Parallel Port Device Driver Implementation

Normally when a computer is turned on, its the kernel who acquire the parallel ports and you can see it by going into proc/ioports. But if we want our device to communicate with the I/O port then we need to acquire those ports from the kernel. There are certain commands in linux by  which we can get those ports . All these commands are defined in <linux/ioport.h> and <linux/io.h>.

    struct resource *request_region(): Registration interface which allows your kernel to claim the ports it need.
    void release_region(): When things are done with the I/O ports, use this command to release them.
    int check_region(): If you want to check whether the requested address regions are available or not , use this command.
    unsigned inb() and void outb(): These commands are used to read/write to and from the parallel port. Used in byte transfer.
    unsigned inw() and void outw(): There are used to access 16 bit ports i.e one work transfer and receive.
    unsigned inl() and void outl(): these are used to access 32 bit ports or 32 bit longword.

// for parallel port check if ports are acquired or not
        //iomem_map_sanity_check(resource_size_t addr, unsigned long size);
        //ppret = check_region(DATA_PORT_ADD, 3);
        ppret = iomem_map_sanity_check(DATA_PORT_ADD, 3);
        if (ppret == 0)
        {
                request_region(DATA_PORT_ADD, 3, DEVNAME);
        }
        else
        {
                release_region(DATA_PORT_ADD,3);
                request_region(DATA_PORT_ADD, 3, DEVNAME);
        }

Since check_region did not work in 4.x kernel, i used
iomem_map_sanity_check(DATA_PORT_ADD, 3);
here DATA_PORT_ADD = 0x378.



Regards
Varun Ojha
Batch 104

Procedure of implementation:(After acquiring parallel port region)

Reader Side

The code is tested with a pulse of 10 usec and above , lower than it is causing synchronisation issues.