RAW-OPLOCK failed

[Peter Somogyi]

Hi,

After setting "kernel oplocks = yes" in script/tests/selftest.sh, I get the 
following error during a "make test" against SAMBA_3_0_RELEASE:

Testing RAW-OPLOCK (0)
TEST OUTPUT:
Unknown parameter encountered: "passdb backend"
Ignoring unknown parameter "passdb backend"
Using seed 1152526195
Running RAW-OPLOCK
open a file with a normal oplock
a 2nd open should not cause a break
Acking to level II in oplock handler
(173) wrong value for break_info.count  got 0x1 - should be 0x0
unlink it - should also be no break
(181) wrong value for break_info.count  got 0x1 - should be 0x0
open with batch oplock
unlink should generate a break
Acking to level II in oplock handler
(207) wrong value for break_info.level  got 0x0 - should be 0x1
2nd unlink should not generate a break
writing should generate a self break to none
(222) wrong value for break_info.count  got 0x0 - should be 0x1
(223) wrong value for break_info.fnum  got 0x0 - should be 0x13e9
open with batch oplocksetting 
unlink should generate a break, which we ack as break to none
Acking to none in oplock handler
(249) wrong value for break_info.level  got 0x0 - should be 0x1
2nd unlink should not generate a break
writing should not generate a break
if we close on break then the unlink can succeed
(287) wrong value for break_info.level  got 0x0 - should be 0x1
a self read should not cause a break
a 2nd open should give a break
Acking to level II in oplock handler
(334) wrong value for break_info.level  got 0x0 - should be 0x1
a 2nd open should give a break to level II if the first open allowed shared 
read
Acking to level II in oplock handler
(357) wrong value for io.ntcreatex.out.oplock_level  got 0x0 - should be 0x3
(361) wrong value for break_info.level  got 0x0 - should be 0x1
write should trigger a break to none on both
(370) wrong value for break_info.count  got 0x0 - should be 0x2
a 2nd open should get an oplock when we close instead of ack
(403) wrong value for break_info.level  got 0x0 - should be 0x1
open with batch oplock
second open with attributes only shouldn't cause oplock break
(428) Incorrect status NT_STATUS_ACCESS_DENIED - should be NT_STATUS_OK
Acking to level II in oplock handler
TEST RAW-OPLOCK FAILED! - ...
RAW-OPLOCK took 1.05887 secs

Why it doesn't have to work with kernel oplocks?

(Or do I have to tune something in my system? - OpenSuSE 10.1, 
2.6.16.13-4-smp)

Peter