[PATCH v4] mtd: gpmi: add NAND write verify support
Huang Shijie
shijie8 at gmail.com
Sat Aug 11 11:15:49 EDT 2012
On Sat, Aug 11, 2012 at 10:10 AM, Marek Vasut <marex at denx.de> wrote:
> Dear Huang Shijie,
>
>> Add NAND write verify support in gpmi-nand driver.
>>
>> This patch fixes a kernel crash when we select
>> CONFIG_MTD_NAND_VERIFY_WRITE=y, and do some write tests:
>>
>> root at freescale /$ dd if=/dev/zero of=/dev/mtd0 bs=128k count=1
>> [ 41.070000] Unable to handle kernel NULL pointer dereference at virtual
>> addr0 [ 41.080000] pgd = c77ac000
>> [ 41.080000] [00000000] *pgd=4773c831, *pte=00000000, *ppte=00000000
>> [ 41.090000] Internal error: Oops: 17 [#1] ARM
>> [ 41.090000] Modules linked in:
>> [ 41.090000] CPU: 0 Not tainted
>> (3.6.0-rc1-next-20120809-00002-ga25d017-d) [ 41.090000] PC is at
>> nand_verify_buf+0x18/0x5c
>> [ 41.090000] LR is at nand_write_page+0xc0/0x140
>> [ 41.090000] pc : [<c023e7d4>] lr : [<c023fcfc>] psr: 20000013
>> [ 41.090000] sp : c7779de4 ip : 00000000 fp : c6e60000
>> [ 41.090000] r10: 00000000 r9 : 00020000 r8 : 00000000
>> [ 41.090000] r7 : 00000000 r6 : c6e60000 r5 : c762cb10 r4 : 00000000
>> [ 41.090000] r3 : c762c8a8 r2 : 00000800 r1 : c6e60000 r0 : 00000000
>> [ 41.090000] Flags: nzCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment
>> user [ 41.090000] Control: 0005317f Table: 477ac000 DAC: 00000015
>> [ 41.090000] Process dd (pid: 427, stack limit = 0xc7778270)
>> [ 41.090000] Stack: (0xc7779de4 to 0xc777a000)
>> ...
>>
>> The crash log is fetched from Fabio's patch. thanks to Fabio's hard
> work.
>>
>> Cc: <stable at vger.kernel.org>
>> Reported-by: Fabio Estevam <fabio.estevam at freescale.com>
>> Signed-off-by: Huang Shijie <shijie8 at gmail.com>
>> ---
>> v3 --> v4:
>> [1] add credit to Fabio.
>>
>> v2 --> v3:
>> [1] free the verify_buf when we exit.
>>
>> v1 --> v2:
>> [1] use the mtd->writesize to allocate the verify buffer.
>> [2] add new field `verify_buf` in gpmi_nand_data{}
>>
>> tested this patch on imx6q-arm2 board.
>> ---
>> drivers/mtd/nand/gpmi-nand/gpmi-nand.c | 25 +++++++++++++++++++++++++
>> drivers/mtd/nand/gpmi-nand/gpmi-nand.h | 1 +
>> 2 files changed, 26 insertions(+), 0 deletions(-)
>>
>> diff --git a/drivers/mtd/nand/gpmi-nand/gpmi-nand.c
>> b/drivers/mtd/nand/gpmi-nand/gpmi-nand.c index 8c0d2f0..e1e4c51 100644
>> --- a/drivers/mtd/nand/gpmi-nand/gpmi-nand.c
>> +++ b/drivers/mtd/nand/gpmi-nand/gpmi-nand.c
>> @@ -1512,12 +1512,34 @@ static int gpmi_pre_bbt_scan(struct gpmi_nand_data
>> *this) return nand_boot_init(this);
>> }
>>
>> +static int gpmi_verify_buf(struct mtd_info *mtd, const uint8_t *buf, int
>> len) +{
>> + struct nand_chip *chip = mtd->priv;
>> + struct gpmi_nand_data *this = chip->priv;
>> + int ret;
>> +
>> + ret = chip->ecc.read_page(mtd, chip, this->verify_buf, 0, 0);
>> + if (ret)
>> + return -EFAULT;
>> + if (memcmp(buf, this->verify_buf, len))
>> + return -EFAULT;
>> + return 0;
>> +}
>> +
>> static int gpmi_scan_bbt(struct mtd_info *mtd)
>> {
>> struct nand_chip *chip = mtd->priv;
>> struct gpmi_nand_data *this = chip->priv;
>> int ret;
>>
>> + /*
>> + * We have gotten the right page size now, use it to allocate
>> + * the proper verify buffer.
>> + */
>> + this->verify_buf = kzalloc(mtd->writesize, GFP_KERNEL);
>
> devm_
emm...
The devm_kzalloc() is the wrong choose in this case. It may makes we
waste much more memory when the NAND page is 8K.
Assume the gpmi uses a 8K page nand chip. The devm_kzalloc() will
allocate (sizeof(struct devres) + 8K).
It's obvious that the size is bigger then the SLUB_MAX_SIZE. So the
kernel gets 2 with the
get_order(sizeof(struct devres) + 8K) which means we have to use 4
memory page for the verify_buf.
But in actually, we only use 8K memory. So the rest nearly 8K(
sizeof(struct devres) is very small) memory
is wasted.
So I prefer to use the kzalloc() here.
thanks
Huang Shijie
>
>> + if (!this->verify_buf)
>> + return -ENOMEM;
>> +
>> /* Prepare for the BBT scan. */
>> ret = gpmi_pre_bbt_scan(this);
>> if (ret)
>> @@ -1531,6 +1553,8 @@ void gpmi_nfc_exit(struct gpmi_nand_data *this)
>> {
>> nand_release(&this->mtd);
>> gpmi_free_dma_buffer(this);
>> + kfree(this->verify_buf);
>> + this->verify_buf = NULL;
>
> Then you won't have to care for this ... I told you at least once before.
>
>> }
>>
>> static int __devinit gpmi_nfc_init(struct gpmi_nand_data *this)
>> @@ -1556,6 +1580,7 @@ static int __devinit gpmi_nfc_init(struct
>> gpmi_nand_data *this) chip->read_byte = gpmi_read_byte;
>> chip->read_buf = gpmi_read_buf;
>> chip->write_buf = gpmi_write_buf;
>> + chip->verify_buf = gpmi_verify_buf;
>> chip->ecc.read_page = gpmi_ecc_read_page;
>> chip->ecc.write_page = gpmi_ecc_write_page;
>> chip->ecc.read_oob = gpmi_ecc_read_oob;
>> diff --git a/drivers/mtd/nand/gpmi-nand/gpmi-nand.h
>> b/drivers/mtd/nand/gpmi-nand/gpmi-nand.h index 1547a60..8ddf115 100644
>> --- a/drivers/mtd/nand/gpmi-nand/gpmi-nand.h
>> +++ b/drivers/mtd/nand/gpmi-nand/gpmi-nand.h
>> @@ -144,6 +144,7 @@ struct gpmi_nand_data {
>> /* MTD / NAND */
>> struct nand_chip nand;
>> struct mtd_info mtd;
>> + uint8_t *verify_buf;
>>
>> /* General-use Variables */
>> int current_chip;
More information about the linux-arm-kernel
mailing list