Re: [PATCH v2 0/2] Lock and Pointer guards

[Peter Zijlstra]

On Tue, May 30, 2023 at 11:23:42AM +0200, Peter Zijlstra wrote:

> Yes, it's a wee bit more involved, but I'm thinking it gives a fair
> amount of flexibility and we don't need to ret rid of
> -Wdeclaration-after-statement.

One other thing I forgot to point put; it allows things like:

	int store_fd(int fd)
	{
		ptr_guard(fdput, f) = fdget(fd);
		void *ret;
		if (!f.file)
			return -EBADF;
		ret = xa_store(&xarray, f.file->private, f);
		if (xa_is_err(ret))
			return xa_err(ret);
		f = null_ptr(fdput); // xarray now owns f
		return 0;
	}

Where we can assign null_ptr() to clear the guard and inhibit the
cleanup function to pass ownership around.