diff options
author | James Bottomley <JBottomley@Parallels.com> | 2012-05-21 12:17:30 +0100 |
---|---|---|
committer | James Bottomley <JBottomley@Parallels.com> | 2012-05-21 12:17:30 +0100 |
commit | e34693336564f02b3e2cc09d8b872aef22a154e9 (patch) | |
tree | 09f51f10f9406042f9176e39b4dc8de850ba712e /drivers/scsi/isci/remote_device.c | |
parent | 76b311fdbdd2e16e5d39cd496a67aa1a1b948914 (diff) | |
parent | de2eb4d5c5c25e8fb75d1e19092f24b83cb7d8d5 (diff) | |
download | blackbird-op-linux-e34693336564f02b3e2cc09d8b872aef22a154e9.tar.gz blackbird-op-linux-e34693336564f02b3e2cc09d8b872aef22a154e9.zip |
Merge tag 'isci-for-3.5' into misc
isci update for 3.5
1/ Rework remote-node-context (RNC) handling for proper management of
the silicon state machine in error handling and hot-plug conditions.
Further details below, suffice to say if the RNC is mismanaged the
silicon state machines may lock up.
2/ Refactor the initialization code to be reused for suspend/resume support
3/ Miscellaneous bug fixes to address discovery issues and hardware
compatibility.
RNC rework details from Jeff Skirvin:
In the controller, devices as they appear on a SAS domain (or
direct-attached SATA devices) are represented by memory structures known
as "Remote Node Contexts" (RNCs). These structures are transferred from
main memory to the controller using a set of register commands; these
commands include setting up the context ("posting"), removing the
context ("invalidating"), and commands to control the scheduling of
commands and connections to that remote device ("suspensions" and
"resumptions"). There is a similar path to control RNC scheduling from
the protocol engine, which interprets the results of command and data
transmission and reception.
In general, the controller chooses among non-suspended RNCs to find one
that has work requiring scheduling the transmission of command and data
frames to a target. Likewise, when a target tries to return data back
to the initiator, the state of the RNC is used by the controller to
determine how to treat the incoming request. As an example, if the RNC
is in the state "TX/RX Suspended", incoming SSP connection requests from
the target will be rejected by the controller hardware. When an RNC is
"TX Suspended", it will not be selected by the controller hardware to
start outgoing command or data operations (with certain priority-based
exceptions).
As mentioned above, there are two sources for management of the RNC
states: commands from driver software, and the result of transmission
and reception conditions of commands and data signaled by the controller
hardware. As an example of the latter, if an outgoing SSP command ends
with a OPEN_REJECT(BAD_DESTINATION) status, the RNC state will
transition to the "TX Suspended" state, and this is signaled by the
controller hardware in the status to the completion of the pending
command as well as signaled in a controller hardware event. Examples of
the former are included in the patch changelogs.
Driver software is required to suspend the RNC in a "TX/RX Suspended"
condition before any outstanding commands can be terminated. Failure to
guarantee this can lead to a complete hardware hang condition. Earlier
versions of the driver software did not guarantee that an RNC was
correctly managed before I/O termination, and so operated in an unsafe
way.
Further, the driver performed unnecessary contortions to preserve the
remote device command state and so was more complicated than it needed
to be. A simplifying driver assumption is that once an I/O has entered
the error handler path without having completed in the target, the
requirement on the driver is that all use of the sas_task must end.
Beyond that, recovery of operation is dependent on libsas and other
components to reset, rediscover and reconfigure the device before normal
operation can restart. In the driver, this simplifying assumption meant
that the RNC management could be reduced to entry into the suspended
state, terminating the targeted I/O request, and resuming the RNC as
needed for device-specific management such as an SSP Abort Task or LUN
Reset Management request.
