Skip to contentSkip to conversation

drm/panthor: validate firmware interface structure sizes

PATCH In review Osama Abdelkader 3 participants 3 messages Jul 20 – Jul 29

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/panthor: validate firmware interface structure sizes

iface_fw_to_cpu_addr() only checks that the firmware-provided MCU virtual address points inside the shared section. The returned pointer is later used as a full firmware interface structure, so accepting an address near the end of the shared section can still lead to out-of-bounds accesses.

Pass the expected object size to iface_fw_to_cpu_addr() and reject ranges that do not fit entirely in the shared section.

Fixes 2718d91816ee ("drm/panthor: Add the FW logical block")
---
 drivers/gpu/drm/panthor/panthor_fw.c | 30 ++++++++++++++++++---------
 1 file changed, 20 insertions(+), 10 deletions(-)
drivers/gpu/drm/panthor/panthor_fw.c +24 -12
index 9d0e1fafdce2..eee2bc7e8541 100644
@@ -870,18 +870,24 @@ static int panthor_fw_load(struct panthor_device *ptdev)
  * iface_fw_to_cpu_addr() - Turn an MCU address into a CPU address
  * @ptdev: Device.
  * @mcu_va: MCU address.
+ * @size: Size of the object pointed to by @mcu_va.
  *
- * Return: NULL if the address is not part of the shared section, non-NULL otherwise.
+ * Return: NULL if the object is not part of the shared section, non-NULL otherwise.
  */
-static void *iface_fw_to_cpu_addr(struct panthor_device *ptdev, u32 mcu_va)
+static void *iface_fw_to_cpu_addr(struct panthor_device *ptdev, u32 mcu_va, size_t size)
 {
 	u64 shared_mem_start = panthor_kernel_bo_gpuva(ptdev->fw->shared_section->mem);
-	u64 shared_mem_end = shared_mem_start +
-			     panthor_kernel_bo_size(ptdev->fw->shared_section->mem);
-	if (mcu_va < shared_mem_start || mcu_va >= shared_mem_end)
+	size_t shared_mem_size = panthor_kernel_bo_size(ptdev->fw->shared_section->mem);
+	u64 offset;
+
+	if (mcu_va < shared_mem_start)
+		return NULL;
+
+	offset = mcu_va - shared_mem_start;
+	if (offset > shared_mem_size || size > shared_mem_size - offset)
 		return NULL;
 
-	return ptdev->fw->shared_section->mem->kmap + (mcu_va - shared_mem_start);
+	return ptdev->fw->shared_section->mem->kmap + offset;
 }
 
 static int panthor_init_cs_iface(struct panthor_device *ptdev,
@@ -906,8 +912,10 @@ static int panthor_init_cs_iface(struct panthor_device *ptdev,
 
 	spin_lock_init(&cs_iface->lock);
 	cs_iface->control = ptdev->fw->shared_section->mem->kmap + iface_offset;
-	cs_iface->input = iface_fw_to_cpu_addr(ptdev, cs_iface->control->input_va);
-	cs_iface->output = iface_fw_to_cpu_addr(ptdev, cs_iface->control->output_va);
+	cs_iface->input = iface_fw_to_cpu_addr(ptdev, cs_iface->control->input_va,
+					       sizeof(*cs_iface->input));
+	cs_iface->output = iface_fw_to_cpu_addr(ptdev, cs_iface->control->output_va,
+						sizeof(*cs_iface->output));
 
 	if (!cs_iface->input || !cs_iface->output) {
 		drm_err(&ptdev->base, "Invalid stream control interface input/output VA");
@@ -957,8 +965,10 @@ static int panthor_init_csg_iface(struct panthor_device *ptdev,
 
 	spin_lock_init(&csg_iface->lock);
 	csg_iface->control = ptdev->fw->shared_section->mem->kmap + iface_offset;
-	csg_iface->input = iface_fw_to_cpu_addr(ptdev, csg_iface->control->input_va);
-	csg_iface->output = iface_fw_to_cpu_addr(ptdev, csg_iface->control->output_va);
+	csg_iface->input = iface_fw_to_cpu_addr(ptdev, csg_iface->control->input_va,
+						sizeof(*csg_iface->input));
+	csg_iface->output = iface_fw_to_cpu_addr(ptdev, csg_iface->control->output_va,
+						 sizeof(*csg_iface->output));
 
 	if (csg_iface->control->stream_num < MIN_CS_PER_CSG ||
 	    csg_iface->control->stream_num > MAX_CS_PER_CSG)
@@ -1015,8 +1025,10 @@ static int panthor_fw_init_ifaces(struct panthor_device *ptdev)
 		return -EINVAL;
 	}
 
