On scst teardown we call rdma_disconnect() on all connections.
this moves the qps to error state and flushes all the wc, including
the drain wc which will results in kref_put of the connection.
In a race condition with the connection request, we might be only in
the init stage of the ref, and calling kref_put will result in refcount
0 and freeing the connection while establishing it.
Call the first kref_get before rdma_accept() to prevent this race.
Signed-off-by: Ariel Nahum <arieln@mellanox.com>
Signed-off-by: Yan Burman <yanb@mellanox.com>
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6486 d57e44dd-8a1f-0410-8b47-8ef2f437770f
The second TM request can come while the old one, response for which is
going to be delayed, is still being processed, hence no response
prepared yet, so the delayed response should be dropped on the stage of
the new response preparing. Otherwise in this place the old delayed
response will trigger BUG_ON().
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6456 d57e44dd-8a1f-0410-8b47-8ef2f437770f
We were submitting work requests sometimes after the drain_wr,
so we would try to process flushes on something that is already
destroyed. This can be seen with very high login/logout load.
Signed-off-by: Yan Burman <yanb@mellanox.com>
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6428 d57e44dd-8a1f-0410-8b47-8ef2f437770f
In order to be able to support large block sizes (larger than 512K),
we allocate required structured in a lazy fashion. This way we both support
large block sizes (over 512K and even over 1M). We also lower memory footprint
when smaller block sizes are used.
Signed-off-by: Yan Burman <yanb@mellanox.com>
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6418 d57e44dd-8a1f-0410-8b47-8ef2f437770f
Preprocessor directives are not allowed in argument lists of a macro.
This patch reverts r6317.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6393 d57e44dd-8a1f-0410-8b47-8ef2f437770f
The put_page_callback patch only improves performance on single
socket systems but not on systems with multiple CPU sockets.
Since this message is confusing to SCST users, do not print a
warning if the put_page_callback patch has not been applied.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6340 d57e44dd-8a1f-0410-8b47-8ef2f437770f
It is annoying that some warnings are only reported in release mode.
Modify scst_debug.h such that the compiler does not report variables
that are only used in debug builds as unused.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6315 d57e44dd-8a1f-0410-8b47-8ef2f437770f
Avoid that the compiler complains that the variable 'pad_bytes'
is not used with CONFIG_LIBCRC32C=n.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6314 d57e44dd-8a1f-0410-8b47-8ef2f437770f
The T10-PI patch (r6162) increases the time needed to run mkfs.ext4
over iSCSI from less than a second to several minutes. This is a
regression so revert the T10-PI patch until there is sufficient time
to find the root cause of this regression and to fix this regression.
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6171 d57e44dd-8a1f-0410-8b47-8ef2f437770f
There is no need for a detect() method - code that is present in such
a method can be moved to module_init(). Hence declare this method
obsolete. Leave the detect pointer member in the target template for
now for out-of-tree SCST drivers.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
with some improvements
git-svn-id: http://svn.code.sf.net/p/scst/svn/trunk@6113 d57e44dd-8a1f-0410-8b47-8ef2f437770f
