Merge tag 'media/v5.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media

Pull media updates from Mauro Carvalho Chehab:

 - uAPI documentation for stateless decoders

 - Added a new CEC ioctl together with its documentation

 - Improved IPU3 documentation

 - New i2c drivers: hi556 and imx290

 - Added support on Vivid driver for meta streams

 - Added de-interlace support for sunxi subdriver

 - Added a few new remote controler keymaps

 - Added H.265 support for Sunxi Cedrus driver

 - Another round of random driver cleanups, fixes and improvements

* tag 'media/v5.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (361 commits)
  media: Revert "media: mtk-vcodec: Remove extra area allocation in an input buffer on encoding"
  media: hantro: Set H264 FIELDPIC_FLAG_E flag correctly
  media: hantro: Remove now unused H264 pic_size
  media: hantro: Use output buffer width and height for H264 decoding
  media: hantro: Reduce H264 extra space for motion vectors
  media: hantro: Fix H264 motion vector buffer offset
  media: ti-vpe: vpe: fix compatible to match bindings
  media: dt-bindings: media: ti-vpe: Document VPE driver
  media: zr364xx: remove redundant assigmnent to idx, clean up code
  media: Documentation: media: *_DEFAULT targets for subdevs
  media: hantro: Fix s_fmt for dynamic resolution changes
  media: i2c: Use the correct style for SPDX License Identifier
  media: siano: Use the correct style for SPDX License Identifier
  media: vicodec: media_device_cleanup was called too early
  media: vim2m: media_device_cleanup was called too early
  media: cedrus: Increase maximum supported size
  media: cedrus: Fix H264 4k support
  media: cedrus: Properly signal size in mode register
  media: v4l2-ctrl: Lock main_hdl on operations of requests_queued.
  media: si470x-i2c: add missed operations in remove
  ...

41 files changed, 1539 insertions, 442 deletions

diff --git a/drivers/staging/media/allegro-dvt/nal-h264.c b/drivers/staging/media/allegro-dvt/nal-h264.c
index 4e14b77851e1..bd48b8883572 100644
--- a/drivers/staging/media/allegro-dvt/nal-h264.c
+++ b/drivers/staging/media/allegro-dvt/nal-h264.c
@@ -235,7 +235,7 @@ static inline int rbsp_write_bit(struct rbsp *rbsp, bool value)
 
 	rbsp->pos++;
 
-	if (value == 1 ||
+	if (value ||
 	    (rbsp->num_consecutive_zeros < 7 && (rbsp->pos % 8 == 0))) {
 		rbsp->num_consecutive_zeros = 0;
 	} else {
diff --git a/drivers/staging/media/hantro/hantro.h b/drivers/staging/media/hantro/hantro.h
index f670bbde4159..deb90ae37859 100644
--- a/drivers/staging/media/hantro/hantro.h
+++ b/drivers/staging/media/hantro/hantro.h
@@ -26,21 +26,9 @@
 
 #include "hantro_hw.h"
 
-#define VP8_MB_DIM			16
-#define VP8_MB_WIDTH(w)			DIV_ROUND_UP(w, VP8_MB_DIM)
-#define VP8_MB_HEIGHT(h)		DIV_ROUND_UP(h, VP8_MB_DIM)
-
-#define H264_MB_DIM			16
-#define H264_MB_WIDTH(w)		DIV_ROUND_UP(w, H264_MB_DIM)
-#define H264_MB_HEIGHT(h)		DIV_ROUND_UP(h, H264_MB_DIM)
-
-#define MPEG2_MB_DIM			16
-#define MPEG2_MB_WIDTH(w)		DIV_ROUND_UP(w, MPEG2_MB_DIM)
-#define MPEG2_MB_HEIGHT(h)		DIV_ROUND_UP(h, MPEG2_MB_DIM)
-
-#define JPEG_MB_DIM			16
-#define JPEG_MB_WIDTH(w)		DIV_ROUND_UP(w, JPEG_MB_DIM)
-#define JPEG_MB_HEIGHT(h)		DIV_ROUND_UP(h, JPEG_MB_DIM)
+#define MB_DIM			16
+#define MB_WIDTH(w)		DIV_ROUND_UP(w, MB_DIM)
+#define MB_HEIGHT(h)		DIV_ROUND_UP(h, MB_DIM)
 
