linux/drivers/media/platform/mediatek/vcodec/encoder, branch master

media: mediatek: vcodec: fix use-after-free in encoder release path

2026-03-23T10:18:34Z

The fops_vcodec_release() function frees the context structure (ctx) without first cancelling any pending or running work in ctx->encode_work. This creates a race window where the workqueue handler (mtk_venc_worker) may still be accessing the context memory after it has been freed. Race condition: CPU 0 (release path) CPU 1 (workqueue) --------------------- ------------------ fops_vcodec_release() v4l2_m2m_ctx_release() v4l2_m2m_cancel_job() // waits for m2m job "done" mtk_venc_worker() v4l2_m2m_job_finish() // m2m job "done" // BUT worker still running! // post-job_finish access: other ctx dereferences // UAF if ctx already freed // returns (job "done") kfree(ctx) // ctx freed Root cause: The v4l2_m2m_ctx_release() only waits for the m2m job lifecycle (via TRANS_RUNNING flag), not the workqueue lifecycle. After v4l2_m2m_job_finish() is called, the m2m framework considers the job complete and v4l2_m2m_ctx_release() returns, but the worker function continues executing and may still access ctx. The work is queued during encode operations via: queue_work(ctx->dev->encode_workqueue, &ctx->encode_work) The worker function accesses ctx->m2m_ctx, ctx->dev, and other ctx fields even after calling v4l2_m2m_job_finish(). This vulnerability was confirmed with KASAN by running an instrumented test module that widens the post-job_finish race window. KASAN detected: BUG: KASAN: slab-use-after-free in mtk_venc_worker+0x159/0x180 Read of size 4 at addr ffff88800326e000 by task kworker/u8:0/12 Workqueue: mtk_vcodec_enc_wq mtk_venc_worker Allocated by task 47: __kasan_kmalloc+0x7f/0x90 fops_vcodec_open+0x85/0x1a0 Freed by task 47: __kasan_slab_free+0x43/0x70 kfree+0xee/0x3a0 fops_vcodec_release+0xb7/0x190 Fix this by calling cancel_work_sync(&ctx->encode_work) before kfree(ctx). This ensures the workqueue handler is both cancelled (if pending) and synchronized (waits for any running handler to complete) before the context is freed. Placement rationale: The fix is placed after v4l2_ctrl_handler_free() and before list_del_init(&ctx->list). At this point, all m2m operations are done (v4l2_m2m_ctx_release() has returned), and we need to ensure the workqueue is synchronized before removing ctx from the list and freeing it. Note: The open error path does NOT need cancel_work_sync() because INIT_WORK() only initializes the work structure - it does not schedule it. Work is only scheduled later during device_run() operations. Fixes: 0934d3759615 ("media: mediatek: vcodec: separate decoder and encoder") Cc: stable@vger.kernel.org Signed-off-by: Fan Wu Reviewed-by: Nicolas Dufresne Signed-off-by: Nicolas Dufresne Signed-off-by: Hans Verkuil

Convert 'alloc_obj' family to use the new default GFP_KERNEL argument

2026-02-22T01:09:51Z

This was done entirely with mindless brute force, using git grep -l '\

treewide: Replace kmalloc with kmalloc_obj for non-scalar types

2026-02-21T09:02:28Z

This is the result of running the Coccinelle script from scripts/coccinelle/api/kmalloc_objs.cocci. The script is designed to avoid scalar types (which need careful case-by-case checking), and instead replace kmalloc-family calls that allocate struct or union object instances: Single allocations: kmalloc(sizeof(TYPE), ...) are replaced with: kmalloc_obj(TYPE, ...) Array allocations: kmalloc_array(COUNT, sizeof(TYPE), ...) are replaced with: kmalloc_objs(TYPE, COUNT, ...) Flex array allocations: kmalloc(struct_size(PTR, FAM, COUNT), ...) are replaced with: kmalloc_flex(*PTR, FAM, COUNT, ...) (where TYPE may also be *VAR) The resulting allocations no longer return "void *", instead returning "TYPE *". Signed-off-by: Kees Cook

media: mediatek: vcodec: Discard pm_runtime_put() return value

2026-01-13T08:41:59Z

Printing error messages on pm_runtime_put() returning negative values is not particularly useful. Returning an error code from pm_runtime_put() merely means that it has not queued up a work item to check whether or not the device can be suspended and there are many perfectly valid situations in which that can happen, like after writing "on" to the devices' runtime PM "control" attribute in sysfs for one example. Accordingly, update mtk_vcodec_enc_pw_off() and mtk_vcodec_dec_pw_off() to simply discard the return value of pm_runtime_put(). This will facilitate a planned change of the pm_runtime_put() return type to void in the future. Signed-off-by: Rafael J. Wysocki Reviewed-by: Nicolas Dufresne Signed-off-by: Nicolas Dufresne Signed-off-by: Hans Verkuil

media: mediatek: vcodec: use = { } instead of memset()

