sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-01-19 10:03:58 +03:00
Code
samba-mirror/source4/torture/raw/oplock.c

4107 lines
130 KiB
C
Raw Normal View History

first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
/*
Unix SMB/CIFS implementation.
basic raw test suite for oplocks
Copyright (C) Andrew Tridgell 2003
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
r23792: convert Samba4 to GPLv3 There are still a few tidyups of old FSF addresses to come (in both s3 and s4). More commits soon. (This used to be commit fcf38a38ac691abd0fa51b89dc951a08e89fdafa)
2007-07-10 02:07:03 +00:00
the Free Software Foundation; either version 3 of the License, or
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
r23792: convert Samba4 to GPLv3 There are still a few tidyups of old FSF addresses to come (in both s3 and s4). More commits soon. (This used to be commit fcf38a38ac691abd0fa51b89dc951a08e89fdafa)
2007-07-10 02:07:03 +00:00
along with this program. If not, see <http://www.gnu.org/licenses/>.
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
*/
#include "includes.h"
r12542: Move some more prototypes out to seperate headers (This used to be commit 0aca5fd5130d980d07398f3291d294202aefe3c2)
2005-12-28 15:38:36 +00:00
#include "libcli/raw/libcliraw.h"
Install public header files again and include required prototypes. (This used to be commit 47ffbbf67435904754469544390b67d34c958343)
2008-04-02 04:53:27 +02:00
#include "libcli/raw/raw_proto.h"
r12694: Move some headers to the directory of the subsystem they belong to. (This used to be commit c722f665c90103f3ed57621c460e32ad33e7a8a3)
2006-01-03 15:40:05 +00:00
#include "libcli/libcli.h"
r14527: Fix build problems. (This used to be commit 863ca4014d9b821706ee90f58ab5d5cf3899a4c7)
2006-03-17 17:59:58 +00:00
#include "torture/util.h"
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
#include "lib/events/events.h"
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
#include "param/param.h"
#include "lib/cmdline/popt_common.h"
#include "libcli/resolve/resolve.h"
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
#define CHECK_VAL(v, correct) do { \
if ((v) != (correct)) { \
RAW-OPLOCK: use torture_result(TORTURE_FAIL) so that the knownfailures file work metze (This used to be commit 656011add338ab87188849c4a716d3c65a102a74)
2008-02-27 16:22:04 +01:00
torture_result(tctx, TORTURE_FAIL, "(%s): wrong value for %s got 0x%x - should be 0x%x\n", \
r24041: After metze's ack, check in torture_comment from Zack Krisch <zack.kirsch@isilon.com> (This used to be commit b1148b7ab84a18d4fea771c887ed7d535841982b)
2007-07-25 09:57:14 +00:00
__location__, #v, (int)v, (int)correct); \
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
ret = false; \
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
}} while (0)
s4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22 Some servers choose to mark a client as bad if they fail an oplock break request by timing out (win7 is an example). Once the client is marked as bad, future oplock requests will timeout instantly. This causes subsequent runs of this test to fail, so rather than erroring out as a failure, a warning is printed instead. There is also a bug in w2k3 where it was incorrectly returning contending a share mode lock. It worked in XP and has been re-fixed in win7. This can also now be run against samba3.
2009-12-02 18:39:05 -08:00
#define CHECK_RANGE(v, min, max) do { \
if ((v) < (min) || (v) > (max)) { \
torture_warning(tctx, "(%s): wrong value for %s got " \
"%d - should be between %d and %d\n", \
__location__, #v, (int)v, (int)min, (int)max); \
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
}} while (0)
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
#define CHECK_STRMATCH(v, correct) do { \
if (!v || strstr((v),(correct)) == NULL) { \
RAW-OPLOCK: use torture_result(TORTURE_FAIL) so that the knownfailures file work metze (This used to be commit 656011add338ab87188849c4a716d3c65a102a74)
2008-02-27 16:22:04 +01:00
torture_result(tctx, TORTURE_FAIL, "(%s): wrong value for %s got '%s' - should be '%s'\n", \
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
__location__, #v, v?v:"NULL", correct); \
ret = false; \
} \
} while (0)
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
#define CHECK_STATUS(tctx, status, correct) do { \
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
if (!NT_STATUS_EQUAL(status, correct)) { \
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_result(tctx, TORTURE_FAIL, __location__": Incorrect status %s - should be %s", \
nt_errstr(status), nt_errstr(correct)); \
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
ret = false; \
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
goto done; \
}} while (0)
static struct {
int fnum;
r943: change samba4 to use 'uint8_t' instead of 'unsigned char' metze (This used to be commit b5378803fdcb3b3afe7c2932a38828e83470f61a)
2004-05-29 08:11:46 +00:00
uint8_t level;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
int count;
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
int failures;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
} break_info;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
#define BASEDIR "\\test_oplock"
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
/*
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
a handler function for oplock break requests. Ack it as a break to level II if possible
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
*/
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
static bool oplock_handler_ack_to_given(struct smbcli_transport *transport,
uint16_t tid, uint16_t fnum,
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
uint8_t level, void *private_data)
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
{
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
struct smbcli_tree *tree = (struct smbcli_tree *)private_data;
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
const char *name;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
break_info.fnum = fnum;
break_info.level = level;
break_info.count++;
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
switch (level) {
case OPLOCK_BREAK_TO_LEVEL_II:
name = "level II";
break;
case OPLOCK_BREAK_TO_NONE:
name = "none";
break;
default:
name = "unknown";
break_info.failures++;
}
printf("Acking to %s [0x%02X] in oplock handler\n",
name, level);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r1654: rename cli_ -> smbcli_ rename CLI_ -> SMBCLI_ metze (This used to be commit 8441750fd9427dd6fe477f27e603821b4026f038)
2004-08-04 13:23:35 +00:00
return smbcli_oplock_ack(tree, fnum, level);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
}
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
/*
a handler function for oplock break requests. Ack it as a break to none
*/
r25000: Fix some more C++ compatibility warnings. (This used to be commit 08bb1ef643ab906f1645cf6f32763dc73b1884e4)
2007-09-07 15:08:14 +00:00
static bool oplock_handler_ack_to_none(struct smbcli_transport *transport,
uint16_t tid, uint16_t fnum,
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
uint8_t level, void *private_data)
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
{
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
struct smbcli_tree *tree = (struct smbcli_tree *)private_data;
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
break_info.fnum = fnum;
break_info.level = level;
break_info.count++;
printf("Acking to none in oplock handler\n");
return smbcli_oplock_ack(tree, fnum, OPLOCK_BREAK_TO_NONE);
}
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
/*
a handler function for oplock break requests. Let it timeout
*/
static bool oplock_handler_timeout(struct smbcli_transport *transport,
uint16_t tid, uint16_t fnum,
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
uint8_t level, void *private_data)
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
{
break_info.fnum = fnum;
break_info.level = level;
break_info.count++;
printf("Let oplock break timeout\n");
return true;
}
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
static void oplock_handler_close_recv(struct smbcli_request *req)
{
NTSTATUS status;
status = smbcli_request_simple_recv(req);
if (!NT_STATUS_IS_OK(status)) {
printf("close failed in oplock_handler_close\n");
break_info.failures++;
}
}
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
/*
a handler function for oplock break requests - close the file
*/
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
static bool oplock_handler_close(struct smbcli_transport *transport, uint16_t tid,
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
uint16_t fnum, uint8_t level, void *private_data)
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
{
union smb_close io;
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
struct smbcli_tree *tree = (struct smbcli_tree *)private_data;
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
struct smbcli_request *req;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
break_info.fnum = fnum;
break_info.level = level;
break_info.count++;
io.close.level = RAW_CLOSE_CLOSE;
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
io.close.in.file.fnum = fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
io.close.in.write_time = 0;
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
req = smb_raw_close_send(tree, &io);
if (req == NULL) {
printf("failed to send close in oplock_handler_close\n");
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return false;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
}
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
req->async.fn = oplock_handler_close_recv;
s4:libcliraw: s/private/private_data metze
2009-02-02 10:17:00 +01:00
req->async.private_data = NULL;
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return true;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
}
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
static bool open_connection_no_level2_oplocks(struct torture_context *tctx,
struct smbcli_state **c)
{
NTSTATUS status;
struct smbcli_options options;
Cope with API changes.
2008-09-30 03:07:08 +02:00
struct smbcli_session_options session_options;
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
lp_smbcli_options(tctx->lp_ctx, &options);
Cope with API changes.
2008-09-30 03:07:08 +02:00
lp_smbcli_session_options(tctx->lp_ctx, &session_options);
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
options.use_level2_oplocks = false;
status = smbcli_full_connection(tctx, c,
torture_setting_string(tctx, "host", NULL),
lp_smb_ports(tctx->lp_ctx),
torture_setting_string(tctx, "share", NULL),
Fix the build.
2008-11-02 02:30:21 +01:00
NULL, lp_socket_options(tctx->lp_ctx), cmdline_credentials,
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
lp_resolve_context(tctx->lp_ctx),
Eliminate another instance of global_loadparm.
2008-10-24 13:13:27 +02:00
tctx->ev, &options, &session_options,
Remove another use of global_loadparm. Eventually, we should move some of these parameters into a separate struct (perhaps into smb_transport_options?), to avoid the long lists of parameters.
2008-11-02 16:07:28 +01:00
lp_gensec_settings(tctx, tctx->lp_ctx));
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
if (!NT_STATUS_IS_OK(status)) {
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "Failed to open connection - %s\n",
nt_errstr(status));
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
return false;
}
return true;
}
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
/*
Timer handler function notifies the registering function that time is up
*/
static void timeout_cb(struct tevent_context *ev,
struct tevent_timer *te,
struct timeval current_time,
void *private_data)
{
bool *timesup = (bool *)private_data;
*timesup = true;
return;
}
/*
Wait a short period of time to receive a single oplock break request
*/
static void torture_wait_for_oplock_break(struct torture_context *tctx)
{
TALLOC_CTX *tmp_ctx = talloc_new(NULL);
struct tevent_timer *te = NULL;
struct timeval ne;
bool timesup = false;
int old_count = break_info.count;
/* Wait .1 seconds for an oplock break */
ne = tevent_timeval_current_ofs(0, 100000);
if ((te = event_add_timed(tctx->ev, tmp_ctx, ne, timeout_cb, &timesup))
== NULL)
{
torture_comment(tctx, "Failed to wait for an oplock break. "
"test results may not be accurate.");
goto done;
}
while (!timesup && break_info.count < old_count + 1) {
if (event_loop_once(tctx->ev) != 0) {
torture_comment(tctx, "Failed to wait for an oplock "
"break. test results may not be "
"accurate.");
goto done;
}
}
done:
/* We don't know if the timed event fired and was freed, we received
* our oplock break, or some other event triggered the loop. Thus,
* we create a tmp_ctx to be able to safely free/remove the timed
* event in all 3 cases. */
talloc_free(tmp_ctx);
return;
}
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
static uint8_t get_break_level1_to_none_count(struct torture_context *tctx)
{
return torture_setting_bool(tctx, "2_step_break_to_none", false) ?