 struct hantro_ctx;
 struct hantro_codec_ops;
@@ -379,7 +367,7 @@ static inline void hantro_reg_write(struct hantro_dev *vpu,
 bool hantro_is_encoder_ctx(const struct hantro_ctx *ctx);
 
 void *hantro_get_ctrl(struct hantro_ctx *ctx, u32 id);
-dma_addr_t hantro_get_ref(struct vb2_queue *q, u64 ts);
+dma_addr_t hantro_get_ref(struct hantro_ctx *ctx, u64 ts);
 
 static inline struct vb2_v4l2_buffer *
 hantro_get_src_buf(struct hantro_ctx *ctx)
diff --git a/drivers/staging/media/hantro/hantro_drv.c b/drivers/staging/media/hantro/hantro_drv.c
index 6d9d41170832..26108c96b674 100644
--- a/drivers/staging/media/hantro/hantro_drv.c
+++ b/drivers/staging/media/hantro/hantro_drv.c
@@ -43,8 +43,9 @@ void *hantro_get_ctrl(struct hantro_ctx *ctx, u32 id)
 	return ctrl ? ctrl->p_cur.p : NULL;
 }
 
-dma_addr_t hantro_get_ref(struct vb2_queue *q, u64 ts)
+dma_addr_t hantro_get_ref(struct hantro_ctx *ctx, u64 ts)
 {
+	struct vb2_queue *q = v4l2_m2m_get_dst_vq(ctx->fh.m2m_ctx);
 	struct vb2_buffer *buf;
 	int index;
 
@@ -413,20 +414,18 @@ static int hantro_open(struct file *filp)
 	if (func->id == MEDIA_ENT_F_PROC_VIDEO_ENCODER) {
 		allowed_codecs = vpu->variant->codec & HANTRO_ENCODERS;
 		ctx->buf_finish = hantro_enc_buf_finish;
-		ctx->fh.m2m_ctx = v4l2_m2m_ctx_init(vpu->m2m_dev, ctx,
-						    queue_init);
 	} else if (func->id == MEDIA_ENT_F_PROC_VIDEO_DECODER) {
 		allowed_codecs = vpu->variant->codec & HANTRO_DECODERS;
 		ctx->buf_finish = hantro_dec_buf_finish;
-		ctx->fh.m2m_ctx = v4l2_m2m_ctx_init(vpu->m2m_dev, ctx,
-						    queue_init);
 	} else {
-		ctx->fh.m2m_ctx = ERR_PTR(-ENODEV);
+		ret = -ENODEV;
+		goto err_ctx_free;
 	}
+
+	ctx->fh.m2m_ctx = v4l2_m2m_ctx_init(vpu->m2m_dev, ctx, queue_init);
 	if (IS_ERR(ctx->fh.m2m_ctx)) {
 		ret = PTR_ERR(ctx->fh.m2m_ctx);
-		kfree(ctx);
-		return ret;
+		goto err_ctx_free;
 	}
 
 	v4l2_fh_init(&ctx->fh, vdev);
@@ -447,6 +446,7 @@ static int hantro_open(struct file *filp)
 err_fh_free:
 	v4l2_fh_del(&ctx->fh);
 	v4l2_fh_exit(&ctx->fh);
+err_ctx_free:
 	kfree(ctx);
 	return ret;
 }
diff --git a/drivers/staging/media/hantro/hantro_g1_h264_dec.c b/drivers/staging/media/hantro/hantro_g1_h264_dec.c
index 7ab534936843..3cd40a8f0daa 100644
--- a/drivers/staging/media/hantro/hantro_g1_h264_dec.c
+++ b/drivers/staging/media/hantro/hantro_g1_h264_dec.c
@@ -34,9 +34,11 @@ static void set_params(struct hantro_ctx *ctx)
 	reg = G1_REG_DEC_CTRL0_DEC_AXI_WR_ID(0x0);
 	if (sps->flags & V4L2_H264_SPS_FLAG_MB_ADAPTIVE_FRAME_FIELD)
 		reg |= G1_REG_DEC_CTRL0_SEQ_MBAFF_E;
-	reg |= G1_REG_DEC_CTRL0_PICORD_COUNT_E;
-	if (dec_param->nal_ref_idc)
-		reg |= G1_REG_DEC_CTRL0_WRITE_MVS_E;
+	if (sps->profile_idc > 66) {
+		reg |= G1_REG_DEC_CTRL0_PICORD_COUNT_E;
+		if (dec_param->nal_ref_idc)
+			reg |= G1_REG_DEC_CTRL0_WRITE_MVS_E;
+	}
 
 	if (!(sps->flags & V4L2_H264_SPS_FLAG_FRAME_MBS_ONLY) &&
 	    (sps->flags & V4L2_H264_SPS_FLAG_MB_ADAPTIVE_FRAME_FIELD ||
@@ -49,8 +51,8 @@ static void set_params(struct hantro_ctx *ctx)
 	vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL0);
 
