[PATCH] MTD: pxa3xx_nand: enable multiple chip select support

Lei Wen adrian.wenl at gmail.com
Thu Jun 23 02:35:31 EDT 2011


Hi Igor,

On Wed, Jun 22, 2011 at 9:45 PM, Igor Grinberg <grinberg at compulab.co.il> wrote:
> Hi Lei,
>
>
> Some comments from a quick look...
>
>
> On 06/22/11 06:17, Lei Wen wrote:
>
>> Current pxa3xx_nand controller has two chip select which
>> both be workable. This patch enable this feature.
>>
>> Update platform driver to support this feature.
>>
>> Another notice should be taken that:
>> When you want to use this feature, you should not enable the
>> keep configuration feature, for two chip select could be
>> attached with different nand chip. The different page size
>> and timing requirement make the keep configuration impossible.
>
> You should _also_ put this comment inside the pxa3xx_nand.h
> may be even inside the pxa3xx_nand_platform_data structure,
> so people would not have to search the git log to find this problem.

How about just throw out a BUG() to force people cannot set keep_config
when supported cs num more than 1? Also appended with the comments above?

>
>> Signed-off-by: Lei Wen <leiwen at marvell.com>
>> ---
>>  arch/arm/mach-mmp/aspenite.c                 |    5 +-
>>  arch/arm/mach-pxa/cm-x300.c                  |    5 +-
>>  arch/arm/mach-pxa/colibri-pxa3xx.c           |    5 +-
>>  arch/arm/mach-pxa/littleton.c                |    5 +-
>>  arch/arm/mach-pxa/mxm8x10.c                  |    9 +-
>>  arch/arm/mach-pxa/raumfeld.c                 |    5 +-
>>  arch/arm/mach-pxa/zylonite.c                 |    5 +-
>>  arch/arm/plat-pxa/include/plat/pxa3xx_nand.h |    8 +-
>>  drivers/mtd/nand/pxa3xx_nand.c               |  735 +++++++++++++++-----------
>>  9 files changed, 444 insertions(+), 338 deletions(-)
>>
>> diff --git a/arch/arm/mach-mmp/aspenite.c b/arch/arm/mach-mmp/aspenite.c
>> index 06b5fa8..b6589d6 100644
>> --- a/arch/arm/mach-mmp/aspenite.c
>> +++ b/arch/arm/mach-mmp/aspenite.c
>> @@ -167,8 +167,9 @@ static struct mtd_partition aspenite_nand_partitions[] = {
>>
>>  static struct pxa3xx_nand_platform_data aspenite_nand_info = {
>>       .enable_arbiter = 1,
>> -     .parts          = aspenite_nand_partitions,
>> -     .nr_parts       = ARRAY_SIZE(aspenite_nand_partitions),
>> +     .cs_num = 1,
>> +     .parts[0]       = aspenite_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(aspenite_nand_partitions),
>>  };
>>
>>  static struct i2c_board_info aspenite_i2c_info[] __initdata = {
>> diff --git a/arch/arm/mach-pxa/cm-x300.c b/arch/arm/mach-pxa/cm-x300.c
>> index b2248e7..d67eb7b 100644
>> --- a/arch/arm/mach-pxa/cm-x300.c
>> +++ b/arch/arm/mach-pxa/cm-x300.c
>> @@ -423,8 +423,9 @@ static struct mtd_partition cm_x300_nand_partitions[] = {
>>  static struct pxa3xx_nand_platform_data cm_x300_nand_info = {
>>       .enable_arbiter = 1,
>>       .keep_config    = 1,
>> -     .parts          = cm_x300_nand_partitions,
>> -     .nr_parts       = ARRAY_SIZE(cm_x300_nand_partitions),
>> +     .cs_num         = 1,
>> +     .parts[0]       = cm_x300_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(cm_x300_nand_partitions),
>>  };
>>
>>  static void __init cm_x300_init_nand(void)
>> diff --git a/arch/arm/mach-pxa/colibri-pxa3xx.c b/arch/arm/mach-pxa/colibri-pxa3xx.c
>> index 3f9be41..ff7a07b 100644
>> --- a/arch/arm/mach-pxa/colibri-pxa3xx.c
>> +++ b/arch/arm/mach-pxa/colibri-pxa3xx.c
>> @@ -139,8 +139,9 @@ static struct mtd_partition colibri_nand_partitions[] = {
>>  static struct pxa3xx_nand_platform_data colibri_nand_info = {
>>       .enable_arbiter = 1,
>>       .keep_config    = 1,
>> -     .parts          = colibri_nand_partitions,
>> -     .nr_parts       = ARRAY_SIZE(colibri_nand_partitions),
>> +     .cs_num         = 1,
>> +     .parts[0]       = colibri_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(colibri_nand_partitions),
>>  };
>>
>>  void __init colibri_pxa3xx_init_nand(void)
>> diff --git a/arch/arm/mach-pxa/littleton.c b/arch/arm/mach-pxa/littleton.c
>> index e5e326d..6eaf852 100644
>> --- a/arch/arm/mach-pxa/littleton.c
>> +++ b/arch/arm/mach-pxa/littleton.c
>> @@ -325,8 +325,9 @@ static struct mtd_partition littleton_nand_partitions[] = {
>>
>>  static struct pxa3xx_nand_platform_data littleton_nand_info = {
>>       .enable_arbiter = 1,
>> -     .parts          = littleton_nand_partitions,
>> -     .nr_parts       = ARRAY_SIZE(littleton_nand_partitions),
>> +     .cs_num         = 1,
>> +     .parts[0]       = littleton_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(littleton_nand_partitions),
>>  };
>>
>>  static void __init littleton_init_nand(void)
>> diff --git a/arch/arm/mach-pxa/mxm8x10.c b/arch/arm/mach-pxa/mxm8x10.c
>> index b5a8fd3..e7ce135 100644
>> --- a/arch/arm/mach-pxa/mxm8x10.c
>> +++ b/arch/arm/mach-pxa/mxm8x10.c
>> @@ -389,10 +389,11 @@ static struct mtd_partition mxm_8x10_nand_partitions[] = {
>>  };
>>
>>  static struct pxa3xx_nand_platform_data mxm_8x10_nand_info = {
>> -     .enable_arbiter = 1,
>> -     .keep_config = 1,
>> -     .parts = mxm_8x10_nand_partitions,
>> -     .nr_parts = ARRAY_SIZE(mxm_8x10_nand_partitions)
>> +     .enable_arbiter = 1,
>> +     .keep_config    = 1,
>> +     .cs_num         = 1,
>> +     .parts[0]       = mxm_8x10_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(mxm_8x10_nand_partitions)
>>  };
>>
>>  static void __init mxm_8x10_nand_init(void)
>> diff --git a/arch/arm/mach-pxa/raumfeld.c b/arch/arm/mach-pxa/raumfeld.c
>> index d130f77..a54846d 100644
>> --- a/arch/arm/mach-pxa/raumfeld.c
>> +++ b/arch/arm/mach-pxa/raumfeld.c
>> @@ -349,8 +349,9 @@ static struct mtd_partition raumfeld_nand_partitions[] = {
>>  static struct pxa3xx_nand_platform_data raumfeld_nand_info = {
>>       .enable_arbiter = 1,
>>       .keep_config    = 1,
>> -     .parts          = raumfeld_nand_partitions,
>> -     .nr_parts       = ARRAY_SIZE(raumfeld_nand_partitions),
>> +     .cs_num         = 1,
>> +     .parts[0]       = raumfeld_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(raumfeld_nand_partitions),
>>  };
>>
>>  /**
>> diff --git a/arch/arm/mach-pxa/zylonite.c b/arch/arm/mach-pxa/zylonite.c
>> index 5821185..ea4752a 100644
>> --- a/arch/arm/mach-pxa/zylonite.c
>> +++ b/arch/arm/mach-pxa/zylonite.c
>> @@ -366,8 +366,9 @@ static struct mtd_partition zylonite_nand_partitions[] = {
>>
>>  static struct pxa3xx_nand_platform_data zylonite_nand_info = {
>>       .enable_arbiter = 1,
>> -     .parts          = zylonite_nand_partitions,
>> -     .nr_parts       = ARRAY_SIZE(zylonite_nand_partitions),
>> +     .cs_num         = 1,
>> +     .parts[0]       = zylonite_nand_partitions,
>> +     .nr_parts[0]    = ARRAY_SIZE(zylonite_nand_partitions),
>>  };
>>
>>  static void __init zylonite_init_nand(void)
>> diff --git a/arch/arm/plat-pxa/include/plat/pxa3xx_nand.h b/arch/arm/plat-pxa/include/plat/pxa3xx_nand.h
>> index 442301f..34a3f52 100644
>> --- a/arch/arm/plat-pxa/include/plat/pxa3xx_nand.h
>> +++ b/arch/arm/plat-pxa/include/plat/pxa3xx_nand.h
>> @@ -41,6 +41,8 @@ struct pxa3xx_nand_flash {
>>       struct pxa3xx_nand_timing *timing;      /* NAND Flash timing */
>>  };
>>
>> +/* The max num of chip select current support */
>
> /* The maximum number of chip selects currently supported */
>
>> +#define NUM_CHIP_SELECT              (2)
>>  struct pxa3xx_nand_platform_data {
>>
>>       /* the data flash bus is shared between the Static Memory
>> @@ -52,8 +54,10 @@ struct pxa3xx_nand_platform_data {
>>       /* allow platform code to keep OBM/bootloader defined NFC config */
>>       int     keep_config;
>>
>> -     const struct mtd_partition              *parts;
>> -     unsigned int                            nr_parts;
>> +     /* indicate how many chip select would be used for this platform */
>
> /* indicate how many chip selects will be used */
>
>> +     int     cs_num;
>
> This name is too confusing, I think even num_cs is better or cs_count?
> Also, may be align it with the others?

