Error in a simple character device module - linux

I am trying to write a simple character device kernel module which allocates 100 bytes of memory in kernel space and has read, write and seek, which operates on the allocated 100 bytes of kernel memory. I have tried to echo the text "Hello World" into its corresponding device file, but get an "Invalid Argument" error and a warning on dmesg.
The device driver c file named "kernelbuffer.c"
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/init.h>
#include <linux/kernel.h> /* printk() */
#include <linux/slab.h> /* kmalloc() */
#include <linux/fs.h> /* everything... */
#include <linux/errno.h> /* error codes */
#include <linux/types.h> /* size_t */
#include <linux/proc_fs.h>
#include <linux/fcntl.h> /* O_ACCMODE */
#include <linux/seq_file.h>
#include <linux/cdev.h>
#include <asm/uaccess.h> /* copy_*_user */
static int major;
struct buffer_dev{
char data[100];
struct semaphore sem;
struct cdev cdev;
};
struct buffer_dev *buff_device;
static void buffer_open(struct inode *inode, struct file *filep){
struct buffer_dev *dev;
printk(KERN_NOTICE "Opening buffer.\n");
dev = container_of(inode->i_cdev, struct buffer_dev, cdev);
filep->private_data = dev;
printk(KERN_NOTICE "Buffer opened.\n");
return 0;
}
ssize_t buffer_read(struct file *filep, char __user *buf, size_t count,
loff_t *f_pos)
{
printk(KERN_NOTICE "Reading from buffer.\n");
struct buffer_dev *dev = filep->private_data;
ssize_t retval = 0;
if (down_interruptible(&dev->sem))
return -ERESTARTSYS;
if (*f_pos >= 100)
goto out;
if (*f_pos + count > 100)
count = 100 - *f_pos;
printk(KERN_NOTICE "First stage fine.\n");
if(copy_to_user(buf, dev->data + *f_pos, count)){
retval = -EFAULT;
goto out;
}
*f_pos += count;
retval = count;
out:
up(&dev->sem);
return retval;
}
ssize_t buffer_write(struct file *filep, char __user *buf, size_t count,
loff_t *f_pos)
{
printk(KERN_NOTICE "Writing to buffer.\n");
struct buffer_dev *dev = filep->private_data;
ssize_t retval = -ENOMEM; /* value used in "goto out" statements */
if (down_interruptible(&dev->sem))
return -ERESTARTSYS;
if (*f_pos >= 100)
goto out;
if (*f_pos + count > 100)
count = 100 - *f_pos;
if (copy_from_user(dev->data + *f_pos, buf, count)) {
retval = -EFAULT;
goto out;
}
printk(KERN_NOTICE "Write success.\n");
*f_pos += count;
retval = count;
out:
up(&dev->sem);
return retval;
}
loff_t buffer_llseek(struct file *filp, loff_t off, int whence)
{
struct buffer_dev *dev = filp->private_data;
loff_t newpos;
switch(whence) {
case 0: /* SEEK_SET */
newpos = off;
break;
case 1: /* SEEK_CUR */
newpos = filp->f_pos + off;
break;
case 2: /* SEEK_END */
newpos = 100 + off;
break;
default: /* can't happen */
return -EINVAL;
}
if (newpos < 0 || newpos >= 100) return -EINVAL;
filp->f_pos = newpos;
return newpos;
}
int buffer_release(struct inode *inode, struct file *filp)
{
return 0;
}
static struct file_operations buffer_fops = {
.owner = THIS_MODULE,
.llseek = buffer_llseek,
.read = buffer_read,
.write = buffer_write,
.open = buffer_open,
.release = buffer_release,
};
static void setup_cdev(struct buffer_dev *dev){
int err, devno = MKDEV(major, 0);
cdev_init(&dev->cdev, &buffer_fops);
dev->cdev.owner = THIS_MODULE;
dev->cdev.ops = &buffer_fops;
err = cdev_add (&dev->cdev, devno, 1);
if (err)
printk(KERN_NOTICE "Error %d adding device\n", err);
}
void cleanup_module_buffer(void)
{
printk(KERN_NOTICE "Removing buffer device.\n");
dev_t devno = MKDEV(major, 0);
if(buff_device){
cdev_del(&buff_device->cdev);
kfree(buff_device);
}
unregister_chrdev_region(devno,1);
}
int module_init_buffer(void)
{
printk(KERN_NOTICE "Initializing buffer device.\n");
int result, i;
dev_t dev = 0;
result = alloc_chrdev_region(&dev, 0, 1,
"buffer");
major = MAJOR(dev);
if (result < 0) {
printk(KERN_WARNING "Can't get major %d\n", major);
return result;
}
buff_device = kmalloc(sizeof(struct buffer_dev), GFP_KERNEL);
if (!buff_device) {
result = -ENOMEM;
goto fail; /* Make this more graceful */
}
memset(buff_device, 0, sizeof(struct buffer_dev));
sema_init(&(buff_device->sem), 1);
setup_cdev(&buff_device);
return 0; /* succeed */
fail:
cleanup_module_buffer();
return result;
}
module_init(module_init_buffer);
module_exit(cleanup_module_buffer);
MODULE_AUTHOR("Gautam Ramakrishnan");
MODULE_LICENSE("GPL");
The makefile used to compile it is:
obj-m := kernelbuffer.o
KDIR := /lib/modules/$(shell uname -r)/build
all: $(MAKE) -C $(KDIR) M=$(shell pwd) modules
clean:
$(MAKE) -C $(KDIR) M=$(shell pwd) clean
$(RM) Module.markers modules.order
I successfully ran insmod on the kernelbuffer.ko file.
On running cat /proc/devices
it shows up with a major number of 247.
I then executed sudo mknod /dev/kbuffer c 247 0 to create the corresponding driver file.