Diffstat (limited to 'drivers/scsi/isci/remote_device.c')
-rw-r--r-- | drivers/scsi/isci/remote_device.c | 576 |
1 files changed, 433 insertions, 143 deletions
diff --git a/drivers/scsi/isci/remote_device.c b/drivers/scsi/isci/remote_device.c index 8f501b0a81d6..c3aa6c5457b9 100644 --- a/drivers/scsi/isci/remote_device.c +++ b/drivers/scsi/isci/remote_device.c @@ -72,46 +72,11 @@ const char *dev_state_name(enum sci_remote_device_states state) } #undef C -/** - * isci_remote_device_not_ready() - This function is called by the ihost when - * the remote device is not ready. We mark the isci device as ready (not - * "ready_for_io") and signal the waiting proccess. - * @isci_host: This parameter specifies the isci host object. - * @isci_device: This parameter specifies the remote device - * - * sci_lock is held on entrance to this function. - */ -static void isci_remote_device_not_ready(struct isci_host *ihost, - struct isci_remote_device *idev, u32 reason) +enum sci_status sci_remote_device_suspend(struct isci_remote_device *idev, + enum sci_remote_node_suspension_reasons reason) { - struct isci_request *ireq; - - dev_dbg(&ihost->pdev->dev, - "%s: isci_device = %p\n", __func__, idev); - - switch (reason) { - case SCIC_REMOTE_DEVICE_NOT_READY_STOP_REQUESTED: - set_bit(IDEV_GONE, &idev->flags); - break; - case SCIC_REMOTE_DEVICE_NOT_READY_SATA_SDB_ERROR_FIS_RECEIVED: - set_bit(IDEV_IO_NCQERROR, &idev->flags); - - /* Kill all outstanding requests for the device. */ - list_for_each_entry(ireq, &idev->reqs_in_process, dev_node) { - - dev_dbg(&ihost->pdev->dev, - "%s: isci_device = %p request = %p\n", - __func__, idev, ireq); - - sci_controller_terminate_request(ihost, - idev, - ireq); - } - /* Fall through into the default case... */ - default: - clear_bit(IDEV_IO_READY, &idev->flags); - break; - } + return sci_remote_node_context_suspend(&idev->rnc, reason, + SCI_SOFTWARE_SUSPEND_EXPECTED_EVENT); } /** @@ -133,18 +98,29 @@ static void isci_remote_device_ready(struct isci_host *ihost, struct isci_remote wake_up(&ihost->eventq); } -/* called once the remote node context is ready to be freed. - * The remote device can now report that its stop operation is complete. none - */ -static void rnc_destruct_done(void *_dev) +static enum sci_status sci_remote_device_terminate_req( + struct isci_host *ihost, + struct isci_remote_device *idev, + int check_abort, + struct isci_request *ireq) { - struct isci_remote_device *idev = _dev; + if (!test_bit(IREQ_ACTIVE, &ireq->flags) || + (ireq->target_device != idev) || + (check_abort && !test_bit(IREQ_PENDING_ABORT, &ireq->flags))) + return SCI_SUCCESS; - BUG_ON(idev->started_request_count != 0); - sci_change_state(&idev->sm, SCI_DEV_STOPPED); + dev_dbg(&ihost->pdev->dev, + "%s: idev=%p; flags=%lx; req=%p; req target=%p\n", + __func__, idev, idev->flags, ireq, ireq->target_device); + + set_bit(IREQ_ABORT_PATH_ACTIVE, &ireq->flags); + + return sci_controller_terminate_request(ihost, idev, ireq); } -static enum sci_status sci_remote_device_terminate_requests(struct isci_remote_device *idev) +static enum sci_status sci_remote_device_terminate_reqs_checkabort( + struct isci_remote_device *idev, + int chk) { struct isci_host *ihost = idev->owning_port->owning_controller; enum sci_status status = SCI_SUCCESS; @@ -154,18 +130,210 @@ static enum sci_status sci_remote_device_terminate_requests(struct isci_remote_d struct isci_request *ireq = ihost->reqs[i]; enum sci_status s; - if (!test_bit(IREQ_ACTIVE, &ireq->flags) || - ireq->target_device != idev) - continue; - - s = sci_controller_terminate_request(ihost, idev, ireq); + s = sci_remote_device_terminate_req(ihost, idev, chk, ireq); if (s != SCI_SUCCESS) status = s; } + return status; +} + +static bool isci_compare_suspendcount( + struct isci_remote_device *idev, + u32 localcount) +{ + smp_rmb(); + + /* Check for a change in the suspend count, or the RNC + * being destroyed. + */ + return (localcount != idev->rnc.suspend_count) + || sci_remote_node_context_is_being_destroyed(&idev->rnc); +} + +static bool isci_check_reqterm( + struct isci_host *ihost, + struct isci_remote_device *idev, + struct isci_request *ireq, + u32 localcount) +{ + unsigned long flags; + bool res; + spin_lock_irqsave(&ihost->scic_lock, flags); + res = isci_compare_suspendcount(idev, localcount) + && !test_bit(IREQ_ABORT_PATH_ACTIVE, &ireq->flags); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + + return res; +} + +static bool isci_check_devempty( + struct isci_host *ihost, + struct isci_remote_device *idev, + u32 localcount) +{ + unsigned long flags; + bool res; + + spin_lock_irqsave(&ihost->scic_lock, flags); + res = isci_compare_suspendcount(idev, localcount) + && idev->started_request_count == 0; + spin_unlock_irqrestore(&ihost->scic_lock, flags); + + return res; +} + +enum sci_status isci_remote_device_terminate_requests( + struct isci_host *ihost, + struct isci_remote_device *idev, + struct isci_request *ireq) +{ + enum sci_status status = SCI_SUCCESS; + unsigned long flags; + u32 rnc_suspend_count; + + spin_lock_irqsave(&ihost->scic_lock, flags); + + if (isci_get_device(idev) == NULL) { + dev_dbg(&ihost->pdev->dev, "%s: failed isci_get_device(idev=%p)\n", + __func__, idev); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + status = SCI_FAILURE; + } else { + /* If already suspended, don't wait for another suspension. */ + smp_rmb(); + rnc_suspend_count + = sci_remote_node_context_is_suspended(&idev->rnc) + ? 0 : idev->rnc.suspend_count; + + dev_dbg(&ihost->pdev->dev, + "%s: idev=%p, ireq=%p; started_request_count=%d, " + "rnc_suspend_count=%d, rnc.suspend_count=%d" + "about to wait\n", + __func__, idev, ireq, idev->started_request_count, + rnc_suspend_count, idev->rnc.suspend_count); + + #define MAX_SUSPEND_MSECS 10000 + if (ireq) { + /* Terminate a specific TC. */ + set_bit(IREQ_NO_AUTO_FREE_TAG, &ireq->flags); + sci_remote_device_terminate_req(ihost, idev, 0, ireq); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + if (!wait_event_timeout(ihost->eventq, + isci_check_reqterm(ihost, idev, ireq, + rnc_suspend_count), + msecs_to_jiffies(MAX_SUSPEND_MSECS))) { + + dev_warn(&ihost->pdev->dev, "%s host%d timeout single\n", + __func__, ihost->id); + dev_dbg(&ihost->pdev->dev, + "%s: ******* Timeout waiting for " + "suspend; idev=%p, current state %s; " + "started_request_count=%d, flags=%lx\n\t" + "rnc_suspend_count=%d, rnc.suspend_count=%d " + "RNC: current state %s, current " + "suspend_type %x dest state %d;\n" + "ireq=%p, ireq->flags = %lx\n", + __func__, idev, + dev_state_name(idev->sm.current_state_id), + idev->started_request_count, idev->flags, + rnc_suspend_count, idev->rnc.suspend_count, + rnc_state_name(idev->rnc.sm.current_state_id), + idev->rnc.suspend_type, + idev->rnc.destination_state, + ireq, ireq->flags); + } + spin_lock_irqsave(&ihost->scic_lock, flags); + clear_bit(IREQ_NO_AUTO_FREE_TAG, &ireq->flags); + if (!test_bit(IREQ_ABORT_PATH_ACTIVE, &ireq->flags)) + isci_free_tag(ihost, ireq->io_tag); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + } else { + /* Terminate all TCs. */ + sci_remote_device_terminate_requests(idev); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + if (!wait_event_timeout(ihost->eventq, + isci_check_devempty(ihost, idev, + rnc_suspend_count), + msecs_to_jiffies(MAX_SUSPEND_MSECS))) { + + dev_warn(&ihost->pdev->dev, "%s host%d timeout all\n", + __func__, ihost->id); + dev_dbg(&ihost->pdev->dev, + "%s: ******* Timeout waiting for " + "suspend; idev=%p, current state %s; " + "started_request_count=%d, flags=%lx\n\t" + "rnc_suspend_count=%d, " + "RNC: current state %s, " + "rnc.suspend_count=%d, current " + "suspend_type %x dest state %d\n", + __func__, idev, + dev_state_name(idev->sm.current_state_id), + idev->started_request_count, idev->flags, + rnc_suspend_count, + rnc_state_name(idev->rnc.sm.current_state_id), + idev->rnc.suspend_count, + idev->rnc.suspend_type, + idev->rnc.destination_state); + } + } + dev_dbg(&ihost->pdev->dev, "%s: idev=%p, wait done\n", + __func__, idev); + isci_put_device(idev); + } return status; } +/** +* isci_remote_device_not_ready() - This function is called by the ihost when +* the remote device is not ready. We mark the isci device as ready (not +* "ready_for_io") and signal the waiting proccess. +* @isci_host: This parameter specifies the isci host object. +* @isci_device: This parameter specifies the remote device +* +* sci_lock is held on entrance to this function. +*/ +static void isci_remote_device_not_ready(struct isci_host *ihost, + struct isci_remote_device *idev, + u32 reason) +{ + dev_dbg(&ihost->pdev->dev, + "%s: isci_device = %p; reason = %d\n", __func__, idev, reason); + + switch (reason) { + case SCIC_REMOTE_DEVICE_NOT_READY_SATA_SDB_ERROR_FIS_RECEIVED: + set_bit(IDEV_IO_NCQERROR, &idev->flags); + + /* Suspend the remote device so the I/O can be terminated. */ + sci_remote_device_suspend(idev, SCI_SW_SUSPEND_NORMAL); + + /* Kill all outstanding requests for the device. */ + sci_remote_device_terminate_requests(idev); + + /* Fall through into the default case... */ + default: + clear_bit(IDEV_IO_READY, &idev->flags); + break; + } +} + +/* called once the remote node context is ready to be freed. + * The remote device can now report that its stop operation is complete. none + */ +static void rnc_destruct_done(void *_dev) +{ + struct isci_remote_device *idev = _dev; + + BUG_ON(idev->started_request_count != 0); + sci_change_state(&idev->sm, SCI_DEV_STOPPED); +} + +enum sci_status sci_remote_device_terminate_requests( + struct isci_remote_device *idev) +{ + return sci_remote_device_terminate_reqs_checkabort(idev, 0); +} + enum sci_status sci_remote_device_stop(struct isci_remote_device *idev, u32 timeout) { @@ -201,13 +369,16 @@ enum sci_status sci_remote_device_stop(struct isci_remote_device *idev, case SCI_SMP_DEV_IDLE: case SCI_SMP_DEV_CMD: sci_change_state(sm, SCI_DEV_STOPPING); - if (idev->started_request_count == 0) { + if (idev->started_request_count == 0) sci_remote_node_context_destruct(&idev->rnc, - rnc_destruct_done, idev); - return SCI_SUCCESS; - } else - return sci_remote_device_terminate_requests(idev); - break; + rnc_destruct_done, + idev); + else { + sci_remote_device_suspend( + idev, SCI_SW_SUSPEND_LINKHANG_DETECT); + sci_remote_device_terminate_requests(idev); + } + return SCI_SUCCESS; case SCI_DEV_STOPPING: /* All requests should have been terminated, but if there is an * attempt to stop a device already in the stopping state, then @@ -265,22 +436,6 @@ enum sci_status sci_remote_device_reset_complete(struct isci_remote_device *idev return SCI_SUCCESS; } -enum sci_status sci_remote_device_suspend(struct