 I prefer the num_cs.

>
>> +     const struct mtd_partition              *parts[NUM_CHIP_SELECT];
>> +     unsigned int                            nr_parts[NUM_CHIP_SELECT];
>>
>>       const struct pxa3xx_nand_flash *        flash;
>>       size_t                                  num_flash;
>> diff --git a/drivers/mtd/nand/pxa3xx_nand.c b/drivers/mtd/nand/pxa3xx_nand.c
>> index 30689cc..259b8d5 100644
>> --- a/drivers/mtd/nand/pxa3xx_nand.c
>> +++ b/drivers/mtd/nand/pxa3xx_nand.c
>> @@ -92,11 +92,13 @@
>>  #define NDCB0_ADDR_CYC_SHIFT (16)
>>
>>  /* macros for registers read/write */
>> -#define nand_writel(info, off, val)  \
>> -     __raw_writel((val), (info)->mmio_base + (off))
>> +#define nand_writel(nand, off, val)  \
>> +     __raw_writel((val), (nand)->mmio_base + (off))
>>
>> -#define nand_readl(info, off)                \
>> -     __raw_readl((info)->mmio_base + (off))
>> +#define nand_readl(nand, off)                \
>> +     __raw_readl((nand)->mmio_base + (off))
>> +#define get_mtd_by_info(info)                \
>> +     (struct mtd_info *)((void *)info - sizeof(struct mtd_info))
>>
>>  /* error code and state */
>>  enum {
>> @@ -110,6 +112,7 @@ enum {
>>
>>  enum {
>>       STATE_IDLE = 0,
>> +     STATE_PREPARED,
>>       STATE_CMD_HANDLE,
>>       STATE_DMA_READING,
>>       STATE_DMA_WRITING,
>> @@ -123,63 +126,63 @@ enum {
>>  struct pxa3xx_nand_info {
>>       struct nand_chip        nand_chip;
>>
>> -     struct nand_hw_control  controller;
>> -     struct platform_device   *pdev;
>>       struct pxa3xx_nand_cmdset *cmdset;
>> +     /* page size of attached chip */
>> +     uint16_t                page_size;
>> +     uint8_t                 chip_select;
>> +     uint8_t                 use_ecc;
>> +
>> +     /* calculated from pxa3xx_nand_flash data */
>> +     uint8_t                 col_addr_cycles;
>> +     uint8_t                 row_addr_cycles;
>> +     uint8_t                 read_id_bytes;
>> +
>> +     /* cached register value */
>> +     uint32_t                reg_ndcr;
>> +     uint32_t                ndtr0cs0;
>> +     uint32_t                ndtr1cs0;
>>
>> +     void                    *nand_data;
>> +};
>> +
>> +struct pxa3xx_nand {
>>       struct clk              *clk;
>>       void __iomem            *mmio_base;
>>       unsigned long           mmio_phys;
>> +     struct nand_hw_control  controller;
>> +     struct completion       cmd_complete;
>> +     struct platform_device   *pdev;
>
> please, align

Do you still mean cs_num? What this comment refer to?

>
>>
>> -     unsigned int            buf_start;
>> -     unsigned int            buf_count;
>> -
>> -     struct mtd_info         *mtd;
>>       /* DMA information */
>>       int                     drcmr_dat;
>>       int                     drcmr_cmd;
>> -
>> -     unsigned char           *data_buff;
>> -     unsigned char           *oob_buff;
>> -     dma_addr_t              data_buff_phys;
>> -     size_t                  data_buff_size;
>>       int                     data_dma_ch;
>> -     struct pxa_dma_desc     *data_desc;
>> +     dma_addr_t              data_buff_phys;
>>       dma_addr_t              data_desc_addr;
>> +     struct pxa_dma_desc     *data_desc;
>>
>> -     uint32_t                reg_ndcr;
>> -
>> -     /* saved column/page_addr during CMD_SEQIN */
>> -     int                     seqin_column;
>> -     int                     seqin_page_addr;
>> +     struct pxa3xx_nand_info *info[NUM_CHIP_SELECT];
>> +     uint32_t                command;
>> +     uint16_t                data_size;      /* data size in FIFO */
>> +     uint16_t                oob_size;
>> +     unsigned char           *data_buff;
>> +     unsigned char           *oob_buff;
>> +     uint32_t                buf_start;
>> +     uint32_t                buf_count;
>>
>>       /* relate to the command */
>>       unsigned int            state;
>> -
>> +     uint8_t                 chip_select;
>>       int                     use_ecc;        /* use HW ECC ? */
>>       int                     use_dma;        /* use DMA ? */
>>       int                     is_ready;
>> -
>> -     unsigned int            page_size;      /* page size of attached chip */
>> -     unsigned int            data_size;      /* data size in FIFO */
>>       int                     retcode;
>> -     struct completion       cmd_complete;
>>
>>       /* generated NDCBx register values */
>> +     uint8_t                 total_cmds;
>>       uint32_t                ndcb0;
>>       uint32_t                ndcb1;
>>       uint32_t                ndcb2;
>> -
>> -     /* timing calcuted from setting */
>> -     uint32_t                ndtr0cs0;
>> -     uint32_t                ndtr1cs0;
>> -
>> -     /* calculated from pxa3xx_nand_flash data */
>> -     size_t          oob_size;
>> -     size_t          read_id_bytes;
>> -
>> -     unsigned int    col_addr_cycles;
>> -     unsigned int    row_addr_cycles;
>>  };
>
> It looks like if you switch the names of the structures above,
> then the patch will be much shorter and cleaner,
> but will it make structures meaning confusion?

Could you elaborate more about this? Do you means no need to create a seperated
structure?