-	glb_iface->input = iface_fw_to_cpu_addr(ptdev, glb_iface->control->input_va);
-	glb_iface->output = iface_fw_to_cpu_addr(ptdev, glb_iface->control->output_va);
+	glb_iface->input = iface_fw_to_cpu_addr(ptdev, glb_iface->control->input_va,
+						sizeof(*glb_iface->input));
+	glb_iface->output = iface_fw_to_cpu_addr(ptdev, glb_iface->control->output_va,
+						 sizeof(*glb_iface->output));
 	if (!glb_iface->input || !glb_iface->output) {
 		drm_err(&ptdev->base, "Invalid global control interface input/output VA");
 		return -EINVAL;
Signature · 1 line
2.43.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Pre-existing issues:
- [Critical] Time-of-Check to Time-of-Use (TOCTOU) vulnerability in firmware interface initialization allows kernel memory corruption.
- [High] Integer overflow and incorrect structure size in firmware bounds checks on 32-bit architectures.
- [High] NULL pointer dereference when firmware provides a 0-sized shared section.
--

--- Patch [1]: [PATCH] drm/panthor: validate firmware interface structure sizes --- commit a126d5ded05975c89ebf69bcb6a208082b1f032d
Author: Osama Abdelkader <[email protected]>

drm/panthor: validate firmware interface structure sizes

iface_fw_to_cpu_addr() only checks that the firmware-provided MCU virtual address points inside the shared section. The returned pointer is later used as a full firmware interface structure, so accepting an address near the end of the shared section can still lead to out-of-bounds accesses.

Pass the expected object size to iface_fw_to_cpu_addr() and reject ranges that do not fit entirely in the shared section.

Quoted 4 lines
drivers/gpu/drm/panthor/panthor_fw.c +0 -0
index 90f59d782a809..1697305c7b0e9 100644

[ ... ]

drivers/gpu/drm/panthor/panthor_fw.c static int panthor_init_cs_iface(struct panthor_device *ptdev, · line 897

Quoted 3 lines
drivers/gpu/drm/panthor/panthor_fw.c +0 -0
@@ -890,8 +896,10 @@ static int panthor_init_cs_iface(struct panthor_device *ptdev,
 
 	spin_lock_init(&cs_iface->lock);

[Severity: High]
This is a pre-existing issue, but could the bounds check earlier in panthor_init_cs_iface() be vulnerable to a 32-bit integer overflow?

panthor_fw.c:panthor_init_cs_iface() {
    ...
    if (iface_offset + sizeof(*cs_iface) >= shared_section_sz)
        return -EINVAL;
    ...
}

If a malicious firmware provides crafted strides, could iface_offset wrap around near U32_MAX on 32-bit architectures, allowing it to bypass this check while still extending beyond shared_section_sz?

Also, does this check incorrectly use the size of the host wrapper struct sizeof(*cs_iface) instead of the actual firmware struct size, allowing an invalid iface_offset to pass the check if the host struct is smaller?

Quoted 7 lines
drivers/gpu/drm/panthor/panthor_fw.c +4 -2
 	cs_iface->control = ptdev->fw->shared_section->mem->kmap + iface_offset;
-	cs_iface->input = iface_fw_to_cpu_addr(ptdev, cs_iface->control->input_va);
-	cs_iface->output = iface_fw_to_cpu_addr(ptdev, cs_iface->control->output_va);
+	cs_iface->input = iface_fw_to_cpu_addr(ptdev, cs_iface->control->input_va,
+					       sizeof(*cs_iface->input));
+	cs_iface->output = iface_fw_to_cpu_addr(ptdev, cs_iface->control->output_va,
+						sizeof(*cs_iface->output));

[ ... ]

drivers/gpu/drm/panthor/panthor_fw.c static int panthor_init_csg_iface(struct panthor_device *ptdev, · line 958

