media: mtk-jpegdec: add jpeg decode worker interface

Add jpeg decoding worker to ensure that three HWs
run in parallel in MT8195.
Signed-off-by: kyrie wu <kyrie.wu@mediatek.com>
Signed-off-by: irui wang <irui.wang@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c

@@ -1100,6 +1100,180 @@ static void mtk_jpeg_multicore_enc_device_run(void *priv)
 	queue_work(jpeg->workqueue, &ctx->jpeg_work);
 }
 
+static int mtk_jpegdec_get_hw(struct mtk_jpeg_ctx *ctx)
+{
+	struct mtk_jpegdec_comp_dev *comp_jpeg;
+	struct mtk_jpeg_dev *jpeg = ctx->jpeg;
+	unsigned long flags;
+	int hw_id = -1;
+	int i;
+
+	spin_lock_irqsave(&jpeg->hw_lock, flags);
+	for (i = 0; i < MTK_JPEGDEC_HW_MAX; i++) {
+		comp_jpeg = jpeg->dec_hw_dev[i];
+		if (comp_jpeg->hw_state == MTK_JPEG_HW_IDLE) {
+			hw_id = i;
+			comp_jpeg->hw_state = MTK_JPEG_HW_BUSY;
+			break;
+		}
+	}
+	spin_unlock_irqrestore(&jpeg->hw_lock, flags);
+
+	return hw_id;
+}
+
+static int mtk_jpegdec_put_hw(struct mtk_jpeg_dev *jpeg, int hw_id)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&jpeg->hw_lock, flags);
+	jpeg->dec_hw_dev[hw_id]->hw_state =
+		MTK_JPEG_HW_IDLE;
+	spin_unlock_irqrestore(&jpeg->hw_lock, flags);
+
+	return 0;
+}
+
+static int mtk_jpegdec_set_hw_param(struct mtk_jpeg_ctx *ctx,
+				    int hw_id,
+				    struct vb2_v4l2_buffer *src_buf,
+				    struct vb2_v4l2_buffer *dst_buf)
+{
+	struct mtk_jpegdec_comp_dev *jpeg =
+		ctx->jpeg->dec_hw_dev[hw_id];
+
+	jpeg->hw_param.curr_ctx = ctx;
+	jpeg->hw_param.src_buffer = src_buf;
+	jpeg->hw_param.dst_buffer = dst_buf;
+
+	return 0;
+}
+
+static void mtk_jpegdec_worker(struct work_struct *work)
+{
+	struct mtk_jpeg_ctx *ctx = container_of(work, struct mtk_jpeg_ctx,
+		jpeg_work);
+	struct mtk_jpegdec_comp_dev *comp_jpeg[MTK_JPEGDEC_HW_MAX];
+	enum vb2_buffer_state buf_state = VB2_BUF_STATE_ERROR;
+	struct mtk_jpeg_src_buf *jpeg_src_buf, *jpeg_dst_buf;
+	struct vb2_v4l2_buffer *src_buf, *dst_buf;
+	struct mtk_jpeg_dev *jpeg = ctx->jpeg;
+	int ret, i, hw_id = 0;
+	struct mtk_jpeg_bs bs;
+	struct mtk_jpeg_fb fb;
+	unsigned long flags;
+
+	for (i = 0; i < MTK_JPEGDEC_HW_MAX; i++)
+		comp_jpeg[i] = jpeg->dec_hw_dev[i];
+	i = 0;
+
+retry_select:
+	hw_id = mtk_jpegdec_get_hw(ctx);
+	if (hw_id < 0) {
+		ret = wait_event_interruptible_timeout(jpeg->dec_hw_wq,
+						       atomic_read(&jpeg->dechw_rdy) > 0,
+						       MTK_JPEG_HW_TIMEOUT_MSEC);
+		if (ret != 0 || (i++ > MTK_JPEG_MAX_RETRY_TIME)) {
+			dev_err(jpeg->dev, "%s : %d, all HW are busy\n",
+				__func__, __LINE__);
+			v4l2_m2m_job_finish(jpeg->m2m_dev, ctx->fh.m2m_ctx);
+			return;
+		}
+
+		goto retry_select;
+	}
+
+	atomic_dec(&jpeg->dechw_rdy);
+	src_buf = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
+	if (!src_buf)
+		goto getbuf_fail;
+
+	dst_buf = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx);
