[dm-devel] [PATCH v3 1/4] dm dust: report some message results back to user directly

[Bryan Gurney]

On Fri, Jun 19, 2020 at 11:45 AM Mike Snitzer <snitzer at redhat.com> wrote:
>
> On Fri, Jun 19 2020 at 11:40am -0400,
> Bryan Gurney <bgurney at redhat.com> wrote:
>
> > On Fri, Jun 19, 2020 at 8:37 AM yangerkun <yangerkun at huawei.com> wrote:
> > >
> > > Some type of message(queryblock/countbadblocks/removebadblock) may better
> > > report results to user directly. Do it with DMEMIT.
> > >
> > > Signed-off-by: yangerkun <yangerkun at huawei.com>
> > > ---
> > >  drivers/md/dm-dust.c | 31 ++++++++++++++++++-------------
> > >  1 file changed, 18 insertions(+), 13 deletions(-)
> > >
> > > diff --git a/drivers/md/dm-dust.c b/drivers/md/dm-dust.c
> > > index ff03b90072c5..a0c75c104de0 100644
> > > --- a/drivers/md/dm-dust.c
> > > +++ b/drivers/md/dm-dust.c
> > > @@ -138,20 +138,22 @@ static int dust_add_block(struct dust_device *dd, unsigned long long block,
> > >         return 0;
> > >  }
> > >
> > > -static int dust_query_block(struct dust_device *dd, unsigned long long block)
> > > +static int dust_query_block(struct dust_device *dd, unsigned long long block, char *result,
> > > +                           unsigned int maxlen, unsigned int *sz_ptr)
> > >  {
> > >         struct badblock *bblock;
> > >         unsigned long flags;
> > > +       unsigned int sz = *sz_ptr;
> > >
> > >         spin_lock_irqsave(&dd->dust_lock, flags);
> > >         bblock = dust_rb_search(&dd->badblocklist, block);
> > >         if (bblock != NULL)
> > > -               DMINFO("%s: block %llu found in badblocklist", __func__, block);
> > > +               DMEMIT("block %llu found in badblocklist", block);
> > >         else
> > > -               DMINFO("%s: block %llu not found in badblocklist", __func__, block);
> > > +               DMEMIT("block %llu not found in badblocklist", block);
> > >         spin_unlock_irqrestore(&dd->dust_lock, flags);
> > >
> > > -       return 0;
> > > +       return 1;
> >
> > First, thank you very much for this patch.  After the concerns to
> > convert some functions to use DMEMIT were brought up, I was trying to
> > start the conversion, when this patch arrived, so I installed it, and
> > tested it.
> >
> > I do have a question, though:
> >
> > First, I see that in dust_query_block() (above) and
> > dust_clear_badblocks(), the "return 0" statement is changed to "return
> > 1".
> >
> > (Additionally, there is a change from "r = 0" to  "r = 1", in the
> > "countbadblocks" message handler)
> >
> > On testing the functions, they still work, but why was this change
> > made?  Is it related to the use of DMEMIT?
>
> It is, but we need to review the returns closer.  Looked to me that 1
> was being returned even if nothing was DMEMIT()'d.. but I could be wrong
> (only looked quickly).
>
> I also noticed that some output was changed to not include __func__.
> Please review that the output reflects what you'd like displayed.
>
> Mike
>

After adding __func__ back into the string, here's what it looks like:

$ sudo dmsetup message dust1 0 addbadblock 60
$ sudo dmsetup message dust1 0 queryblock 60
dust_query_block: block 60 found in badblocklist
$ sudo dmsetup message dust1 0 queryblock 61
dust_query_block: block 61 not found in badblocklist

I feel that the output's origin is more clear, so I'd like to leave
__func__ in the output.

I took a look at the DMEMIT calls, and in all three cases, it looks
like there's something DMEMIT()'d:

dust_query_block: either "block found", or "block not found".
dust_clear_badblocks: either "no badblocks found", or "badblocks cleared".
result of "message ... countbadblocks": always prints "%llu badblocks found".


Thanks,

Bryan