Quoted 13 lines
drivers/gpu/drm/panthor/panthor_fw.c +4 -2
@@ -941,8 +949,10 @@ static int panthor_init_csg_iface(struct panthor_device *ptdev,
 
 	spin_lock_init(&csg_iface->lock);
 	csg_iface->control = ptdev->fw->shared_section->mem->kmap + iface_offset;
-	csg_iface->input = iface_fw_to_cpu_addr(ptdev, csg_iface->control->input_va);
-	csg_iface->output = iface_fw_to_cpu_addr(ptdev, csg_iface->control->output_va);
+	csg_iface->input = iface_fw_to_cpu_addr(ptdev, csg_iface->control->input_va,
+						sizeof(*csg_iface->input));
+	csg_iface->output = iface_fw_to_cpu_addr(ptdev, csg_iface->control->output_va,
+						 sizeof(*csg_iface->output));
 
 	if (csg_iface->control->stream_num < MIN_CS_PER_CSG ||
 	    csg_iface->control->stream_num > MAX_CS_PER_CSG)

[Severity: Critical]
This is a pre-existing issue, but is there a Time-of-Check to Time-of-Use (TOCTOU) vulnerability here and in panthor_fw_init_ifaces()?

Since panthor_fw_start() starts the GPU MCU before panthor_fw_init_ifaces() is called, could compromised GPU firmware mutate stream_num (or group_num) directly in shared memory after this bounds check?

If the compiler reloads the limit from memory on each iteration of the subsequent loop:

    for (i = 0; i < csg_iface->control->stream_num; i++) {
        int ret = panthor_init_cs_iface(ptdev, csg_idx, i);
        ...
    }

could the loop overrun and index out of bounds, overwriting adjacent kernel structures?

[ ... ]

drivers/gpu/drm/panthor/panthor_fw.c static int panthor_fw_init_ifaces(struct panthor_device *ptdev) · line 1003

Quoted 6 lines
drivers/gpu/drm/panthor/panthor_fw.c +0 -2
@@ -999,8 +1009,10 @@ static int panthor_fw_init_ifaces(struct panthor_device *ptdev)
 		return -EINVAL;
 	}
 
-	glb_iface->input = iface_fw_to_cpu_addr(ptdev, glb_iface->control->input_va);
-	glb_iface->output = iface_fw_to_cpu_addr(ptdev, glb_iface->control->output_va);

[Severity: High]
This isn't a bug introduced by this patch, but earlier in this function could we dereference a NULL pointer if the firmware provides a 0-sized shared section?

In panthor_fw_load_section_entry(), if the firmware provides a CSF_MCU_SHARED_REGION_START section header where hdr.va.end == hdr.va.start, the memory allocation is skipped:

panthor_fw.c:panthor_fw_load_section_entry() {
    ...
    section_size = hdr.va.end - hdr.va.start;
    if (section_size) {
        ...
    }
    ptdev->fw->shared_section = section;
    ...
}

This leaves section->mem as NULL. When panthor_fw_init_ifaces() runs, it checks:

    if (!ptdev->fw->shared_section->mem->kmap)

Will this unconditionally dereference mem when it is NULL, causing a kernel panic during device probe?

Quoted 7 lines
drivers/gpu/drm/panthor/panthor_fw.c +4 -0
+	glb_iface->input = iface_fw_to_cpu_addr(ptdev, glb_iface->control->input_va,
+						sizeof(*glb_iface->input));
+	glb_iface->output = iface_fw_to_cpu_addr(ptdev, glb_iface->control->output_va,
+						 sizeof(*glb_iface->output));
 	if (!glb_iface->input || !glb_iface->output) {
 		drm_err(&ptdev->base, "Invalid global control interface input/output VA");
 		return -EINVAL;
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
maintainer Reviewed-by on v1 1/1
#2 lore raw
In reply to Osama Abdelkader
Quoted 11 lines · Osama Abdelkader

iface_fw_to_cpu_addr() only checks that the firmware-provided MCU virtual address points inside the shared section. The returned pointer is later used as a full firmware interface structure, so accepting an address near the end of the shared section can still lead to out-of-bounds accesses.

Pass the expected object size to iface_fw_to_cpu_addr() and reject ranges that do not fit entirely in the shared section.

Fixes 2718d91816ee ("drm/panthor: Add the FW logical block")

I'll push this to drm-misc-fixes.