+	if (!dst_buf)
+		goto getbuf_fail;
+
+	v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
+	v4l2_m2m_dst_buf_remove(ctx->fh.m2m_ctx);
+
+	v4l2_m2m_buf_copy_metadata(src_buf, dst_buf, true);
+	jpeg_src_buf = mtk_jpeg_vb2_to_srcbuf(&src_buf->vb2_buf);
+	jpeg_dst_buf = mtk_jpeg_vb2_to_srcbuf(&dst_buf->vb2_buf);
+
+	if (mtk_jpeg_check_resolution_change(ctx,
+					     &jpeg_src_buf->dec_param)) {
+		mtk_jpeg_queue_src_chg_event(ctx);
+		ctx->state = MTK_JPEG_SOURCE_CHANGE;
+		goto dec_end;
+	}
+
+	jpeg_src_buf->curr_ctx = ctx;
+	jpeg_src_buf->frame_num = ctx->total_frame_num;
+	jpeg_dst_buf->curr_ctx = ctx;
+	jpeg_dst_buf->frame_num = ctx->total_frame_num;
+	ctx->total_frame_num++;
+
+	mtk_jpegdec_set_hw_param(ctx, hw_id, src_buf, dst_buf);
+	ret = pm_runtime_get_sync(comp_jpeg[hw_id]->dev);
+	if (ret < 0) {
+		dev_err(jpeg->dev, "%s : %d, pm_runtime_get_sync fail !!!\n",
+			__func__, __LINE__);
+		goto dec_end;
+	}
+
+	ret = clk_prepare_enable(comp_jpeg[hw_id]->jdec_clk.clks->clk);
+	if (ret) {
+		dev_err(jpeg->dev, "%s : %d, jpegdec clk_prepare_enable fail\n",
+			__func__, __LINE__);
+		goto clk_end;
+	}
+
+	schedule_delayed_work(&comp_jpeg[hw_id]->job_timeout_work,
+			      msecs_to_jiffies(MTK_JPEG_HW_TIMEOUT_MSEC));
+
+	mtk_jpeg_set_dec_src(ctx, &src_buf->vb2_buf, &bs);
+	if (mtk_jpeg_set_dec_dst(ctx,
+				 &jpeg_src_buf->dec_param,
+				 &dst_buf->vb2_buf, &fb)) {
+		dev_err(jpeg->dev, "%s : %d, mtk_jpeg_set_dec_dst fail\n",
+			__func__, __LINE__);
+		goto setdst_end;
+	}
+
+	spin_lock_irqsave(&comp_jpeg[hw_id]->hw_lock, flags);
+	mtk_jpeg_dec_reset(comp_jpeg[hw_id]->reg_base);
+	mtk_jpeg_dec_set_config(jpeg->reg_base,
+				&jpeg_src_buf->dec_param,
+				&bs,
+				&fb);
+	mtk_jpeg_dec_start(comp_jpeg[hw_id]->reg_base);
+	v4l2_m2m_job_finish(jpeg->m2m_dev, ctx->fh.m2m_ctx);
+	spin_unlock_irqrestore(&comp_jpeg[hw_id]->hw_lock, flags);
+
+	return;
+
+setdst_end:
+	clk_disable_unprepare(comp_jpeg[hw_id]->jdec_clk.clks->clk);
+clk_end:
+	pm_runtime_put(comp_jpeg[hw_id]->dev);
+dec_end:
+	v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
+	v4l2_m2m_dst_buf_remove(ctx->fh.m2m_ctx);
+	v4l2_m2m_buf_done(src_buf, buf_state);
+	v4l2_m2m_buf_done(dst_buf, buf_state);
+getbuf_fail:
+	atomic_inc(&jpeg->dechw_rdy);
+	mtk_jpegdec_put_hw(jpeg, hw_id);
+	v4l2_m2m_job_finish(jpeg->m2m_dev, ctx->fh.m2m_ctx);
+}
+
+static void mtk_jpeg_multicore_dec_device_run(void *priv)
+{
+	struct mtk_jpeg_ctx *ctx = priv;
+	struct mtk_jpeg_dev *jpeg = ctx->jpeg;
+
+	queue_work(jpeg->workqueue, &ctx->jpeg_work);
+}
+
 static void mtk_jpeg_dec_device_run(void *priv)
 {
 	struct mtk_jpeg_ctx *ctx = priv;
@@ -1166,6 +1340,10 @@ static const struct v4l2_m2m_ops mtk_jpeg_multicore_enc_m2m_ops = {
 	.device_run = mtk_jpeg_multicore_enc_device_run,
 };
 
