[PATCH 01/10] EDAC/altera: Fix NULL of_node dereference altr_edac_device_probe()

[PATCH 01/10] EDAC/altera: Fix NULL of_node dereference altr_edac_device_probe()

[Dinh Nguyen]

In altr_edac_device_probe() dereferenced pdev->dev.of_node and
of_match_node() before validating it, which could cause a NULL pointer
dereference when the device tree node is missing. Check the result for
a NULL and bail out with -ENODEV before using it.

Assisted-by: Cursor:claude-4.8-opus
Fixes: c3eea1942a16 ("EDAC, altera: Add Altera L2 cache and OCRAM support")
Closes: https://sashiko.dev/#/patchset/20260719211238.589402-1-rosenp%40gmail.com
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index 68846f583eeef..d6ca68d74f78a 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -710,9 +710,18 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 	struct resource *r;
 	int res = 0;
 	struct device_node *np = pdev->dev.of_node;
-	char *ecc_name = (char *)np->name;
+	const struct of_device_id *edac_dev_match;
+	char *ecc_name;
 	static int dev_instance;
 
+	if (!np) {
+		edac_printk(KERN_ERR, EDAC_DEVICE,
+			    "Unable to get device tree node\n");
+		return -ENODEV;
+	}
+
+	ecc_name = (char *)np->name;
+
 	if (!devres_open_group(&pdev->dev, NULL, GFP_KERNEL)) {
 		edac_printk(KERN_ERR, EDAC_DEVICE,
 			    "Unable to open devm\n");
@@ -757,7 +766,14 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 	}
 
 	/* Get driver specific data for this EDAC device */
-	drvdata->data = of_match_node(altr_edac_device_of_match, np)->data;
+	edac_dev_match = of_match_node(altr_edac_device_of_match, np);
+	if (!edac_dev_match) {
+		edac_printk(KERN_ERR, EDAC_DEVICE,
+			    "%s: Unable to match device\n", ecc_name);
+		res = -ENODEV;
+		goto fail1;
+	}
+	drvdata->data = edac_dev_match->data;
 
 	/* Check specific dependencies for the module */
 	if (drvdata->data->setup) {
-- 
2.42.0.411.g813d9a9188

Re: [PATCH 01/10] EDAC/altera: Fix NULL of_node dereference altr_edac_device_probe()

[Borislav Petkov]

On Mon, Jul 27, 2026 at 08:24:07AM -0500, Dinh Nguyen wrote:
> In altr_edac_device_probe() dereferenced pdev->dev.of_node and
> of_match_node() before validating it,

How did that AI let you fumble that sentence? :-P

> which could cause a NULL pointer
> dereference when the device tree node is missing. Check the result for
> a NULL and bail out with -ENODEV before using it.
> 
> Assisted-by: Cursor:claude-4.8-opus
> Fixes: c3eea1942a16 ("EDAC, altera: Add Altera L2 cache and OCRAM support")
> Closes: https://sashiko.dev/#/patchset/20260719211238.589402-1-rosenp%40gmail.com

So the AI review for this one here is this one, I think:

"This isn't a bug introduced by this patch, but can np be NULL here if the
driver is bound via sysfs?
If a privileged user unbinds and rebinds the driver via sysfs to a platform
device that lacks an of_node, pdev->dev.of_node will be NULL, which would
cause an unconditional dereference when accessing np->name."

Can that happen?

Have you been able to trigger it?

If so, then the mechanism how this happens should be in the commit message,
explaining why this fix is needed and why it is needed now, all of a sudden.

AI is helpful in some cases but I would like us to sanity-check everything it
hallucinates pls.

> Cc: stable@vger.kernel.org
> Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
> ---
>  drivers/edac/altera_edac.c | 20 ++++++++++++++++++--
>  1 file changed, 18 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
> index 68846f583eeef..d6ca68d74f78a 100644
> --- a/drivers/edac/altera_edac.c
> +++ b/drivers/edac/altera_edac.c
> @@ -710,9 +710,18 @@ static int altr_edac_device_probe(struct platform_device *pdev)
>  	struct resource *r;
>  	int res = 0;
>  	struct device_node *np = pdev->dev.of_node;
> -	char *ecc_name = (char *)np->name;
> +	const struct of_device_id *edac_dev_match;
> +	char *ecc_name;
>  	static int dev_instance;

I know this is not something you should do now but let's pay attention to
this:

The preferred ordering of variable declarations at the beginning of a function
is reverse fir tree order::

	struct long_struct_name *descriptive_name;
	unsigned long foo, bar;
	unsigned int tmp;
	int ret;