>
>>
>>  static int use_dma = 1;
>> @@ -225,7 +228,7 @@ static struct pxa3xx_nand_flash builtin_flash_types[] = {
>>  /* Define a default flash type setting serve as flash detecting only */
>>  #define DEFAULT_FLASH_TYPE (&builtin_flash_types[0])
>>
>> -const char *mtd_names[] = {"pxa3xx_nand-0", NULL};
>> +const char *mtd_names[] = {"pxa3xx_nand-0", "pxa3xx_nand-1", NULL};
>>
>>  #define NDTR0_tCH(c) (min((c), 7) << 19)
>>  #define NDTR0_tCS(c) (min((c), 7) << 16)
>> @@ -244,9 +247,11 @@ const char *mtd_names[] = {"pxa3xx_nand-0", NULL};
>>  static void pxa3xx_nand_set_timing(struct pxa3xx_nand_info *info,
>>                                  const struct pxa3xx_nand_timing *t)
>>  {
>> -     unsigned long nand_clk = clk_get_rate(info->clk);
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     unsigned long nand_clk;
>>       uint32_t ndtr0, ndtr1;
>>
>> +     nand_clk = clk_get_rate(nand->clk);
>>       ndtr0 = NDTR0_tCH(ns2cycle(t->tCH, nand_clk)) |
>>               NDTR0_tCS(ns2cycle(t->tCS, nand_clk)) |
>>               NDTR0_tWH(ns2cycle(t->tWH, nand_clk)) |
>> @@ -260,26 +265,27 @@ static void pxa3xx_nand_set_timing(struct pxa3xx_nand_info *info,
>>
>>       info->ndtr0cs0 = ndtr0;
>>       info->ndtr1cs0 = ndtr1;
>> -     nand_writel(info, NDTR0CS0, ndtr0);
>> -     nand_writel(info, NDTR1CS0, ndtr1);
>> +     nand_writel(nand, NDTR0CS0, ndtr0);
>> +     nand_writel(nand, NDTR1CS0, ndtr1);
>>  }
>>
>>  static void pxa3xx_set_datasize(struct pxa3xx_nand_info *info)
>>  {
>> +     struct pxa3xx_nand *nand = info->nand_data;
>>       int oob_enable = info->reg_ndcr & NDCR_SPARE_EN;
>>
>> -     info->data_size = info->page_size;
>> +     nand->data_size = info->page_size;
>>       if (!oob_enable) {
>> -             info->oob_size = 0;
>> +             nand->oob_size = 0;
>>               return;
>>       }
>>
>>       switch (info->page_size) {
>>       case 2048:
>> -             info->oob_size = (info->use_ecc) ? 40 : 64;
>> +             nand->oob_size = (info->use_ecc) ? 40 : 64;
>>               break;
>>       case 512:
>> -             info->oob_size = (info->use_ecc) ? 8 : 16;
>> +             nand->oob_size = (info->use_ecc) ? 8 : 16;
>>               break;
>>       }
>>  }
>> @@ -290,185 +296,189 @@ static void pxa3xx_set_datasize(struct pxa3xx_nand_info *info)
>>   * We enable all the interrupt at the same time, and
>>   * let pxa3xx_nand_irq to handle all logic.
>>   */
>> -static void pxa3xx_nand_start(struct pxa3xx_nand_info *info)
>> +static void pxa3xx_nand_start(struct pxa3xx_nand *nand)
>>  {
>> +     struct pxa3xx_nand_info *info;
>>       uint32_t ndcr;
>>
>> +     info = nand->info[nand->chip_select];
>>       ndcr = info->reg_ndcr;
>> -     ndcr |= info->use_ecc ? NDCR_ECC_EN : 0;
>> -     ndcr |= info->use_dma ? NDCR_DMA_EN : 0;
>> +     ndcr |= nand->use_ecc ? NDCR_ECC_EN : 0;
>> +     ndcr |= nand->use_dma ? NDCR_DMA_EN : 0;
>>       ndcr |= NDCR_ND_RUN;
>>
>>       /* clear status bits and run */
>> -     nand_writel(info, NDCR, 0);
>> -     nand_writel(info, NDSR, NDSR_MASK);
>> -     nand_writel(info, NDCR, ndcr);
>> +     nand_writel(nand, NDCR, 0);
>> +     nand_writel(nand, NDSR, NDSR_MASK);
>> +     nand_writel(nand, NDCR, ndcr);
>>  }
>>
>> -static void pxa3xx_nand_stop(struct pxa3xx_nand_info *info)
>> +static void pxa3xx_nand_stop(struct pxa3xx_nand *nand)
>>  {
>>       uint32_t ndcr;
>>       int timeout = NAND_STOP_DELAY;
>>
>>       /* wait RUN bit in NDCR become 0 */
>> -     ndcr = nand_readl(info, NDCR);
>> +     ndcr = nand_readl(nand, NDCR);
>>       while ((ndcr & NDCR_ND_RUN) && (timeout-- > 0)) {
>> -             ndcr = nand_readl(info, NDCR);
>> +             ndcr = nand_readl(nand, NDCR);
>>               udelay(1);
>>       }
>>
>>       if (timeout <= 0) {
>>               ndcr &= ~NDCR_ND_RUN;
>> -             nand_writel(info, NDCR, ndcr);
>> +             nand_writel(nand, NDCR, ndcr);
>>       }
>>       /* clear status bits */
>> -     nand_writel(info, NDSR, NDSR_MASK);
>> +     nand_writel(nand, NDSR, NDSR_MASK);
>>  }
>>
>> -static void enable_int(struct pxa3xx_nand_info *info, uint32_t int_mask)
>> +static void enable_int(struct pxa3xx_nand *nand, uint32_t int_mask)
>>  {
>>       uint32_t ndcr;
>>
>> -     ndcr = nand_readl(info, NDCR);
>> -     nand_writel(info, NDCR, ndcr & ~int_mask);
>> +     ndcr = nand_readl(nand, NDCR);
>> +     nand_writel(nand, NDCR, ndcr & ~int_mask);
>>  }
>>
>> -static void disable_int(struct pxa3xx_nand_info *info, uint32_t int_mask)
>> +static void disable_int(struct pxa3xx_nand *nand, uint32_t int_mask)
>>  {
>>       uint32_t ndcr;
>>
>> -     ndcr = nand_readl(info, NDCR);
>> -     nand_writel(info, NDCR, ndcr | int_mask);
>> +     ndcr = nand_readl(nand, NDCR);
>> +     nand_writel(nand, NDCR, ndcr | int_mask);
>>  }
>>
>> -static void handle_data_pio(struct pxa3xx_nand_info *info)
>> +static void handle_data_pio(struct pxa3xx_nand *nand)
>>  {
>> -     switch (info->state) {
>> +     switch (nand->state) {
>>       case STATE_PIO_WRITING:
>> -             __raw_writesl(info->mmio_base + NDDB, info->data_buff,
>> -                             DIV_ROUND_UP(info->data_size, 4));
>> -             if (info->oob_size > 0)
>> -                     __raw_writesl(info->mmio_base + NDDB, info->oob_buff,
>> -                                     DIV_ROUND_UP(info->oob_size, 4));
>> +             __raw_writesl(nand->mmio_base + NDDB, nand->data_buff,
>> +                             DIV_ROUND_UP(nand->data_size, 4));
>> +             if (nand->oob_size > 0)
>> +                     __raw_writesl(nand->mmio_base + NDDB, nand->oob_buff,
>> +                                     DIV_ROUND_UP(nand->oob_size, 4));
>>               break;
>>       case STATE_PIO_READING:
>> -             __raw_readsl(info->mmio_base + NDDB, info->data_buff,
>> -                             DIV_ROUND_UP(info->data_size, 4));
>> -             if (info->oob_size > 0)
>> -                     __raw_readsl(info->mmio_base + NDDB, info->oob_buff,
>> -                                     DIV_ROUND_UP(info->oob_size, 4));
>> +             __raw_readsl(nand->mmio_base + NDDB, nand->data_buff,
>> +                             DIV_ROUND_UP(nand->data_size, 4));
>> +             if (nand->oob_size > 0)
>> +                     __raw_readsl(nand->mmio_base + NDDB, nand->oob_buff,
>> +                                     DIV_ROUND_UP(nand->oob_size, 4));
>>               break;
>>       default:
>>               printk(KERN_ERR "%s: invalid state %d\n", __func__,
>> -                             info->state);
>> +                             nand->state);
>>               BUG();
>>       }
>>  }
>>
>> -static void start_data_dma(struct pxa3xx_nand_info *info)
>> +static void start_data_dma(struct pxa3xx_nand *nand)
>>  {
>> -     struct pxa_dma_desc *desc = info->data_desc;
>> -     int dma_len = ALIGN(info->data_size + info->oob_size, 32);
>> +     struct pxa_dma_desc *desc = nand->data_desc;
>> +     int dma_len = ALIGN(nand->data_size + nand->oob_size, 32);
>>
>>       desc->ddadr = DDADR_STOP;
>>       desc->dcmd = DCMD_ENDIRQEN | DCMD_WIDTH4 | DCMD_BURST32 | dma_len;
>>
>> -     switch (info->state) {
>> +     switch (nand->state) {
>>       case STATE_DMA_WRITING:
>> -             desc->dsadr = info->data_buff_phys;
>> -             desc->dtadr = info->mmio_phys + NDDB;
>> +             desc->dsadr = nand->data_buff_phys;
>> +             desc->dtadr = nand->mmio_phys + NDDB;
>>               desc->dcmd |= DCMD_INCSRCADDR | DCMD_FLOWTRG;
>>               break;
>>       case STATE_DMA_READING:
>> -             desc->dtadr = info->data_buff_phys;
>> -             desc->dsadr = info->mmio_phys + NDDB;
>> +             desc->dtadr = nand->data_buff_phys;
>> +             desc->dsadr = nand->mmio_phys + NDDB;
>>               desc->dcmd |= DCMD_INCTRGADDR | DCMD_FLOWSRC;
>>               break;
>>       default:
>>               printk(KERN_ERR "%s: invalid state %d\n", __func__,
>> -                             info->state);
>> +                             nand->state);
>>               BUG();
>>       }
>>
>> -     DRCMR(info->drcmr_dat) = DRCMR_MAPVLD | info->data_dma_ch;
>> -     DDADR(info->data_dma_ch) = info->data_desc_addr;
>> -     DCSR(info->data_dma_ch) |= DCSR_RUN;
>> +     DRCMR(nand->drcmr_dat) = DRCMR_MAPVLD | nand->data_dma_ch;
>> +     DDADR(nand->data_dma_ch) = nand->data_desc_addr;
>> +     DCSR(nand->data_dma_ch) |= DCSR_RUN;
>>  }
>>
>>  static void pxa3xx_nand_data_dma_irq(int channel, void *data)
>>  {
>> -     struct pxa3xx_nand_info *info = data;
>> +     struct pxa3xx_nand *nand = data;
>>       uint32_t dcsr;
>>
>>       dcsr = DCSR(channel);
>>       DCSR(channel) = dcsr;
>>
>>       if (dcsr & DCSR_BUSERR) {
>> -             info->retcode = ERR_DMABUSERR;
>> +             nand->retcode = ERR_DMABUSERR;
>>       }
>>
>> -     info->state = STATE_DMA_DONE;
>> -     enable_int(info, NDCR_INT_MASK);
>> -     nand_writel(info, NDSR, NDSR_WRDREQ | NDSR_RDDREQ);
>> +     nand->state = STATE_DMA_DONE;
>> +     enable_int(nand, NDCR_INT_MASK);
>> +     nand_writel(nand, NDSR, NDSR_WRDREQ | NDSR_RDDREQ);
>>  }
>>
>>  static irqreturn_t pxa3xx_nand_irq(int irq, void *devid)
>>  {
>> -     struct pxa3xx_nand_info *info = devid;
>> -     unsigned int status, is_completed = 0;
>> +     struct pxa3xx_nand *nand = devid;
>> +     struct pxa3xx_nand_info *info;
>> +     unsigned int status, is_completed = 0, cs;
>> +     unsigned int ready, cmd_done, page_done, badblock_detect;
>>
>> -     status = nand_readl(info, NDSR);
>> +     cs              = nand->chip_select;
>> +     ready           = (cs) ? NDSR_RDY : NDSR_FLASH_RDY;
>> +     cmd_done        = (cs) ? NDSR_CS1_CMDD : NDSR_CS0_CMDD;
>> +     page_done       = (cs) ? NDSR_CS1_PAGED : NDSR_CS0_PAGED;
>> +     badblock_detect = (cs) ? NDSR_CS1_BBD : NDSR_CS0_BBD;
>
> This is confusing... do you use to ?: operator for differentiating between
> cs = 0 and cs = 1?
> I think this is a bad idea.
> Moreover, the use of ?: is discouraged among the kernel.