On running echo "Hello World > /dev/kbuffer as a superuser, I get the reply bash: /dev/kbuffer: Invalid argument
On running dmesg, I get the following output:
[ 583.814364] Opening buffer.
[ 583.814367] Buffer opened.
[ 583.814369] ------------[ cut here ]------------
[ 583.814392] WARNING: CPU: 0 PID: 2673 at /build/linux-dcxD3m/linux-4.4.0/fs/namei.c:3206 path_openat+0xb19/0x1330()
[ 583.814393] Modules linked in: kernelbuffer(OE) nls_utf8 isofs vboxsf(OE) crct10dif_pclmul crc32_pclmul snd_intel8x0 snd_ac97_codec ac97_bus aesni_intel joydev aes_x86_64 snd_pcm lrw gf128mul snd_seq_midi glue_helper ablk_helper snd_seq_midi_event input_leds vboxvideo(OE) cryptd snd_rawmidi ttm snd_seq serio_raw snd_seq_device drm_kms_helper snd_timer drm snd soundcore fb_sys_fops syscopyarea sysfillrect 8250_fintek sysimgblt i2c_piix4 vboxguest(OE) mac_hid parport_pc ppdev lp parport autofs4 hid_generic usbhid hid psmouse ahci libahci e1000 pata_acpi fjes video
[ 583.814427] CPU: 0 PID: 2673 Comm: bash Tainted: G OE 4.4.0-31-generic #50-Ubuntu
[ 583.814429] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
[ 583.814430] 0000000000000286 00000000c588cd8a ffff8800593ebcb8 ffffffff813f1143
[ 583.814433] 0000000000000000 ffffffff81cd8430 ffff8800593ebcf0 ffffffff81081102
[ 583.814435] ffff8800593ebdd0 0000000000008241 ffff8800593ebef4 00000000ffffffea
[ 583.814437] Call Trace:
[ 583.814442] [<ffffffff813f1143>] dump_stack+0x63/0x90
[ 583.814445] [<ffffffff81081102>] warn_slowpath_common+0x82/0xc0
[ 583.814447] [<ffffffff8108124a>] warn_slowpath_null+0x1a/0x20
[ 583.814449] [<ffffffff8121bfb9>] path_openat+0xb19/0x1330
[ 583.814452] [<ffffffff814f0a41>] ? copy_termios+0x71/0x80
[ 583.814455] [<ffffffff8121d9c1>] do_filp_open+0x91/0x100
[ 583.814457] [<ffffffff810ca961>] ? __raw_callee_save___pv_queued_spin_unlock+0x11/0x20
[ 583.814459] [<ffffffff8122b2d8>] ? __alloc_fd+0xc8/0x190
[ 583.814462] [<ffffffff8120c248>] do_sys_open+0x138/0x2a0
[ 583.814464] [<ffffffff81092230>] ? SyS_rt_sigaction+0xa0/0xd0
[ 583.814466] [<ffffffff8120c3ce>] SyS_open+0x1e/0x20
[ 583.814469] [<ffffffff8182db32>] entry_SYSCALL_64_fastpath+0x16/0x71
[ 583.814470] ---[ end trace ad5061abc2861b8c ]---
I have not been able to find any solution to this problem. Would be very helpful if anyone could provide a reason as to why my code does not work and any fixes, if possible.

Related

Which Linux kernel version support uffdio_writeprotect structure and how to compile and install that linux kernel?

I have try to compile a program which manage userfault-fd to collect some dirty pages on memory. In this program i have used the uffdio_writeprotect structure to survey a memory region where program try to access. I am using Ubuntu 18.04 with linux kernel version 5.14.0. But when i compile a program with the command gcc -o with_userfault -I. with_userfault.c
i still have the same errors such as: with_userfault.c:108:34: error: storage size of ‘wp’ isn’t known
struct uffdio_writeprotect wp;
with_userfault.c:114:21: error: ‘UFFDIO_WRITEPROTECT’ undeclared (first use in this function); did you mean ‘UFFDIO_REGISTER’?
if (ioctl(fd, UFFDIO_WRITEPROTECT, &wp) == -1)
I don't know how to solve this problem. Please I need your help to solve this issue because i don't know what i am suppose to do now. thanks!!!!!
this is a source code:
The header file:
#ifndef __RDTSC_H_DEFINED__
#define __RDTSC_H_DEFINED__
#if defined(__i386__)
static __inline__ unsigned long long rdtsc(void)
{
unsigned long long int x;
__asm__ volatile (".byte 0x0f, 0x31" : "=A" (x));
return x;
}
#elif defined(__x86_64__)
static __inline__ unsigned long long rdtsc(void)
{
unsigned hi, lo;
__asm__ __volatile__ ("rdtsc" : "=a"(lo), "=d"(hi));
return ( (unsigned long long)lo)|( ((unsigned long long)hi)<<32 );
}
#elif defined(__powerpc__)
static __inline__ unsigned long long rdtsc(void)
{
unsigned long long int result=0;
unsigned long int upper, lower,tmp;
__asm__ volatile(
"0: \n"
"\tmftbu %0 \n"
"\tmftb %1 \n"
"\tmftbu %2 \n"
"\tcmpw %2,%0 \n"
"\tbne 0b \n"
: "=r"(upper),"=r"(lower),"=r"(tmp)
);
result = upper;
result = result<<32;
result = result|lower;
return(result);
}
#else
#error "No tick counter is available!"
#endif
/* $RCSfile: $ $Author: kazutomo $
* $Revision: 1.6 $ $Date: 2005/04/13 18:49:58 $
*/
#endif
The source file
/*
* Example program about using userfaultfd(2) for garbage collection.
*
* This establishes a couple pages, all of which are filled from
* compressed files on disk when first accessed. For simplicity
* these are
* one file per page. Files are written at the beginning of the
* program.
