranks) {
temp1 &= dimm_params[i].burst_lengths_bitmask;
}
}
outpdimm->all_DIMMs_burst_lengths_bitmask = temp1;
/* Determine if all DIMMs registered buffered. */
temp1 = temp2 = 0;
for (i = 0; i < number_of_dimms; i++) {
if (dimm_params[i].n_ranks) {
if (dimm_params[i].registered_dimm) {
temp1 = 1;
printf("Detected RDIMM %s\n",
dimm_params[i].mpart);
} else {
temp2 = 1;
printf("Detected UDIMM %s\n",
dimm_params[i].mpart);
}
}
}
outpdimm->all_DIMMs_registered = 0;
outpdimm->all_DIMMs_unbuffered = 0;
if (temp1 && !temp2) {
outpdimm->all_DIMMs_registered = 1;
} else if (!temp1 && temp2) {
outpdimm->all_DIMMs_unbuffered = 1;
} else {
printf("ERROR: Mix of registered buffered and unbuffered "
"DIMMs detected!\n");
}
temp1 = 0;
if (outpdimm->all_DIMMs_registered)
for (j = 0; j < 16; j++) {
outpdimm->rcw[j] = dimm_params[0].rcw[j];
for (i = 1; i < number_of_dimms; i++)
if (dimm_params[i].rcw[j] != dimm_params[0].rcw[j]) {
temp1 = 1;
break;
}
}
if (temp1 != 0)
printf("ERROR: Mix different RDIMM detected!\n");
#if defined(CONFIG_FSL_DDR3)
if (compute_cas_latency_ddr3(dimm_params, outpdimm, number_of_dimms))
return 1;
#else
/*
* Compute a CAS latency suitable for all DIMMs
*
* Strategy for SPD-defined latencies: compute only
* CAS latency defined by all DIMMs.
*/
/*
* Step 1: find CAS latency common to all DIMMs using bitwise
* operation.
*/
temp1 = 0xFF;
for (i = 0; i < number_of_dimms; i++) {
if (dimm_params[i].n_ranks) {
temp2 = 0;
temp2 |= 1 << dimm_params[i].caslat_X;
temp2 |= 1 << dimm_params[i].caslat_X_minus_1;
temp2 |= 1 << dimm_params[i].caslat_X_minus_2;
/*
* FIXME: If there was no entry for X-2 (X-1) in
* the SPD, then caslat_X_minus_2
* (caslat_X_minus_1) contains either 255 or
* 0xFFFFFFFF because that's what the glorious
* __ilog2 function returns for an input of 0.
* On 32-bit PowerPC, left shift counts with bit
* 26 set (that the value of 255 or 0xFFFFFFFF
* will have), cause the destination register to
* be 0. That is why this works.
*/
temp1 &= temp2;
}
}
/*
* Step 2: check each common CAS latency against tCK of each
* DIMM's SPD.
*/
lowest_good_caslat = 0;
temp2 = 0;
while (temp1) {
not_ok = 0;
temp2 = __ilog2(temp1);
debug("checking common caslat = %u\n", temp2);
/* Check if this CAS latency will work on all DIMMs at tCK. */
for (i = 0; i < number_of_dimms; i++) {
if (!dimm_params[i].n_ranks) {
continue;
}
if (dimm_params[i].caslat_X == temp2) {
if (mclk_ps >= dimm_params[i].tCKmin_X_ps) {
debug("CL = %u ok on DIMM %u at tCK=%u"
" ps with its tCKmin_X_ps of %u\n",
temp2, i, mclk_ps,
dimm_params[i].tCKmin_X_ps);
continue;
} else {
not_ok++;
}
}
if (dimm_params[i].caslat_X_minus_1 == temp2) {
unsigned int tCKmin_X_minus_1_ps
= dimm_params[i].tCKmin_X_minus_1_ps;
if (mclk_ps >= tCKmin_X_minus_1_ps) {
debug("CL = %u ok on DIMM %u at "
"tCK=%u ps with its "
"tCKmin_X_minus_1_ps of %u\n",
temp2, i, mclk_ps,
tCKmin_X_minus_1_ps);
continue;
} else {
not_ok++;
}
}
if (dimm_params[i].caslat_X_minus_2 == temp2) {
unsigned int tCKmin_X_minus_2_ps
= dimm_params[i].tCKmin_X_minus_2_ps;
if (mclk_ps >= tCKmin_X_minus_2_ps) {
debug("CL = %u ok on DIMM %u at "
"tCK=%u ps with its "
"tCKmin_X_minus_2_ps of %u\n",
temp2, i, mclk_ps,
tCKmin_X_minus_2_ps);
continue;
} else {
not_ok++;
}
}
}
if (!not_ok) {
lowest_good_caslat = temp2;
}
temp1 &= ~(1 << temp2);
}
debug("lowest common SPD-defined CAS latency = %u\n",
lowest_good_caslat);
outpdimm->lowest_common_SPD_caslat = lowest_good_caslat;
/*
* Compute a common 'de-rated' CAS latency.
