[PATCH] power management for NAND flash

[PATCH] power management for NAND flash

[Vitaly Wool]

Greetings,
the patch inlined below allows to suspend/resume NAND flash.
It's similar to what is implemented for NOR flashes in cfi_cmdset_0001.c/cfi_cmdset_0002.c.
It's based on 2.6.10 kernel but applies to the current CVS snapshot.

Best regards,
   Vitaly

Index: linux-2.6.10/include/linux/mtd/nand.h
===================================================================
--- linux-2.6.10.orig/include/linux/mtd/nand.h
+++ linux-2.6.10/include/linux/mtd/nand.h
@@ -244,6 +244,7 @@
 	FL_ERASING,
 	FL_SYNCING,
 	FL_CACHEDPRG,
+	FL_PM_SUSPENDED,
 } nand_state_t;
 
 /* Keep gcc happy */
Index: linux-2.6.10/drivers/mtd/nand/nand_base.c
===================================================================
--- linux-2.6.10.orig/drivers/mtd/nand/nand_base.c
+++ linux-2.6.10/drivers/mtd/nand/nand_base.c
@@ -46,6 +46,8 @@
  *		perform extra error status checks on erase and write failures.  This required
  *		adding a wrapper function for nand_read_ecc.
  *
+ * 08-20-2005	vwool: suspend/resume added
+ *
  * Credits:
  *	David Woodhouse for adding multichip support  
  *	
@@ -153,7 +155,7 @@
 #define nand_verify_pages(...) (0)
 #endif
 		
-static void nand_get_device (struct nand_chip *this, struct mtd_info *mtd, int new_state);
+static int nand_get_device (struct nand_chip *this, struct mtd_info *mtd, int new_state);
 
 /**
  * nand_release_device - [GENERIC] release chip
@@ -755,7 +757,7 @@
  *
  * Get the device and lock it for exclusive access
  */
-static void nand_get_device (struct nand_chip *this, struct mtd_info *mtd, int new_state)
+static int nand_get_device (struct nand_chip *this, struct mtd_info *mtd, int new_state)
 {
 	struct nand_chip *active;
 	spinlock_t *lock;
@@ -778,7 +780,11 @@
 	if (active == this && this->state == FL_READY) {
 		this->state = new_state;
 		spin_unlock(lock);
-		return;
+		return 0;
+	}
+	if (new_state == FL_PM_SUSPENDED) {
+		spin_unlock(lock);
+		return (this->state == FL_PM_SUSPENDED) ? 0 : -EAGAIN;
 	}
 	set_current_state(TASK_UNINTERRUPTIBLE);
 	add_wait_queue(wq, &wait);
@@ -2284,6 +2290,34 @@
 }
 
 /**
+ * nand_suspend - [MTD Interface] Suspend the NAND flash
+ * @mtd:	MTD device structure
+ */
+static int nand_suspend(struct mtd_info *mtd)
+{
+	struct nand_chip *this = mtd->priv;
+
+	return nand_get_device (this, mtd, FL_PM_SUSPENDED);
+}
+
+/**
+ * nand_resume - [MTD Interface] Resume the NAND flash
+ * @mtd:	MTD device structure
+ */
+static void nand_resume(struct mtd_info *mtd)
+{
+	struct nand_chip *this = mtd->priv;
+
+	if (this->state == FL_PM_SUSPENDED)
+		nand_release_device(mtd);
+	else
+		printk(KERN_ERR "resume() called for the chip which is not "
+				"in suspended state\n");
+
+}
+
+
+/**
  * nand_scan - [NAND Interface] Scan for the NAND device
  * @mtd:	MTD device structure
  * @maxchips:	Number of chips to scan for
@@ -2642,8 +2676,8 @@
 	mtd->sync = nand_sync;
 	mtd->lock = NULL;
 	mtd->unlock = NULL;
-	mtd->suspend = NULL;
-	mtd->resume = NULL;
+	mtd->suspend = nand_suspend;
+	mtd->resume = nand_resume;
 	mtd->block_isbad = nand_block_isbad;
 	mtd->block_markbad = nand_block_markbad;
 

Re: [PATCH] power management for NAND flash

[Martin Waitz]

[-- Attachment #1: Type: text/plain, Size: 262 bytes --]

hoi :)

On Mon, Sep 12, 2005 at 03:12:44PM +0400, Vitaly Wool wrote:
> the patch inlined below allows to suspend/resume NAND flash.

shouldn't we call the board driver while suspending,
to allow to actually power down the NAND flash?

-- 
Martin Waitz

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH] power management for NAND flash

[Vitaly Wool]

Hi Martin,

Exactly. We should.
But prior to shut off the clock/power down the flash we need to put the 
flash driver into the corresponding state (suspended).

Thus, a typical suspend (pseudo) code for NAND flash will look the 
following way:

static int chip_nand_suspend(struct device *dev, u32 state, u32 level)
{
        int retval = 0;

#ifdef CONFIG_PM
        struct mtd_info *mtd = dev_get_drvdata(dev);
        if (mtd && mtd->suspend && level == SUSPEND_SAVE_STATE) {
                retval = mtd->suspend(mtd);
                if (retval == 0) {
                        /* shut off the clocks and power down the 
controller here */
                }
        }
#endif
        return retval;
}

with a resume routine doing just the oppsite.


Hope that clarifies the whole thing :)

Best regards,
   Vitaly

Martin Waitz wrote:

>hoi :)
>
>On Mon, Sep 12, 2005 at 03:12:44PM +0400, Vitaly Wool wrote:
>  
>
>>the patch inlined below allows to suspend/resume NAND flash.
>>    
>>
>
>shouldn't we call the board driver while suspending,
>to allow to actually power down the NAND flash?
>
>  
>

Re: [PATCH] power management for NAND flash

[Martin Waitz]

[-- Attachment #1: Type: text/plain, Size: 208 bytes --]

hoi :)

On Mon, Sep 12, 2005 at 08:08:16PM +0400, Vitaly Wool wrote:
> Hope that clarifies the whole thing :)

yeah, thanks a lot.
now it's clear how to integrate it into drivers.

-- 
Martin Waitz

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]