Modified the event coalescing code for iSCSI offload to combat both
corner cases and optimize performance as follows:
1. Added mechanism to loop back a second time to process any leftover
CQEs that was generated by the hardware during the time the driver is
busy processing previous CQEs in the bh. This not only helps the
performance but also fixes the corner case when no more CQEs are being
generated in the pipeline; so those leftover CQEs will get a a chance
to be processed.
2. Added ARM_CQE_FP to distinguish between fast path arming versus
slow path arming. This change will guarantee that the CQEs will
always get a chance to be re-armed during fast path completions.
3. Removed the inline event coalescing division for perf optimization.
Also fixed a division-by-zero error when the event_coal_div module
param was set to 0.
4. Changed the default SQ WQEs size from 256 to 128 to match chip
default.
5. Changed the cmd_per_lun from 32 to 24.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <jbottomley@parallels.com>
Modified the 10s wait time for inflight offload connections to
advance to the next state to 2s based on test result.
Modified the 20s shutdown timeout to 30s based on test result.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <jbottomley@parallels.com>
The number of chip's internal command cell, which is use to generate
SCSI cmd packets to the target, was not initialized correctly by
the driver when the sq_size is changed from the default 128.
This, in turn, will create a problem where the chip's transmit pipe
will erroneously reuse an old command cell that is no longer valid.
The fix is to correctly initialize the chip's command cell upon setup.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <jbottomley@parallels.com>
bnx2i has some checks to try and make sure the ep
is not destroyed while the addr/port is getting
read. However, if after this check:
if (!(bnx2i_conn && bnx2i_conn->ep && bnx2i_conn->ep->hba))
goto out;
bnx2i_conn->ep is cleared by ep_disconnect then we will
oops.
This patches fixes the problem by having the driver
use the get_ep_param callback instead of get_conn_param.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This is part of an effort to support send target discovery via
the iSCSI offload path.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Allow CNIC to go through the proper cleanup procedure for an endpoint
which failed to connect. Proper cleanup is necessary for the chip
to reset back to the initial state for the offloaded endpoint.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Various error conditions inside ep_connect and ep_disconnect were
either not being handled or not being handled correctly. This patch
fixes all those issues.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Added the handling for cases when a chip request is made to the
CNIC module but the hardware is not ready to accept. This would
lead to many unnecessary wait timeouts.
This code adds check in the connect establishment and destruction
path.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The stop path has been augmented to wait a max of 10s for all in
progress offload and destroy activities to complete before proceeding
to terminate all active connections (via iscsid or forcefully).
Note that any new offload and destroy requests are now blocked and
return to the caller immediately.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The code no longer needs to dynamically register and unregister
the CNIC device. The CNIC device will be kept registered until
module unload.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Added net_dev mutex lock protection before accessing the csk
parameters.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
In the situation where the connect completion response arrives after
the connect request has already timed out, the connection was not being
aborted but only the resource was being freed. This creates a problem
for 5771X (10g) as the chip flags this with an assertion.
This change will properly aborts the connection before freeing the
resource.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Modified the handling of the remote TCP RST code so the chip can now
flush the tx pipe accordingly upon a remote TCP RST reception.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
A cid leak issue was found when the connect destroy request exceeded
the driver's disconnection timeout. This will lead to a cid resource
leak issue.
The fix is to allow the cid cleanup even when this happens.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
In the case when an ep_connect request is interrupted due to route
request stall, if the iSCSI daemon is terminated by the user, the chip
will be left in a state which will not get cleaned up upon module
removal. Upon module reload, when the same context id is used for a
new connection, chip panic would occur.
This patch adds chip cleanup in the module removal path.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Acked-by: Benjamin Li <benli@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Specific to the Broadcom 10g chipset, the CFC delete operation must be
coupled with the cm_abort/close with does the SRC delete/terminate offload
operation prior.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Acked-by: Benjamin Li <benli@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
According to RFC3720, nopout packet sent in response to unsolicited
nopin packet requesting a response must retain the TTT of the requester.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This sysfs attribute is proven to be useful during pivot_root.
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
In compliance to RFC793, a TCP graceful termination will be used
instead of an abortive termination for the case where the remote
has initiated the close of the connection.
Additionally, a TCP abortive termination will be used to close the
connection when a logout response is not received in time after a
logout request has been initiated.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Added variables to separate the fine tuned timeout values for
connection destroy and context destroy for both 1g and 10g devices.
v2: Extended the 5771X disconnect timeout from 10s to 20s as the firmware
has a retransmission timeout of 16s. This fixes one of the iscsi_endpoint
leak issues when the target is slow or non-responsive to our TCP FIN.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
For cases where the iSCSI disconnection procedure times out due to
the iSCSI daemon being slow or unresponsive, the bnx2i_stop routine
will now perform hardware cleanup via bnx2i_hw_ep_disconnect on all
active endpoints so that subsequent operations will perform properly.