The above is faster to parse than the reverse ordering::

	int ret;
	unsigned int tmp;
	unsigned long foo, bar;
	struct long_struct_name *descriptive_name;

And even more so than random ordering::

	unsigned long foo, bar;
	int ret;
	struct long_struct_name *descriptive_name;
	unsigned int tmp;
  
> +	if (!np) {
> +		edac_printk(KERN_ERR, EDAC_DEVICE,
> +			    "Unable to get device tree node\n");
> +		return -ENODEV;
> +	}

Pls write it like this:

	np = pdev->dev.of_node;
	if (!np)
		...

> +
> +	ecc_name = (char *)np->name;
> +
>  	if (!devres_open_group(&pdev->dev, NULL, GFP_KERNEL)) {
>  		edac_printk(KERN_ERR, EDAC_DEVICE,
>  			    "Unable to open devm\n");

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [PATCH 01/10] EDAC/altera: Fix NULL of_node dereference altr_edac_device_probe()

[Dinh Nguyen]

On 7/28/26 13:56, Borislav Petkov wrote:
> On Mon, Jul 27, 2026 at 08:24:07AM -0500, Dinh Nguyen wrote:
>> In altr_edac_device_probe() dereferenced pdev->dev.of_node and
>> of_match_node() before validating it,
> 
> How did that AI let you fumble that sentence? :-P

sorry about that...

> 
>> which could cause a NULL pointer
>> dereference when the device tree node is missing. Check the result for
>> a NULL and bail out with -ENODEV before using it.
>>
>> Assisted-by: Cursor:claude-4.8-opus
>> Fixes: c3eea1942a16 ("EDAC, altera: Add Altera L2 cache and OCRAM support")
>> Closes: https://sashiko.dev/#/patchset/20260719211238.589402-1-rosenp%40gmail.com
> 
> So the AI review for this one here is this one, I think:
> 
> "This isn't a bug introduced by this patch, but can np be NULL here if the
> driver is bound via sysfs?
> If a privileged user unbinds and rebinds the driver via sysfs to a platform
> device that lacks an of_node, pdev->dev.of_node will be NULL, which would
> cause an unconditional dereference when accessing np->name."
> 
> Can that happen?

I don't think this can ever happen. The SoCFPGA platform is strictly OF 
enabled, so not having an of_node would be impossible.

Let me rework this down some more. Sorry for the noise!

Dinh

[PATCH 02/10] EDAC/altera: Fix use-after-free in error paths

[Dinh Nguyen]

In both altr_edac_a10_device_add() and altr_portb_setup(), the error path
freed dci before releasing the devres group. Since the managed SB/DB IRQ
handlers use altdev(dci->pvt_info) as their data, an IRQ firing between
freeing dci and unregistering the IRQs could dereference freed memory.

Release the devres group first so the managed IRQs are unregistered
before dci is freed.

Assisted-by: Cursor:claude-4.8-opus
Fixes: 911049845d70 ("EDAC, altera: Add Arria10 SD-MMC EDAC support")
Fixes: 588cb03ea208 ("EDAC, altera: Add Arria10 L2 Cache ECC handling")
Closes: https://sashiko.dev/#/patchset/20260719211238.589402-1-rosenp%40gmail.com
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index d6ca68d74f78a..34b33298cf86e 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -1625,8 +1625,13 @@ static int altr_portb_setup(struct altr_edac_device_dev *device)
 	return 0;
 
 err_release_group_1:
-	edac_device_free_ctl_info(dci);
+	/*
+	 * Release the devres group first so the managed IRQs are
+	 * unregistered before dci (which contains the IRQ handler's
+	 * data via dci->pvt_info) is freed, avoiding a use-after-free.
+	 */
 	devres_release_group(device->edac->dev, altr_portb_setup);
+	edac_device_free_ctl_info(dci);
 	edac_printk(KERN_ERR, EDAC_DEVICE,
 		    "%s:Error setting up EDAC device: %d\n", ecc_name, rc);
 	return rc;
@@ -2029,9 +2034,17 @@ static int altr_edac_a10_device_add(struct altr_arria10_edac *edac,
 	return 0;
 
 err_release_group1:
+	/*
+	 * Release the devres group first so the managed IRQs are
+	 * unregistered before dci (which contains the IRQ handler's
+	 * data via dci->pvt_info) is freed, avoiding a use-after-free.
+	 */
+	devres_release_group(edac->dev, NULL);
 	edac_device_free_ctl_info(dci);
+	goto err_print;
 err_release_group:
 	devres_release_group(edac->dev, NULL);
+err_print:
 	edac_printk(KERN_ERR, EDAC_DEVICE,
 		    "%s:Error setting up EDAC device: %d\n", ecc_name, rc);
 