 	/* Decoder control register 1. */
-	reg = G1_REG_DEC_CTRL1_PIC_MB_WIDTH(sps->pic_width_in_mbs_minus1 + 1) |
-	      G1_REG_DEC_CTRL1_PIC_MB_HEIGHT_P(sps->pic_height_in_map_units_minus1 + 1) |
+	reg = G1_REG_DEC_CTRL1_PIC_MB_WIDTH(MB_WIDTH(ctx->src_fmt.width)) |
+	      G1_REG_DEC_CTRL1_PIC_MB_HEIGHT_P(MB_HEIGHT(ctx->src_fmt.height)) |
 	      G1_REG_DEC_CTRL1_REF_FRAMES(sps->max_num_ref_frames);
 	vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL1);
 
@@ -61,7 +63,7 @@ static void set_params(struct hantro_ctx *ctx)
 	/* always use the matrix sent from userspace */
 	reg |= G1_REG_DEC_CTRL2_TYPE1_QUANT_E;
 
-	if (slices[0].flags &  V4L2_H264_SLICE_FLAG_FIELD_PIC)
+	if (!(sps->flags & V4L2_H264_SPS_FLAG_FRAME_MBS_ONLY))
 		reg |= G1_REG_DEC_CTRL2_FIELDPIC_FLAG_E;
 	vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL2);
 
@@ -79,7 +81,7 @@ static void set_params(struct hantro_ctx *ctx)
 		reg |= G1_REG_DEC_CTRL4_CABAC_E;
 	if (sps->flags & V4L2_H264_SPS_FLAG_DIRECT_8X8_INFERENCE)
 		reg |= G1_REG_DEC_CTRL4_DIR_8X8_INFER_E;
-	if (sps->chroma_format_idc == 0)
+	if (sps->profile_idc >= 100 && sps->chroma_format_idc == 0)
 		reg |= G1_REG_DEC_CTRL4_BLACKWHITE_E;
 	if (pps->flags & V4L2_H264_PPS_FLAG_WEIGHTED_PRED)
 		reg |= G1_REG_DEC_CTRL4_WEIGHT_PRED_E;
@@ -220,10 +222,9 @@ static void set_ref(struct hantro_ctx *ctx)
 
 	/* Set up addresses of DPB buffers. */
 	for (i = 0; i < HANTRO_H264_DPB_SIZE; i++) {
-		struct vb2_buffer *buf =  hantro_h264_get_ref_buf(ctx, i);
+		dma_addr_t dma_addr = hantro_h264_get_ref_buf(ctx, i);
 
-		vdpu_write_relaxed(vpu, vb2_dma_contig_plane_dma_addr(buf, 0),
-				   G1_REG_ADDR_REF(i));
+		vdpu_write_relaxed(vpu, dma_addr, G1_REG_ADDR_REF(i));
 	}
 }
 
@@ -233,6 +234,7 @@ static void set_buffers(struct hantro_ctx *ctx)
 	struct vb2_v4l2_buffer *src_buf, *dst_buf;
 	struct hantro_dev *vpu = ctx->dev;
 	dma_addr_t src_dma, dst_dma;
+	size_t offset = 0;
 
 	src_buf = hantro_get_src_buf(ctx);
 	dst_buf = hantro_get_dst_buf(ctx);
@@ -243,18 +245,30 @@ static void set_buffers(struct hantro_ctx *ctx)
 
 	/* Destination (decoded frame) buffer. */
 	dst_dma = vb2_dma_contig_plane_dma_addr(&dst_buf->vb2_buf, 0);
-	vdpu_write_relaxed(vpu, dst_dma, G1_REG_ADDR_DST);
+	/* Adjust dma addr to start at second line for bottom field */
+	if (ctrls->slices[0].flags & V4L2_H264_SLICE_FLAG_BOTTOM_FIELD)
+		offset = ALIGN(ctx->src_fmt.width, MB_DIM);
+	vdpu_write_relaxed(vpu, dst_dma + offset, G1_REG_ADDR_DST);
 
