Re: [PATCH 2/3] media: mtk-jpegenc: use component framework to manage jpg HW

From: Tzung-Bi Shih
Date: Fri Jun 25 2021 - 05:18:53 EST


On Wed, Jun 23, 2021 at 2:06 PM kyrie.wu <kyrie.wu@xxxxxxxxxxxx> wrote:
> Mtk jpeg encoder has several hardware, one HW may register a device
> node; use component framework to manage jpg HW device node,
> in this case, one device node could represent all jpg HW.
Can roughly understand. But could you rephrase your sentences?

> #include <media/videobuf2-core.h>
> #include <media/videobuf2-dma-contig.h>
> #include <soc/mediatek/smi.h>
> +#include <linux/component.h>
Maintain the alphabetical order.

> +void mtk_jpeg_put_buf(struct mtk_jpeg_dev *jpeg)
> +{
> + struct mtk_jpeg_ctx *ctx = NULL;
> + struct vb2_v4l2_buffer *dst_buffer = NULL;
> + struct list_head *temp_entry = NULL;
> + struct list_head *pos = NULL;
> + struct mtk_jpeg_src_buf *dst_done_buf = NULL, *tmp_dst_done_buf = NULL;
Remove the initialization if they don't need to.

> + unsigned long flags;
> +
> + ctx = jpeg->hw_param.curr_ctx;
> + if (!ctx) {
> + pr_err("%s : %d, comp_jpeg ctx fail !!!\n", __func__, __LINE__);
Use dev_err().

> + return;
> + }
> +
> + dst_buffer = jpeg->hw_param.dst_buffer;
> + if (!dst_buffer) {
> + pr_err("%s : %d, comp_jpeg dst_buffer fail !!!\n",
> + __func__, __LINE__);
Use dev_err().

> + if (!(irq_status & JPEG_ENC_INT_STATUS_DONE)) {
> + pr_err("%s : %d, jpeg encode failed\n", __func__, __LINE__);
Use dev_err().

> +void mtk_jpegenc_timeout_work(struct work_struct *work)
> +{
> + struct mtk_jpeg_dev *jpeg = container_of(work, struct mtk_jpeg_dev,
> + job_timeout_work.work);
> + struct mtk_jpeg_ctx *ctx = NULL;
It doesn't need to initialize.

> +static const struct of_device_id mtk_jpegenc_drv_ids[] = {
Remove the extra space between "static" and "const".

> + {
> + .compatible = "mediatek,mt8195-jpgenc0",
> + .data = (void *)MTK_JPEGENC_HW0,
> + },
> + {
> + .compatible = "mediatek,mt8195-jpgenc1",
> + .data = (void *)MTK_JPEGENC_HW1,
> + },
Using compatible strings to separate them doesn't sound like a scalable method.

> #include <linux/kernel.h>
> #include <media/videobuf2-core.h>
> #include <media/videobuf2-dma-contig.h>
> +#include <linux/interrupt.h>
> +#include <linux/irq.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/of_device.h>
> +#include <linux/slab.h>
> +#include <linux/component.h>
> +#include <linux/clk.h>
> +#include <linux/pm_runtime.h>
Maintain the alphabetical order.

> #include "mtk_jpeg_enc_hw.h"
> +#include "mtk_jpeg_core.h"
Maintain the alphabetical order.

> +int mtk_jpegenc_init_pm(struct mtk_jpeg_dev *mtkdev)
> +{
> + struct platform_device *pdev;
> + struct mtk_jpegenc_pm *pm;
> + struct mtk_jpegenc_clk *jpegenc_clk;
> + struct mtk_jpegenc_clk_info *clk_info;
> + int i = 0, ret = 0;
They don't need to initialize.

> + pdev = mtkdev->plat_dev;
> + pm = &mtkdev->pm;
To be concise, can inline to above when declaring the variables.

