Jeff Skirvin [Wed, 14 Mar 2012 00:03:00 +0000 (17:03 -0700)]
isci: Fixed RNC bug that lost the suspension or resumption during destroy
This fix corrects the saving of resume parameters when the destruction
of the RNC has already been directed, and makes sure not to overwrite
the RNC destruction callbacks.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The RNC state machine would incorrectly transition from
SCI_RNC_AWAIT_SUSPENSION directly to SCI_RNC_INVALIDATING when a destruct
request was made. This would skip the increment of the suspension count
and the abort of pending TCs (although the invalidating state would at
least cleanup outstanding TCs).
Instead, the RNC will transition to SCI_RNC_SUSPENDED and then start the
destruction process.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Tue, 13 Mar 2012 00:29:51 +0000 (17:29 -0700)]
isci: Manage the IREQ_NO_AUTO_FREE_TAG under scic_lock.
Since there is a possibilty of a timeout waiting for the RNC suspension,
handle the exit case from the task termination under scic_lock, and leave
the tag allocated if the termination timed-out.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Sun, 4 Mar 2012 12:44:53 +0000 (12:44 +0000)]
isci: Remove obviated host callback list.
Since the callbacks to libsas now occur under scic_lock, there is no
longer any reason to save the completed requests in a separate list
for completion to libsas.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Sat, 10 Mar 2012 05:46:46 +0000 (05:46 +0000)]
isci: Check IDEV_GONE before performing abort path operations.
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>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:07 +0000 (22:42 -0800)]
isci: Change the phy control and link reset interface for HW reasons.
There is an apparent HW lockup caused when the PE is disabled while there
is an outstanding TC in progress. This change puts the link into OOB to
force the TC to end before the PE is disabled.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:06 +0000 (22:42 -0800)]
isci: Added timeouts to RNC suspensions in the abort path.
This change adds timeouts to the RNC suspension wait. It makes the
suspend and resume timeouts the same.
The previous resume timeout of 5 ms was too short, and timeouts were
seen in resumptions of devices in the abort task/LUN reset path - which
would receive an RNC resumed message within a tenth of a second later.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:06 +0000 (22:42 -0800)]
isci: Add protocol indicator for TMF requests.
Requests contructed as task management requests need to have the protocol
indicator set so the completion decode can observe any RNC suspension
conditions.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:05 +0000 (22:42 -0800)]
isci: Directly control IREQ_ABORT_PATH_ACTIVE when completing TMFs.
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>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:04 +0000 (22:42 -0800)]
isci: Wait for RNC resumption before leaving the abort path.
In the case of TMF execution, or device resets, wait for the RNC to fully
resume before returning to the caller. This ensures that the remote
device will not fail I/O requests while waiting for the RNC resumption to
complete.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:04 +0000 (22:42 -0800)]
isci: Fix RNC suspend call for SCI_RESUMING state.
Instead of immediately transitioning to the SCI_RNC_AWAIT_SUSPENSION
state, handle the SCI_RNC_RESUMING suspend transition from the
SCI_RNC_READY state like the SCI_RNC_INVALIDATING --> SCI_RNC_POSTING
transitions do now, by setting the destination state for the entry
into the READY state.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:03 +0000 (22:42 -0800)]
isci: Callbacks to libsas occur under scic_lock and are synchronized.
This patch changes the callback mechanism to libsas to only occur while
the scic_lock is held; the abort path cleanup of I/Os also checks to make
sure IREQ_ABORT_PATH_ACTIVE is clear before proceding.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:02 +0000 (22:42 -0800)]
isci: When in the abort path, defeat other resume calls until done.
Completion of I/Os during the one of the abort path interface calls
from libsas can drive remote device state changes and the resumption
of the device RNC. This is a problem when the abort path is
attempting to cleanup outstanding I/O at the same time - the resumption
can prevent the termination from occuring correctly.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:01 +0000 (22:42 -0800)]
isci: Implement waiting for suspend in the abort path.
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>
Jeff Skirvin [Fri, 9 Mar 2012 06:42:00 +0000 (22:42 -0800)]
isci: Make sure all TCs are terminated and cleaned in LUN reset.
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>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:59 +0000 (22:41 -0800)]
isci: Save the suspension hint for upcoming suspensions.
