Fix zfs_copy_file_range() for handling of RLIMIT_FSIZE

[rmacklem]

Summary

Commit 52360ca32ff9 fixed the handling of RLIMIT_FSIZE
for vn_generic_copy_file_range() and nfs_copy_file_range().
I think the same fix is needed for zfs_copy_file_range(),
which is what this patch does.

I have no way of testing this and have no idea how ZFS commits
are done so, hopefully, someone else will handle this.

Test Plan

Someone else needs to test this.

[rmacklem]

Oh, and I didn't include it in the review, but the
comment for zfs_copy_file_range() should probably
be updated since it now gets calls for different mounts.
ie. The TODO has been done.

[asomers]

This isn't necessary. My test program passes on ZFS even without this change. Here's the program:

#include <sys/types.h>

#include <sys/resource.h>
#include <sys/time.h>

#include <assert.h>
#include <err.h>
#include <fcntl.h>
#include <stdlib.h>
#include <unistd.h>

size_t FSIZE = 65536;

int main(int argc, char **argv) {
	struct rlimit rl;
	int infd, outfd;
	char *buf;
	size_t nbytes = FSIZE;
	ssize_t r;
	off_t ioff = 0, ooff = 0;

	buf = calloc(FSIZE, 1);

	infd = open("copy_file_range_rlimit_fsize_infile", O_RDWR | O_CREAT,
		0644);
	if (infd < 0)
		err(1, "open");

	while (nbytes > 0) {
		int r;
		if ((r = write(infd, buf, nbytes)) < 0)
			err(1, "write");
		nbytes -= r;
	}

	if (getrlimit(RLIMIT_FSIZE, &rl) != 0)
		err(1, "getrlimit");

	rl.rlim_cur = FSIZE / 2;
	if (setrlimit(RLIMIT_FSIZE, &rl) != 0)
		err(1, "setrlimit");

	outfd = open("copy_file_range_rlimit_fsize_outfile", O_WRONLY | O_CREAT,
		0644);
	if (outfd < 0)
		err(1, "open");

	r = copy_file_range(infd, &ioff, outfd, &ooff, FSIZE, 0);
	if (r < 0)
		err(1, "copy_file_range");
	if (r > rl.rlim_cur)
		errx(1, "copy_file_range exceeded RLIMIT_FSIZE");

	return (0);
}