2 : 1;
}
RAW-OPLOCK: rename test (NORMAL => EXCLUSIVE1) metze (This used to be commit 546536dde07d82dbba6724df95cdd1ba50caf8c5)
2008-02-26 14:58:00 +01:00
static bool test_raw_oplock_exclusive1(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
{
RAW-OPLOCK: rename test (NORMAL => EXCLUSIVE1) metze (This used to be commit 546536dde07d82dbba6724df95cdd1ba50caf8c5)
2008-02-26 14:58:00 +01:00
const char *fname = BASEDIR "\\test_exclusive1.dat";
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
NTSTATUS status;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
bool ret = true;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
union smb_open io;
r14173: change smb interface structures to always use a union smb_file, to abtract - const char *path fot qpathinfo and setpathinfo - uint16_t fnum for SMB - smb2_handle handle for SMB2 the idea is to later add a struct ntvfs_handle *ntvfs so that the ntvfs subsystem don't need to know the difference between SMB and SMB2 metze (This used to be commit 2ef3f5970901b5accdb50f0d0115b5d46b0c788f)
2006-03-10 20:49:20 +00:00
union smb_unlink unl;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
/* cleanup */
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_unlink(cli1->tree, fname);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r4035: more effort on consistent naming of the access mask bits. This removes the duplicate named SEC_RIGHTS_MAXIMUM_ALLOWED and SEC_RIGHTS_FULL_CONTROL, which are just other names for SEC_FLAG_MAXIMUM_ALLOWED and SEC_RIGHTS_FILE_ALL. The latter names match the new naming conventions in security.idl Also added names for the generic->specific mappings for files are directories (This used to be commit 17a4e0b3aca227b40957ed1e0c57e498debc6ddf)
2004-12-02 04:37:36 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "EXCLUSIVE1: open a file with an exclusive oplock (share mode: none)\n");
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "a 2nd open should not cause a break\n");
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15016: add a test for a 2nd open with an exclusive oplock. It does not cause a break, as exclusive oplocks don't cache close requests (This used to be commit 9f2617a062e99805c42fe9b2fa0179b1672b25e5)
2006-04-10 06:06:26 +00:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "unlink it - should also be no break\n");
r14173: change smb interface structures to always use a union smb_file, to abtract - const char *path fot qpathinfo and setpathinfo - uint16_t fnum for SMB - smb2_handle handle for SMB2 the idea is to later add a struct ntvfs_handle *ntvfs so that the ntvfs subsystem don't need to know the difference between SMB and SMB2 metze (This used to be commit 2ef3f5970901b5accdb50f0d0115b5d46b0c788f)
2006-03-10 20:49:20 +00:00
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_unlink(cli2->tree, &unl);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.count, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
static bool test_raw_oplock_exclusive2(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_exclusive2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_unlink unl;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "EXCLUSIVE2: open a file with an exclusive oplock (share mode: all)\n");
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
torture_comment(tctx, "a 2nd open should cause a break to level 2\n");
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
ZERO_STRUCT(break_info);
/* now we have 2 level II oplocks... */
torture_comment(tctx, "try to unlink it - should not cause a break, but a sharing violation\n");
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
status = smb_raw_unlink(cli2->tree, &unl);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
torture_comment(tctx, "close 1st handle\n");
smbcli_close(cli1->tree, fnum);
torture_comment(tctx, "try to unlink it - should not cause a break, but a sharing violation\n");
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
status = smb_raw_unlink(cli2->tree, &unl);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_comment(tctx, "close 2nd handle\n");
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
smbcli_close(cli2->tree, fnum2);
torture_comment(tctx, "unlink it\n");
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
status = smb_raw_unlink(cli2->tree, &unl);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
static bool test_raw_oplock_exclusive3(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_exclusive3.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_setfileinfo sfi;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
s4 torture: Change RAW-OPLOCK to use the documented version of SET_END_OF_FILE_INFO The passtrhough version of SET_END_OF_FILE_INFO is tested in RAW-SFILEINFO-END-OF-FILE. Additionally, the first opener is changed to use SHARE_WRITE for the share mode since SET_END_OF_FILE_INFO actually writes data to the file via truncating/extending.
2009-11-25 08:03:36 -08:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_WRITE;
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "EXCLUSIVE3: open a file with an exclusive oplock (share mode: none)\n");
Samba3 now passes the test_raw_oplock_exclusive3 test. Jeremy. (This used to be commit b2007956aa4534f22ad7fd85b0aee0be769548ae)
2008-03-11 12:37:20 -07:00
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
torture_comment(tctx, "setpathinfo EOF should trigger a break to none\n");
ZERO_STRUCT(sfi);
s4 torture: Change oplock to use passthrough for exclusive3/batch11 In light of the INVALID_LEVEL that is seen for RAW_SFILEINFO_END_OF_FILE_INFO requests on a path, I'm changing these back to using the passthrough RAW_SFILEINFO_END_OF_FILE_INFORMATION to test the oplock break behavior as originally intended
2009-11-30 16:43:00 -08:00
sfi.generic.level = RAW_SFILEINFO_END_OF_FILE_INFORMATION;
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
sfi.generic.in.file.path = fname;
sfi.end_of_file_info.in.size = 100;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
static bool test_raw_oplock_exclusive4(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_exclusive4.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "EXCLUSIVE4: open with exclusive oplock\n");
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
torture_comment(tctx, "second open with attributes only shouldn't cause oplock break\n");
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, NO_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
static bool test_raw_oplock_exclusive5(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_exclusive5.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "EXCLUSIVE5: open with exclusive oplock\n");
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
torture_comment(tctx, "second open with attributes only and NTCREATEX_DISP_OVERWRITE_IF dispostion causes oplock break\n");
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OVERWRITE_IF;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
static bool test_raw_oplock_exclusive6(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_exclusive6_1.dat";
const char *fname2 = BASEDIR "\\test_exclusive6_2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_rename rn;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "EXCLUSIVE6: open a file with an exclusive "
"oplock (share mode: none)\n");
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED | NTCREATEX_FLAGS_REQUEST_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "rename should not generate a break but get a "
"sharing violation\n");
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_RENAME;
rn.rename.in.pattern1 = fname1;
rn.rename.in.pattern2 = fname2;
rn.rename.in.attrib = 0;
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "trying rename while first file open\n");
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
status = smb_raw_rename(cli2->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
s4 torture: Add RAW-OPLOCK-EXCLUSIVE7 which is similar to BATCH19
2009-12-02 11:39:50 -08:00
/**
* Exclusive version of batch19
*/
static bool test_raw_oplock_exclusive7(struct torture_context *tctx,
struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_exclusiv6_1.dat";
const char *fname2 = BASEDIR "\\test_exclusiv6_2.dat";
const char *fname3 = BASEDIR "\\test_exclusiv6_3.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_fileinfo qfi;
union smb_setfileinfo sfi;
uint16_t fnum=0;
uint16_t fnum2 = 0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
smbcli_unlink(cli1->tree, fname3);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given,
cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
io.ntcreatex.in.root_fid.fnum = 0;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE | NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
torture_comment(tctx, "open a file with an exclusive oplock (share "
"mode: none)\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
torture_comment(tctx, "setpathinfo rename info should trigger a break "
"to none\n");
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
sfi.generic.in.file.path = fname1;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname2+strlen(BASEDIR)+1;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
torture_wait_for_oplock_break(tctx);
CHECK_VAL(break_info.failures, 0);
if (TARGET_IS_WINXP(tctx)) {
/* XP incorrectly breaks to level2. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
} else {
/* Exclusive oplocks should not be broken on rename. */
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.count, 0);
}
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname2);
/* Try breaking to level2 and then see if rename breaks the level2.*/
ZERO_STRUCT(break_info);
io.ntcreatex.in.fname = fname2;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
torture_wait_for_oplock_break(tctx);
CHECK_VAL(break_info.failures, 0);
if (TARGET_IS_WINXP(tctx)) {
/* XP already broke to level2. */
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.count, 0);
} else {
/* Break to level 2 expected. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
}
ZERO_STRUCT(break_info);
sfi.generic.in.file.path = fname2;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname1+strlen(BASEDIR)+1;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
/* Level2 oplocks are not broken on rename. */
torture_wait_for_oplock_break(tctx);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.count, 0);
/* Close and re-open file with oplock. */
smbcli_close(cli1->tree, fnum);
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
torture_comment(tctx, "setfileinfo rename info on a client's own fid "
"should not trigger a break nor a violation\n");
ZERO_STRUCT(break_info);
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
sfi.generic.in.file.fnum = fnum;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname3+strlen(BASEDIR)+1;
status = smb_raw_setfileinfo(cli1->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
torture_wait_for_oplock_break(tctx);
if (TARGET_IS_WINXP(tctx)) {
/* XP incorrectly breaks to level2. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
} else {
CHECK_VAL(break_info.count, 0);
}
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname3);
done:
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
static bool test_raw_oplock_batch1(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch1.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_unlink unl;
uint16_t fnum=0;
char c = 0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
/*
with a batch oplock we get a break
*/
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH1: open with batch oplock\n");
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "unlink should generate a break\n");
r14173: change smb interface structures to always use a union smb_file, to abtract - const char *path fot qpathinfo and setpathinfo - uint16_t fnum for SMB - smb2_handle handle for SMB2 the idea is to later add a struct ntvfs_handle *ntvfs so that the ntvfs subsystem don't need to know the difference between SMB and SMB2 metze (This used to be commit 2ef3f5970901b5accdb50f0d0115b5d46b0c788f)
2006-03-10 20:49:20 +00:00
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_unlink(cli2->tree, &unl);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 1);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.fnum, fnum);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "2nd unlink should not generate a break\n");
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
ZERO_STRUCT(break_info);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_unlink(cli2->tree, &unl);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 0);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "writing should generate a self break to none\n");
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_write(cli1->tree, fnum, 0, &c, 0, 1);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
torture_wait_for_oplock_break(tctx);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.count, 1);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch2(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_unlink unl;
uint16_t fnum=0;
char c = 0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH2: open with batch oplock\n");
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "unlink should generate a break, which we ack as break to none\n");
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_none, cli1->tree);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_unlink(cli2->tree, &unl);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "2nd unlink should not generate a break\n");
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
ZERO_STRUCT(break_info);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_unlink(cli2->tree, &unl);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 0);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "writing should not generate a break\n");
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_write(cli1->tree, fnum, 0, &c, 0, 1);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 0);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch3(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch3.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_unlink unl;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH3: if we close on break then the unlink can succeed\n");
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_oplock_handler(cli1->transport, oplock_handler_close, cli1->tree);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
r14173: change smb interface structures to always use a union smb_file, to abtract - const char *path fot qpathinfo and setpathinfo - uint16_t fnum for SMB - smb2_handle handle for SMB2 the idea is to later add a struct ntvfs_handle *ntvfs so that the ntvfs subsystem don't need to know the difference between SMB and SMB2 metze (This used to be commit 2ef3f5970901b5accdb50f0d0115b5d46b0c788f)
2006-03-10 20:49:20 +00:00
unl.unlink.in.pattern = fname;
unl.unlink.in.attrib = 0;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_unlink(cli2->tree, &unl);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 1);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.fnum, fnum);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
CHECK_VAL(break_info.level, 1);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch4(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch4.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_read rd;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH4: a self read should not cause a break\n");
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
s4 torture: Convert to a more modern version of read in RAW-OPLOCK-BATCH4
2009-11-20 11:25:43 -08:00
rd.readx.level = RAW_READ_READX;
rd.readx.in.file.fnum = fnum;
rd.readx.in.mincnt = 1;
rd.readx.in.maxcnt = 1;
rd.readx.in.offset = 0;
rd.readx.in.remaining = 0;
rd.readx.in.read_for_execute = false;
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_read(cli1->tree, &rd);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.count, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch5(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch5.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH5: a 2nd open should give a break\n");
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
CHECK_VAL(break_info.level, 1);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch6(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch6.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
char c = 0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli2->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH6: a 2nd open should give a break to level II if the first open allowed shared read\n");
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_READ | SEC_RIGHTS_FILE_WRITE;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ | NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
//torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.level, 1);
CHECK_VAL(break_info.failures, 0);
ZERO_STRUCT(break_info);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "write should trigger a break to none on both\n");
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_write(cli1->tree, fnum, 0, &c, 0, 1);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
/* We expect two breaks */
torture_wait_for_oplock_break(tctx);
torture_wait_for_oplock_break(tctx);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.count, 2);
CHECK_VAL(break_info.level, 0);
CHECK_VAL(break_info.failures, 0);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch7(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch7.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH7: a 2nd open should get an oplock when we close instead of ack\n");
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
ZERO_STRUCT(break_info);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_oplock_handler(cli1->transport, oplock_handler_close, cli1->tree);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum2);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
CHECK_VAL(break_info.level, 1);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli2->tree, fnum);
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch8(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch8.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH8: open with batch oplock\n");
