Re: [PATCH v4 07/10] scsi: scsi_debug: Do not write a partial physical block to a zoned device

[Niklas Cassel <cassel@kernel.org>]

On Fri, Sep 18, 2026 at 04:31:23PM +0700, Damien Le Moal wrote:
> On 2026/09/18 13:29, Niklas Cassel wrote:
> > do_device_access() copies one logical block at a time and stops at the
> > first short copy, so when the data-out buffer is smaller than the
> > transfer length of the command it can write part of a physical block.
> > An initiator can arrange that with SG_IO.
> > 
> > A write that does not end on a physical block boundary is perfectly
> > acceptable to a device that is not zoned, and to a conventional zone,
> > where the device reads, modifies and writes the physical block that the
> > write falls in. ZBC-3 r06 (T10/BSR INCITS 579), 4.5.3.3.2, does require
> > a write to a sequential write required zone to end on a physical block
> > boundary, though, so a partly written physical block is not a state that
> > such a zone can be left in.
> > 
> > Stop at the last whole physical block that the buffer holds, for a
> > sequential write required zone only. The bytes that are left over are
> > not written, and are reported to the initiator as part of the residual.
> > 
> > With the default physblk_exp=0 the physical block size equals the
> > logical block size and this changes nothing. Nothing changes either when
> > the buffer holds all of the data that the command asks for, so a command
> > that transfers fewer logical blocks than a physical block is unaffected
> > wherever it is legal.
> > 
> > Assisted-by: LLM
> > Signed-off-by: Niklas Cassel <cassel@kernel.org>
> 
> A bit shift would be nicer than a division... But nevertheless, looks good.
> 
> Reviewed-by: Damien Le Moal <dlemoal@kernel.org>

Sure, will apply the following change for this patch:

diff --git a/drivers/scsi/scsi_debug.c b/drivers/scsi/scsi_debug.c
index 5ae1241e4d8b..bbd675ca47f2 100644
--- a/drivers/scsi/scsi_debug.c
+++ b/drivers/scsi/scsi_debug.c
@@ -4313,7 +4313,8 @@ static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp,
                struct sdeb_zone_state *zsp = zbc_zone(devip, lba);

                if (zsp->z_type == ZBC_ZTYPE_SWR) {
-                       u32 avail = (sdb->length - sg_skip) / sdebug_sector_size;
+                       u32 avail = (sdb->length - sg_skip)
+                               >> ilog2(sdebug_sector_size);

                        if (avail < num)
                                num = round_down(avail, 1U << sdebug_physblk_exp);


And the following change for patch "scsi: scsi_debug: Advance the write
pointer over the data written":

@@ -5184,7 +5185,7 @@ static int resp_write_dt0(struct scsi_cmnd *scp, struct sdebug_dev_info *devip)
 
        /* If ZBC zone then bump its write pointer over the data written */
        if (sdebug_dev_is_zoned(devip) && ret > 0)
-               zbc_inc_wp(devip, lba, ret / sdebug_sector_size);
+               zbc_inc_wp(devip, lba, ret >> ilog2(sdebug_sector_size));
        if (meta_data_locked)
                sdeb_meta_write_unlock(sip);
 
@@ -5350,7 +5351,8 @@ static int resp_write_scat(struct scsi_cmnd *scp,
                ret = do_device_access(sip, scp, sg_off, lba, num, group, true, true);
                /* If ZBC zone then bump its write pointer over the data written */
                if (sdebug_dev_is_zoned(devip) && ret > 0)
-                       zbc_inc_wp(devip, lba, ret / sdebug_sector_size);
+                       zbc_inc_wp(devip, lba,
+                                  ret >> ilog2(sdebug_sector_size));
                if (unlikely(scsi_debug_lbp()))
                        map_region(sip, lba, num);
                if (unlikely(-1 == ret)) {


when respinning.


Kind regards,
Niklas