From patchwork Tue Jun 1 23:55:09 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: James Smart X-Patchwork-Id: 451687 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-15.8 required=3.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI, SPF_HELO_NONE, SPF_PASS, USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 56754C47093 for ; Tue, 1 Jun 2021 23:55:55 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 426C661351 for ; Tue, 1 Jun 2021 23:55:55 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S235226AbhFAX5g (ORCPT ); Tue, 1 Jun 2021 19:57:36 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:40460 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S235286AbhFAX5Y (ORCPT ); Tue, 1 Jun 2021 19:57:24 -0400 Received: from mail-pf1-x42a.google.com (mail-pf1-x42a.google.com [IPv6:2607:f8b0:4864:20::42a]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 1C1ECC06174A for ; Tue, 1 Jun 2021 16:55:41 -0700 (PDT) Received: by mail-pf1-x42a.google.com with SMTP id u18so738865pfk.11 for ; Tue, 01 Jun 2021 16:55:41 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=tvpPt6Ysdbtec21kJeIyQWGvujsd9n1+qPg+86e9Fug=; b=sW46V0xxHP6gVlJgoQRoOPj+WvX5Srb97vvH3aH1Fj/Iy3rwwsDnJu+zufyWqYobNE 9uAQHN6+GfBJwh+EHrOqRFS1BYn9t42xEJty5OkcE0UvJ4oHyve/VSc2jN4ZxxBqoJw9 Z2j040CVtJ+uJ6RAzUnb7qlgCgxMMYOVHC6FiPrgVFtjWZqZOrTu4UVYrLI6DClp/Xxq GjPA905ZYkL440sdnPfdSeVWuyv2IFXDRHvlbnIudaKnPXnrPfKSrxF++gJWEtGIt1Cm znINn3v4ZzusMsFAcySxu2HlaWxbzUuWuGOsoBQmyloeXTEdpIvhffpoLf6Vw+yZkUhr lTew== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=tvpPt6Ysdbtec21kJeIyQWGvujsd9n1+qPg+86e9Fug=; b=GabHheQuHOPzkkLaqdR/vJyHp6HufhlAuFk5dlifZMTKZMt1v7eVLYb6N2bORG7fN7 /+6mhCOtHCtZfif+UIl/kQJ5mkhe/sgUN8TJZtZLqXf2nTzCDaPkQYpdaI6NXO67aAf9 gIUWynIlQeO+otTaW8OdO/d1AWl015LdPBxrTo8nl/MJfz02q3jfuRpaA0HsFfS13gP7 2t9cYAzCk6Q7DyKKExZTTIxw2Lwj/FJiBt8txcmJyy4H1iwAQsGsoh15jRZKlU5B62JH Xq56ZbsibVb7rzvDEct0PxGfrNIBX/fCT1S24KL/MRRcL+bc4DsR6AVuRbt0WLv/E+Jd cYfQ== X-Gm-Message-State: AOAM533TjO66TYJBIgGGCjCWWPHv12hgXlMBM+WK5Fw1nR+bOwSlz2Ks amlPeiDcA7Vb+xsGvEGugnVcXWX9aYE= X-Google-Smtp-Source: ABdhPJx74Nq1ZfRCxeY9a8+39FxNmvCxrtUIOfZyEfxOceA/lwlFyPXQWbsxpUg5iHLWjT7mAh/+EA== X-Received: by 2002:a63:4b43:: with SMTP id k3mr30886194pgl.450.1622591740344; Tue, 01 Jun 2021 16:55:40 -0700 (PDT) Received: from localhost.localdomain ([192.19.223.252]) by smtp.gmail.com with ESMTPSA id v15sm13915357pfm.187.2021.06.01.16.55.39 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 01 Jun 2021 16:55:40 -0700 (PDT) From: James Smart To: linux-scsi@vger.kernel.org Cc: James Smart , Ram Vegesna , Daniel Wagner , Hannes Reinecke Subject: [PATCH v9 28/31] elx: efct: xport and hardware teardown routines Date: Tue, 1 Jun 2021 16:55:09 -0700 Message-Id: <20210601235512.20104-29-jsmart2021@gmail.com> X-Mailer: git-send-email 2.26.2 In-Reply-To: <20210601235512.20104-1-jsmart2021@gmail.com> References: <20210601235512.20104-1-jsmart2021@gmail.