From patchwork Wed Jul 26 05:36:39 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Anup Patel X-Patchwork-Id: 9864189 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id CC2D26038C for ; Wed, 26 Jul 2017 05:38:23 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id C0F382866D for ; Wed, 26 Jul 2017 05:38:23 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id B58E828678; Wed, 26 Jul 2017 05:38:23 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-1.9 required=2.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,RCVD_IN_DNSWL_NONE autolearn=unavailable version=3.3.1 Received: from bombadil.infradead.org (bombadil.infradead.org [65.50.211.133]) (using TLSv1.2 with cipher AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id 7AAA520004 for ; Wed, 26 Jul 2017 05:38:22 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender: Content-Transfer-Encoding:Content-Type:MIME-Version:Cc:List-Subscribe: List-Help:List-Post:List-Archive:List-Unsubscribe:List-Id:References: In-Reply-To:Message-Id:Date:Subject:To:From:Reply-To:Content-ID: Content-Description:Resent-Date:Resent-From:Resent-Sender:Resent-To:Resent-Cc :Resent-Message-ID:List-Owner; bh=nMn8DFRMVBvwdwZfHOGRaTg09AvEpdxkRJdK1a7Ze5g=; b=Pm7u2TTKzkxun3FSlGJlLA6zVe Su3aTCY1qdkMhwlVo7+XKsvEUvI+WC436R713HRmQKvUd8V+Y16xr7hz6DA1xoh6AQoWAm9SiZZuM 2IBMzhDfGeFI+//NNmvnpxOluK4TQzjUZGp2d18shncUXO5zLa7iDPTDgvyAYKjmXrdB39x38b/ko hmXDk/27efionnJ7UzGxp/tR0ueGPP1bZDK8K4t+JvFXBevMkMsnHEW/Lc6hI/e0xdoRDiSvzYFDz 2dYy4qTX2mGWhaeIoJ12EIPhlGNi6U2s8V4aAtzrJ8qhme782PrrFxo+rzZEqu2VRyT81KIIPrKs+ U7jTZQ9Q==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.87 #1 (Red Hat Linux)) id 1daF1S-0004Hw-I4; Wed, 26 Jul 2017 05:38:14 +0000 Received: from mail-wm0-x230.google.com ([2a00:1450:400c:c09::230]) by bombadil.infradead.org with esmtps (Exim 4.87 #1 (Red Hat Linux)) id 1daF18-0003wF-Jn for linux-arm-kernel@lists.infradead.org; Wed, 26 Jul 2017 05:38:09 +0000 Received: by mail-wm0-x230.google.com with SMTP id c184so68487460wmd.0 for ; Tue, 25 Jul 2017 22:37:34 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=broadcom.com; s=google; h=from:to:cc:subject:date:message-id:in-reply-to:references; bh=koG1apK62n5v+cxosbYhC5fw2ZXWqb6seNZgFSWOHhs=; b=G8ExciwsbWAd16lcCCpDw3q7b52fYli8KYd24+56rs1t2zsKmGEKraigDyrfvztDb2 dkwqdkFWwlsm7gGwxdxSUdYfXV55ACAqFLd2DkcwN7Oku61Ofgl4Tmo7WvrOzS6gcQM0 gbOFROp7qEcglGcmf8PazJEc72pr3S2mm1YFI= 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; bh=koG1apK62n5v+cxosbYhC5fw2ZXWqb6seNZgFSWOHhs=; b=MVsLeTajOwPra8fOGNwNPClFKD6TS2P9VwLxJTQ5XoteJ6R3HINPEOedbhr2fi0V2h t+avnLcIt6YSfboCCOZL0fQjfFje3XkgAu73NRcsP176xIcrTUI2QfS1vcGwEe0kH5rL k/rwt5U+GR/kVrMRxEWoCBBE1V/Y1XZeFHXZqa81I963qnGwPJww9KOTSkZpCOvQzx3+ k4M+3q/6kKc+3Oz00WLi2/elNxbKdlcgsHa5BWUPaK881Zkr4Qoz9r11j+9mFywC8xe3 9Wu0VVjP3sl4neMkE2Ik+3AkQUh30LPghqhfLCAKZe+wnF99DPAQqwp5toyAh/QRpMEA 5KQA== X-Gm-Message-State: AIVw113Fh+yCxnKWt9BTHkE6Zh6wEKkVwniqKPB1Jg9OhOefoTCfgak2 Flv4MQalQRxKfVKH X-Received: by 10.28.145.12 with SMTP id t12mr9906157wmd.11.1501047452510; Tue, 25 Jul 2017 22:37:32 -0700 (PDT) Received: from anup-HP-Compaq-8100-Elite-CMT-PC.dhcp.avagotech.net ([192.19.237.250]) by smtp.gmail.com with ESMTPSA id 34sm17417595wrt.36.2017.07.25.22.37.25 (version=TLS1_2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Tue, 25 Jul 2017 22:37:32 -0700 (PDT) From: Anup Patel To: Rob Herring , Mark Rutland , Vinod Koul , Dan Williams Subject: [PATCH 1/6] dma: bcm-sba-raid: Improve memory allocation in SBA RAID driver Date: Wed, 26 Jul 2017 11:06:39 +0530 Message-Id: <1501047404-14456-2-git-send-email-anup.patel@broadcom.