Re: [PATCH v2] xfsprogs: xfs_quota command error message improvement

[Bill O'Donnell <billodo@redhat.com>]

On Tue, Jun 23, 2020 at 05:30:38PM -0500, Eric Sandeen wrote:
> On 6/22/20 8:13 AM, Bill O'Donnell wrote:
> > @@ -350,8 +365,15 @@ quotaoff(
> >  		return;
> >  	}
> >  	dir = mount->fs_name;
> > -	if (xfsquotactl(XFS_QUOTAOFF, dir, type, 0, (void *)&qflags) < 0)
> > -		perror("XFS_QUOTAOFF");
> > +	if (xfsquotactl(XFS_QUOTAOFF, dir, type, 0, (void *)&qflags) < 0) {
> > +		if (errno == EEXIST)
> > +			fprintf(stderr, _("Quota already off.\n"));
> > +		else if (errno == EINVAL)
> > +			fprintf(stderr,
> > +				_("Can't disable when quotas are off.\n"));
> 
> Is this the right message here?  We get here from off_f(), which disables
> enforcement and accounting, so I'm not sure "can't disable" makes sense
> if "disable" means "disable enforcement" as it did in disable_enforcement()...?
> 
> (IOWs, have can you provoke EINVAL?  How?  Sorry, this just kind of jumps out
> at me because "can't disable" seems a little out of place in quotaoff() so
> I want to double check.)

You're right. It was a copy/paste from the disable case. You also have a good point
in that I don't think EINVAL is ever provoked for the OFF case. I'll recheck it.

Thanks-
Bill

> 
> Thanks,
> -Eric
>