[PATCH 0/3] EDAC: Balance PCI device enablement in three Intel drivers

[PATCH 0/3] EDAC: Balance PCI device enablement in three Intel drivers

[Myeonghun Pak]

These drivers leave PCI enable references held on probe failure or removal.

Use managed PCI enablement for e752x. Keep explicit cleanup for i3000
and x38 because their module initialization can invoke probe outside
PCI driver binding. Each patch is independent.

These issues were identified during our ongoing static-analysis research
while reviewing kernel code.

Assisted-by: LLM

Myeonghun Pak (3):
  EDAC/e752x: Use managed PCI device enablement
  EDAC/i3000: Balance PCI device enablement
  EDAC/x38: Balance PCI device enablement

 drivers/edac/e752x_edac.c | 2 +-
 drivers/edac/i3000_edac.c | 4 ++++
 drivers/edac/x38_edac.c   | 4 ++++
 3 files changed, 9 insertions(+), 1 deletion(-)

[PATCH 1/3] EDAC/e752x: Use managed PCI device enablement

[Myeonghun Pak]

e752x_init_one() enables the PCI device without balancing the enable
reference on probe failure or removal. Use pcim_enable_device() so the
PCI device is disabled after failed probing and after driver removal.

This issue was identified during our ongoing static-analysis research
while reviewing kernel code.

Fixes: 806c35f5057a ("[PATCH] EDAC: drivers for AMD 76x and Intel E750x, E752x")
Cc: stable@vger.kernel.org
Assisted-by: LLM
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
---
 drivers/edac/e752x_edac.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/edac/e752x_edac.c b/drivers/edac/e752x_edac.c
--- a/drivers/edac/e752x_edac.c
+++ b/drivers/edac/e752x_edac.c
@@ -1387,7 +1387,7 @@ static int e752x_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	edac_dbg(0, "\n");
 
 	/* wake up and enable device */
-	if (pci_enable_device(pdev) < 0)
+	if (pcim_enable_device(pdev) < 0)
 		return -EIO;
 
 	return e752x_probe1(pdev, ent->driver_data);

Re: [PATCH 1/3] EDAC/e752x: Use managed PCI device enablement

[Borislav Petkov]

This one too?

If no objections, I'd orphan them all and stop taking patches for them.

On Mon, Sep 14, 2026 at 09:55:17PM -0400, Myeonghun Pak wrote:
> e752x_init_one() enables the PCI device without balancing the enable
> reference on probe failure or removal. Use pcim_enable_device() so the
> PCI device is disabled after failed probing and after driver removal.
> 
> This issue was identified during our ongoing static-analysis research
> while reviewing kernel code.
> 
> Fixes: 806c35f5057a ("[PATCH] EDAC: drivers for AMD 76x and Intel E750x, E752x")
> Cc: stable@vger.kernel.org
> Assisted-by: LLM
> Co-developed-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
> ---
>  drivers/edac/e752x_edac.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/edac/e752x_edac.c b/drivers/edac/e752x_edac.c
> --- a/drivers/edac/e752x_edac.c
> +++ b/drivers/edac/e752x_edac.c
> @@ -1387,7 +1387,7 @@ static int e752x_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  	edac_dbg(0, "\n");
>  
>  	/* wake up and enable device */
> -	if (pci_enable_device(pdev) < 0)
> +	if (pcim_enable_device(pdev) < 0)
>  		return -EIO;
>  
>  	return e752x_probe1(pdev, ent->driver_data);

-- 
Regards/Gruss,
    Boris.

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

RE: [PATCH 1/3] EDAC/e752x: Use managed PCI device enablement

[Luck, Tony]

> This one too?
>
> If no objections, I'd orphan them all and stop taking patches for them.

Agreed. Even if someone is still using these old systems, they aren't going to be running v7.x kernels on them.

-Tony

Re: [PATCH 1/3] EDAC/e752x: Use managed PCI device enablement

[Borislav Petkov]

On Wed, Sep 16, 2026 at 03:31:18PM +0000, Luck, Tony wrote:
> Agreed. Even if someone is still using these old systems, they aren't going
> to be running v7.x kernels on them.

Yeah, and it's not like we cannot resurrect those drivers. But for some of
them we're simply keeping them in shape and considering how there's no
feedback or anything else showing that at least someone is using them, we're
simply wasting effort.

Lemme mark them as orphaned and queue them - we'll see who cries.

Thx.

-- 
Regards/Gruss,
    Boris.