I could use if/else to replace this statement.

>
>> +     info            = nand->info[cs];
>>
>> +     status = nand_readl(nand, NDSR);
>>       if (status & NDSR_DBERR)
>> -             info->retcode = ERR_DBERR;
>> +             nand->retcode = ERR_DBERR;
>>       if (status & NDSR_SBERR)
>> -             info->retcode = ERR_SBERR;
>> +             nand->retcode = ERR_SBERR;
>>       if (status & (NDSR_RDDREQ | NDSR_WRDREQ)) {
>>               /* whether use dma to transfer data */
>> -             if (info->use_dma) {
>> -                     disable_int(info, NDCR_INT_MASK);
>> -                     info->state = (status & NDSR_RDDREQ) ?
>> +             if (nand->use_dma) {
>> +                     disable_int(nand, NDCR_INT_MASK);
>> +                     nand->state = (status & NDSR_RDDREQ) ?
>>                                     STATE_DMA_READING : STATE_DMA_WRITING;
>> -                     start_data_dma(info);
>> +                     start_data_dma(nand);
>>                       goto NORMAL_IRQ_EXIT;
>>               } else {
>> -                     info->state = (status & NDSR_RDDREQ) ?
>> +                     nand->state = (status & NDSR_RDDREQ) ?
>>                                     STATE_PIO_READING : STATE_PIO_WRITING;
>> -                     handle_data_pio(info);
>> +                     handle_data_pio(nand);
>>               }
>>       }
>> -     if (status & NDSR_CS0_CMDD) {
>> -             info->state = STATE_CMD_DONE;
>> +     if (status & cmd_done) {
>> +             nand->state = STATE_CMD_DONE;
>>               is_completed = 1;
>>       }
>> -     if (status & NDSR_FLASH_RDY) {
>> -             info->is_ready = 1;
>> -             info->state = STATE_READY;
>> +     if (status & ready) {
>> +             nand->is_ready = 1;
>> +             nand->state = STATE_READY;
>>       }
>>
>>       if (status & NDSR_WRCMDREQ) {
>> -             nand_writel(info, NDSR, NDSR_WRCMDREQ);
>> +             nand_writel(nand, NDSR, NDSR_WRCMDREQ);
>>               status &= ~NDSR_WRCMDREQ;
>> -             info->state = STATE_CMD_HANDLE;
>> -             nand_writel(info, NDCB0, info->ndcb0);
>> -             nand_writel(info, NDCB0, info->ndcb1);
>> -             nand_writel(info, NDCB0, info->ndcb2);
>> +             nand->state = STATE_CMD_HANDLE;
>> +             nand_writel(nand, NDCB0, nand->ndcb0);
>> +             nand_writel(nand, NDCB0, nand->ndcb1);
>> +             nand_writel(nand, NDCB0, nand->ndcb2);
>>       }
>>
>>       /* clear NDSR to let the controller exit the IRQ */
>> -     nand_writel(info, NDSR, status);
>> +     nand_writel(nand, NDSR, status);
>>       if (is_completed)
>> -             complete(&info->cmd_complete);
>> +             complete(&nand->cmd_complete);
>>  NORMAL_IRQ_EXIT:
>>       return IRQ_HANDLED;
>>  }
>>
>> -static int pxa3xx_nand_dev_ready(struct mtd_info *mtd)
>> -{
>> -     struct pxa3xx_nand_info *info = mtd->priv;
>> -     return (nand_readl(info, NDSR) & NDSR_RDY) ? 1 : 0;
>> -}
>> -
>>  static inline int is_buf_blank(uint8_t *buf, size_t len)
>>  {
>>       for (; len > 0; len--)
>> @@ -477,42 +487,49 @@ static inline int is_buf_blank(uint8_t *buf, size_t len)
>>       return 1;
>>  }
>>
>> -static int prepare_command_pool(struct pxa3xx_nand_info *info, int command,
>> +static int prepare_command_pool(struct pxa3xx_nand *nand, int command,
>>               uint16_t column, int page_addr)
>>  {
>>       uint16_t cmd;
>>       int addr_cycle, exec_cmd, ndcb0;
>> -     struct mtd_info *mtd = info->mtd;
>> +     struct mtd_info *mtd;
>> +     struct pxa3xx_nand_info *info = nand->info[nand->chip_select];
>>
>> -     ndcb0 = 0;
>> +     mtd = get_mtd_by_info(info);
>> +     ndcb0 = (nand->chip_select) ? NDCB0_CSEL : 0;
>
> This one is confusing too...
> Besides, you don't need the parenthesis.

Got that. I would fix it.

