On Thu, Feb 06, 2020 at 02:26:31PM +0900, Damien Le Moal wrote: > Add the new file Documentation/filesystems/zonefs.txt to document > zonefs principles and user-space tool usage. > > Signed-off-by: Damien Le Moal <damien.lemoal@xxxxxxx> > --- > Documentation/filesystems/zonefs.txt | 404 +++++++++++++++++++++++++++ > MAINTAINERS | 1 + > 2 files changed, 405 insertions(+) > create mode 100644 Documentation/filesystems/zonefs.txt Looks largely OK to me. A few small nits below in the new error handling text, but otherwise Reviewed-by: Dave Chinner <dchinner@xxxxxxxxxx> > +IO error handling > +----------------- > + > +Zoned block devices may fail I/O requests for reasons similar to regular block > +devices, e.g. due to bad sectors. However, in addition to such known I/O > +failure pattern, the standards governing zoned block devices behavior define > +additional conditions that result in I/O errors. > + > +* A zone may transition to the read-only condition (BLK_ZONE_COND_READONLY): > + While the data already written in the zone is still readable, the zone can > + no longer be written. No user action on the zone (zone management command or > + read/write access) can change the zone condition back to a normal read/write > + state. While the reasons for the device to transition a zone to read-only > + state are not defined by the standards, a typical cause for such transition > + would be a defective write head on an HDD (all zones under this head are > + changed to read-only). > + > +* A zone may transition to the offline condition (BLK_ZONE_COND_OFFLINE): > + An offline zone cannot be read nor written. No user action can transition an > + offline zone back to an operational good state. Similarly to zone read-only > + transitions, the reasons for a drive to transition a zone to the offline > + condition are undefined. A typical cause would be a defective read-write head > + on an HDD causing all zones on the platter under the broken head to be > + inaccessible. > + > +* Unaligned write errors: These errors result from the host issuing write > + requests with a start sector that does not correspond to a zone write pointer > + position when the write request is executed by the device. Even though zonefs > + enforces sequential file write for sequential zones, unaligned write errors > + may still happen in the case of a partial failure of a very large direct I/O > + operation split into multiple BIOs/requests or asynchronous I/O operations. > + If one of the write request within the set of sequential write requests > + issued to the device fails, all write requests after queued after it will > + become unaligned and fail. > + > +* Delayed write errors: similarly to regular block devices, if the device side > + write cache is enabled, write errors may occur in ranges of previously > + completed writes when the device write cache is flushed, e.g. on fsync(). > + Similarly to the previous immediate unaligned write error case, delayed write > + errors can propagate through a stream of cached sequential data for a zone > + causing all data to be dropped after the sector that caused the error. > + > +All I/O errors detected by zonefs are always notified to the user with an error s/always// > +code return for the system call that trigered or detected the error. The > +recovery actions taken by zonefs in response to I/O errors depend on the I/O > +type (read vs write) and on the reason for the error (bad sector, unaligned > +writes or zone condition change). > + > +* For read I/O errors, zonefs does not execute any particular recovery action, > + but only if the file zone is still in a good condition and there is no > + inconsistency between the file inode size and its zone write pointer position. > + If a problem is detected, I/O error recovery is executed (see below table). > + > +* For write I/O errors, zonefs I/O error recovery is always executed. > + > +* A zone condition change to read-only or offline also always triggers zonefs > + I/O error recovery. > + > +Zonefs minimal I/O error recovery may change a file size and a file access > +permissions. > + > +* File size changes: > + Immediate or delayed write errors in a sequential zone file may cause the file > + inode size to be inconsistent with the amount of data successfully written in > + the file zone. For instance, the partial failure of a multi-BIO large write > + operation will cause the zone write pointer to advance partially, eventhough "even though" > + the entire write operation will be reported as failed to the user. In such > + case, the file inode size must be advanced to reflect the zone write pointer > + change and eventually allow the user to restart writing at the end of the > + file. > + A file size may also be reduced to reflect a delayed write error detected on > + fsync(): in this case, the amount of data effectively written in the zone may > + be less than originally indicated by the file inode size. After such I/O > + error zonefs always fixes a file inode size to reflect the amount of data "error, zonefs" ? Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx
