From patchwork Fri Mar 5 10:28:58 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Suganath Prabu S X-Patchwork-Id: 12118295 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=-18.8 required=3.0 tests=BAYES_00,DKIMWL_WL_HIGH, DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,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 3585DC433E6 for ; Fri, 5 Mar 2021 10:30:25 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id F10ED64F52 for ; Fri, 5 Mar 2021 10:30:24 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229772AbhCEK3y (ORCPT ); Fri, 5 Mar 2021 05:29:54 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39476 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229723AbhCEK3j (ORCPT ); Fri, 5 Mar 2021 05:29:39 -0500 Received: from mail-pl1-x62d.google.com (mail-pl1-x62d.google.com [IPv6:2607:f8b0:4864:20::62d]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 78A3FC061574 for ; Fri, 5 Mar 2021 02:29:39 -0800 (PST) Received: by mail-pl1-x62d.google.com with SMTP id s7so1197756plg.5 for ; Fri, 05 Mar 2021 02:29:39 -0800 (PST) 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 :mime-version; bh=zymnQ1kPQOf1fHfZARBSYBG0ZDdDVTOB/qKs7iU2j4U=; b=EqdqR++08HAkz9r/x/rW+VAF2NPUJQ+g1XcEXrZeEsALIjkA0J5q/bOgx/m59Ss0E0 SYlutqonfUfpC+AgO9nKcHDZxg30KNGCqReGvN2NG4g8XkWH0up4MXPyJLqNEqE3u57N I5fosrqaJ1gUls434YlVMjpTyFGlrcXBX7KZ8= 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; bh=zymnQ1kPQOf1fHfZARBSYBG0ZDdDVTOB/qKs7iU2j4U=; b=O9eaSazA/IF2Nr2pIqr/Qqi+chN5oI9E6ApMhoOhp2tvdYK4Z0a9ViLTeiVpyTftLQ OlS4xgUBVawjVs2f2AYq1+xkIg2nEmKSR64724cYaMJq7JO8N6NpkaF/Kohp/FErBT4r 23g4teST3WzktP/s7dpfTEq8VAx+t8EpvL5DrbagovzbWIXwo7A//3YOyDVa/6hxoerB KTEjsnT+ZG3cJD7o/tK5NZXZQRuhXPnwBY9LPz4H3a8boGkpZs9IaN+4IeD34MFrdXL0 Ix7paY6AfH9E3s/2hzOmZg/15VOVHrLFDybBSkKt6PMm+MB0jR0GuIjz9yHF0XrSuDM/ OCSw== X-Gm-Message-State: AOAM530R5rUfRo13CvX1BXfHMKIKeJuA/CD3mBsKaPjGjKmdLhAg1F3L QcPq2hthkfOCoDKP0IENlFPgYdaBHVjIcZtBXYz1XmDxIRIMff8o/C3oqpINdeKYc1eOMKsLeKr or6+PWll6S/0E6OWUWMoAlxssNh2ghxSTr0dQioxyFTDBJ5noOmJ01n5DU2fZr035HMMOuoDoHf nQJQ50ddfcYo6Vpzmoxxmb X-Google-Smtp-Source: ABdhPJxBNrEp5irxz9+3vfr7JqoMvH9dnHfNgItcIaIfiH0IObJVA/5NZ3VG6WKuEzGxcYdU4I6Svw== X-Received: by 2002:a17:902:9894:b029:e5:ce48:5808 with SMTP id s20-20020a1709029894b02900e5ce485808mr8005332plp.31.1614940178117; Fri, 05 Mar 2021 02:29:38 -0800 (PST) Received: from dhcp-10-123-20-76.dhcp.broadcom.net ([192.19.234.250]) by smtp.gmail.com with ESMTPSA id v15sm2015983pgl.44.2021.03.05.02.29.35 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Fri, 05 Mar 2021 02:29:37 -0800 (PST) From: Suganath Prabu S To: linux-scsi@vger.kernel.org, martin.petersen@oracle.com Cc: Sathya.Prakash@broadcom.com, sreekanth.reddy@broadcom.com, Suganath Prabu S Subject: [PATCH 1/7] mpt3sas: Handle PCIe sgl's in same 4G region. Date: Fri, 5 Mar 2021 15:58:58 +0530 Message-Id: <20210305102904.7560-2-suganath-prabu.subramani@broadcom.