Patchwork mtd: mxc_nand: do not depend on disabling the irq in the interrupt handler

login
register
mail settings
Submitter Sascha Hauer
Date Sept. 21, 2010, 7:40 a.m.
Message ID <1285054802-21388-2-git-send-email-s.hauer@pengutronix.de>
Download mbox | patch
Permalink /patch/65279/
State Accepted
Commit 63f1474c69c061f923068a25f136dca0c49cd478
Headers show

Comments

Sascha Hauer - Sept. 21, 2010, 7:40 a.m.
This patch reverts the driver to enabling/disabling the NFC interrupt
mask rather than enabling/disabling the system interrupt. This cleans
up the driver so that it doesn't rely on interrupts being disabled
within the interrupt handler.
For i.MX21 we keep the current behaviour, that is calling
enable_irq/disable_irq_nosync to enable/disable interrupts.
This patch is based on earlier work by John Ogness.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: John Ogness <john.ogness@linutronix.de>
Tested-by: John Ogness <john.ogness@linutronix.de>
---
 drivers/mtd/nand/mxc_nand.c |   92 ++++++++++++++++++++++++++++++++++++++----
 1 files changed, 83 insertions(+), 9 deletions(-)
Artem Bityutskiy - Sept. 23, 2010, 11:16 a.m.
On Tue, 2010-09-21 at 09:40 +0200, Sascha Hauer wrote:
> This patch reverts the driver to enabling/disabling the NFC interrupt
> mask rather than enabling/disabling the system interrupt. This cleans
> up the driver so that it doesn't rely on interrupts being disabled
> within the interrupt handler.
> For i.MX21 we keep the current behaviour, that is calling
> enable_irq/disable_irq_nosync to enable/disable interrupts.
> This patch is based on earlier work by John Ogness.
> 
> Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> Acked-by: John Ogness <john.ogness@linutronix.de>
> Tested-by: John Ogness <john.ogness@linutronix.de>

I've created l2-mtd-2.6.git/for-2.6.36 branch, pushed your patch there,
pinged dwmw2 for you. I do not give you any guarantees this will reach
2.6.36, you have to ping dwmw2.
Sascha Hauer - Sept. 29, 2010, 7:28 a.m.
David,

On Thu, Sep 23, 2010 at 02:16:52PM +0300, Artem Bityutskiy wrote:
> On Tue, 2010-09-21 at 09:40 +0200, Sascha Hauer wrote:
> > This patch reverts the driver to enabling/disabling the NFC interrupt
> > mask rather than enabling/disabling the system interrupt. This cleans
> > up the driver so that it doesn't rely on interrupts being disabled
> > within the interrupt handler.
> > For i.MX21 we keep the current behaviour, that is calling
> > enable_irq/disable_irq_nosync to enable/disable interrupts.
> > This patch is based on earlier work by John Ogness.
> > 
> > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> > Acked-by: John Ogness <john.ogness@linutronix.de>
> > Tested-by: John Ogness <john.ogness@linutronix.de>
> 
> I've created l2-mtd-2.6.git/for-2.6.36 branch, pushed your patch there,
> pinged dwmw2 for you. I do not give you any guarantees this will reach
> 2.6.36, you have to ping dwmw2.

Any chance to get this in? The mxc-nand driver is nonfunctional at the
moment and 2.6.36 is getting closer.

Sascha
Eric Benard - Sept. 30, 2010, 7:21 p.m.
Le 29/09/2010 09:28, Sascha Hauer a écrit :
> David,
>
> On Thu, Sep 23, 2010 at 02:16:52PM +0300, Artem Bityutskiy wrote:
>> On Tue, 2010-09-21 at 09:40 +0200, Sascha Hauer wrote:
>>> This patch reverts the driver to enabling/disabling the NFC interrupt
>>> mask rather than enabling/disabling the system interrupt. This cleans
>>> up the driver so that it doesn't rely on interrupts being disabled
>>> within the interrupt handler.
>>> For i.MX21 we keep the current behaviour, that is calling
>>> enable_irq/disable_irq_nosync to enable/disable interrupts.
>>> This patch is based on earlier work by John Ogness.
>>>
>>> Signed-off-by: Sascha Hauer<s.hauer@pengutronix.de>
>>> Acked-by: John Ogness<john.ogness@linutronix.de>
>>> Tested-by: John Ogness<john.ogness@linutronix.de>
>>
>> I've created l2-mtd-2.6.git/for-2.6.36 branch, pushed your patch there,
>> pinged dwmw2 for you. I do not give you any guarantees this will reach
>> 2.6.36, you have to ping dwmw2.
>
> Any chance to get this in? The mxc-nand driver is nonfunctional at the
> moment and 2.6.36 is getting closer.
>
Acked-by: Eric Bénard <eric@eukrea.com>
Tested-by: Eric B2nard <eric@eukrea.com>