In the case of a suspend call while in SCI_RNC_POSTING or INVALIDATING
states, the LLHANG detect needed to be saved so the upcoming suspension
would enable it correctly. The unused suspend callback parameters were
removed.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:58 +0000 (22:41 -0800)]
isci: Fix the terminated I/O to not call sas_task_abort().
This addresses a regression from the commit "isci: Redesign
device suspension, abort, cleanup." in which the sas_task end
condition for terminated I/Os was made to call back on
sas_task_abort()".
This commit will be rolled into the original.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:54 +0000 (22:41 -0800)]
isci: Add suspension cases for RNC INVALIDATING, POSTING states.
The RNC can be any of the states in the loop from suspended to
ready when the API "suspend" or "resume" are called. This change
adds destination states parameters that control the suspension /
resumption action of the RNC statemachine for those transition states.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:54 +0000 (22:41 -0800)]
isci: Redesign device suspension, abort, cleanup.
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>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:51 +0000 (22:41 -0800)]
isci: Manage device suspensions during TC terminations.
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>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:50 +0000 (22:41 -0800)]
isci: Handle all suspending TC completions
Add comprehensive decode for all TC completions that generate RNC
suspensions.
Note that this commit also removes unconditional resumptions of ATAPI
devices when in the SCI_STP_DEV_ATAPI_ERROR state, and STP devices
when in the SCI_STP_DEV_IDLE state. This is because the SCI_STP_DEV_IDLE
and SCI_STP_DEV_ATAPI state entry functions manage the RNC resumption.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Fri, 9 Mar 2012 06:41:48 +0000 (22:41 -0800)]
isci: Manage the link layer hang detect timer for RNC suspensions.
For STP devices under certain protocol conditions, an RNC will not
suspend until the current transfer state is broken with a SYNC/ESC
sequence from the SCU. The SYNC/ESC driven by expiration of the
SCU link layer hang detect timer, which has too small a dynamic
range to support slow SATA devices, so normally it is disabled.
This change enables the timer with the minimum period at the point
when the suspension is requested.
Note that there is potential collateral damage to other open
connections to slow SATA devices on the same port, since there
is no alternative but to enable the LLHANG timer on every phy in
the port for the current suspension request - there is no way to
tell on which phy the RNC in question is currently active.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Wed, 15 Feb 2012 21:58:42 +0000 (13:58 -0800)]
isci: refactor initialization for S3/S4
Based on an original implementation by Ed Nadolski and Artur Wojcik
In preparation for S3/S4 support refactor initialization so that
driver-load and resume-from-suspend can share the common init path of
isci_host_init(). Organize the initialization into objects that are
self-contained to the driver (initialized by isci_host_init) versus
those that have some upward registration (initialized at allocation time
asd_sas_phy, asd_sas_port, dma allocations). The largest change is
moving the the validation of the oem and module parameters from
isci_host_init() to isci_host_alloc().
The S3/S4 approach being taken is that libsas will be tasked with
remembering the state of the domain and the lldd is free to be
forgetful. In the case of isci we'll just re-init using a subset of the
normal driver load path.
[clean up some unused / mis-indented function definitions in host.h]
Signed-off-by: Ed Nadolski <edmund.nadolski@intel.com> Signed-off-by: Artur Wojcik <artur.wojcik@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Tom Jackson [Fri, 24 Feb 2012 09:38:49 +0000 (09:38 +0000)]
isci: Don't filter BROADCAST CHANGE primitives
Per the SAS spec, several types of BROADCAST CHANGE primitives
must cause re-discovery of the originating expander.
Only the standard BROADCAST CHANGE primitive was being
sent to the LIBSAS layer. The other BC primitives have been
added to the sci_phy_event_handler()
Signed-off-by: Tom Jackson <thomas.p.jackson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Fri, 10 Feb 2012 09:05:43 +0000 (01:05 -0800)]
isci: improve 'invalid state' warnings
Convert controller state machine warnings to emit the state number (it
missed the number to string conversion, but since these error rarely
happen not much motivation to go further).
Fix up the rnc warnings to use the state name.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Thu, 22 Mar 2012 04:09:12 +0000 (21:09 -0700)]
libsas: suspend / resume support
libsas power management routines to suspend and recover the sas domain
based on a model where the lldd is allowed and expected to be
"forgetful".
sas_suspend_ha - disable event processing allowing the lldd to take down
links without concern for causing hotplug events.