*
* Later, this program demonstrates the use of write protection to
* get
* a notification on write access, analogous to using
* mprotect(!PROT_WRITE)
* and doing the bookkeeping in a SIGSEGV handler.
*
*/
#include <linux/userfaultfd.h>
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <sys/mman.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/ioctl.h>
#include <time.h>
#include <math.h>
#include <unistd.h>
#include <strings.h>
#include <string.h>
#include <unistd.h>
#include <asm/unistd.h>
#include <poll.h>
#include <pthread.h>
#include "rdtsc.h"
//#define size 102400
#define FILE_BUF 25
#define PAGE_SIZE sysconf(_SC_PAGE_SIZE)
#define n_pages 102400
//102400
#define iterations 1
pthread_t tracker;
void *tracker_task();
int i;
unsigned long main_tsc_start, main_tsc_end;
// This is doing the work in the uffd handler thread
void *tracker_task(void *data)
{
int fd = *(int *)(data);
for (;;)
{
struct uffd_msg msg;
struct pollfd pollfd;
int pollres, readret;
unsigned long addr, page_begin, whichpage;
unsigned long handler_tsc_start, handler_tsc_end;
pollfd.fd = fd;
pollfd.events = POLLIN;
pollres = poll(&pollfd, 1, -1);
if( pollres == -1 )
perror("poll");
if (pollfd.revents & POLLERR)
{
fprintf(stderr, "POLLERR on userfaultfd\n");
exit(1);
}
readret = read(fd, &msg, sizeof(msg));
if (readret == -1)
perror("read userfaultfd");
if (readret != sizeof(msg))
{
fprintf(stderr, "short read, not expected, exiting\n");
exit(1);
}
/*
* Proper sequence is important here.
*
* For the GC we expect that write-protected pages can only
* be pages already backed by physical pages.
* Regular writes into unprotected pages that come before
* reads need the page be filled.
*
* So we do the WP case first and get it out of the way.
* Then both of the other cases need the page read.
*/
if (msg.arg.pagefault.flags & UFFD_PAGEFAULT_FLAG_WP)
{
// send write unlock
struct uffdio_writeprotect wp;
wp.range.start = msg.arg.pagefault.address;
wp.range.len = PAGE_SIZE;
wp.mode = 0;
//printf("sending !UFFDIO_WRITEPROTECT event to
//userfaultfd\n");
if (ioctl(fd, UFFDIO_WRITEPROTECT, &wp) == -1)
perror("ioctl(UFFDIO_WRITEPROTECT)");
//continue;
}
}
printf("end\n");
return NULL;
}
int main(int argc, char *argv[])
{
unsigned long *region;
int uffd, uffd_flags, expected, t_create;
void *status;
struct uffdio_writeprotect wp;
struct uffdio_api uffdio_api;
struct uffdio_register uffdio_register;
char *clear = "3", *drop_caches_path = "/proc/sys/vm/drop_caches";
FILE *drop_caches_file;
main_tsc_start = rdtsc();
uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
if (uffd == -1)
{
perror("syscall");
exit(2);
}
uffdio_api.api = UFFD_API;
uffdio_api.features = 1;
if (ioctl(uffd, UFFDIO_API, &uffdio_api))
{
fprintf(stderr, "UFFDIO_API\n");
return 1;
}
//printf("Features: 0x%llx\n", uffdio_api.features);
if (uffdio_api.api != UFFD_API)
{
fprintf(stderr, "UFFDIO_API error %Lu\n", uffdio_api.api);
return 1;
}
/* Allocate memory that will be tracked */
region = (unsigned long *) mmap(NULL, PAGE_SIZE * n_pages, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, -1, 0);
if (!region)
{
perror("mmap");
exit(2);
}
/* Force alignment of contiguous pages */
if (posix_memalign((void **)region, PAGE_SIZE, PAGE_SIZE * (n_pages - 1)))
{
fprintf(stderr, "cannot align by PAGE_SIZE %ld\n", PAGE_SIZE);
exit(1);
}
uffdio_register.range.start = (unsigned long)region;
uffdio_register.range.len = PAGE_SIZE * n_pages;
uffdio_register.mode = UFFDIO_REGISTER_MODE_WP;
//main_tsc_start = rdtsc();
if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register) == -1)
{
perror("ioctl(UFFDIO_REGISTER)");
exit(1);
}
expected = UFFD_API_RANGE_IOCTLS;
if ((uffdio_register.ioctls & expected) != expected)
{
fprintf(stderr, "ioctl set is incorrect\n");
exit(1);
}
if( (t_create = pthread_create(&tracker, NULL, tracker_task,
&uffd)) != 0 )
{
errno = t_create;
perror("pthread_create");
}
//printf("mainline writing writable pages.\n");
for (unsigned long i = 0; i < n_pages; i++)
{
unsigned long entry = (i * PAGE_SIZE / sizeof(unsigned long)) + ((rand()%10000) % 512);//(i % 512);
region[entry] = i;
}
drop_caches_file = fopen(drop_caches_path, "w");
for (i = 0; i < iterations; i++)//
{
/* Indicate the range of pages to be write-protected */
//sync();
fwrite(clear, sizeof(clear), 1, drop_caches_file);
wp.range.start = (unsigned long long)region;
wp.range.len = PAGE_SIZE * n_pages;
wp.mode = UFFDIO_WRITEPROTECT_MODE_WP;
/* Write-protect pages */
if (ioctl(uffd, UFFDIO_WRITEPROTECT, &wp) == -1)
{
perror("ioctl(UFFDIO_WRITEPROTECT)");
exit(1);
}
/* Now try to "touch" the pages to trigger page faults and handling by the tracker thread */
for (unsigned long i = 0; i < n_pages; i++)
{
unsigned long entry = (i * PAGE_SIZE / sizeof(unsigned long)) + ((rand()%10000) % 512);//(i % 512);
region[entry] = i;
}
}
fclose(drop_caches_file);
if (ioctl(uffd, UFFDIO_UNREGISTER, &uffdio_register.range)) {
fprintf(stderr, "ioctl unregister failure\n");
return 1;
}
main_tsc_end = rdtsc();
printf("Apllication TSC : %lu\n", (unsigned long) (main_tsc_end-main_tsc_start));
return 0;
}

write_proc is not invoked when written from userspace

I am trying to understand procfs for communication between userspace and kernel module. My module has basic two functions for procfs write_proc, driver_mmap.
