IS_ERR() macro in Linux

Question

While getting to know how to write device drivers I came across IS_ERR() macro. But I couldn't find how is it working. I have include the code below:

majorNumber = register_chrdev(0, DEVICE_NAME, &fops);

if (majorNumber<0)
{
    printk(KERN_ALERT "Failed to register a major number\n");
    return majorNumber;
}
printk(KERN_INFO "Registered correctly with major number %d\n", majorNumber);

// Register the device class
ebbcharClass = class_create(THIS_MODULE, CLASS_NAME);

if (IS_ERR(ebbcharClass))
{               
  unregister_chrdev(majorNumber, DEVICE_NAME);
  printk(KERN_ALERT "Failed to register device class\n");
  return PTR_ERR(ebbcharClass);          
}

So what does the IS_ERR() macro expand to and how it gets executed.

It is unclear what you are asking. If you only wish to see what the macro evaluates to, compile with `-E` option — Eregrith, May 19 '15 at 08:38
I have edited my question to give a clear meaning of what I need to know. — S.I.J, May 19 '15 at 09:01
Btw, use `pr_*()` macros for print and check the [CodingStyle](https://www.kernel.org/doc/Documentation/CodingStyle) just in case. — 0andriy, Jun 14 '15 at 16:51

score 9 · Accepted Answer · edited Jun 21 '19 at 16:40

9

Tests if the supplied pointer should be considered an error value.

It does not check if the pointer is valid.

In your code IS_ERR is used to check if class_create succeded creating ebbcharClass. If an error occurs unregister the char driver and signal the error.

You can find MACROs and inline funcs in err.h

edited Jun 21 '19 at 16:40

mcsim

1,647
2
16
35

answered May 19 '15 at 08:44

LPs

16,045
8
30
61

I would add that is defined in `err.h`. – 0andriy Jun 14 '15 at 16:50
I saw somewhere IS_ERR is used like this : if (IS_ERR(dev_ret = class_create(THIS_MODULE, "chardev"))) . so IS_ERR works both for the `return value of class_create function` and the `assignment operation of class_create return value`? from https://testkernel.tistory.com/91 – Chan Kim Apr 28 '21 at 11:53

score 8 · Answer 2 · edited Oct 17 '17 at 20:52

8

Be careful for pitfall:

#define IS_ERR_VALUE(x) unlikely((x) >= (unsigned long)-MAX_ERRNO)
#define MAX_ERRNO       4095

This covers -1 to -4095, which represents error code, not number below 4096, nor NULL (0). Every value from 0 to 4294963201 (0xfffff001) is considered no error. Do not use it to cover NULL checking.

edited Oct 17 '17 at 20:52

msrd0

7,816
9
47
82

answered Oct 17 '17 at 20:44

bsd unix

81
1
1

Range typo: good value should be from 0 to 4294963200 (0xfffff000). – bsd unix Oct 17 '17 at 20:58

score 1 · Answer 3 · answered May 19 '15 at 09:08

If you want to know what the macro expands to, just compile your file using the -E option of gcc, which will only do pre-processing. It will include all headers and expand all macros.

The macro does not "get executed" per se, it's just a "search and replace" type of thing.

IS_ERR() macro in Linux

3 Answers3