svn commit: r343118 - in head/usr.sbin: . trim

[Warner Losh]

On Thu, Jan 17, 2019 at 5:48 PM Rodney W. Grimes <
freebsd at pdx.rh.cn85.dnsmgr.net> wrote:

> > On Thu, Jan 17, 2019 at 3:43 PM Conrad Meyer <cem at freebsd.org> wrote:
> >
> > > On Thu, Jan 17, 2019 at 1:16 PM Eugene Grosbein <eugen at freebsd.org>
> wrote:
> > > >
> > > > 18.01.2019 3:23, Conrad Meyer ?????:
> > > >
> > > > > Please back it out; stop attributing code review to "hackers@,"
> which
> > > > > can not (it's a list, not individuals) and did not review this
> > > > > changeset; and put it on phabricator for actual review.
> > > >
> > > > There is already https://reviews.freebsd.org/D18380 by imp
> > > > and there were over 6 weeks since it had an update.
> > > > Newly committed code has most of its changes.
> > >
> > > Your response does not address *any* of the above concerns.  It's just
> > > unrelated.
> > >
> >
> > Yes.
> >
> >
> > > The review you linked to isn't one you submitted for this change; it's
> > > Warner's, and that one stalled because you were such a jerk to him
> > > last time that he needed time off from you.
> > >
> >
> > I abandoned it because dd is the better way to go because it can transfer
> > images with 0's and TRIM the zero'd pages (and only the zero'd pages).
> Sure
> > dd is hard to use, but it's not that hard to use.
>
> I strongly disagree that dd is the proper place for this tool,
> dd is far to easy a way to destroy data when trying to do this
> type of operation and should be the last choice for implementing
> what is normally a very safe thing to do.
>
> And as eugene points out dd can not easily be made
> to do "trim ada0 ada1 ada4"  which is probably the
> far more common use of this command
>

for i in ada0 ada1 ada4; do dd of=/dev/$i if=/dev/zero conv=trim &; done;
wait

if that's your objection. It's a weak counter argument, but this really
isn't the place for the debate.


> dd's ability to do sparse(ing) of a file should NOT be
> overloaded with a device layer block trimming function.
>

It's not overloaded. You have to specifically ask for the conversion. It's
the right place for it.


> >
> >
> > > If you want to socialize, revive, or expedite someone else's review,
> > > maybe add a new comment to the review, or post a discussion hackers@,
> > > or something like that.  You still can't attribute code review to
> > > hackers@, especially as no such review happened.
> > >
> >
> > I think it should be backed out. Especially since it hijacks my work and
> my
> > name to try to ram it into the tree.
> >
> > I'm very much not amused.
>
> Nor am I with some of the reactoin to the commit :-(
>

I'm sorry, but I fire back when people abuse me. Especially when it's a
repeat offender who clearly didn't internalize the feedback given the first
time.

Warner