-- 
2.42.0.411.g813d9a9188

[PATCH 03/10] EDAC/altera: Guard against NULL match data in altr_sdram_probe()

[Dinh Nguyen]

sashiko reports: If a device is manually bound and lacks a valid match entry,
of_match_node() or device_get_match_data() could return NULL.

altr_sdram_probe() dereferenced the pointer returned by of_match_node()
and device_get_match_data() without checking it, which could cause a NULL
pointer dereference when no match data is found. Add a guard that logs
an error and returns -ENODEV.

Assisted-by: Cursor:claude-4.8-opus
Fixes: 143f4a5ac5af ("EDAC, altera: Refactor for Altera CycloneV SoC")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index 34b33298cf86e..8cbf71ce5d2d8 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -300,6 +300,10 @@ static int altr_sdram_probe(struct platform_device *pdev)
 
 	/* Check specific dependencies for the module */
 	priv = device_get_match_data(&pdev->dev);
+	if (!priv) {
+		edac_printk(KERN_ERR, EDAC_MC, "No match data found\n");
+		return -ENODEV;
+	}
 
 	/* Validate the SDRAM controller has ECC enabled */
 	if (regmap_read(mc_vbase, priv->ecc_ctrl_offset, &read_reg) ||
-- 
2.42.0.411.g813d9a9188

[PATCH 04/10] EDAC/altera: Fix IRQ domain leak on platform_get_irq() failure

[Dinh Nguyen]

sashiko reports: If platform_get_irq() fails, the function returns without
freeing the previously allocated edac->domain.

altr_edac_a10_probe() created an IRQ domain via irq_domain_create_linear()
but returned without freeing it when the subsequent platform_get_irq()
calls for the SB or DB IRQ failed, leaking the domain. Call
irq_domain_remove() before returning on those error paths.

Assisted-by: Cursor:claude-4.8-opus
Fixes: 13ab8448d2c9 ("EDAC, altera: Add ECC Manager IRQ controller support")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index 8cbf71ce5d2d8..e5d225f799f39 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -2179,8 +2179,10 @@ static int altr_edac_a10_probe(struct platform_device *pdev)
 	}
 
 	edac->sb_irq = platform_get_irq(pdev, 0);
-	if (edac->sb_irq < 0)
+	if (edac->sb_irq < 0) {
+		irq_domain_remove(edac->domain);
 		return edac->sb_irq;
+	}
 
 	irq_set_chained_handler_and_data(edac->sb_irq,
 					 altr_edac_a10_irq_handler,
@@ -2210,8 +2212,10 @@ static int altr_edac_a10_probe(struct platform_device *pdev)
 		}
 	} else {
 		edac->db_irq = platform_get_irq(pdev, 1);
-		if (edac->db_irq < 0)
+		if (edac->db_irq < 0) {
+			irq_domain_remove(edac->domain);
 			return edac->db_irq;
+		}
 
 		irq_set_chained_handler_and_data(edac->db_irq,
 						 altr_edac_a10_irq_handler, edac);
-- 
2.42.0.411.g813d9a9188

[PATCH 05/10] EDAC/altera: Fix code leak on dci allocation failure

[Dinh Nguyen]

sashiko reports: If devres_open_group() fails, the function returns -ENOMEM
without freeing the dci structure allocated earlier with edac_device_alloc_ctl_info().

Free the dci structure if devres_open_group() fails.

Fixes: c3eea1942a16 ("EDAC, altera: Add Altera L2 cache and OCRAM support")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index e5d225f799f39..09e040b244673 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -1554,8 +1554,10 @@ static int altr_portb_setup(struct altr_edac_device_dev *device)
 	altdev = dci->pvt_info;
 	*altdev = *device;
 
-	if (!devres_open_group(device->edac->dev, altr_portb_setup, GFP_KERNEL))
+	if (!devres_open_group(device->edac->dev, altr_portb_setup, GFP_KERNEL)) {
+		edac_device_free_ctl_info(dci);
 		return -ENOMEM;
+	}
 
 	/* Update PortB specific values */
 	altdev->edac_dev_name = ecc_name;
-- 
2.42.0.411.g813d9a9188

[PATCH 06/10] EDAC/altera: Add suppress_bind_attrs so driver cannot get unbound

[Dinh Nguyen]

The EDAC driver should not be able to be removed from userspace because a
re-bind would clear out existing memory contents of an active running system.

