On 1/3/21 6:11 PM, James Smart wrote:
This patch continues the efct driver population.
This patch adds driver definitions for:
Routines to detach xport and hardware objects.
Co-developed-by: Ram Vegesna <ram.vegesna@xxxxxxxxxxxx>
Signed-off-by: Ram Vegesna <ram.vegesna@xxxxxxxxxxxx>
Signed-off-by: James Smart <jsmart2021@xxxxxxxxx>
---
drivers/scsi/elx/efct/efct_hw.c | 275 +++++++++++++++++++++++++++++
drivers/scsi/elx/efct/efct_hw.h | 27 +++
drivers/scsi/elx/efct/efct_xport.c | 261 +++++++++++++++++++++++++++
3 files changed, 563 insertions(+)
diff --git a/drivers/scsi/elx/efct/efct_hw.c b/drivers/scsi/elx/efct/efct_hw.c
index a2d60a837329..683f1d29a155 100644
--- a/drivers/scsi/elx/efct/efct_hw.c
+++ b/drivers/scsi/elx/efct/efct_hw.c
@@ -3356,3 +3356,278 @@ efct_hw_firmware_write(struct efct_hw *hw, struct efc_dma *dma, u32 size,
return rc;
}
+
+static int
+efct_hw_cb_port_control(struct efct_hw *hw, int status, u8 *mqe,
+ void *arg)
+{
+ return EFC_SUCCESS;
+}
+
+enum efct_hw_rtn
+efct_hw_port_control(struct efct_hw *hw, enum efct_hw_port ctrl,
+ uintptr_t value,
+ void (*cb)(int status, uintptr_t value, void *arg),
+ void *arg)
+{
+ enum efct_hw_rtn rc = EFCT_HW_RTN_ERROR;
+ u8 link[SLI4_BMBX_SIZE];
+ u32 speed = 0;
+ u8 reset_alpa = 0;
+
+ switch (ctrl) {
+ case EFCT_HW_PORT_INIT:
+ if (!sli_cmd_config_link(&hw->sli, link))
+ rc = efct_hw_command(hw, link, EFCT_CMD_NOWAIT,
+ efct_hw_cb_port_control, NULL);
+
+ if (rc != EFCT_HW_RTN_SUCCESS) {
+ efc_log_err(hw->os, "CONFIG_LINK failed\n");
+ break;
+ }
+ speed = hw->config.speed;
+ reset_alpa = (u8)(value & 0xff);
+
+ rc = EFCT_HW_RTN_ERROR;
+ if (!sli_cmd_init_link(&hw->sli, link, speed, reset_alpa))
+ rc = efct_hw_command(hw, link, EFCT_CMD_NOWAIT,
+ efct_hw_cb_port_control, NULL);
+ /* Free buffer on error, since no callback is coming */
+ if (rc)
+ efc_log_err(hw->os, "INIT_LINK failed\n");
+ break;
+
+ case EFCT_HW_PORT_SHUTDOWN:
+ if (!sli_cmd_down_link(&hw->sli, link))
+ rc = efct_hw_command(hw, link, EFCT_CMD_NOWAIT,
+ efct_hw_cb_port_control, NULL);
+ /* Free buffer on error, since no callback is coming */
+ if (rc)
+ efc_log_err(hw->os, "DOWN_LINK failed\n");
+ break;
+
+ default:
+ efc_log_debug(hw->os, "unhandled control %#x\n", ctrl);
+ break;
+ }
+
+ return rc;
+}
+
+enum efct_hw_rtn
+efct_hw_teardown(struct efct_hw *hw)
+{
+ u32 i = 0;
+ u32 iters = 10;
+ u32 destroy_queues;
+ u32 free_memory;
+ struct efc_dma *dma;
+ struct efct *efct = hw->os;
+
+ destroy_queues = (hw->state == EFCT_HW_STATE_ACTIVE);
+ free_memory = (hw->state != EFCT_HW_STATE_UNINITIALIZED);
+
+ /* Cancel Sliport Healthcheck */
+ if (hw->sliport_healthcheck) {
+ hw->sliport_healthcheck = 0;
+ efct_hw_config_sli_port_health_check(hw, 0, 0);
+ }
+
+ if (hw->state != EFCT_HW_STATE_QUEUES_ALLOCATED) {
+ hw->state = EFCT_HW_STATE_TEARDOWN_IN_PROGRESS;
+
+ efct_hw_flush(hw);
+
+ /*
+ * If there are outstanding commands, wait for them to complete
+ */
+ while (!list_empty(&hw->cmd_head) && iters) {
+ mdelay(10);
+ efct_hw_flush(hw);
