Attached is the list of ZFS improvements / bugs fixes in S10U4.

--
Prabahar.
4894692 caching data in heap inflates crash dump
6269805 properties should be set via an nvlist.
6276925 option to sort 'zfs list' output
6281585 user defined properties
6341569 zio_alloc_blk() vdev distribution performs badly
6343741 want to store a command history on disk
6349494 'zfs list' output annoying for even moderately long dataset names
6349987 lzjb.c lived longer than expected?
6351954 zfs missing noxattr mount flag
6366244 'canmount' option for container-like functionality
6367103 create-time properties
6368751 libzfs interface for mount/umounting all the file systems for a given 
pool
6385349 zpool import -d /dev hangs
6393525 vdev_reopen() should verify that it's still the same device
6397052 unmounting datasets should process /etc/mnttab instead of traverse DSL
6403510 zfs remount,noatime option broken
6413510 zfs: writing to ZFS filesystem slows down fsync() on other files in the 
same FS
6414648 zfs allows overlapping devices to be added
6416639 RFE: provide zfs get -a
6420135 zfs(1m) should display properties of snapshots that affect their 
behavior
6421992 'zfs send -i' requires redundant input
6423412 Two spaces lines are unnecessary after 'zpool import -a'
6424466 "panic: data after EOF" when unmounting abused pool
6428639 large writes to zvol synchs too much, better cut down a little
6431818 ZFS receive should print a better error message on failure
6434054 'zfs destroy' core dumps if clone is namespace-parent of origin
6435943 assertion failed: spare != 0L, file: ../../common/fs/zfs/spa_misc.c
6436000 import of actively spared device returns EBUSY
6437472 want 'zfs recv -F' to force rollback
6437808 ZFS module version should match on-disk version
6438643 'zfs rename <snapshot> <snapshot>' requires redundant arguments
6438702 error handling in zfs_getpage() can trigger "page not locked" panic
6438947 znode_t's z_active should be removed
6440515 namespace_reload() can leak memory on allocation faiure
6440592 nvlist_dup() should not fill in destination pointer on error
6444692 Need to flush disk write cache for dmu_sync buffers
6445680 Having write_acl allowed in an ACL doesn't give the ability to set the 
mode via chmod
6446060 zfs get does not consistently report temporary properties
6446512 zfs send does not catch malformed snapshot name
6447701 ZFS hangs when iSCSI Target attempts to initialize its backing store
6448326 zfs(1) 'list' command crashes if hidden property createtxg is requested
6450653 get_dependents() has poor error semantics
6450672 chmod acl parse error gave poor help for incorrect delimeter (,) 
instead of (/)
6453026 typo in zfs clone error message
6453172 ztest turns into a sloth due to massive arc_min_prefetch_lifespan
6454551 'zfs create -b blocksize filesystem' should fail.
6455228 zpool_mount_datasets() should take an additional flag
6455234 libzfs_error_description() should not abort if no error
6456642 Bug in libzfs_init()
6457478 unrecognized character in error message with 'zpool create -R' command
6457865 missing device name in the error message of 'zpool clear' command
6458058 invalid snapshot error message
6458571 zfs_ioc_set_prop() doesn't validate input
6458614 zfs ACL #defines should use prefix
6458638 get_configs() accesses bogus memory
6458678 zvol functions should be moved out of zfs_ioctl.h
6458683 zfs_cmd_t could use more cleanup
6458691 common routines to manage zfs_cmd_t nvlists
6460043 zfs_write()/zfs_getpage() range lock deadlock
6460059 zfs destroy <snapshot> leaves behind kruft
6460398 zpool import cores on zfs_prop_get
6461029 zpool status -x noexisting-pool has incorrect error message.
