On Wed, 2021-08-04 at 06:32 +0000, Shinichiro Kawasaki wrote:
> Dmitry, thank you for the review comments.
>
> On Aug 03, 2021 / 19:36, Dmitry Fomichev wrote:
> > On Wed, 2021-07-28 at 19:47 +0900, Shin'ichiro Kawasaki wrote:
> > > Enable trim workload for zonemode=zbd by modifying do_io_u_trim()
> > > to
> > > call zoned block device unique function zbd_do_io_u_trim() which
> > > resets
> > > target zone. This allows fio to emulate workloads which mix data
> > > read/write and zone resets with zonemode=zbd.
> > >
> > > To call reset zones, the trim I/O shall have offset aligned to
> > > zone
> > > start and block size same as zone size. Reset zone is called only
> > > to
> > > sequential write required zones and sequential write preferred
> > > zones.
> > > Conventional zones are handled in same manner as regular block
> > > devices
> > > by calling os_trim() function.
> > >
> > > When zones are reset with random trim workload, choose only non-
> > > empty
> > > zones as trim target. This avoids meaningless trim to empty zones
> > > and
> > > makes the workload more realistic. To find the non-empty zones,
> > > utilize
> > > zbd_find_zone() helper function which is already used for read
> > > workload,
> > > specifying 1 byte as the minimum valid data size.
> > >
> > > Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@xxxxxxx>
> > > ---
> > > HOWTO | 3 +++
> > > fio.1 | 2 ++
> > > io_u.c | 9 ++++++++
> > > zbd.c | 71
> > > ++++++++++++++++++++++++++++++++++++++++++++++++++++++----
> > > zbd.h | 2 ++
> > > 5 files changed, 83 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/HOWTO b/HOWTO
> > > index 59c7f1ff..f1fd2045 100644
> > > --- a/HOWTO
> > > +++ b/HOWTO
> > > @@ -992,6 +992,9 @@ Target file/device
> > > single zone. The
> > > :option:`zoneskip`
> > > parameter
> > > is ignored. :option:`zonerange`
> > > and
> > > :option:`zonesize` must be
> > > identical.
> > > + Trim is handled using a zone
> > > reset
> > > operation.
> > > + Trim only considers non-empty
> > > sequential write
> > > + required and sequential write
> > > preferred
> > > zones.
> >
> > The documentation changes could be moved to a separate patch. While
> > looking at HOWTO file, I notice that libzbc is absent from the list
> > of
> > i/o engines there. Perhaps, since you are making changes to this
> > file,
> > you could add a small description of libzbc there and mention that
> > it
> > supports read, write and trim workloads when run against zoned
> > block
> > devices.
>
> Okay, I will separate out changes in HOWTO and fio.1 into a patch.
> Will
> add the description of libzbc i/o engine also.
>
> >
> > >
> > > .. option:: zonerange=int
> > >
> > > diff --git a/fio.1 b/fio.1
> > > index 6cc82542..ef319062 100644
> > > --- a/fio.1
> > > +++ b/fio.1
> > > @@ -766,6 +766,8 @@ starts. The \fBzonecapacity\fR parameter is
> > > ignored.
> > > Zoned block device mode. I/O happens sequentially in each zone,
> > > even
> > > if random32ad4373
> > > I/O has been selected. Random I/O happens across all zones
> > > instead of
> > > being
> > > restricted to a single zone.
> > > +Trim is handled using a zone reset operation. Trim only
> > > considers non-
> > > empty
> > > +sequential write required and sequential write preferred zones.
> > > .RE
> > > .RE
> > > .TP
> > > diff --git a/io_u.c b/io_u.c
> > > index 9a1cd547..696d25cd 100644
> > > --- a/io_u.c
> > > +++ b/io_u.c
> > > @@ -2317,10 +2317,19 @@ int do_io_u_trim(const struct thread_data
> > > *td,
> > > struct io_u *io_u)
> > > struct fio_file *f = io_u->file;
> > > int ret;
> > >
> > > + if (td->o.zone_mode == ZONE_MODE_ZBD) {
> > > + ret = zbd_do_io_u_trim(td, io_u);
> > > + if (ret == io_u_completed)
> > > + return io_u->xfer_buflen;
> > > + if (ret)
> > > + goto err;
> > > + }
> > > +
> > > ret = os_trim(f, io_u->offset, io_u->xfer_buflen);
> > > if (!ret)
> > > return io_u->xfer_buflen;
> > >
> > > +err:
> > > io_u->error = ret;
> > > return 0;
> > > #endif
> > > diff --git a/zbd.c b/zbd.c
> > > index f10b3267..39060ecd 100644
> > > --- a/zbd.cpassing in an offset
> > modifier with a value of 8. If the suffix is used with a
> > sequential I/O
>
> I did not catch this part. May I ask to rephrase?
Please disregard this, I think my email client (Evolution) had a little
glitch while formatting the reply...
>