Regardless of whether the lldd actually posts link down
messages libsas notifies the lldd that all
domain_devices are gone.
sas_prep_resume_ha - on the way back up before the lldd starts link
training clean out any spurious events that were
generated on the way down, and re-enable event
processing
sas_resume_ha - after the lldd has started and decided that all phys
have posted link-up events this routine is called to let
libsas start it's own timeout of any phys that did not
resume. After the timeout an lldd can cancel the
phy teardown by posting a link-up event.
Storage for ex_change_count (u16) and phy_change_count (u8) are changed
to int so they can be set to -1 to indicate 'invalidated'.
Cc: Alan Stern <stern@rowland.harvard.edu> Reviewed-by: Jacek Danecki <jacek.danecki@intel.com> Tested-by: Maciej Patelczyk <maciej.patelczyk@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Thu, 22 Mar 2012 04:09:11 +0000 (21:09 -0700)]
scsi, sd: limit the scope of the async probe domain
sd injects and synchronizes probe work on the global kernel-wide domain.
This runs into conflict with PM that wants to perform resume actions in
async context:
Provide a 'scsi_sd_probe_domain' so that async probe actions actions can
be flushed without regard for the state of PM, and allow for the resume
path to handle devices that have transitioned from SDEV_QUIESCE to
SDEV_DEL prior to resume.
Acked-by: Alan Stern <stern@rowland.harvard.edu>
[alan: uplevel scsi_sd_probe_domain, clarify scsi_device_resume] Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Thu, 22 Mar 2012 04:09:10 +0000 (21:09 -0700)]
libsas: drop sata port multiplier infrastructure
On the way to add a new sata_device field, noticed that libsas is
carrying port multiplier infrastructure that is explicitly disabled by
sas_discover_sata(). The aic94xx touches the unused port_no, so leave
that field in case there was some use for it.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Thu, 22 Mar 2012 04:09:08 +0000 (21:09 -0700)]
libata: export ata_port suspend/resume infrastructure for sas
Reuse ata_port_{suspend|resume}_common for sas. This path is chosen
over adding coordination between ata-tranport and sas-transport because
libsas wants to revalidate the domain at resume-time at the host level.
It can not validate links have resumed properly until libata has had a
chance to perform its revalidation, and any sane placing of an ata_port
in the sas-transport model would delay it's resumption until after the
host.
Export the common portion of port suspend/resume (bypass pm_runtime),
and allow sas to perform these operations asynchronously (similar to the
libsas async-ata probe implmentation). Async operation is determined by
having an external, rather than stack based, location for storing the
result of the operation.
Reviewed-by: Jacek Danecki <jacek.danecki@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 20 Mar 2012 17:58:35 +0000 (10:58 -0700)]
libata: reset once
Hotplug testing with libsas currently encounters a 55 second wait for
link recovery to give up. In the case where the user trusts the
response time of their devices permit the recovery attempts to be
limited to one.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Jeff Skirvin [Sat, 10 Mar 2012 05:55:50 +0000 (05:55 +0000)]
libsas: sas_rediscover_dev did not look at the SMP exec status.
The discovery function "sas_rediscover_dev" had two bugs: 1) it did
not pay attention to the return status from the SMP task execution;
2) the stack variable used for the returned SAS address was compared
against 0 without being initialized.
Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
[djbw: todo sanitize smp_execute_task return values (see: residue)] Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Thu, 9 Feb 2012 04:52:40 +0000 (20:52 -0800)]
libsas: trim sas_task of slow path infrastructure
The timer and the completion are only used for slow path tasks (smp, and
lldd tmfs), yet we incur the allocation space and cpu setup time for
every fast path task.
Cc: Christoph Hellwig <hch@lst.de> Acked-by: Jack Wang <jack_wang@usish.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Wed, 1 Feb 2012 09:12:23 +0000 (01:12 -0800)]
libsas: use ->lldd_I_T_nexus_reset for ->eh_bus_reset_handler
sas_eh_bus_reset_handler() amounts to sas_phy_reset() without
notification of the reset to the lldd. If this is triggered from
eh-cmnd recovery there may be sas_tasks for the lldd to terminate, so
->lldd_I_T_nexus_reset is warranted.