6461223 index translations should live with property definitions
6461424 zpool_unmount_datasets() has some busted logic
6461427 zfs_realloc() would be useful
6461438 zfs send/recv code should live in its own file
6461609 zfs delete permissions are not working correctly
6461757 'zpool status' can report the wrong number of persistent errors
6461784 recursive zfs_snapshot() leaks memory
6462174 zap_update() likes to return 0
6463140 zfs recv with a snapshot name that has 2 @@ in a row succeeds
6463348 ZFS code could be more portable
6463349 error message from zpool(1M) is missing a newline
6463350 getcomponent() has spurious and wrong check
6463788 'zfs recv -d' fails if some ancestors already exist
6464310 Stressing trunc can induce file corruption
6464897 assertion failed: "BP_GET_COMPRESS(bp) == compress" zio.c, line:897
6465634 zvol: dmu_sync() should be issued in parallel
6468453 Incorrect usage of rw_enter() in zfs_getpage()
6468731 lwb_state_t can be nuked
6468748 assertion failure in dnode_sync
6469119 race between arc_buf_clone() and arc_buf_add_ref() results in NULL 
pointer dereference
6469385 zfs_set_prop_nvlist range checking is busted
6469830 'zfs set' panics non-debug systems
6470042 parallel dmu_sync() isn't being used
6470764 zfs_clone() confuses filesystems and volumes
6471679 stash blocksize in zvol_state_t rather than reading in every 
zvol_log_write
6472021 vdev knobs can not be turned
6472230 ZIL vdev management is inefficient
6472843 panic when write to zvol snapshot
6473775 zil_commit changes in snv_48 make it hot for O_DSYNC workloads
6475506 panic in dmu_recvbackup due to NULL pointer dereference
6475942 need more assertions in dnode_destroy()
6476196 spa_sync_spares() is missing a nvlist_free()
6476457 Assigning ACL to a ZFS filesystem using chmod failed with "Missing 
fields after user/group"
6476663 bad trap possible when dereferencing b_hdr after buf eviction
6477102 recvbackup ioctl does not advance file offset
6477103 read-only properties should be passed as nvlist
6477900 want more /etc/system tunables for ZFS performance analysis
6478388 ZIL replay takes too long causing issues while booting
6478928 buffer recycling causes long arc mutex spins
6479497 ::abuf_find is broken
6480230 ::walk spa|::walk zms_freelist cause mdb to core dump
6480245 renaming a dataset to something with '%s' will cause segfault
6483675 want a private property to return number of clones
6484044 destroying a clone takes O(data)
6484731 zfs import only mounts and shares the first layer of hierarchy
6485204 more tuneable tweakin
6485728 zpool iostat should flush output periodically
6485955 need more dtrace probes
6486390 zil_commit could push more transactions
6486496 zil_replay() useful debug
6489548 zap_leaf_array_create() relies on architecture-specific behavior
6489714 JNI: getDataset() returns null for pool snapshot name
6490104 'zfs recv' can be very slow when processing large DRR_FREEOBJECTS record
6490105 assertion failure in dbuf_verify when doing zfs recv
6490569 verify arc bufs are not modified when they shouldn't be
6490829 panic in zfs_hash() while 'zfs recv' a zvol
6492686 NFSv4 client got EACCES over ZFS when trying to OPENATTR without 
createdir
6493146 get_numeric_property() has redundant code
6493750 zfs and iSCSI do not work togetherly correctly
6493923 nfsfind on ZFS filesystem quickly depletes memory in a 1GB system
6494072 A device which was set as spare disk is not detach
6494075 destroying a pool shouldn't touch the DSL
6494638 Panic - buffer modified while frozen! when running zpool_create_005
6494840 libzfs should dlopen libiscsitgt rather than linking to it
6496365 assertion failed: refcount_count(&tx->tx_space_written) + delta <= 
tx->tx_space_towrite
6497563 zfs double-spared an already-spared disk on reboot
6499454 time to increase size of kmem default allocation caches
6499459 vm should stop checking kvp directly
6499476 some zfs_iter_* users leak memory