I call multiple times write_proc by calling fputs("123456789",fd). where fd is file descriptor to procfs entry in /proc directory. But I don't see write_proc called multiple time. Code is attached by here.
<code>
#include <linux/init.h>
#include <linux/pci.h>
#include <linux/slab.h>
#include <sound/core.h>
#include <sound/initval.h>
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/proc_fs.h>
#include <asm/uaccess.h>
#include <linux/interrupt.h>
#include <linux/io.h>
#include <linux/mm.h> /* mmap related stuff */
#define BUF_SIZE 64 * 1024
int *MmapBuffer;
int Factor = 1;
static int write_proc(struct file *filp, int *buf, size_t count, loff_t *offp)
{
int rc,i;
printk("in Write \n");
for (i = 1; i <= 16*1024 ; i++)
MmapBuffer[i-1] = (i+1)*Factor;
Factor++;
return count;
}
static int driver_mmap(struct file *file, struct vm_area_struct *vma)
{
int ret;
vma->vm_flags |= VM_LOCKED|VM_SHARED;
ret = remap_pfn_range(vma, vma->vm_start,
virt_to_phys(MmapBuffer) >> PAGE_SHIFT,
vma->vm_end-vma->vm_start, vma->vm_page_prot);
if(ret != 0)
printk("MMAP Failed \n");
SetPageReserved(virt_to_page(MmapBuffer));
printk("MMAP Succeeded \n");
return 0;
}
// file operations
struct file_operations proc_fops =
{
.write = write_proc,
.mmap = driver_mmap,
};
// init module
int init_module_test(void)
{
printk("<1>Hello world\n");
MmapBuffer = kzalloc(BUF_SIZE,__GFP_COLD|GFP_DMA);
if(MmapBuffer == NULL)
printk("Kzalloc failed. reduce buffer size \n");
proc_create ("Test_fs",0,NULL, &proc_fops);
return 0;
}
// exit module
void cleanup_module_test(void)
{
kfree(MmapBuffer);
remove_proc_entry ("Test_fs", NULL);
printk("Goodbye world\n");
}
module_init(init_module_test);
module_exit(cleanup_module_test);
MODULE_LICENSE("GPL");
</code>
Application code
<code>
#include<stdio.h>
#include<stdlib.h>
#include<sys/mman.h>
#include<errno.h>
#include <fcntl.h>
int main(void)
{
int fd;
int i,j;
int *msg ;
printf("Allocation started \n ");
msg=(int*)malloc(64*1024);
if(msg == NULL)
printf("Allocation failed \n");
//unsigned int *addr;
printf("Starting opening \n ");
if((fd=open("/proc/Test_fs", O_RDONLY ))<0)
{
printf("File not opened ");
}
printf("Starting mapping \n ");
msg = mmap(NULL, 64*1024, PROT_READ, MAP_SHARED , fd, 0);
printf("done from module \n ");
if(msg == MAP_FAILED)
{
printf("MAP failed and error is %s", strerror(errno));
return 0;
}
close(fd);
printf("Successful mapping");
FILE *f;
f=fopen("/proc/Test_fs", "wr");
if(!f)
{
printf("File not opened ");
}
for (j = 0; j < 10 ; j++)
{
if(fputs("1234567890,",f) <= 0)
printf("write failed, ");
for (i = 0; i < 16*1024 ; i++)
printf("%d, ", msg[i]);
printf("\n \n done \n \n ");
}
fclose(f);
return 0;
}
</code>

work with parallel port interrupt

I am working on parallel port driver. Now i had seen the methods to get interrupt from parallel port.
by one of them,
First make 4th pin of control reg 1(IRQ).
then make nACK low.
so i make a switch in between data pin 8 and nACK. so, if i write some data which has msb 1 then it will be interrupted, if that switch is on. Now i have a problem. If i disconnected that switch and again connect then it will not give me interrupt.
So, how can i do that thing by which i got interrupt by means of switch is connect or not.