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

[PATCH 2/3] EDAC/i3000: Balance PCI device enablement

[Myeonghun Pak]

i3000_init_one() leaves the PCI enable reference held when probing fails
and when the controller is removed. Disable the device on probe failure
and after removing the EDAC controller. Keep explicit cleanup because
module initialization can call i3000_init_one() outside PCI driver binding.

This issue was identified during our ongoing static-analysis research
while reviewing kernel code.

Fixes: 535c6a53035d ("drivers/edac: new inte 30x0 MC driver")
Cc: stable@vger.kernel.org
Assisted-by: LLM
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
---
 drivers/edac/i3000_edac.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/edac/i3000_edac.c b/drivers/edac/i3000_edac.c
--- a/drivers/edac/i3000_edac.c
+++ b/drivers/edac/i3000_edac.c
@@ -461,6 +461,8 @@ static int i3000_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 		return -EIO;
 
 	rc = i3000_probe1(pdev, ent->driver_data);
+	if (rc)
+		pci_disable_device(pdev);
 	if (!mci_pdev)
 		mci_pdev = pci_dev_get(pdev);
 
@@ -480,7 +482,9 @@ static void i3000_remove_one(struct pci_dev *pdev)
 	if (!mci)
 		return;
 
 	edac_mc_free(mci);
+
+	pci_disable_device(pdev);
 }
 
 static const struct pci_device_id i3000_pci_tbl[] = {

Re: [PATCH 2/3] EDAC/i3000: Balance PCI device enablement

[Borislav Petkov]

Tony,

I have recently started marking obsolete EDAC drivers as orphaned because,
well, they don't deserve the effort.

How about this one?

On Mon, Sep 14, 2026 at 09:55:18PM -0400, Myeonghun Pak wrote:
> i3000_init_one() leaves the PCI enable reference held when probing fails
> and when the controller is removed. Disable the device on probe failure
> and after removing the EDAC controller. Keep explicit cleanup because
> module initialization can call i3000_init_one() outside PCI driver binding.
> 
> This issue was identified during our ongoing static-analysis research
> while reviewing kernel code.
> 
> Fixes: 535c6a53035d ("drivers/edac: new inte 30x0 MC driver")
> Cc: stable@vger.kernel.org
> Assisted-by: LLM
> Co-developed-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
> ---
>  drivers/edac/i3000_edac.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/edac/i3000_edac.c b/drivers/edac/i3000_edac.c
> --- a/drivers/edac/i3000_edac.c
> +++ b/drivers/edac/i3000_edac.c
> @@ -461,6 +461,8 @@ static int i3000_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  		return -EIO;
>  
>  	rc = i3000_probe1(pdev, ent->driver_data);
> +	if (rc)
> +		pci_disable_device(pdev);
>  	if (!mci_pdev)
>  		mci_pdev = pci_dev_get(pdev);
>  
> @@ -480,7 +482,9 @@ static void i3000_remove_one(struct pci_dev *pdev)
>  	if (!mci)
>  		return;
>  
>  	edac_mc_free(mci);
> +
> +	pci_disable_device(pdev);
>  }
>  
>  static const struct pci_device_id i3000_pci_tbl[] = {

-- 
Regards/Gruss,
    Boris.

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

Re: [PATCH 2/3] EDAC/i3000: Balance PCI device enablement

[Borislav Petkov]

On Tue, Sep 15, 2026 at 06:15:56PM -0700, Borislav Petkov wrote:
> Tony,
> 
> I have recently started marking obsolete EDAC drivers as orphaned because,
> well, they don't deserve the effort.
> 
> How about this one?
> 
> On Mon, Sep 14, 2026 at 09:55:18PM -0400, Myeonghun Pak wrote:
> > i3000_init_one() leaves the PCI enable reference held when probing fails
> > and when the controller is removed. Disable the device on probe failure
> > and after removing the EDAC controller. Keep explicit cleanup because
> > module initialization can call i3000_init_one() outside PCI driver binding.
> > 
> > This issue was identified during our ongoing static-analysis research
> > while reviewing kernel code.
> > 
> > Fixes: 535c6a53035d ("drivers/edac: new inte 30x0 MC driver")
> > Cc: stable@vger.kernel.org
> > Assisted-by: LLM
> > Co-developed-by: Ijae Kim <ae878000@gmail.com>
> > Signed-off-by: Ijae Kim <ae878000@gmail.com>
> > Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
> > ---
> >  drivers/edac/i3000_edac.c | 4 ++++
> >  1 file changed, 4 insertions(+)

Oh, that one is an orphan already:

./scripts/get_maintainer.pl -f drivers/edac/i3000_edac.c
EDAC-I3000 status: Orphan

You can save your energy doing fixes for it.