Cc: Xiangliang Yu <yuxiangl@marvell.com> Cc: Luben Tuikov <ltuikov@yahoo.com> Cc: Jack Wang <jack_wang@usish.com> Reviewed-by: Jacek Danecki <jacek.danecki@intel.com>
[jacek: modify pm8001_I_T_nexus_reset to return -ENODEV] Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 31 Jan 2012 07:39:11 +0000 (23:39 -0800)]
libsas: enforce eh strategy handlers only in eh context
The strategy handlers may be called in places that are problematic for
libsas (i.e. sata resets outside of domain revalidation filtering /
libata link recovery), or problematic for userspace (non-blocking ioctl
to sleeping reset functions). However, these routines are also called
for eh escalations and recovery of scsi_eh_prep_cmnd(), so permit them
as long as we are running in the host's error handler, otherwise arrange
for them to be triggered in eh_context.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
...teach sas_rphy_remove to wait for async scanning to quiesce before
removing the end_device. It seems this is a more general problem [1],
but this patch only addresses sas transport.
[1]: 23edb6e [SCSI] mpt2sas: Do not set sas_device->starget to NULL from
the slave_destroy callback when all the LUNS have been deleted
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Normalize phy->attached_sas_addr to return a zero-address in the case
when device-type == NO_DEVICE or the linkrate is invalid to handle
expanders that put non-zero sas addresses in the discovery response:
Dan Williams [Fri, 6 Apr 2012 23:35:36 +0000 (16:35 -0700)]
scsi: fix eh wakeup (scsi_schedule_eh vs scsi_restart_operations)
Rapid ata hotplug on a libsas controller results in cases where libsas
is waiting indefinitely on eh to perform an ata probe.
A race exists between scsi_schedule_eh() and scsi_restart_operations()
in the case when scsi_restart_operations() issues i/o to other devices
in the sas domain. When this happens the host state transitions from
SHOST_RECOVERY (set by scsi_schedule_eh) back to SHOST_RUNNING and
->host_busy is non-zero so we put the eh thread to sleep even though
->host_eh_scheduled is active.
Before putting the error handler to sleep we need to check if the
host_state needs to return to SHOST_RECOVERY for another trip through
eh.
Cc: Tejun Heo <tj@kernel.org> Reported-by: Tom Jackson <thomas.p.jackson@intel.com> Tested-by: Tom Jackson <thomas.p.jackson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Thu, 22 Mar 2012 04:09:07 +0000 (21:09 -0700)]
libsas, libata: fix start of life for a sas ata_port
This changes the ordering of initialization and probing events from:
1/ allocate rphy in PORTE_BYTES_DMAED, DISCE_REVALIDATE_DOMAIN
2/ allocate ata_port and schedule port probe in DISCE_PROBE
...to:
1/ allocate ata_port in PORTE_BYTES_DMAED, DISCE_REVALIDATE_DOMAIN
2/ allocate rphy in PORTE_BYTES_DMAED, DISCE_REVALIDATE_DOMAIN
3/ schedule port probe in DISCE_PROBE
This ordering prevents PHYE_SIGNAL_LOSS_EVENTS from sneaking in to
destrory ata devices before they have been fully initialized:
Dan Williams [Thu, 22 Mar 2012 04:09:05 +0000 (21:09 -0700)]
libata: make ata_print_id atomic
This variable is incremented from multiple contexts (module_init via
libata-lldds and the libsas discovery thread). Make it atomic to head
off any chance of libsas and libata creating duplicate ids.
Acked-by: Jacek Danecki <jacek.danecki@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Tue, 20 Mar 2012 20:24:29 +0000 (13:24 -0700)]
libsas: fix ata_eh clobbering ex_phys via smp_ata_check_ready
The check_ready implementation in the expander-attached ata device case
polls on sas_ex_phy_discover(). The effect is that the ex_phy fields
(critically ->attached_sas_addr) can change. When ata_eh ends and
libsas comes along to revalidate the domain
sas_unregister_devs_sas_addr() can fail to lookup devices to remove, or
fail to re-add an ata device that ata_eh marked as disabled. So change
the code to skip the sas_address and change count updates when ata_eh is
active.
