ed1c3e1129
TESTS: Fix all 0block tests
...
Remove the test for accessing 0blocks (transfer length==0) at the end of the device since they were invalid.
SBC says that :
If the LBA plus the transfer length exceeds the capacity
of the medium, then the device server shall terminate the
command with CHECK CONDITION status with the sense key
set to ILLEGAL REQUEST and the additional sense code
set to LOGICAL BLOCK ADDRESS OUT OF RANGE.
This means that if TL==0 then a LBA that is the first LBA beyond the end of
device is actually valid.
Improve output for error conditions and harminize all tests
2012-09-10 19:08:57 -07:00
9d15c1ea40
TESTS: 0100: change command names to upper-case in logging
2012-09-10 17:49:17 -07:00
ca3ac9cafd
TESTS: Improve error reporting for *beyondeol* tests
...
Beyondeol tests: Add missing checks for optional opcodes
to SKIP tests where the target does not implement the opcode.
Improve the error reporting and make sure we always show the Sense we
received from the target when we get an unexpected error.
Check the size of the LUN and skip the test for luns that are too big.
For example skip testing READ10 on LUNs > 2^31 blocks etc.
2012-09-10 08:38:21 -07:00
ef5a1e20a4
Merge pull request #16 from bonzini/iscsi-test
...
Iscsi test
2012-09-09 16:34:49 -07:00
d155231341
test-tool: 0171 is using different 0-block semantics than other tests
...
0171 is expecting a 0-block unmap at end-of-volume (i.e. start block
one beyond what READ CAPACITY returns) to pass, while every other 0blocks
test is expecting that address to be invalid.
2012-09-06 16:58:21 +02:00
a60c23887c
test-tool: return SKIPPED exit code from 0191 when LBPPB = 1
...
Missed on my previous round.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-09-06 16:58:21 +02:00
3af45a0677
TEST: add test for prefetching 0 blocks from a device
2012-09-05 20:17:50 -07:00
8f11b127fe
TEST: Add test for PREFETCH10/16 beyond eol
2012-09-05 19:42:14 -07:00
ca6875e69d
TESTS: Add tests for the flags for PREFETCH10/16
2012-09-05 19:17:16 -07:00
e78c5ae8a4
Merge pull request #14 from bonzini/iscsi-test
...
test-tool: 0214 was reading twice from -1
2012-09-04 19:15:18 -07:00
0abacf8fa4
test-tool: 0214 was reading twice from -1
...
It should read once from -1, once from 2^31.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-09-04 18:34:38 +02:00
61488fd381
New version 1.0.6
2012-09-03 10:17:43 -07:00
d6bbe8fe44
TESTS: PREFETCH10/16, add test for prefetch-lengt==0
2012-09-03 10:00:53 -07:00
81e935ffb0
TESTS: PREFETCH10/16 are optional. Skip test if not supported on target
2012-09-03 09:52:43 -07:00
116af01156
TYPO: Change periperal_device_type to just device_type
...
but leave the old name as a define for backward compatibility
2012-09-03 09:45:59 -07:00
77fc2497f7
TESTS: Change all tests to use 'block_size' instead of hardcoded 512
...
This should allow the testst to work corectly on block devices with 4k blocksize too.
2012-09-03 08:58:56 -07:00
5639b92d95
Improve documentation for when tasks may be cancelled
2012-08-31 16:18:59 -07:00
6ef4bbd3e1
TEST: add back extra test for invalid buffer offsets
2012-08-31 16:15:53 -07:00
6a24b283f8
TEST: If we timeout a test, drop all wait/out pdus from the list
...
we dont want the callbacks anymore.
2012-08-31 16:12:21 -07:00
c526b9e954
TESTS: remove compiler warnings
2012-08-31 15:26:23 -07:00
09573068ad
TESTS: Skipped tests should return -2
2012-08-31 14:19:11 -07:00
4df1525415
TESTS: Additional checks for missing opcodes and skipping tests
2012-08-31 11:42:44 -07:00
f3aacb3a5d
TESTS: skip COMPARE-AND-WRITE tests on targets missing this opcode
2012-08-31 11:18:55 -07:00
98d18d5c42
TESTS: Add remaining checks for *verify* opcode not implemented skip test
2012-08-30 20:30:28 -07:00
8bc3d51cd4
TESTS: improve verify tests to detect/abort on opcode not supported furthter
2012-08-30 20:22:16 -07:00
b4594c1ff3
TESTS: Add check and skip tests for targets that do not implement VERIFY*
2012-08-30 20:09:53 -07:00
dcc41d5c43
Merge branch 'master' of github.com:sahlberg/libiscsi
2012-08-30 16:08:57 -07:00
01d3450ef5
Merge pull request #13 from bonzini/master
...
iscsi-test improvements
2012-08-30 16:08:10 -07:00
e4377e7d91
iscsi-test: return -2 when skipping tests
...