Eric
Uwe Kleine-König - Oct. 4, 2010, 6:58 a.m.
On Wed, Sep 29, 2010 at 09:28:43AM +0200, Sascha Hauer wrote:
> David,
> 
> On Thu, Sep 23, 2010 at 02:16:52PM +0300, Artem Bityutskiy wrote:
> > On Tue, 2010-09-21 at 09:40 +0200, Sascha Hauer wrote:
> > > This patch reverts the driver to enabling/disabling the NFC interrupt
> > > mask rather than enabling/disabling the system interrupt. This cleans
> > > up the driver so that it doesn't rely on interrupts being disabled
> > > within the interrupt handler.
> > > For i.MX21 we keep the current behaviour, that is calling
> > > enable_irq/disable_irq_nosync to enable/disable interrupts.
> > > This patch is based on earlier work by John Ogness.
> > > 
> > > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> > > Acked-by: John Ogness <john.ogness@linutronix.de>
> > > Tested-by: John Ogness <john.ogness@linutronix.de>
> > 
> > I've created l2-mtd-2.6.git/for-2.6.36 branch, pushed your patch there,
> > pinged dwmw2 for you. I do not give you any guarantees this will reach
> > 2.6.36, you have to ping dwmw2.
> 
> Any chance to get this in? The mxc-nand driver is nonfunctional at the
> moment and 2.6.36 is getting closer.

Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

(for the patch and the request to get it in before 2.6.36.)

Uwe
Eric Benard - Oct. 15, 2010, 7:34 p.m.
Hi David,

Le 23/09/2010 13:16, Artem Bityutskiy a écrit :
> On Tue, 2010-09-21 at 09:40 +0200, Sascha Hauer wrote:
>> This patch reverts the driver to enabling/disabling the NFC interrupt
>> mask rather than enabling/disabling the system interrupt. This cleans
>> up the driver so that it doesn't rely on interrupts being disabled
>> within the interrupt handler.
>> For i.MX21 we keep the current behaviour, that is calling
>> enable_irq/disable_irq_nosync to enable/disable interrupts.
>> This patch is based on earlier work by John Ogness.
>>
>> Signed-off-by: Sascha Hauer<s.hauer@pengutronix.de>
>> Acked-by: John Ogness<john.ogness@linutronix.de>
>> Tested-by: John Ogness<john.ogness@linutronix.de>
>
> I've created l2-mtd-2.6.git/for-2.6.36 branch, pushed your patch there,
> pinged dwmw2 for you. I do not give you any guarantees this will reach
> 2.6.36, you have to ping dwmw2.
>
Would that be possible to merge Artem's l2-mtd-2.6.git/for-2.6.36 branch for 
2.6.36 (or at least cherry-pick 34e15930f2c322389d4f946bf891601a2d1e4280 from 
this branch) ? Without it mxc-nand is broken in 2.6.36.

Many thanks
Eric

Patch

diff --git a/drivers/mtd/nand/mxc_nand.c b/drivers/mtd/nand/mxc_nand.c
index b2828e8..214b03a 100644
--- a/drivers/mtd/nand/mxc_nand.c
+++ b/drivers/mtd/nand/mxc_nand.c
@@ -30,6 +30,8 @@ 
 #include <linux/clk.h>
 #include <linux/err.h>
 #include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/completion.h>
 
 #include <asm/mach/flash.h>
 #include <mach/mxc_nand.h>
@@ -151,7 +153,7 @@  struct mxc_nand_host {
 	int			irq;
 	int			eccsize;
 
-	wait_queue_head_t	irq_waitq;
+	struct completion	op_completion;
 
 	uint8_t			*data_buf;
 	unsigned int		buf_start;
@@ -164,6 +166,7 @@  struct mxc_nand_host {
 	void			(*send_read_id)(struct mxc_nand_host *);
 	uint16_t		(*get_dev_status)(struct mxc_nand_host *);
 	int			(*check_int)(struct mxc_nand_host *);
+	void			(*irq_control)(struct mxc_nand_host *, int);
 };
 
 /* OOB placement block for use with hardware ecc generation */
@@ -216,9 +219,12 @@  static irqreturn_t mxc_nfc_irq(int irq, void *dev_id)
 {
 	struct mxc_nand_host *host = dev_id;
 
-	disable_irq_nosync(irq);
+	if (!host->check_int(host))
+		return IRQ_NONE;
 
-	wake_up(&host->irq_waitq);
+	host->irq_control(host, 0);
+
+	complete(&host->op_completion);
 
 	return IRQ_HANDLED;
 }
@@ -245,11 +251,54 @@  static int check_int_v1_v2(struct mxc_nand_host *host)
 	if (!(tmp & NFC_V1_V2_CONFIG2_INT))
 		return 0;
 