 	/* Higher profiles require DMV buffer appended to reference frames. */
-	if (ctrls->sps->profile_idc > 66) {
-		size_t pic_size = ctx->h264_dec.pic_size;
-		size_t mv_offset = round_up(pic_size, 8);
-
+	if (ctrls->sps->profile_idc > 66 && ctrls->decode->nal_ref_idc) {
+		unsigned int bytes_per_mb = 384;
+
+		/* DMV buffer for monochrome start directly after Y-plane */
+		if (ctrls->sps->profile_idc >= 100 &&
+		    ctrls->sps->chroma_format_idc == 0)
+			bytes_per_mb = 256;
+		offset = bytes_per_mb * MB_WIDTH(ctx->src_fmt.width) *
+			 MB_HEIGHT(ctx->src_fmt.height);
+
+		/*
+		 * DMV buffer is split in two for field encoded frames,
+		 * adjust offset for bottom field
+		 */
 		if (ctrls->slices[0].flags & V4L2_H264_SLICE_FLAG_BOTTOM_FIELD)
-			mv_offset += 32 * H264_MB_WIDTH(ctx->dst_fmt.width);
-
-		vdpu_write_relaxed(vpu, dst_dma + mv_offset,
-				   G1_REG_ADDR_DIR_MV);
+			offset += 32 * MB_WIDTH(ctx->src_fmt.width) *
+				  MB_HEIGHT(ctx->src_fmt.height);
+		vdpu_write_relaxed(vpu, dst_dma + offset, G1_REG_ADDR_DIR_MV);
 	}
 
 	/* Auxiliary buffer prepared in hantro_g1_h264_dec_prepare_table(). */
diff --git a/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c b/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c
index 80f0e94f8afa..f3bf67d8a289 100644
--- a/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c
+++ b/drivers/staging/media/hantro/hantro_g1_mpeg2_dec.c
@@ -105,17 +105,14 @@ hantro_g1_mpeg2_dec_set_buffers(struct hantro_dev *vpu, struct hantro_ctx *ctx,
 {
 	dma_addr_t forward_addr = 0, backward_addr = 0;
 	dma_addr_t current_addr, addr;
-	struct vb2_queue *vq;
-
-	vq = v4l2_m2m_get_dst_vq(ctx->fh.m2m_ctx);
 
 	switch (picture->picture_coding_type) {
 	case V4L2_MPEG2_PICTURE_CODING_TYPE_B:
-		backward_addr = hantro_get_ref(vq,
+		backward_addr = hantro_get_ref(ctx,
 					       slice_params->backward_ref_ts);
 		/* fall-through */
 	case V4L2_MPEG2_PICTURE_CODING_TYPE_P:
-		forward_addr = hantro_get_ref(vq,
+		forward_addr = hantro_get_ref(ctx,
 					      slice_params->forward_ref_ts);
 	}
 
@@ -207,8 +204,8 @@ void hantro_g1_mpeg2_dec_run(struct hantro_ctx *ctx)
 	      G1_REG_DEC_AXI_WR_ID(0);
 	vdpu_write_relaxed(vpu, reg, G1_SWREG(3));
 
-	reg = G1_REG_PIC_MB_WIDTH(MPEG2_MB_WIDTH(ctx->dst_fmt.width)) |
-	      G1_REG_PIC_MB_HEIGHT_P(MPEG2_MB_HEIGHT(ctx->dst_fmt.height)) |
+	reg = G1_REG_PIC_MB_WIDTH(MB_WIDTH(ctx->dst_fmt.width)) |
+	      G1_REG_PIC_MB_HEIGHT_P(MB_HEIGHT(ctx->dst_fmt.height)) |
 	      G1_REG_ALT_SCAN_E(picture->alternate_scan) |
 	      G1_REG_TOPFIELDFIRST_E(picture->top_field_first);
 	vdpu_write_relaxed(vpu, reg, G1_SWREG(4));
diff --git a/drivers/staging/media/hantro/hantro_g1_vp8_dec.c b/drivers/staging/media/hantro/hantro_g1_vp8_dec.c
index 6d99c2be01cf..cad18094fee0 100644
--- a/drivers/staging/media/hantro/hantro_g1_vp8_dec.c
+++ b/drivers/staging/media/hantro/hantro_g1_vp8_dec.c
@@ -370,19 +370,18 @@ static void cfg_tap(struct hantro_ctx *ctx,
 static void cfg_ref(struct hantro_ctx *ctx,
 		    const struct v4l2_ctrl_vp8_frame_header *hdr)
 {
-	struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q;
 	struct hantro_dev *vpu = ctx->dev;
 	struct vb2_v4l2_buffer *vb2_dst;
 	dma_addr_t ref;
 
 	vb2_dst = hantro_get_dst_buf(ctx);
 