>
>>       addr_cycle = 0;
>>       exec_cmd = 1;
>>
>>       /* reset data and oob column point to handle data */
>> -     info->buf_start         = 0;
>> -     info->buf_count         = 0;
>> -     info->oob_size          = 0;
>> -     info->use_ecc           = 0;
>> -     info->is_ready          = 0;
>> -     info->retcode           = ERR_NONE;
>> +     nand->buf_start         = 0;
>> +     nand->buf_count         = 0;
>> +     nand->oob_size          = 0;
>> +     nand->use_ecc           = 0;
>> +     nand->is_ready          = 0;
>> +     nand->retcode           = ERR_NONE;
>> +     nand->data_size         = 0;
>> +     nand->use_dma           = 0;
>> +     nand->command           = command;
>>
>>       switch (command) {
>>       case NAND_CMD_READ0:
>>       case NAND_CMD_PAGEPROG:
>> -             info->use_ecc = 1;
>> +             nand->use_ecc = 1;
>>       case NAND_CMD_READOOB:
>>               pxa3xx_set_datasize(info);
>> +             nand->oob_buff = nand->data_buff + nand->data_size;
>> +             nand->use_dma = use_dma;
>>               break;
>>       case NAND_CMD_SEQIN:
>>               exec_cmd = 0;
>>               break;
>>       default:
>> -             info->ndcb1 = 0;
>> -             info->ndcb2 = 0;
>> +             nand->ndcb1 = 0;
>> +             nand->ndcb2 = 0;
>>               break;
>>       }
>>
>> -     info->ndcb0 = ndcb0;
>> +     nand->ndcb0 = ndcb0;
>>       addr_cycle = NDCB0_ADDR_CYC(info->row_addr_cycles
>>                                   + info->col_addr_cycles);
>>
>> @@ -521,16 +538,16 @@ static int prepare_command_pool(struct pxa3xx_nand_info *info, int command,
>>       case NAND_CMD_READ0:
>>               cmd = info->cmdset->read1;
>>               if (command == NAND_CMD_READOOB)
>> -                     info->buf_start = mtd->writesize + column;
>> +                     nand->buf_start = mtd->writesize + column;
>>               else
>> -                     info->buf_start = column;
>> +                     nand->buf_start = column;
>>
>>               if (unlikely(info->page_size < PAGE_CHUNK_SIZE))
>> -                     info->ndcb0 |= NDCB0_CMD_TYPE(0)
>> +                     nand->ndcb0 |= NDCB0_CMD_TYPE(0)
>>                                       | addr_cycle
>>                                       | (cmd & NDCB0_CMD1_MASK);
>>               else
>> -                     info->ndcb0 |= NDCB0_CMD_TYPE(0)
>> +                     nand->ndcb0 |= NDCB0_CMD_TYPE(0)
>>                                       | NDCB0_DBC
>>                                       | addr_cycle
>>                                       | cmd;
>> @@ -538,34 +555,34 @@ static int prepare_command_pool(struct pxa3xx_nand_info *info, int command,
>>       case NAND_CMD_SEQIN:
>>               /* small page addr setting */
>>               if (unlikely(info->page_size < PAGE_CHUNK_SIZE)) {
>> -                     info->ndcb1 = ((page_addr & 0xFFFFFF) << 8)
>> +                     nand->ndcb1 = ((page_addr & 0xFFFFFF) << 8)
>>                                       | (column & 0xFF);
>>
>> -                     info->ndcb2 = 0;
>> +                     nand->ndcb2 = 0;
>>               } else {
>> -                     info->ndcb1 = ((page_addr & 0xFFFF) << 16)
>> +                     nand->ndcb1 = ((page_addr & 0xFFFF) << 16)
>>                                       | (column & 0xFFFF);
>>
>>                       if (page_addr & 0xFF0000)
>> -                             info->ndcb2 = (page_addr & 0xFF0000) >> 16;
>> +                             nand->ndcb2 = (page_addr & 0xFF0000) >> 16;
>>                       else
>> -                             info->ndcb2 = 0;
>> +                             nand->ndcb2 = 0;
>>               }
>>
>> -             info->buf_count = mtd->writesize + mtd->oobsize;
>> -             memset(info->data_buff, 0xFF, info->buf_count);
>> +             nand->buf_count = mtd->writesize + mtd->oobsize;
>> +             memset(nand->data_buff, 0xFF, nand->buf_count);
>>
>>               break;
>>
>>       case NAND_CMD_PAGEPROG:
>> -             if (is_buf_blank(info->data_buff,
>> +             if (is_buf_blank(nand->data_buff,
>>                                       (mtd->writesize + mtd->oobsize))) {
>>                       exec_cmd = 0;
>>                       break;
>>               }
>>
>>               cmd = info->cmdset->program;
>> -             info->ndcb0 |= NDCB0_CMD_TYPE(0x1)
>> +             nand->ndcb0 |= NDCB0_CMD_TYPE(0x1)
>>                               | NDCB0_AUTO_RS
>>                               | NDCB0_ST_ROW_EN
>>                               | NDCB0_DBC
>> @@ -575,37 +592,37 @@ static int prepare_command_pool(struct pxa3xx_nand_info *info, int command,
>>
>>       case NAND_CMD_READID:
>>               cmd = info->cmdset->read_id;
>> -             info->buf_count = info->read_id_bytes;
>> -             info->ndcb0 |= NDCB0_CMD_TYPE(3)
>> +             nand->buf_count = info->read_id_bytes;
>> +             nand->ndcb0 |= NDCB0_CMD_TYPE(3)
>>                               | NDCB0_ADDR_CYC(1)
>>                               | cmd;
>>
>> -             info->data_size = 8;
>> +             nand->data_size = 8;
>>               break;
>>       case NAND_CMD_STATUS:
>>               cmd = info->cmdset->read_status;
>> -             info->buf_count = 1;
>> -             info->ndcb0 |= NDCB0_CMD_TYPE(4)
>> +             nand->buf_count = 1;
>> +             nand->ndcb0 |= NDCB0_CMD_TYPE(4)
>>                               | NDCB0_ADDR_CYC(1)
>>                               | cmd;
>>
>> -             info->data_size = 8;
>> +             nand->data_size = 8;
>>               break;
>>
>>       case NAND_CMD_ERASE1:
>>               cmd = info->cmdset->erase;
>> -             info->ndcb0 |= NDCB0_CMD_TYPE(2)
>> +             nand->ndcb0 |= NDCB0_CMD_TYPE(2)
>>                               | NDCB0_AUTO_RS
>>                               | NDCB0_ADDR_CYC(3)
>>                               | NDCB0_DBC
>>                               | cmd;
>> -             info->ndcb1 = page_addr;
>> -             info->ndcb2 = 0;
>> +             nand->ndcb1 = page_addr;
>> +             nand->ndcb2 = 0;
>>
>>               break;
>>       case NAND_CMD_RESET:
>>               cmd = info->cmdset->reset;
>> -             info->ndcb0 |= NDCB0_CMD_TYPE(5)
>> +             nand->ndcb0 |= NDCB0_CMD_TYPE(5)
>>                               | cmd;
>>
>>               break;
>> @@ -628,6 +645,7 @@ static void pxa3xx_nand_cmdfunc(struct mtd_info *mtd, unsigned command,
>>                               int column, int page_addr)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>>       int ret, exec_cmd;
>>
>>       /*
>> @@ -638,20 +656,32 @@ static void pxa3xx_nand_cmdfunc(struct mtd_info *mtd, unsigned command,
>>       if (info->reg_ndcr & NDCR_DWIDTH_M)
>>               column /= 2;
>>
>> -     exec_cmd = prepare_command_pool(info, command, column, page_addr);
>> +     /*
>> +      * There may be different NAND chip hooked to
>> +      * different chip select, so check whether
>> +      * chip select has been changed, if yes, reset the timing
>> +      */
>> +     if (nand->chip_select != info->chip_select) {
>> +             nand->chip_select = info->chip_select;
>> +             nand_writel(nand, NDTR0CS0, info->ndtr0cs0);
>> +             nand_writel(nand, NDTR1CS0, info->ndtr1cs0);
>> +     }
>> +
>> +     nand->state = STATE_PREPARED;
>> +     exec_cmd = prepare_command_pool(nand, command, column, page_addr);
>>       if (exec_cmd) {
>> -             init_completion(&info->cmd_complete);
>> -             pxa3xx_nand_start(info);
>> +             init_completion(&nand->cmd_complete);
>> +             pxa3xx_nand_start(nand);
>>
>> -             ret = wait_for_completion_timeout(&info->cmd_complete,
>> +             ret = wait_for_completion_timeout(&nand->cmd_complete,
>>                               CHIP_DELAY_TIMEOUT);
>>               if (!ret) {
>>                       printk(KERN_ERR "Wait time out!!!\n");
>>                       /* Stop State Machine for next command cycle */
>> -                     pxa3xx_nand_stop(info);
>> +                     pxa3xx_nand_stop(nand);
>>               }
>> -             info->state = STATE_IDLE;
>>       }
>> +     nand->state = STATE_IDLE;
>>  }
>>
>>  static void pxa3xx_nand_write_page_hwecc(struct mtd_info *mtd,
>> @@ -665,11 +695,12 @@ static int pxa3xx_nand_read_page_hwecc(struct mtd_info *mtd,
>>               struct nand_chip *chip, uint8_t *buf, int page)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>>
>>       chip->read_buf(mtd, buf, mtd->writesize);
>>       chip->read_buf(mtd, chip->oob_poi, mtd->oobsize);
>>
>> -     if (info->retcode == ERR_SBERR) {
>> +     if (nand->retcode == ERR_SBERR) {
>>               switch (info->use_ecc) {
>>               case 1:
>>                       mtd->ecc_stats.corrected++;
>> @@ -678,14 +709,14 @@ static int pxa3xx_nand_read_page_hwecc(struct mtd_info *mtd,
>>               default:
>>                       break;
>>               }
>> -     } else if (info->retcode == ERR_DBERR) {
>> +     } else if (nand->retcode == ERR_DBERR) {
>>               /*
>>                * for blank page (all 0xff), HW will calculate its ECC as
>>                * 0, which is different from the ECC information within
>>                * OOB, ignore such double bit errors
>>                */
>>               if (is_buf_blank(buf, mtd->writesize))
>> -                     info->retcode = ERR_NONE;
>> +                     nand->retcode = ERR_NONE;
>>               else
>>                       mtd->ecc_stats.failed++;
>>       }
>> @@ -696,11 +727,12 @@ static int pxa3xx_nand_read_page_hwecc(struct mtd_info *mtd,
>>  static uint8_t pxa3xx_nand_read_byte(struct mtd_info *mtd)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>>       char retval = 0xFF;
>>
>> -     if (info->buf_start < info->buf_count)
>> +     if (nand->buf_start < nand->buf_count)
>>               /* Has just send a new command? */
>> -             retval = info->data_buff[info->buf_start++];
>> +             retval = nand->data_buff[nand->buf_start++];
>>
>>       return retval;
>>  }
>> @@ -708,11 +740,12 @@ static uint8_t pxa3xx_nand_read_byte(struct mtd_info *mtd)
>>  static u16 pxa3xx_nand_read_word(struct mtd_info *mtd)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>>       u16 retval = 0xFFFF;
>>
>> -     if (!(info->buf_start & 0x01) && info->buf_start < info->buf_count) {
>> -             retval = *((u16 *)(info->data_buff+info->buf_start));
>> -             info->buf_start += 2;
>> +     if (!(nand->buf_start & 0x01) && nand->buf_start < nand->buf_count) {
>> +             retval = *((u16 *)(nand->data_buff+nand->buf_start));
>> +             nand->buf_start += 2;
>>       }
>>       return retval;
>>  }
>> @@ -720,20 +753,22 @@ static u16 pxa3xx_nand_read_word(struct mtd_info *mtd)
>>  static void pxa3xx_nand_read_buf(struct mtd_info *mtd, uint8_t *buf, int len)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> -     int real_len = min_t(size_t, len, info->buf_count - info->buf_start);
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     int real_len = min_t(size_t, len, nand->buf_count - nand->buf_start);
>>
>> -     memcpy(buf, info->data_buff + info->buf_start, real_len);
>> -     info->buf_start += real_len;
>> +     memcpy(buf, nand->data_buff + nand->buf_start, real_len);
>> +     nand->buf_start += real_len;
>>  }
>>
>>  static void pxa3xx_nand_write_buf(struct mtd_info *mtd,
>>               const uint8_t *buf, int len)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> -     int real_len = min_t(size_t, len, info->buf_count - info->buf_start);
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     int real_len = min_t(size_t, len, nand->buf_count - nand->buf_start);
>>
>> -     memcpy(info->data_buff + info->buf_start, buf, real_len);
>> -     info->buf_start += real_len;
>> +     memcpy(nand->data_buff + nand->buf_start, buf, real_len);
>> +     nand->buf_start += real_len;
>>  }
>>
>>  static int pxa3xx_nand_verify_buf(struct mtd_info *mtd,
>> @@ -750,10 +785,11 @@ static void pxa3xx_nand_select_chip(struct mtd_info *mtd, int chip)
>>  static int pxa3xx_nand_waitfunc(struct mtd_info *mtd, struct nand_chip *this)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>>
>>       /* pxa3xx_nand_send_command has waited for command complete */
>>       if (this->state == FL_WRITING || this->state == FL_ERASING) {
>> -             if (info->retcode == ERR_NONE)
>> +             if (nand->retcode == ERR_NONE)
>>                       return 0;
>>               else {
>>                       /*
>> @@ -770,7 +806,8 @@ static int pxa3xx_nand_waitfunc(struct mtd_info *mtd, struct nand_chip *this)
>>  static int pxa3xx_nand_config_flash(struct pxa3xx_nand_info *info,
>>                                   const struct pxa3xx_nand_flash *f)
>>  {
>> -     struct platform_device *pdev = info->pdev;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     struct platform_device *pdev = nand->pdev;
>>       struct pxa3xx_nand_platform_data *pdata = pdev->dev.platform_data;
>>       uint32_t ndcr = 0x0; /* enable all interrupts */
>>
>> @@ -804,6 +841,7 @@ static int pxa3xx_nand_config_flash(struct pxa3xx_nand_info *info,
>>       ndcr |= NDCR_SPARE_EN; /* enable spare by default */
>>
>>       info->reg_ndcr = ndcr;
>> +     info->use_ecc = 1;
>>
>>       pxa3xx_nand_set_timing(info, f->timing);
>>       return 0;
>> @@ -811,15 +849,22 @@ static int pxa3xx_nand_config_flash(struct pxa3xx_nand_info *info,
>>
>>  static int pxa3xx_nand_detect_config(struct pxa3xx_nand_info *info)
>>  {
>> -     uint32_t ndcr = nand_readl(info, NDCR);
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     uint32_t ndcr = nand_readl(nand, NDCR);
>> +
>> +     if (info->chip_select > 0) {
>> +             printk(KERN_ERR "We could not detect configure"
>> +                             " if more than one cs is supported!!\n");
>> +             BUG();
>
> like Daniel already noticed, may be dev_err() is enough?
Would change to dev_err();