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "second open with attributes only shouldn't cause oplock break\n");
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r4011: get rid of rpc_secdes.h and replace it with a single sane set of definitions for security access masks, in security.idl The previous definitions were inconsistently named, and contained many duplicate and misleading entries. I kept finding myself tripping up while using them. (This used to be commit 01c0fa722f80ceeb3f81f01987de95f365a2ed3d)
2004-11-30 04:33:27 +00:00
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, NO_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r865: Regression test for attribute-only opens not causing an oplock break. Samba3 server used to get this wrong. Jeremy. (This used to be commit 4120b3bdf6668909efa0af97b2c63adea22c35da)
2004-05-25 00:05:51 +00:00
CHECK_VAL(break_info.count, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch9(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch9.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
char c = 0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_unlink(cli1->tree, fname);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH9: open with attributes only can create file\n");
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r4011: get rid of rpc_secdes.h and replace it with a single sane set of definitions for security access masks, in security.idl The previous definitions were inconsistently named, and contained many duplicate and misleading entries. I kept finding myself tripping up while using them. (This used to be commit 01c0fa722f80ceeb3f81f01987de95f365a2ed3d)
2004-11-30 04:33:27 +00:00
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_CREATE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "Subsequent normal open should break oplock on attribute only open to level II\n");
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r4035: more effort on consistent naming of the access mask bits. This removes the duplicate named SEC_RIGHTS_MAXIMUM_ALLOWED and SEC_RIGHTS_FULL_CONTROL, which are just other names for SEC_FLAG_MAXIMUM_ALLOWED and SEC_RIGHTS_FILE_ALL. The latter names match the new naming conventions in security.idl Also added names for the generic->specific mappings for files are directories (This used to be commit 17a4e0b3aca227b40957ed1e0c57e498debc6ddf)
2004-12-02 04:37:36 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
r15033: expanded the RAW-OPLOCK test to include the different cases where the client chooses to break to none or break to level II, when offered a level II break from exclusive (This used to be commit 9a35b14422980fb9a8049058fd30e79d19a1e242)
2006-04-11 04:40:11 +00:00
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli2->tree, fnum2);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "third oplocked open should grant level2 without break\n");
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli2->tree);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(break_info.count, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "write should trigger a break to none on both\n");
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_write(cli2->tree, fnum2, 0, &c, 0, 1);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
/* We expect two breaks */
torture_wait_for_oplock_break(tctx);
torture_wait_for_oplock_break(tctx);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(break_info.count, 2);
CHECK_VAL(break_info.level, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch10(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch10.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH10: Open with oplock after a non-oplock open should grant level2\n");
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
ZERO_STRUCT(break_info);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(break_info.count, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, 0);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli2->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(break_info.count, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "write should trigger a break to none\n");
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
{
union smb_write wr;
wr.write.level = RAW_WRITE_WRITE;
r14256: - rename smb_file -> smb_handle - move it into the in/out substructs again - allow file.path only on smb_fileinfo/smb_setfileinfo metze (This used to be commit be6d5298a2cdb7e7c61d70471bad445645af5963)
2006-03-12 22:48:25 +00:00
wr.write.in.file.fnum = fnum;
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
wr.write.in.count = 1;
wr.write.in.offset = 0;
wr.write.in.remaining = 0;
r13004: fix compiler warnings metze (This used to be commit 833efdf8a943b210ba8e5b219dc754260001bedb)
2006-01-18 16:20:33 +00:00
wr.write.in.data = (const uint8_t *)"x";
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_write(cli1->tree, &wr);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
}
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r8229: A level2 oplock does not conflict a no-oplock open, but is broken on first write. Test that. Volker (This used to be commit 2ad02d5e09ca00ca998e363b535b16af45c82a5f)
2005-07-08 06:29:06 +00:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum2);
CHECK_VAL(break_info.level, 0);
r12549: fixed the problem with serialisation and the RAW-OPLOCK test (This used to be commit ad3b8286636f4101fa6528cb447aeeb7ee7507a4)
2005-12-28 22:18:45 +00:00
CHECK_VAL(break_info.failures, 0);
r946: Updated attribute only oplock tester to cover more cases. Jeremy. (This used to be commit 6c9d32173b26353e8ae00f7f6b666290651883e6)
2004-05-30 08:12:17 +00:00
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch11(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch11.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_setfileinfo sfi;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_unlink(cli1->tree, fname);
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
s4 torture: Change RAW-OPLOCK to use the documented version of SET_END_OF_FILE_INFO The passtrhough version of SET_END_OF_FILE_INFO is tested in RAW-SFILEINFO-END-OF-FILE. Additionally, the first opener is changed to use SHARE_WRITE for the share mode since SET_END_OF_FILE_INFO actually writes data to the file via truncating/extending.
2009-11-25 08:03:36 -08:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_WRITE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
/* Test if a set-eof on pathname breaks an exclusive oplock. */
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH11: Test if setpathinfo set EOF breaks oplocks.\n");
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
NTCREATEX_FLAGS_REQUEST_OPLOCK |
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_CREATE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
fnum = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
ZERO_STRUCT(sfi);
s4 torture: Change oplock to use passthrough for exclusive3/batch11 In light of the INVALID_LEVEL that is seen for RAW_SFILEINFO_END_OF_FILE_INFO requests on a path, I'm changing these back to using the passthrough RAW_SFILEINFO_END_OF_FILE_INFORMATION to test the oplock break behavior as originally intended
2009-11-30 16:43:00 -08:00
sfi.generic.level = RAW_SFILEINFO_END_OF_FILE_INFORMATION;
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
sfi.generic.in.file.path = fname;
sfi.end_of_file_info.in.size = 100;
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_setpathinfo(cli2->tree, &sfi);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
CHECK_VAL(break_info.failures, 0);
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
CHECK_VAL(break_info.level, 0);
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_close(cli1->tree, fnum);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch12(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch12.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_setfileinfo sfi;
RAW-OPLOCK: remove unused var (This used to be commit 6820da44828172769d9fdfa161acf762e3937a88)
2008-02-23 11:51:43 +01:00
uint16_t fnum=0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
smbcli_unlink(cli1->tree, fname);
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
/* Test if a set-allocation size on pathname breaks an exclusive oplock. */
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH12: Test if setpathinfo allocation size breaks oplocks.\n");
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
ZERO_STRUCT(break_info);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
NTCREATEX_FLAGS_REQUEST_OPLOCK |
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_CREATE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
fnum = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
ZERO_STRUCT(sfi);
sfi.generic.level = SMB_SFILEINFO_ALLOCATION_INFORMATION;
sfi.generic.in.file.path = fname;
sfi.allocation_info.in.alloc_size = 65536 * 8;
r17139: Checking in Volker's extended oplock test. Jeremy (This used to be commit 720c425af167373aaedf739d1d7794ada3fcddf6)
2006-07-19 16:44:50 +00:00
status = smb_raw_setpathinfo(cli2->tree, &sfi);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
r15947: Do the same test on setting allocation info. Jeremy. (This used to be commit b8f6f62acd2ae79c87303b72e8bea22abc3a6b6e)
2006-05-30 03:57:43 +00:00
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.level, 0);
r15946: Added check that proves that setpathinfo on an oplocked open file generates a break. Jeremy. (This used to be commit 061b519b27cecbeb0687426201caaa7e8d95d856)
2006-05-30 03:36:51 +00:00
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
smbcli_close(cli1->tree, fnum);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch13(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch13.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH13: open with batch oplock\n");
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
ZERO_STRUCT(break_info);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "second open with attributes only and NTCREATEX_DISP_OVERWRITE dispostion causes oplock break\n");
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OVERWRITE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
static bool test_raw_oplock_batch14(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch14.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH14: open with batch oplock\n");
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(tctx, "second open with attributes only and NTCREATEX_DISP_SUPERSEDE dispostion causes oplock break\n");
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OVERWRITE;
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
status = smb_raw_open(cli2->tree, tctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(tctx, status, NT_STATUS_OK);
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
r17141: Prove that create dispositions with truncate and overwrite break oplocks. Jeremy. (This used to be commit 94886f8c4029a9661a0bce2f7d9c41a3a5eaa366)
2006-07-19 18:35:07 +00:00
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
done:
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
return ret;
}
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
static bool test_raw_oplock_batch15(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch15.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_fileinfo qfi;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
/* Test if a qpathinfo all info on pathname breaks a batch oplock. */
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH15: Test if qpathinfo all info breaks a batch oplock (should not).\n");
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_CREATE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.path = fname;
status = smb_raw_pathinfo(cli2->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
CHECK_VAL(break_info.count, 0);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH16 and tests attribute only with OVERWRITE_IF metze (This used to be commit 2453628623795c34ece18a285e6c2de40c4518de)
2008-02-27 09:41:44 +01:00
static bool test_raw_oplock_batch16(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch16.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH16 and tests attribute only with OVERWRITE_IF metze (This used to be commit 2453628623795c34ece18a285e6c2de40c4518de)
2008-02-27 09:41:44 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH16 and tests attribute only with OVERWRITE_IF metze (This used to be commit 2453628623795c34ece18a285e6c2de40c4518de)
2008-02-27 09:41:44 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH16: open with batch oplock\n");
RAW-OPLOCK: add BATCH16 and tests attribute only with OVERWRITE_IF metze (This used to be commit 2453628623795c34ece18a285e6c2de40c4518de)
2008-02-27 09:41:44 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
torture_comment(tctx, "second open with attributes only and NTCREATEX_DISP_OVERWRITE_IF dispostion causes oplock break\n");
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.access_mask = SEC_FILE_READ_ATTRIBUTE|SEC_FILE_WRITE_ATTRIBUTE|SEC_STD_SYNCHRONIZE;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OVERWRITE_IF;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Allow some implementations to break from level1 to none with two breaks
2009-12-01 15:29:00 -08:00
CHECK_VAL(break_info.count, get_break_level1_to_none_count(tctx));
RAW-OPLOCK: add BATCH16 and tests attribute only with OVERWRITE_IF metze (This used to be commit 2453628623795c34ece18a285e6c2de40c4518de)
2008-02-27 09:41:44 +01:00
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
static bool test_raw_oplock_batch17(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_batch17_1.dat";
const char *fname2 = BASEDIR "\\test_batch17_2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_rename rn;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH17: open a file with an batch oplock (share mode: none)\n");
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
torture_comment(tctx, "rename should trigger a break\n");
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_RENAME;
rn.rename.in.pattern1 = fname1;
rn.rename.in.pattern2 = fname2;
rn.rename.in.attrib = 0;
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "trying rename while first file open\n");
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
status = smb_raw_rename(cli2->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
static bool test_raw_oplock_batch18(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_batch18_1.dat";
const char *fname2 = BASEDIR "\\test_batch18_2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_rename rn;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH18: open a file with an batch oplock (share mode: none)\n");
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
torture_comment(tctx, "ntrename should trigger a break\n");
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_NTRENAME;
rn.ntrename.in.attrib = 0;
rn.ntrename.in.flags = RENAME_FLAG_RENAME;
rn.ntrename.in.old_name = fname1;
rn.ntrename.in.new_name = fname2;
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "trying rename while first file open\n");
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
status = smb_raw_rename(cli2->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
static bool test_raw_oplock_batch19(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_batch19_1.dat";
const char *fname2 = BASEDIR "\\test_batch19_2.dat";
const char *fname3 = BASEDIR "\\test_batch19_3.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_fileinfo qfi;
union smb_setfileinfo sfi;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
smbcli_unlink(cli1->tree, fname3);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
s4 torture: Be more permissive with share modes for oplock testing Share modes are tested elsewhere, and there is currently an outstanding issue about share mode contention for nt-passthrough levels: http://lists.samba.org/archive/cifs-protocol/2009-December/001227.html
2009-12-21 14:48:43 -08:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE | NTCREATEX_SHARE_ACCESS_DELETE;
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH19: open a file with an batch oplock (share mode: none)\n");
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
torture_comment(tctx, "setpathinfo rename info should trigger a break "
"to none\n");
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
sfi.generic.in.file.path = fname1;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname2+strlen(BASEDIR)+1;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
CHECK_VAL(break_info.failures, 0);
if (TARGET_IS_WINXP(tctx)) {
/* Win XP breaks to level2. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
} else if (TARGET_IS_W2K3(tctx) || TARGET_IS_W2K8(tctx) ||
TARGET_IS_SAMBA3(tctx) || TARGET_IS_SAMBA4(tctx)) {
/* Win2K3/2k8 incorrectly doesn't break at all. */
CHECK_VAL(break_info.count, 0);
} else {
/* win7/2k8r2 break to none. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
}
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname2);
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
/* Close and re-open file with oplock. */
smbcli_close(cli1->tree, fnum);
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
torture_comment(tctx, "setfileinfo rename info on a client's own fid "
"should not trigger a break nor a violation\n");
ZERO_STRUCT(break_info);
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
Test trans2 renames with oplocks. It looks like trans2 renames ignore share modes and oplock breaks :-(. Jeremy. (This used to be commit a8a5339cf2ca218668f888eced5ffe7ce059553c)
2008-03-11 10:21:09 -07:00
sfi.generic.in.file.fnum = fnum;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname3+strlen(BASEDIR)+1;
status = smb_raw_setfileinfo(cli1->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
if (TARGET_IS_WINXP(tctx)) {
/* XP incorrectly breaks to level2. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
} else {
CHECK_VAL(break_info.count, 0);
}
Test trans2 renames with oplocks. It looks like trans2 renames ignore share modes and oplock breaks :-(. Jeremy. (This used to be commit a8a5339cf2ca218668f888eced5ffe7ce059553c)
2008-03-11 10:21:09 -07:00
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname3);
done:
s4 torture: Do a better job of closing open files in RAW-OPLOCK.
