On Fri, 16 Oct 2026 19:54:36 +0800, Stanley Jhu wrote: > On Fri, 16 Oct 2026 18:33:43 +0900, Daejun Park wrote: >> + case WRITE_SAME_10: > > As far as I can see, JESD220H does not list WRITE SAME among the UFS > commands. Would it make sense to reject WRITE SAME(10/16) on a zoned > LU?
Thanks, agreed. Linux ufshcd sets no_write_same on every UFS logical unit, so Linux guests do not depend on it. v3 will fail WRITE SAME(10) and WRITE SAME(16) on a zoned LU with INVALID COMMAND OPERATION CODE, as it does UNMAP. This also removes a case where the write pointer runs ahead of the data: scsi-hd completes a WRITE SAME with NDOB=1 with GOOD without writing anything, after the zone check has moved the write pointer. Regular LUs keep passing WRITE SAME to scsi-hd as today; if they should reject it too, I can send that as a separate patch. Daejun