-	ref = hantro_get_ref(cap_q, hdr->last_frame_ts);
+	ref = hantro_get_ref(ctx, hdr->last_frame_ts);
 	if (!ref)
 		ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0);
 	vdpu_write_relaxed(vpu, ref, G1_REG_ADDR_REF(0));
 
-	ref = hantro_get_ref(cap_q, hdr->golden_frame_ts);
+	ref = hantro_get_ref(ctx, hdr->golden_frame_ts);
 	WARN_ON(!ref && hdr->golden_frame_ts);
 	if (!ref)
 		ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0);
@@ -390,7 +389,7 @@ static void cfg_ref(struct hantro_ctx *ctx,
 		ref |= G1_REG_ADDR_REF_TOPC_E;
 	vdpu_write_relaxed(vpu, ref, G1_REG_ADDR_REF(4));
 
-	ref = hantro_get_ref(cap_q, hdr->alt_frame_ts);
+	ref = hantro_get_ref(ctx, hdr->alt_frame_ts);
 	WARN_ON(!ref && hdr->alt_frame_ts);
 	if (!ref)
 		ref = vb2_dma_contig_plane_dma_addr(&vb2_dst->vb2_buf, 0);
@@ -470,8 +469,8 @@ void hantro_g1_vp8_dec_run(struct hantro_ctx *ctx)
 	vdpu_write_relaxed(vpu, reg, G1_REG_DEC_CTRL0);
 
 	/* Frame dimensions */
-	mb_width = VP8_MB_WIDTH(width);
-	mb_height = VP8_MB_HEIGHT(height);
+	mb_width = MB_WIDTH(width);
+	mb_height = MB_HEIGHT(height);
 	reg = G1_REG_DEC_CTRL1_PIC_MB_WIDTH(mb_width) |
 	      G1_REG_DEC_CTRL1_PIC_MB_HEIGHT_P(mb_height) |
 	      G1_REG_DEC_CTRL1_PIC_MB_W_EXT(mb_width >> 9) |
diff --git a/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c b/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c
index ecd34a7db190..938b48d4d3d9 100644
--- a/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c
+++ b/drivers/staging/media/hantro/hantro_h1_jpeg_enc.c
@@ -116,8 +116,8 @@ void hantro_h1_jpeg_enc_run(struct hantro_ctx *ctx)
 	/* Make sure that all registers are written at this point. */
 	vepu_write(vpu, reg, H1_REG_AXI_CTRL);
 
-	reg = H1_REG_ENC_CTRL_WIDTH(JPEG_MB_WIDTH(ctx->src_fmt.width))
-		| H1_REG_ENC_CTRL_HEIGHT(JPEG_MB_HEIGHT(ctx->src_fmt.height))
+	reg = H1_REG_ENC_CTRL_WIDTH(MB_WIDTH(ctx->src_fmt.width))
+		| H1_REG_ENC_CTRL_HEIGHT(MB_HEIGHT(ctx->src_fmt.height))
 		| H1_REG_ENC_CTRL_ENC_MODE_JPEG
 		| H1_REG_ENC_PIC_INTRA
 		| H1_REG_ENC_CTRL_EN_BIT;
diff --git a/drivers/staging/media/hantro/hantro_h264.c b/drivers/staging/media/hantro/hantro_h264.c
index 0d758e0c0f99..568640eab3a6 100644
--- a/drivers/staging/media/hantro/hantro_h264.c
+++ b/drivers/staging/media/hantro/hantro_h264.c
@@ -20,9 +20,9 @@
 /* Size with u32 units. */
 #define CABAC_INIT_BUFFER_SIZE		(460 * 2)
 #define POC_BUFFER_SIZE			34
-#define SCALING_LIST_SIZE		(6 * 16 + 6 * 64)
+#define SCALING_LIST_SIZE		(6 * 16 + 2 * 64)
 
-#define POC_CMP(p0, p1) ((p0) < (p1) ? -1 : 1)
+#define HANTRO_CMP(a, b) ((a) < (b) ? -1 : 1)
 