2009-12-21 14:38:45 -08:00
smbcli_close(cli1->tree, fnum);
Test trans2 renames with oplocks. It looks like trans2 renames ignore share modes and oplock breaks :-(. Jeremy. (This used to be commit a8a5339cf2ca218668f888eced5ffe7ce059553c)
2008-03-11 10:21:09 -07:00
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
/****************************************************
Called from raw-rename - we need oplock handling for
this test so this is why it's in oplock.c, not rename.c
****************************************************/
bool test_trans2rename(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_trans2rename_1.dat";
const char *fname2 = BASEDIR "\\test_trans2rename_2.dat";
const char *fname3 = BASEDIR "\\test_trans2rename_3.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_fileinfo qfi;
union smb_setfileinfo sfi;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
smbcli_unlink(cli1->tree, fname3);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
Test trans2 renames with oplocks. It looks like trans2 renames ignore share modes and oplock breaks :-(. Jeremy. (This used to be commit a8a5339cf2ca218668f888eced5ffe7ce059553c)
2008-03-11 10:21:09 -07:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
s4 torture: Be more permissive with share modes for oplock testing Share modes are tested elsewhere, and there is currently an outstanding issue about share mode contention for nt-passthrough levels: http://lists.samba.org/archive/cifs-protocol/2009-December/001227.html
2009-12-21 14:48:43 -08:00
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE | NTCREATEX_SHARE_ACCESS_DELETE;
Test trans2 renames with oplocks. It looks like trans2 renames ignore share modes and oplock breaks :-(. Jeremy. (This used to be commit a8a5339cf2ca218668f888eced5ffe7ce059553c)
2008-03-11 10:21:09 -07:00
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
torture_comment(tctx, "open a file with an exclusive oplock (share mode: none)\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
torture_comment(tctx, "setpathinfo rename info should not trigger a break nor a violation\n");
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
sfi.generic.in.file.path = fname1;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname2+strlen(BASEDIR)+1;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
Test trans2 renames with oplocks. It looks like trans2 renames ignore share modes and oplock breaks :-(. Jeremy. (This used to be commit a8a5339cf2ca218668f888eced5ffe7ce059553c)
2008-03-11 10:21:09 -07:00
CHECK_VAL(break_info.count, 0);
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname2);
torture_comment(tctx, "setfileinfo rename info should not trigger a break nor a violation\n");
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
sfi.generic.in.file.fnum = fnum;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname3+strlen(BASEDIR)+1;
status = smb_raw_setfileinfo(cli1->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
CHECK_VAL(break_info.count, 0);
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname3);
done:
s4 torture: Do a better job of closing open files in RAW-OPLOCK.
2009-12-21 14:38:45 -08:00
smbcli_close(cli1->tree, fnum);
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
/****************************************************
Called from raw-rename - we need oplock handling for
this test so this is why it's in oplock.c, not rename.c
****************************************************/
bool test_nttransrename(struct torture_context *tctx, struct smbcli_state *cli1)
{
const char *fname1 = BASEDIR "\\test_nttransrename_1.dat";
const char *fname2 = BASEDIR "\\test_nttransrename_2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
RAW-RENAME: w2k3 just ignores a NTTRANS-RENAME! metze (This used to be commit cd1b8efc5d8dc1eec03fe1bf1eb58dbded9584eb)
2008-03-12 15:10:57 +01:00
union smb_fileinfo qfi, qpi;
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
union smb_rename rn;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
s4 torture: Change nttransrename to use a normal access mask batch9 already tests for stat-only opens
2009-02-10 16:51:47 -08:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
torture_comment(tctx, "nttrans_rename: open a file with an exclusive oplock (share mode: none)\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
RAW-RENAME: w2k3 just ignores a NTTRANS-RENAME! metze (This used to be commit cd1b8efc5d8dc1eec03fe1bf1eb58dbded9584eb)
2008-03-12 15:10:57 +01:00
torture_comment(tctx, "nttrans_rename: should not trigger a break nor a share mode violation\n");
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_NTTRANS;
rn.nttrans.in.file.fnum = fnum;
rn.nttrans.in.flags = 0;
rn.nttrans.in.new_name = fname2+strlen(BASEDIR)+1;
status = smb_raw_rename(cli1->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
CHECK_VAL(break_info.count, 0);
RAW-RENAME: w2k3 just ignores a NTTRANS-RENAME! metze (This used to be commit cd1b8efc5d8dc1eec03fe1bf1eb58dbded9584eb)
2008-03-12 15:10:57 +01:00
/* w2k3 does nothing, it doesn't rename the file */
torture_comment(tctx, "nttrans_rename: the server should have done nothing\n");
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
RAW-RENAME: w2k3 just ignores a NTTRANS-RENAME! metze (This used to be commit cd1b8efc5d8dc1eec03fe1bf1eb58dbded9584eb)
2008-03-12 15:10:57 +01:00
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname1);
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
RAW-RENAME: w2k3 just ignores a NTTRANS-RENAME! metze (This used to be commit cd1b8efc5d8dc1eec03fe1bf1eb58dbded9584eb)
2008-03-12 15:10:57 +01:00
ZERO_STRUCT(qpi);
qpi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qpi.generic.in.file.path = fname1;
status = smb_raw_pathinfo(cli1->tree, tctx, &qpi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qpi.all_info.out.fname.s, fname1);
ZERO_STRUCT(qpi);
qpi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qpi.generic.in.file.path = fname2;
status = smb_raw_pathinfo(cli1->tree, tctx, &qpi);
CHECK_STATUS(tctx, status, NT_STATUS_OBJECT_NAME_NOT_FOUND);
torture_comment(tctx, "nttrans_rename: after closing the file the file is still not renamed\n");
status = smbcli_close(cli1->tree, fnum);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
ZERO_STRUCT(qpi);
qpi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qpi.generic.in.file.path = fname1;
status = smb_raw_pathinfo(cli1->tree, tctx, &qpi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qpi.all_info.out.fname.s, fname1);
ZERO_STRUCT(qpi);
qpi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qpi.generic.in.file.path = fname2;
status = smb_raw_pathinfo(cli1->tree, tctx, &qpi);
CHECK_STATUS(tctx, status, NT_STATUS_OBJECT_NAME_NOT_FOUND);
torture_comment(tctx, "nttrans_rename: rename with an invalid handle gives NT_STATUS_INVALID_HANDLE\n");
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_NTTRANS;
rn.nttrans.in.file.fnum = fnum+1;
rn.nttrans.in.flags = 0;
rn.nttrans.in.new_name = fname2+strlen(BASEDIR)+1;
status = smb_raw_rename(cli1->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_INVALID_HANDLE);
Added test_nttransrename() to test Metze's new code. Jeremy. (This used to be commit 4556fafede8691c6a12670695ff108e9e59aff98)
2008-03-11 15:27:08 -07:00
done:
smb_raw_exit(cli1->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
static bool test_raw_oplock_batch20(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_batch20_1.dat";
const char *fname2 = BASEDIR "\\test_batch20_2.dat";
const char *fname3 = BASEDIR "\\test_batch20_3.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_fileinfo qfi;
union smb_setfileinfo sfi;
uint16_t fnum=0,fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
smbcli_unlink(cli1->tree, fname3);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH20: open a file with an batch oplock (share mode: all)\n");
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
sfi.generic.in.file.path = fname1;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname2+strlen(BASEDIR)+1;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
CHECK_VAL(break_info.failures, 0);
if (TARGET_IS_WINXP(tctx)) {
/* Win XP breaks to level2. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
} else if (TARGET_IS_W2K3(tctx) || TARGET_IS_W2K8(tctx) ||
TARGET_IS_SAMBA3(tctx) || TARGET_IS_SAMBA4(tctx)) {
/* Win2K3/2k8 incorrectly doesn't break at all. */
CHECK_VAL(break_info.count, 0);
} else {
/* win7/2k8r2 break to none. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
}
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname2);
torture_comment(tctx, "open a file with the new name an batch oplock (share mode: all)\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.fname = fname2;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
if (TARGET_IS_WINXP(tctx)) {
/* XP broke to level2, and doesn't break again. */
CHECK_VAL(break_info.count, 0);
} else if (TARGET_IS_W2K3(tctx) || TARGET_IS_W2K8(tctx) ||
TARGET_IS_SAMBA3(tctx) || TARGET_IS_SAMBA4(tctx)) {
/* Win2K3 incorrectly didn't break before so break now. */
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
} else {
/* win7/2k8r2 broke to none, and doesn't break again. */
CHECK_VAL(break_info.count, 0);
}
ZERO_STRUCT(break_info);
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_RENAME_INFORMATION;
sfi.generic.in.file.fnum = fnum;
sfi.rename_information.in.overwrite = 0;
sfi.rename_information.in.root_fid = 0;
sfi.rename_information.in.new_name = fname3+strlen(BASEDIR)+1;
status = smb_raw_setfileinfo(cli1->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Update raw oplock to use win7 as the baseline for rename oplock break behavior
2009-12-01 17:40:30 -08:00
CHECK_VAL(break_info.count, 0);
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum;
status = smb_raw_fileinfo(cli1->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname3);
ZERO_STRUCT(qfi);
qfi.generic.level = RAW_FILEINFO_ALL_INFORMATION;
qfi.generic.in.file.fnum = fnum2;
status = smb_raw_fileinfo(cli2->tree, tctx, &qfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_STRMATCH(qfi.all_info.out.fname.s, fname3);
done:
s4 torture: Do a better job of closing open files in RAW-OPLOCK.
