[PATCH] mtd: m25p80: Add Power Management support
Brian Norris
computersforpeace at gmail.com
Fri Jan 3 14:00:49 EST 2014
On Wed, Dec 11, 2013 at 04:19:30PM +0800, Hou Zhiqiang wrote:
> Add PM support using callback function suspend and resume in .driver of
> spi_driver.
>
> Signed-off-by: Hou Zhiqiang <b48286 at freescale.com>
> ---
> drivers/mtd/devices/m25p80.c | 37 +++++++++++++++++++++++++++++++++++++
> 1 file changed, 37 insertions(+)
>
> diff --git a/drivers/mtd/devices/m25p80.c b/drivers/mtd/devices/m25p80.c
> index 7eda71d..b0c2b8c 100644
> --- a/drivers/mtd/devices/m25p80.c
> +++ b/drivers/mtd/devices/m25p80.c
> @@ -66,6 +66,8 @@
>
> /* Used for Spansion flashes only. */
> #define OPCODE_BRWR 0x17 /* Bank register write */
> +#define OPCODE_DP 0xb9 /* Enter deep power down mode */
> +#define OPCODE_RES 0xab /* Exit deep power down mode */
Where did you get these opcodes from? They are not in the Spansion
datasheets I'm reading. And in fact, they are overloaded as RES (Read
Electronic Signature, 0xab) and Bank Register Access (0xb9) in the
datasheet I'm looking at. So this patch is wrong.
Also, can you describe the purpose of these "deep power down" modes?
I've never seen PM states where the *flash* needs to be put into a lower
power mode. Typically the flash is pretty low-power when idle, and it
may even be powered off completely when the system enters a lower-power
state. Anyway, please describe why this patch is needed.
>
> /* Status Register bits. */
> #define SR_WIP 1 /* Write in progress */
> @@ -1128,11 +1130,46 @@ static int m25p_remove(struct spi_device *spi)
> return mtd_device_unregister(&flash->mtd);
> }
>
> +#ifdef CONFIG_PM
> +static int m25p_suspend(struct device *dev, pm_message_t mesg)
> +{
> + struct m25p *flash = dev_get_drvdata(dev);
> + int ret;
> +
> + flash->command[0] = OPCODE_DP;
As mentioned above, this opcode is not recognized by many flash
supported in this driver. So we might want one or more of the following:
(1) to assign different suspend/resume opcodes for use in
m25p_suspend/resume
(2) to provide over-loadable callbacks so that different flash could
use different suspend/resume routines
And of course, we need to avoid sending these commands at all to
unsupported flash.
> + mutex_lock(&flash->lock);
> + /* Wait until finished previous write/erase command. */
> + ret = wait_till_ready(flash);
> + if (ret) {
> + mutex_unlock(&flash->lock);
> + return ret;
> + }
> + ret = spi_write(flash->spi, flash->command, 1);
> + mutex_unlock(&flash->lock);
> +
> + return ret;
> +}
> +
> +static int m25p_resume(struct device *dev)
> +{
> + struct m25p *flash = dev_get_drvdata(dev);
> + int ret;
> +
> + flash->command[0] = OPCODE_RES;
> + ret = spi_write(flash->spi, flash->command, 1);
> +
> + return ret;
> +}
> +#endif /* CONFIG_PM */
>
> static struct spi_driver m25p80_driver = {
> .driver = {
> .name = "m25p80",
> .owner = THIS_MODULE,
> +#ifdef CONFIG_PM
> + .suspend = m25p_suspend,
> + .resume = m25p_resume,
> +#endif
> },
> .id_table = m25p_ids,
> .probe = m25p_probe,
Brian
More information about the linux-mtd
mailing list