Remove an erroneous BUG_ON() in the case of a hard reset timeout. The
reset timeout handler puts the port into the "awaiting link-up" state.
The timeout causes the device to be disconnected and we need to be in
the awaiting link-up state to re-connect the port. The BUG_ON() made
the incorrect assumption that resets never timeout and we always
complete the reset in the "resetting" state.
Testing this patch also uncovered that libata continues to attempt to
reset the port long after the driver has torn down the context. Once
the driver has committed to abandoning the link it must indicate to
libata that recovery ends by returning -ENODEV from
->lldd_I_T_nexus_reset().
Cc: <stable@vger.kernel.org>
Acked-by: Lukasz Dorau <lukasz.dorau@intel.com>
Reported-by: David Milburn <dmilburn@redhat.com>
Reported-by: Xun Ni <xun.ni@intel.com>
Tested-by: Xun Ni <xun.ni@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
This commit fixes a race condition in the isci driver abort task and SSP
device task management path. The race is caused when an I/O termination
in the SCU hardware is necessary because of an SSP target timeout condition,
and the check of the I/O end state races against the HW-termination-driven
end state. The failure of the race meant that no TMF was sent to the device
to clean-up the pending I/O.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Reviewed-by: Lukasz Dorau <lukasz.dorau@intel.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
These enums have been separate since the dawn of SAS, mainly because the
latter is a procotol only enum and the former includes additional state
for libsas. The dichotomy causes endless confusion about which one you
should use where and leads to pointless warnings like this:
drivers/scsi/mvsas/mv_sas.c: In function 'mvs_update_phyinfo':
drivers/scsi/mvsas/mv_sas.c:1162:34: warning: comparison between 'enum sas_device_type' and 'enum sas_dev_type' [-Wenum-compare]
Fix by eliminating one of them. The one kept is effectively the sas.h
one, but call it sas_device_type and make sure the enums are all
properly namespaced with the SAS_ prefix.
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
In the link fail path, set IDEV_GONE for every device on the domain
when the last link in the port fails.
In the abort path functions like isci_reset_device, make sure that
there has not already been a detected domain failure with the device
by checking IDEV_GONE, before performing any kind of hard reset, SMP
phy control, or TMF operation.
The check for IDEV_GONE makes sure that the device in the abort path
really has control of the port with which it is associated. This
prevents starting hard resets at incorrect times and scheduling
unnecessary LUN resets for SATA devices.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
TMF requests, unlike normal I/O requests, need to handle I/O management
conditions in the completion function because TMFs are not handled in the
completion tasklet.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
When an individual request is being terminated, the request's tag
is managed in the terminate function.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
In order to prevent a device from receiving an I/O request while still
in an RNC suspending or resuming state (and therefore failing that
I/O back to libsas with a reset required status) wait for the RNC state
change before proceding in the abort path.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
In the libsas error path, SATA disks require extra handling in
libata to recover operation. However, libsas expects to be able
to immediately recover all outstanding I/O once the error handler
escalation stops. This patch fixes the condition where the libata
error handler is scheduled for operation but libsas has already
deleted the outstanding sas_tasks.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
This commit changes the means by which outstanding I/Os are handled
for cleanup.
The likelihood is that this commit will be broken into smaller pieces,
however that will be a later revision. Among the changes:
- All completion structures have been removed from the tmf and
abort paths.
- Now using one completed I/O list, with the I/O completed in host bit being
used to select error or normal callback paths.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
If LUN reset sees that the device is gone, it returns TMF_RESP_FUNC_FAILED
to cause libsas to escalate to an I_T_Nexus_Reset.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
TCs must be terminated only while the RNC is suspended. This commit
adds remote device suspensions and resumptions in the abort, reset and
termination paths.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Prior to commit 61aaff49 "isci: filter broadcast change notifications
during SMP phy resets" we borrowed the MVS_DEV_EH approach from the
mvsas driver for preventing ->lldd_I_T_nexus_reset() events during ata
discovery. This hack was protecting against the old ->phy_reset() in
ata_bus_probe(), but since the conversion to the new error handling this
hack is preventing resets from reaching ata devices.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Remove ->eh_device_reset_handler() and ->eh_bus_reset_handler() for the
same reason they are not implemented for libata hosts, they cannot be
implemented reliably with ata-eh. ATA error recovery wants to divert
all resets to the eh thread and wait for completion, these handlers may
be invoked from a non-blocking ioctl.
The other path they are called from is libsas-eh, and if we escalate
past I_T_nexus reset we have larger problems i.e. tear down all
in-flight commands in the domain potentially without notification to the
lldd if it has chosen not to implement ->lldd_clear_nexus_port() /
->lldd_clear_nexus_ha().
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Driving resets from libsas-eh is pre-mature as libata will make a
decision about performing a softreset. Currently libata determines
whether to perform a softreset based on ata_eh_followup_srst_needed(),
and none of those conditions apply to isci.