2009-12-21 14:38:45 -08:00
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH21: a self write with an oplock doesn't break it metze (This used to be commit a5476ee41c140123db160b2e36c8c7084619a738)
2008-03-04 12:15:32 +01:00
static bool test_raw_oplock_batch21(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch21.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
struct smb_echo e;
uint16_t fnum=0;
char c = 0;
ssize_t wr;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH21: a self write with an oplock doesn't break it metze (This used to be commit a5476ee41c140123db160b2e36c8c7084619a738)
2008-03-04 12:15:32 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH21: a self write with an oplock doesn't break it metze (This used to be commit a5476ee41c140123db160b2e36c8c7084619a738)
2008-03-04 12:15:32 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
/*
with a batch oplock we get a break
*/
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH21: open with batch oplock\n");
RAW-OPLOCK: add BATCH21: a self write with an oplock doesn't break it metze (This used to be commit a5476ee41c140123db160b2e36c8c7084619a738)
2008-03-04 12:15:32 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
torture_comment(tctx, "writing should not generate a break\n");
wr = smbcli_write(cli1->tree, fnum, 0, &c, 0, 1);
CHECK_VAL(wr, 1);
CHECK_STATUS(tctx, smbcli_nt_error(cli1->tree), NT_STATUS_OK);
ZERO_STRUCT(e);
e.in.repeat_count = 1;
status = smb_raw_echo(cli1->transport, &e);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH21: a self write with an oplock doesn't break it metze (This used to be commit a5476ee41c140123db160b2e36c8c7084619a738)
2008-03-04 12:15:32 +01:00
CHECK_VAL(break_info.count, 0);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
static bool test_raw_oplock_batch22(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch22.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
s4 torture: Close the third handle in RAW-OPLOCK-BATCH22
2009-12-03 13:16:18 -08:00
uint16_t fnum = 0, fnum2 = 0, fnum3 = 0;
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
struct timeval tv;
int timeout = torture_setting_int(tctx, "oplocktimeout", 30);
int te;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
/*
with a batch oplock we get a break
*/
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH22: open with batch oplock\n");
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ|
NTCREATEX_SHARE_ACCESS_WRITE|
NTCREATEX_SHARE_ACCESS_DELETE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "a 2nd open should not succeed after the oplock "
"break timeout\n");
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
tv = timeval_current();
smbcli_oplock_handler(cli1->transport, oplock_handler_timeout, cli1->tree);
status = smb_raw_open(cli1->tree, tctx, &io);
s4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22 Some servers choose to mark a client as bad if they fail an oplock break request by timing out (win7 is an example). Once the client is marked as bad, future oplock requests will timeout instantly. This causes subsequent runs of this test to fail, so rather than erroring out as a failure, a warning is printed instead. There is also a bug in w2k3 where it was incorrectly returning contending a share mode lock. It worked in XP and has been re-fixed in win7. This can also now be run against samba3.
2009-12-02 18:39:05 -08:00
if (TARGET_IS_W2K3(tctx)) {
/* 2k3 has an issue here. xp/win7 are ok. */
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
} else {
CHECK_STATUS(tctx, status, NT_STATUS_OK);
}
s4 torture: Close the third handle in RAW-OPLOCK-BATCH22
2009-12-03 13:16:18 -08:00
fnum2 = io.ntcreatex.out.file.fnum;
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
te = (int)timeval_elapsed(&tv);
s4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22 Some servers choose to mark a client as bad if they fail an oplock break request by timing out (win7 is an example). Once the client is marked as bad, future oplock requests will timeout instantly. This causes subsequent runs of this test to fail, so rather than erroring out as a failure, a warning is printed instead. There is also a bug in w2k3 where it was incorrectly returning contending a share mode lock. It worked in XP and has been re-fixed in win7. This can also now be run against samba3.
2009-12-02 18:39:05 -08:00
/*
* Some servers detect clients that let oplocks timeout, so this check
* only shows a warning message instead failing the test to eliminate
* failures from repeated runs of the test. This isn't ideal, but
* it's better than not running the test at all.
*/
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
CHECK_RANGE(te, timeout - 1, timeout + 15);
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
ZERO_STRUCT(break_info);
s4/torture: convert printf to torture_comment() Allows "make test" and other harnesses to print cleaner output.
2009-09-08 13:12:26 -07:00
torture_comment(tctx, "a 2nd open should succeed after the oplock "
"release without break\n");
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
tv = timeval_current();
RAW-OPLOCK: rename _ack_to_levelII() -> ack_to_given() Also improve the output. metze (This used to be commit d0b641a9ee36939468beb427bf6b15892342a33d)
2008-03-06 16:27:13 +01:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
Now we behave as Windows does, remove a Samba3 specific test return. Jeremy.
2010-05-13 21:27:24 -07:00
#if 0
/* Samba 3.6.0 and above behave as Windows. */
s4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22 Some servers choose to mark a client as bad if they fail an oplock break request by timing out (win7 is an example). Once the client is marked as bad, future oplock requests will timeout instantly. This causes subsequent runs of this test to fail, so rather than erroring out as a failure, a warning is printed instead. There is also a bug in w2k3 where it was incorrectly returning contending a share mode lock. It worked in XP and has been re-fixed in win7. This can also now be run against samba3.
2009-12-02 18:39:05 -08:00
if (TARGET_IS_SAMBA3(tctx)) {
/* samba3 doesn't grant additional oplocks to bad clients. */
CHECK_VAL(io.ntcreatex.out.oplock_level, NO_OPLOCK_RETURN);
} else {
CHECK_VAL(io.ntcreatex.out.oplock_level,
Now we behave as Windows does, remove a Samba3 specific test return. Jeremy.
2010-05-13 21:27:24 -07:00
LEVEL_II_OPLOCK_RETURN);
s4 torture: Convert an error to a warning in RAW-OPLOCK-BATCH22 Some servers choose to mark a client as bad if they fail an oplock break request by timing out (win7 is an example). Once the client is marked as bad, future oplock requests will timeout instantly. This causes subsequent runs of this test to fail, so rather than erroring out as a failure, a warning is printed instead. There is also a bug in w2k3 where it was incorrectly returning contending a share mode lock. It worked in XP and has been re-fixed in win7. This can also now be run against samba3.
2009-12-02 18:39:05 -08:00
}
Now we behave as Windows does, remove a Samba3 specific test return. Jeremy.
2010-05-13 21:27:24 -07:00
#else
CHECK_VAL(io.ntcreatex.out.oplock_level,
LEVEL_II_OPLOCK_RETURN);
#endif
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
te = (int)timeval_elapsed(&tv);
/* it should come in without delay */
CHECK_RANGE(te+1, 0, timeout);
s4 torture: Close the third handle in RAW-OPLOCK-BATCH22
2009-12-03 13:16:18 -08:00
fnum3 = io.ntcreatex.out.file.fnum;
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
CHECK_VAL(break_info.count, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli1->tree, fnum2);
s4 torture: Close the third handle in RAW-OPLOCK-BATCH22
2009-12-03 13:16:18 -08:00
smbcli_close(cli1->tree, fnum3);
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
static bool test_raw_oplock_batch23(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch23.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0, fnum2=0,fnum3=0;
struct smbcli_state *cli3 = NULL;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
ret = open_connection_no_level2_oplocks(tctx, &cli3);
CHECK_VAL(ret, true);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli2->tree);
smbcli_oplock_handler(cli3->transport, oplock_handler_ack_to_given, cli3->tree);
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH23: a open and ask for a batch oplock\n");
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_READ | SEC_RIGHTS_FILE_WRITE;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ | NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
torture_comment(tctx, "a 2nd open without level2 oplock support should generate a break to level2\n");
status = smb_raw_open(cli3->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum3 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, NO_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
ZERO_STRUCT(break_info);
torture_comment(tctx, "a 3rd open with level2 oplock support should not generate a break\n");
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
CHECK_VAL(break_info.count, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
smbcli_close(cli3->tree, fnum3);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smb_raw_exit(cli3->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
RAW-OPLOCK: add BATCH24 test another case with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 4fb2c9fb1f4e3ee23281ca83f8b91d252cbf53c7)
2008-03-06 15:56:03 +01:00
static bool test_raw_oplock_batch24(struct torture_context *tctx, struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch24.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum2=0,fnum3=0;
struct smbcli_state *cli3 = NULL;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
ret = open_connection_no_level2_oplocks(tctx, &cli3);
CHECK_VAL(ret, true);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli2->tree);
smbcli_oplock_handler(cli3->transport, oplock_handler_ack_to_given, cli3->tree);
RAW-OPLOCK: add BATCH24 test another case with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 4fb2c9fb1f4e3ee23281ca83f8b91d252cbf53c7)
2008-03-06 15:56:03 +01:00
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
RAW-OPLOCK: add BATCH24 test another case with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 4fb2c9fb1f4e3ee23281ca83f8b91d252cbf53c7)
2008-03-06 15:56:03 +01:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
Samba3.2 now passes a lot more tests - remove the S3 specific exemptions. Jeremy. (This used to be commit a70719d579a7eefbfd973267b95a87aaa6b649b0)
2008-03-11 13:39:04 -07:00
torture_comment(tctx, "BATCH24: a open without level support and ask for a batch oplock\n");
RAW-OPLOCK: add BATCH24 test another case with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 4fb2c9fb1f4e3ee23281ca83f8b91d252cbf53c7)
2008-03-06 15:56:03 +01:00
ZERO_STRUCT(break_info);
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_READ | SEC_RIGHTS_FILE_WRITE;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ | NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli3->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum3 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
ZERO_STRUCT(break_info);
torture_comment(tctx, "a 2nd open with level2 oplock support should generate a break to none\n");
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, LEVEL_II_OPLOCK_RETURN);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
RAW-OPLOCK: add BATCH24 test another case with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 4fb2c9fb1f4e3ee23281ca83f8b91d252cbf53c7)
2008-03-06 15:56:03 +01:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.fnum, fnum3);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli3->tree, fnum3);
smbcli_close(cli2->tree, fnum2);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smb_raw_exit(cli3->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
Add the oplock-batch25 test showing that a writeattr does not break an oplock Probably that's already somewhere else, but I did not find it.
2008-10-14 13:29:03 +02:00
static bool test_raw_oplock_batch25(struct torture_context *tctx,
struct smbcli_state *cli1,
struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch25.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_setfileinfo sfi;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
Add the oplock-batch25 test showing that a writeattr does not break an oplock Probably that's already somewhere else, but I did not find it.
2008-10-14 13:29:03 +02:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
torture_comment(tctx, "BATCH25: open a file with an batch oplock "
"(share mode: none)\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
torture_comment(tctx, "setpathinfo attribute info should not trigger "
"a break nor a violation\n");
ZERO_STRUCT(sfi);
sfi.generic.level = RAW_SFILEINFO_SETATTR;
sfi.generic.in.file.path = fname;
sfi.setattr.in.attrib = FILE_ATTRIBUTE_HIDDEN;
sfi.setattr.in.write_time = 0;
status = smb_raw_setpathinfo(cli2->tree, &sfi);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
Add the oplock-batch25 test showing that a writeattr does not break an oplock Probably that's already somewhere else, but I did not find it.
2008-10-14 13:29:03 +02:00
CHECK_VAL(break_info.count, 0);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
s4 torture: Add a new RAW-OPLOCK test: BATCH26 Try a rename with a wide-open share mode on an already open file and the there is still share mode contention. For the reason why see: http://social.msdn.microsoft.com/Forums/en-US/os_fileservices/thread/3ca14dc9-da1f-4786-a8f7-a86e9903db0c Msft's anser: After further review, The reason for server to fail with sharing violation is that the windows server that executes a path-based rename request opens the file for DELETE access, but only with FILE_SHARED_READ as ShareAccess . Therefore, the existing open(frame 76), which has shared read/write/delete , is compatible with the Windows servers access mode (DELETE), but Windows servers open is not compatible with access mode in existing open. Note that it is correct to state that the logic in Windows server could have been written to allow shared read/write/delete in which case it would succeed as you mention. The behavior here is historical based on the existing implementation.
2009-12-03 13:46:11 -08:00
/**
* Similar to batch17/18, but test with open share mode rather than
* share_none.
