From patchwork Tue Sep 22 08:21:04 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Leon Romanovsky X-Patchwork-Id: 11791817 X-Patchwork-Delegate: jgg@ziepe.ca Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 27486112C for ; Tue, 22 Sep 2020 08:21:23 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 09ED223A1E for ; Tue, 22 Sep 2020 08:21:22 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1600762883; bh=lfO6wozy3t9vkjCRw4UsYQFeZ0fvbF2+UjBfiz0Er0w=; h=From:To:Cc:Subject:Date:In-Reply-To:References:List-ID:From; b=rQmDm59oquSob+98igAN/hUqDc1fXChkR9ZHmPdRclf98pVftmFi6/RH2Ry6wHv14 HPpANLtYoDgcE2b6yrRBsWpRKEo2r8rR2pEQUHTGDgVB+/TavzO4ACqABunK4rY4QO eF4ha9QaEYYFkhN9VPY1eQxv3yUMw2Fgj5cyDnGU= Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726716AbgIVIVW (ORCPT ); Tue, 22 Sep 2020 04:21:22 -0400 Received: from mail.kernel.org ([198.145.29.99]:35946 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726649AbgIVIVW (ORCPT ); Tue, 22 Sep 2020 04:21:22 -0400 Received: from localhost (unknown [213.57.247.131]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPSA id CDD7F239E5; Tue, 22 Sep 2020 08:21:20 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1600762881; bh=lfO6wozy3t9vkjCRw4UsYQFeZ0fvbF2+UjBfiz0Er0w=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=Surrj9TCni9KdyhzWgXqk3LhBpxPdGTTx6WV0Ze+pqoRPanwd4EUO5FwNz8gzDQCR pJPAgd2D22KgmsZB8f5d5iGOIshIqY5q34W9FefPKRhKP18kdHCXhWZB6T8/nU74s5 86q54UiqR1SGRZBdRv8PsBb7IArCGUsm8f+0ipy0= From: Leon Romanovsky To: Doug Ledford , Jason Gunthorpe Cc: Yishai Hadas , linux-rdma@vger.kernel.org, Christoph Hellwig Subject: [PATCH rdma-next v2 4/4] RDMA/mlx5: Sync device with CPU pages upon ODP MR registration Date: Tue, 22 Sep 2020 11:21:04 +0300 Message-Id: <20200922082104.2148873-5-leon@kernel.org> X-Mailer: git-send-email 2.26.2 In-Reply-To: <20200922082104.2148873-1-leon@kernel.org> References: <20200922082104.2148873-1-leon@kernel.org> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org From: Yishai Hadas Sync device with CPU pages upon ODP MR registration. This reduce potential page faults down the road and improve performance. Signed-off-by: Yishai Hadas Signed-off-by: Leon Romanovsky --- drivers/infiniband/hw/mlx5/mlx5_ib.h | 5 +++++ drivers/infiniband/hw/mlx5/mr.c | 11 +++++++---- drivers/infiniband/hw/mlx5/odp.c | 21 ++++++++++++++++++++- 3 files changed, 32 insertions(+), 5 deletions(-) -- 2.26.2 diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index 6ab3efb75b21..b1f2b34e5955 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -1283,6 +1283,7 @@ void mlx5_odp_populate_xlt(void *xlt, size_t idx, size_t nentries, int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice, u32 flags, struct ib_sge *sg_list, u32 num_sge); +int mlx5_ib_init_odp_mr(struct mlx5_ib_mr *mr, bool enable); #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */ static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) { @@ -1304,6 +1305,10 @@ mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, { return -EOPNOTSUPP; } +static inline int mlx5_ib_init_odp_mr(struct mlx5_ib_mr *mr, bool enable) +{ + return -EOPNOTSUPP; +} #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */ extern const struct mmu_interval_notifier_ops mlx5_mn_ops; diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index dea65e511a3e..d6264687f08c 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -1431,7 +1431,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, mr->umem = umem; set_mr_fields(dev, mr, npages, length, access_flags); - if (xlt_with_umr) { + if (xlt_with_umr && !(access_flags & IB_ACCESS_ON_DEMAND)) { /* * If the MR was created with reg_create then it will be * configured properly but left disabled. It is safe to go ahead @@ -1439,9 +1439,6 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, */ int update_xlt_flags = MLX5_IB_UPD_XLT_ENABLE; - if (access_flags & IB_ACCESS_ON_DEMAND) - update_xlt_flags |= MLX5_IB_UPD_XLT_ZAP; - err = mlx5_ib_update_xlt(mr, 0, ncont, page_shift, update_xlt_flags); if (err) { @@ -1467,6 +1464,12 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, dereg_mr(dev, mr); return ERR_PTR(err); } + + err = mlx5_ib_init_odp_mr(mr, xlt_with_umr); + if (err) { + dereg_mr(dev, mr); + return ERR_PTR(err); + } } return &mr->ibmr; diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 28b7227d31bf..15fd6d224527 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -666,6 +666,7 @@ void mlx5_ib_fence_odp_mr(struct mlx5_ib_mr *mr) #define MLX5_PF_FLAGS_DOWNGRADE BIT(1) #define MLX5_PF_FLAGS_SNAPSHOT BIT(2) +#define MLX5_PF_FLAGS_ENABLE BIT(3) static int pagefault_real_mr(struct mlx5_ib_mr *mr, struct ib_umem_odp *odp, u64 user_va, size_t bcnt, u32 *bytes_mapped, u32 flags) @@ -675,6 +676,10 @@ static int pagefault_real_mr(struct mlx5_ib_mr *mr, struct ib_umem_odp *odp, u64 access_mask; u64 start_idx; bool fault = !(flags & MLX5_PF_FLAGS_SNAPSHOT); + u32 xlt_flags = MLX5_IB_UPD_XLT_ATOMIC; + + if (flags & MLX5_PF_FLAGS_ENABLE) + xlt_flags |= MLX5_IB_UPD_XLT_ENABLE; page_shift = odp->page_shift; start_idx = (user_va - ib_umem_start(odp)) >> page_shift; @@ -692,7 +697,7 @@ static int pagefault_real_mr(struct mlx5_ib_mr *mr, struct ib_umem_odp *odp, * ib_umem_odp_map_dma_and_lock already checks this. */ ret = mlx5_ib_update_xlt(mr, start_idx, np, - page_shift, MLX5_IB_UPD_XLT_ATOMIC); + page_shift, xlt_flags); mutex_unlock(&odp->umem_mutex); if (ret < 0) { @@ -827,6 +832,20 @@ static int pagefault_mr(struct mlx5_ib_mr *mr, u64 io_virt, size_t bcnt, flags); } +int mlx5_ib_init_odp_mr(struct mlx5_ib_mr *mr, bool enable) +{ + u32 flags = MLX5_PF_FLAGS_SNAPSHOT; + int ret; + + if (enable) + flags |= MLX5_PF_FLAGS_ENABLE; + + ret = pagefault_real_mr(mr, to_ib_umem_odp(mr->umem), + mr->umem->address, mr->umem->length, NULL, + flags); + return ret >= 0 ? 0 : ret; +} + struct pf_frame { struct pf_frame *next; u32 key;