#include <linux/init.h>
#include <linux/module.h>
#include <linux/fs.h>
#include <linux/parport.h>
#include <asm/uaccess.h>
#include <linux/platform_device.h>
#include <linux/interrupt.h>
#include <asm/io.h>
#include <linux/errno.h>
#include <asm/irq.h>
#include <linux/kthread.h>
#define DEVICE_NAME "parlelport"
struct pardevice *pdev;
static int dummy;
int ret;
static irqreturn_t recv_handler(int irq, void *dev_id)
{
printk("we inside if isr");
return 0;
}
int led_open(struct inode *inode, struct file *file)
{
printk("1\n");
printk("Device File Opened\n");
char byte1;
byte1=inb(0x37A);
printk("%d \n",byte1);
return 0;
}
ssize_t led_write(struct file *file, const char *buf, size_t count, loff_t *ppos)
{
printk("2\n");
char byte=inb(0x37A);
printk("%d",byte);
byte = byte | 0x10; // 0x10= 00010000, 4th pin of CTRL reg
outb(byte, 0x37A); //which enable IRQ
char kbuf;
copy_from_user(&kbuf, buf, 1);
parport_claim_or_block(pdev); /* Claim the port */
parport_write_data(pdev->port, kbuf); /* Write to the device */
//parport_release (pdev);
return count;
}
int led_release(struct inode *inode, struct file *file)
{
printk("3\n");
printk("Device File Released\n");
char byte;
byte=inb(0x37A);
printk("%d", byte);
return 0;
}
static struct file_operations led_fops = {
.owner = THIS_MODULE,
.open = led_open,
.write = led_write,
.release = led_release,
};
static int led_preempt(void *handle)
{
printk("4\n");
return 1;
}
static void led_attach(struct parport *port)
{
printk("5\n");
pdev = parport_register_device(port, DEVICE_NAME, led_preempt, NULL, NULL, 0, NULL);
printk("Port attached\n");
char byte1;
byte1=inb(0x37A);
printk("%d \n",byte1);
}
static void led_detach(struct parport *port)
{
printk("6\n");
parport_unregister_device (pdev);
printk("Port Deattached\n");
}
static struct parport_driver led_driver = {
.name= "led",
.attach = led_attach,
.detach = led_detach,
};
int __init led_init(void)
{
printk("7\n");
if (register_chrdev(89, DEVICE_NAME, &led_fops))
{
printk("Can't register device\n");
return -1;
}
char byte=inb(0x37A);
printk("%d",byte);
byte = byte | 0x10;
outb(byte, 0x37A);
char byte1;
byte1=inb(0x37A);
printk("%d %d \n",byte,byte1);
parport_register_driver(&led_driver);
ret= request_irq(7, recv_handler, IRQF_SHARED, "parlelport", &dummy);
printk("%d",ret);
return 0;
}
void __exit led_cleanup(void)
{
printk("8\n");
unregister_chrdev(89, DEVICE_NAME);
if(!ret)
free_irq(7, &dummy);
parport_unregister_driver(&led_driver);
printk("LED Driver unregistered.\n");
return;
}
module_init(led_init);
module_exit(led_cleanup);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Vikrant Patel");
Test.c file
int main()
{
int fd=open("/dev/parlelport",O_RDWR);
char byte;
printf("Enter Value to send on parallel port");
scanf("%c",&byte);
printf("Byte value is %c\n",byte);
if(write(fd,&byte,sizeof(char)))
{
printf("\nSuccessfully written on port");
}
getchar();
getchar();
close(fd);
}
I got it.
First make a thread
put the Enable IRQ code in that thread
so it will continuously execute it
whenever i connect pins at my hardware then it will be interrupted.
check this code for your ref.
#include <linux/module.h>
#include <linux/parport.h>
#include <asm/io.h>
#include <linux/interrupt.h>
#include <linux/kthread.h>
#define DEVICE_NAME "parlelport"
#define DATA 0x378
#define STATUS 0x379
#define CONTROL 0x37A
struct pardevice *pdev;
struct task_struct *ts1, *ts2;
int dummy;
char buf1='1',buf2='2';
char byte='0';
int thread1(void *data)
{
while(1)
{
outb(byte, CONTROL); /* 0x30 = 0011 0000 , makes IRQ pin(5th bit) enable */
printk("Thread1\n");
parport_claim_or_block(pdev); /* Claim the port */
parport_write_data(pdev->port, buf1); /* Write to the device */
parport_release(pdev); /* Release the port */
msleep(4000);
if (kthread_should_stop())
break;
}
return 0;
}
int thread2(void *data)
{
while(1)
{
outb(byte,CONTROL); /* 0x30 = 0011 0000 , makes IRQ pin(5th bit) enable */
printk("Thread2\n");
parport_claim_or_block(pdev); /* Claim the port */
parport_write_data(pdev->port, buf2); /* Write to the device */
parport_release(pdev); /* Release the port */
msleep(4000);
if (kthread_should_stop())
break;
}
return 0;
}
int led_open(struct inode *inode, struct file *file)
{
printk("Device File Opened\n");
ts1=kthread_run(thread1,NULL,"kthread"); /* Initiation of thread 1 */
msleep(2000);
ts2=kthread_run(thread2,NULL,"kthread"); /* Initiation of thread 2 */
return 0;
}
ssize_t led_write(struct file *file, const char *buf, size_t count, loff_t *ppos)
{
return count;
}
int led_release(struct inode *inode, struct file *file)
{
printk("Device File Released\n");
kthread_stop(ts1);
kthread_stop(ts2);
buf1='1';
buf2='2';
outb_p(0x00,DATA);
return 0;
}
static irqreturn_t recv_handler(int irq, void *unused)
{
printk("we inside of isr");
buf1= buf1 ^ 0x7F;
buf2= buf2 ^ 0x7F;
return 0;
}
static struct file_operations led_fops = {
.owner = THIS_MODULE,
.open = led_open,
.write = led_write,
.release = led_release,
};
static int led_preempt(void *handle)
{
return 1;
}
static void led_attach(struct parport *port)
{
pdev = parport_register_device(port, DEVICE_NAME, led_preempt, NULL, NULL, 0, NULL);
printk("Port attached\n");
}
static void led_detach(struct parport *port)
{
parport_unregister_device (pdev);
printk("Port Deattached\n");
}
static struct parport_driver led_driver = {
.name= "led",
.attach = led_attach,
.detach = led_detach,
};
int __init led_init(void)
{
/*Register our ISR with the kernel for PARALLEL_IRQ */
if (request_irq(7, recv_handler, IRQF_SHARED, DEVICE_NAME ,&dummy))
{
printk("Registering ISR failed\n");
return -ENODEV;
}
/*Register Character Device Driver at 89 Major number*/
if (register_chrdev(89, DEVICE_NAME, &led_fops))
{
printk("Can't register device\n");
return -1;
}
/*Register parallel port driver with parport structure led_driver*/
parport_register_driver(&led_driver);
return 0;
}
void __exit led_cleanup(void)
{
unregister_chrdev(89, DEVICE_NAME); /* Unregister char driver */
free_irq(7, &dummy); /* Free the ISR from IRQ7 */
parport_unregister_driver(&led_driver); /* Unregister the parallel port driver */
printk("LED Driver unregistered.\n");
return;
}
module_init(led_init);
module_exit(led_cleanup);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Vikrant Patel");

Net Link Linux User code bind socket call always fail for multicast group Id (non zero value)

Hi am trying to implement net link user code and kernel code every thing works fine for unicast (src_addr.nl_groups = 0;). For mulicast, user code bind call always fails for non zero src_addr.nl_groups value. Really am not sure what value to put for multicast and how to proceed further. I checked the usage of netlink_broadcast in kernel source tree, so I put the same group Id value (RTMGRP_LINK) here. For unicast I found good number of help in internet but for multicast I don't think so . So Please help me to proceed further.
