On 08/21/2012 04:25 PM, Richard W.M. Jones wrote:
On Tue, Aug 21, 2012 at 04:04:54PM +0800, Wanlong Gao wrote:
[...]
>>> + spid = fork ();
>>> + if (spid == -1) {
>>> + perror ("fork");
>>> + exit (EXIT_FAILURE);
>>> + }
>>> + if (spid == 0) {
>>> + if (guestfs_mount_local_run (sg) == -1)
>>> + exit (EXIT_FAILURE);
>>> + _exit (EXIT_SUCCESS);
>>> + }
>>> +
>>> + dpid = fork();
>>> + if (dpid == -1) {
>>> + perror ("fork");
>>> + exit (EXIT_FAILURE);
>>> + }
>>> + if (dpid == 0) {
>>> + if (guestfs_mount_local_run (dg) == -1)
>>> + exit (EXIT_FAILURE);
>>> + _exit (EXIT_SUCCESS);
>>> + }
>>> +
>>> + const char *dir = argv[optind];
>>> + char system_arg[BUFSIZ];
>>> + sprintf (system_arg, "diff -urN %s%s %s%s", stempdir, dir,
>>> + dtempdir, dir);
>>> + sleep (5);
>>
>> ?
>
> I think I'm waiting for guest launching?
No, that's the wrong thing to do.
The API is in two parts: First guestfs_mount_local prepares the mount
point. Second guestfs_mount_local_run causes requests to be
processed.
You can access the mount point after guestfs_mount_local has returned.
Your process will block until guestfs_mount_local_run is called, but
that doesn't matter in this case (in fact, it's what you want).
Yeah, got it.
Another thing: Don't use sprintf, ever. With the wrong $TMPDIR that
code above is a security hole. Since you need to quote 'dir' (since
it comes from user input) you're going to have to rewrite the whole
call to system(3) so that it uses fork + exec instead.
OK, let me have a try. ;)
Thanks,
Wanlong Gao
Rich.