*
* The strategy here is to find the *highest* dereated cas latency
* with the assumption that all of the DIMMs will support a dereated
* CAS latency higher than or equal to their lowest dereated value.
*/
temp1 = 0;
for (i = 0; i < number_of_dimms; i++) {
temp1 = max(temp1, dimm_params[i].caslat_lowest_derated);
}
outpdimm->highest_common_derated_caslat = temp1;
debug("highest common dereated CAS latency = %u\n", temp1);
#endif /* #if defined(CONFIG_FSL_DDR3) */
/* Determine if all DIMMs ECC capable. */
temp1 = 1;
for (i = 0; i < number_of_dimms; i++) {
if (dimm_params[i].n_ranks &&
!(dimm_params[i].edc_config & EDC_ECC)) {
temp1 = 0;
break;
}
}
if (temp1) {
debug("all DIMMs ECC capable\n");
} else {
debug("Warning: not all DIMMs ECC capable, cant enable ECC\n");
}
outpdimm->all_DIMMs_ECC_capable = temp1;
#ifndef CONFIG_FSL_DDR3
/* FIXME: move to somewhere else to validate. */
if (mclk_ps > tCKmax_max_ps) {
printf("Warning: some of the installed DIMMs "
"can not operate this slowly.\n");
return 1;
}
#endif
/*
* Compute additive latency.
*
* For DDR1, additive latency should be 0.
*
* For DDR2, with ODT enabled, use "a value" less than ACTTORW,
* which comes from Trcd, and also note that:
* add_lat + caslat must be >= 4
*
* For DDR3, we use the AL=0
*
* When to use additive latency for DDR2:
*
* I. Because you are using CL=3 and need to do ODT on writes and
* want functionality.
* 1. Are you going to use ODT? (Does your board not have
* additional termination circuitry for DQ, DQS, DQS_,
* DM, RDQS, RDQS_ for x4/x8 configs?)
* 2. If so, is your lowest supported CL going to be 3?
* 3. If so, then you must set AL=1 because
*
* WL >= 3 for ODT on writes
* RL = AL + CL
* WL = RL - 1
* ->
* WL = AL + CL - 1
* AL + CL - 1 >= 3
* AL + CL >= 4
* QED
*
* RL >= 3 for ODT on reads
* RL = AL + CL
*
* Since CL aren't usually less than 2, AL=0 is a minimum,
* so the WL-derived AL should be the -- FIXME?
*
* II. Because you are using auto-precharge globally and want to
* use additive latency (posted CAS) to get more bandwidth.
* 1. Are you going to use auto-precharge mode globally?
*
* Use addtivie latency and compute AL to be 1 cycle less than
* tRCD, i.e. the READ or WRITE command is in the cycle
* immediately following the ACTIVATE command..
*
* III. Because you feel like it or want to do some sort of
* degraded-performance experiment.
* 1. Do you just want to use additive latency because you feel
* like it?
*
* Validation: AL is less than tRCD, and within the other
* read-to-precharge constraints.
*/
additive_latency = 0;
#if defined(CONFIG_FSL_DDR2)
if (lowest_good_caslat < 4) {
additive_latency = picos_to_mclk(tRCD_ps) - lowest_good_caslat;
if (mclk_to_picos(additive_latency) > tRCD_ps) {
additive_latency = picos_to_mclk(tRCD_ps);
debug("setting additive_latency to %u because it was "
" greater than tRCD_ps\n", additive_latency);
}
}
#elif defined(CONFIG_FSL_DDR3)
/*
* The system will not use the global auto-precharge mode.
* However, it uses the page mode, so we set AL=0
*/
additive_latency = 0;
#endif
/*
* Validate additive latency
* FIXME: move to somewhere else to validate
*
* AL <= tRCD(min)
*/
if (mclk_to_picos(additive_latency) > tRCD_ps) {
printf("Error: invalid additive latency exceeds tRCD(min).\n");
return 1;
}
/*
* RL = CL + AL; RL >= 3 for ODT_RD_CFG to be enabled
* WL = RL - 1; WL >= 3 for ODT_WL_CFG to be enabled
* ADD_LAT (the register) must be set to a value less
* than ACTTORW if WL = 1, then AL must be set to 1
* RD_TO_PRE (the register) must be set to a minimum
* tRTP + AL if AL is nonzero
*/
/*
* Additive latency will be applied only if the memctl option to
* use it.
*/
outpdimm->additive_latency = additive_latency;
return 0;
}