isci_remote_device *idev, - u32 suspend_type) -{ - struct sci_base_state_machine *sm = &idev->sm; - enum sci_remote_device_states state = sm->current_state_id; - - if (state != SCI_STP_DEV_CMD) { - dev_warn(scirdev_to_dev(idev), "%s: in wrong state: %s\n", - __func__, dev_state_name(state)); - return SCI_FAILURE_INVALID_STATE; - } - - return sci_remote_node_context_suspend(&idev->rnc, - suspend_type, NULL, NULL); -} - enum sci_status sci_remote_device_frame_handler(struct isci_remote_device *idev, u32 frame_index) { @@ -412,9 +567,9 @@ static void atapi_remote_device_resume_done(void *_dev) enum sci_status sci_remote_device_event_handler(struct isci_remote_device *idev, u32 event_code) { + enum sci_status status; struct sci_base_state_machine *sm = &idev->sm; enum sci_remote_device_states state = sm->current_state_id; - enum sci_status status; switch (scu_get_event_type(event_code)) { case SCU_EVENT_TYPE_RNC_OPS_MISC: @@ -427,9 +582,7 @@ enum sci_status sci_remote_device_event_handler(struct isci_remote_device *idev, status = SCI_SUCCESS; /* Suspend the associated RNC */ - sci_remote_node_context_suspend(&idev->rnc, - SCI_SOFTWARE_SUSPENSION, - NULL, NULL); + sci_remote_device_suspend(idev, SCI_SW_SUSPEND_NORMAL); dev_dbg(scirdev_to_dev(idev), "%s: device: %p event code: %x: %s\n", @@ -455,6 +608,10 @@ enum sci_status sci_remote_device_event_handler(struct isci_remote_device *idev, if (status != SCI_SUCCESS) return status; + /* Decode device-specific states that may require an RNC resume during + * normal operation. When the abort path is active, these resumes are + * managed when the abort path exits. + */ if (state == SCI_STP_DEV_ATAPI_ERROR) { /* For ATAPI error state resume the RNC right away. */ if (scu_get_event_type(event_code) == SCU_EVENT_TYPE_RNC_SUSPEND_TX || @@ -743,10 +900,6 @@ enum sci_status sci_remote_device_start_task(struct isci_host *ihost, if (status != SCI_SUCCESS) return status; - status = sci_remote_node_context_start_task(&idev->rnc, ireq); - if (status != SCI_SUCCESS) - goto out; - status = sci_request_start(ireq); if (status != SCI_SUCCESS) goto out; @@ -765,11 +918,11 @@ enum sci_status sci_remote_device_start_task(struct isci_host *ihost, * the correct action when the remote node context is suspended * and later resumed. */ - sci_remote_node_context_suspend(&idev->rnc, - SCI_SOFTWARE_SUSPENSION, NULL, NULL); - sci_remote_node_context_resume(&idev->rnc, - sci_remote_device_continue_request, - idev); + sci_remote_device_suspend(idev, + SCI_SW_SUSPEND_LINKHANG_DETECT); + + status = sci_remote_node_context_start_task(&idev->rnc, ireq, + sci_remote_device_continue_request, idev); out: sci_remote_device_start_request(idev, ireq, status); @@ -783,7 +936,9 @@ enum sci_status sci_remote_device_start_task(struct isci_host *ihost, if (status != SCI_SUCCESS) return status; - status = sci_remote_node_context_start_task(&idev->rnc, ireq); + /* Resume the RNC as needed: */ + status = sci_remote_node_context_start_task(&idev->rnc, ireq, + NULL, NULL); if (status != SCI_SUCCESS) break; @@ -892,7 +1047,7 @@ static void isci_remote_device_deconstruct(struct isci_host *ihost, struct isci_ * here should go through isci_remote_device_nuke_requests. * If we hit this condition, we will need a way to complete * io requests in process */ - BUG_ON(!list_empty(&idev->reqs_in_process)); + BUG_ON(idev->started_request_count > 0); sci_remote_device_destruct(idev); list_del_init(&idev->node); @@ -954,14 +1109,21 @@ static void sci_remote_device_ready_state_exit(struct sci_base_state_machine *sm static void sci_remote_device_resetting_state_enter(struct sci_base_state_machine *sm) { struct isci_remote_device *idev = container_of(sm, typeof(*idev), sm); + struct isci_host *ihost = idev->owning_port->owning_controller; - sci_remote_node_context_suspend( - &idev->rnc, SCI_SOFTWARE_SUSPENSION, NULL, NULL); + dev_dbg(&ihost->pdev->dev, + "%s: isci_device = %p\n", __func__, idev); + + sci_remote_device_suspend(idev, SCI_SW_SUSPEND_LINKHANG_DETECT); } static void sci_remote_device_resetting_state_exit(struct sci_base_state_machine *sm) { struct isci_remote_device *idev = container_of(sm, typeof(*idev), sm); + struct isci_host *ihost = idev->owning_port->owning_controller; + + dev_dbg(&ihost->pdev->dev, + "%s: isci_device = %p\n", __func__, idev); sci_remote_node_context_resume(&idev->rnc, NULL, NULL); } @@ -1113,33 +1275,20 @@ static enum sci_status sci_remote_device_da_construct(struct isci_port *iport, { enum sci_status status; struct sci_port_properties properties; - struct domain_device *dev = idev->domain_dev; sci_remote_device_construct(iport, idev); - /* - * This information is request to determine how many remote node context - * entries will be needed to store the remote node. - */ - idev->is_direct_attached = true; - sci_port_get_properties(iport, &properties); /* Get accurate port width from port's phy mask for a DA device. */ idev->device_port_width = hweight32(properties.phy_mask); status = sci_controller_allocate_remote_node_context(iport->owning_controller, - idev, - &idev->rnc.remote_node_index); + idev, + &idev->rnc.remote_node_index); if (status != SCI_SUCCESS) return status; - if (dev->dev_type == SAS_END_DEV || dev->dev_type == SATA_DEV || - (dev->tproto & SAS_PROTOCOL_STP) || dev_is_expander(dev)) - /* pass */; - else - return SCI_FAILURE_UNSUPPORTED_PROTOCOL; - idev->connection_rate = sci_port_get_max_allowed_speed(iport); return SCI_SUCCESS; @@ -1171,19 +1320,13 @@ static enum sci_status sci_remote_device_ea_construct(struct isci_port *iport, if (status != SCI_SUCCESS) return status; - if (dev->dev_type == SAS_END_DEV || dev->dev_type == SATA_DEV || - (dev->tproto & SAS_PROTOCOL_STP) || dev_is_expander(dev)) - /* pass */; - else - return SCI_FAILURE_UNSUPPORTED_PROTOCOL; - - /* - * For SAS-2 the physical link rate is actually a logical link + /* For SAS-2 the physical link rate is actually a logical link * rate that incorporates multiplexing. The SCU doesn't * incorporate multiplexing and for the purposes of the * connection the logical link rate is that same as the * physical. Furthermore, the SAS-2 and SAS-1.1 fields overlay - * one another, so this code works for both situations. */ + * one another, so this code works for both situations. + */ idev->connection_rate = min_t(u16, sci_port_get_max_allowed_speed(iport), dev->linkrate); @@ -1193,6 +1336,105 @@ static enum sci_status sci_remote_device_ea_construct(struct isci_port *iport, return SCI_SUCCESS; } +enum sci_status sci_remote_device_resume( + struct isci_remote_device *idev, + scics_sds_remote_node_context_callback cb_fn, + void *cb_p) +{ + enum sci_status status; + + status = sci_remote_node_context_resume(&idev->rnc, cb_fn, cb_p); + if (status != SCI_SUCCESS) + dev_dbg(scirdev_to_dev(idev), "%s: failed to resume: %d\n", + __func__, status); + return status; +} + +static void isci_remote_device_resume_from_abort_complete(void *cbparam) +{ + struct isci_remote_device *idev = cbparam; + struct isci_host *ihost = idev->owning_port->owning_controller; + scics_sds_remote_node_context_callback