mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH v2 0/2] drm/imagination: Fix leaks in HWRT data set handling
@ 2026-11-21 15:47 Suyuan Wang via B4 Relay
  2026-11-21 15:47 ` [PATCH v2 1/2] drm/imagination: Destroy the common FW object when HWRT data setup fails Suyuan Wang via B4 Relay
  2026-11-21 15:47 ` [PATCH v2 2/2] drm/imagination: Don't truncate the layer count in max_rts Suyuan Wang via B4 Relay
  0 siblings, 2 replies; 3+ messages in thread
From: Suyuan Wang via B4 Relay @ 2026-11-21 15:47 UTC (permalink / raw)
  To: alessio.belle, luigi.santivetti
  Cc: donald.robson, mripard, dri-devel, maarten.lankhorst, airlied,
	simona, tzimmermann, linux-kernel, imagination

Patch 1 is v1 with the Fixes tag corrected: the error path that skips
hwrt_fini_common_fw_structure() came with f1f55ed3ffe4, not with the
commit that added the ioctl (thanks to the Sashiko review).

Patch 2 fixes the other leak that review pointed out: the 32-bit layer
count is truncated into the 16-bit max_rts, so the SRTC and RAA objects
allocated for layers > 1 can be skipped on destroy.

Found with a static analysis tool and checked by hand. Compile tested on
arm64 (W=1, all files that include pvr_hwrt.h); I don't have the
hardware.

Changes in v2:
- 1/2: Fixes tag now points to f1f55ed3ffe4. No code change.
- 2/2: new.

v1: https://lore.kernel.org/all/20261121-pvr-hwrt-common-v1-1-784ce45b782a@gmail.com/
---
Suyuan Wang (2):
      drm/imagination: Destroy the common FW object when HWRT data setup fails
      drm/imagination: Don't truncate the layer count in max_rts

 drivers/gpu/drm/imagination/pvr_hwrt.c | 2 ++
 drivers/gpu/drm/imagination/pvr_hwrt.h | 2 +-
 2 files changed, 3 insertions(+), 1 deletion(-)



^ permalink raw reply	[flat|nested] 3+ messages in thread

* [PATCH v2 1/2] drm/imagination: Destroy the common FW object when HWRT data setup fails
  2026-11-21 15:47 [PATCH v2 0/2] drm/imagination: Fix leaks in HWRT data set handling Suyuan Wang via B4 Relay
@ 2026-11-21 15:47 ` Suyuan Wang via B4 Relay
  2026-11-21 15:47 ` [PATCH v2 2/2] drm/imagination: Don't truncate the layer count in max_rts Suyuan Wang via B4 Relay
  1 sibling, 0 replies; 3+ messages in thread
From: Suyuan Wang via B4 Relay @ 2026-11-21 15:47 UTC (permalink / raw)
  To: alessio.belle, luigi.santivetti
  Cc: donald.robson, mripard, dri-devel, maarten.lankhorst, airlied,
	simona, tzimmermann, linux-kernel, imagination

From: Suyuan Wang <stevewong@hust.edu.cn>

If hwrt_data_init_fw_structure() fails in pvr_hwrt_dataset_create(),
the error path frees the per-data FW structures and then jumps straight
to err_fini_kernel_structure, skipping hwrt_fini_common_fw_structure().
The common FW object created just before is never destroyed, and its
init callback data is the hwrt that is then freed:

	err = pvr_fw_object_create(pvr_dev, ...,
				   hwrtdata_common_init, hwrt,
				   &hwrt->common_fw_obj);

Besides leaking the object, this leaves it on the device's FW object
list, so a later pvr_fw_hard_reset() calls hwrtdata_common_init() with
the freed hwrt, and pvr_fw_fini() warns that the list is not empty.

Destroy the common FW object on that path, in the same order as
pvr_hwrt_dataset_release().

Found with a static analysis tool and checked by hand.