Cc: Jack Wang <jack_wang@usish.com> Tested-by: Maciej Patelczyk <maciej.patelczyk@intel.com> Tested-by: Bartek Nowakowski <bartek.nowakowski@intel.com> Tested-by: Jacek Danecki <jacek.danecki@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reported-by: Tom Jackson <thomas.p.jackson@intel.com> Tested-by: Tom Jackson <thomas.p.jackson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Mon, 12 Mar 2012 18:38:26 +0000 (11:38 -0700)]
libsas: fix sas_get_port_device regression
Commit 899fcf4 "[SCSI] libsas: set attached device type and target
protocols for local phys" setup 'phy' to be dereferenced after
list_for_each_entry(phy, &port->phy_list, port_phy_el) (i.e. phy ==
&port->phy_list) resulting in reports like:
BUG: unable to handle kernel NULL pointer dereference at 00000000000002b0
IP: [<ffffffffa00ce948>] sas_discover_domain+0x29e/0x4fb [libsas]
...fix by deferring sas_phy_set_target() to the end of
sas_get_port_device().
Reported-by: Tom Jackson <thomas.p.jackson@intel.com> Tested-by: Tom Jackson <thomas.p.jackson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Thomas Jackson [Sat, 18 Feb 2012 02:33:10 +0000 (18:33 -0800)]
libsas: fix sas_find_bcast_phy() in the presence of 'vacant' phys
If an expander reports 'PHY VACANT' for a phy index prior to the one
that generated a BCN libsas fails rediscovery. Since a vacant phy is
defined as a valid phy index that will never have an attached device
just continue the search.
Cc: <stable@vger.kernel.org> Signed-off-by: Thomas Jackson <thomas.p.jackson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Fri, 2 Mar 2012 02:44:25 +0000 (18:44 -0800)]
libata, libsas: introduce sched_eh and end_eh port ops
When managing shost->host_eh_scheduled libata assumes that there is a
1:1 shost-to-ata_port relationship. libsas creates a 1:N relationship
so it needs to manage host_eh_scheduled cumulatively at the host level.
The sched_eh and end_eh port port ops allow libsas to track when domain
devices enter/leave the "eh-pending" state under ha->lock (previously
named ha->state_lock, but it is no longer just a lock for ha->state
changes).
Since host_eh_scheduled indicates eh without backing commands pinning
the device it can be deallocated at any time. Move the taking of the
domain_device reference under the port_lock to guarantee that the
ata_port stays around for the duration of eh.
Cc: Tejun Heo <tj@kernel.org> Acked-by: Jacek Danecki <jacek.danecki@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Maciej Trela [Mon, 5 Mar 2012 01:58:55 +0000 (17:58 -0800)]
libsas: cleanup spurious calls to scsi_schedule_eh
eh is woken up automatically by the presence of failed commands,
scsi_schedule_eh is reserved for cases where there are no failed
commands. This guarantees that host_eh_sceduled is only incremented
when an explicit eh request is made.
Reviewed-by: Jacek Danecki <jacek.danecki@intel.com> Signed-off-by: Maciej Trela <maciej.trela@intel.com>
[fixed spurious delete of sas_ata_task_abort] Signed-off-by: Artur Wojcik <artur.wojcik@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Dan Williams [Fri, 9 Mar 2012 19:00:06 +0000 (11:00 -0800)]
libsas: introduce sas_work to fix sas_drain_work vs sas_queue_work
When requeuing work to a draining workqueue the last work instance may
not be idle, so sas_queue_work() must not touch work->entry. Introduce
sas_work with a drain_node list_head to have a private list for
collecting work deferred due to drain collision.
Fixes reports like:
BUG: unable to handle kernel NULL pointer dereference at (null)
IP: [<ffffffff810410d4>] process_one_work+0x2e/0x338
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Nilesh Javali [Mon, 27 Feb 2012 11:08:52 +0000 (03:08 -0800)]
[SCSI] qla4xxx: Add support to display CHAP list and delete CHAP entry
For offload iSCSI like qla4xxx CHAP entries are stored in FLASH.
This patch adds support to list CHAP entries stored in FLASH and
delete specified CHAP entry from FLASH using iscsi tools.
Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com> Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Nilesh Javali [Mon, 27 Feb 2012 11:08:51 +0000 (03:08 -0800)]
[SCSI] iscsi_transport: Add support to display CHAP list and delete CHAP entry
For offload iSCSI like qla4xxx CHAP entries are stored in FLASH.
This patch adds support to list CHAP entries stored in FLASH and
delete specified CHAP entry from FLASH using iscsi tools.
