On Thu, Feb 16, 2012 at 05:59:24PM +0800, Wanlong Gao wrote: > we wipe the filesystem signatures from device before do mkfs, > since we can't trust mkfs can remove the filesystem signature absolutely. > > Signed-off-by: Wanlong Gao <gaowanlong(a)cn.fujitsu.com&gt; > --- > daemon/mkfs.c | 9 +++++++++ > 1 files changed, 9 insertions(+), 0 deletions(-) > > diff --git a/daemon/mkfs.c b/daemon/mkfs.c > index 5475582..19224ec 100644 > --- a/daemon/mkfs.c > +++ b/daemon/mkfs.c > @@ -181,6 +181,15 @@ do_mkfs_opts (const char *fstype, const char *device, int blocksize, > ADD_ARG (argv, i, device); > ADD_ARG (argv, i, NULL); > > + /* we wipe the filesystem signatures from device before do mkfs, > + * since we can't trust mkfs can remove the filesystem signature absolutely. > + */ > + if (do_wipefs (device)) { You should test do_wipefs () == -1 here. > + reply_with_error ("%s: %s: %s", fstype, device, err); Calling reply_with_error will cause an error to be sent twice (once from do_wipefs, once here), and you'll lose protocol synchronization. In any case, I'm doubtful that this patch is useful. If callers want to wipe the filesystem before doing mkfs, they can just call wipefs themselves. It sounds like this will just slow down mkfs and make it more likely to fail.

Rich.