+static const struct v4l2_m2m_ops mtk_jpeg_multicore_dec_m2m_ops = {
+	.device_run = mtk_jpeg_multicore_dec_device_run,
+};
+
 static const struct v4l2_m2m_ops mtk_jpeg_dec_m2m_ops = {
 	.device_run = mtk_jpeg_dec_device_run,
 	.job_ready  = mtk_jpeg_dec_job_ready,
@@ -1369,6 +1547,9 @@ static int mtk_jpeg_open(struct file *file)
 	if (jpeg->is_jpgenc_multihw)
 		INIT_WORK(&ctx->jpeg_work, mtk_jpegenc_worker);
 
+	if (jpeg->is_jpgdec_multihw)
+		INIT_WORK(&ctx->jpeg_work, mtk_jpegdec_worker);
+
 	INIT_LIST_HEAD(&ctx->dst_done_queue);
 	spin_lock_init(&ctx->done_queue_lock);
 	v4l2_fh_init(&ctx->fh, vfd);
@@ -1675,6 +1856,17 @@ static struct mtk_jpeg_variant mtk8195_jpegenc_drvdata = {
 	.cap_q_default_fourcc = V4L2_PIX_FMT_JPEG,
 };
 
+static const struct mtk_jpeg_variant mtk8195_jpegdec_drvdata = {
+	.formats = mtk_jpeg_dec_formats,
+	.num_formats = MTK_JPEG_DEC_NUM_FORMATS,
+	.qops = &mtk_jpeg_dec_qops,
+	.m2m_ops = &mtk_jpeg_multicore_dec_m2m_ops,
+	.dev_name = "mtk-jpeg-dec",
+	.ioctl_ops = &mtk_jpeg_dec_ioctl_ops,
+	.out_q_default_fourcc = V4L2_PIX_FMT_JPEG,
+	.cap_q_default_fourcc = V4L2_PIX_FMT_YUV420M,
+};
+
 #if defined(CONFIG_OF)
 static const struct of_device_id mtk_jpeg_match[] = {
 	{
@@ -1693,6 +1885,10 @@ static const struct of_device_id mtk_jpeg_match[] = {
 		.compatible = "mediatek,mt8195-jpgenc",
 		.data = &mtk8195_jpegenc_drvdata,
 	},
+	{
+		.compatible = "mediatek,mt8195-jpgdec",
+		.data = &mtk8195_jpegdec_drvdata,
+	},
 	{},
 };
 

drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.h

@@ -167,6 +167,8 @@ struct mtk_jpegenc_comp_dev {
  * @jpegdec_irq:		jpeg decode irq num
  * @job_timeout_work:		decode timeout workqueue
  * @hw_param:			jpeg decode hw parameters
+ * @hw_state:			record hw state
+ * @hw_lock:			spinlock protecting hw
  */
 struct mtk_jpegdec_comp_dev {
 	struct device *dev;
@@ -177,6 +179,9 @@ struct mtk_jpegdec_comp_dev {
 	int jpegdec_irq;
 	struct delayed_work job_timeout_work;
 	struct mtk_jpeg_hw_param hw_param;
+	enum mtk_jpeg_hw_state hw_state;
+	/* spinlock protecting the hw device resource */
+	spinlock_t hw_lock;
 };
 
 /**
@@ -200,6 +205,9 @@ struct mtk_jpegdec_comp_dev {
  * @reg_decbase:	jpg decode register base addr
  * @dec_hw_dev:		jpg decode hardware device
  * @is_jpgdec_multihw:	the flag of dec multi-hw core
+ * @dec_hw_wq:		jpg decode wait queue
+ * @dec_workqueue:	jpg decode work queue
+ * @dechw_rdy:		jpg decode hw ready flag
  */
 struct mtk_jpeg_dev {
 	struct mutex		lock;
@@ -223,6 +231,9 @@ struct mtk_jpeg_dev {
 	void __iomem *reg_decbase[MTK_JPEGDEC_HW_MAX];
 	struct mtk_jpegdec_comp_dev *dec_hw_dev[MTK_JPEGDEC_HW_MAX];
 	bool is_jpgdec_multihw;
+	wait_queue_head_t dec_hw_wq;
+	struct workqueue_struct	*dec_workqueue;
+	atomic_t dechw_rdy;
 };
 
