Reason to reset the offset about lkmpg HOT 2 CLOSED

Hi,
It's just making sure that every time we read to the end of the buffer we go back to the beginning.
Even though, in most cases, we just open (create the empty file) the file, read it, and close it once time, the reset might be unnecessary.
When we open the file, the function calls in v5.4 will be like this:

[113739.851760]  device_open.cold+0x18/0x4b [chardev]
[113739.851762]  chrdev_open+0xd3/0x1c0
[113739.851764]  ? cdev_default_release+0x20/0x20
[113739.851765]  do_dentry_open+0x143/0x3a0
[113739.851767]  vfs_open+0x2d/0x30
[113739.851769]  do_last+0x194/0x900
[113739.851770]  path_openat+0x8d/0x290
[113739.851771]  do_filp_open+0x91/0x100
[113739.851774]  ? __alloc_fd+0x46/0x150
[113739.851775]  do_sys_open+0x17e/0x290
[113739.851777]  __x64_sys_openat+0x20/0x30
[113739.851779]  do_syscall_64+0x57/0x190
[113739.851780]  entry_SYSCALL_64_after_hwframe+0x44/0xa9

See the path_openat(), it will allocate the empty file:

static struct file *path_openat(struct nameidata *nd,
			const struct open_flags *op, unsigned flags)
{
	struct file *file;
	int error;

	file = alloc_empty_file(op->open_flag, current_cred());
	if (IS_ERR(file))
		return file;

So, every time we do the open-read-close, the offset will be reset implicitly.
However, if we do something like open-read-read-close, we need to reset the offset by ourselves, otherwise, we will not get any data on the subsequent read calls.

Hopefully, this will help you.
Thanks :)

from lkmpg.

I understand now. Thanks for the detailed explanation.

from lkmpg.