With this change, remove the .remove functions because they will not
ever get called.

Fixes: 588cb03ea208 ("EDAC, altera: Add Arria10 L2 Cache ECC handling")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 25 ++++---------------------
 1 file changed, 4 insertions(+), 21 deletions(-)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index 09e040b244673..cd2c2d11214d4 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -457,15 +457,6 @@ static int altr_sdram_probe(struct platform_device *pdev)
 	return res;
 }
 
-static void altr_sdram_remove(struct platform_device *pdev)
-{
-	struct mem_ctl_info *mci = platform_get_drvdata(pdev);
-
-	edac_mc_del_mc(&pdev->dev);
-	edac_mc_free(mci);
-	platform_set_drvdata(pdev, NULL);
-}
-
 /*
  * If you want to suspend, need to disable EDAC by removing it
  * from the device tree or defconfig.
@@ -485,13 +476,13 @@ static const struct dev_pm_ops altr_sdram_pm_ops = {
 
 static struct platform_driver altr_sdram_edac_driver = {
 	.probe = altr_sdram_probe,
-	.remove = altr_sdram_remove,
 	.driver = {
 		.name = "altr_sdram_edac",
 #ifdef CONFIG_PM
 		.pm = &altr_sdram_pm_ops,
 #endif
 		.of_match_table = altr_sdram_ctrl_of_match,
+		.suppress_bind_attrs = true,
 	},
 };
 
@@ -521,6 +512,7 @@ static struct platform_driver altr_edac_driver = {
 	.driver = {
 		.name = "socfpga_ecc_manager",
 		.of_match_table = altr_edac_of_match,
+		.suppress_bind_attrs = true,
 	},
 };
 module_platform_driver(altr_edac_driver);
@@ -823,22 +815,12 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 	return res;
 }
 
-static void altr_edac_device_remove(struct platform_device *pdev)
-{
-	struct edac_device_ctl_info *dci = platform_get_drvdata(pdev);
-	struct altr_edac_device_dev *drvdata = dci->pvt_info;
-
-	debugfs_remove_recursive(drvdata->debugfs_dir);
-	edac_device_del_device(&pdev->dev);
-	edac_device_free_ctl_info(dci);
-}
-
 static struct platform_driver altr_edac_device_driver = {
 	.probe =  altr_edac_device_probe,
-	.remove = altr_edac_device_remove,
 	.driver = {
 		.name = "altr_edac_device",
 		.of_match_table = altr_edac_device_of_match,
+		.suppress_bind_attrs = true,
 	},
 };
 module_platform_driver(altr_edac_device_driver);
@@ -2253,6 +2235,7 @@ static struct platform_driver altr_edac_a10_driver = {
 	.driver = {
 		.name = "socfpga_a10_ecc_manager",
 		.of_match_table = altr_edac_a10_of_match,
+		.suppress_bind_attrs = true,
 	},
 };
 module_platform_driver(altr_edac_a10_driver);
-- 
2.42.0.411.g813d9a9188

[PATCH 08/10] EDAC/altera: Skip ECC memory init if already initialized

[Dinh Nguyen]

altr_init_memory_port() unconditionally triggered the HW memory-init
block, which clears the entire memory and ECC data. On a re-probe (e.g.
a probe deferred via -EPROBE_DEFER that only succeeds after the
protected peripheral is active, or a rebind) this would wipe live
peripheral memory. Skip the destructive init when the INITCOMPLETE
status bit is already set, so it runs only once.

Assisted-by: Cursor:claude-4.8-opus
Fixes: 1166fde93d5b ("EDAC, altera: Add Arria10 ECC memory init functions")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index d8d2adc9c39ac..5087aa5ae5612 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -930,6 +930,18 @@ static int __maybe_unused altr_init_memory_port(void __iomem *ioaddr, int port)
 		clear_mask = ALTR_A10_ECC_ERRPENA_MASK;
 	}
 
