This patch avoids that Valgrind complains about the memset() call
that was used to initialize the array that it triggers an invalid
write.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
all sync commands that return an integer value are
supposed to return a negative value on error.
However, state.status is positive non-zero on error.
Fix this by returning -1 if the state.status is
not SCSI_STATUS_GOOD.
Reported-by: Sitsofe Wheeler <sitsofe@yahoo.com>
Signed-off-by: Peter Lieven <pl@kamp.de>
If with int defintion, this statement:
num_blocks = client->num_blocks - client->pos;
May not get the exact value we want.
Signed-off-by: optimistyzy <optimistyzy@gmail.com>
we mangled the PDU header after calculating the checksum which
effectively broke CRC32C header digests completely.
Signed-off-by: Peter Lieven <pl@kamp.de>
commit bc64420 introduced an extra smalloc for the in->hdr,
however it did use iscsi_free instead of iscsi_sfree to free it.
Signed-off-by: Peter Lieven <pl@kamp.de>
EXTENDED COPY can be triggered with the new --xcopy/-x parameter. When
invoked, (--max) EXTENDED COPY requests are dispatched in parallel, with
each request attempting to copy (--blocks) from source to destination.
Signed-off-by: David Disseldorp <ddiss@suse.de>
Build on existing scsi_cdb_extended_copy() functionality. This operation
can be used to offload inter and intra LU copies to the target.
The API is rather primitive, in that the caller needs to construct the
parameter buffer, including CSCD and segment descriptor lists, etc.
Signed-off-by: David Disseldorp <ddiss@suse.de>
Build on existing scsi_cdb_receive_copy_results() functionality. This
request is most commonly used to determine target-side EXTENDED COPY
operational parameters.
Signed-off-by: David Disseldorp <ddiss@suse.de>
If the length of iscsi->waitpdu and iscsi->inqueue are the same
then except for any target initiated NOPs or async messages
we should have received any and all possible pdus from this
socket and can abort early.
This avoids running the loop one more time just to fail with EAGAIN
at the recs/readv. Just avoiding that recv/readv syscall will shave
at least 10us off this function and thus the latency.
Suggested-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Signed-off-by: Peter Lieven <pl@kamp.de>
iscsi_read_from_socket can currently only read one PDU in each iscsi_service invocation even
if there is more data available on the socket. This patch reads all PDUs until the socket
would block. It enqueues all complete read PDUs and then processes them in order of arrival.
Signed-off-by: Peter Lieven <pl@kamp.de>
With SPC-4, the 0xE4 descriptor LU ID TYPE field should be ignored, and
the NUL bit is obsolete. Update the ValidTgtDescr accordingly.
Signed-off-by: David Disseldorp <ddiss@suse.de>
ExtendedCopy.DescrLimits currently attempts to test Maximum Descriptor
List Length bounds checking by issuing a request with
(max_target_desc_count + 1 + max_segment_desc_count + 1) descriptors.
Maximum Descriptor List Length is a separately advertised field, that
may not be exceeded with the above logic, e.g. LIO advertises:
- Max Target Desc Count = 2
- Max Segment Desc Count = 1
- Max Desc List Len = 1024
Calculate the number of descriptors using the advertised Maximum
Descriptor List Length value, to explicitly test bounds checking.
Signed-off-by: David Disseldorp <ddiss@suse.de>
We need to use iovectors for iSER to work as it requires that the vectors
are attached to the task before we queue the task.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