> + jpegenc_clk->clk_num =
> + of_property_count_strings(pdev->dev.of_node, "clock-names");
> + if (jpegenc_clk->clk_num > 0) {
> + jpegenc_clk->clk_info = devm_kcalloc(&pdev->dev,
> + jpegenc_clk->clk_num,
> + sizeof(*clk_info),
> + GFP_KERNEL);
> + if (!jpegenc_clk->clk_info)
> + return -ENOMEM;
> + } else {
> + pr_err("Failed to get jpegenc clock count\n");
Use dev_err().
> + return -EINVAL;
> + }
Would prefer the block turn to be:

if (... <= 0) {
...
return -EINVAL;
}

... = devm_kcalloc(...);
if (!...)
return -ENOMEM;

> + for (i = 0; i < jpegenc_clk->clk_num; i++) {
> + clk_info = &jpegenc_clk->clk_info[i];
> + ret = of_property_read_string_index(pdev->dev.of_node,
> + "clock-names", i,
> + &clk_info->clk_name);
> + if (ret) {
> + pr_err("Failed to get jpegenc clock name id = %d", i);
Use dev_err().

> + return ret;
> + }
> +
> + clk_info->jpegenc_clk = devm_clk_get(&pdev->dev,
> + clk_info->clk_name);
> + if (IS_ERR(clk_info->jpegenc_clk)) {
> + pr_err("devm_clk_get (%d)%s fail",
> + i, clk_info->clk_name);
Use dev_err().

> + pm_runtime_enable(&pdev->dev);
> + return ret;
return 0;

> +void mtk_jpegenc_release_pm(struct mtk_jpeg_dev *dev)
> +{
> + pm_runtime_disable(dev->pm.dev);
> +}
Would prefer this function to be more "symmetric" to mtk_jpegenc_init_pm().

For example:

void mtk_jpegenc_release_pm(struct mtk_jpeg_dev *mtkdev)
{
struct platform_device *pdev = mtkdev->plat_dev;
pm_runtime_disable(&pdev->dev);
}

That way, it doesn't rely on whether mtkdev->pm is set or not.

> + ret = devm_request_irq(&pdev->dev, dev->jpegenc_irq,
> + irq_handler, 0, pdev->name, dev);
> + if (ret) {
> + dev_err(&pdev->dev, "Failed to install dev->jpegenc_irq %d (%d)",
> + dev->jpegenc_irq, ret);
> +
> + return -ENOENT;
How about just returning ret?

> + }
> +
> + //disable_irq(dev->jpegenc_irq);
Remove it.

> + ret = component_add(&pdev->dev, &mtk_jpegenc_hw_component_ops);
> + if (ret) {
> + dev_err(&pdev->dev, "Failed to component_add: %d\n", ret);
> + goto err;
> + }
How about just returning component_add(...)?

> +err:
> + mtk_jpegenc_release_pm(dev);
Would expect the platform driver to have a .remove() callback and
invoke the mtk_jpegenc_release_pm() too.

> +static const struct of_device_id mtk_jpegenc_hw_ids[] = {
> + {
> + .compatible = "mediatek,mt8195-jpgenc0",
> + .data = mtk_jpegenc_hw_irq_handler,
> + },
> + { .compatible = "mediatek,mt8195-jpgenc1",
> + .data = mtk_jpegenc_hw_irq_handler,
> + },
> + {},
> +};
> +MODULE_DEVICE_TABLE(of, mtk_jpegenc_hw_ids);
Had the same concern in dt-bindings patch. Does it really need
multiple compatible strings to separate?

Also, the block should guard by using CONFIG_OF.

> +struct platform_driver mtk_jpegenc_hw_driver = {
> + .probe = mtk_jpegenc_hw_probe,
> + .driver = {
> + .name = "mtk-jpegenc-hw",
> + .of_match_table = mtk_jpegenc_hw_ids,
Should guard by using of_match_ptr().


Hi, after reading the patch for a while, I realized it is way too big
to me so that I didn't go through too much detail (especially the
component framework part). Could you further divide the series into
smaller pieces? For example:
- part i. refactor to make modifying code easier
- part ii. leverage component framework
- part iii. add new code for MT8195
I would expect part i and ii don't change the original behavior.