Error am getting is:
bind: No such file or directory
./a.out: can't bind socket (3)and err : -1: No such file or directory
#include <sys/socket.h>
#include <unistd.h>
#include <stdlib.h>
#include <stdio.h>
#include <sys/types.h>
#include <stdio.h>
#include <linux/netlink.h>
#include <linux/rtnetlink.h>
#define NETLINK_TEST 28
#define GROUP_IB 1
#define MAX_PAYLOAD 1024
struct sockaddr_nl src_addr, dst_addr;
struct nlmsghdr *nlh = NULL;
struct msghdr msg;
struct iovec iov;
int sock_fd;
int main(int argc, char ** argv)
{
int err;
sock_fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_TEST);
if (sock_fd<0) {
char s[BUFSIZ];
sprintf( s, "%s: can't assign fd for socket", argv[0] );
perror(s);
return -1;
}
memset(&src_addr, 0, sizeof(src_addr));
src_addr.nl_family = AF_NETLINK;
src_addr.nl_pid = getpid();
src_addr.nl_groups = 0; // Unicast
//src_addr.nl_groups = RTMGRP_LINK; /* Multicast, bind call always fails for non zero values */
err = bind(sock_fd, (struct sockaddr*)&src_addr, sizeof(src_addr));
perror("bind");
if (err<0) {
char s[BUFSIZ];
sprintf( s, "%s: can't bind socket (%d)and err : %d", argv[0], sock_fd,err );
perror(s);
return -1;
}
memset(&dst_addr, 0, sizeof(dst_addr));
nlh = (struct nlhmsghdr *) malloc(NLMSG_SPACE(MAX_PAYLOAD));
memset(nlh, 0, NLMSG_SPACE(MAX_PAYLOAD));
iov.iov_base = (void *)nlh;
iov.iov_len = NLMSG_SPACE(MAX_PAYLOAD);
msg.msg_name = (void *)&dst_addr;
msg.msg_namelen = sizeof(dst_addr);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
printf("pid : %d\n Waiting for messages from kernel...\n",getpid());
recvmsg(sock_fd, &msg, 0);
printf("Message : %s\n", NLMSG_DATA(nlh));
close(sock_fd);
return 0;
}
Netlink socket binds are sensitive to what USER you are- I've seen them reliably fail if you are not running the program in question as 'root', at least on RedHat 6.
Try running as root 1st, before changing your logic. If you get the same failure as you do in normal operation, then you know it isn't (necessarily) a permissions issue.
The issue is
sock_fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_TEST);
Does you kernel module define the NETLINK_TEST family? your own family might must be supported at kernel module and it should post the message in the proper group using nlmsg_multicast()
RTMGRP_LINK is group defined in NETLINK_ROUTE.
This sample code is example for multicast
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <sys/socket.h>
#include <linux/netlink.h>
#include <unistd.h>
#define MYPROTO NETLINK_USERSOCK
#define MYMGRP 21
int open_netlink(void)
{
int sock;
struct sockaddr_nl addr;
int group = MYMGRP;
sock = socket(AF_NETLINK, SOCK_RAW, MYPROTO);
if (sock < 0) {
printf("sock < 0.\n");
return sock;
}
memset((void *) &addr, 0, sizeof(addr));
addr.nl_family = AF_NETLINK;
addr.nl_pid = getpid();
/* addr.nl_groups = MYMGRP; */
if (bind(sock, (struct sockaddr *) &addr, sizeof(addr)) < 0) {
printf("bind < 0.\n");
return -1;
}
if (setsockopt(sock, 270, NETLINK_ADD_MEMBERSHIP, &group, sizeof(group)) < 0) {
printf("setsockopt < 0\n");
return -1;
}
return sock;
}
void read_event(int sock)
{
struct sockaddr_nl nladdr;
struct msghdr msg;
struct iovec iov;
char buffer[65536];
int ret;
iov.iov_base = (void *) buffer;
iov.iov_len = sizeof(buffer);
msg.msg_name = (void *) &(nladdr);
msg.msg_namelen = sizeof(nladdr);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
printf("Ok, listening.\n");
ret = recvmsg(sock, &msg, 0);
if (ret < 0)
printf("ret < 0.\n");
else
printf("Received message payload: %s\n", NLMSG_DATA((struct nlmsghdr *) &buffer));
}
int main(int argc, char *argv[])
{
int nls;
nls = open_netlink();
if (nls < 0)
return nls;
while (1)
read_event(nls);
return 0;
}
kernel module:
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/netlink.h>
#include <net/netlink.h>
#include <net/net_namespace.h>
#define MYPROTO NETLINK_USERSOCK
#define MYGRP 21
static struct sock *nl_sk = NULL;
static void send_to_user(void)
{
struct sk_buff *skb;
struct nlmsghdr *nlh;
char *msg = "Hello from kernel";
int msg_size = strlen(msg) + 1;
int res;
pr_info("Creating skb.\n");
skb = nlmsg_new(NLMSG_ALIGN(msg_size + 1), GFP_KERNEL);
if (!skb) {
pr_err("Allocation failure.\n");
return;
}
nlh = nlmsg_put(skb, 0, 1, NLMSG_DONE, msg_size + 1, 0);
strcpy(nlmsg_data(nlh), msg);
pr_info("Sending skb.\n");
res = nlmsg_multicast(nl_sk, skb, 0, MYGRP, GFP_KERNEL);
if (res < 0)
pr_info("nlmsg_multicast() error: %d\n", res);
else
pr_info("Success.\n");
}
static int __init hello_init(void)
{
pr_info("Inserting hello module.\n");
nl_sk = netlink_kernel_create(&init_net, MYPROTO, NULL);
if (!nl_sk) {
pr_err("Error creating socket.\n");
return -10;
}
send_to_user();
netlink_kernel_release(nl_sk);
return 0;
}
static void __exit hello_exit(void)
{
pr_info("Exiting hello module.\n");
}
module_init(hello_init);
module_exit(hello_exit);
MODULE_LICENSE("GPL");

How to mmap a Linux kernel buffer to user space?

