[PATCH] net: eepro testing positive EBUSY return by request_irq()?

[PATCH] net: eepro testing positive EBUSY return by request_irq()?

[roel kluin]

Fix -EBUSY test for request_irq().

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
---
 drivers/net/eepro.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

I just found this in the code, no bug was observed. Is this patch needed? the test
for an -EBUSY return by request_irq() occurs much more frequently in kernel code.

diff --git a/drivers/net/eepro.c b/drivers/net/eepro.c
index 7c82631..47cfecb 100644
--- a/drivers/net/eepro.c
+++ b/drivers/net/eepro.c
@@ -920,7 +920,7 @@ static int	eepro_grab_irq(struct net_device *dev)
 
 		eepro_sw2bank0(ioaddr); /* Switch back to Bank 0 */
 
-		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != EBUSY) {
+		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != -EBUSY) {
 			unsigned long irq_mask;
 			/* Twinkle the interrupt, and check if it's seen */
 			irq_mask = probe_irq_on();

Re: [PATCH] net: eepro testing positive EBUSY return by request_irq()?

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1400 bytes --]

On Fri, 2010-12-31 at 15:47 +0100, roel kluin wrote:
> Fix -EBUSY test for request_irq().
> 
> Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
> ---
>  drivers/net/eepro.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> I just found this in the code, no bug was observed. Is this patch needed? the test
> for an -EBUSY return by request_irq() occurs much more frequently in kernel code.
> 
> diff --git a/drivers/net/eepro.c b/drivers/net/eepro.c
> index 7c82631..47cfecb 100644
> --- a/drivers/net/eepro.c
> +++ b/drivers/net/eepro.c
> @@ -920,7 +920,7 @@ static int	eepro_grab_irq(struct net_device *dev)
>  
>  		eepro_sw2bank0(ioaddr); /* Switch back to Bank 0 */
>  
> -		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != EBUSY) {
> +		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != -EBUSY) {
>  			unsigned long irq_mask;
>  			/* Twinkle the interrupt, and check if it's seen */
>  			irq_mask = probe_irq_on();

This condition is completely bogus - request_irq() with a NULL handler
now returns -EINVAL before even checking whether the IRQ is in use.  The
code should be fixed along the lines of what I did for 3c503 in commit
b0cf4dfb7cd21556efd9a6a67edcba0840b4d98d.

The e2100 and hp net drivers have the same bug.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: [PATCH] net: eepro testing positive EBUSY return by request_irq()?

[roel kluin]

>> -		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != EBUSY) {
>> +		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != -EBUSY) {

> This condition is completely bogus - request_irq() with a NULL handler
> now returns -EINVAL before even checking whether the IRQ is in use.  The
> code should be fixed along the lines of what I did for 3c503 in commit
> b0cf4dfb7cd21556efd9a6a67edcba0840b4d98d.
> 
> The e2100 and hp net drivers have the same bug.
> 
> Ben.

I've made an attempt to fix the mentioned issues, but I'm not sure it is complete.
The commit made some other changes as well which didn't seem appropriate here to me.
And I didn't compile or run test this.

As a side note, in the function changed by the commit you mentioned, el2_open(), it
appears that in current git, boolean `seen' cannot ever become true, can it? So it
loops forever until the request_irq() returns an error other than -EBUSY. Not sure
how it should be fixed though.

------------->8-------------------8<----------------
Handle errors returned by request_irq() appropriately

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
---
 drivers/net/e2100.c |   14 +++++++++-----
 drivers/net/eepro.c |   26 +++++++++++++++-----------
 drivers/net/hp.c    |   25 ++++++++++++++-----------
 3 files changed, 38 insertions(+), 27 deletions(-)

diff --git a/drivers/net/e2100.c b/drivers/net/e2100.c
index 06e72fb..115fa16 100644
--- a/drivers/net/e2100.c
+++ b/drivers/net/e2100.c
@@ -217,11 +217,15 @@ static int __init e21_probe1(struct net_device *dev, int ioaddr)
 
 	if (dev->irq < 2) {
 		int irqlist[] = {15, 11, 10, 12, 5, 9, 3, 4};
-		for (i = 0; i < ARRAY_SIZE(irqlist); i++)
-			if (request_irq (irqlist[i], NULL, 0, "bogus", NULL) != -EBUSY) {
-				dev->irq = irqlist[i];
-				break;
-			}
+		for (i = 0; i < ARRAY_SIZE(irqlist); i++) {
+			retval = request_irq (irqlist[i], NULL, 0, "bogus", NULL);
+			if (retval != -EBUSY)
+				continue;
+			if (retval < 0)
+				goto out;
+			dev->irq = irqlist[i];
+			break;
+		}
 		if (i >= ARRAY_SIZE(irqlist)) {
 			printk(" unable to get IRQ %d.\n", dev->irq);
 			retval = -EAGAIN;
diff --git a/drivers/net/eepro.c b/drivers/net/eepro.c
index 7c82631..24ff522 100644
--- a/drivers/net/eepro.c
+++ b/drivers/net/eepro.c
@@ -897,6 +897,7 @@ static int	eepro_grab_irq(struct net_device *dev)
 {
 	int irqlist[] = { 3, 4, 5, 7, 9, 10, 11, 12, 0 };
 	int *irqp = irqlist, temp_reg, ioaddr = dev->base_addr;
+	int retval;
 
 	eepro_sw2bank1(ioaddr); /* be CAREFUL, BANK 1 now */
 
@@ -919,21 +920,24 @@ static int	eepro_grab_irq(struct net_device *dev)
 		outb((temp_reg & 0xf8) | irqrmap[*irqp], ioaddr + INT_NO_REG);
 
 		eepro_sw2bank0(ioaddr); /* Switch back to Bank 0 */
+		retval = request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev);
+		if (retval == -EBUSY)
+			continue;
+		if (retval < 0)
+			return retval;
 
