How can I read large data from proc file?

Question

I'm trying to write a kernel module which writes some data to a proc file. I'm trying to write something like 5000 characters but when I say $>cat /proc/myentry I can read only 1000 characters.

int procfile_read(char *buffer,  char **buffer_location,  off_t offset, int buffer_length, int *eof, void *data){
int ret;
static char my_buffer[4096];

if (offset > 0) {

    ret  = 0;
} else {

    ret = sprintf(my_buffer, LARGE STRING HERE);
}

*buffer_location=my_buffer;
return ret;
}

This is my code. Thanks in advance.

Answer 1

I had exactly this problem.

One issue in the original post, the if (offset>0) is used many times in examples of small proc files. The read is called multiple times until we return a 0 to indicate that there is no more data. So the if (offset>0) means we return (length of the buffer) as 0.

There are 3 ways to return data with this function. Look at the source code comments, line 75 onwards :

For large files (method 2 from comments), I did the following :-

• For each lump of your large data, copy 'buffer_length' of data to 'buffer'.
• Set '*start' (or in your case *buffer_location) to 'buffer'.
• return the amount of data you wrote (typically 'buffer_length')

Finally, all data will be written and you return 0.

This worked for me with several Meg of data.

Answer 2

I am not a kernel expert, but in linux-3.1.6/fs/proc/task_mmu.c I see some code like

    seq_printf(m,
            "VmPeak:\t%8lu kB\n"
            "VmSize:\t%8lu kB\n"
            "VmLck:\t%8lu kB\n"
            "VmHWM:\t%8lu kB\n"
            "VmRSS:\t%8lu kB\n"
            "VmData:\t%8lu kB\n"
            "VmStk:\t%8lu kB\n"

so this suggests that you might want to use seq_printf not sprintf .... The m is a struct seq_file * pointer.

As a general rule, you'll learn a lot by studying the free software source code which you are extending. In your case, it is the Linux kernel source code