>
>> +     }
>>       info->page_size = ndcr & NDCR_PAGE_SZ ? 2048 : 512;
>>       /* set info fields needed to read id */
>>       info->read_id_bytes = (info->page_size == 2048) ? 4 : 2;
>>       info->reg_ndcr = ndcr & ~NDCR_INT_MASK;
>>       info->cmdset = &default_cmdset;
>>
>> -     info->ndtr0cs0 = nand_readl(info, NDTR0CS0);
>> -     info->ndtr1cs0 = nand_readl(info, NDTR1CS0);
>> +     info->ndtr0cs0 = nand_readl(nand, NDTR0CS0);
>> +     info->ndtr1cs0 = nand_readl(nand, NDTR1CS0);
>>
>>       return 0;
>>  }
>> @@ -830,50 +875,31 @@ static int pxa3xx_nand_detect_config(struct pxa3xx_nand_info *info)
>>   */
>>  #define MAX_BUFF_SIZE        PAGE_SIZE
>>
>> -static int pxa3xx_nand_init_buff(struct pxa3xx_nand_info *info)
>> +static void free_cs_resource(struct pxa3xx_nand_info *info, int cs)
>>  {
>> -     struct platform_device *pdev = info->pdev;
>> -     int data_desc_offset = MAX_BUFF_SIZE - sizeof(struct pxa_dma_desc);
>> -
>> -     if (use_dma == 0) {
>> -             info->data_buff = kmalloc(MAX_BUFF_SIZE, GFP_KERNEL);
>> -             if (info->data_buff == NULL)
>> -                     return -ENOMEM;
>> -             return 0;
>> -     }
>> -
>> -     info->data_buff = dma_alloc_coherent(&pdev->dev, MAX_BUFF_SIZE,
>> -                             &info->data_buff_phys, GFP_KERNEL);
>> -     if (info->data_buff == NULL) {
>> -             dev_err(&pdev->dev, "failed to allocate dma buffer\n");
>> -             return -ENOMEM;
>> -     }
>> -
>> -     info->data_buff_size = MAX_BUFF_SIZE;
>> -     info->data_desc = (void *)info->data_buff + data_desc_offset;
>> -     info->data_desc_addr = info->data_buff_phys + data_desc_offset;
>> +     struct pxa3xx_nand *nand;
>> +     struct mtd_info *mtd;
>>
>> -     info->data_dma_ch = pxa_request_dma("nand-data", DMA_PRIO_LOW,
>> -                             pxa3xx_nand_data_dma_irq, info);
>> -     if (info->data_dma_ch < 0) {
>> -             dev_err(&pdev->dev, "failed to request data dma\n");
>> -             dma_free_coherent(&pdev->dev, info->data_buff_size,
>> -                             info->data_buff, info->data_buff_phys);
>> -             return info->data_dma_ch;
>> -     }
>> +     if (!info)
>> +             return;
>>
>> -     return 0;
>> +     nand = info->nand_data;
>> +     mtd = get_mtd_by_info(info);
>> +     kfree(mtd);
>> +     nand->info[cs] = NULL;
>>  }
>>
>>  static int pxa3xx_nand_sensing(struct pxa3xx_nand_info *info)
>>  {
>> -     struct mtd_info *mtd = info->mtd;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     struct mtd_info *mtd = get_mtd_by_info(info);
>>       struct nand_chip *chip = mtd->priv;
>>
>>       /* use the common timing to make a try */
>> -     pxa3xx_nand_config_flash(info, &builtin_flash_types[0]);
>> +     if (pxa3xx_nand_config_flash(info, &builtin_flash_types[0]))
>> +             return 0;
>>       chip->cmdfunc(mtd, NAND_CMD_RESET, 0, 0);
>> -     if (info->is_ready)
>> +     if (nand->is_ready)
>>               return 1;
>>       else
>>               return 0;
>
> I think it is time to change this function return convention to propagate
> errors and not just 0 or 1, (may be in separate patch) what do you think?

How about return 0 when sensing the READY signal, or return -ENODEV?

>
>> @@ -882,7 +908,8 @@ static int pxa3xx_nand_sensing(struct pxa3xx_nand_info *info)
>>  static int pxa3xx_nand_scan(struct mtd_info *mtd)
>>  {
>>       struct pxa3xx_nand_info *info = mtd->priv;
>> -     struct platform_device *pdev = info->pdev;
>> +     struct pxa3xx_nand *nand = info->nand_data;
>> +     struct platform_device *pdev = nand->pdev;
>>       struct pxa3xx_nand_platform_data *pdata = pdev->dev.platform_data;
>>       struct nand_flash_dev pxa3xx_flash_ids[2], *def = NULL;
>>       const struct pxa3xx_nand_flash *f = NULL;
>> @@ -891,27 +918,25 @@ static int pxa3xx_nand_scan(struct mtd_info *mtd)
>>       uint64_t chipsize;
>>       int i, ret, num;
>>
>> +     nand->chip_select = info->chip_select;
>>       if (pdata->keep_config && !pxa3xx_nand_detect_config(info))
>>               goto KEEP_CONFIG;
>>
>>       ret = pxa3xx_nand_sensing(info);
>>       if (!ret) {
>> -             kfree(mtd);
>> -             info->mtd = NULL;
>> -             printk(KERN_INFO "There is no nand chip on cs 0!\n");
>> +             free_cs_resource(info, nand->chip_select);
>> +             printk(KERN_INFO "There is no nand chip on cs %d!\n",
>> +                             nand->chip_select);
>>
>>               return -EINVAL;
>>       }
>>
>>       chip->cmdfunc(mtd, NAND_CMD_READID, 0, 0);
>> -     id = *((uint16_t *)(info->data_buff));
>> +     id = *((uint16_t *)(nand->data_buff));
>>       if (id != 0)
>>               printk(KERN_INFO "Detect a flash id %x\n", id);
>>       else {
>> -             kfree(mtd);
>> -             info->mtd = NULL;
>> -             printk(KERN_WARNING "Read out ID 0, potential timing set wrong!!\n");
>
> Is this warning no longer needed?
Would readd it in the next post.