+ iters--;
+ }
+
+ if (list_empty(&hw->cmd_head))
+ efc_log_debug(hw->os,
+ "All commands completed on MQ queue\n");
+ else
+ efc_log_debug(hw->os,
+ "Some cmds still pending on MQ queue\n");
+
+ /* Cancel any remaining commands */
+ efct_hw_command_cancel(hw);
+ } else {
+ hw->state = EFCT_HW_STATE_TEARDOWN_IN_PROGRESS;
+ }
+
+ dma_free_coherent(&efct->pci->dev,
+ hw->rnode_mem.size, hw->rnode_mem.virt,
+ hw->rnode_mem.phys);
+ memset(&hw->rnode_mem, 0, sizeof(struct efc_dma));
+
+ if (hw->io) {
+ for (i = 0; i < hw->config.n_io; i++) {
+ if (hw->io[i] && hw->io[i]->sgl &&
+ hw->io[i]->sgl->virt) {
+ dma_free_coherent(&efct->pci->dev,
+ hw->io[i]->sgl->size,
+ hw->io[i]->sgl->virt,
+ hw->io[i]->sgl->phys);
+ }
+ kfree(hw->io[i]);
+ hw->io[i] = NULL;
+ }
+ kfree(hw->io);
+ hw->io = NULL;
+ kfree(hw->wqe_buffs);
+ hw->wqe_buffs = NULL;
+ }
+
+ dma = &hw->xfer_rdy;
+ dma_free_coherent(&efct->pci->dev,
+ dma->size, dma->virt, dma->phys);
+ memset(dma, 0, sizeof(struct efc_dma));
+
+ dma = &hw->loop_map;
+ dma_free_coherent(&efct->pci->dev,
+ dma->size, dma->virt, dma->phys);
+ memset(dma, 0, sizeof(struct efc_dma));
+
+ for (i = 0; i < hw->wq_count; i++)
+ sli_queue_free(&hw->sli, &hw->wq[i], destroy_queues,
+ free_memory);
+
+ for (i = 0; i < hw->rq_count; i++)
+ sli_queue_free(&hw->sli, &hw->rq[i], destroy_queues,
+ free_memory);
+
+ for (i = 0; i < hw->mq_count; i++)
+ sli_queue_free(&hw->sli, &hw->mq[i], destroy_queues,
+ free_memory);
+
+ for (i = 0; i < hw->cq_count; i++)
+ sli_queue_free(&hw->sli, &hw->cq[i], destroy_queues,
+ free_memory);
+
+ for (i = 0; i < hw->eq_count; i++)
+ sli_queue_free(&hw->sli, &hw->eq[i], destroy_queues,
+ free_memory);
+
+ /* Free rq buffers */
+ efct_hw_rx_free(hw);
+
+ efct_hw_queue_teardown(hw);
+
+ sli_teardown(&hw->sli);
+
+ /* record the fact that the queues are non-functional */
+ hw->state = EFCT_HW_STATE_UNINITIALIZED;
+
+ /* free sequence free pool */
+ kfree(hw->seq_pool);
+ hw->seq_pool = NULL;
+
+ /* free hw_wq_callback pool */
+ efct_hw_reqtag_pool_free(hw);
+
+ mempool_destroy(hw->cmd_ctx_pool);
+ mempool_destroy(hw->mbox_rqst_pool);
+
+ /* Mark HW setup as not having been called */
+ hw->hw_setup_called = false;
+
+ return EFCT_HW_RTN_SUCCESS;
+}
The point of this return value is what exactly?
Please make this a void function.
+
+static enum efct_hw_rtn
+efct_hw_sli_reset(struct efct_hw *hw, enum efct_hw_reset reset,
+ enum efct_hw_state prev_state)
+{
+ enum efct_hw_rtn rc = EFCT_HW_RTN_SUCCESS;
+
+ switch (reset) {
+ case EFCT_HW_RESET_FUNCTION:
+ efc_log_debug(hw->os, "issuing function level reset\n");
+ if (sli_reset(&hw->sli)) {
+ efc_log_err(hw->os, "sli_reset failed\n");
+ rc = EFCT_HW_RTN_ERROR;
+ }
+ break;
+ case EFCT_HW_RESET_FIRMWARE:
+ efc_log_debug(hw->os, "issuing firmware reset\n");
+ if (sli_fw_reset(&hw->sli)) {
+ efc_log_err(hw->os, "sli_soft_reset failed\n");
+ rc = EFCT_HW_RTN_ERROR;
+ }
+ /*
+ * Because the FW reset leaves the FW in a non-running state,
+ * follow that with a regular reset.