Signed-off-by: Nilesh Javali <nilesh.javali@qlogic.com> Signed-off-by: Vikas Chaudhary <vikas.chaudhary@qlogic.com> Reviewed-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Santosh Nayak [Sun, 26 Feb 2012 14:44:46 +0000 (20:14 +0530)]
[SCSI] pm8001: fix endian issue with code optimization.
1. Fix endian issue.
2. Fix the following warning :
" drivers/scsi/pm8001/pm8001_hwi.c:2932:32: warning: comparison
between ‘enum sas_device_type’ and ‘enum sas_dev_type’".
3. Few code optimization.
Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com> Acked-by: Jack Wang <jack_wang@usish.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
In 'mpi_sata_completion', "pm8001_ha->lock" is first released.
It means lock is taken before, which is true for
the context A, as 'pm8001_ha->lock' is taken in 'pm8001_chip_isr()'
But for context B there is no lock taken before and pm8001_ha->lock
is unlocked in 'mpi_sata_completion()'. This may unlock the lock
taken in context A. Possible racing ??
If 'pm8001_ha->lock' is taken in 'process_oq()' instead of
'pm8001_chip_isr' then the above issue can be avoided.
Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com> Acked-by: Jack Wang <jack_wang@usish.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Santosh Nayak [Sun, 26 Feb 2012 13:33:30 +0000 (19:03 +0530)]
[SCSI] pm8001: Fix bogus interrupt state flag issue.
Static checker is giving following warning:
" error: calling 'spin_unlock_irqrestore()' with bogus flags"
The code flow is as shown below:
process_oq() --> process_one_iomb --> mpi_sata_completion
In 'mpi_sata_completion'
the first call for 'spin_unlock_irqrestore()' is with flags=0,
which is as good as 'spin_unlock_irq()' ( unconditional interrupt
enabling).
So for better performance 'spin_unlock_irqrestore()' can be replaced
with 'spin_unlock_irq()' and 'spin_lock_irqsave()' can be replaced by
'spin_lock_irq()'.
Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com> Acked-by: Jack Wang <jack_wang@usish.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Wayne Boyer [Thu, 23 Feb 2012 19:54:55 +0000 (11:54 -0800)]
[SCSI] ipr: update PCI ID definitions for new adapters
This patch updates some PCI ID definitions for new adapters based on the next
generation 64 bit IOA PCI interface chip.
Signed-off-by: Wayne Boyer <wayneb@linux.vnet.ibm.com> Acked-by: Brian King <brking@linux.vnet.ibm.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Carpenter [Tue, 21 Feb 2012 07:29:40 +0000 (10:29 +0300)]
[SCSI] qla2xxx: handle default case in qla2x00_request_firmware()
This silences a static checker warning. Also we're always adding new
types of firmware, so it might fix a bug in real life some day.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Acked-by: Chad Dupuis <chad.dupuis@qlogic.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Andrzej Jakowski [Fri, 10 Feb 2012 09:18:54 +0000 (01:18 -0800)]
[SCSI] isci: improvements in driver unloading routine
This patch fixes scenario where driver removal should be possible
only when driver is in READY state. Also it removes redundant
invocation of routine disabling SCU interrupts - this method is
called somewhere else in driver deinitialization path.
Signed-off-by: Andrzej Jakowski <andrzej.jakowski@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Williams [Tue, 31 Jan 2012 05:40:45 +0000 (21:40 -0800)]
[SCSI] libsas: don't recover end devices attached to disabled phys
If userspace has decided to disable a phy the kernel should honor that
and not inadvertantly re-enable the phy via error recovery. This is
more straightforward in the sata case where link recovery (via
libata-eh) is separate from sas_task cancelling in libsas-eh. Teach
libsas to accept -ENODEV as a successful response from I_T_nexus_reset
('successful' in terms of not escalating further).