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-scsi@vger.kernel.org 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 Signed-off-by: Ram Vegesna Signed-off-by: James Smart Reviewed-by: Daniel Wagner Reviewed-by: Hannes Reinecke --- v9: Non-functional changes: Remove EFC_SUCCESS/EFC_FAIL defines and use 0 and -Exxx errno values. Remove EFCT_xxx/EFCT_HW_RTN_xxx defines and use appropriate -Exxx errno values. Correct indentation on line continuations. --- drivers/scsi/elx/efct/efct_hw.c | 251 +++++++++++++++++++++++++++++ drivers/scsi/elx/efct/efct_hw.h | 27 ++++ drivers/scsi/elx/efct/efct_xport.c | 178 ++++++++++++++++++++ 3 files changed, 456 insertions(+) diff --git a/drivers/scsi/elx/efct/efct_hw.c b/drivers/scsi/elx/efct/efct_hw.c index ddfbb74dbcbd..e2d9a57a4dcb 100644 --- a/drivers/scsi/elx/efct/efct_hw.c +++ b/drivers/scsi/elx/efct/efct_hw.c @@ -3329,3 +3329,254 @@ 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 0; +} + +int +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) +{ + int rc = -EIO; + 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 != 0) { + efc_log_err(hw->os, "CONFIG_LINK failed\n"); + break; + } + speed = hw->config.speed; + reset_alpa = (u8)(value & 0xff); + + rc = -EIO; + 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; +} + +void +efct_hw_teardown(struct efct_hw *hw) +{ + u32 i = 0; + 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 (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); + + kfree(hw->wq_cpu_array); + + 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; +} + +static int +efct_hw_sli_reset(struct efct_hw *hw, enum efct_hw_reset reset, + enum efct_hw_state prev_state) +{ + int rc = 0; + + 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 = -EIO; + } + 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 = -EIO; + } + /* + * 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 = -EIO; + } + break; + default: + efc_log_err(hw->os, "unknown type - no reset performed\n"); + hw->state = prev_state; + rc = -EINVAL; + break; + } + + return rc; +} + +int +efct_hw_reset(struct efct_hw *hw, enum efct_hw_reset reset) +{ + int rc = 0; + 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 (list_empty(&hw->cmd_head)) + efc_log_debug(hw->os, + "All commands completed on MQ queue\n"); + else + efc_log_err(hw->os, + "Some commands still pending on MQ queue\n"); + } + + /* Reset the chip */ + rc = efct_hw_sli_reset(hw, reset, prev_state); + if (rc == -EINVAL) + return -EIO; + + return rc; +} diff --git a/drivers/scsi/elx/efct/efct_hw.h b/drivers/scsi/elx/efct/efct_hw.h index 016ba6ac7b96..32cff5548f0e 100644 --- a/drivers/scsi/elx/efct/efct_hw.h +++ b/drivers/scsi/elx/efct/efct_hw.h @@ -734,4 +734,31 @@ 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); +void efct_hw_teardown(struct efct_hw *hw); +int +efct_hw_reset(struct efct_hw *hw, enum efct_hw_reset reset); + +int +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 58dad8af9e72..e1040d507175 100644 --- a/drivers/scsi/elx/efct/efct_xport.c +++ b/drivers/scsi/elx/efct/efct_xport.c @@ -498,3 +498,181 @@ efct_scsi_release_fc_transport(void) efct_vport_fc_tt = NULL; } + +void +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); +} + +static void +efct_xport_domain_free_cb(struct efc *efc, void *arg) +{ + struct completion *done = arg; + + complete(done); +} + +int +efct_xport_control(struct efct_xport *xport, enum efct_xport_ctrl cmd, ...) +{ + 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; + } + + /* + * 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); +} + +void +efct_scsi_del_device(struct efct *efct) +{ + if (!efct->shost) + return; + + 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; +}