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20210305102904.7560-1-suganath-prabu.subramani@broadcom.com> References: <20210305102904.7560-1-suganath-prabu.subramani@broadcom.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-scsi@vger.kernel.org According to MPI Specification PCIe SGL buffers should not cross 4GB boundary. So while allocating PCIe SGL buffers, if any buffer crosses the 4GB boundary then, * Release the already allocated memory pools and * Reallocate them by changing the DMA coherent mask to 32 bit. Signed-off-by: Suganath Prabu S --- drivers/scsi/mpt3sas/mpt3sas_base.c | 159 ++++++++++++++++++++-------- drivers/scsi/mpt3sas/mpt3sas_base.h | 1 + 2 files changed, 113 insertions(+), 47 deletions(-) diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.c b/drivers/scsi/mpt3sas/mpt3sas_base.c index ac066f8..f9e6f8e 100644 --- a/drivers/scsi/mpt3sas/mpt3sas_base.c +++ b/drivers/scsi/mpt3sas/mpt3sas_base.c @@ -2905,23 +2905,22 @@ static int _base_config_dma_addressing(struct MPT3SAS_ADAPTER *ioc, struct pci_dev *pdev) { struct sysinfo s; - int dma_mask; if (ioc->is_mcpu_endpoint || sizeof(dma_addr_t) == 4 || ioc->use_32bit_dma || dma_get_required_mask(&pdev->dev) <= 32) - dma_mask = 32; + ioc->dma_mask = 32; /* Set 63 bit DMA mask for all SAS3 and SAS35 controllers */ else if (ioc->hba_mpi_version_belonged > MPI2_VERSION) - dma_mask = 63; + ioc->dma_mask = 63; else - dma_mask = 64; + ioc->dma_mask = 64; - if (dma_set_mask(&pdev->dev, DMA_BIT_MASK(dma_mask)) || - dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(dma_mask))) + if (dma_set_mask(&pdev->dev, DMA_BIT_MASK(ioc->dma_mask)) || + dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(ioc->dma_mask))) return -ENODEV; - if (dma_mask > 32) { + if (ioc->dma_mask > 32) { ioc->base_add_sg_single = &_base_add_sg_single_64; ioc->sge_size = sizeof(Mpi2SGESimple64_t); } else { @@ -2931,7 +2930,7 @@ _base_config_dma_addressing(struct MPT3SAS_ADAPTER *ioc, struct pci_dev *pdev) si_meminfo(&s); ioc_info(ioc, "%d BIT PCI BUS DMA ADDRESSING SUPPORTED, total mem (%ld kB)\n", - dma_mask, convert_to_kb(s.totalram)); + ioc->dma_mask, convert_to_kb(s.totalram)); return 0; } @@ -5338,10 +5337,10 @@ _base_release_memory_pools(struct MPT3SAS_ADAPTER *ioc) dma_pool_free(ioc->pcie_sgl_dma_pool, ioc->pcie_sg_lookup[i].pcie_sgl, ioc->pcie_sg_lookup[i].pcie_sgl_dma); + ioc->pcie_sg_lookup[i].pcie_sgl = NULL; } dma_pool_destroy(ioc->pcie_sgl_dma_pool); } - if (ioc->config_page) { dexitprintk(ioc, ioc_info(ioc, "config_page(0x%p): free\n", @@ -5399,6 +5398,89 @@ mpt3sas_check_same_4gb_region(long reply_pool_start_address, u32 pool_sz) return 0; } +/** + * _base_reduce_hba_queue_depth- Retry with reduced queue depth + * @ioc: Adapter object + * + * Return: 0 for success, non-zero for failure. + **/ +static inline int +_base_reduce_hba_queue_depth(struct MPT3SAS_ADAPTER *ioc) +{ + int reduce_sz = 64; + + if ((ioc->hba_queue_depth - reduce_sz) > + (ioc->internal_depth + INTERNAL_SCSIIO_CMDS_COUNT)) { + ioc->hba_queue_depth -= reduce_sz; + return 0; + } else + return -ENOMEM; +} + +/** + * _base_allocate_pcie_sgl_pool - Allocating DMA'able memory + * for pcie sgl pools. + * @ioc: Adapter object + * @sz: DMA Pool size + * @ct: Chain tracker + * Return: 0 for success, non-zero for failure. + */ + +static int +_base_allocate_pcie_sgl_pool(struct MPT3SAS_ADAPTER *ioc, u32 sz) +{ + int i = 0, j = 0; + struct chain_tracker *ct; + + ioc->pcie_sgl_dma_pool = + dma_pool_create("PCIe SGL pool", &ioc->pdev->dev, sz, + ioc->page_size, 0); + if (!ioc->pcie_sgl_dma_pool) { + ioc_err(ioc, "PCIe SGL pool: dma_pool_create failed\n"); + return -ENOMEM; + } + + ioc->chains_per_prp_buffer = sz/ioc->chain_segment_sz; + ioc->chains_per_prp_buffer = + min(ioc->chains_per_prp_buffer, ioc->chains_needed_per_io); + for (i = 0; i < ioc->scsiio_depth; i++) { + ioc->pcie_sg_lookup[i].pcie_sgl = + dma_pool_alloc(ioc->pcie_sgl_dma_pool, GFP_KERNEL, + &ioc->pcie_sg_lookup[i].pcie_sgl_dma); + if (!ioc->pcie_sg_lookup[i].pcie_sgl) { + ioc_err(ioc, "PCIe SGL pool: dma_pool_alloc failed\n"); + return -EAGAIN; + } + + if (!mpt3sas_check_same_4gb_region( + (long)ioc->pcie_sg_lookup[i].pcie_sgl, sz)) { + ioc_err(ioc, "PCIE SGLs are not in same 4G !! pcie sgl (0x%p) dma = (0x%llx)\n", + ioc->pcie_sg_lookup[i].pcie_sgl, + (unsigned long long) + ioc->pcie_sg_lookup[i].pcie_sgl_dma); + ioc->use_32bit_dma = true; + return -EAGAIN; + } + + for (j = 0; j < ioc->chains_per_prp_buffer; j++) { + ct = &ioc->chain_lookup[i].chains_per_smid[j]; + ct->chain_buffer = + ioc->pcie_sg_lookup[i].pcie_sgl + + (j * ioc->chain_segment_sz); + ct->chain_buffer_dma = + ioc->pcie_sg_lookup[i].pcie_sgl_dma + + (j * ioc->chain_segment_sz); + } + } + dinitprintk(ioc, ioc_info(ioc, + "PCIe sgl pool depth(%d), element_size(%d), pool_size(%d kB)\n", + ioc->scsiio_depth, sz, (sz * ioc->scsiio_depth)/1024)); + dinitprintk(ioc, ioc_info(ioc, + "Number of chains can fit in a PRP page(%d)\n", + ioc->chains_per_prp_buffer)); + return 0; +} + /** * base_alloc_rdpq_dma_pool - Allocating DMA'able memory * for reply queues. @@ -5497,7 +5579,7 @@ _base_allocate_memory_pools(struct MPT3SAS_ADAPTER *ioc) unsigned short sg_tablesize; u16 sge_size; int i, j; - int ret = 0; + int ret = 0, rc = 0; struct chain_tracker *ct; dinitprintk(ioc, ioc_info(ioc, "%s\n", __func__)); @@ -5802,6 +5884,7 @@ _base_allocate_memory_pools(struct MPT3SAS_ADAPTER *ioc) * be required for NVMe PRP's, only each set of NVMe blocks will be * contiguous, so a new set is allocated for each possible