 /**

drivers/media/platform/mediatek/jpeg/mtk_jpeg_dec_hw.c

@@ -446,6 +446,7 @@ static void mtk_jpegdec_timeout_work(struct work_struct *work)
 	struct mtk_jpegdec_comp_dev *cjpeg =
 		container_of(work, struct mtk_jpegdec_comp_dev,
 			     job_timeout_work.work);
+	struct mtk_jpeg_dev *master_jpeg = cjpeg->master_dev;
 	struct vb2_v4l2_buffer *src_buf, *dst_buf;
 
 	src_buf = cjpeg->hw_param.src_buffer;
@@ -455,6 +456,9 @@ static void mtk_jpegdec_timeout_work(struct work_struct *work)
 	mtk_jpeg_dec_reset(cjpeg->reg_base);
 	clk_disable_unprepare(cjpeg->jdec_clk.clks->clk);
 	pm_runtime_put(cjpeg->dev);
+	cjpeg->hw_state = MTK_JPEG_HW_IDLE;
+	atomic_inc(&master_jpeg->dechw_rdy);
+	wake_up(&master_jpeg->dec_hw_wq);
 	v4l2_m2m_buf_done(src_buf, buf_state);
 	v4l2_m2m_buf_done(dst_buf, buf_state);
 }
@@ -474,22 +478,19 @@ static irqreturn_t mtk_jpegdec_hw_irq_handler(int irq, void *priv)
 
 	cancel_delayed_work(&jpeg->job_timeout_work);
 
+	ctx = jpeg->hw_param.curr_ctx;
+	src_buf = jpeg->hw_param.src_buffer;
+	dst_buf = jpeg->hw_param.dst_buffer;
+	v4l2_m2m_buf_copy_metadata(src_buf, dst_buf, true);
+
 	irq_status = mtk_jpeg_dec_get_int_status(jpeg->reg_base);
 	dec_irq_ret = mtk_jpeg_dec_enum_result(irq_status);
 	if (dec_irq_ret >= MTK_JPEG_DEC_RESULT_UNDERFLOW)
 		mtk_jpeg_dec_reset(jpeg->reg_base);
-	if (dec_irq_ret != MTK_JPEG_DEC_RESULT_EOF_DONE)
-		return IRQ_NONE;
 
-	ctx = v4l2_m2m_get_curr_priv(master_jpeg->m2m_dev);
-	if (!ctx) {
-		dev_err(jpeg->dev, "Context is NULL\n");
-		return IRQ_HANDLED;
-	}
+	if (dec_irq_ret != MTK_JPEG_DEC_RESULT_EOF_DONE)
+		dev_warn(jpeg->dev, "Jpg Dec occurs unknown Err.");
 
-	src_buf = v4l2_m2m_src_buf_remove(ctx->fh.m2m_ctx);
-	dst_buf = v4l2_m2m_dst_buf_remove(ctx->fh.m2m_ctx);
-	v4l2_m2m_buf_copy_metadata(src_buf, dst_buf, true);
 	jpeg_src_buf =
 		container_of(src_buf, struct mtk_jpeg_src_buf, b);
 
@@ -498,11 +499,14 @@ static irqreturn_t mtk_jpegdec_hw_irq_handler(int irq, void *priv)
 				      jpeg_src_buf->dec_param.comp_size[i]);
 
 	buf_state = VB2_BUF_STATE_DONE;
-
 	v4l2_m2m_buf_done(src_buf, buf_state);
 	v4l2_m2m_buf_done(dst_buf, buf_state);
-	v4l2_m2m_job_finish(master_jpeg->m2m_dev, ctx->fh.m2m_ctx);
 	pm_runtime_put(ctx->jpeg->dev);
+	clk_disable_unprepare(jpeg->jdec_clk.clks->clk);
+
+	jpeg->hw_state = MTK_JPEG_HW_IDLE;
+	wake_up(&master_jpeg->dec_hw_wq);
+	atomic_inc(&master_jpeg->dechw_rdy);
 
 	return IRQ_HANDLED;
 }
@@ -557,8 +561,19 @@ static int mtk_jpegdec_hw_probe(struct platform_device *pdev)
 		master_dev->is_jpgdec_multihw = true;
 		for (i = 0; i < MTK_JPEGDEC_HW_MAX; i++)
 			master_dev->dec_hw_dev[i] = NULL;
+
+		init_waitqueue_head(&master_dev->dec_hw_wq);
+		master_dev->workqueue = alloc_ordered_workqueue(MTK_JPEG_NAME,
+								WQ_MEM_RECLAIM
+								| WQ_FREEZABLE);
+		if (!master_dev->workqueue)
+			return -EINVAL;
 	}
 
+	atomic_set(&master_dev->dechw_rdy, MTK_JPEGDEC_HW_MAX);
+	spin_lock_init(&dev->hw_lock);
+	dev->hw_state = MTK_JPEG_HW_IDLE;
+
 	INIT_DELAYED_WORK(&dev->job_timeout_work,
 			  mtk_jpegdec_timeout_work);