[PATCH v3,2/2] media: mediatek: vcodec: Fix possible invalid memory access for encoder

Yunfei Dong yunfei.dong at mediatek.com
Fri Jul 28 20:47:35 PDT 2023


The vpu maybe null pointer or unreasonable value when scp crash, need to
validate that the vpu pointer and the vpu instance within this context is
valid in case of leading to kernel reboot.

Fixes: 27a274db6b4c6 ("[media] vcodec: mediatek: Add Mediatek VP8 Video Encoder Driver")
Signed-off-by: Yunfei Dong <yunfei.dong at mediatek.com>
Reported-by: Steve Cho <stevecho at google.com>
---
 .../vcodec/encoder/mtk_vcodec_enc_drv.h       |  2 +
 .../mediatek/vcodec/encoder/venc_vpu_if.c     | 39 +++++++++++++++++--
 2 files changed, 37 insertions(+), 4 deletions(-)

diff --git a/drivers/media/platform/mediatek/vcodec/encoder/mtk_vcodec_enc_drv.h b/drivers/media/platform/mediatek/vcodec/encoder/mtk_vcodec_enc_drv.h
index c07010e566492..a042f607ed8d1 100644
--- a/drivers/media/platform/mediatek/vcodec/encoder/mtk_vcodec_enc_drv.h
+++ b/drivers/media/platform/mediatek/vcodec/encoder/mtk_vcodec_enc_drv.h
@@ -123,6 +123,7 @@ struct mtk_enc_params {
  * @xfer_func: enum v4l2_xfer_func, colorspace transfer function
  *
  * @q_mutex: vb2_queue mutex.
+ * @vpu_inst: vpu instance pointer.
  */
 struct mtk_vcodec_enc_ctx {
 	enum mtk_instance_type type;
@@ -156,6 +157,7 @@ struct mtk_vcodec_enc_ctx {
 	enum v4l2_xfer_func xfer_func;
 
 	struct mutex q_mutex;
+	void *vpu_inst;
 };
 
 /**
diff --git a/drivers/media/platform/mediatek/vcodec/encoder/venc_vpu_if.c b/drivers/media/platform/mediatek/vcodec/encoder/venc_vpu_if.c
index 708db1bb32d44..b10f3544b2411 100644
--- a/drivers/media/platform/mediatek/vcodec/encoder/venc_vpu_if.c
+++ b/drivers/media/platform/mediatek/vcodec/encoder/venc_vpu_if.c
@@ -42,19 +42,46 @@ static void handle_enc_encode_msg(struct venc_vpu_inst *vpu, const void *data)
 	vpu->is_key_frm = msg->is_key_frm;
 }
 
+static bool vpu_enc_check_ap_inst(struct mtk_vcodec_enc_dev *enc_dev, struct venc_vpu_inst *vpu)
+{
+	struct mtk_vcodec_enc_ctx *ctx;
+	int ret = false;
+
+	list_for_each_entry(ctx, &enc_dev->ctx_list, list) {
+		if (!IS_ERR_OR_NULL(ctx) && ctx->vpu_inst == vpu) {
+			ret = true;
+			break;
+		}
+	}
+
+	return ret;
+}
+
 static void vpu_enc_ipi_handler(void *data, unsigned int len, void *priv)
 {
+	struct mtk_vcodec_enc_dev *enc_dev;
 	const struct venc_vpu_ipi_msg_common *msg = data;
-	struct venc_vpu_inst *vpu =
-		(struct venc_vpu_inst *)(unsigned long)msg->venc_inst;
+	struct venc_vpu_inst *vpu;
+
+	enc_dev = (struct mtk_vcodec_enc_dev *)priv;
+	vpu = (struct venc_vpu_inst *)(unsigned long)msg->venc_inst;
+	if (!priv || !vpu) {
+		mtk_v4l2_venc_err(vpu->ctx, "venc_inst is NULL, did the SCP hang or crash?");
+		return;
+	}
 
 	mtk_venc_debug(vpu->ctx, "msg_id %x inst %p status %d", msg->msg_id, vpu, msg->status);
+	if (!vpu_enc_check_ap_inst(enc_dev, vpu) || msg->msg_id < VPU_IPIMSG_ENC_INIT_DONE ||
+	    msg->msg_id > VPU_IPIMSG_ENC_DEINIT_DONE) {
+		mtk_v4l2_venc_err(vpu->ctx, "venc msg id not correctly => 0x%x", msg->msg_id);
+		vpu->failure = -EINVAL;
+		goto error;
+	}
 
-	vpu->signaled = 1;
 	vpu->failure = (msg->status != VENC_IPI_MSG_STATUS_OK);
 	if (vpu->failure) {
 		mtk_venc_err(vpu->ctx, "vpu enc status failure %d", vpu->failure);
-		return;
+		goto error;
 	}
 
 	switch (msg->msg_id) {
@@ -72,6 +99,9 @@ static void vpu_enc_ipi_handler(void *data, unsigned int len, void *priv)
 		mtk_venc_err(vpu->ctx, "unknown msg id %x", msg->msg_id);
 		break;
 	}
+
+error:
+	vpu->signaled = 1;
 }
 
 static int vpu_enc_send_msg(struct venc_vpu_inst *vpu, void *msg,
@@ -105,6 +135,7 @@ int vpu_enc_init(struct venc_vpu_inst *vpu)
 	init_waitqueue_head(&vpu->wq_hd);
 	vpu->signaled = 0;
 	vpu->failure = 0;
+	vpu->ctx->vpu_inst = vpu;
 
 	status = mtk_vcodec_fw_ipi_register(vpu->ctx->dev->fw_handler, vpu->id,
 					    vpu_enc_ipi_handler, "venc", NULL);
-- 
2.18.0




More information about the linux-arm-kernel mailing list