I/O. */ + ioc->chains_per_prp_buffer = 0; if (ioc->facts.ProtocolFlags & MPI2_IOCFACTS_PROTOCOL_NVME_DEVICES) { nvme_blocks_needed = @@ -5816,43 +5899,11 @@ _base_allocate_memory_pools(struct MPT3SAS_ADAPTER *ioc) goto out; } sz = nvme_blocks_needed * ioc->page_size; - ioc->pcie_sgl_dma_pool = - dma_pool_create("PCIe SGL pool", &ioc->pdev->dev, sz, 16, 0); - if (!ioc->pcie_sgl_dma_pool) { - ioc_info(ioc, "PCIe SGL pool: dma_pool_create failed\n"); - goto out; - } - - ioc->chains_per_prp_buffer = sz/ioc->chain_segment_sz; - ioc->chains_per_prp_buffer = min(ioc->chains_per_prp_buffer, - ioc->chains_needed_per_io); - - for (i = 0; i < ioc->scsiio_depth; i++) { - ioc->pcie_sg_lookup[i].pcie_sgl = dma_pool_alloc( - ioc->pcie_sgl_dma_pool, GFP_KERNEL, - &ioc->pcie_sg_lookup[i].pcie_sgl_dma); - if (!ioc->pcie_sg_lookup[i].pcie_sgl) { - ioc_info(ioc, "PCIe SGL pool: dma_pool_alloc failed\n"); - goto out; - } - for (j = 0; j < ioc->chains_per_prp_buffer; j++) { - ct = &ioc->chain_lookup[i].chains_per_smid[j]; - ct->chain_buffer = - ioc->pcie_sg_lookup[i].pcie_sgl + - (j * ioc->chain_segment_sz); - ct->chain_buffer_dma = - ioc->pcie_sg_lookup[i].pcie_sgl_dma + - (j * ioc->chain_segment_sz); - } - } - - dinitprintk(ioc, - ioc_info(ioc, "PCIe sgl pool depth(%d), element_size(%d), pool_size(%d kB)\n", - ioc->scsiio_depth, sz, - (sz * ioc->scsiio_depth) / 1024)); - dinitprintk(ioc, - ioc_info(ioc, "Number of chains can fit in a PRP page(%d)\n", - ioc->chains_per_prp_buffer)); + rc = _base_allocate_pcie_sgl_pool(ioc, sz); + if (rc == -ENOMEM) + return -ENOMEM; + else if (rc == -EAGAIN) + goto try_32bit_dma; total_sz += sz * ioc->scsiio_depth; } @@ -6022,6 +6073,19 @@ _base_allocate_memory_pools(struct MPT3SAS_ADAPTER *ioc) ioc->shost->sg_tablesize); return 0; +try_32bit_dma: + _base_release_memory_pools(ioc); + if (ioc->use_32bit_dma && (ioc->dma_mask > 32)) { + /* Change dma coherent mask to 32 bit and reallocate */ + if (_base_config_dma_addressing(ioc, ioc->pdev) != 0) { + pr_err("Setting 32 bit coherent DMA mask Failed %s\n", + pci_name(ioc->pdev)); + return -ENODEV; + } + } else if (_base_reduce_hba_queue_depth(ioc) != 0) + return -ENOMEM; + goto retry_allocation; + out: return -ENOMEM; } @@ -7682,6 +7746,7 @@ mpt3sas_base_attach(struct MPT3SAS_ADAPTER *ioc) ioc->rdpq_array_enable_assigned = 0; ioc->use_32bit_dma = false; + ioc->dma_mask = 64; if (ioc->is_aero_ioc) ioc->base_readl = &_base_readl_aero; else diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.h b/drivers/scsi/mpt3sas/mpt3sas_base.h index 315aee6..b86eced 100644 --- a/drivers/scsi/mpt3sas/mpt3sas_base.h +++ b/drivers/scsi/mpt3sas/mpt3sas_base.h @@ -1371,6 +1371,7 @@ struct MPT3SAS_ADAPTER { u16 thresh_hold; u8 high_iops_queues; u32 drv_support_bitmap; + u32 dma_mask; bool enable_sdev_max_qd; bool use_32bit_dma;