Thanks,
Steve

Quoted 78 lines
---
 drivers/gpu/drm/panthor/panthor_fw.c | 30 ++++++++++++++++++---------
 1 file changed, 20 insertions(+), 10 deletions(-)
drivers/gpu/drm/panthor/panthor_fw.c +24 -12
index 9d0e1fafdce2..eee2bc7e8541 100644
@@ -870,18 +870,24 @@ static int panthor_fw_load(struct panthor_device *ptdev)
  * iface_fw_to_cpu_addr() - Turn an MCU address into a CPU address
  * @ptdev: Device.
  * @mcu_va: MCU address.
+ * @size: Size of the object pointed to by @mcu_va.
  *
- * Return: NULL if the address is not part of the shared section, non-NULL otherwise.
+ * Return: NULL if the object is not part of the shared section, non-NULL otherwise.
  */
-static void *iface_fw_to_cpu_addr(struct panthor_device *ptdev, u32 mcu_va)
+static void *iface_fw_to_cpu_addr(struct panthor_device *ptdev, u32 mcu_va, size_t size)
 {
 	u64 shared_mem_start = panthor_kernel_bo_gpuva(ptdev->fw->shared_section->mem);
-	u64 shared_mem_end = shared_mem_start +
-			     panthor_kernel_bo_size(ptdev->fw->shared_section->mem);
-	if (mcu_va < shared_mem_start || mcu_va >= shared_mem_end)
+	size_t shared_mem_size = panthor_kernel_bo_size(ptdev->fw->shared_section->mem);
+	u64 offset;
+
+	if (mcu_va < shared_mem_start)
+		return NULL;
+
+	offset = mcu_va - shared_mem_start;
+	if (offset > shared_mem_size || size > shared_mem_size - offset)
 		return NULL;
 
-	return ptdev->fw->shared_section->mem->kmap + (mcu_va - shared_mem_start);
+	return ptdev->fw->shared_section->mem->kmap + offset;
 }
 
 static int panthor_init_cs_iface(struct panthor_device *ptdev,
@@ -906,8 +912,10 @@ static int panthor_init_cs_iface(struct panthor_device *ptdev,
 
 	spin_lock_init(&cs_iface->lock);
 	cs_iface->control = ptdev->fw->shared_section->mem->kmap + iface_offset;
-	cs_iface->input = iface_fw_to_cpu_addr(ptdev, cs_iface->control->input_va);
-	cs_iface->output = iface_fw_to_cpu_addr(ptdev, cs_iface->control->output_va);
+	cs_iface->input = iface_fw_to_cpu_addr(ptdev, cs_iface->control->input_va,
+					       sizeof(*cs_iface->input));
+	cs_iface->output = iface_fw_to_cpu_addr(ptdev, cs_iface->control->output_va,
+						sizeof(*cs_iface->output));
 
 	if (!cs_iface->input || !cs_iface->output) {
 		drm_err(&ptdev->base, "Invalid stream control interface input/output VA");
@@ -957,8 +965,10 @@ static int panthor_init_csg_iface(struct panthor_device *ptdev,
 
 	spin_lock_init(&csg_iface->lock);
 	csg_iface->control = ptdev->fw->shared_section->mem->kmap + iface_offset;
-	csg_iface->input = iface_fw_to_cpu_addr(ptdev, csg_iface->control->input_va);
-	csg_iface->output = iface_fw_to_cpu_addr(ptdev, csg_iface->control->output_va);
+	csg_iface->input = iface_fw_to_cpu_addr(ptdev, csg_iface->control->input_va,
+						sizeof(*csg_iface->input));
+	csg_iface->output = iface_fw_to_cpu_addr(ptdev, csg_iface->control->output_va,
+						 sizeof(*csg_iface->output));
 
 	if (csg_iface->control->stream_num < MIN_CS_PER_CSG ||
 	    csg_iface->control->stream_num > MAX_CS_PER_CSG)
@@ -1015,8 +1025,10 @@ static int panthor_fw_init_ifaces(struct panthor_device *ptdev)
 		return -EINVAL;
 	}
 
-	glb_iface->input = iface_fw_to_cpu_addr(ptdev, glb_iface->control->input_va);
-	glb_iface->output = iface_fw_to_cpu_addr(ptdev, glb_iface->control->output_va);
+	glb_iface->input = iface_fw_to_cpu_addr(ptdev, glb_iface->control->input_va,
+						sizeof(*glb_iface->input));
+	glb_iface->output = iface_fw_to_cpu_addr(ptdev, glb_iface->control->output_va,
+						 sizeof(*glb_iface->output));
 	if (!glb_iface->input || !glb_iface->output) {
 		drm_err(&ptdev->base, "Invalid global control interface input/output VA");
 		return -EINVAL;