*/
static bool test_raw_oplock_batch26(struct torture_context *tctx,
struct smbcli_state *cli1, struct smbcli_state *cli2)
{
const char *fname1 = BASEDIR "\\test_batch26_1.dat";
const char *fname2 = BASEDIR "\\test_batch26_2.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
union smb_rename rn;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname1);
smbcli_unlink(cli1->tree, fname2);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given,
cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
io.ntcreatex.in.root_fid.fnum = 0;
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE | NTCREATEX_SHARE_ACCESS_DELETE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname1;
torture_comment(tctx, "BATCH26: open a file with an batch oplock "
"(share mode: none)\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
torture_comment(tctx, "rename should trigger a break\n");
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_RENAME;
rn.rename.in.pattern1 = fname1;
rn.rename.in.pattern2 = fname2;
rn.rename.in.attrib = 0;
torture_comment(tctx, "trying rename while first file open\n");
status = smb_raw_rename(cli2->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
torture_wait_for_oplock_break(tctx);
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
/* Close and reopen with batch again. */
smbcli_close(cli1->tree, fnum);
ZERO_STRUCT(break_info);
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
/* Now try ntrename. */
torture_comment(tctx, "ntrename should trigger a break\n");
ZERO_STRUCT(rn);
rn.generic.level = RAW_RENAME_NTRENAME;
rn.ntrename.in.attrib = 0;
rn.ntrename.in.flags = RENAME_FLAG_RENAME;
rn.ntrename.in.old_name = fname1;
rn.ntrename.in.new_name = fname2;
torture_comment(tctx, "trying rename while first file open\n");
status = smb_raw_rename(cli2->tree, &rn);
CHECK_STATUS(tctx, status, NT_STATUS_SHARING_VIOLATION);
torture_wait_for_oplock_break(tctx);
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
/* Test how oplocks work on streams. */
static bool test_raw_oplock_stream1(struct torture_context *tctx,
struct smbcli_state *cli1,
struct smbcli_state *cli2)
{
NTSTATUS status;
union smb_open io;
const char *fname_base = BASEDIR "\\test_stream1.txt";
const char *stream = "Stream One:$DATA";
const char *fname_stream, *fname_default_stream;
const char *default_stream = "::$DATA";
bool ret = true;
int fnum = -1;
int i;
int stream_fnum = -1;
uint32_t batch_req = NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK | NTCREATEX_FLAGS_EXTENDED;
uint32_t exclusive_req = NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_EXTENDED;
s4/torture: fix RAW-OPLOCK-STREAM1 test after commit a11bb14 We were pulling junk memory for our stream names after the reordering of the struct definition.
2009-09-29 19:03:20 -07:00
s4/torture: don't mix declarations and code
2009-09-10 12:14:53 +10:00
#define NSTREAM_OPLOCK_RESULTS 8
struct {
s4/torture: fix RAW-OPLOCK-STREAM1 test after commit a11bb14 We were pulling junk memory for our stream names after the reordering of the struct definition.
2009-09-29 19:03:20 -07:00
const char **fname;
s4/torture: don't mix declarations and code
2009-09-10 12:14:53 +10:00
bool open_base_file;
uint32_t oplock_req;
uint32_t oplock_granted;
} stream_oplock_results[NSTREAM_OPLOCK_RESULTS] = {
/* Request oplock on stream without the base file open. */
s4/torture: fix RAW-OPLOCK-STREAM1 test after commit a11bb14 We were pulling junk memory for our stream names after the reordering of the struct definition.
2009-09-29 19:03:20 -07:00
{&fname_stream, false, batch_req, NO_OPLOCK_RETURN},
{&fname_default_stream, false, batch_req, NO_OPLOCK_RETURN},
{&fname_stream, false, exclusive_req, EXCLUSIVE_OPLOCK_RETURN},
{&fname_default_stream, false, exclusive_req, EXCLUSIVE_OPLOCK_RETURN},
s4/torture: don't mix declarations and code
2009-09-10 12:14:53 +10:00
/* Request oplock on stream with the base file open. */
s4/torture: fix RAW-OPLOCK-STREAM1 test after commit a11bb14 We were pulling junk memory for our stream names after the reordering of the struct definition.
2009-09-29 19:03:20 -07:00
{&fname_stream, true, batch_req, NO_OPLOCK_RETURN},
{&fname_default_stream, true, batch_req, NO_OPLOCK_RETURN},
{&fname_stream, true, exclusive_req, EXCLUSIVE_OPLOCK_RETURN},
{&fname_default_stream, true, exclusive_req, LEVEL_II_OPLOCK_RETURN},
s4/torture: don't mix declarations and code
2009-09-10 12:14:53 +10:00
};
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
/* Only passes against windows at the moment. */
if (torture_setting_bool(tctx, "samba3", false) ||
torture_setting_bool(tctx, "samba4", false)) {
torture_skip(tctx, "STREAM1 disabled against samba3+4\n");
}
fname_stream = talloc_asprintf(tctx, "%s:%s", fname_base, stream);
fname_default_stream = talloc_asprintf(tctx, "%s%s", fname_base,
default_stream);
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
smbcli_unlink(cli1->tree, fname_base);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given, cli1->tree);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
smbcli_oplock_handler(cli2->transport, oplock_handler_ack_to_given, cli2->tree);
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
/* Setup generic open parameters. */
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
io.ntcreatex.in.access_mask = (SEC_FILE_READ_DATA|SEC_FILE_WRITE_DATA|
SEC_FILE_APPEND_DATA|SEC_STD_READ_CONTROL);
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
/* Create the file with a stream */
io.ntcreatex.in.fname = fname_stream;
io.ntcreatex.in.flags = 0;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_CREATE;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
smbcli_close(cli1->tree, io.ntcreatex.out.file.fnum);
/* Change the disposition to open now that the file has been created. */
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN;
s4/torture: fix RAW-OPLOCK-STREAM1 test after commit a11bb14 We were pulling junk memory for our stream names after the reordering of the struct definition.
2009-09-29 19:03:20 -07:00
/* Try some permutations of taking oplocks on streams. */
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
for (i = 0; i < NSTREAM_OPLOCK_RESULTS; i++) {
s4/torture: fix RAW-OPLOCK-STREAM1 test after commit a11bb14 We were pulling junk memory for our stream names after the reordering of the struct definition.
2009-09-29 19:03:20 -07:00
const char *fname = *stream_oplock_results[i].fname;
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
bool open_base_file = stream_oplock_results[i].open_base_file;
uint32_t oplock_req = stream_oplock_results[i].oplock_req;
uint32_t oplock_granted =
stream_oplock_results[i].oplock_granted;
int base_fnum = -1;
if (open_base_file) {
torture_comment(tctx, "Opening base file: %s with "
s4/torture: fix typo in test comment
2009-10-01 11:35:17 -07:00
"%d\n", fname_base, batch_req);
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
io.ntcreatex.in.fname = fname_base;
io.ntcreatex.in.flags = batch_req;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(io.ntcreatex.out.oplock_level,
BATCH_OPLOCK_RETURN);
base_fnum = io.ntcreatex.out.file.fnum;
}
torture_comment(tctx, "%d: Opening stream: %s with %d\n", i,
fname, oplock_req);
io.ntcreatex.in.fname = fname;
io.ntcreatex.in.flags = oplock_req;
/* Do the open with the desired oplock on the stream. */
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(io.ntcreatex.out.oplock_level, oplock_granted);
smbcli_close(cli1->tree, io.ntcreatex.out.file.fnum);
/* Cleanup the base file if it was opened. */
if (base_fnum != -1) {
smbcli_close(cli2->tree, base_fnum);
}
}
/* Open the stream with an exclusive oplock. */
torture_comment(tctx, "Opening stream: %s with %d\n",
fname_stream, exclusive_req);
io.ntcreatex.in.fname = fname_stream;
io.ntcreatex.in.flags = exclusive_req;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(io.ntcreatex.out.oplock_level, EXCLUSIVE_OPLOCK_RETURN);
stream_fnum = io.ntcreatex.out.file.fnum;
/* Open the base file and see if it contends. */
ZERO_STRUCT(break_info);
torture_comment(tctx, "Opening base file: %s with "
"%d\n", fname_base, batch_req);
io.ntcreatex.in.fname = fname_base;
io.ntcreatex.in.flags = batch_req;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(io.ntcreatex.out.oplock_level,
BATCH_OPLOCK_RETURN);
smbcli_close(cli2->tree, io.ntcreatex.out.file.fnum);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.failures, 0);
/* Open the stream again to see if it contends. */
ZERO_STRUCT(break_info);
torture_comment(tctx, "Opening stream again: %s with "
"%d\n", fname_base, batch_req);
io.ntcreatex.in.fname = fname_stream;
io.ntcreatex.in.flags = exclusive_req;
status = smb_raw_open(cli2->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(io.ntcreatex.out.oplock_level,
LEVEL_II_OPLOCK_RETURN);
smbcli_close(cli2->tree, io.ntcreatex.out.file.fnum);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
/* Close the stream. */
if (stream_fnum != -1) {
smbcli_close(cli1->tree, stream_fnum);
}
done:
smbcli_close(cli1->tree, fnum);
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
static bool test_raw_oplock_doc(struct torture_context *tctx,
struct smbcli_state *cli)
{
const char *fname = BASEDIR "\\test_oplock_doc.dat";
NTSTATUS status;
bool ret = true;
union smb_open io;
uint16_t fnum=0;
if (!torture_setup_dir(cli, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli->tree, fname);
smbcli_oplock_handler(cli->transport, oplock_handler_ack_to_given,
cli->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = NTCREATEX_OPTIONS_DELETE_ON_CLOSE;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
torture_comment(tctx, "open a delete-on-close file with a batch "
"oplock\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
smbcli_close(cli->tree, fnum);
done:
smb_raw_exit(cli->session);
smbcli_deltree(cli->tree, BASEDIR);
return ret;
}
/* Open a file with a batch oplock, then open it again from a second client
* requesting no oplock. Having two open file handles should break our own
* oplock during BRL acquisition.
*/
static bool test_raw_oplock_brl1(struct torture_context *tctx,
struct smbcli_state *cli1,
struct smbcli_state *cli2)
{
const char *fname = BASEDIR "\\test_batch_brl.dat";
/*int fname, f;*/
bool ret = true;
uint8_t buf[1000];
bool correct = true;
union smb_open io;
NTSTATUS status;
uint16_t fnum=0;
uint16_t fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given,
cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_READ |
SEC_RIGHTS_FILE_WRITE;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
/*
with a batch oplock we get a break
*/
torture_comment(tctx, "open with batch oplock\n");
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
/* create a file with bogus data */
memset(buf, 0, sizeof(buf));
if (smbcli_write(cli1->tree, fnum, 0, buf, 0, sizeof(buf)) !=
sizeof(buf))
{
torture_comment(tctx, "Failed to create file\n");
correct = false;
goto done;
}
torture_comment(tctx, "a 2nd open should give a break\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED;
status = smb_raw_open(cli2->tree, tctx, &io);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.fnum, fnum);
ZERO_STRUCT(break_info);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_comment(tctx, "a self BRL acquisition should break to none\n");
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
status = smbcli_lock(cli1->tree, fnum, 0, 4, 0, WRITE_LOCK);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.failures, 0);
/* expect no oplock break */
ZERO_STRUCT(break_info);
status = smbcli_lock(cli1->tree, fnum, 2, 4, 0, WRITE_LOCK);
CHECK_STATUS(tctx, status, NT_STATUS_LOCK_NOT_GRANTED);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.level, 0);
CHECK_VAL(break_info.fnum, 0);
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli2->tree, fnum2);
done:
smb_raw_exit(cli1->session);
smb_raw_exit(cli2->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
/* Open a file with a batch oplock on one client and then acquire a brl.
* We should not contend our own oplock.
*/
static bool test_raw_oplock_brl2(struct torture_context *tctx, struct smbcli_state *cli1)
{
const char *fname = BASEDIR "\\test_batch_brl.dat";
/*int fname, f;*/
bool ret = true;
uint8_t buf[1000];
bool correct = true;
union smb_open io;
NTSTATUS status;
uint16_t fnum=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given,
cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_READ |
SEC_RIGHTS_FILE_WRITE;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
/*
with a batch oplock we get a break
*/
torture_comment(tctx, "open with batch oplock\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
/* create a file with bogus data */
memset(buf, 0, sizeof(buf));
if (smbcli_write(cli1->tree, fnum, 0, buf, 0, sizeof(buf)) !=
sizeof(buf))
{
torture_comment(tctx, "Failed to create file\n");
correct = false;
goto done;
}
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_comment(tctx, "a self BRL acquisition should not break to "
"none\n");
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
status = smbcli_lock(cli1->tree, fnum, 0, 4, 0, WRITE_LOCK);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
status = smbcli_lock(cli1->tree, fnum, 2, 4, 0, WRITE_LOCK);
CHECK_STATUS(tctx, status, NT_STATUS_LOCK_NOT_GRANTED);
/* With one file handle open a BRL should not contend our oplock.