2026-01-05T14:56:31Z

Based on testing and recommendations by David Lechner et al. [1][2], using = { } to initialize a structure or array is the preferred way to do this in the kernel. Converts memset() to = { }, thereby: - Eliminating the risk of sizeof() mismatches. - Simplifying the code. [1]: https://lore.kernel.org/linux-iio/202505090942.48EBF01B@keescook/ [2]: https://lore.kernel.org/lkml/20250614151844.50524610@jic23-huawei/ Signed-off-by: Qianfeng Rong Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Nicolas Dufresne Signed-off-by: Hans Verkuil

media: mediatek: encoder: Fix uninitialized scalar variable issue

2026-01-05T14:56:31Z

UNINIT checker finds some instances of variables that are used without being initialized, for example using the uninitialized value enc_result.is_key_frm can result in unpredictable behavior, so initialize these variables after declaring. Fixes: 4e855a6efa54 ("[media] vcodec: mediatek: Add Mediatek V4L2 Video Encoder Driver") Cc: stable@vger.kernel.org Signed-off-by: Irui Wang Reviewed-by: AngeloGioacchino Del Regno Signed-off-by: Nicolas Dufresne Signed-off-by: Hans Verkuil

media: mediatek: vcodec: Use spinlock for context list protection lock

2025-10-20T07:23:17Z

Previously a mutex was added to protect the encoder and decoder context lists from unexpected changes originating from the SCP IP block, causing the context pointer to go invalid, resulting in a NULL pointer dereference in the IPI handler. Turns out on the MT8173, the VPU IPI handler is called from hard IRQ context. This causes a big warning from the scheduler. This was first reported downstream on the ChromeOS kernels, but is also reproducible on mainline using Fluster with the FFmpeg v4l2m2m decoders. Even though the actual capture format is not supported, the affected code paths are triggered. Since this lock just protects the context list and operations on it are very fast, it should be OK to switch to a spinlock. Fixes: 6467cda18c9f ("media: mediatek: vcodec: adding lock to protect decoder context list") Fixes: afaaf3a0f647 ("media: mediatek: vcodec: adding lock to protect encoder context list") Cc: Yunfei Dong Cc: stable@vger.kernel.org Signed-off-by: Chen-Yu Tsai Reviewed-by: Fei Shao Reviewed-by: Tomasz Figa Signed-off-by: Nicolas Dufresne Signed-off-by: Hans Verkuil

media: mediatek: vcodec: Drop unneeded v4l2_m2m_get_vq() NULL check

2025-10-17T09:31:14Z

The v4l2_m2m_get_vq() function never returns NULL. In the set format handlers, the check may have been intended to catch invalid format types, but that's not needed as the V4L2 core picks the appropriate VIDIOC_S_FMT ioctl handler based on the format type, so the type can't be incorrect. In the get format handlers, the return value is not used for any purpose other than the NULL check, which was therefore probably intended to catch invalid format types. That's not needed for the same reason as in the set format handler. In other locations the v4l2_m2m_get_vq() function is called with a hardcoded V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE type, so the NULL check can't have been an attempt to catch an invalid type there either. Drop the unneeded return value checks and, as the function has no side effect, the unneeded function call as well. Signed-off-by: Laurent Pinchart Signed-off-by: Hans Verkuil

media: mtk: vcodec: Access v4l2_fh from file

2025-08-13T06:33:49Z

The v4l2_fh associated with an open file handle is now guaranteed to be available in file->private_data, initialised by v4l2_fh_add(). Access the v4l2_fh, and from there the driver-specific structure, from the file * in all ioctl handlers. While at modify mtk_vcodec_enc_get_chip_name() to accept a ctx instead of a raw void *. Signed-off-by: Jacopo Mondi Co-developed-by: Laurent Pinchart Signed-off-by: Laurent Pinchart Signed-off-by: Hans Verkuil

media: Reset file->private_data to NULL in v4l2_fh_del()

2025-08-13T06:33:44Z

Multiple drivers that use v4l2_fh and call v4l2_fh_del() manually reset the file->private_data pointer to NULL in their video device .release() file operation handler. Move the code to the v4l2_fh_del() function to avoid direct access to file->private_data in drivers. This requires adding a file pointer argument to the function. Changes to drivers have been generated with the following coccinelle semantic patch: @@ expression fh; identifier filp; identifier release; type ret; @@ ret release(..., struct file *filp, ...) { <... - filp->private_data = NULL; ... - v4l2_fh_del(fh); + v4l2_fh_del(fh, filp); ...> } @@ expression fh; identifier filp; identifier release; type ret; @@ ret release(..., struct file *filp, ...) { <... - v4l2_fh_del(fh); + v4l2_fh_del(fh, filp); ... - filp->private_data = NULL; ...> } @@ expression fh; identifier filp; identifier release; type ret; @@ ret release(..., struct file *filp, ...) { <... - v4l2_fh_del(fh); + v4l2_fh_del(fh, filp); ...> } Manual changes have been applied to Documentation/ to update the usage patterns, to drivers/media/v4l2-core/v4l2-fh.c to update the v4l2_fh_del() prototype and reset file->private_data, and to include/media/v4l2-fh.h to update the v4l2_fh_del() function prototype and its documentation. Additionally, white space issues have been fixed manually in drivers/usb/gadget/function/uvc_v4l2.c Signed-off-by: Laurent Pinchart Signed-off-by: Hans Verkuil