Uwe Kleine-König wrote: > Hello Jason, > > On Thu, Sep 02, 2010 at 03:51:59PM +0800, Jason Wang wrote: > >> There are 3 SPI controllers on i.MX51, one is called CSPI and is >> 100% compatible with the one on i.MX35, the other two are called >> eCSPI and are not compatible with existing controllers on other >> i.MX platforms, here we add support of these three controllers in >> the imx spi driver. >> >> Signed-off-by: Jason Wang <jason77.w...@gmail.com> >> --- >> drivers/spi/spi_imx.c | 135 >> +++++++++++++++++++++++++++++++++++++++++++++++-- >> 1 files changed, 131 insertions(+), 4 deletions(-) >> >> diff --git a/drivers/spi/spi_imx.c b/drivers/spi/spi_imx.c >> index 7972e90..8d9c9da 100644 >> --- a/drivers/spi/spi_imx.c >> +++ b/drivers/spi/spi_imx.c >> @@ -155,6 +155,120 @@ static unsigned int spi_imx_clkdiv_2(unsigned int fin, >> return 7; >> } >> >> +/* MX51 eCSPI post divider */ >> +static unsigned int spi_imx_clkdiv_3(unsigned int fin, >> + unsigned int fspi) >> +{ >> + int i, div = 1; >> + >> + for (i = 0; i < 15; i++) { >> + if (fspi * div >= fin) >> + return i; >> + div <<= 1; >> + } >> + >> + return 15; >> +} >> > This is different in Sascha's patch. Didn't try yet to understand the > difference. > > To change the spi clock frequency, we can set pre-divider and post-divider, here, to be simple, i only change post-divider(set pre-divider to 1), while sascha handle both pre-divider and post-divider. >> + >> +#define MX51_INTREG_TEEN (1 << 0) >> +#define MX51_INTREG_RREN (1 << 3) >> + >> +#define MX51_CSPICTRL_ENABLE (1 << 0) >> +#define MX51_CSPICTRL_XCH (1 << 2) >> + >> +#define MX51_CSPICTRL_BL_SHIFT 20 >> +#define MX51_CSPICTRL_CS_SHIFT 18 >> +#define MX51_CSPICTRL_DR_SHIFT 8 >> +#define MX51_CSPICTRL_MODE_SHIFT 4 >> +#define MX51_CSPICONF_PHA_SHIFT 0 >> +#define MX51_CSPICONF_POL_SHIFT 4 >> +#define MX51_CSPICONF_SSPOL_SHIFT 12 >> +#define MX51_CSPICONF_SSCTL_SHIFT 8 >> + >> +#define MX51_CSPICTRL_CSMASK 0x3 >> +#define MX51_CSPIINT 0x10 >> +#define MX51_CSPICONF 0xC >> +#define MX51_CSPISTATUS 0x18 >> +#define MX51_STATUS_RR (1 << 3) >> + >> +#define MAX_CHIPSELECT_NUM 4 >> + >> +static int get_chipselect(struct spi_imx_data *spi_imx, >> + struct spi_imx_config *config) >> +{ >> + int i; >> + >> + for (i = 0; i < MAX_CHIPSELECT_NUM; i++) { >> + if (config->cs == spi_imx->chipselect[i]) >> + return i; >> + } >> + >> + return -EINVAL; >> +} >> +static void mx51_intctrl(struct spi_imx_data *spi_imx, int enable) >> +{ >> + unsigned int val = 0; >> + >> + if (enable & MXC_INT_TE) >> + val |= MX51_INTREG_TEEN; >> + if (enable & MXC_INT_RR) >> + val |= MX51_INTREG_RREN; >> + >> + writel(val, spi_imx->base + MX51_CSPIINT); >> +} >> + >> +static void mx51_trigger(struct spi_imx_data *spi_imx) >> +{ >> + unsigned int reg; >> + >> + reg = readl(spi_imx->base + MXC_CSPICTRL); >> + reg |= MX51_CSPICTRL_XCH; >> + writel(reg, spi_imx->base + MXC_CSPICTRL); >> +} >> + >> +static int mx51_config(struct spi_imx_data *spi_imx, >> + struct spi_imx_config *config) >> +{ >> + unsigned int config_reg = 0; >> + unsigned int ctrl_reg = MX51_CSPICTRL_ENABLE; >> + int chan; >> + >> + chan = get_chipselect(spi_imx, config); >> + if (chan < 0) >> + return chan; >> > can this happen? Does that have to do with using a GPIO for CS? > If we use 4 GPIOs for 4 SPI client devices, we can assign each to a separate channel. each channel has its own transfer mode(like clk edge...). > >> + ctrl_reg |= (chan & MX51_CSPICTRL_CSMASK) << MX51_CSPICTRL_CS_SHIFT; >> + ctrl_reg |= (1 << (chan & MX51_CSPICTRL_CSMASK)) << >> + MX51_CSPICTRL_MODE_SHIFT; >> + ctrl_reg |= spi_imx_clkdiv_3(spi_imx->spi_clk, config->speed_hz) << >> + MX51_CSPICTRL_DR_SHIFT; >> + >> + ctrl_reg |= (config->bpw - 1) << MX51_CSPICTRL_BL_SHIFT; >> + >> + >> > duplicated empty line > > OK. >> + if (config->mode & SPI_CPHA) >> + config_reg |= (1 << (chan & MX51_CSPICTRL_CSMASK)) << >> + MX51_CSPICONF_PHA_SHIFT; >> + if (config->mode & SPI_CPOL) >> + config_reg |= (1 << (chan & MX51_CSPICTRL_CSMASK)) << >> + MX51_CSPICONF_POL_SHIFT; >> + if (config->mode & SPI_CS_HIGH) >> + config_reg |= (1 << (chan & MX51_CSPICTRL_CSMASK)) << >> + MX51_CSPICONF_SSPOL_SHIFT; >> + >> + config_reg |= (1 << (chan & MX51_CSPICTRL_CSMASK)) << >> + MX51_CSPICONF_SSCTL_SHIFT; >> + >> + writel(ctrl_reg, spi_imx->base + MXC_CSPICTRL); >> + writel(config_reg, spi_imx->base + MX51_CSPICONF); >> + >> + return 0; >> +} >> + >> +static int mx51_rx_available(struct spi_imx_data *spi_imx) >> +{ >> + return readl(spi_imx->base + MX51_CSPISTATUS) & MX51_STATUS_RR; >> +} >> + >> #define MX31_INTREG_TEEN (1 << 0) >> #define MX31_INTREG_RREN (1 << 3) >> >> @@ -209,7 +323,7 @@ static int mx31_config(struct spi_imx_data *spi_imx, >> >> if (cpu_is_mx31()) >> reg |= (config->bpw - 1) << MX31_CSPICTRL_BC_SHIFT; >> - else if (cpu_is_mx25() || cpu_is_mx35()) { >> + else if (cpu_is_mx25() || cpu_is_mx35() || cpu_is_mx51()) { >> reg |= (config->bpw - 1) << MX35_CSPICTRL_BL_SHIFT; >> reg |= MX31_CSPICTRL_SSCTL; >> } >> @@ -223,7 +337,7 @@ static int mx31_config(struct spi_imx_data *spi_imx, >> if (config->cs < 0) { >> if (cpu_is_mx31()) >> reg |= (config->cs + 32) << MX31_CSPICTRL_CS_SHIFT; >> - else if (cpu_is_mx25() || cpu_is_mx35()) >> + else if (cpu_is_mx25() || cpu_is_mx35() || cpu_is_mx51()) >> reg |= (config->cs + 32) << MX35_CSPICTRL_CS_SHIFT; >> } >> >> @@ -567,7 +681,14 @@ static int __devinit spi_imx_probe(struct >> platform_device *pdev) >> goto out_iounmap; >> } >> >> - if (cpu_is_mx25() || cpu_is_mx31() || cpu_is_mx35()) { >> + /* i.MX51 has two eCSPI and one CSPI controllers, eCSPI controllers are >> + * not compatible with existing SPI controllers on other i.MX platforms, >> + * while CSPI controller is 100% compatible with the one on the i.MX35. >> + * We set the platform device id to 2 for this CSPI at i.MX51 board init >> + * level to distinguish it from two eCSPI controllers. >> + */ >> > This comment is missing in Sascha's driver. I like it. > BTW, I'd like to make use of platform ids in this driver. This would > make this ugly "on imx51 id2 is a cspi" distinction unnecessary. > > agree, i like both your and lothar's solution. Either platform ids or flags in platform_data.
>> + if (cpu_is_mx25() || cpu_is_mx31() || cpu_is_mx35() || >> + (cpu_is_mx51() && (pdev->id == 2))) { >> spi_imx->intctrl = mx31_intctrl; >> spi_imx->config = mx31_config; >> spi_imx->trigger = mx31_trigger; >> @@ -582,6 +703,11 @@ static int __devinit spi_imx_probe(struct >> platform_device *pdev) >> spi_imx->config = mx1_config; >> spi_imx->trigger = mx1_trigger; >> spi_imx->rx_available = mx1_rx_available; >> + } else if (cpu_is_mx51()) { >> + spi_imx->intctrl = mx51_intctrl; >> + spi_imx->config = mx51_config; >> + spi_imx->trigger = mx51_trigger; >> + spi_imx->rx_available = mx51_rx_available; >> } else >> BUG(); >> >> @@ -599,7 +725,8 @@ static int __devinit spi_imx_probe(struct >> platform_device *pdev) >> writel(1, spi_imx->base + MXC_RESET); >> >> /* drain receive buffer */ >> - if (cpu_is_mx25() || cpu_is_mx31() || cpu_is_mx35()) >> + if (cpu_is_mx25() || cpu_is_mx31() || cpu_is_mx35() || >> + (cpu_is_mx51() && (pdev->id == 2))) >> while (readl(spi_imx->base + MX3_CSPISTAT) & MX3_CSPISTAT_RR) >> readl(spi_imx->base + MXC_CSPIRXDATA); >> > Best regards > Uwe > > ------------------------------------------------------------------------------ This SF.net Dev2Dev email is sponsored by: Show off your parallel programming skills. Enter the Intel(R) Threading Challenge 2010. http://p.sf.net/sfu/intel-thread-sfd _______________________________________________ spi-devel-general mailing list spi-devel-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/spi-devel-general