Skip to contentSkip to conversation

drm/i915: Guard against NULL driver_data in i915_pci_probe()

PATCH Superseded Deepanshu Kartikey 5 participants 6 messages Aug 9 – Aug 13

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/i915: Guard against NULL driver_data in i915_pci_probe()

pci_match_device() can return the dummy pci_device_id_any entry when a device is force-bound via sysfs driver_override, in which case ->driver_data is unset (NULL). i915_pci_probe() casts it to struct intel_device_info * unconditionally and dereferences intel_info->require_force_probe, causing a NULL-ptr-deref.

Reported-by [email protected]
Fixes 42f5551d2769 ("drm/i915: Split out the PCI driver interface to i915_pci.c")
---
 drivers/gpu/drm/i915/i915_pci.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/i915/i915_pci.c +3 -0
index 82415af47d54..2f03f95945f1 100644
@@ -958,6 +958,9 @@ static int i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		(struct intel_device_info *) ent->driver_data;
 	int err;
 
+	if (!intel_info)
+		return -ENODEV;
+
 	if (intel_info->require_force_probe && !id_forced(pdev->device)) {
 		dev_info(&pdev->dev,
 			 "Your graphics device %04x is not properly supported by i915 in this\n"
Signature · 1 line
2.43.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] The exact same NULL-pointer dereference vulnerability fixed in drm/i915 also exists in the sibling driver drm/xe. --

--- Patch [1]: [PATCH] drm/i915: Guard against NULL driver_data in i915_pci_probe() --- commit 88147f0f2fd48d4b9d650095dee423ce2e3c8939
Author: Deepanshu Kartikey <[email protected]>

drm/i915: Guard against NULL driver_data in i915_pci_probe()

This commit fixes a NULL pointer dereference in i915_pci_probe() when a device is force-bound via sysfs driver_override, which results in a NULL driver_data being returned by pci_match_device().

drivers/gpu/drm/i915/i915_pci.c static int i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) · line 962

Quoted 10 lines
drivers/gpu/drm/i915/i915_pci.c +2 -0
index 82415af47d540..2f03f95945f1b 100644
@@ -958,6 +958,9 @@ static int i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		(struct intel_device_info *) ent->driver_data;
 	int err;
 
+	if (!intel_info)
+		return -ENODEV;

[Severity: High]
This is a pre-existing issue, but does the exact same vulnerability exist in the sibling drm/xe driver?

When a device is force-bound via the sysfs driver_override interface, xe_pci_probe() in drivers/gpu/drm/xe/xe_pci.c will also receive a dummy entry where ent->driver_data is NULL. This is cast to desc and passed directly to find_subplatform():

drivers/gpu/drm/xe/xe_pci.c:xe_pci_probe() {
    ...
    const struct xe_device_desc *desc = (const void *)ent->driver_data;
    ...
    subplatform_desc = find_subplatform(desc, pdev->device);
    ...
}

Inside find_subplatform(), this pointer is unconditionally dereferenced:

drivers/gpu/drm/xe/xe_pci.c:find_subplatform() {
    ...
    for (sp = desc->subplatforms; sp && sp->subplatform; sp++)
    ...
}

Should a similar guard be added to xe_pci_probe() to prevent a kernel panic there as well?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#2 lore raw

✗ LGCI.VerificationFailed: failure for drm/i915: Guard against NULL driver_data in i915_pci_probe()

In reply to Deepanshu Kartikey
== Series Details ==

== Series Details ==

Series: drm/i915: Guard against NULL driver_data in i915_pci_probe()
URL : https://patchwork.freedesktop.org/series/171944/
State : failure

== Summary ==

Series author address '[email protected]' is not on the allowlist, which prevents CI from being automatically triggered. If you want CI to run for this series, ask Patchwork project owners to click 'retest' on the series in Patchwork. Exception occurred during validation, bailing out!
Build URL: http://gfx-ci.igk.intel.com:8080/job/CI_PW_kernel/183932/ (on built-in)

#3 lore raw
In reply to Deepanshu Kartikey

Hi Deepanshu,

Quoted 10 lines · Deepanshu Kartikey

pci_match_device() can return the dummy pci_device_id_any entry when a device is force-bound via sysfs driver_override, in which case ->driver_data is unset (NULL). i915_pci_probe() casts it to struct intel_device_info * unconditionally and dereferences intel_info->require_force_probe, causing a NULL-ptr-deref.

Reported-by [email protected]
Fixes 42f5551d2769 ("drm/i915: Split out the PCI driver interface to i915_pci.c")

This "Fixes" tag should point to the commit that breaks stuff,
I believe, so what you are looking for is
7ef5ef5cdead ("drm/i915: add force_probe module parameter to replace alpha_support") since this patch introduced that unconditional dereference.