Fixes: f1f55ed3ffe4 ("drm/imagination: Fixed oops when misusing ioctl CREATE_HWRT_DATASET")
Cc: stable@vger.kernel.org
Co-developed-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Suyuan Wang <stevewong@hust.edu.cn>
---
 drivers/gpu/drm/imagination/pvr_hwrt.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/imagination/pvr_hwrt.c b/drivers/gpu/drm/imagination/pvr_hwrt.c
index 3f7c4346c52c..aa25f6adaa3b 100644
--- a/drivers/gpu/drm/imagination/pvr_hwrt.c
+++ b/drivers/gpu/drm/imagination/pvr_hwrt.c
@@ -483,6 +483,8 @@ pvr_hwrt_dataset_create(struct pvr_file *pvr_file,
 	while (--i >= 0)
 		hwrt_data_fini_fw_structure(hwrt, i);
 
+	hwrt_fini_common_fw_structure(hwrt);
+
 err_fini_kernel_structure:
 	hwrt_fini_kernel_structure(hwrt);
 

-- 
2.34.1



^ permalink raw reply	[flat|nested] 3+ messages in thread

* [PATCH v2 2/2] drm/imagination: Don't truncate the layer count in max_rts
  2026-11-21 15:47 [PATCH v2 0/2] drm/imagination: Fix leaks in HWRT data set handling Suyuan Wang via B4 Relay
  2026-11-21 15:47 ` [PATCH v2 1/2] drm/imagination: Destroy the common FW object when HWRT data setup fails Suyuan Wang via B4 Relay
@ 2026-11-21 15:47 ` Suyuan Wang via B4 Relay
  1 sibling, 0 replies; 3+ messages in thread
From: Suyuan Wang via B4 Relay @ 2026-11-21 15:47 UTC (permalink / raw)
  To: alessio.belle, luigi.santivetti
  Cc: donald.robson, mripard, dri-devel, maarten.lankhorst, airlied,
	simona, tzimmermann, linux-kernel, imagination

From: Suyuan Wang <stevewong@hust.edu.cn>

hwrt_init_kernel_structure() stores the 32-bit args->layers from
userspace in the 16-bit hwrt->max_rts:

	hwrt->max_rts = args->layers;

hwrt_data_init_fw_structure() then allocates the SRTC and RAA objects
when args->layers > 1, but hwrt_data_fini_fw_structure() only frees them
when hwrt->max_rts > 1. With layers = 0x10000 or 0x10001, max_rts ends
up as 0 or 1, so both firmware objects are leaked when the data set is
destroyed or its creation fails. Any user who can open the render node
can do this with DRM_IOCTL_PVR_CREATE_HWRT_DATASET, and the objects
grow with the layer count.

Make max_rts a u32, like args->layers and the max_rts field of the
firmware's struct rogue_fwif_rta_ctl, so that both checks see the same
value.

Found with a static analysis tool and checked by hand.

Fixes: 6eedddab733b ("drm/imagination: Implement free list and HWRT create and destroy ioctls")
Cc: stable@vger.kernel.org
Co-developed-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Kaixuan Li <kaixuan.li@ntu.edu.sg>
Signed-off-by: Suyuan Wang <stevewong@hust.edu.cn>
---
 drivers/gpu/drm/imagination/pvr_hwrt.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/imagination/pvr_hwrt.h b/drivers/gpu/drm/imagination/pvr_hwrt.h
index 676070b20c3b..7c215e2bd35a 100644
--- a/drivers/gpu/drm/imagination/pvr_hwrt.h
+++ b/drivers/gpu/drm/imagination/pvr_hwrt.h
@@ -74,7 +74,7 @@ struct pvr_hwrt_dataset {
 	struct pvr_free_list *free_lists[ROGUE_FWIF_NUM_RTDATA_FREELISTS];
 
 	/** @max_rts: Maximum render targets for this HWRT data set. */
-	u16 max_rts;
+	u32 max_rts;
 };
 
 struct pvr_hwrt_dataset *

-- 
2.34.1



^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2026-11-21 15:47 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-21 15:47 [PATCH v2 0/2] drm/imagination: Fix leaks in HWRT data set handling Suyuan Wang via B4 Relay
2026-11-21 15:47 ` [PATCH v2 1/2] drm/imagination: Destroy the common FW object when HWRT data setup fails Suyuan Wang via B4 Relay
2026-11-21 15:47 ` [PATCH v2 2/2] drm/imagination: Don't truncate the layer count in max_rts Suyuan Wang via B4 Relay

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®