Let's say the buffer is allocated using a page based scheme. One way to implement mmap would be to use remap_pfn_range but LDD3 says this does not work for conventional memory. It appears we can work around this by marking the page(s) reserved using SetPageReserved so that it gets locked in memory. But isn't all kernel memory already non-swappable i.e. already reserved? Why the need to set the reserved bit explicitly?
Does this have something to do with pages allocated from HIGH_MEM?
The simplest way to map a set of pages from the kernel in your mmap method is to use the fault handler to map the pages. Basically you end up with something like:
static int my_mmap(struct file *filp, struct vm_area_struct *vma)
{
vma->vm_ops = &my_vm_ops;
return 0;
}
static const struct file_operations my_fops = {
.owner = THIS_MODULE,
.open = nonseekable_open,
.mmap = my_mmap,
.llseek = no_llseek,
};
(where the other file operations are whatever your module needs). Also in my_mmap you do whatever range checking etc. is needed to validate the mmap parameters.
Then the vm_ops look like:
static int my_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
vmf->page = my_page_at_index(vmf->pgoff);
get_page(vmf->page);
return 0;
}
static const struct vm_operations_struct my_vm_ops = {
.fault = my_fault
}
where you just need to figure out for a given vma / vmf passed to your fault function which page to map into userspace. This depends on exactly how your module works. For example, if you did
my_buf = vmalloc_user(MY_BUF_SIZE);
then the page you use would be something like
vmalloc_to_page(my_buf + (vmf->pgoff << PAGE_SHIFT));
But you could easily create an array and allocate a page for each entry, use kmalloc, whatever.
[just noticed that my_fault is a slightly amusing name for a function]
Minimal runnable example and userland test
Kernel module:
#include <linux/fs.h>
#include <linux/init.h>
#include <linux/kernel.h> /* min */
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/proc_fs.h>
#include <linux/uaccess.h> /* copy_from_user, copy_to_user */
#include <linux/slab.h>
static const char *filename = "lkmc_mmap";
enum { BUFFER_SIZE = 4 };
struct mmap_info {
char *data;
};
/* After unmap. */
static void vm_close(struct vm_area_struct *vma)
{
pr_info("vm_close\n");
}
/* First page access. */
static vm_fault_t vm_fault(struct vm_fault *vmf)
{
struct page *page;
struct mmap_info *info;
pr_info("vm_fault\n");
info = (struct mmap_info *)vmf->vma->vm_private_data;
if (info->data) {
page = virt_to_page(info->data);
get_page(page);
vmf->page = page;
}
return 0;
}
/* After mmap. TODO vs mmap, when can this happen at a different time than mmap? */
static void vm_open(struct vm_area_struct *vma)
{
pr_info("vm_open\n");
}
static struct vm_operations_struct vm_ops =
{
.close = vm_close,
.fault = vm_fault,
.open = vm_open,
};
static int mmap(struct file *filp, struct vm_area_struct *vma)
{
pr_info("mmap\n");
vma->vm_ops = &vm_ops;
vma->vm_flags |= VM_DONTEXPAND | VM_DONTDUMP;
vma->vm_private_data = filp->private_data;
vm_open(vma);
return 0;
}
static int open(struct inode *inode, struct file *filp)
{
struct mmap_info *info;
pr_info("open\n");
info = kmalloc(sizeof(struct mmap_info), GFP_KERNEL);
pr_info("virt_to_phys = 0x%llx\n", (unsigned long long)virt_to_phys((void *)info));
info->data = (char *)get_zeroed_page(GFP_KERNEL);
memcpy(info->data, "asdf", BUFFER_SIZE);
filp->private_data = info;
return 0;
}
static ssize_t read(struct file *filp, char __user *buf, size_t len, loff_t *off)
{
struct mmap_info *info;
ssize_t ret;
pr_info("read\n");
if ((size_t)BUFFER_SIZE <= *off) {
ret = 0;
} else {
info = filp->private_data;
ret = min(len, (size_t)BUFFER_SIZE - (size_t)*off);
if (copy_to_user(buf, info->data + *off, ret)) {
ret = -EFAULT;
} else {
*off += ret;
}
}
return ret;
}
static ssize_t write(struct file *filp, const char __user *buf, size_t len, loff_t *off)
{
struct mmap_info *info;
pr_info("write\n");
info = filp->private_data;
if (copy_from_user(info->data, buf, min(len, (size_t)BUFFER_SIZE))) {
return -EFAULT;
} else {
return len;
}
}
static int release(struct inode *inode, struct file *filp)
{
struct mmap_info *info;
pr_info("release\n");
info = filp->private_data;
free_page((unsigned long)info->data);
kfree(info);
filp->private_data = NULL;
return 0;
}
static const struct file_operations fops = {
.mmap = mmap,
.open = open,
.release = release,
.read = read,
.write = write,
};
static int myinit(void)
{
proc_create(filename, 0, NULL, &fops);
return 0;
}
static void myexit(void)
{
remove_proc_entry(filename, NULL);
}
module_init(myinit)
module_exit(myexit)
MODULE_LICENSE("GPL");
GitHub upstream.
