zfs crypto discuss - Mar 2008 - [Bug 697] New: zpool import dumps core when key available

http://defect.opensolaris.org/bz/show_bug.cgi?id=697

           Summary: zpool import dumps core when key available
    Classification: Development
           Product: zfs-crypto
           Version: unspecified
          Platform: Other
        OS/Version: Solaris
            Status: NEW
          Severity: major
          Priority: P2
         Component: other
        AssignedTo: ajscarp at yahoo.com
        ReportedBy: darrenm at opensolaris.org
         QAContact: hua.tang at sun.com
                CC: zfs-crypto-discuss at opensolaris.org
   Estimated Hours: 0.0


The pool does actually import but since zpool_load_key core dumps none of the
datasets (not even the ones with encryption=off) get mounted.

# zpool export tank
# zpool import tank
....

/1 at 1:   -> libzfs:zpool_load_key(0x813f4c8, 0x0, 0x1)
/1 at 1:     -> libzfs:zpool_get_prop_int(0x813f4c8, 0x8, 0x0)
/1 at 1:       -> libzfs:zpool_get_all_props(0x813f4c8)
/1 at 1:         -> libzfs:zcmd_alloc_dst_nvlist(0x80e0548, 0x8041654, 0x0)
/1 at 1:           -> libzfs:zfs_alloc(0x80e0548, 0x800)
/1 at 1:           <- libzfs:zfs_alloc() = 0x80dfc08
/1 at 1:         <- libzfs:zcmd_alloc_dst_nvlist() = 0
/1:     ioctl(3, ZFS_IOC_POOL_GET_PROPS, 0x08041654)    = 0
/1 at 1:         -> libzfs:zcmd_read_dst_nvlist(0x80e0548, 0x8041654,
0x813f5dc)
/1 at 1:         <- libzfs:zcmd_read_dst_nvlist() = 0
/1 at 1:         -> libzfs:zcmd_free_nvlists(0x8041654)
/1 at 1:         <- libzfs:zcmd_free_nvlists() = 1
/1 at 1:       <- libzfs:zpool_get_all_props() = 0
/1 at 1:       -> libzfs:zpool_prop_to_name(0x8)
/1 at 1:       <- libzfs:zpool_prop_to_name() = 0xfedf4fc4
/1 at 1:     <- libzfs:zpool_get_prop_int() = 11
/1 at 1:     -> libzfs:zpool_get_prop_int(0x813f4c8, 0x11, 0x0)
/1 at 1:       -> libzfs:zpool_prop_to_name(0x11)
/1 at 1:       <- libzfs:zpool_prop_to_name() = 0xfedf503c
/1 at 1:     <- libzfs:zpool_get_prop_int() = 2
/1 at 1:     -> libzfs:zpool_get_prop_int(0x813f4c8, 0x12, 0x0)
/1 at 1:       -> libzfs:zpool_prop_to_name(0x12)
/1 at 1:       <- libzfs:zpool_prop_to_name() = 0xfedf4fec
/1 at 1:     <- libzfs:zpool_get_prop_int() = 0x36e8ce31
/1:         Incurred fault #6, FLTBOUNDS  %pc = 0xFEDA82B3
/1:           siginfo: SIGSEGV SEGV_MAPERR addr=0x0000001C
/1:         Received signal #11, SIGSEGV [default]
/1:           siginfo: SIGSEGV SEGV_MAPERR addr=0x0000001C

0x12 is ZPOOL_PROP_SALT and the failing line of source is this one:

        default:
            cry->salt = zpool_get_prop_int(zhp, ZPOOL_PROP_SALT, NULL);

cry is NULL during a zpool import, in do_import():

        (void) zpool_load_key(zhp, NULL, B_TRUE);

That style of calling zpool_load_key should be allowed (similarly for
zfs_key_load().

I would rather see the salt only retrieved or generated after we actually know
that the format of the keysource is "passphrase" - there is no point
in storing
a salt when the key format is hex or raw.  

This possibly means moving the code currently in zpool_key_load() for dealing
with the slat down into use_key_material() function or similar in the
KEY_FORMAT_PASSPHRASE case of the switch.

Unfortunately though that may take some of the genericness away from that
function since it doesn''t currently know if it is dealing with a
dataset or
a pool.  I''m sure this can be resolved though.

-- 
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

http://defect.opensolaris.org/bz/show_bug.cgi?id=697





--- Comment #1 from Darren J Moffat <darrenm at opensolaris.org> 
2008-03-07 06:59:22 ---
I''m integrating into the gate a temporary fix for this issue by having
zpool_main.c call zpool_key_load with a non NULL zfs_crypt_t.  This will avoid
the issue for now.

This bug may well get closed by other means later.

-- 
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

http://defect.opensolaris.org/bz/show_bug.cgi?id=697


ajscarp at yahoo.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from ajscarp at yahoo.com  2008-03-12 00:06:30 ---
This has been completely taken care of by the March 11th fix

-- 
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.