Re: [RFC PATCH] fpga: Introduce new fpga subsystem

[Michal Simek]

Hi Joe,

On 09/18/2013 06:11 PM, Joe Perches wrote:
> On Wed, 2013-09-18 at 17:56 +0200, Michal Simek wrote:
>> This new subsystem should unify all fpga drivers which
>> do the same things. Load configuration data to fpga
>> or another programmable logic through common interface.
>> It doesn't matter if it is MMIO device, gpio bitbanging,
>> etc. connection. The point is to have the same
>> inteface for these drivers.
> 
> Is this really a "driver".
> Perhaps it's more of a core kernel function
> and belongs in kernel.

This is not a driver just kernel subsystem and real drivers
will register itself in this subsystem.

> 
>> diff --git a/drivers/fpga/fpga-mgr.c b/drivers/fpga/fpga-mgr.c
> 
> The error messages are a little shouty with all
> the unnecessary "!" uses.  
> 
> []

Fixed.

> 
>> +/**
>> + * fpga_mgr_read - Read data from fpga
>> + * @mgr: Pointer to the fpga manager structure
>> + * @buf: Pointer to the buffer location
>> + * @count: Pointer to the number of copied bytes
>> + *
>> + * Function reads fpga bitstream and copy them to output buffer.
>> + *
>> + * Returns the number of bytes copied to @buf, a negative error number otherwise
>> + */
>> +static int fpga_mgr_read(struct fpga_manager *mgr, char *buf, ssize_t *count)
>> +{
>> +	int ret;
>> +
>> +	if (test_and_set_bit_lock(FPGA_MGR_DEV_BUSY, &mgr->flags))
>> +		return -EBUSY;
>> +
>> +	if (!mgr->mops || !mgr->mops->read) {
>> +		dev_err(mgr->dev,
>> +			"Controller doesn't support read operations!\n");
>> +		return -EPERM;
>> +	}
>> +
>> +	if (mgr->mops->read_init) {
>> +		ret = mgr->mops->read_init(mgr);
>> +		if (ret) {
>> +			dev_err(mgr->dev, "Failed in read-init!\n");
>> +			return ret;
>> +		}
>> +	}
>> +
>> +	if (mgr->mops->read) {
>> +		ret = mgr->mops->read(mgr, buf, count);
>> +		if (ret) {
>> +			dev_err(mgr->dev, "Failed to read firmware!\n");
>> +			return ret;
>> +		}
>> +	}
>> +
>> +	if (mgr->mops->read_complete) {
>> +		ret = mgr->mops->read_complete(mgr);
>> +		if (ret) {
>> +			dev_err(mgr->dev, "Failed in read-complete!\n");
>> +			return ret;
>> +		}
>> +	}
>> +
>> +	clear_bit_unlock(FPGA_MGR_DEV_BUSY, &mgr->flags);
>> +
>> +	return 0;
>> +}
>> +
>> +/**
>> + * fpga_mgr_attr_read - Read data from fpga
>> + * @dev: Pointer to the device structure
>> + * @attr: Pointer to the device attribute structure
>> + * @buf: Pointer to the buffer location
>> + *
>> + * Function reads fpga bitstream and copy them to output buffer
>> + *
>> + * Returns the number of bytes copied to @buf, a negative error number otherwise
>> + */
>> +static ssize_t fpga_mgr_attr_read(struct device *dev,
>> +				   struct device_attribute *attr, char *buf)
>> +{
>> +	struct fpga_manager *mgr = dev_get_drvdata(dev);
>> +	ssize_t count;
>> +	int ret;
>> +
>> +	if (mgr && mgr->fpga_read)
>> +		ret = mgr->fpga_read(mgr, buf, &count);
>> +
>> +	return ret == 0 ? count : -EPERM;
> 
> EPERM isn't the only error return from fpga_read.

Yeah I know. Will revisit all return codes.


>> +}
>> +
>> +/**
>> + * fpga_mgr_write - Write data to fpga
>> + * @mgr: Pointer to the fpga manager structure
>> + * @fw_name: Pointer to the buffer location with bistream firmware filename
>> + *
>> + * @buf contains firmware filename which is loading through firmware
>> + * interface and passed to the fpga driver.
>> + *
>> + * Returns string lenght added to @fw_name, a negative error number otherwise
>> + */
>> +static int fpga_mgr_write(struct fpga_manager *mgr, const char *fw_name)
>> +{
>> +	int ret = 0;
>> +	const struct firmware *fw;
>> +
>> +	if (!fw_name || !strlen(fw_name)) {
>> +		dev_err(mgr->dev, "Firmware name is not specified!\n");
>> +		return -EINVAL;
>> +	}
>> +
>> +	if (!mgr->mops || !mgr->mops->write) {
>> +		dev_err(mgr->dev,
>> +			"Controller doesn't support write operations!\n");
>> +		return -EPERM;
> 
> I think you should revisit the return codes.

yap

> 
> 
>> +/**
>> + * fpga_mgr_attr_write - Write data to fpga
>> + * @dev: Pointer to the device structure
>> + * @attr: Pointer to the device attribute structure
>> + * @buf: Pointer to the buffer location with bistream firmware filename
>> + * @count: Number of characters in @buf
>> + *
>> + * @buf contains firmware filename which is loading through firmware
>> + * interface and passed to the fpga driver.
>> + *
>> + * Returns string lenght added to @buf, a negative error number otherwise
>> + */
>> +static ssize_t fpga_mgr_attr_write(struct device *dev,
>> +				   struct device_attribute *attr,
>> +				   const char *buf, size_t count)
>> +{
>> +	struct fpga_manager *mgr = dev_get_drvdata(dev);
>> +	int ret;
>> +
>> +	if (mgr && mgr->fpga_write)
>> +		ret = mgr->fpga_write(mgr, buf);
>> +
>> +	return ret == 0 ? strlen(buf) : -EPERM;
>> +}
> 
> Same -EPERM issue as read

yap

> 
>> +	mgr = devm_kzalloc(&pdev->dev, sizeof(*mgr), GFP_KERNEL);
>> +	if (!mgr) {
>> +		dev_err(&pdev->dev, "Failed to alloc mgr struct!\n");
> 
> Unnecessary OOM message as there's a general dump_stack()
> already done on any OOM without GFP_NOWARN
> 
>> diff --git a/include/linux/fpga.h b/include/linux/fpga.h
> []
>> +struct fpga_manager {
>> +	char name[48];
> 
> Maybe a #define instead of 48?

There is another comment to fix this by pointer which is sensible solution too.

Thanks for review,
Michal

-- 
Michal Simek, Ing. (M.Eng), OpenPGP -> KeyID: FE3D1F91
w: www.monstr.eu p: +42-0-721842854
Maintainer of Linux kernel - Microblaze cpu - http://www.monstr.eu/fdt/
Maintainer of Linux kernel - Xilinx Zynq ARM architecture
Microblaze U-BOOT custodian and responsible for u-boot arm zynq platform

Attachment: signature.asc
Description: OpenPGP digital signature