6501789 'zfs recv pool/fs@' causes assertion failure in mzap_addent
6502193 race in zfs_rename
6503724 adding spare that is in use in another pool should be allowed
6503788 memory leak in is_spare()
6503790 memory leak in zpool_get_errlog()
6504386 Need to deprecate libzfs.so.1 in s10u4
6505225 zpool(1M) can give misleading error when removing active spare
6505625 large regression with zfs in snv_53
6505658 target MRU size (arc.p) needs to be adjusted more aggressively
6506748 segzio uses unfeasibly large amount of VA
6508883 libzfs should not use VERIFY()
6510448 zfs_unshare_iscsi() gives bogus error message when libiscsitgt cannot 
be opened or is missing
6260523 want 'zfs list <path>'
6363303 numerous source files misspell "relevant"
6402600 administrative actions while doing 'zfs recv' or 'zfs destroy' can 
cause confusing situation
6410433 'zpool status -v' would be more useful with filenames
6437750 panic: db->db_buf==0||arc_referenced(db->db_buf), file: dbuf.c,line:1539
6441384 zpool import action message is not correctly localized
6456888 zpool scrubbing leads to memory exhaustion and system hang
6458218 assertion failed: ss == NULL
6460622 zio_nowait() doesn't live up to its name
6471255 Some ZFS error messages contain spelling errors
6473418 setting user defined property on full filesystem should report error
6480626 contention on spa_config_lock on niagra
6480849 source of canmount property needs improvement
6483718 need a stable interface to get all native properties
6488665 Invalid argument cause Abort (core dumped).
6489169 zvol: Incorrect ordering of biodone() and ensuring data is stable
6494654 zfs inherit can't handle property short names
6495013 Loops and recursion in metaslab_ff_alloc can kill performance, even on 
a pool with lots of free data
6496341 zvol performance can be improved
6496344 zvol async routines are synchronous
6496346 zfs_write with multiple iovs logs wrong data
6496347 zfs_write can be made much simpler thanks to range locks
6496350 zvol needs to log all writes to honor fsync
6496356 zvol should behave like a write-caching device
6496357 spec_fsync() is useless on devices that do write caching
6496359 zfs_read does unnecessary work on mmaped files with cached data
6497211 Creating a zpool mirror with a large number of devices will core dump
6504702 zdb -dddv <poolname> chokes on xattrs
6506506 spa_history.c: LE_64(reclen) needs to be cast to uint64_t in case of 
32-bit big-endian kernel
6509749 RAIDZ reads unnecessarily generate parity
6510807 ARC statistics should be exported via kstat
6511976 memory leak in dmu_sendbackup()
6511978 memory leak in dmu_recvbackup()
6512311 zio_done() is chunky and needs to go on a diet
6512391 DMU should leverage ZIO dependencies to achieve greater parallelism
6512661 unowned mutex panic during stress test
6513011 libzfs .po files are not being generated
6513020 zio pipeline went out to lunch
6513953 Unable to create global hotspare in ja locale
6514071 recent changes to zfs.h breaks Sun Cluster
6514331 in-memory delete queue is not needed
6516175 zfs_send()/zfs_receive() should take file descriptor arguments
6520051 segzio's sizing policy needs further optimization
6523936 BAD TRAP: type=8 (#df Double fault) rp=fec266f8 addr=0
6525008 panic: dr->dt.dl.dr_override_state != DR_IN_DMU_SYNC, file: 
../../common/fs/zfs/dbuf.c, line: 676
6526196 ZFS: assertion failed: zp->z_phys->zp_links > zp_is_dir
6526874 ARC is using ints instead of uint64s in some places
6527134 'zpool status -v' can report duplicate entries duplicate entries
6527325 want more assertions in space map code
6528274 arc_reclaim_needed's zio code has a pair of incorrect assumptions
6529406 zpool history needs to bump the on-disk version
6532056 missing dmu_tx_commit() in zfs_write() eventually hangs zfs
6545015 RAID-Z resilver broken
_______________________________________________
zfs-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to