Remove the srst implementation and translate ->lldd_lu_reset() for ata
devices as a request to drive a reset via libata-eh.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
A hard reset to isci in the direct-attached case is one where the driver
internally manages debouncing the link. In the sas-expander-attached
case a hard reset is one that clears affiliations. The driver should
not be prematurely dropping affiliations at run time, that decision (to
force expander hard resets to ata devices) is left to userspace to
manage. So, arrange for I_T_nexus resets to be sas-link-resets in the
expander-attached case and isci-hard-resets in the direct-attached case.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
In the direct-attached case this routine returns the phy on which this
device was first discovered. Which is broken if we want to support
wide-targets, as this phy reference can become stale even though the
port is still active.
In the expander-attached case this routine tries to lookup the phy by
scanning the attached sas addresses of the parent expander, and BUG_ONs
if it can't find it. However since eh and the libsas workqueue run
independently we can still be attempting device recovery via eh after
libsas has recorded the device as detached. This is even easier to hit
now that eh is blocked while device domain rediscovery takes place, and
that libata is fed more timed out commands increasing the chances that
it will try to recover the ata device.
Arrange for dev->phy to always point to a last known good phy, it may be
stale after the port is torn down, but it will catch up for wide port
reconfigurations, and never be NULL.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Each libsas driver (mvsas, pm8001, and isci) has invented a different
method for managing the ap->lock. The lock is held by the ata
->queuecommand() path. mvsas drops it prior to acquiring any internal
locks which allows it to hold its internal lock across calls to
task->task_done(). This capability is important as it is the only way
the driver can flush task->task_done() instances to guarantee that it no
longer has any in-flight references to a domain_device at
->lldd_dev_gone() time.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
As the field was never set, isci_print_tmf() using 'isci_tmf->device'
sometimes causes a kernel crash if the dev_dbg() statement is enabled.
Remove the unused field both from isci_tmf struct definition and from
isci_print_tmf()
Signed-off-by: Maciej Trela <maciej.trela@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
The initial bcn filtering implementation was validated on a kernel
baseline that predated the switch to new libata error handling. Also,
prior to that conversion we borrowed the mvsas MVS_DEV_EH approach to
prevent the unwanted extra ap->ops->phy_reset(ap) that occurred in the
ata_bus_probe() path.
After the conversion to new libata eh resets at discovery are more
frequent and get filtered prematurely by IDEV_EH. The result is that
our bcn filtering has been blocked from running and at discovery and it
appears to stall discovery completion to the point of triggering hung
task timeouts. So, revert the implementation for now. When it returns
it will go into libsas proper.
The domain rediscovery that takes place due to ->lldd_I_T_nexus_reset()
events should now be properly waited for by the ata_port_wait_eh() call
in ata_port_probe(). So the hard coded delay in the isci
->lldd_I_T_nexus_reset() and other libsas drivers should help debounce
the libsas thread from seeing temporary device removals.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
The lldd does not need to look at or manage the pending device
reset bit in pending sas_tasks.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Use the existing IREQ_TMF flag as a request type indicator.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
libsas uses the LLDD abort task interface to handle I/O timeouts
in the SATA/STP and SMP discovery paths, so this change will terminate
STP/SMP requests. Also, if the device is gone, the lldd will prevent
libsas from further escalations in the error handler.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
libsas will cleanup pending sas_tasks after error handler
path functions are called; do not call task_done callbacks.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
In the case where "task" requests timeout (note that this class of
requests can also include SATA/STP soft reset FIS transmissions),
handle the case where the task was being managed by some call to
terminate the task request by completing both the tmf and the aborting
process.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Make sure terminated requests and completed task tags are freed.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
In the case where an I/O fails to start in isci_request_execute,
only allow retries if the device is not already gone.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Now that isci has added a 3rd open coded user of this functionality just
share the libsas version.
Acked-by: Jack Wang <jack_wang@usish.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
The hard_reset parameter passed to the LLDD in the direct-attached
phy control case allows the LLDD to filter link failure events
while the direct-attached device reset is executing.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The messages emitted from task.c and some from request.c likely
duplicate (in a less undertandable way) what is reported by the
midlayer.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Most of these simple dereference macros are longer than their open coded
equivalent. Deleting enum sci_controller_mode is thrown in for good
measure.
Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The distinction between scic_sds_ scic_ and sci_ are no longer relevant
so just unify the prefixes on sci_. The distinction between isci_ and
sci_ is historically significant, and useful for comparing the old
'core' to the current Linux driver. 'sci_' represents the former core as
well as the routines that are closer to the hardware and protocol than
their 'isci_' brethren. sci == sas controller interface.
Also unwind the 'sds1' out of the parameter structs.
Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Remove the distinction between these two implementations and unify on
isci_host (local instances named ihost). Hmmm, we had two
'oem_parameters' instances, one was unused... nice.
Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Remove the distinction between these two implementations and unify on
isci_remote_device (local instances named idev).
Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
They are one in the same object so remove the distinction. The near
duplicate fields (owning_controller, and isci_host) will be cleaned up
after the scic_sds_contoller isci_host unification.
Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
the dma_pool interface is optimized for object_size << page_size which
is not the case with isci_request objects and the dma_pool routines show
up in the top of the profile.
The old io_request_table which tracked whether tci slots were in-flight
or not is replaced with an IREQ_ACTIVE flag per request.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Combine three bools into one unsigned long 'flags'. Doesn't increase the
request size due to packing. (to do: optimize the structure layout).
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The tci_pool tracks our outstanding command slots which are also the 'index'
portion of our tags. Grabbing the tag early in ->lldd_execute_task let's us
drop the isci_host_can_queue() and ->was_tag_assigned_by_user infrastructure.
->was_tag_assigned_by_user required the task context to be duplicated in
request-local buffer. With the tci established early we can build the
task_context directly into its final location and skip a memcpy.
With the task context buffer at a known address at request construction we
have the opportunity/obligation to also fix sgl handling. This rework feels
like it belongs in another patch but the sgl handling and task_context are too
intertwined.
1/ fix the 'ab' pair embedded in the task context to point to the 'cd' pair in
the task context (previously we were prematurely linking to the staging
buffer).
2/ fix the broken iteration of pio sgls that assumes all sgls are relative to
the request, and does a dangerous looking reverse lookup of physical
address to virtual address.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
When the remote device transitions to a not-ready state because of
an NCQ error condition, all outstanding requests to that device
are terminated and completed to libsas on the normal path. The
device then waits for a READ LOG EXT command to issue on the task
management path.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
1/ fix the timeout for wait_for_completion_timeout
2/ In the tmf timeout case we need to wait for our termination callback
3/ Once the request is successfully started it will be freed according to the
normal lifetime for requests.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Now that "stopping/stopped" are one in the same and signalled by a NULL device
pointer the rest of the device status infrastructure can be removed (->status
and ->state_lock). The "not ready for i/o state" is replaced with a state
flag, and is evaluated under scic_lock so that we don't see transients from
taking the device reference to submitting the i/o.
This also fixes a potential leakage of can_queue slots in the rare case that
SAS_TASK_ABORTED is set at submission.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
We have unsafe references to remote devices that are notified to
disappear at lldd_dev_gone. In order to clean this up we need a single
canonical source for device lookups and stable references once a lookup
succeeds. Towards that end guarantee that domain_device.lldd_dev is
NULL as soon as we start the process of stopping a device. Any code
path that wants to safely lookup a remote device must do so through
task->dev->lldd_dev (isci_lookup_device()).
For in-flight references outside of scic_lock we need reference counting
to ensure that the device is not recycled before we are done with it.
Simplify device back references to just scic_sds_request.target_device
which is now the only permissible internal reference that is maintained
relative to the reference count.
There were two occasions where we wanted new i/o's to be treated as
SAS_TASK_UNDELIVERED but where the domain_dev->lldd_dev link is still
intact. Introduce a 'gone' flag to prevent i/o while waiting for libsas
to take action on the port down event.
One 'core' leftover is that we currently call
scic_remote_device_destruct() from isci_remote_device_deconstruct()
which is called when the 'core' says the device is stopped. It would be
more natural for the final put to trigger
isci_remote_device_deconstruct() but this implementation is deferred as
it requires other changes.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Rather than return an error code and update a pointer that was passed by
reference just return the request object directly (or null if allocation
failed).
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Walk through the list of pending requests being careful to consider that
multiple requests can be terminated when the lock is dropped (i.e.
invalidating the 'next' reference established by
list_for_each_entry_safe).
Also noticed that all callers to isci_terminate_pending_requests()
specifying terminating, so just drop the parameter.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
In the situation where a termination of an I/O times-out,
make sure that the linkage from the request to the task
is severed completely. Also make sure that the selection
of tasks to terminate occurs under scic_lock.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
When resetting a sata device in the domain we have seen occasions where
libsas prematurely marks a device gone in the time it takes for the
device to re-establish the link. This plays badly with software raid
arrays. Other libsas drivers have non-uniform delays in their reset
handlers to try to cover this condition, but not sufficient to close the
hole. Given that a sata device can take many seconds to recover we
filter bcns and poll for the device reattach state before notifying
libsas that the port needs the domain to be rediscovered. Once this has
been proven out at the lldd level we can think about uplevelling this
feature to a common implementation in libsas.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
[ use kzalloc instead of kmem_cache ]
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
[ use eventq and time macros ]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams
Jeff Skirvin
James Bottomley
Anatol Pomozov
Maciej Trela
James Bottomley