Re: [PATCH V2] mkfs.btrfs: allow UUID specification at mkfs time

[Goffredo Baroncelli <kreijack@libero.it>]

On 05/14/2014 06:01 PM, David Sterba wrote:
> On Wed, May 14, 2014 at 10:35:05AM -0500, Eric Sandeen wrote:
>> > @@ -125,7 +154,19 @@ int make_btrfs(int fd, const char *device, const char *label,
>> >  	memset(&super, 0, sizeof(super));
>> >  
>> >  	num_bytes = (num_bytes / sectorsize) * sectorsize;
>> > -	uuid_generate(super.fsid);
>> > +	if (fs_uuid) {
>> > +		if (uuid_parse(fs_uuid, super.fsid) != 0) {
>> > +			fprintf(stderr, "could not parse UUID: %s\n", fs_uuid);
>> > +			ret = -EINVAL;
>> > +			goto out;
> I think the uuid validity check comes too late, IMHO it should be done
> after the while/getopt block outside of make_btrfs. At this point eg.
> the discard or device zeroing is already done.


Pay attention that, if you don't change the test_uuid_unique() you need to perform the check after zeroing the disk, or otherwise you are not able to mkfs the same disk with the same UUID (I suspect that this is a real use case for testing).

I am still convinced that a Warning is a better way to handle these kind of situations.

As reported before, forcing a unique UUID to be not unique is a thing to skilled person. The problem is that BTRFS in this regard behaves differently respect other file-systems, and even a skilled person could not be aware of the possible problem.
In this case is better to provide a complete information, instead of complicating the things adding further check.

BR
G.Baroncelli
-- 
gpg @keyserver.linux.it: Goffredo Baroncelli (kreijackATinwind.it>
Key fingerprint BBF5 1610 0B64 DAC6 5F7D  17B2 0EDA 9B37 8B82 E0B5