abort_resume_cb = + idev->abort_resume_cb; + + dev_dbg(scirdev_to_dev(idev), "%s: passing-along resume: %p\n", + __func__, abort_resume_cb); + + if (abort_resume_cb != NULL) { + idev->abort_resume_cb = NULL; + abort_resume_cb(idev->abort_resume_cbparam); + } + clear_bit(IDEV_ABORT_PATH_RESUME_PENDING, &idev->flags); + wake_up(&ihost->eventq); +} + +static bool isci_remote_device_test_resume_done( + struct isci_host *ihost, + struct isci_remote_device *idev) +{ + unsigned long flags; + bool done; + + spin_lock_irqsave(&ihost->scic_lock, flags); + done = !test_bit(IDEV_ABORT_PATH_RESUME_PENDING, &idev->flags) + || test_bit(IDEV_STOP_PENDING, &idev->flags) + || sci_remote_node_context_is_being_destroyed(&idev->rnc); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + + return done; +} + +void isci_remote_device_wait_for_resume_from_abort( + struct isci_host *ihost, + struct isci_remote_device *idev) +{ + dev_dbg(&ihost->pdev->dev, "%s: starting resume wait: %p\n", + __func__, idev); + + #define MAX_RESUME_MSECS 10000 + if (!wait_event_timeout(ihost->eventq, + isci_remote_device_test_resume_done(ihost, idev), + msecs_to_jiffies(MAX_RESUME_MSECS))) { + + dev_warn(&ihost->pdev->dev, "%s: #### Timeout waiting for " + "resume: %p\n", __func__, idev); + } + clear_bit(IDEV_ABORT_PATH_RESUME_PENDING, &idev->flags); + + dev_dbg(&ihost->pdev->dev, "%s: resume wait done: %p\n", + __func__, idev); +} + +enum sci_status isci_remote_device_resume_from_abort( + struct isci_host *ihost, + struct isci_remote_device *idev) +{ + unsigned long flags; + enum sci_status status = SCI_SUCCESS; + int destroyed; + + spin_lock_irqsave(&ihost->scic_lock, flags); + /* Preserve any current resume callbacks, for instance from other + * resumptions. + */ + idev->abort_resume_cb = idev->rnc.user_callback; + idev->abort_resume_cbparam = idev->rnc.user_cookie; + set_bit(IDEV_ABORT_PATH_RESUME_PENDING, &idev->flags); + clear_bit(IDEV_ABORT_PATH_ACTIVE, &idev->flags); + destroyed = sci_remote_node_context_is_being_destroyed(&idev->rnc); + if (!destroyed) + status = sci_remote_device_resume( + idev, isci_remote_device_resume_from_abort_complete, + idev); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + if (!destroyed && (status == SCI_SUCCESS)) + isci_remote_device_wait_for_resume_from_abort(ihost, idev); + else + clear_bit(IDEV_ABORT_PATH_RESUME_PENDING, &idev->flags); + + return status; +} + /** * sci_remote_device_start() - This method will start the supplied remote * device. This method enables normal IO requests to flow through to the @@ -1207,7 +1449,7 @@ static enum sci_status sci_remote_device_ea_construct(struct isci_port *iport, * the device when there have been no phys added to it. */ static enum sci_status sci_remote_device_start(struct isci_remote_device *idev, - u32 timeout) + u32 timeout) { struct sci_base_state_machine *sm = &idev->sm; enum sci_remote_device_states state = sm->current_state_id; @@ -1219,9 +1461,8 @@ static enum sci_status sci_remote_device_start(struct isci_remote_device *idev, return SCI_FAILURE_INVALID_STATE; } - status = sci_remote_node_context_resume(&idev->rnc, - remote_device_resume_done, - idev); + status = sci_remote_device_resume(idev, remote_device_resume_done, + idev); if (status != SCI_SUCCESS) return status; @@ -1259,20 +1500,6 @@ static enum sci_status isci_remote_device_construct(struct isci_port *iport, return status; } -void isci_remote_device_nuke_requests(struct isci_host *ihost, struct isci_remote_device *idev) -{ - DECLARE_COMPLETION_ONSTACK(aborted_task_completion); - - dev_dbg(&ihost->pdev->dev, - "%s: idev = %p\n", __func__, idev); - - /* Cleanup all requests pending for this device. */ - isci_terminate_pending_requests(ihost, idev); - - dev_dbg(&ihost->pdev->dev, - "%s: idev = %p, done\n", __func__, idev); -} - /** * This function builds the isci_remote_device when a libsas dev_found message * is received. @@ -1297,10 +1524,6 @@ isci_remote_device_alloc(struct isci_host *ihost, struct isci_port *iport) dev_warn(&ihost->pdev->dev, "%s: failed\n", __func__); return NULL; } - - if (WARN_ONCE(!list_empty(&idev->reqs_in_process), "found requests in process\n")) - return NULL; - if (WARN_ONCE(!list_empty(&idev->node), "found non-idle remote device\n")) return NULL; @@ -1342,14 +1565,8 @@ enum sci_status isci_remote_device_stop(struct isci_host *ihost, struct isci_rem spin_lock_irqsave(&ihost->scic_lock, flags); idev->domain_dev->lldd_dev = NULL; /* disable new lookups */ set_bit(IDEV_GONE, &idev->flags); - spin_unlock_irqrestore(&ihost->scic_lock, flags); - - /* Kill all outstanding requests. */ - isci_remote_device_nuke_requests(ihost, idev); set_bit(IDEV_STOP_PENDING, &idev->flags); - - spin_lock_irqsave(&ihost->scic_lock, flags); status = sci_remote_device_stop(idev, 50); spin_unlock_irqrestore(&ihost->scic_lock, flags); @@ -1359,6 +1576,9 @@ enum sci_status isci_remote_device_stop(struct isci_host *ihost, struct isci_rem else wait_for_device_stop(ihost, idev); + dev_dbg(&ihost->pdev->dev, + "%s: isci_device = %p, waiting done.\n", __func__, idev); + return status; } @@ -1434,3 +1654,73 @@ int isci_remote_device_found(struct domain_device *dev) return status == SCI_SUCCESS ? 0 : -ENODEV; } + +enum sci_status isci_remote_device_suspend_terminate( + struct isci_host *ihost, + struct isci_remote_device *idev, + struct isci_request *ireq) +{ + unsigned long flags; + enum sci_status status; + + /* Put the device into suspension. */ + spin_lock_irqsave(&ihost->scic_lock, flags); + set_bit(IDEV_ABORT_PATH_ACTIVE, &idev->flags); + sci_remote_device_suspend(idev, SCI_SW_SUSPEND_LINKHANG_DETECT); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + + /* Terminate and wait for the completions. */ + status = isci_remote_device_terminate_requests(ihost, idev, ireq); + if (status != SCI_SUCCESS) + dev_dbg(&ihost->pdev->dev, + "%s: isci_remote_device_terminate_requests(%p) " + "returned %d!\n", + __func__, idev, status); + + /* NOTE: RNC resumption is left to the caller! */ + return status; +} + +int isci_remote_device_is_safe_to_abort( + struct isci_remote_device *idev) +{ + return sci_remote_node_context_is_safe_to_abort(&idev->rnc); +} + +enum sci_status sci_remote_device_abort_requests_pending_abort( + struct isci_remote_device *idev) +{ + return sci_remote_device_terminate_reqs_checkabort(idev, 1); +} + +enum sci_status isci_remote_device_reset_complete( + struct isci_host *ihost, + struct isci_remote_device *idev) +{ + unsigned long flags; + enum sci_status status; + + spin_lock_irqsave(&ihost->scic_lock, flags); + status = sci_remote_device_reset_complete(idev); + spin_unlock_irqrestore(&ihost->scic_lock, flags); + + return status; +} + +void isci_dev_set_hang_detection_timeout( + struct isci_remote_device *idev, + u32 timeout) +{ + if (dev_is_sata(idev->domain_dev)) { + if (timeout) { + if (test_and_set_bit(IDEV_RNC_LLHANG_ENABLED, + &idev->flags)) + return; /* Already enabled. */ + } else if (!test_and_clear_bit(IDEV_RNC_LLHANG_ENABLED, + &idev->flags)) + return; /* Not enabled. */ + + sci_port_set_hang_detection_timeout(idev->owning_port, + timeout); + } +} |