-	writew(tmp & ~NFC_V1_V2_CONFIG2_INT, NFC_V1_V2_CONFIG2);
+	if (!cpu_is_mx21())
+		writew(tmp & ~NFC_V1_V2_CONFIG2_INT, NFC_V1_V2_CONFIG2);
 
 	return 1;
 }
 
+/*
+ * It has been observed that the i.MX21 cannot read the CONFIG2:INT bit
+ * if interrupts are masked (CONFIG1:INT_MSK is set). To handle this, the
+ * driver can enable/disable the irq line rather than simply masking the
+ * interrupts.
+ */
+static void irq_control_mx21(struct mxc_nand_host *host, int activate)
+{
+	if (activate)
+		enable_irq(host->irq);
+	else
+		disable_irq_nosync(host->irq);
+}
+
+static void irq_control_v1_v2(struct mxc_nand_host *host, int activate)
+{
+	uint16_t tmp;
+
+	tmp = readw(NFC_V1_V2_CONFIG1);
+
+	if (activate)
+		tmp &= ~NFC_V1_V2_CONFIG1_INT_MSK;
+	else
+		tmp |= NFC_V1_V2_CONFIG1_INT_MSK;
+
+	writew(tmp, NFC_V1_V2_CONFIG1);
+}
+
+static void irq_control_v3(struct mxc_nand_host *host, int activate)
+{
+	uint32_t tmp;
+
+	tmp = readl(NFC_V3_CONFIG2);
+
+	if (activate)
+		tmp &= ~NFC_V3_CONFIG2_INT_MSK;
+	else
+		tmp |= NFC_V3_CONFIG2_INT_MSK;
+
+	writel(tmp, NFC_V3_CONFIG2);
+}
+
 /* This function polls the NANDFC to wait for the basic operation to
  * complete by checking the INT bit of config2 register.
  */
@@ -259,10 +308,9 @@  static void wait_op_done(struct mxc_nand_host *host, int useirq)
 
 	if (useirq) {
 		if (!host->check_int(host)) {
-
-			enable_irq(host->irq);
-
-			wait_event(host->irq_waitq, host->check_int(host));
+			INIT_COMPLETION(host->op_completion);
+			host->irq_control(host, 1);
+			wait_for_completion(&host->op_completion);
 		}
 	} else {
 		while (max_retries-- > 0) {
@@ -799,6 +847,7 @@  static void preset_v3(struct mtd_info *mtd)
 		NFC_V3_CONFIG2_2CMD_PHASES |
 		NFC_V3_CONFIG2_SPAS(mtd->oobsize >> 1) |
 		NFC_V3_CONFIG2_ST_CMD(0x70) |
+		NFC_V3_CONFIG2_INT_MSK |
 		NFC_V3_CONFIG2_NUM_ADDR_PHASE0;
 
 	if (chip->ecc.mode == NAND_ECC_HW)
@@ -1024,6 +1073,10 @@  static int __init mxcnd_probe(struct platform_device *pdev)
 		host->send_read_id = send_read_id_v1_v2;
 		host->get_dev_status = get_dev_status_v1_v2;
 		host->check_int = check_int_v1_v2;
+		if (cpu_is_mx21())
+			host->irq_control = irq_control_mx21;
+		else
+			host->irq_control = irq_control_v1_v2;
 	}
 
 	if (nfc_is_v21()) {
@@ -1062,6 +1115,7 @@  static int __init mxcnd_probe(struct platform_device *pdev)
 		host->send_read_id = send_read_id_v3;
 		host->check_int = check_int_v3;
 		host->get_dev_status = get_dev_status_v3;
+		host->irq_control = irq_control_v3;
 		oob_smallpage = &nandv2_hw_eccoob_smallpage;
 		oob_largepage = &nandv2_hw_eccoob_largepage;
 	} else
@@ -1093,14 +1147,34 @@  static int __init mxcnd_probe(struct platform_device *pdev)
 		this->options |= NAND_USE_FLASH_BBT;
 	}
 
-	init_waitqueue_head(&host->irq_waitq);
+	init_completion(&host->op_completion);
 
 	host->irq = platform_get_irq(pdev, 0);
 
+	/*
+	 * mask the interrupt. For i.MX21 explicitely call
+	 * irq_control_v1_v2 to use the mask bit. We can't call
+	 * disable_irq_nosync() for an interrupt we do not own yet.
+	 */
+	if (cpu_is_mx21())
+		irq_control_v1_v2(host, 0);
+	else
+		host->irq_control(host, 0);
+
 	err = request_irq(host->irq, mxc_nfc_irq, IRQF_DISABLED, DRIVER_NAME, host);
 	if (err)
 		goto eirq;
 
+	host->irq_control(host, 0);
+
+	/*
+	 * Now that the interrupt is disabled make sure the interrupt
+	 * mask bit is cleared on i.MX21. Otherwise we can't read
+	 * the interrupt status bit on this machine.
+	 */
+	if (cpu_is_mx21())
+		irq_control_v1_v2(host, 1);
+
 	/* first scan to find the device and get the page size */
 	if (nand_scan_ident(mtd, 1, NULL)) {
 		err = -ENXIO;