Re: [PATCH v2011.2] fs: symlink restrictions on sticky directories

[Ingo Molnar <mingo@elte.hu>]

* Kees Cook <keescook@chromium.org> wrote:

> +	/* Check parent directory mode and owner. */
> +	spin_lock(&dentry->d_lock);
> +	parent = dentry->d_parent->d_inode;
> +	if ((parent->i_mode & (S_ISVTX|S_IWOTH)) == (S_ISVTX|S_IWOTH) &&
> +	    parent->i_uid != inode->i_uid) {
> +		error = -EACCES;
> +	}
> +	spin_unlock(&dentry->d_lock);
> +
> +#ifdef CONFIG_AUDIT
> +	if (error) {
> +		struct audit_buffer *ab;
> +
> +		ab = audit_log_start(current->audit_context,
> +				     GFP_ATOMIC, AUDIT_AVC);
> +		audit_log_format(ab, "op=follow_link action=denied");
> +		audit_log_d_path(ab, "path=", &nameidata->path);
> +		audit_log_format(ab, " name=");
> +		audit_log_untrustedstring(ab, dentry->d_name.name);
> +		audit_log_format(ab, " dev=%s ino=%lu",
> +				 inode->i_sb->s_id,
> +				 inode->i_ino);
> +		audit_log_end(ab);
> +	}
> +#endif

Hm, is GFP_ATOMIC really necessary here? Why not GFP_KERNEL? 
This is in VFS process context, not in some atomic context that 
has to be careful about allocations, right?

Thanks,

	Ingo