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1501047404-14456-1-git-send-email-anup.patel@broadcom.com> References: <1501047404-14456-1-git-send-email-anup.patel@broadcom.com> X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20170725_223755_353361_696D554B X-CRM114-Status: GOOD ( 20.92 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: devicetree@vger.kernel.org, Florian Fainelli , Anup Patel , Scott Branden , Ray Jui , linux-kernel@vger.kernel.org, bcm-kernel-feedback-list@broadcom.com, dmaengine@vger.kernel.org, linux-arm-kernel@lists.infradead.org MIME-Version: 1.0 Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+patchwork-linux-arm=patchwork.kernel.org@lists.infradead.org X-Virus-Scanned: ClamAV using ClamSMTP This patch improves memory allocation in SBA RAID driver in following ways: 1. Simplify struct sba_request to reduce memory consumption 2. Allocate sba resources before registering dma device Signed-off-by: Anup Patel Reviewed-by: Scott Branden Reviewed-by: Ray Jui Reviewed-by: Vikram Prakash --- drivers/dma/bcm-sba-raid.c | 439 +++++++++++++++++++++++---------------------- 1 file changed, 226 insertions(+), 213 deletions(-) diff --git a/drivers/dma/bcm-sba-raid.c b/drivers/dma/bcm-sba-raid.c index e41bbc7..6d15fed 100644 --- a/drivers/dma/bcm-sba-raid.c +++ b/drivers/dma/bcm-sba-raid.c @@ -48,7 +48,8 @@ #include "dmaengine.h" -/* SBA command related defines */ +/* ====== Driver macros and defines ===== */ + #define SBA_TYPE_SHIFT 48 #define SBA_TYPE_MASK GENMASK(1, 0) #define SBA_TYPE_A 0x0 @@ -82,39 +83,41 @@ #define SBA_CMD_WRITE_BUFFER 0xc #define SBA_CMD_GALOIS 0xe -/* Driver helper macros */ +#define SBA_MAX_REQ_PER_MBOX_CHANNEL 8192 + #define to_sba_request(tx) \ container_of(tx, struct sba_request, tx) #define to_sba_device(dchan) \ container_of(dchan, struct sba_device, dma_chan) -enum sba_request_state { - SBA_REQUEST_STATE_FREE = 1, - SBA_REQUEST_STATE_ALLOCED = 2, - SBA_REQUEST_STATE_PENDING = 3, - SBA_REQUEST_STATE_ACTIVE = 4, - SBA_REQUEST_STATE_RECEIVED = 5, - SBA_REQUEST_STATE_COMPLETED = 6, - SBA_REQUEST_STATE_ABORTED = 7, +/* ===== Driver data structures ===== */ + +enum sba_request_flags { + SBA_REQUEST_STATE_FREE = 0x001, + SBA_REQUEST_STATE_ALLOCED = 0x002, + SBA_REQUEST_STATE_PENDING = 0x004, + SBA_REQUEST_STATE_ACTIVE = 0x008, + SBA_REQUEST_STATE_RECEIVED = 0x010, + SBA_REQUEST_STATE_COMPLETED = 0x020, + SBA_REQUEST_STATE_ABORTED = 0x040, + SBA_REQUEST_STATE_MASK = 0x0ff, + SBA_REQUEST_FENCE = 0x100, }; struct sba_request { /* Global state */ struct list_head node; struct sba_device *sba; - enum sba_request_state state; - bool fence; + u32 flags; /* Chained requests management */ struct sba_request *first; struct list_head next; - unsigned int next_count; atomic_t next_pending_count; /* BRCM message data */ - void *resp; - dma_addr_t resp_dma; - struct brcm_sba_command *cmds; struct brcm_message msg; struct dma_async_tx_descriptor tx; + /* SBA commands */ + struct brcm_sba_command cmds[0]; }; enum sba_version { @@ -128,11 +131,11 @@ struct sba_device { /* DT configuration parameters */ enum sba_version ver; /* Derived configuration parameters */ - u32 max_req; u32 hw_buf_size; u32 hw_resp_size; u32 max_pq_coefs; u32 max_pq_srcs; + u32 max_req; u32 max_cmd_per_req; u32 max_xor_srcs; u32 max_resp_pool_size; @@ -152,7 +155,6 @@ struct sba_device { void *cmds_base; dma_addr_t cmds_dma_base; spinlock_t reqs_lock; - struct sba_request *reqs; bool reqs_fence; struct list_head reqs_alloc_list; struct list_head reqs_pending_list; @@ -161,10 +163,9 @@ struct sba_device { struct list_head reqs_completed_list; struct list_head reqs_aborted_list; struct list_head reqs_free_list; - int