[MTD] [NAND] FSL-UPM: add multi chip support
Wolfgang Grandegger [Mon, 30 Mar 2009 10:02:42 +0000 (12:02 +0200)]
This patch adds support for multi-chip NAND devices to the FSL-UPM
driver. This requires support for multiple GPIOs for the RNB pins.
The NAND chips are selected through address lines defined by the
FDT property "fsl,upm-addr-line-cs-offsets".

Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Acked-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

arch/powerpc/sysdev/fsl_lbc.c
drivers/mtd/nand/fsl_upm.c

index 0494ee5..dceb8d1 100644 (file)
@@ -150,7 +150,7 @@ int fsl_upm_run_pattern(struct fsl_upm *upm, void __iomem *io_base, u32 mar)
 
        spin_lock_irqsave(&fsl_lbc_lock, flags);
 
-       out_be32(&fsl_lbc_regs->mar, mar << (32 - upm->width));
+       out_be32(&fsl_lbc_regs->mar, mar);
 
        switch (upm->width) {
        case 8:
index 7815a40..430de6d 100644 (file)
@@ -36,7 +36,10 @@ struct fsl_upm_nand {
        uint8_t upm_addr_offset;
        uint8_t upm_cmd_offset;
        void __iomem *io_base;
-       int rnb_gpio;
+       int rnb_gpio[NAND_MAX_CHIPS];
+       uint32_t mchip_offsets[NAND_MAX_CHIPS];
+       uint32_t mchip_count;
+       uint32_t mchip_number;
        int chip_delay;
 };
 
@@ -46,7 +49,7 @@ static int fun_chip_ready(struct mtd_info *mtd)
 {
        struct fsl_upm_nand *fun = to_fsl_upm_nand(mtd);
 
-       if (gpio_get_value(fun->rnb_gpio))
+       if (gpio_get_value(fun->rnb_gpio[fun->mchip_number]))
                return 1;
 
        dev_vdbg(fun->dev, "busy\n");
@@ -55,9 +58,9 @@ static int fun_chip_ready(struct mtd_info *mtd)
 
 static void fun_wait_rnb(struct fsl_upm_nand *fun)
 {
-       int cnt = 1000000;
+       if (fun->rnb_gpio[fun->mchip_number] >= 0) {
+               int cnt = 1000000;
 
-       if (fun->rnb_gpio >= 0) {
                while (--cnt && !fun_chip_ready(&fun->mtd))
                        cpu_relax();
                if (!cnt)
@@ -69,7 +72,9 @@ static void fun_wait_rnb(struct fsl_upm_nand *fun)
 
 static void fun_cmd_ctrl(struct mtd_info *mtd, int cmd, unsigned int ctrl)
 {
+       struct nand_chip *chip = mtd->priv;
        struct fsl_upm_nand *fun = to_fsl_upm_nand(mtd);
+       u32 mar;
 
        if (!(ctrl & fun->last_ctrl)) {
                fsl_upm_end_pattern(&fun->upm);
@@ -87,11 +92,29 @@ static void fun_cmd_ctrl(struct mtd_info *mtd, int cmd, unsigned int ctrl)
                        fsl_upm_start_pattern(&fun->upm, fun->upm_cmd_offset);
        }
 
-       fsl_upm_run_pattern(&fun->upm, fun->io_base, cmd);
+       mar = (cmd << (32 - fun->upm.width)) |
+               fun->mchip_offsets[fun->mchip_number];
+       fsl_upm_run_pattern(&fun->upm, chip->IO_ADDR_R, mar);
 
        fun_wait_rnb(fun);
 }
 
+static void fun_select_chip(struct mtd_info *mtd, int mchip_nr)
+{
+       struct nand_chip *chip = mtd->priv;
+       struct fsl_upm_nand *fun = to_fsl_upm_nand(mtd);
+
+       if (mchip_nr == -1) {
+               chip->cmd_ctrl(mtd, NAND_CMD_NONE, 0 | NAND_CTRL_CHANGE);
+       } else if (mchip_nr >= 0) {
+               fun->mchip_number = mchip_nr;
+               chip->IO_ADDR_R = fun->io_base + fun->mchip_offsets[mchip_nr];
+               chip->IO_ADDR_W = chip->IO_ADDR_R;
+       } else {
+               BUG();
+       }
+}
+
 static uint8_t fun_read_byte(struct mtd_info *mtd)
 {
        struct fsl_upm_nand *fun = to_fsl_upm_nand(mtd);
@@ -137,8 +160,10 @@ static int __devinit fun_chip_init(struct fsl_upm_nand *fun,
        fun->chip.read_buf = fun_read_buf;
        fun->chip.write_buf = fun_write_buf;
        fun->chip.ecc.mode = NAND_ECC_SOFT;
+       if (fun->mchip_count > 1)
+               fun->chip.select_chip = fun_select_chip;
 