This is a more comprehensive fix then "libsas: don't recover 'gone'
devices in sas_ata_hard_reset()", as it is no longer sata-specific.
aic94xx does check the return value from sas_phy_reset() so if the phy
is disabled we proceed with clearing the I_T_nexus.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Williams [Thu, 9 Feb 2012 07:20:41 +0000 (23:20 -0800)]
[SCSI] libsas: fixup target_port_protocols for expanders that don't report sata
If discovery returns 0 for target_port_protocols but shows an attached
sata device, just report SAS_PROTOCOL_SATA in the identify data so
userspace can reliably search for sata devices in the domain.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Williams [Fri, 20 Jan 2012 23:26:03 +0000 (15:26 -0800)]
[SCSI] libsas: revert ata srst
libata issues follow up srsts when the controller has a hard time
recording the signature-fis after a reset, or if the link supports port
multipliers. libsas does not support port multipliers and no current
libsas lldds appear to need help retrieving the signature fis. Revert
it for now to remove confusion.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Williams [Fri, 20 Jan 2012 23:23:07 +0000 (15:23 -0800)]
[SCSI] libsas: fix lifetime of SAS_HA_FROZEN
Until all sas_tasks are known to no longer be in-flight this flag gates late
completions from colliding with error handling. However, it must be cleared
prior to the submission of scsi_send_eh_cmnd() requests, otherwise those
commands will never be completed correctly.
This was spotted by slub debug:
=============================================================================
BUG sas_task: Objects remaining on kmem_cache_close()
-----------------------------------------------------------------------------
Dan Williams [Thu, 19 Jan 2012 04:47:01 +0000 (20:47 -0800)]
[SCSI] libsas: async ata scanning
libsas ata error handling is already async but this does not help the
scan case. Move initial link recovery out from under host->scan_mutex,
and delay synchronization with eh until after all port probe/recovery
work has been queued.
Device ordering is maintained with scan order by still calling
sas_rphy_add() in order of domain discovery.
Since we now scan the domain list when invoking libata-eh we need to be
careful to check for fully initialized ata ports.
Acked-by: Jack Wang <jack_wang@usish.com> Acked-by: Jeff Garzik <jgarzik@redhat.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Williams [Thu, 19 Jan 2012 04:14:01 +0000 (20:14 -0800)]
[SCSI] libsas: restore scan order
ata devices are always scanned after ssp. Prior to the ata error
handling reworks libsas would tend to scan devices in ascending expander
phy order. Restore this ordering by deferring ssp discovery to a
DISCE_PROBE event, and keep the probe order consistent with the
discovery order, not the placement of sata devices.
Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Dan Williams [Fri, 13 Jan 2012 01:57:35 +0000 (17:57 -0800)]
[SCSI] libsas: let libata recover links that fail to transmit initial sig-fis
libsas fails to discover all sata devices in the domain. If a device fails
negotiation and does not transmit a signature fis the link needs recovery.
libata already understands how to manage slow to come up links, so treat these
conditions as ata device attach events for the purposes of creating an
ata_port. This allows libata to manage retrying link bring up.
Rediscovery is modified to be careful about checking changes in dev_type. It
looks like libsas leaks old devices if the sas address changes, but that's a
fix for another patch.
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>
Dan Williams [Mon, 16 Jan 2012 21:54:28 +0000 (13:54 -0800)]
[SCSI] libsas: fix sas port naming
Make sas-port naming consistent with the expander-attached case whereby
the phy-id is the last digit in the port name. Otherwise we get the
random behavior of the allocation order.
Reported-by: Patrick Thomson <patrick.s.thomson@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
which shows attached link rate, device type, and associates a
domain_device with its ata_port id to correlate messages emitted from
libata-eh.
As Doug notes, we can also take the opportunity to clarify expander phy
routing capabilities.
[dgilbert@interlog.com: clarify table2table with 'U'] Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Maciej Trela [Fri, 13 Jan 2012 21:52:38 +0000 (21:52 +0000)]
[SCSI] libsas: kill spurious sas_put_device
Holdover from a patch rework, prior to the addition of SAS_DEV_DESTROY
we were holding a reference while the destruct was pending in case the
domain was torn down before the desctruct event ran. That case is
covered by SAS_DEV_DESTROY, and the sas_put_device() just corrupts freed
memory, or worse frees the memory while another agent holds a reference.
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>
Dan Williams [Thu, 12 Jan 2012 19:47:24 +0000 (11:47 -0800)]
[SCSI] libsas: fix sas_unregister_ports vs sas_drain_work
We need to hold drain_mutex across the unregistration as port down events
queue device removal as chained events, so we need to make sure no other
drainers are active.
Reported-by: Jacek Danecki <jacek.danecki@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: James Bottomley <JBottomley@Parallels.com>
projects / karo-tx-linux.git / log