reqs_free_count; }; -/* ====== SBA command helper routines ===== */ +/* ====== Command helper routines ===== */ static inline u64 __pure sba_cmd_enc(u64 cmd, u32 val, u32 shift, u32 mask) { @@ -196,7 +197,7 @@ static inline u32 __pure sba_cmd_pq_c_mdata(u32 d, u32 b1, u32 b0) ((d & SBA_C_MDATA_DNUM_MASK) << SBA_C_MDATA_DNUM_SHIFT); } -/* ====== Channel resource management routines ===== */ +/* ====== General helper routines ===== */ static struct sba_request *sba_alloc_request(struct sba_device *sba) { @@ -204,24 +205,20 @@ static struct sba_request *sba_alloc_request(struct sba_device *sba) struct sba_request *req = NULL; spin_lock_irqsave(&sba->reqs_lock, flags); - req = list_first_entry_or_null(&sba->reqs_free_list, struct sba_request, node); - if (req) { + if (req) list_move_tail(&req->node, &sba->reqs_alloc_list); - req->state = SBA_REQUEST_STATE_ALLOCED; - req->fence = false; - req->first = req; - INIT_LIST_HEAD(&req->next); - req->next_count = 1; - atomic_set(&req->next_pending_count, 1); - - sba->reqs_free_count--; + spin_unlock_irqrestore(&sba->reqs_lock, flags); + if (!req) + return NULL; - dma_async_tx_descriptor_init(&req->tx, &sba->dma_chan); - } + req->flags = SBA_REQUEST_STATE_ALLOCED; + req->first = req; + INIT_LIST_HEAD(&req->next); + atomic_set(&req->next_pending_count, 1); - spin_unlock_irqrestore(&sba->reqs_lock, flags); + dma_async_tx_descriptor_init(&req->tx, &sba->dma_chan); return req; } @@ -231,7 +228,8 @@ static void _sba_pending_request(struct sba_device *sba, struct sba_request *req) { lockdep_assert_held(&sba->reqs_lock); - req->state = SBA_REQUEST_STATE_PENDING; + req->flags &= ~SBA_REQUEST_STATE_MASK; + req->flags |= SBA_REQUEST_STATE_PENDING; list_move_tail(&req->node, &sba->reqs_pending_list); if (list_empty(&sba->reqs_active_list)) sba->reqs_fence = false; @@ -246,9 +244,10 @@ static bool _sba_active_request(struct sba_device *sba, sba->reqs_fence = false; if (sba->reqs_fence) return false; - req->state = SBA_REQUEST_STATE_ACTIVE; + req->flags &= ~SBA_REQUEST_STATE_MASK; + req->flags |= SBA_REQUEST_STATE_ACTIVE; list_move_tail(&req->node, &sba->reqs_active_list); - if (req->fence) + if (req->flags & SBA_REQUEST_FENCE) sba->reqs_fence = true; return true; } @@ -258,7 +257,8 @@ static void _sba_abort_request(struct sba_device *sba, struct sba_request *req) { lockdep_assert_held(&sba->reqs_lock); - req->state = SBA_REQUEST_STATE_ABORTED; + req->flags &= ~SBA_REQUEST_STATE_MASK; + req->flags |= SBA_REQUEST_STATE_ABORTED; list_move_tail(&req->node, &sba->reqs_aborted_list); if (list_empty(&sba->reqs_active_list)) sba->reqs_fence = false; @@ -269,42 +269,34 @@ static void _sba_free_request(struct sba_device *sba, struct sba_request *req) { lockdep_assert_held(&sba->reqs_lock); - req->state = SBA_REQUEST_STATE_FREE; + req->flags &= ~SBA_REQUEST_STATE_MASK; + req->flags |= SBA_REQUEST_STATE_FREE; list_move_tail(&req->node, &sba->reqs_free_list); if (list_empty(&sba->reqs_active_list)) sba->reqs_fence = false; - sba->reqs_free_count++; } -static void sba_received_request(struct sba_request *req) +/* Note: Must be called with sba->reqs_lock held */ +static void _sba_complete_request(struct sba_device *sba, + struct sba_request *req) { - unsigned long flags; - struct sba_device *sba = req->sba; - - spin_lock_irqsave(&sba->reqs_lock, flags); - req->state = SBA_REQUEST_STATE_RECEIVED; - list_move_tail(&req->node, &sba->reqs_received_list); - spin_unlock_irqrestore(&sba->reqs_lock, flags); + lockdep_assert_held(&sba->reqs_lock); + req->flags &= ~SBA_REQUEST_STATE_MASK; + req->flags |= SBA_REQUEST_STATE_COMPLETED; + list_move_tail(&req->node, &sba->reqs_completed_list); + if (list_empty(&sba->reqs_active_list)) + sba->reqs_fence = false; } -static void sba_complete_chained_requests(struct sba_request *req) +/* Note: Must be called with sba->reqs_lock held */ +static void _sba_received_request(struct sba_device *sba, + struct sba_request *req) { - unsigned long flags; - struct sba_request *nreq; - struct sba_device *sba = req->sba; - - spin_lock_irqsave(&sba->reqs_lock, flags); - - req->state = SBA_REQUEST_STATE_COMPLETED; - list_move_tail(&req->node, &sba->reqs_completed_list); - list_for_each_entry(nreq, &req->next, next) { - nreq->state = SBA_REQUEST_STATE_COMPLETED; - list_move_tail(&nreq->node, &sba->reqs_completed_list); - } + lockdep_assert_held(&sba->reqs_lock); + req->flags = SBA_REQUEST_STATE_RECEIVED; + list_move_tail(&req->node, &sba->reqs_received_list); if (list_empty(&sba->reqs_active_list)) sba->reqs_fence = false; - - spin_unlock_irqrestore(&sba->reqs_lock, flags); } static void sba_free_chained_requests(struct sba_request *req) @@ -332,8 +324,7 @@ static void sba_chain_request(struct sba_request *first, list_add_tail(&req->next, &first->next); req->first = first; - first->next_count++; - atomic_set(&first->next_pending_count, first->next_count); + atomic_inc(&first->next_pending_count); spin_unlock_irqrestore(&sba->reqs_lock, flags); } @@ -383,26 +374,6 @@ static void sba_cleanup_pending_requests(struct sba_device *sba) spin_unlock_irqrestore(&sba->reqs_lock, flags); } -/* ====== DMAENGINE callbacks ===== */ - -static void sba_free_chan_resources(struct dma_chan *dchan) -{ - /* - * Channel resources are pre-alloced so we just free-up - * whatever we can so that we can re-use pre-alloced - * channel resources next time. - */ - sba_cleanup_nonpending_requests(to_sba_device(dchan)); -} - -static int sba_device_terminate_all(struct dma_chan *dchan) -{ - /* Cleanup all pending requests */ - sba_cleanup_pending_requests(to_sba_device(dchan)); - - return 0; -} - static int sba_send_mbox_request(struct sba_device *sba, struct sba_request *req) { @@ -428,17 +399,27 @@ static int sba_send_mbox_request(struct sba_device *sba, return 0; } -static void sba_issue_pending(struct dma_chan *dchan) +static void sba_process_deferred_requests(struct sba_device *sba) { int ret; + u32 count; unsigned long flags; - struct sba_request *req, *req1; - struct sba_device *sba = to_sba_device(dchan); + struct sba_request *req; + struct dma_async_tx_descriptor *tx; spin_lock_irqsave(&sba->reqs_lock, flags); - /* Process all pending request */ - list_for_each_entry_safe(req, req1, &sba->reqs_pending_list, node) { + /* Count pending requests */ + count = 0; + list_for_each_entry(req, &sba->reqs_pending_list, node) + count++; + + /* Process pending requests */ + while (!list_empty(&sba->reqs_pending_list) && count) { + /* Get the first pending request */ + req = list_first_entry(&sba->reqs_pending_list, + struct sba_request, node); + /* Try to make request active */ if (!_sba_active_request(sba, req)) break; @@ -453,11 +434,102 @@ static void sba_issue_pending(struct dma_chan *dchan) _sba_pending_request(sba, req); break; } + + count--; } + /* Count completed requests */ + count = 0; + list_for_each_entry(req, &sba->reqs_completed_list, node) + count++; + + /* Process completed requests */ + while (!list_empty(&sba->reqs_completed_list) && count) { + req = list_first_entry(&sba->reqs_completed_list, + struct sba_request, node); + list_del_init(&req->node); + tx = &req->tx; + + spin_unlock_irqrestore(&sba->reqs_lock, flags); + + WARN_ON(tx->cookie < 0); + if (tx->cookie > 0) { + dma_cookie_complete(tx); + dmaengine_desc_get_callback_invoke(tx, NULL); + dma_descriptor_unmap(tx); + tx->callback = NULL; + tx->callback_result = NULL; + } + + dma_run_dependencies(tx); + + spin_lock_irqsave(&sba->reqs_lock, flags); + + /* If waiting for 'ack' then move to completed list */ + if (!async_tx_test_ack(&req->tx)) + _sba_complete_request(sba, req); + else + _sba_free_request(sba, req); + + count--; + } + + /* Re-check pending and completed work */ + count = 0; + if (!list_empty(&sba->reqs_pending_list) || + !list_empty(&sba->reqs_completed_list)) + count = 1; + spin_unlock_irqrestore(&sba->reqs_lock, flags); } +static void sba_process_received_request(struct sba_device *sba, + struct sba_request *req) +{ + unsigned long flags; + + spin_lock_irqsave(&sba->reqs_lock, flags); + + /* Mark request as received */ + _sba_received_request(sba, req); + + /* Update request */ + if (!atomic_dec_return(&req->first->next_pending_count)) + _sba_complete_request(sba, req->first); + if (req->first != req) + _sba_free_request(sba, req); + + spin_unlock_irqrestore(&sba->reqs_lock, flags); +} + +/* ====== DMAENGINE callbacks ===== */ + +static void sba_free_chan_resources(struct dma_chan *dchan) +{ + /* + * Channel resources are pre-alloced so we just free-up + * whatever we can so that we can re-use pre-alloced + * channel resources next time. + */ + sba_cleanup_nonpending_requests(to_sba_device(dchan)); +} + +static int sba_device_terminate_all(struct dma_chan *dchan) +{ + /* Cleanup all pending requests */ + sba_cleanup_pending_requests(to_sba_device(dchan)); + + return 0; +} + +static void sba_issue_pending(struct dma_chan *dchan) +{ + struct sba_device *sba = to_sba_device(dchan); + + /* Process deferred requests */ + sba_process_deferred_requests(sba); +} + static dma_cookie_t sba_tx_submit(struct dma_async_tx_descriptor *tx) { unsigned long flags; @@ -506,6 +578,7 @@ static void sba_fillup_interrupt_msg(struct sba_request *req, { u64 cmd; u32 c_mdata; + dma_addr_t resp_dma = req->tx.phys; struct brcm_sba_command *cmdsp = cmds; /* Type-B command to load dummy data into buf0 */ @@ -521,7 +594,7 @@ static void sba_fillup_interrupt_msg(struct sba_request *req, cmdsp->cmd = cmd; *cmdsp->cmd_dma = cpu_to_le64(cmd); cmdsp->flags = BRCM_SBA_CMD_TYPE_B; - cmdsp->data = req->resp_dma; + cmdsp->data = resp_dma; cmdsp->data_len = req->sba->hw_resp_size; cmdsp++; @@ -542,11 +615,11 @@ static void sba_fillup_interrupt_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; - cmdsp->data = req->resp_dma; + cmdsp->data = resp_dma; cmdsp->data_len = req->sba->hw_resp_size; cmdsp++; @@ -573,7 +646,7 @@ sba_prep_dma_interrupt(struct dma_chan *dchan, unsigned long flags) * Force fence so that no requests are submitted * until DMA callback for this request is invoked. */ - req->fence = true; + req->flags |= SBA_REQUEST_FENCE; /* Fillup request message */ sba_fillup_interrupt_msg(req, req->cmds, &req->msg); @@ -593,6 +666,7 @@ static void sba_fillup_memcpy_msg(struct sba_request *req, { u64 cmd; u32 c_mdata; + dma_addr_t resp_dma = req->tx.phys; struct brcm_sba_command *cmdsp = cmds; /* Type-B command to load data into buf0 */ @@ -629,7 +703,7 @@ static void sba_fillup_memcpy_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; @@ -656,7 +730,8 @@ sba_prep_dma_memcpy_req(struct sba_device *sba, req = sba_alloc_request(sba); if (!req) return NULL; - req->fence = (flags & DMA_PREP_FENCE) ? true : false; + if (flags & DMA_PREP_FENCE) + req->flags |= SBA_REQUEST_FENCE; /* Fillup request message */ sba_fillup_memcpy_msg(req, req->cmds, &req->msg, @@ -711,6 +786,7 @@ static void sba_fillup_xor_msg(struct sba_request *req, u64 cmd; u32 c_mdata; unsigned int i; + dma_addr_t resp_dma = req->tx.phys; struct brcm_sba_command *cmdsp = cmds; /* Type-B command to load data into buf0 */ @@ -766,7 +842,7 @@ static void sba_fillup_xor_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; @@ -793,7 +869,8 @@ sba_prep_dma_xor_req(struct sba_device *sba, req = sba_alloc_request(sba); if (!req) return NULL; - req->fence = (flags & DMA_PREP_FENCE) ? true : false; + if (flags & DMA_PREP_FENCE) + req->flags |= SBA_REQUEST_FENCE; /* Fillup request message */ sba_fillup_xor_msg(req, req->cmds, &req->msg, @@ -854,6 +931,7 @@ static void sba_fillup_pq_msg(struct sba_request *req, u64 cmd; u32 c_mdata; unsigned int i; + dma_addr_t resp_dma = req->tx.phys; struct brcm_sba_command *cmdsp = cmds; if (pq_continue) { @@ -947,7 +1025,7 @@ static void sba_fillup_pq_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; @@ -974,7 +1052,7 @@ static void sba_fillup_pq_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; @@ -1002,7 +1080,8 @@ sba_prep_dma_pq_req(struct sba_device *sba, dma_addr_t off, req = sba_alloc_request(sba); if (!req) return NULL; - req->fence = (flags & DMA_PREP_FENCE) ? true : false; + if (flags & DMA_PREP_FENCE) + req->flags |= SBA_REQUEST_FENCE; /* Fillup request messages */ sba_fillup_pq_msg(req, dmaf_continue(flags), @@ -1027,6 +1106,7 @@ static void sba_fillup_pq_single_msg(struct sba_request *req, u64 cmd; u32 c_mdata; u8 pos, dpos = raid6_gflog[scf]; + dma_addr_t resp_dma = req->tx.phys; struct brcm_sba_command *cmdsp = cmds; if (!dst_p) @@ -1105,7 +1185,7 @@ static void sba_fillup_pq_single_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; @@ -1226,7 +1306,7 @@ static void sba_fillup_pq_single_msg(struct sba_request *req, cmdsp->flags = BRCM_SBA_CMD_TYPE_A; if (req->sba->hw_resp_size) { cmdsp->flags |= BRCM_SBA_CMD_HAS_RESP; - cmdsp->resp = req->resp_dma; + cmdsp->resp = resp_dma; cmdsp->resp_len = req->sba->hw_resp_size; } cmdsp->flags |= BRCM_SBA_CMD_HAS_OUTPUT; @@ -1255,7 +1335,8 @@ sba_prep_dma_pq_single_req(struct sba_device *sba, dma_addr_t off, req = sba_alloc_request(sba); if (!req) return NULL; - req->fence = (flags & DMA_PREP_FENCE) ? true : false; + if (flags & DMA_PREP_FENCE) + req->flags |= SBA_REQUEST_FENCE; /* Fillup request messages */ sba_fillup_pq_single_msg(req, dmaf_continue(flags), @@ -1370,40 +1451,10 @@ sba_prep_dma_pq(struct dma_chan *dchan, dma_addr_t *dst, dma_addr_t *src, /* ====== Mailbox callbacks ===== */ -static void sba_dma_tx_actions(struct sba_request *req) -{ - struct dma_async_tx_descriptor *tx = &req->tx; - - WARN_ON(tx->cookie < 0); - - if (tx->cookie > 0) { - dma_cookie_complete(tx); - - /* - * Call the callback (must not sleep or submit new - * operations to this channel) - */ - if (tx->callback) - tx->callback(tx->callback_param); - - dma_descriptor_unmap(tx); - } - - /* Run dependent operations */ - dma_run_dependencies(tx); - - /* If waiting for 'ack' then move to completed list */ - if (!async_tx_test_ack(&req->tx)) - sba_complete_chained_requests(req); - else - sba_free_chained_requests(req); -} - static void sba_receive_message(struct mbox_client *cl, void *msg) { - unsigned long flags; struct brcm_message *m = msg; - struct sba_request *req = m->ctx, *req1; + struct sba_request *req = m->ctx; struct sba_device *sba = req->sba; /* Error count if message has error */ @@ -1411,36 +1462,11 @@ static void sba_receive_message(struct mbox_client *cl, void *msg) dev_err(sba->dev, "%s got message with error %d", dma_chan_name(&sba->dma_chan), m->error); - /* Mark request as received */ - sba_received_request(req); - - /* Wait for all chained requests to be completed */ - if (atomic_dec_return(&req->first->next_pending_count)) - goto