* Thus, no oplock break will be received and the entire break_info
* struct will be 0 */
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
CHECK_VAL(break_info.fnum, 0);
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.level, 0);
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
done:
smb_raw_exit(cli1->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
/* Open a file with a batch oplock twice from one client and then acquire a
* brl. BRL acquisition should break our own oplock.
*/
static bool test_raw_oplock_brl3(struct torture_context *tctx,
struct smbcli_state *cli1)
{
const char *fname = BASEDIR "\\test_batch_brl.dat";
bool ret = true;
uint8_t buf[1000];
bool correct = true;
union smb_open io;
NTSTATUS status;
uint16_t fnum=0;
uint16_t fnum2=0;
if (!torture_setup_dir(cli1, BASEDIR)) {
return false;
}
/* cleanup */
smbcli_unlink(cli1->tree, fname);
smbcli_oplock_handler(cli1->transport, oplock_handler_ack_to_given,
cli1->tree);
/*
base ntcreatex parms
*/
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_READ |
SEC_RIGHTS_FILE_WRITE;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_READ |
NTCREATEX_SHARE_ACCESS_WRITE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = fname;
/*
with a batch oplock we get a break
*/
torture_comment(tctx, "open with batch oplock\n");
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
status = smb_raw_open(cli1->tree, tctx, &io);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
fnum = io.ntcreatex.out.file.fnum;
CHECK_VAL(io.ntcreatex.out.oplock_level, BATCH_OPLOCK_RETURN);
/* create a file with bogus data */
memset(buf, 0, sizeof(buf));
if (smbcli_write(cli1->tree, fnum, 0, buf, 0, sizeof(buf)) !=
sizeof(buf))
{
torture_comment(tctx, "Failed to create file\n");
correct = false;
goto done;
}
torture_comment(tctx, "a 2nd open should give a break\n");
ZERO_STRUCT(break_info);
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED;
status = smb_raw_open(cli1->tree, tctx, &io);
fnum2 = io.ntcreatex.out.file.fnum;
CHECK_STATUS(tctx, status, NT_STATUS_OK);
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_LEVEL_II);
CHECK_VAL(break_info.failures, 0);
CHECK_VAL(break_info.fnum, fnum);
ZERO_STRUCT(break_info);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_comment(tctx, "a self BRL acquisition should break to none\n");
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
status = smbcli_lock(cli1->tree, fnum, 0, 4, 0, WRITE_LOCK);
CHECK_STATUS(tctx, status, NT_STATUS_OK);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
CHECK_VAL(break_info.count, 1);
CHECK_VAL(break_info.level, OPLOCK_BREAK_TO_NONE);
CHECK_VAL(break_info.fnum, fnum);
CHECK_VAL(break_info.failures, 0);
/* expect no oplock break */
ZERO_STRUCT(break_info);
status = smbcli_lock(cli1->tree, fnum, 2, 4, 0, WRITE_LOCK);
CHECK_STATUS(tctx, status, NT_STATUS_LOCK_NOT_GRANTED);
s4/torture: Allow receiving of oplock break requests in any order Previously, the oplock torture tests, being single threaded, required the server to return oplock break requests, and other SMB packets in a specific order for us to verify "correctness". Of course, in several cases the protocol allows the break packets, especially breaks to levelII to come back in any order. With tevent we're now able to wait for oplock breaks in the middle of a torture test. I've added a helper to do this, and modified all oplock tests to allow returning of oplock breaks in any order.
2009-09-25 16:44:51 -07:00
torture_wait_for_oplock_break(tctx);
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
CHECK_VAL(break_info.count, 0);
CHECK_VAL(break_info.level, 0);
CHECK_VAL(break_info.fnum, 0);
CHECK_VAL(break_info.failures, 0);
smbcli_close(cli1->tree, fnum);
smbcli_close(cli1->tree, fnum2);
done:
smb_raw_exit(cli1->session);
smbcli_deltree(cli1->tree, BASEDIR);
return ret;
}
/*
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
basic testing of oplocks
*/
r24735: Use torture API in more places. (This used to be commit 1319d88c099496be29dd9214fa2492c81e848369)
2007-08-28 12:54:27 +00:00
struct torture_suite *torture_raw_oplock(TALLOC_CTX *mem_ctx)
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
{
r24735: Use torture API in more places. (This used to be commit 1319d88c099496be29dd9214fa2492c81e848369)
2007-08-28 12:54:27 +00:00
struct torture_suite *suite = torture_suite_create(mem_ctx, "OPLOCK");
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
RAW-OPLOCK: rename test (NORMAL => EXCLUSIVE1) metze (This used to be commit 546536dde07d82dbba6724df95cdd1ba50caf8c5)
2008-02-26 14:58:00 +01:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE1", test_raw_oplock_exclusive1);
RAW-OPLOCK: add EXCLUSIVE2 Exclusive oplocks break to LEVEL2 metze (This used to be commit 1cddadb052106394046f03d65297ea68232e7492)
2008-02-26 15:51:17 +01:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE2", test_raw_oplock_exclusive2);
RAW-OPLOCK: add EXCLUSIVE3 This tests that a setpathinfo EOF breaks an exclusive oplock to none. metze (This used to be commit fd5b4b3d4ec52572619dba065d689d58bec71ae0)
2008-02-26 16:16:31 +01:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE3", test_raw_oplock_exclusive3);
RAW-OPLOCK: add EXCLUSIVE4 a attribute only open doesn't break an exclusive oplock metze (This used to be commit b22587a5db4a1e2d5fa1a944896330f1a86eef11)
2008-02-27 08:51:25 +01:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE4", test_raw_oplock_exclusive4);
RAW-OPLOCK: add EXCLUSIVE5 to test attribute only with OVERWRITE_IF metze (This used to be commit 06547036e6554dc94f683263b54d528eac227ea8)
2008-02-27 10:02:52 +01:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE5", test_raw_oplock_exclusive5);
RAW-OPLOCK: add EXCLISIVE6 test a rename on a file with an oplock metze (This used to be commit 3886a5dea2fa26eeed8bc1d20c44ef012ea1f746)
2008-02-27 10:40:19 +01:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE6", test_raw_oplock_exclusive6);
s4 torture: Add RAW-OPLOCK-EXCLUSIVE7 which is similar to BATCH19
2009-12-02 11:39:50 -08:00
torture_suite_add_2smb_test(suite, "EXCLUSIVE7", test_raw_oplock_exclusive7);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
torture_suite_add_2smb_test(suite, "BATCH1", test_raw_oplock_batch1);
torture_suite_add_2smb_test(suite, "BATCH2", test_raw_oplock_batch2);
torture_suite_add_2smb_test(suite, "BATCH3", test_raw_oplock_batch3);
torture_suite_add_2smb_test(suite, "BATCH4", test_raw_oplock_batch4);
torture_suite_add_2smb_test(suite, "BATCH5", test_raw_oplock_batch5);
torture_suite_add_2smb_test(suite, "BATCH6", test_raw_oplock_batch6);
torture_suite_add_2smb_test(suite, "BATCH7", test_raw_oplock_batch7);
torture_suite_add_2smb_test(suite, "BATCH8", test_raw_oplock_batch8);
torture_suite_add_2smb_test(suite, "BATCH9", test_raw_oplock_batch9);
torture_suite_add_2smb_test(suite, "BATCH10", test_raw_oplock_batch10);
torture_suite_add_2smb_test(suite, "BATCH11", test_raw_oplock_batch11);
torture_suite_add_2smb_test(suite, "BATCH12", test_raw_oplock_batch12);
torture_suite_add_2smb_test(suite, "BATCH13", test_raw_oplock_batch13);
torture_suite_add_2smb_test(suite, "BATCH14", test_raw_oplock_batch14);
RAW-OPLOCK: BATCH15: make sure a qpathinfo doesn't break batch oplocks metze (This used to be commit 4d17d44d3f418216a40c80f290f97a4dc29147b0)
2008-02-26 14:52:50 +01:00
torture_suite_add_2smb_test(suite, "BATCH15", test_raw_oplock_batch15);
RAW-OPLOCK: add BATCH16 and tests attribute only with OVERWRITE_IF metze (This used to be commit 2453628623795c34ece18a285e6c2de40c4518de)
2008-02-27 09:41:44 +01:00
torture_suite_add_2smb_test(suite, "BATCH16", test_raw_oplock_batch16);
RAW-OPLOCK: add BATCH17 test rename on a file with oplock metze (This used to be commit 32f138d4b313a6c5e8ef8e15945df02e3b4db864)
2008-02-27 10:41:47 +01:00
torture_suite_add_2smb_test(suite, "BATCH17", test_raw_oplock_batch17);
RAW-OPLOCK: add BATCH18 test a ntrename with FLAG_RENAME metze (This used to be commit 22c35d9ee7858c0ff58d5d0b4a9bf69fe0b8653f)
2008-02-27 11:07:53 +01:00
torture_suite_add_2smb_test(suite, "BATCH18", test_raw_oplock_batch18);
RAW-OPLOCK: add BATCH19 test renaming via setpathinfo This is really strange as, it seems to complete ignore oplocks and share modes and doesn't match the behavior of RENAME and NTRENAME. metze (This used to be commit 7bac95db5edcc7432aea067e0a9ba9230c7dc1b0)
2008-02-27 15:43:04 +01:00
torture_suite_add_2smb_test(suite, "BATCH19", test_raw_oplock_batch19);
RAW-OPLOCK: add BATCH20 test renaming via setpathinfo() and setfileinfo() metze (This used to be commit b66c246c5db78db897bcf21a6ea3c777d65cfc6a)
2008-02-27 15:45:08 +01:00
torture_suite_add_2smb_test(suite, "BATCH20", test_raw_oplock_batch20);
RAW-OPLOCK: add BATCH21: a self write with an oplock doesn't break it metze (This used to be commit a5476ee41c140123db160b2e36c8c7084619a738)
2008-03-04 12:15:32 +01:00
torture_suite_add_2smb_test(suite, "BATCH21", test_raw_oplock_batch21);
RAW-OPLOCK: add BATCH22 and test the behavior of oplock break timeouts metze (This used to be commit c459885898c9912df1ae5afff4fef2ff809dc15e)
2008-03-04 14:08:32 +01:00
torture_suite_add_2smb_test(suite, "BATCH22", test_raw_oplock_batch22);
RAW-OPLOCK: add BATCH23 and test with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 2192d6d2bda2afd2ba1ed0cb68bdfe590b9d50af)
2008-03-06 15:54:07 +01:00
torture_suite_add_2smb_test(suite, "BATCH23", test_raw_oplock_batch23);
RAW-OPLOCK: add BATCH24 test another case with a connection with no CAP_LEVEL_II_OPLOCKS metze (This used to be commit 4fb2c9fb1f4e3ee23281ca83f8b91d252cbf53c7)
2008-03-06 15:56:03 +01:00
torture_suite_add_2smb_test(suite, "BATCH24", test_raw_oplock_batch24);
Add the oplock-batch25 test showing that a writeattr does not break an oplock Probably that's already somewhere else, but I did not find it.
2008-10-14 13:29:03 +02:00
torture_suite_add_2smb_test(suite, "BATCH25", test_raw_oplock_batch25);
s4 torture: Add a new RAW-OPLOCK test: BATCH26 Try a rename with a wide-open share mode on an already open file and the there is still share mode contention. For the reason why see: http://social.msdn.microsoft.com/Forums/en-US/os_fileservices/thread/3ca14dc9-da1f-4786-a8f7-a86e9903db0c Msft's anser: After further review, The reason for server to fail with sharing violation is that the windows server that executes a path-based rename request opens the file for DELETE access, but only with FILE_SHARED_READ as ShareAccess . Therefore, the existing open(frame 76), which has shared read/write/delete , is compatible with the Windows servers access mode (DELETE), but Windows servers open is not compatible with access mode in existing open. Note that it is correct to state that the logic in Windows server could have been written to allow shared read/write/delete in which case it would succeed as you mention. The behavior here is historical based on the existing implementation.