 /* Data structure describing auxiliary buffer format. */
 struct hantro_h264_dec_priv_tbl {
@@ -194,23 +194,6 @@ static const u32 h264_cabac_table[] = {
 	0x1f0c2517, 0x1f261440
 };
 
-/*
- * NOTE: The scaling lists are in zig-zag order, apply inverse scanning process
- * to get the values in matrix order. In addition, the hardware requires bytes
- * swapped within each subsequent 4 bytes. Both arrays below include both
- * transformations.
- */
-static const u32 zig_zag_4x4[] = {
-	3, 2, 7, 11, 6, 1, 0, 5, 10, 15, 14, 9, 4, 8, 13, 12
-};
-
-static const u32 zig_zag_8x8[] = {
-	3, 2, 11, 19, 10, 1, 0, 9, 18, 27, 35, 26, 17, 8, 7, 6,
-	15, 16, 25, 34, 43, 51, 42, 33, 24, 23, 14, 5, 4, 13, 22, 31,
-	32, 41, 50, 59, 58, 49, 40, 39, 30, 21, 12, 20, 29, 38, 47, 48,
-	57, 56, 55, 46, 37, 28, 36, 45, 54, 63, 62, 53, 44, 52, 61, 60
-};
-
 static void
 reorder_scaling_list(struct hantro_ctx *ctx)
 {
@@ -218,33 +201,23 @@ reorder_scaling_list(struct hantro_ctx *ctx)
 	const struct v4l2_ctrl_h264_scaling_matrix *scaling = ctrls->scaling;
 	const size_t num_list_4x4 = ARRAY_SIZE(scaling->scaling_list_4x4);
 	const size_t list_len_4x4 = ARRAY_SIZE(scaling->scaling_list_4x4[0]);
-	const size_t num_list_8x8 = ARRAY_SIZE(scaling->scaling_list_8x8);
 	const size_t list_len_8x8 = ARRAY_SIZE(scaling->scaling_list_8x8[0]);
 	struct hantro_h264_dec_priv_tbl *tbl = ctx->h264_dec.priv.cpu;
-	u8 *dst = tbl->scaling_list;
-	const u8 *src;
+	u32 *dst = (u32 *)tbl->scaling_list;
+	const u32 *src;
 	int i, j;
 
-	BUILD_BUG_ON(ARRAY_SIZE(zig_zag_4x4) != list_len_4x4);
-	BUILD_BUG_ON(ARRAY_SIZE(zig_zag_8x8) != list_len_8x8);
-	BUILD_BUG_ON(ARRAY_SIZE(tbl->scaling_list) !=
-		     num_list_4x4 * list_len_4x4 +
-		     num_list_8x8 * list_len_8x8);
-
-	src = &scaling->scaling_list_4x4[0][0];
-	for (i = 0; i < num_list_4x4; ++i) {
-		for (j = 0; j < list_len_4x4; ++j)
-			dst[zig_zag_4x4[j]] = src[j];
-		src += list_len_4x4;
-		dst += list_len_4x4;
+	for (i = 0; i < num_list_4x4; i++) {
+		src = (u32 *)&scaling->scaling_list_4x4[i];
+		for (j = 0; j < list_len_4x4 / 4; j++)
+			*dst++ = swab32(src[j]);
 	}
 
-	src = &scaling->scaling_list_8x8[0][0];
-	for (i = 0; i < num_list_8x8; ++i) {
-		for (j = 0; j < list_len_8x8; ++j)
-			dst[zig_zag_8x8[j]] = src[j];
-		src += list_len_8x8;
-		dst += list_len_8x8;
+	/* Only Intra/Inter Y lists */
+	for (i = 0; i < 2; i++) {
+		src = (u32 *)&scaling->scaling_list_8x8[i];
+		for (j = 0; j < list_len_8x8 / 4; j++)
+			*dst++ = swab32(src[j]);
 	}
 }
 
