Re: [PATCH v2] Fix i_mutex handling in nfsd readdir

[hooanon05@yahoo.co.jp]

David Woodhouse:
> This patch fixes it by locking the directory's i_mutex again before
> calling the filldir functions. The original deadlocks which commit
	:::

An entry may be removed between the first mutex_unlock and the second
mutex_lock. In this case, lookup_one_len() in compose_entry_fh() will
return a negative dentry.
Currently the inode test (positive/negative) is done AFTER fh_compose().
Isn't it better to test it BEFORE fh_compose()?

compose_entry_fh()
{
	:::
		dchild = lookup_one_len(name, dparent, namlen);
	if (IS_ERR(dchild))
		return 1;
	if (d_mountpoint(dchild) ||
	    fh_compose(fhp, exp, dchild, &cd->fh) != 0 ||
	    !dchild->d_inode)
		rv = 1;
	:::
}


J. R. Okajima