>
>> -
>> +             free_cs_resource(info, nand->chip_select);
>>               return -EINVAL;
>>       }
>>
>> @@ -928,14 +953,16 @@ static int pxa3xx_nand_scan(struct mtd_info *mtd)
>>       }
>>
>>       if (i >= (ARRAY_SIZE(builtin_flash_types) + pdata->num_flash - 1)) {
>> -             kfree(mtd);
>> -             info->mtd = NULL;
>> +             free_cs_resource(info, nand->chip_select);
>>               printk(KERN_ERR "ERROR!! flash not defined!!!\n");
>>
>>               return -EINVAL;
>>       }
>>
>> -     pxa3xx_nand_config_flash(info, f);
>> +     if (pxa3xx_nand_config_flash(info, f)) {
>> +             printk(KERN_ERR "ERROR! Configure failed\n");
>> +             return -EINVAL;
>> +     }
>
> Although, the pxa3xx_nand_config_flash() returns only 0 or -EINVAL,
> it is better to propagate its return value.

Ok, would apply this.

>
>>       pxa3xx_flash_ids[0].name = f->name;
>>       pxa3xx_flash_ids[0].id = (f->chip_id >> 8) & 0xffff;
>>       pxa3xx_flash_ids[0].pagesize = f->page_size;
>> @@ -950,13 +977,13 @@ KEEP_CONFIG:
>>       if (nand_scan_ident(mtd, 1, def))
>>               return -ENODEV;
>>       /* calculate addressing information */
>> +     nand->oob_buff = nand->data_buff + mtd->writesize;
>>       info->col_addr_cycles = (mtd->writesize >= 2048) ? 2 : 1;
>> -     info->oob_buff = info->data_buff + mtd->writesize;
>>       if ((mtd->size >> chip->page_shift) > 65536)
>>               info->row_addr_cycles = 3;
>>       else
>>               info->row_addr_cycles = 2;
>> -     mtd->name = mtd_names[0];
>> +     mtd->name = mtd_names[nand->chip_select];
>>       chip->ecc.mode = NAND_ECC_HW;
>>       chip->ecc.size = info->page_size;
>>
>> @@ -967,51 +994,33 @@ KEEP_CONFIG:
>>       return nand_scan_tail(mtd);
>>  }
>>
>> -static
>> -struct pxa3xx_nand_info *alloc_nand_resource(struct platform_device *pdev)
>> +static int alloc_nand_resource(struct platform_device *pdev)
>>  {
>> +     struct pxa3xx_nand_platform_data *pdata;
>>       struct pxa3xx_nand_info *info;
>>       struct nand_chip *chip;
>>       struct mtd_info *mtd;
>> +     struct pxa3xx_nand *nand;
>>       struct resource *r;
>> -     int ret, irq;
>> +     int ret, irq, cs;
>> +     int data_desc_offset = MAX_BUFF_SIZE - sizeof(struct pxa_dma_desc);
>>
>> -     mtd = kzalloc(sizeof(struct mtd_info) + sizeof(struct pxa3xx_nand_info),
>> -                     GFP_KERNEL);
>> -     if (!mtd) {
>> +     pdata = pdev->dev.platform_data;
>> +     nand = kzalloc(sizeof(struct mtd_info)
>> +                     + sizeof(struct pxa3xx_nand_info), GFP_KERNEL);
>> +     if (!nand) {
>>               dev_err(&pdev->dev, "failed to allocate memory\n");
>> -             return NULL;
>> +             return -ENOMEM;
>>       }
>>
>> -     info = (struct pxa3xx_nand_info *)(&mtd[1]);
>> -     chip = (struct nand_chip *)(&mtd[1]);
>> -     info->pdev = pdev;
>> -     info->mtd = mtd;
>> -     mtd->priv = info;
>> -     mtd->owner = THIS_MODULE;
>> -
>> -     chip->ecc.read_page     = pxa3xx_nand_read_page_hwecc;
>> -     chip->ecc.write_page    = pxa3xx_nand_write_page_hwecc;
>> -     chip->controller        = &info->controller;
>> -     chip->waitfunc          = pxa3xx_nand_waitfunc;
>> -     chip->select_chip       = pxa3xx_nand_select_chip;
>> -     chip->dev_ready         = pxa3xx_nand_dev_ready;
>> -     chip->cmdfunc           = pxa3xx_nand_cmdfunc;
>> -     chip->read_word         = pxa3xx_nand_read_word;
>> -     chip->read_byte         = pxa3xx_nand_read_byte;
>> -     chip->read_buf          = pxa3xx_nand_read_buf;
>> -     chip->write_buf         = pxa3xx_nand_write_buf;
>> -     chip->verify_buf        = pxa3xx_nand_verify_buf;
>> -
>> -     spin_lock_init(&chip->controller->lock);
>> -     init_waitqueue_head(&chip->controller->wq);
>> -     info->clk = clk_get(&pdev->dev, NULL);
>> -     if (IS_ERR(info->clk)) {
>> +     nand->pdev = pdev;
>> +     nand->clk = clk_get(&pdev->dev, NULL);
>> +     if (IS_ERR(nand->clk)) {
>>               dev_err(&pdev->dev, "failed to get nand clock\n");
>> -             ret = PTR_ERR(info->clk);
>> -             goto fail_free_mtd;
>> +             ret = PTR_ERR(nand->clk);
>> +             goto fail_alloc;
>>       }
>> -     clk_enable(info->clk);
>> +     clk_enable(nand->clk);
>>
>>       r = platform_get_resource(pdev, IORESOURCE_DMA, 0);
>>       if (r == NULL) {
>> @@ -1019,7 +1028,7 @@ struct pxa3xx_nand_info *alloc_nand_resource(struct platform_device *pdev)
>>               ret = -ENXIO;
>>               goto fail_put_clk;
>>       }
>> -     info->drcmr_dat = r->start;
>> +     nand->drcmr_dat = r->start;
>>
>>       r = platform_get_resource(pdev, IORESOURCE_DMA, 1);
>>       if (r == NULL) {
>> @@ -1027,7 +1036,7 @@ struct pxa3xx_nand_info *alloc_nand_resource(struct platform_device *pdev)
>>               ret = -ENXIO;
>>               goto fail_put_clk;
>>       }
>> -     info->drcmr_cmd = r->start;
>> +     nand->drcmr_cmd = r->start;
>>
>>       irq = platform_get_irq(pdev, 0);
>>       if (irq < 0) {
>> @@ -1050,81 +1059,147 @@ struct pxa3xx_nand_info *alloc_nand_resource(struct platform_device *pdev)
>>               goto fail_put_clk;
>>       }
>>
>> -     info->mmio_base = ioremap(r->start, resource_size(r));
>> -     if (info->mmio_base == NULL) {
>> +     nand->mmio_base = ioremap(r->start, resource_size(r));
>> +     if (nand->mmio_base == NULL) {
>>               dev_err(&pdev->dev, "ioremap() failed\n");
>>               ret = -ENODEV;
>>               goto fail_free_res;
>>       }
>> -     info->mmio_phys = r->start;
>> -
>> -     ret = pxa3xx_nand_init_buff(info);
>> -     if (ret)
>> -             goto fail_free_io;
>> +     nand->mmio_phys = r->start;
>>
>>       /* initialize all interrupts to be disabled */
>> -     disable_int(info, NDSR_MASK);
>> +     disable_int(nand, NDSR_MASK);
>>
>>       ret = request_irq(irq, pxa3xx_nand_irq, IRQF_DISABLED,
>> -                       pdev->name, info);
>> +                       pdev->name, nand);
>>       if (ret < 0) {
>>               dev_err(&pdev->dev, "failed to request IRQ\n");
>> -             goto fail_free_buf;
>> +             ret = -ENXIO;
>> +             goto fail_free_io;
>> +     }
>> +
>> +     platform_set_drvdata(pdev, nand);
>> +
>> +     spin_lock_init(&nand->controller.lock);
>> +     init_waitqueue_head(&nand->controller.wq);
>> +     for (cs = 0; cs < pdata->cs_num; cs++) {
>> +             mtd = kzalloc(sizeof(struct mtd_info)
>> +                             + sizeof(struct pxa3xx_nand_info),
>> +                             GFP_KERNEL);
>> +             if (!mtd) {
>> +                     dev_err(&pdev->dev, "failed to allocate memory\n");
>> +                     ret = -ENOMEM;
>> +                     goto fail_free_irq;
>> +             }
>> +
>> +             info = (struct pxa3xx_nand_info *)(&mtd[1]);
>> +             info->nand_data = nand;
>> +             info->chip_select = cs;
>> +             mtd->priv = info;
>> +             mtd->owner = THIS_MODULE;
>> +             nand->info[cs] = info;
>> +
>> +             chip = (struct nand_chip *)(&mtd[1]);
>> +             chip->controller        = &nand->controller;
>> +             chip->ecc.read_page     = pxa3xx_nand_read_page_hwecc;
>> +             chip->ecc.write_page    = pxa3xx_nand_write_page_hwecc;
>> +             chip->waitfunc          = pxa3xx_nand_waitfunc;
>> +             chip->select_chip       = pxa3xx_nand_select_chip;
>> +             chip->cmdfunc           = pxa3xx_nand_cmdfunc;
>> +             chip->read_word         = pxa3xx_nand_read_word;
>> +             chip->read_byte         = pxa3xx_nand_read_byte;
>> +             chip->read_buf          = pxa3xx_nand_read_buf;
>> +             chip->write_buf         = pxa3xx_nand_write_buf;
>> +             chip->verify_buf        = pxa3xx_nand_verify_buf;
>>       }
>>
>> -     platform_set_drvdata(pdev, info);
>> +     if (use_dma == 0) {
>> +             nand->data_buff = kmalloc(MAX_BUFF_SIZE, GFP_KERNEL);
>> +             if (nand->data_buff == NULL) {
>> +                     ret = -ENOMEM;
>> +                     goto fail_free_buf;
>> +             }
>> +             goto success_exit;
>> +     }
>>
>> -     return info;
>> +     nand->data_buff = dma_alloc_coherent(&pdev->dev, MAX_BUFF_SIZE,
>> +                     &nand->data_buff_phys, GFP_KERNEL);
>> +     if (nand->data_buff == NULL) {
>> +             dev_err(&pdev->dev, "failed to allocate dma buffer\n");
>> +             ret = -ENOMEM;
>> +             goto fail_free_buf;
>> +     }
>>
>> +     nand->data_desc = (void *)nand->data_buff + data_desc_offset;
>> +     nand->data_desc_addr = nand->data_buff_phys + data_desc_offset;
>> +     nand->data_dma_ch = pxa_request_dma("nand-data", DMA_PRIO_LOW,
>> +                     pxa3xx_nand_data_dma_irq, nand);
>> +     if (nand->data_dma_ch < 0) {
>> +             dev_err(&pdev->dev, "failed to request data dma\n");
>> +             ret = -ENXIO;
>> +             goto fail_free_dma_buf;
>> +     }
>> +success_exit:
>> +     return 0;
>> +
>> +fail_free_dma_buf:
>> +     dma_free_writecombine(&pdev->dev, MAX_BUFF_SIZE,
>> +                     nand->data_buff, nand->data_buff_phys);
>>  fail_free_buf:
>> -     free_irq(irq, info);
>> -     if (use_dma) {
>> -             pxa_free_dma(info->data_dma_ch);
>> -             dma_free_coherent(&pdev->dev, info->data_buff_size,
>> -                     info->data_buff, info->data_buff_phys);
>> -     } else
>> -             kfree(info->data_buff);
>> +     for (cs = 0; cs < pdata->cs_num; cs++) {
>> +             info = nand->info[cs];
>> +             free_cs_resource(info, cs);
>> +     }
>> +fail_free_irq:
>> +     free_irq(irq, nand);
>>  fail_free_io:
>> -     iounmap(info->mmio_base);
>> +     iounmap(nand->mmio_base);
>>  fail_free_res:
>>       release_mem_region(r->start, resource_size(r));
>>  fail_put_clk:
>> -     clk_disable(info->clk);
>> -     clk_put(info->clk);
>> -fail_free_mtd:
>> -     kfree(mtd);
>> -     return NULL;
>> +     clk_disable(nand->clk);
>> +     clk_put(nand->clk);
>> +fail_alloc:
>> +     kfree(nand);
>> +     return ret;
>>  }
>>
>>  static int pxa3xx_nand_remove(struct platform_device *pdev)
>>  {
>> -     struct pxa3xx_nand_info *info = platform_get_drvdata(pdev);
>> -     struct mtd_info *mtd = info->mtd;
>> +     struct pxa3xx_nand *nand = platform_get_drvdata(pdev);
>> +     struct pxa3xx_nand_platform_data *pdata;
>> +     struct pxa3xx_nand_info *info;
>> +     struct mtd_info *mtd;
>>       struct resource *r;
>> -     int irq;
>> +     int irq, cs;
>>
>>       platform_set_drvdata(pdev, NULL);
>> +     pdata = pdev->dev.platform_data;
>>
>>       irq = platform_get_irq(pdev, 0);
>>       if (irq >= 0)
>> -             free_irq(irq, info);
>> +             free_irq(irq, nand);
>>       if (use_dma) {
>> -             pxa_free_dma(info->data_dma_ch);
>> -             dma_free_writecombine(&pdev->dev, info->data_buff_size,
>> -                             info->data_buff, info->data_buff_phys);
>> +             pxa_free_dma(nand->data_dma_ch);
>> +             dma_free_writecombine(&pdev->dev, MAX_BUFF_SIZE,
>> +                             nand->data_buff, nand->data_buff_phys);
>>       } else
>> -             kfree(info->data_buff);
>> +             kfree(nand->data_buff);
>>
>> -     iounmap(info->mmio_base);
>> +     iounmap(nand->mmio_base);
>>       r = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>>       release_mem_region(r->start, resource_size(r));
>>
>> -     clk_disable(info->clk);
>> -     clk_put(info->clk);
>> +     clk_disable(nand->clk);
>> +     clk_put(nand->clk);
>>
>> -     if (mtd) {
>> +     for (cs = 0; cs < pdata->cs_num; cs++) {
>> +             info = nand->info[cs];
>> +             if (!info)
>> +                     continue;
>> +             mtd = get_mtd_by_info(info);
>>               mtd_device_unregister(mtd);
>> -             kfree(mtd);
>> +             free_cs_resource(info, cs);
>>       }
>>       return 0;
>>  }
>> @@ -1134,34 +1209,54 @@ static int pxa3xx_nand_probe(struct platform_device *pdev)
>>       struct pxa3xx_nand_platform_data *pdata;
>>       struct pxa3xx_nand_info *info;
>>
>> +     struct pxa3xx_nand *nand;
>> +     struct mtd_info *mtd;
>> +     int cs, ret, nr_parts, probe_success;
>> +
>> +     probe_success = 0;
>
> Can this be done along with declaration?
Ok, would apply to the next post.