@@ -271,6 +244,7 @@ struct hantro_h264_reflist_builder {
 	const struct v4l2_h264_dpb_entry *dpb;
 	s32 pocs[HANTRO_H264_DPB_SIZE];
 	u8 unordered_reflist[HANTRO_H264_DPB_SIZE];
+	int frame_nums[HANTRO_H264_DPB_SIZE];
 	s32 curpoc;
 	u8 num_valid;
 };
@@ -294,13 +268,20 @@ static void
 init_reflist_builder(struct hantro_ctx *ctx,
 		     struct hantro_h264_reflist_builder *b)
 {
+	const struct v4l2_ctrl_h264_slice_params *slice_params;
 	const struct v4l2_ctrl_h264_decode_params *dec_param;
+	const struct v4l2_ctrl_h264_sps *sps;
 	struct vb2_v4l2_buffer *buf = hantro_get_dst_buf(ctx);
 	const struct v4l2_h264_dpb_entry *dpb = ctx->h264_dec.dpb;
 	struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q;
+	int cur_frame_num, max_frame_num;
 	unsigned int i;
 
 	dec_param = ctx->h264_dec.ctrls.decode;
+	slice_params = &ctx->h264_dec.ctrls.slices[0];
+	sps = ctx->h264_dec.ctrls.sps;
+	max_frame_num = 1 << (sps->log2_max_frame_num_minus4 + 4);
+	cur_frame_num = slice_params->frame_num;
 
 	memset(b, 0, sizeof(*b));
 	b->dpb = dpb;
@@ -318,6 +299,18 @@ init_reflist_builder(struct hantro_ctx *ctx,
 			continue;
 
 		buf = to_vb2_v4l2_buffer(vb2_get_buffer(cap_q, buf_idx));
+
+		/*
+		 * Handle frame_num wraparound as described in section
+		 * '8.2.4.1 Decoding process for picture numbers' of the spec.
+		 * TODO: This logic will have to be adjusted when we start
+		 * supporting interlaced content.
+		 */
+		if (dpb[i].frame_num > cur_frame_num)
+			b->frame_nums[i] = (int)dpb[i].frame_num - max_frame_num;
+		else
+			b->frame_nums[i] = dpb[i].frame_num;
+
 		b->pocs[i] = get_poc(buf->field, dpb[i].top_field_order_cnt,
 				     dpb[i].bottom_field_order_cnt);
 		b->unordered_reflist[b->num_valid] = i;
@@ -353,9 +346,10 @@ static int p_ref_list_cmp(const void *ptra, const void *ptrb, const void *data)
 	 * ascending order.
 	 */
 	if (!(a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM))
-		return b->frame_num - a->frame_num;
+		return HANTRO_CMP(builder->frame_nums[idxb],
+				  builder->frame_nums[idxa]);
 
-	return a->pic_num - b->pic_num;
+	return HANTRO_CMP(a->pic_num, b->pic_num);
 }
 
 static int b0_ref_list_cmp(const void *ptra, const void *ptrb, const void *data)
@@ -381,7 +375,7 @@ static int b0_ref_list_cmp(const void *ptra, const void *ptrb, const void *data)
 
 	/* Long term pics in ascending pic num order. */
 	if (a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)
-		return a->pic_num - b->pic_num;
+		return HANTRO_CMP(a->pic_num, b->pic_num);
 
 	poca = builder->pocs[idxa];
 	pocb = builder->pocs[idxb];
@@ -392,11 +386,11 @@ static int b0_ref_list_cmp(const void *ptra, const void *ptrb, const void *data)
 	 * order.
 	 */
 	if ((poca < builder->curpoc) != (pocb < builder->curpoc))
-		return POC_CMP(poca, pocb);
+		return HANTRO_CMP(poca, pocb);
 	else if (poca < builder->curpoc)
-		return POC_CMP(pocb, poca);
+		return HANTRO_CMP(pocb, poca);
 
-	return POC_CMP(poca, pocb);
+	return HANTRO_CMP(poca, pocb);
 }
 
 static int b1_ref_list_cmp(const void *ptra, const void *ptrb, const void *data)
@@ -422,22 +416,22 @@ static int b1_ref_list_cmp(const void *ptra, const void *ptrb, const void *data)
 
 	/* Long term pics in ascending pic num order. */
 	if (a->flags & V4L2_H264_DPB_ENTRY_FLAG_LONG_TERM)
-		return a->pic_num - b->pic_num;
+		return HANTRO_CMP(a->pic_num, b->pic_num);
 
 	poca = builder->pocs[idxa];
 	pocb = builder->pocs[idxb];
 
 	/*
 	 * Short term pics with POC > cur POC first in POC ascending order
-	 * followed by short term pics with POC > cur POC in POC descending
+	 * followed by short term pics with POC < cur POC in POC descending
 	 * order.
 	 */
 	if ((poca < builder->curpoc) != (pocb < builder->curpoc))
-		return POC_CMP(pocb, poca);
+		return HANTRO_CMP(pocb, poca);
 	else if (poca < builder->curpoc)
-		return POC_CMP(pocb, poca);
+		return HANTRO_CMP(pocb, poca);
 