Quoted 22 lines
---
 drivers/gpu/drm/i915/i915_pci.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/i915/i915_pci.c +3 -0
index 82415af47d54..2f03f95945f1 100644
@@ -958,6 +958,9 @@ static int i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		(struct intel_device_info *) ent->driver_data;
 	int err;
 
+	if (!intel_info)
+		return -ENODEV;
+
 	if (intel_info->require_force_probe && !id_forced(pdev->device)) {
 		dev_info(&pdev->dev,
 			 "Your graphics device %04x is not properly supported by i915 in this\n"
Signature · 1 line
2.43.0
Signature · 2 lines
Best Regards,
Krzysztof
maintainer
#4 lore raw
Quoted 9 lines · Krzysztof Karas

Hi Deepanshu,

Quoted 6 lines · Deepanshu Kartikey

pci_match_device() can return the dummy pci_device_id_any entry when a device is force-bound via sysfs driver_override, in which case ->driver_data is unset (NULL). i915_pci_probe() casts it to struct intel_device_info * unconditionally and dereferences intel_info->require_force_probe, causing a NULL-ptr-deref.

Quoted 5 lines
Quoted 4 lines
Reported-by [email protected]
Fixes 42f5551d2769 ("drm/i915: Split out the PCI driver interface to i915_pci.c")

This "Fixes" tag should point to the commit that breaks stuff,

In general, agreed...

Quoted 3 lines

I believe, so what you are looking for is
7ef5ef5cdead ("drm/i915: add force_probe module parameter to replace alpha_support") since this patch introduced that unconditional dereference.

...but that didn't introduce the unconditional reference. I think it was always there. We've always expected .driver_data to point at whatever we have specified in MODULE_DEVICE_TABLE().

I'd just slap a Cc: stable on it without Fixes.

BR,
Jani.

Quoted 23 lines
Quoted 23 lines
---
 drivers/gpu/drm/i915/i915_pci.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/i915/i915_pci.c +3 -0
index 82415af47d54..2f03f95945f1 100644
@@ -958,6 +958,9 @@ static int i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 		(struct intel_device_info *) ent->driver_data;
 	int err;
 
+	if (!intel_info)
+		return -ENODEV;
+
 	if (intel_info->require_force_probe && !id_forced(pdev->device)) {
 		dev_info(&pdev->dev,
 			 "Your graphics device %04x is not properly supported by i915 in this\n"
Signature · 1 line
2.43.0
Signature · 1 line
Jani Nikula, Intel
#5 lore raw
Quoted 47 lines · Jani Nikula
Quoted 9 lines · Krzysztof Karas

Hi Deepanshu,

Quoted 6 lines · Deepanshu Kartikey

pci_match_device() can return the dummy pci_device_id_any entry when a device is force-bound via sysfs driver_override, in which case ->driver_data is unset (NULL). i915_pci_probe() casts it to struct intel_device_info * unconditionally and dereferences intel_info->require_force_probe, causing a NULL-ptr-deref.

Quoted 5 lines
Quoted 4 lines
Reported-by [email protected]
Fixes 42f5551d2769 ("drm/i915: Split out the PCI driver interface to i915_pci.c")

This "Fixes" tag should point to the commit that breaks stuff,

In general, agreed...

Quoted 3 lines

I believe, so what you are looking for is
7ef5ef5cdead ("drm/i915: add force_probe module parameter to replace alpha_support") since this patch introduced that unconditional dereference.

...but that didn't introduce the unconditional reference. I think it was always there. We've always expected .driver_data to point at whatever we have specified in MODULE_DEVICE_TABLE().

I'd just slap a Cc: stable on it without Fixes.

BR,
Jani.

Quoted 14 lines
Quoted 14 lines
---
 drivers/gpu/drm/i915/i915_pci.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/i915/i915_pci.c +0 -0
index 82415af47d54..2f03f95945f1 100644
@@ -958,6 +958,9 @@ static int i915_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
             (struct intel_device_info *) ent->driver_data;
     int err;
 
Quoted 12 lines
Quoted 9 lines
Quoted 9 lines
drivers/gpu/drm/i915/i915_pci.c +3 -1
+    if (!intel_info)
+            return -ENODEV;
+
     if (intel_info->require_force_probe && !id_forced(pdev->device)) {
             dev_info(&pdev->dev,
                      "Your graphics device %04x is not properly supported by i915 in this\n"
--

2.43.0

--
Jani Nikula, Intel

Thanks for the review. I have sent patch v2.

Deepanshu