Also moved the mutex locks inside ep_connect and ep_disconnect so
that proper exclusivity can resolve simultaneous calls to the
ep_disconnect routine.
v2: Removed the unnecessary read lock in the bnx2i_stop
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This introduces a new active linklist which would link up all active
bnx2i_endpoints. This will be used by subsequent patches that
follows.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This patch introduces a new bnx2i_hw_ep_disconnect routine which
contains all chip related disconnect and clean up procedure of
iSCSI offload connections. This separation is intended as a
preparation for the subsequent bnx2i_stop patch.
Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
bnx2i driver has to wait and cleanup all iscsi endpoints before
returning from bnx2i_stop(). This is to make sure all chip resources
are freed before chip is reset.
As the requirements for 1G and 10G chipsets is different, added
per-device 'hba_shutdown_tmo' parameter to adapter structure
If the connections are not torn down by the daemon within this timeout
period, 'cid's will be leaked in 10G device. 1G devices are more
flexible and do not leak any resources because the whole chip ports
gets reset when MTU is changed or ethtool selftest is run
fixed a minor issue in bnx2i_ep_poll() which unnecessarily forced
error return code when driver timed out waiting for TCP connect
request to complete
Signed-off-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
The iscsi_eh_target_reset has been modified to attempt
target reset only. If it fails, then iscsi_eh_session_reset
will be called.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Jayamohan Kallickal <jayamohank@serverengines.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
No reason that we cannot set the change_queue_depth
function for bnx2i. We just forgot to when the
driver was created.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Remove uses of NIPQUAD, use %pI4
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Removed duplicate function call and not-so-useful comment line
Signed-off-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This implements warm target reset tmf support for
the scsi-ml target reset callback. Previously we would
just drop the session in that callback. This patch will
now try a target reset and if that fails drop the session.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This just has bnx2i use the iscsi_suspend_queue helper.
The suspend works as follows:
When ep_poll has succeeed iscsid will call conn_bind, the LLD will
then call iscsi_conn_bind which will clear the suspend bit.
When ep_disconnect is called (or if there is a conn error) we set
the suspend bit. For the ep_disconnect case I added a helper
in the previous kernel that will take the session lock to make sure
iscsi_queuecommand/xmit_task is not running and it will set
the suspend bit.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Acked-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
fix some typos and punctuation in comments
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
This patch contains changes that allow iscsi_session_setup
to allocate private space for LLD's
Signed-off-by: Jayamohan Kallickal <jayamohank@serverengines.com>
Acked-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
When bnx2i_adapter_ready() fails, connection handle(cid) = 0 is wrongly freed
because 'cid' is not yet allocated for the endpoint. Fix is to initialize
bnx2i_ep->ep_iscsi_cid to '-1' in bnx2i_alloc_ep() and not in
bnx2i_ep_connect() to avoid releasing invalid 'cid'. There is already a check
in bnx2i_free_iscsi_cid() not to free invalid iscsi connection handle (-1)
Signed-off-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Without the fix bnx2i would fail tt->xmit_task() when link is down and
libiscsi would have already incremented session->cmdsn before calling bnx2i's
xmit_task() entry point and will just return the command to SCSI-ML when
xmit_task() fails. libiscsi does not retract the session->cmdsn as the command
was never sent on wire. It is generally good idea for LLD, bnx2i to accept
the scsi cmnd/nopout and let upper layer timeout and go though normal session
recovery process. When link is down, unsolicited nopout will not be accepted
by bnx2i and connection will never enter recovery state. This fix is required
for MPIO to work corectly
Signed-off-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
When using iface, bnx2i was unable to offload further connections after all
active sessions are logged out. bnx2i will unregister the device from cnic
when the last connection is torn down. Next call to ep_connect() will fail
because the device is not registered. This issue is not seen if shost == NULL
is passed to ep_connect() call because in that case bnx2i will registers all
known devices with cnic before doing a route look-up. When shost != NULL,
bnx2i knows the device on which to offload the connection and has to register
this device before attempting to offload the connection
Signed-off-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
New iSCSI driver for Broadcom BNX2 devices. The driver interfaces with
the CNIC driver to access the hardware.
Signed-off-by: Anil Veerabhadrappa <anilgv@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Eddie Wai