-	return POC_CMP(poca, pocb);
+	return HANTRO_CMP(poca, pocb);
 }
 
 static void
@@ -537,22 +531,18 @@ static void update_dpb(struct hantro_ctx *ctx)
 	}
 }
 
-struct vb2_buffer *hantro_h264_get_ref_buf(struct hantro_ctx *ctx,
-					   unsigned int dpb_idx)
+dma_addr_t hantro_h264_get_ref_buf(struct hantro_ctx *ctx,
+				   unsigned int dpb_idx)
 {
-	struct vb2_queue *cap_q = &ctx->fh.m2m_ctx->cap_q_ctx.q;
 	struct v4l2_h264_dpb_entry *dpb = ctx->h264_dec.dpb;
-	struct vb2_buffer *buf;
-	int buf_idx = -1;
+	dma_addr_t dma_addr = 0;
 
 	if (dpb[dpb_idx].flags & V4L2_H264_DPB_ENTRY_FLAG_ACTIVE)
-		buf_idx = vb2_find_timestamp(cap_q,
-					     dpb[dpb_idx].reference_ts, 0);
+		dma_addr = hantro_get_ref(ctx, dpb[dpb_idx].reference_ts);
 
-	if (buf_idx >= 0) {
-		buf = vb2_get_buffer(cap_q, buf_idx);
-	} else {
+	if (!dma_addr) {
 		struct vb2_v4l2_buffer *dst_buf;
+		struct vb2_buffer *buf;
 
 		/*
 		 * If a DPB entry is unused or invalid, address of current
@@ -560,9 +550,10 @@ struct vb2_buffer *hantro_h264_get_ref_buf(struct hantro_ctx *ctx,
 		 */
 		dst_buf = hantro_get_dst_buf(ctx);
 		buf = &dst_buf->vb2_buf;
+		dma_addr = vb2_dma_contig_plane_dma_addr(buf, 0);
 	}
 
-	return buf;
+	return dma_addr;
 }
 
 int hantro_h264_dec_prepare_run(struct hantro_ctx *ctx)
@@ -627,7 +618,6 @@ int hantro_h264_dec_init(struct hantro_ctx *ctx)
 	struct hantro_h264_dec_hw_ctx *h264_dec = &ctx->h264_dec;
 	struct hantro_aux_buf *priv = &h264_dec->priv;
 	struct hantro_h264_dec_priv_tbl *tbl;
-	struct v4l2_pix_format_mplane pix_mp;
 
 	priv->cpu = dma_alloc_coherent(vpu->dev, sizeof(*tbl), &priv->dma,
 				       GFP_KERNEL);
@@ -638,9 +628,5 @@ int hantro_h264_dec_init(struct hantro_ctx *ctx)
 	tbl = priv->cpu;
 	memcpy(tbl->cabac_table, h264_cabac_table, sizeof(tbl->cabac_table));
 
-	v4l2_fill_pixfmt_mp(&pix_mp, ctx->dst_fmt.pixelformat,
-			    ctx->dst_fmt.width, ctx->dst_fmt.height);
-	h264_dec->pic_size = pix_mp.plane_fmt[0].sizeimage;
-
 	return 0;
 }
diff --git a/drivers/staging/media/hantro/hantro_hw.h b/drivers/staging/media/hantro/hantro_hw.h
index 2fab655bf098..fa91dd1848b7 100644
--- a/drivers/staging/media/hantro/hantro_hw.h
+++ b/drivers/staging/media/hantro/hantro_hw.h
@@ -80,15 +80,12 @@ struct hantro_h264_dec_reflists {
  * @dpb:	DPB
  * @reflists:	P/B0/B1 reflists
  * @ctrls:	V4L2 controls attached to a run
- * @pic_size:	Size in bytes of decoded picture, this is needed
- *		to pass the location of motion vectors.
  */
 struct hantro_h264_dec_hw_ctx {
 	struct hantro_aux_buf priv;
 	struct v4l2_h264_dpb_entry dpb[HANTRO_H264_DPB_SIZE];
 	struct hantro_h264_dec_reflists reflists;
 	struct hantro_h264_dec_ctrls ctrls;
-	size_t pic_size;
 };
 
 /**
@@ -158,8 +155,8 @@ v