This lets a test driver distinguish failure from skipping unsupported
tests.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-30 17:39:00 +02:00
c7b519524c
iscsi-test: prepare to return exit status 77 for skipped tests
...
Teach the test harness to distinguish skipped tests from failed tests.
Return -2 from a skipped test and, if no test was failed but some were
skipped, return 77 from the main program.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-30 17:38:12 +02:00
f04471d642
iscsi-test: do not test read-only for non-SBC devices
...
The device-specific parameter is not defined in MMC.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-30 17:38:05 +02:00
418058e52a
iscsi-test: do not test DPO/FUA for non-SBC devices
...
These bits are not defined in MMC.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-30 17:37:53 +02:00
560241b494
iscsi-test: avoid declaring struct inside parameter list
...
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-30 17:37:50 +02:00
778aacb47f
fix building out-of-tree
...
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-30 17:37:38 +02:00
7b3c04a43c
Use a better iqn name for ld_iscsi.so
2012-08-17 14:50:40 +10:00
d1c7b17cf5
TEST: Start adding tests for invalid BUFFEROFFSET in DATA-IN
2012-08-12 12:51:37 +10:00
7c2fada513
TEST: Add test that DATASN in reverse order is detected as an error
2012-08-12 11:12:12 +10:00
5754851b9d
TESTS: More tests for invalid DATASN
2012-08-10 16:48:17 +10:00
6412834754
TEST: Add test that invalid DATASN for DATAIN is detected as an error
...
Send a write containing two DATA-IN PDUs in a sequence but both PDUs have DATASN==0
This is an error and the target should react to this as an invalid command
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com >
2012-08-10 16:08:55 +10:00
0cf42c3662
TESTS: Add a test for ISCSI CMDSN out of range
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com >
2012-08-10 15:36:10 +10:00
1a387632a9
TEST: Use better default iscsi name for the test tool
2012-08-10 11:27:10 +10:00
b7f342906f
Change to proper default IQN names for the iscsi-ls/-inq utilities
...
Change the default iqn name for the utilities to
iqn.2007-10.com.github:sahlberg:libiscsi:iscsi-ls/-inq
2012-08-10 10:17:47 +10:00
8eedcfed60
Merge pull request #12 from bonzini/libiscsi-tarball
...
include lib/libiscsi.syms in tarball
2012-08-09 14:17:27 -07:00
b5d6a9ff6f
include lib/libiscsi.syms in tarball
...
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com >
2012-08-09 15:38:46 +02:00
185649144d
Update the MTX patch for libiscsi.
...
MTX sets XFER-direction to "WRITE" even if the xfer length is 0
Catch this condition so that we pass XFER-direction==NONE to libiscsi for this case
2012-08-09 20:46:54 +10:00
08034c3335
SYNC CONNECT: make sure state is non-NULL before we dereference it
2012-08-09 08:46:29 +10:00
06cc2d2ece
CONNECT: connect data is not always malloc()ed so we can demand it is free()able
...
After a sync connection, make sure to clear connect_data since it will otherwise
point to a structure on the stack.
2012-08-09 08:27:48 +10:00
a98b6e63f0
MAXCMDSN: Stop sending new PDUs to the target when maxcmdsn is reached
...
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com >
2012-08-02 09:46:42 +10:00
71be26ad03
Add tracking of MAXCMDSN
2012-08-02 09:09:22 +10:00
439f68e555
CONNECTION: Track the connection state and callback via the context structure
...
Dont free the connection state when login is finished, instead track it via the
iscsi context structure and free it once the context is destroyed
2012-08-02 08:51:25 +10:00
Ronnie Sahlberg