+ */
+ efc_log_debug(hw->os, "issuing function level reset\n");
+ if (sli_reset(&hw->sli)) {
+ efc_log_err(hw->os, "sli_reset failed\n");
+ rc = EFCT_HW_RTN_ERROR;
+ }
+ break;
+ default:
+ efc_log_err(hw->os,
+ "unknown reset type - no reset performed\n");
+ hw->state = prev_state;
+ rc = EFCT_HW_RTN_INVALID_ARG;
+ break;
+ }
+
+ return rc;
+}
+
+enum efct_hw_rtn
+efct_hw_reset(struct efct_hw *hw, enum efct_hw_reset reset)
+{
+ enum efct_hw_rtn rc = EFCT_HW_RTN_SUCCESS;
+ u32 iters;
+ enum efct_hw_state prev_state = hw->state;
+
+ if (hw->state != EFCT_HW_STATE_ACTIVE)
+ efc_log_debug(hw->os,
+ "HW state %d is not active\n", hw->state);
+
+ hw->state = EFCT_HW_STATE_RESET_IN_PROGRESS;
+
+ /*
+ * If the prev_state is already reset/teardown in progress,
+ * don't continue further
+ */
+ if (prev_state == EFCT_HW_STATE_RESET_IN_PROGRESS ||
+ prev_state == EFCT_HW_STATE_TEARDOWN_IN_PROGRESS)
+ return efct_hw_sli_reset(hw, reset, prev_state);
+
+ if (prev_state != EFCT_HW_STATE_UNINITIALIZED) {
+ efct_hw_flush(hw);
+
+ /*
+ * If an mailbox command requiring a DMA is outstanding
+ * (SFP/DDM), then the FW will UE when the reset is issued.
+ * So attempt to complete all mailbox commands.
+ */
+ iters = 10;
+ while (!list_empty(&hw->cmd_head) && iters) {
+ mdelay(10);
+ efct_hw_flush(hw);
+ iters--;
+ }
+
+ if (list_empty(&hw->cmd_head))
+ efc_log_debug(hw->os,
+ "All commands completed on MQ queue\n");
+ else
+ efc_log_debug(hw->os,
+ "Some commands still pending on MQ queue\n");
A-ha. Doesn't UE mean the chip is unavailable and needs to reset?
But don't we reset the chip _anyway_ (it's the reset routine, after
all). So why do we need to wait for pending commands if we're continue
with pending commands?
+ }
+
+ /* Reset the chip */
+ rc = efct_hw_sli_reset(hw, reset, prev_state);
+ if (rc == EFCT_HW_RTN_INVALID_ARG)
+ return EFCT_HW_RTN_ERROR;
+
+ return rc;
+}
... begging the question what happens if we fail here ...
HW done in?
diff --git a/drivers/scsi/elx/efct/efct_hw.h b/drivers/scsi/elx/efct/efct_hw.h
index 457ee065a7ca..a8ed3cef71dc 100644
--- a/drivers/scsi/elx/efct/efct_hw.h
+++ b/drivers/scsi/elx/efct/efct_hw.h
@@ -752,4 +752,31 @@ typedef void (*efct_hw_async_cb_t)(struct efct_hw *hw, int status,
int
efct_hw_async_call(struct efct_hw *hw, efct_hw_async_cb_t callback, void *arg);
+struct hw_eq *efct_hw_new_eq(struct efct_hw *hw, u32 entry_count);
+struct hw_cq *efct_hw_new_cq(struct hw_eq *eq, u32 entry_count);
+u32
+efct_hw_new_cq_set(struct hw_eq *eqs[], struct hw_cq *cqs[],
+ u32 num_cqs, u32 entry_count);
+struct hw_mq *efct_hw_new_mq(struct hw_cq *cq, u32 entry_count);
+struct hw_wq
+*efct_hw_new_wq(struct hw_cq *cq, u32 entry_count);
+u32
+efct_hw_new_rq_set(struct hw_cq *cqs[], struct hw_rq *rqs[],
+ u32 num_rq_pairs, u32 entry_count);
+void efct_hw_del_eq(struct hw_eq *eq);
+void efct_hw_del_cq(struct hw_cq *cq);
+void efct_hw_del_mq(struct hw_mq *mq);
+void efct_hw_del_wq(struct hw_wq *wq);
+void efct_hw_del_rq(struct hw_rq *rq);
+void efct_hw_queue_teardown(struct efct_hw *hw);
+enum efct_hw_rtn efct_hw_teardown(struct efct_hw *hw);
+enum efct_hw_rtn
+efct_hw_reset(struct efct_hw *hw, enum efct_hw_reset reset);
+
+enum efct_hw_rtn
+efct_hw_port_control(struct efct_hw *hw, enum efct_hw_port ctrl,
+ uintptr_t value,
+ void (*cb)(int status, uintptr_t value, void *arg),
+ void *arg);
+
#endif /* __EFCT_H__ */
diff --git a/drivers/scsi/elx/efct/efct_xport.c b/drivers/scsi/elx/efct/efct_xport.c
index b19304d6febb..0cf558c03547 100644
--- a/drivers/scsi/elx/efct/efct_xport.c
+++ b/drivers/scsi/elx/efct/efct_xport.c
@@ -517,3 +517,264 @@ efct_scsi_release_fc_transport(void)
return EFC_SUCCESS;
}
+
+int
+efct_xport_detach(struct efct_xport *xport)
+{
+ struct efct *efct = xport->efct;
+
+ /* free resources associated with target-server and initiator-client */
+ efct_scsi_tgt_del_device(efct);
+
+ efct_scsi_del_device(efct);
+
+ /*Shutdown FC Statistics timer*/
+ if (timer_pending(&xport->stats_timer))
+ del_timer(&xport->stats_timer);
+
+ efct_hw_teardown(&efct->hw);
+
+ efct_xport_delete_debugfs(efct);
+
+ return EFC_SUCCESS;
+}
+
Same here. If we always return SUCCESS we should make it a void function.
+static void
+efct_xport_domain_free_cb(struct efc *efc, void *arg)
+{
+ struct completion *done = arg;
+
+ complete(done);
+}
+
+static void
+efct_xport_post_node_event_cb(struct efct_hw *hw, int status,
+ u8 *mqe, void *arg)
+{
+ struct efct_xport_post_node_event *payload = arg;
+
+ if (payload) {
+ efc_node_post_shutdown(payload->node, payload->evt,
+ payload->context);
+ complete(&payload->done);
+ if (atomic_sub_and_test(1, &payload->refcnt))
+ kfree(payload);
+ }
+}
+
+int
+efct_xport_control(struct efct_xport *xport, enum efct_xport_ctrl cmd, ...)
+{
Why is this an 'int' function?
Shouldn't it be using the standard calling convention?
+ u32 rc = 0;
+ struct efct *efct = NULL;
+ va_list argp;
+
+ efct = xport->efct;
+
+ switch (cmd) {
+ case EFCT_XPORT_PORT_ONLINE: {
+ /* Bring the port on-line */
+ rc = efct_hw_port_control(&efct->hw, EFCT_HW_PORT_INIT, 0,
+ NULL, NULL);
+ if (rc)
+ efc_log_err(efct,
+ "%s: Can't init port\n", efct->desc);
+ else
+ xport->configured_link_state = cmd;
+ break;
+ }
+ case EFCT_XPORT_PORT_OFFLINE: {
+ if (efct_hw_port_control(&efct->hw, EFCT_HW_PORT_SHUTDOWN, 0,
+ NULL, NULL))
+ efc_log_err(efct, "port shutdown failed\n");
+ else
+ xport->configured_link_state = cmd;
+ break;
+ }
+
+ case EFCT_XPORT_SHUTDOWN: {
+ struct completion done;
+ unsigned long timeout;
+
+ /* if a PHYSDEV reset was performed (e.g. hw dump), will affect
+ * all PCI functions; orderly shutdown won't work,
+ * just force free
+ */
+ if (sli_reset_required(&efct->hw.sli)) {
+ struct efc_domain *domain = efct->efcport->domain;
+
+ if (domain)
+ efc_domain_cb(efct->efcport, EFC_HW_DOMAIN_LOST,
+ domain);
+ } else {
+ efct_hw_port_control(&efct->hw, EFCT_HW_PORT_SHUTDOWN,
+ 0, NULL, NULL);
+ }
+
+ init_completion(&done);
+
+ efc_register_domain_free_cb(efct->efcport,
+ efct_xport_domain_free_cb, &done);
+
+ efc_log_debug(efct, "Waiting %d seconds for domain shutdown\n",
+ (EFC_SHUTDOWN_TIMEOUT_USEC / 1000000));
+
+ timeout = usecs_to_jiffies(EFC_SHUTDOWN_TIMEOUT_USEC);
+ if (!wait_for_completion_timeout(&done, timeout)) {
+ efc_log_err(efct, "Domain shutdown timed out!!\n");
+ WARN_ON(1);
+ }
+
+ efc_register_domain_free_cb(efct->efcport, NULL, NULL);
+
+ /* Free up any saved virtual ports */
+ efc_vport_del_all(efct->efcport);
+ break;
+ }
+
+ /*
+ * POST_NODE_EVENT: post an event to a node object
+ *
+ * This transport function is used to post an event to a node object.