Userland test:
#define _XOPEN_SOURCE 700
#include <assert.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h> /* uintmax_t */
#include <string.h>
#include <sys/mman.h>
#include <unistd.h> /* sysconf */
/* Format documented at:
* https://github.com/torvalds/linux/blob/v4.9/Documentation/vm/pagemap.txt
*/
typedef struct {
uint64_t pfn : 54;
unsigned int soft_dirty : 1;
unsigned int file_page : 1;
unsigned int swapped : 1;
unsigned int present : 1;
} PagemapEntry;
/* Parse the pagemap entry for the given virtual address.
*
* #param[out] entry the parsed entry
* #param[in] pagemap_fd file descriptor to an open /proc/pid/pagemap file
* #param[in] vaddr virtual address to get entry for
* #return 0 for success, 1 for failure
*/
int pagemap_get_entry(PagemapEntry *entry, int pagemap_fd, uintptr_t vaddr)
{
size_t nread;
ssize_t ret;
uint64_t data;
nread = 0;
while (nread < sizeof(data)) {
ret = pread(pagemap_fd, ((uint8_t*)&data) + nread, sizeof(data),
(vaddr / sysconf(_SC_PAGE_SIZE)) * sizeof(data) + nread);
nread += ret;
if (ret <= 0) {
return 1;
}
}
entry->pfn = data & (((uint64_t)1 << 54) - 1);
entry->soft_dirty = (data >> 54) & 1;
entry->file_page = (data >> 61) & 1;
entry->swapped = (data >> 62) & 1;
entry->present = (data >> 63) & 1;
return 0;
}
/* Convert the given virtual address to physical using /proc/PID/pagemap.
*
* #param[out] paddr physical address
* #param[in] pid process to convert for
* #param[in] vaddr virtual address to get entry for
* #return 0 for success, 1 for failure
*/
int virt_to_phys_user(uintptr_t *paddr, pid_t pid, uintptr_t vaddr)
{
char pagemap_file[BUFSIZ];
int pagemap_fd;
snprintf(pagemap_file, sizeof(pagemap_file), "/proc/%ju/pagemap", (uintmax_t)pid);
pagemap_fd = open(pagemap_file, O_RDONLY);
if (pagemap_fd < 0) {
return 1;
}
PagemapEntry entry;
if (pagemap_get_entry(&entry, pagemap_fd, vaddr)) {
return 1;
}
close(pagemap_fd);
*paddr = (entry.pfn * sysconf(_SC_PAGE_SIZE)) + (vaddr % sysconf(_SC_PAGE_SIZE));
return 0;
}
enum { BUFFER_SIZE = 4 };
int main(int argc, char **argv)
{
int fd;
long page_size;
char *address1, *address2;
char buf[BUFFER_SIZE];
uintptr_t paddr;
if (argc < 2) {
printf("Usage: %s <mmap_file>\n", argv[0]);
return EXIT_FAILURE;
}
page_size = sysconf(_SC_PAGE_SIZE);
printf("open pathname = %s\n", argv[1]);
fd = open(argv[1], O_RDWR | O_SYNC);
if (fd < 0) {
perror("open");
assert(0);
}
printf("fd = %d\n", fd);
/* mmap twice for double fun. */
puts("mmap 1");
address1 = mmap(NULL, page_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
if (address1 == MAP_FAILED) {
perror("mmap");
assert(0);
}
puts("mmap 2");
address2 = mmap(NULL, page_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
if (address2 == MAP_FAILED) {
perror("mmap");
return EXIT_FAILURE;
}
assert(address1 != address2);
/* Read and modify memory. */
puts("access 1");
assert(!strcmp(address1, "asdf"));
/* vm_fault */
puts("access 2");
assert(!strcmp(address2, "asdf"));
/* vm_fault */
strcpy(address1, "qwer");
/* Also modified. So both virtual addresses point to the same physical address. */
assert(!strcmp(address2, "qwer"));
/* Check that the physical addresses are the same.
* They are, but TODO why virt_to_phys on kernel gives a different value? */
assert(!virt_to_phys_user(&paddr, getpid(), (uintptr_t)address1));
printf("paddr1 = 0x%jx\n", (uintmax_t)paddr);
assert(!virt_to_phys_user(&paddr, getpid(), (uintptr_t)address2));
printf("paddr2 = 0x%jx\n", (uintmax_t)paddr);
/* Check that modifications made from userland are also visible from the kernel. */
read(fd, buf, BUFFER_SIZE);
assert(!memcmp(buf, "qwer", BUFFER_SIZE));
/* Modify the data from the kernel, and check that the change is visible from userland. */
write(fd, "zxcv", 4);
assert(!strcmp(address1, "zxcv"));
assert(!strcmp(address2, "zxcv"));
/* Cleanup. */
puts("munmap 1");
if (munmap(address1, page_size)) {
perror("munmap");
assert(0);
}
puts("munmap 2");
if (munmap(address2, page_size)) {
perror("munmap");
assert(0);
}
puts("close");
close(fd);
return EXIT_SUCCESS;
}
GitHub upstream.
Tested on kernel 5.4.3.
Though the pages are reserved via a kernel driver, it is meant to be accessed via user space. As a result, the PTE (page table entries) do not know if the pfn belongs to user space or kernel space (even though they are allocated via kernel driver).
This is why they are marked with SetPageReserved.

Resources