2009-12-03 13:46:11 -08:00
torture_suite_add_2smb_test(suite, "BATCH26", test_raw_oplock_batch26);
s4 torture: Add new streams oplock test
2009-04-01 18:44:15 -07:00
torture_suite_add_2smb_test(suite, "STREAM1", test_raw_oplock_stream1);
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
torture_suite_add_1smb_test(suite, "DOC1", test_raw_oplock_doc);
torture_suite_add_2smb_test(suite, "BRL1", test_raw_oplock_brl1);
torture_suite_add_1smb_test(suite, "BRL2", test_raw_oplock_brl2);
torture_suite_add_1smb_test(suite, "BRL3", test_raw_oplock_brl3);
r23144: splitup RAW-OPLOCK into small units metze (This used to be commit 0cfe152cb143165b908ef023a06c4107efbdecb9)
2007-05-25 13:03:33 +00:00
return suite;
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
}
s4/torture: add new SMB oplock tests * test if oplocks are granted when requesting delete-on-close * test how oplocks are broken by byte-range-lock requests
2009-09-08 13:19:44 -07:00
/*
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
stress testing of oplocks
*/
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
bool torture_bench_oplock(struct torture_context *torture)
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
{
struct smbcli_state **cli;
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
bool ret = true;
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
TALLOC_CTX *mem_ctx = talloc_new(torture);
r24670: Use torture functions for settings. (This used to be commit a6906676309f3b1bec1a69427e7ffd0a768061b8)
2007-08-26 18:24:12 +00:00
int torture_nprocs = torture_setting_int(torture, "nprocs", 4);
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
int i, count=0;
r19392: Use torture_setting_* rather than lp_parm_* where possible. (This used to be commit b28860978fe29c5b10abfb8c59d7182864e21dd6)
2006-10-18 14:23:19 +00:00
int timelimit = torture_setting_int(torture, "timelimit", 10);
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
union smb_open io;
struct timeval tv;
cli = talloc_array(mem_ctx, struct smbcli_state *, torture_nprocs);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(torture, "Opening %d connections\n", torture_nprocs);
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
for (i=0;i<torture_nprocs;i++) {
Avoid event_find_context when a event context is already available. (This used to be commit 4ca264679ecfd938c538a93f4efff1bfa23c3744)
2008-04-16 23:06:27 +02:00
if (!torture_open_connection_ev(&cli[i], i, torture, torture->ev)) {
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return false;
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
}
talloc_steal(mem_ctx, cli[i]);
smbcli_oplock_handler(cli[i]->transport, oplock_handler_close,
cli[i]->tree);
}
if (!torture_setup_dir(cli[0], BASEDIR)) {
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
ret = false;
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
goto done;
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
}
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
io.ntcreatex.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = NTCREATEX_SHARE_ACCESS_NONE;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = BASEDIR "\\test.dat";
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
tv = timeval_current();
r15015: add an explanation for this test (This used to be commit 9fdeec506fe5005601c0589574d2b8ae3c6166e5)
2006-04-10 05:12:49 +00:00
/*
we open the same file with SHARE_ACCESS_NONE from all the
connections in a round robin fashion. Each open causes an
oplock break on the previous connection, which is answered
by the oplock_handler_close() to close the file.
This measures how fast we can pass on oplocks, and stresses
the oplock handling code
*/
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(torture, "Running for %d seconds\n", timelimit);
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
while (timeval_elapsed(&tv) < timelimit) {
for (i=0;i<torture_nprocs;i++) {
NTSTATUS status;
status = smb_raw_open(cli[i]->tree, mem_ctx, &io);
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
CHECK_STATUS(torture, status, NT_STATUS_OK);
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
count++;
}
r22579: disable progress printing in the build-farm metze (This used to be commit 93089ad5e8b6e20c4fa92bf13b0137765aeac689)
2007-04-29 21:37:29 +00:00
if (torture_setting_bool(torture, "progress", true)) {
torture_comment(torture, "%.2f ops/second\r", count/timeval_elapsed(&tv));
}
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
}
r21707: Finally merge my (long-living) perlselftest branch. This changes the main selftest code to be in perl rather than in shell script. The selftest script is now no longer a black box but a regular executable that takes --help. This adds the following features: * "make test TESTS=foo" will run only the tests that match the regex "foo" * ability to deal with expected failures. the suite will not warn about tests that fail and are known to fail, but will warn about other failing tests and tests that are succeeding tests but incorrectly marked as failing. * ability to print a summary with all failures at the end of the run It also opens up the way to the following features, which I hope to implement later: * "environments", for example having a complete domains with DCs and domain members in a testenvironment * only set up smbd if necessary (not when running LOCAL tests, for example) * different mktestsetup scripts per target. except for the mktestsetup script, we can use the same infrastructure for samba 3 or windows. (This used to be commit 38f867880beb40c691e9713f854426031310629c)
2007-03-05 21:28:55 +00:00
torture_comment(torture, "%.2f ops/second\n", count/timeval_elapsed(&tv));
r15014: added a simple batch oplock benchmark Jeremy, Samba3 currently fails this test with a 2.6.16-rc6 kernel and kernel oplocks enabled. With kernel oplock off it passes, though its a bit slow (This used to be commit 99df20246049c2645b04d5790b9e41c607872dc7)
2006-04-10 04:51:52 +00:00
smb_raw_exit(cli[torture_nprocs-1]->session);
done:
smb_raw_exit(cli[0]->session);
smbcli_deltree(cli[0]->tree, BASEDIR);
r5037: got rid of all of the TALLOC_DEPRECATED stuff. My apologies for the large commit. I thought this was worthwhile to get done for consistency. (This used to be commit ec32b22ed5ec224f6324f5e069d15e92e38e15c0)
2005-01-27 07:08:20 +00:00
talloc_free(mem_ctx);
first public release of samba4 code (This used to be commit b0510b5428b3461aeb9bbe3cc95f62fc73e2b97f)
2003-08-13 01:53:07 +00:00
return ret;
}
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
static struct hold_oplock_info {
const char *fname;
bool close_on_break;
uint32_t share_access;
uint16_t fnum;
} hold_info[] = {
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
{ BASEDIR "\\notshared_close", true,
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
NTCREATEX_SHARE_ACCESS_NONE, },
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
{ BASEDIR "\\notshared_noclose", false,
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
NTCREATEX_SHARE_ACCESS_NONE, },
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
{ BASEDIR "\\shared_close", true,
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
NTCREATEX_SHARE_ACCESS_READ|NTCREATEX_SHARE_ACCESS_WRITE|NTCREATEX_SHARE_ACCESS_DELETE, },
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
{ BASEDIR "\\shared_noclose", false,
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
NTCREATEX_SHARE_ACCESS_READ|NTCREATEX_SHARE_ACCESS_WRITE|NTCREATEX_SHARE_ACCESS_DELETE, },
};
r25000: Fix some more C++ compatibility warnings. (This used to be commit 08bb1ef643ab906f1645cf6f32763dc73b1884e4)
2007-09-07 15:08:14 +00:00
static bool oplock_handler_hold(struct smbcli_transport *transport,
uint16_t tid, uint16_t fnum, uint8_t level,
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
void *private_data)
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
{
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
struct smbcli_tree *tree = (struct smbcli_tree *)private_data;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
struct hold_oplock_info *info;
int i;
for (i=0;i<ARRAY_SIZE(hold_info);i++) {
if (hold_info[i].fnum == fnum) break;
}
if (i == ARRAY_SIZE(hold_info)) {
printf("oplock break for unknown fnum %u\n", fnum);
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return false;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
info = &hold_info[i];
if (info->close_on_break) {
printf("oplock break on %s - closing\n",
info->fname);
s4:torture: s/private/private_data metze
2009-02-02 08:44:44 +01:00
oplock_handler_close(transport, tid, fnum, level, private_data);
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return true;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
printf("oplock break on %s - acking break\n", info->fname);
return smbcli_oplock_ack(tree, fnum, OPLOCK_BREAK_TO_NONE);
}
/*
used for manual testing of oplocks - especially interaction with
other filesystems (such as NFS and local access)
*/
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
bool torture_hold_oplock(struct torture_context *torture,
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
struct smbcli_state *cli)
{
s4:lib/tevent: rename structs list="" list="$list event_context:tevent_context" list="$list fd_event:tevent_fd" list="$list timed_event:tevent_timer" for s in $list; do o=`echo $s | cut -d ':' -f1` n=`echo $s | cut -d ':' -f2` r=`git grep "struct $o" |cut -d ':' -f1 |sort -u` files=`echo "$r" | grep -v source3 | grep -v nsswitch | grep -v packaging4` for f in $files; do cat $f | sed -e "s/struct $o/struct $n/g" > $f.tmp mv $f.tmp $f done done metze
2008-12-29 20:24:57 +01:00
struct tevent_context *ev =
(struct tevent_context *)cli->transport->socket->event.ctx;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
int i;
printf("Setting up open files with oplocks in %s\n", BASEDIR);
if (!torture_setup_dir(cli, BASEDIR)) {
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return false;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
smbcli_oplock_handler(cli->transport, oplock_handler_hold, cli->tree);
/* setup the files */
for (i=0;i<ARRAY_SIZE(hold_info);i++) {
union smb_open io;
NTSTATUS status;
r25689: test with a non-zero size file (This used to be commit 2a6713aa239d07f6898f507ddb3631504c379792)
2007-10-18 01:37:46 +02:00
char c = 1;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
io.generic.level = RAW_OPEN_NTCREATEX;
s4-smb: declare root_fid as a file handle In order to implement root_fid in the s4 SMB server we need to declare it as a handle type, just as for other fnum values in SMB. This required some extensive (but simple) changes in many bits of code.
2009-10-15 18:26:19 +11:00
io.ntcreatex.in.root_fid.fnum = 0;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
io.ntcreatex.in.access_mask = SEC_RIGHTS_FILE_ALL;
io.ntcreatex.in.alloc_size = 0;
io.ntcreatex.in.file_attr = FILE_ATTRIBUTE_NORMAL;
io.ntcreatex.in.share_access = hold_info[i].share_access;
io.ntcreatex.in.open_disposition = NTCREATEX_DISP_OPEN_IF;
io.ntcreatex.in.create_options = 0;
io.ntcreatex.in.impersonation = NTCREATEX_IMPERSONATION_ANONYMOUS;
io.ntcreatex.in.security_flags = 0;
io.ntcreatex.in.fname = hold_info[i].fname;
io.ntcreatex.in.flags = NTCREATEX_FLAGS_EXTENDED |
NTCREATEX_FLAGS_REQUEST_OPLOCK |
NTCREATEX_FLAGS_REQUEST_BATCH_OPLOCK;
printf("opening %s\n", hold_info[i].fname);
status = smb_raw_open(cli->tree, cli, &io);
if (!NT_STATUS_IS_OK(status)) {
printf("Failed to open %s - %s\n",
hold_info[i].fname, nt_errstr(status));
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return false;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
if (io.ntcreatex.out.oplock_level != BATCH_OPLOCK_RETURN) {
printf("Oplock not granted for %s - expected %d but got %d\n",
hold_info[i].fname, BATCH_OPLOCK_RETURN,
io.ntcreatex.out.oplock_level);
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return false;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
hold_info[i].fnum = io.ntcreatex.out.file.fnum;
r25689: test with a non-zero size file (This used to be commit 2a6713aa239d07f6898f507ddb3631504c379792)
2007-10-18 01:37:46 +02:00
/* make the file non-zero size */
if (smbcli_write(cli->tree, hold_info[i].fnum, 0, &c, 0, 1) != 1) {
printf("Failed to write to file\n");
return false;
}
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
printf("Waiting for oplock events\n");
event_loop_wait(ev);
r25554: Convert last instances of BOOL, True and False to the standard types. (This used to be commit 566aa14139510788548a874e9213d91317f83ca9)
2007-10-06 22:28:14 +00:00
return true;
r24765: added a RAW-HOLD-OPLOCK test. This is a manual test, meant to be used to test the interaction of oplocks with local filesystem or NFS access. You start it, then manually access the 4 files via another mechanism (such as a shell) and see what oplock breaks you get (This used to be commit 4c3e4564260e55434a54a013051db609c92cabe8)
2007-08-29 04:33:26 +00:00
}
Copy Permalink