Re: [PATCH v6 4/9] media: venus: adding core part and helper functions

From: Bjorn Andersson
Date: Tue Feb 07 2017 - 18:40:22 EST


On Tue 07 Feb 05:10 PST 2017, Stanimir Varbanov wrote:

> * firmware loader
>

I like the way this turns out, just some style comments below.

[..]
> diff --git a/drivers/media/platform/qcom/venus/firmware.c b/drivers/media/platform/qcom/venus/firmware.c
> new file mode 100644
> index 000000000000..4057696abaf5
> --- /dev/null
> +++ b/drivers/media/platform/qcom/venus/firmware.c
> @@ -0,0 +1,151 @@
> +/*
> + * Copyright (C) 2017 Linaro Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + *
> + */
> +
> +#include <linux/dma-mapping.h>
> +#include <linux/firmware.h>
> +#include <linux/kernel.h>
> +#include <linux/of.h>
> +#include <linux/of_reserved_mem.h>
> +#include <linux/slab.h>
> +#include <linux/qcom_scm.h>
> +#include <linux/soc/qcom/mdt_loader.h>
> +
> +#define VENUS_FIRMWARE_NAME "venus.mdt"
> +#define VENUS_PAS_ID 9
> +#define VENUS_FW_MEM_SIZE SZ_8M
> +
> +struct firmware_mem {
> + struct device dev;
> + void *mem_va;
> + phys_addr_t mem_phys;
> + size_t mem_size;
> +};
> +
> +static struct firmware_mem fw;

Rather than operating on a global variable I think you should either
return your firmware_mem pointer or the device pointer to the caller of
venus_boot() and have the core pass that back into venus_shutdown().

> +
> +static void device_release_dummy(struct device *dev)
> +{
> +}
> +
> +static int firmware_alloc_mem(struct device *parent, struct firmware_mem *fw)
> +{
> + struct device_node *np;
> + struct device *dev = &fw->dev;
> + int ret;
> +
> + np = of_get_child_by_name(parent->of_node, "video-firmware");
> + if (!np)
> + return -ENODEV;
> +
> + memset(fw, 0, sizeof(*fw));

This should not be necessary.

Further more, if it's already initialized it's safe to assume that the
allocation below will fail - regardless of you clearing it or not.

> +
> + dev->of_node = np;
> + dev->parent = parent;
> + dev->release = device_release_dummy;
> +
> + ret = dev_set_name(dev, "venus-fw");
> + if (ret)
> + return ret;
> +
> + ret = device_register(dev);
> + if (ret < 0)
> + return ret;
> +
> + ret = of_reserved_mem_device_init(dev);
> + if (ret)
> + goto err_unreg_device;
> +
> + fw->mem_size = VENUS_FW_MEM_SIZE;
> +
> + fw->mem_va = dma_alloc_coherent(dev, fw->mem_size, &fw->mem_phys,
> + GFP_KERNEL);

As this should follow the life of dev you can use dmam_alloc_coherent()
to reduce the clean up paths.

> + if (!fw->mem_va) {
> + ret = -ENOMEM;
> + goto err_mem_device_release;
> + }
> +
> + return 0;
> +
> +err_mem_device_release:
> + of_reserved_mem_device_release(dev);
> +err_unreg_device:
> + device_unregister(dev);
> + return ret;
> +}
> +
> +static void firmware_free_mem(struct firmware_mem *fw)
> +{
> + dma_free_coherent(&fw->dev, fw->mem_size, fw->mem_va, fw->mem_phys);

If you use dmam_alloc_coherent() this goes.

> + of_reserved_mem_device_release(&fw->dev);

And I would suggest that as this is related to the device you should
release it in the dev->release function; turning this function into
device_unregister(&fw->dev).


(The devres allocation will be freed right before the release function
is called)

> + device_unregister(&fw->dev);
> + memset(fw, 0, sizeof(*fw));

This should not be necessary.

> +}
> +
> +static int firmware_load(struct firmware_mem *fw)
> +{
> + struct device *dev = &fw->dev;
> + const struct firmware *mdt;
> + ssize_t fw_size;
> + int ret;
> +
> + ret = request_firmware(&mdt, VENUS_FIRMWARE_NAME, dev);
> + if (ret < 0)
> + return ret;
> +
> + fw_size = qcom_mdt_get_size(mdt);
> + if (fw_size < 0) {
> + ret = fw_size;
> + goto err_release_fw;
> + } else if (fw_size > VENUS_FW_MEM_SIZE) {

You can skip this this check, as qcom_mdt_load() will fail if any part
of the firmware doesn't fit - and we would benefit from making that
error message more verbose.

> + ret = -ENOMEM;
> + goto err_release_fw;
> + }
> +
> + ret = qcom_mdt_load(&fw->dev, mdt, VENUS_FIRMWARE_NAME, VENUS_PAS_ID,
> + fw->mem_va, fw->mem_phys, fw->mem_size);
> +
> +err_release_fw:

This is not only the error path, so "release_fw" would be better.

> + release_firmware(mdt);
> +
> + return ret;
> +}
> +
> +int venus_boot(struct device *parent)
> +{
> + int ret;
> +
> + if (!qcom_scm_is_available())
> + return -EPROBE_DEFER;
> +
> + ret = firmware_alloc_mem(parent, &fw);
> + if (ret)
> + return ret;
> +
> + ret = firmware_load(&fw);
> + if (ret) {
> + firmware_free_mem(&fw);
> + return ret;
> + }
> +
> + return qcom_scm_pas_auth_and_reset(VENUS_PAS_ID);
> +}
> +
> +int venus_shutdown(void)
> +{
> + int ret;
> +
> + ret = qcom_scm_pas_shutdown(VENUS_PAS_ID);
> + firmware_free_mem(&fw);
> + return ret;
> +}

Regards,
Bjorn