+ * It does this by submitting a NOP mailbox command to defer execution
+ * to the interrupt context (thereby enforcing the serialized execution
+ * of event posting to the node state machine instances)
+ */
+ case EFCT_XPORT_POST_NODE_EVENT: {
+ struct efc_node *node;
+ u32 evt;
+ void *context;
+ struct efct_xport_post_node_event *payload = NULL;
+ struct efct_hw *hw;
+
+ /* Retrieve arguments */
+ va_start(argp, cmd);
+ node = va_arg(argp, struct efc_node *);
+ evt = va_arg(argp, u32);
+ context = va_arg(argp, void *);
+ va_end(argp);
+
+ payload = kzalloc(sizeof(*payload), GFP_KERNEL);
+ if (!payload)
+ return EFC_FAIL;
And if it needs to be an 'int' function: why do we return EFC_FAIL here?
+
+ hw = &efct->hw;
+
+ /* if node's state machine is disabled,
+ * don't bother continuing
+ */
+ if (!node->sm.current_state) {
+ efc_log_debug(efct, "node %p state machine disabled\n",
+ node);
+ kfree(payload);
+ rc = -1;
See? Here we _can_ return -1 ...
+ break;
+ }
+
+ /* Setup payload */
+ init_completion(&payload->done);
+
+ /* one for self and one for callback */
+ atomic_set(&payload->refcnt, 2);
+ payload->node = node;
+ payload->evt = evt;
+ payload->context = context;
+
+ if (efct_hw_async_call(hw, efct_xport_post_node_event_cb,
+ payload)) {
+ efc_log_debug(efct, "efct_hw_async_call failed\n");
+ kfree(payload);
+ rc = -1;
+ break;
+ }
+
+ /* Wait for completion */
+ if (wait_for_completion_interruptible(&payload->done)) {
+ efc_log_debug(efct,
+ "POST_NODE_EVENT: completion failed\n");
+ rc = -1;
+ }
+ if (atomic_sub_and_test(1, &payload->refcnt))
+ kfree(payload);
+
+ break;
+ }
+ /*
+ * Set wwnn for the port. This will be used instead of the default
+ * provided by FW.
+ */
+ case EFCT_XPORT_WWNN_SET: {
+ u64 wwnn;
+
+ /* Retrieve arguments */
+ va_start(argp, cmd);
+ wwnn = va_arg(argp, uint64_t);
+ va_end(argp);
+
+ efc_log_debug(efct, " WWNN %016llx\n", wwnn);
+ xport->req_wwnn = wwnn;
+
+ break;
+ }
+ /*
+ * Set wwpn for the port. This will be used instead of the default
+ * provided by FW.
+ */
+ case EFCT_XPORT_WWPN_SET: {
+ u64 wwpn;
+
+ /* Retrieve arguments */
+ va_start(argp, cmd);
+ wwpn = va_arg(argp, uint64_t);
+ va_end(argp);
+
+ efc_log_debug(efct, " WWPN %016llx\n", wwpn);
+ xport->req_wwpn = wwpn;
+
+ break;
+ }
+
+ default:
+ break;
+ }
+ return rc;
+}
+
+void
+efct_xport_free(struct efct_xport *xport)
+{
+ if (xport) {
+ efct_io_pool_free(xport->io_pool);
+
+ kfree(xport);
+ }
+}
+
+void
+efct_release_fc_transport(struct scsi_transport_template *transport_template)
+{
+ if (transport_template)
+ pr_err("releasing transport layer\n");
+
+ /* Releasing FC transport */
+ fc_release_transport(transport_template);
+}
+
+static void
+efct_xport_remove_host(struct Scsi_Host *shost)
+{
+ fc_remove_host(shost);
+}
+
+int efct_scsi_del_device(struct efct *efct)
+{
+ if (efct->shost) {
+ efc_log_debug(efct, "Unregistering with Transport Layer\n");
+ efct_xport_remove_host(efct->shost);
+ efc_log_debug(efct, "Unregistering with SCSI Midlayer\n");
+ scsi_remove_host(efct->shost);
+ scsi_host_put(efct->shost);
+ efct->shost = NULL;
+ }
+ return EFC_SUCCESS;
+}
And again, make this a void function.
Cheers,
Hannes
--
Dr. Hannes Reinecke Kernel Storage Architect
hare@xxxxxxx +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), Geschäftsführer: Felix Imendörffer