>
>>       pdata = pdev->dev.platform_data;
>>       if (!pdata) {
>>               dev_err(&pdev->dev, "no platform data defined\n");
>>               return -ENODEV;
>>       }
>>
>> -     info = alloc_nand_resource(pdev);
>> -     if (info == NULL)
>> +     ret = alloc_nand_resource(pdev);
>> +     if (ret)
>>               return -ENOMEM;
>
> Why not propagate the return value of alloc_nand_resource()?

Would apply propagate error to the next post.

>
>>
>> -     if (pxa3xx_nand_scan(info->mtd)) {
>> -             dev_err(&pdev->dev, "failed to scan nand\n");
>> -             pxa3xx_nand_remove(pdev);
>> -             return -ENODEV;
>> -     }
>> +     nand = platform_get_drvdata(pdev);
>> +     for (cs = 0; cs < pdata->cs_num; cs++) {
>> +             info = nand->info[cs];
>> +             mtd = get_mtd_by_info(info);
>> +             if (pxa3xx_nand_scan(mtd)) {
>> +                     dev_err(&pdev->dev, "failed to scan nand\n");
>
> I think, it would be useful also to print here the return value of pxa3xx_nand_scan().
Ok, would apply to the next post

>
>> +                     continue;
>> +             }
>> +
>> +             ret = 0;
>> +             nr_parts = 0;
>> +             if (mtd_has_cmdlinepart()) {
>> +                     const char *probes[] = { "cmdlinepart", NULL };
>> +                     struct mtd_partition *parts;
>>
>> -     if (mtd_has_cmdlinepart()) {
>> -             const char *probes[] = { "cmdlinepart", NULL };
>> -             struct mtd_partition *parts;
>> -             int nr_parts;
>> +                     nr_parts = parse_mtd_partitions(mtd, probes, &parts, 0);
>>
>> -             nr_parts = parse_mtd_partitions(info->mtd, probes, &parts, 0);
>> +                     if (nr_parts)
>> +                             ret = mtd_device_register(mtd, parts, nr_parts);
>> +             }
>>
>> -             if (nr_parts)
>> -                     return mtd_device_register(info->mtd, parts, nr_parts);
>> +             if (!nr_parts)
>> +                     ret = mtd_device_register(mtd, pdata->parts[cs],
>> +                                     pdata->nr_parts[cs]);
>> +             if (!ret)
>> +                     probe_success = 1;
>>       }
>>
>> -     return mtd_device_register(info->mtd, pdata->parts, pdata->nr_parts);
>> +     if (!probe_success) {
>> +             pxa3xx_nand_remove(pdev);
>> +             return -ENODEV;
>> +     } else
>
> You don't need this else statement

Would apply to the next post.

>
>> +             return 0;
>>  }
>>
>>  #ifdef CONFIG_PM
>> @@ -1183,8 +1278,8 @@ static int pxa3xx_nand_resume(struct platform_device *pdev)
>>       struct pxa3xx_nand_info *info = platform_get_drvdata(pdev);
>>       struct mtd_info *mtd = info->mtd;
>>
>> -     nand_writel(info, NDTR0CS0, info->ndtr0cs0);
>> -     nand_writel(info, NDTR1CS0, info->ndtr1cs0);
>> +     nand_writel(nand, NDTR0CS0, info->ndtr0cs0);
>> +     nand_writel(nand, NDTR1CS0, info->ndtr1cs0);
>>       clk_enable(info->clk);
>>
>>       return 0;
>
> I won't be able to test the patch in the near future, sorry...
>

Thanks for comments.

Best regards,
Lei



More information about the linux-mtd mailing list