-- 
Regards/Gruss,
    Boris.

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

[PATCH 3/3] EDAC/x38: Balance PCI device enablement

[Myeonghun Pak]

x38_init_one() leaves the PCI enable reference held when probing fails
and when the controller is removed. Disable the device on probe failure
and after removing the EDAC controller. Keep explicit cleanup because
module initialization can call x38_init_one() outside PCI driver binding.

This issue was identified during our ongoing static-analysis research
while reviewing kernel code.

Fixes: df8bc08c192f ("edac x38: new MC driver module")
Cc: stable@vger.kernel.org
Assisted-by: LLM
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
---
 drivers/edac/x38_edac.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/edac/x38_edac.c b/drivers/edac/x38_edac.c
--- a/drivers/edac/x38_edac.c
+++ b/drivers/edac/x38_edac.c
@@ -423,6 +423,8 @@ static int x38_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 		return -EIO;
 
 	rc = x38_probe1(pdev, ent->driver_data);
+	if (rc)
+		pci_disable_device(pdev);
 	if (!mci_pdev)
 		mci_pdev = pci_dev_get(pdev);
 
@@ -439,9 +441,11 @@ static void x38_remove_one(struct pci_dev *pdev)
 	if (!mci)
 		return;
 
 	iounmap(mci->pvt_info);
 
 	edac_mc_free(mci);
+
+	pci_disable_device(pdev);
 }
 
 static const struct pci_device_id x38_pci_tbl[] = {

Re: [PATCH 3/3] EDAC/x38: Balance PCI device enablement

[Borislav Petkov]

This one looks old too.

On Mon, Sep 14, 2026 at 09:55:19PM -0400, Myeonghun Pak wrote:
> x38_init_one() leaves the PCI enable reference held when probing fails
> and when the controller is removed. Disable the device on probe failure
> and after removing the EDAC controller. Keep explicit cleanup because
> module initialization can call x38_init_one() outside PCI driver binding.
> 
> This issue was identified during our ongoing static-analysis research
> while reviewing kernel code.
> 
> Fixes: df8bc08c192f ("edac x38: new MC driver module")
> Cc: stable@vger.kernel.org
> Assisted-by: LLM
> Co-developed-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Ijae Kim <ae878000@gmail.com>
> Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
> ---
>  drivers/edac/x38_edac.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/edac/x38_edac.c b/drivers/edac/x38_edac.c
> --- a/drivers/edac/x38_edac.c
> +++ b/drivers/edac/x38_edac.c
> @@ -423,6 +423,8 @@ static int x38_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  		return -EIO;
>  
>  	rc = x38_probe1(pdev, ent->driver_data);
> +	if (rc)
> +		pci_disable_device(pdev);
>  	if (!mci_pdev)
>  		mci_pdev = pci_dev_get(pdev);
>  
> @@ -439,9 +441,11 @@ static void x38_remove_one(struct pci_dev *pdev)
>  	if (!mci)
>  		return;
>  
>  	iounmap(mci->pvt_info);
>  
>  	edac_mc_free(mci);
> +
> +	pci_disable_device(pdev);
>  }
>  
>  static const struct pci_device_id x38_pci_tbl[] = {
> 

-- 
Regards/Gruss,
    Boris.

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

[PATCH] MAINTAINERS: Orphan drivers/edac/x38_edac.c

[Borislav Petkov]

From: "Borislav Petkov (AMD)" <bp@alien8.de>
Date: Wed, 16 Sep 2026 12:25:37 -0700

This driver has been receiving odd fixes and API adjustments without
anyone actually using it, TTBOMK. So orphan it.

Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
---
 MAINTAINERS | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 4f57a3234fb8..9497c4e5b1f9 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -9546,6 +9546,11 @@ L:	linux-edac@vger.kernel.org
 S:	Odd Fixes
 F:	drivers/edac/ti_edac.c
 
+EDAC-X38
+L:	linux-edac@vger.kernel.org
+S:	Orphan
+F:	drivers/edac/x38_edac.c
+
 EDIROL UA-101/UA-1000 DRIVER
 M:	Clemens Ladisch <clemens@ladisch.de>
 L:	linux-sound@vger.kernel.org
-- 
2.53.0


-- 
Regards/Gruss,
    Boris.

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