done; - - /* Point to first request */ - req = req->first; - - /* Update request */ - if (req->state == SBA_REQUEST_STATE_RECEIVED) - sba_dma_tx_actions(req); - else - sba_free_chained_requests(req); - - spin_lock_irqsave(&sba->reqs_lock, flags); - - /* Re-check all completed request waiting for 'ack' */ - list_for_each_entry_safe(req, req1, &sba->reqs_completed_list, node) { - spin_unlock_irqrestore(&sba->reqs_lock, flags); - sba_dma_tx_actions(req); - spin_lock_irqsave(&sba->reqs_lock, flags); - } - - spin_unlock_irqrestore(&sba->reqs_lock, flags); + /* Process received request */ + sba_process_received_request(sba, req); -done: - /* Try to submit pending request */ - sba_issue_pending(&sba->dma_chan); + /* Process deferred requests */ + sba_process_deferred_requests(sba); } /* ====== Platform driver routines ===== */ @@ -1450,13 +1476,13 @@ static int sba_prealloc_channel_resources(struct sba_device *sba) int i, j, p, ret = 0; struct sba_request *req = NULL; - sba->resp_base = dma_alloc_coherent(sba->dma_dev.dev, + sba->resp_base = dma_alloc_coherent(sba->mbox_dev, sba->max_resp_pool_size, &sba->resp_dma_base, GFP_KERNEL); if (!sba->resp_base) return -ENOMEM; - sba->cmds_base = dma_alloc_coherent(sba->dma_dev.dev, + sba->cmds_base = dma_alloc_coherent(sba->mbox_dev, sba->max_cmds_pool_size, &sba->cmds_dma_base, GFP_KERNEL); if (!sba->cmds_base) { @@ -1474,31 +1500,21 @@ static int sba_prealloc_channel_resources(struct sba_device *sba) INIT_LIST_HEAD(&sba->reqs_aborted_list); INIT_LIST_HEAD(&sba->reqs_free_list); - sba->reqs = devm_kcalloc(sba->dev, sba->max_req, - sizeof(*req), GFP_KERNEL); - if (!sba->reqs) { - ret = -ENOMEM; - goto fail_free_cmds_pool; - } - for (i = 0, p = 0; i < sba->max_req; i++) { - req = &sba->reqs[i]; + req = devm_kzalloc(sba->dev, + sizeof(*req) + + sba->max_cmd_per_req * sizeof(req->cmds[0]), + GFP_KERNEL); + if (!req) { + ret = -ENOMEM; + goto fail_free_cmds_pool; + } INIT_LIST_HEAD(&req->node); req->sba = sba; - req->state = SBA_REQUEST_STATE_FREE; + req->flags = SBA_REQUEST_STATE_FREE; INIT_LIST_HEAD(&req->next); - req->next_count = 1; atomic_set(&req->next_pending_count, 0); - req->fence = false; - req->resp = sba->resp_base + p; - req->resp_dma = sba->resp_dma_base + p; p += sba->hw_resp_size; - req->cmds = devm_kcalloc(sba->dev, sba->max_cmd_per_req, - sizeof(*req->cmds), GFP_KERNEL); - if (!req->cmds) { - ret = -ENOMEM; - goto fail_free_cmds_pool; - } for (j = 0; j < sba->max_cmd_per_req; j++) { req->cmds[j].cmd = 0; req->cmds[j].cmd_dma = sba->cmds_base + @@ -1510,20 +1526,18 @@ static int sba_prealloc_channel_resources(struct sba_device *sba) memset(&req->msg, 0, sizeof(req->msg)); dma_async_tx_descriptor_init(&req->tx, &sba->dma_chan); req->tx.tx_submit = sba_tx_submit; - req->tx.phys = req->resp_dma; + req->tx.phys = sba->resp_dma_base + i * sba->hw_resp_size; list_add_tail(&req->node, &sba->reqs_free_list); } - sba->reqs_free_count = sba->max_req; - return 0; fail_free_cmds_pool: - dma_free_coherent(sba->dma_dev.dev, + dma_free_coherent(sba->mbox_dev, sba->max_cmds_pool_size, sba->cmds_base, sba->cmds_dma_base); fail_free_resp_pool: - dma_free_coherent(sba->dma_dev.dev, + dma_free_coherent(sba->mbox_dev, sba->max_resp_pool_size, sba->resp_base, sba->resp_dma_base); return ret; @@ -1532,9 +1546,9 @@ static int sba_prealloc_channel_resources(struct sba_device *sba) static void sba_freeup_channel_resources(struct sba_device *sba) { dmaengine_terminate_all(&sba->dma_chan); - dma_free_coherent(sba->dma_dev.dev, sba->max_cmds_pool_size, + dma_free_coherent(sba->mbox_dev, sba->max_cmds_pool_size, sba->cmds_base, sba->cmds_dma_base); - dma_free_coherent(sba->dma_dev.dev, sba->max_resp_pool_size, + dma_free_coherent(sba->mbox_dev, sba->max_resp_pool_size, sba->resp_base, sba->resp_dma_base); sba->resp_base = NULL; sba->resp_dma_base = 0; @@ -1625,6 +1639,13 @@ static int sba_probe(struct platform_device *pdev) sba->dev = &pdev->dev; platform_set_drvdata(pdev, sba); + /* Number of channels equals number of mailbox channels */ + ret = of_count_phandle_with_args(pdev->dev.of_node, + "mboxes", "#mbox-cells"); + if (ret <= 0) + return -ENODEV; + mchans_count = ret; + /* Determine SBA version from DT compatible string */ if (of_device_is_compatible(sba->dev->of_node, "brcm,iproc-sba")) sba->ver = SBA_VER_1; @@ -1637,14 +1658,12 @@ static int sba_probe(struct platform_device *pdev) /* Derived Configuration parameters */ switch (sba->ver) { case SBA_VER_1: - sba->max_req = 1024; sba->hw_buf_size = 4096; sba->hw_resp_size = 8; sba->max_pq_coefs = 6; sba->max_pq_srcs = 6; break; case SBA_VER_2: - sba->max_req = 1024; sba->hw_buf_size = 4096; sba->hw_resp_size = 8; sba->max_pq_coefs = 30; @@ -1658,6 +1677,7 @@ static int sba_probe(struct platform_device *pdev) default: return -EINVAL; } + sba->max_req = SBA_MAX_REQ_PER_MBOX_CHANNEL * mchans_count; sba->max_cmd_per_req = sba->max_pq_srcs + 3; sba->max_xor_srcs = sba->max_cmd_per_req - 1; sba->max_resp_pool_size = sba->max_req * sba->hw_resp_size; @@ -1671,22 +1691,14 @@ static int sba_probe(struct platform_device *pdev) sba->client.knows_txdone = false; sba->client.tx_tout = 0; - /* Number of channels equals number of mailbox channels */ - ret = of_count_phandle_with_args(pdev->dev.of_node, - "mboxes", "#mbox-cells"); - if (ret <= 0) - return -ENODEV; - mchans_count = ret; - sba->mchans_count = 0; - atomic_set(&sba->mchans_current, 0); - /* Allocate mailbox channel array */ - sba->mchans = devm_kcalloc(&pdev->dev, sba->mchans_count, + sba->mchans = devm_kcalloc(&pdev->dev, mchans_count, sizeof(*sba->mchans), GFP_KERNEL); if (!sba->mchans) return -ENOMEM; /* Request mailbox channels */ + sba->mchans_count = 0; for (i = 0; i < mchans_count; i++) { sba->mchans[i] = mbox_request_channel(&sba->client, i); if (IS_ERR(sba->mchans[i])) { @@ -1695,6 +1707,7 @@ static int sba_probe(struct platform_device *pdev) } sba->mchans_count++; } + atomic_set(&sba->mchans_current, 0); /* Find-out underlying mailbox device */ ret = of_parse_phandle_with_args(pdev->dev.of_node, @@ -1723,15 +1736,15 @@ static int sba_probe(struct platform_device *pdev) } } - /* Register DMA device with linux async framework */ - ret = sba_async_register(sba); + /* Prealloc channel resource */ + ret = sba_prealloc_channel_resources(sba); if (ret) goto fail_free_mchans; - /* Prealloc channel resource */ - ret = sba_prealloc_channel_resources(sba); + /* Register DMA device with linux async framework */ + ret = sba_async_register(sba); if (ret) - goto fail_async_dev_unreg; + goto fail_free_resources; /* Print device info */ dev_info(sba->dev, "%s using SBAv%d and %d mailbox channels", @@ -1740,8 +1753,8 @@ static int sba_probe(struct platform_device *pdev) return 0; -fail_async_dev_unreg: - dma_async_device_unregister(&sba->dma_dev); +fail_free_resources: + sba_freeup_channel_resources(sba); fail_free_mchans: for (i = 0; i < sba->mchans_count; i++) mbox_free_channel(sba->mchans[i]); @@ -1753,10 +1766,10 @@ static int sba_remove(struct platform_device *pdev) int i; struct sba_device *sba = platform_get_drvdata(pdev); - sba_freeup_channel_resources(sba); - dma_async_device_unregister(&sba->dma_dev); + sba_freeup_channel_resources(sba); + for (i = 0; i < sba->mchans_count; i++) mbox_free_channel(sba->mchans[i]);