+	/*
+	 * The HW init block clears the entire memory and ECC data, so only
+	 * run it once. If initialization already completed (e.g. on an
+	 * earlier probe or in the bootloader), skip it so a re-probe caused
+	 * by deferred probing or a rebind does not wipe an active peripheral.
+	 */
+	if (ecc_test_bits(stat_mask, (ioaddr + ALTR_A10_ECC_INITSTAT_OFST))) {
+		/* Clear any pending ECC interrupts */
+		writel(clear_mask, (ioaddr + ALTR_A10_ECC_INTSTAT_OFST));
+		return 0;
+	}
+
 	ecc_set_bits(init_mask, (ioaddr + ALTR_A10_ECC_CTRL_OFST));
 	while (limit--) {
 		if (ecc_test_bits(stat_mask,
-- 
2.42.0.411.g813d9a9188

[PATCH 09/10] EDAC/altera: Detach sb_irq handler when db_irq setup fails

[Dinh Nguyen]

altr_edac_a10_probe() installs the sb_irq chained handler before fetching
db_irq. If platform_get_irq() for db_irq fails, the probe returned
without detaching the sb_irq handler, leaving it pointing at the
devm-freed edac struct. Unlink the sb_irq chained handler before
removing the IRQ domain and returning.

Assisted-by: Cursor:claude-4.8-opus
Fixes: d5fc9125566c ("EDAC, altera: Combine Stratix10 and Arria10 probe functions")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index 5087aa5ae5612..80ca9f511ac98 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -2209,6 +2209,8 @@ static int altr_edac_a10_probe(struct platform_device *pdev)
 	} else {
 		edac->db_irq = platform_get_irq(pdev, 1);
 		if (edac->db_irq < 0) {
+			irq_set_chained_handler_and_data(edac->sb_irq,
+							 NULL, NULL);
 			irq_domain_remove(edac->domain);
 			return edac->db_irq;
 		}
-- 
2.42.0.411.g813d9a9188

[PATCH 10/10] EDAC/altera: Fix error cleanup path in altr_edac_device_probe()

[Dinh Nguyen]

sashiko reports: Does the error cleanup path in
altr_edac_device_probe() free data before unregistering the interrupt
handler? If a failure occurs, dci is explicitly freed via
edac_device_free_ctl_info(dci) before the managed IRQ handler is unregistered
via devres_release_group(). If an interrupt fires in this window, could it
dereference the freed dci?

Clean up the fail1 path to always call devres_release_group() and
conditionally free the dci.

Fixes: c3eea1942a16 ("EDAC, altera: Add Altera L2 cache and OCRAM support")
Cc: stable@vger.kernel.org
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
---
 drivers/edac/altera_edac.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/edac/altera_edac.c b/drivers/edac/altera_edac.c
index 80ca9f511ac98..b6048a1884f43 100644
--- a/drivers/edac/altera_edac.c
+++ b/drivers/edac/altera_edac.c
@@ -701,7 +701,7 @@ MODULE_DEVICE_TABLE(of, altr_edac_device_of_match);
  */
 static int altr_edac_device_probe(struct platform_device *pdev)
 {
-	struct edac_device_ctl_info *dci;
+	struct edac_device_ctl_info *dci = NULL;
 	struct altr_edac_device_dev *drvdata;
 	struct resource *r;
 	int res = 0;
@@ -729,7 +729,7 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 		edac_printk(KERN_ERR, EDAC_DEVICE,
 			    "Unable to get mem resource\n");
 		res = -ENODEV;
-		goto fail;
+		goto fail1;
 	}
 
 	if (!devm_request_mem_region(&pdev->dev, r->start, resource_size(r),
@@ -737,7 +737,7 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 		edac_printk(KERN_ERR, EDAC_DEVICE,
 			    "%s:Error requesting mem region\n", ecc_name);
 		res = -EBUSY;
-		goto fail;
+		goto fail1;
 	}
 
 	dci = edac_device_alloc_ctl_info(sizeof(*drvdata), ecc_name,
@@ -747,7 +747,7 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 		edac_printk(KERN_ERR, EDAC_DEVICE,
 			    "%s: Unable to allocate EDAC device\n", ecc_name);
 		res = -ENOMEM;
-		goto fail;
+		goto fail1;
 	}
 
 	drvdata = dci->pvt_info;
@@ -806,9 +806,9 @@ static int altr_edac_device_probe(struct platform_device *pdev)
 	return 0;
 
 fail1:
-	edac_device_free_ctl_info(dci);
-fail:
 	devres_release_group(&pdev->dev, NULL);
+	if (dci)
+		edac_device_free_ctl_info(dci);
 	edac_printk(KERN_ERR, EDAC_DEVICE,
 		    "%s:Error setting up EDAC device: %d\n", ecc_name, res);
 
-- 
2.42.0.411.g813d9a9188