-		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != EBUSY) {
-			unsigned long irq_mask;
-			/* Twinkle the interrupt, and check if it's seen */
-			irq_mask = probe_irq_on();
+		unsigned long irq_mask;
+		/* Twinkle the interrupt, and check if it's seen */
+		irq_mask = probe_irq_on();
 
-			eepro_diag(ioaddr); /* RESET the 82595 */
-			mdelay(20);
+		eepro_diag(ioaddr); /* RESET the 82595 */
+		mdelay(20);
 
-			if (*irqp == probe_irq_off(irq_mask))  /* It's a good IRQ line */
-				break;
+		if (*irqp == probe_irq_off(irq_mask))  /* It's a good IRQ line */
+			break;
 
-			/* clear all interrupts */
-			eepro_clear_int(ioaddr);
-		}
+		/* clear all interrupts */
+		eepro_clear_int(ioaddr);
 	} while (*++irqp);
 
 	eepro_sw2bank1(ioaddr); /* Switch back to Bank 1 */
diff --git a/drivers/net/hp.c b/drivers/net/hp.c
index d15d2f2..aeb68a0 100644
--- a/drivers/net/hp.c
+++ b/drivers/net/hp.c
@@ -167,17 +167,20 @@ static int __init hp_probe1(struct net_device *dev, int ioaddr)
 		int *irqp = wordmode ? irq_16list : irq_8list;
 		do {
 			int irq = *irqp;
-			if (request_irq (irq, NULL, 0, "bogus", NULL) != -EBUSY) {
-				unsigned long cookie = probe_irq_on();
-				/* Twinkle the interrupt, and check if it's seen. */
-				outb_p(irqmap[irq] | HP_RUN, ioaddr + HP_CONFIGURE);
-				outb_p( 0x00 | HP_RUN, ioaddr + HP_CONFIGURE);
-				if (irq == probe_irq_off(cookie)		 /* It's a good IRQ line! */
-					&& request_irq (irq, eip_interrupt, 0, DRV_NAME, dev) == 0) {
-					printk(" selecting IRQ %d.\n", irq);
-					dev->irq = *irqp;
-					break;
-				}
+			retval = request_irq (irq, NULL, 0, "bogus", NULL);
+			if (retval == -EBUSY)
+				continue;
+			if (retval < 0)
+				goto out;
+			unsigned long cookie = probe_irq_on();
+			/* Twinkle the interrupt, and check if it's seen. */
+			outb_p(irqmap[irq] | HP_RUN, ioaddr + HP_CONFIGURE);
+			outb_p( 0x00 | HP_RUN, ioaddr + HP_CONFIGURE);
+			if (irq == probe_irq_off(cookie)		 /* It's a good IRQ line! */
+				&& request_irq (irq, eip_interrupt, 0, DRV_NAME, dev) == 0) {
+				printk(" selecting IRQ %d.\n", irq);
+				dev->irq = *irqp;
+				break;
 			}
 		} while (*++irqp);
 		if (*irqp == 0) {

Re: [PATCH] net: eepro testing positive EBUSY return by request_irq()?

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 2115 bytes --]

On Sun, 2011-01-02 at 15:52 +0100, roel kluin wrote:
> >> -		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != EBUSY) {
> >> +		if (request_irq (*irqp, NULL, IRQF_SHARED, "bogus", dev) != -EBUSY) {
> 
> > This condition is completely bogus - request_irq() with a NULL handler
> > now returns -EINVAL before even checking whether the IRQ is in use.  The
> > code should be fixed along the lines of what I did for 3c503 in commit
> > b0cf4dfb7cd21556efd9a6a67edcba0840b4d98d.
> > 
> > The e2100 and hp net drivers have the same bug.
> > 
> > Ben.
> 
> I've made an attempt to fix the mentioned issues, but I'm not sure it is complete.
> The commit made some other changes as well which didn't seem appropriate here to me.

They are absolutely necessary.

In the bad old days before plug-and-play, some expansion cards would
have jumpers to configure the IRQ number and no software mechanism to
set or query it.  Users were expected to specify the IRQ to the driver
explicitly.  probe_irq_{on,off}() allow drivers to avoid this by
triggering an IRQ and checking which was received.  That assumes that
there are no IRQ conflicts, and it only works reliably in a quiescent
system.

These drivers have been abusing request_irq(), probe_irq_{on,off}() to
do resource allocation for hardware that *does* have a software
mechanism to set the IRQ number.  This has become less reliable as
driver loading has been made more dynamic, and is now broken completely
due to the change in request_irq().  I believe the correct way to do
this allocation is what I did in 3c503.  Possibly it's worth adding a
generic function.

> And I didn't compile or run test this.

I compiled but didn't have hardware to test.

> As a side note, in the function changed by the commit you mentioned, el2_open(), it
> appears that in current git, boolean `seen' cannot ever become true, can it?
[...]

The IRQ handler data is set to &seen and el2_probe_interrupt() uses that
pointer to set it.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: [PATCH] net: eepro testing positive EBUSY return by request_irq()?

[David Miller]

From: roel kluin <roel.kluin@gmail.com>
Date: Sun, 02 Jan 2011 15:52:23 +0100

> +		for (i = 0; i < ARRAY_SIZE(irqlist); i++) {
> +			retval = request_irq (irqlist[i], NULL, 0, "bogus", NULL);
> +			if (retval != -EBUSY)
> +				continue;
> +			if (retval < 0)
> +				goto out;
> +			dev->irq = irqlist[i];
> +			break;

This series of tests don't make much sense.

If we get to the "retval < 0" check, retval must be -EBUSY.  So at
best it's superfluous, at worst it's confusing.