socket: need to malloc hdr
include/iscsi-private: changing iscsi_in_pdu hdr to char*
instead of static array for more convinient iser
pdu creation.
To use iscsi_in_pdu in iSER without making a copy of it
we need to change hdr to pointer from static array,
Because of that, iscsi_tcp flow need to do szmalloc (small zero malloc)
hdr when creating new iscsi_in_pdu. iscsi_in_pdu is being malloced
once per each received pdu. This change is reducing iscsi_in_pdu struct
size but adding extra allocation of the same size we reduced
from the struct.
Signed-off-by: Roy Shterman <roysh@mellanox.com>
include/iscsi-private: adding queue_pdu in transport function pointers
struct
include/iscsi: declaration of tcp_queue_pdu function
socket: adding queue_pdu function to transport initialization
all_library: changing iscsi_queue_pdu into iscsi->t->queue_pdu
Signed-off-by: Roy Shterman <roysh@mellanox.com>
socket: adding tcp_connect function and implement it
in the last common part of iSER and TCP in iscsi_connect_async
Signed-off-by: Roy Shterman <roysh@mellanox.com>
lib/init: initializing connection transport
lib/socket: Adding function to initialize tcp transport.
future commits will include adding iSER transport layer,
so each transport option will has it's own template of functions.
all_utils: All utils stay the same, in future iSER commits in case of
iSER transport we will override tcp_transport in
iscsi_context.
connect: adding transport initialization when creating context
include/iscsi-private: Adding iscsi_transport attribute in iscsi_context
Signed-off-by: Roy Shterman <roysh@mellanox.com>
Make iscsi_init_tcp_transport private
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
This function allows the caller to determine the number of PDUs queued
for transmission for a given iscsi context.
Signed-off-by: David Disseldorp <ddiss@suse.de>
If the server closes the TCP connection before we call send() in
iscsi_write_to_socket() we will get SIGPIPE unless we pass this
argument to send.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Add a test to verify the pdu timeout handling.
Fix numerous bugs in the timeout handling. It was really broken.
Add test for non-SCSI task PDUs too and verify they works.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
the original idea of writing to the socket before reading was to put data
on the wire as early as possible and avoid potential latency increase due
to long taking callbacks invoked in iscsi_read_from_socket.
Benchmarking with libiscsi userland tools have shown that changing the
order increases throughput by about 5%.
The reason is that during POLLIN we might receive an updated MaxCmdSN
which might us allow to send new data out during POLLOUT.
For Qemu the change doesn't matter since POLLIN and POLLOUT are processed
independently.
Signed-off-by: Peter Lieven <pl@kamp.de>
the check was not quite correct. iscsi->pending_reconnect can be true but
we are still connected (and thus can disconnect). This is the case when
we receive an async logout request.
Signed-off-by: Peter Lieven <pl@kamp.de>
if reconnect is disabled then a failure in iscsi_service_reconnect_if_loggedin
can leak back to the application by iscsi_service returning an error
but the error string is not being updated.
Thus the applucation will print the last prevous error message that was set.
Change iscsi_service_reconnect_if_loggedin to print a less confusing
error message.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
If we are reconnecting and the connection is already established we
end up looping forever if there is a socket error before we finish
the login process. Fix this by invoking the reconnect callback if
there is an error and no new reconnect is scheduled.
Signed-off-by: Peter Lieven <pl@kamp.de>
this is meant as a hint that there can be no event at the moment
and in this special case iscsi_service will return immediately.
We use this at the moment in the case of a wait for a reconnect
in which case we would enter a busy loop waiting for that
reconnect. The 0 means do not poll for the client application.
Do not poll, just wait a little (a few hundred ms) and call iscsi_which_events
again. The wait should be timer driven or at least non busy looping,
of course.
Signed-off-by: Peter Lieven <pl@kamp.de>
Don't reset nalloc when resetting an iovector during reconnect.
Resetting offset/consumed should be sufficient.
Also, don't reset the iovectors when we detect an error condition in
iscsi_iovector_readv_writev. If there is a bug feeding an invalid
pos into this function, where pos suddently points before the amount
of data we have already read/written, then singlan this as an error
and return -EINVAL.
Previosly we did not reset the iovectors correctly when we re-queued
PDUs after a reconnect and thus relied on the iovectors being automatically
reset if/when we detected this type of error in readv_writev.
Now we do reset the iovectors properly so we do nt need to atuo reset them
here anymore and we can change this check to test for and abort the transfer
if an error is detected.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
From Adam Endrodi <adam.endrodi@nsn.com>
Fix a bug and clear is_corked during socket disconnect so that we
can re-use the context for a new connection.
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
set_tcp_sockopt in particular conflicts with libnfs's function of the same
name and prevents a a program from statically linking against both libnfs and
libiscsi.
Similar fix should also go into libnfs.
This issue was introduced via patch "pdu: introduce ISCSI_PDU_CORK_WHEN_SENT"
on June 13, 2014 (commit 99585b6996).
Valgrind reported this use-after-free as follows:
Invalid read of size 4
at 0x5267606: iscsi_write_to_socket (socket.c:721)
by 0x5267A72: iscsi_service (socket.c:823)
by 0x526827C: event_loop (sync.c:67)
by 0x52698A4: iscsi_compareandwrite_sync (sync.c:823)
by 0x408111: compareandwrite (iscsi-support.c:1752)
by 0x4139E2: test_compareandwrite_simple (test_compareandwrite_simple.c:88)
by 0x503D260: ??? (in /usr/lib64/libcunit.so.1.0.1)
by 0x503D578: ??? (in /usr/lib64/libcunit.so.1.0.1)
by 0x503D8B5: CU_run_all_tests (in /usr/lib64/libcunit.so.1.0.1)
by 0x4046C6: main (iscsi-test-cu.c:1241)
Address 0x639f258 is 8 bytes inside a block of size 256 free'd
at 0x4C291E7: free (vg_replace_malloc.c:473)
by 0x525321B: iscsi_free (init.c:68)
by 0x52532F0: iscsi_sfree (init.c:110)
by 0x5257AD9: iscsi_free_pdu (pdu.c:179)
by 0x5267601: iscsi_write_to_socket (socket.c:719)
by 0x5267A72: iscsi_service (socket.c:823)
by 0x526827C: event_loop (sync.c:67)
by 0x52698A4: iscsi_compareandwrite_sync (sync.c:823)
by 0x408111: compareandwrite (iscsi-support.c:1752)
by 0x4139E2: test_compareandwrite_simple (test_compareandwrite_simple.c:88)
by 0x503D260: ??? (in /usr/lib64/libcunit.so.1.0.1)
by 0x503D578: ??? (in /usr/lib64/libcunit.so.1.0.1)
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Cc: Peter Lieven <pl@kamp.de>
Since writing headers and payload in a single iov has never been
implementend and after thinking about it several times seems to
be very hairy I would like to revert this change since
the original implementation is in O(1) while the changed one
is in O(n). This results in a complexity of O(n^2) instead of
O(n) for the whole send operation.
This reverts commit 06eab264f6.
if the socket is corked we otherwise set POLLOUT and then
do not sent. Depending on the event loop implementation this
can result in a busy wait.
Signed-off-by: Peter Lieven <pl@kamp.de>
Rename the macros for managing the linked lists from SLIST_* to ISCSI_LIST_*
to avoid a clash on *BSD which already have other macros SLIST_*
Signed-off-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