-       if (fun->rnb_gpio >= 0)
+       if (fun->rnb_gpio[0] >= 0)
                fun->chip.dev_ready = fun_chip_ready;
 
        fun->mtd.priv = &fun->chip;
@@ -155,7 +180,7 @@ static int __devinit fun_chip_init(struct fsl_upm_nand *fun,
                goto err;
        }
 
-       ret = nand_scan(&fun->mtd, 1);
+       ret = nand_scan(&fun->mtd, fun->mchip_count);
        if (ret)
                goto err;
 
@@ -185,8 +210,10 @@ static int __devinit fun_probe(struct of_device *ofdev,
        struct fsl_upm_nand *fun;
        struct resource io_res;
        const uint32_t *prop;
+       int rnb_gpio;
        int ret;
        int size;
+       int i;
 
        fun = kzalloc(sizeof(*fun), GFP_KERNEL);
        if (!fun)
@@ -208,7 +235,7 @@ static int __devinit fun_probe(struct of_device *ofdev,
        if (!prop || size != sizeof(uint32_t)) {
                dev_err(&ofdev->dev, "can't get UPM address offset\n");
                ret = -EINVAL;
-               goto err2;
+               goto err1;
        }
        fun->upm_addr_offset = *prop;
 
@@ -216,21 +243,40 @@ static int __devinit fun_probe(struct of_device *ofdev,
        if (!prop || size != sizeof(uint32_t)) {
                dev_err(&ofdev->dev, "can't get UPM command offset\n");
                ret = -EINVAL;
-               goto err2;
+               goto err1;
        }
        fun->upm_cmd_offset = *prop;
 
-       fun->rnb_gpio = of_get_gpio(ofdev->node, 0);
-       if (fun->rnb_gpio >= 0) {
-               ret = gpio_request(fun->rnb_gpio, dev_name(&ofdev->dev));
-               if (ret) {
-                       dev_err(&ofdev->dev, "can't request RNB gpio\n");
+       prop = of_get_property(ofdev->node,
+                              "fsl,upm-addr-line-cs-offsets", &size);
+       if (prop && (size / sizeof(uint32_t)) > 0) {
+               fun->mchip_count = size / sizeof(uint32_t);
+               if (fun->mchip_count >= NAND_MAX_CHIPS) {
+                       dev_err(&ofdev->dev, "too much multiple chips\n");
+                       goto err1;
+               }
+               for (i = 0; i < fun->mchip_count; i++)
+                       fun->mchip_offsets[i] = prop[i];
+       } else {
+               fun->mchip_count = 1;
+       }
+
+       for (i = 0; i < fun->mchip_count; i++) {
+               fun->rnb_gpio[i] = -1;
+               rnb_gpio = of_get_gpio(ofdev->node, i);
+               if (rnb_gpio >= 0) {
+                       ret = gpio_request(rnb_gpio, dev_name(&ofdev->dev));
+                       if (ret) {
+                               dev_err(&ofdev->dev,
+                                       "can't request RNB gpio #%d\n", i);
+                               goto err2;
+                       }
+                       gpio_direction_input(rnb_gpio);
+                       fun->rnb_gpio[i] = rnb_gpio;
+               } else if (rnb_gpio == -EINVAL) {
+                       dev_err(&ofdev->dev, "RNB gpio #%d is invalid\n", i);
                        goto err2;
                }
-               gpio_direction_input(fun->rnb_gpio);
-       } else if (fun->rnb_gpio == -EINVAL) {
-               dev_err(&ofdev->dev, "specified RNB gpio is invalid\n");
-               goto err2;
        }
 
        prop = of_get_property(ofdev->node, "chip-delay", NULL);
@@ -240,7 +286,7 @@ static int __devinit fun_probe(struct of_device *ofdev,
                fun->chip_delay = 50;
 
        fun->io_base = devm_ioremap_nocache(&ofdev->dev, io_res.start,
-                                         io_res.end - io_res.start + 1);
+                                           io_res.end - io_res.start + 1);
        if (!fun->io_base) {
                ret = -ENOMEM;
                goto err2;
@@ -257,8 +303,11 @@ static int __devinit fun_probe(struct of_device *ofdev,
 
        return 0;
 err2:
-       if (fun->rnb_gpio >= 0)
-               gpio_free(fun->rnb_gpio);
+       for (i = 0; i < fun->mchip_count; i++) {
+               if (fun->rnb_gpio[i] < 0)
+                       break;
+               gpio_free(fun->rnb_gpio[i]);
+       }
 err1:
        kfree(fun);
 
@@ -268,12 +317,16 @@ err1:
 static int __devexit fun_remove(struct of_device *ofdev)
 {
        struct fsl_upm_nand *fun = dev_get_drvdata(&ofdev->dev);
+       int i;
 
        nand_release(&fun->mtd);
        kfree(fun->mtd.name);
 
-       if (fun->rnb_gpio >= 0)
-               gpio_free(fun->rnb_gpio);
+       for (i = 0; i < fun->mchip_count; i++) {
+               if (fun->rnb_gpio[i] < 0)
+                       break;
+               gpio_free(fun->rnb_gpio[i]);
+       }
 
        kfree(fun);