From nobody Thu Sep 19 16:23:05 2024 Received: from szxga04-in.huawei.com (szxga04-in.huawei.com [45.249.212.190]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id D38A41DA0E4; Fri, 13 Sep 2024 12:35:57 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=45.249.212.190 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1726230961; cv=none; b=uF+vsJImDiZB4pc8l9PHX6NBcCPqt0iTPdHX5K+v0y77t4nx1Se72NZBF0VOVZgDl+jpkTpHwhpKYzHakEyg84wUoagsR/7vGVjIFoMsv6w+Ks40aaxkfq2eZn7m9OGFCtTCawPH6fnw/spD442A/TmzBxPss9aq2LZmBDd3roY= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1726230961; c=relaxed/simple; bh=gpQAwDKw8Kf+lbReKVEjgA+iHO84l3ALkdLM5z4g1Fo=; h=From:To:CC:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version:Content-Type; b=fW8sZMIT443HjyL34djh+lt+wohI1T0HdWi7F4V4Ubqlll7p3FUnjwXddQgdjFFoOY0a2jCRxTOZUe4AklSJY7Vi4gDIJmasUBHFvXFc2x+kSkKTrEqNEJM1cD6A2f3YCOMWN7AeNlb2mMkhoFd6lNSK3m2WsA6c11Bx7RAOpng= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=quarantine dis=none) header.from=hisilicon.com; spf=pass smtp.mailfrom=hisilicon.com; arc=none smtp.client-ip=45.249.212.190 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=quarantine dis=none) header.from=hisilicon.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=hisilicon.com Received: from mail.maildlp.com (unknown [172.19.163.44]) by szxga04-in.huawei.com (SkyGuard) with ESMTP id 4X4v0P4Kfwz2CpX9; Fri, 13 Sep 2024 20:35:17 +0800 (CST) Received: from kwepemf100018.china.huawei.com (unknown [7.202.181.17]) by mail.maildlp.com (Postfix) with ESMTPS id 995431402CD; Fri, 13 Sep 2024 20:35:49 +0800 (CST) Received: from localhost.localdomain (10.90.30.45) by kwepemf100018.china.huawei.com (7.202.181.17) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.2.1544.11; Fri, 13 Sep 2024 20:35:49 +0800 From: Junxian Huang To: , CC: , , , Subject: [PATCH v5 for-next 1/2] RDMA/core: Provide rdma_user_mmap_disassociate() to disassociate mmap pages Date: Fri, 13 Sep 2024 20:29:54 +0800 Message-ID: <20240913122955.1283597-2-huangjunxian6@hisilicon.com> X-Mailer: git-send-email 2.30.0 In-Reply-To: <20240913122955.1283597-1-huangjunxian6@hisilicon.com> References: <20240913122955.1283597-1-huangjunxian6@hisilicon.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-ClientProxiedBy: dggems702-chm.china.huawei.com (10.3.19.179) To kwepemf100018.china.huawei.com (7.202.181.17) Content-Type: text/plain; charset="utf-8" From: Chengchang Tang Provide a new api rdma_user_mmap_disassociate() for drivers to disassociate mmap pages for a device. Since drivers can now disassociate mmaps by calling this api, introduce a new disassociation_lock to specifically prevent races between this disassociation process and new mmaps. And thus the old hw_destroy_rwsem is not needed in this api. Signed-off-by: Chengchang Tang Signed-off-by: Junxian Huang --- drivers/infiniband/core/uverbs.h | 2 ++ drivers/infiniband/core/uverbs_main.c | 47 ++++++++++++++++++++++++--- include/rdma/ib_verbs.h | 8 +++++ 3 files changed, 53 insertions(+), 4 deletions(-) diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uve= rbs.h index 821d93c8f712..dfd2e5a86e6f 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -160,6 +160,8 @@ struct ib_uverbs_file { struct page *disassociate_page; =20 struct xarray idr; + + struct mutex disassociation_lock; }; =20 struct ib_uverbs_event { diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/cor= e/uverbs_main.c index bc099287de9a..9b2073c8cc6e 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -76,6 +76,7 @@ static dev_t dynamic_uverbs_dev; static DEFINE_IDA(uverbs_ida); static int ib_uverbs_add_one(struct ib_device *device); static void ib_uverbs_remove_one(struct ib_device *device, void *client_da= ta); +static struct ib_client uverbs_client; =20 static char *uverbs_devnode(const struct device *dev, umode_t *mode) { @@ -217,6 +218,7 @@ void ib_uverbs_release_file(struct kref *ref) =20 if (file->disassociate_page) __free_pages(file->disassociate_page, 0); + mutex_destroy(&file->disassociation_lock); mutex_destroy(&file->umap_lock); mutex_destroy(&file->ucontext_lock); kfree(file); @@ -698,11 +700,16 @@ static int ib_uverbs_mmap(struct file *filp, struct v= m_area_struct *vma) ucontext =3D ib_uverbs_get_ucontext_file(file); if (IS_ERR(ucontext)) { ret =3D PTR_ERR(ucontext); - goto out; + goto out_srcu; } + + mutex_lock(&file->disassociation_lock); + vma->vm_ops =3D &rdma_umap_ops; ret =3D ucontext->device->ops.mmap(ucontext, vma); -out: +out_mutex: + mutex_unlock(&file->disassociation_lock); +out_srcu: srcu_read_unlock(&file->device->disassociate_srcu, srcu_key); return ret; } @@ -723,6 +730,8 @@ static void rdma_umap_open(struct vm_area_struct *vma) /* We are racing with disassociation */ if (!down_read_trylock(&ufile->hw_destroy_rwsem)) goto out_zap; + mutex_lock(&ufile->disassociation_lock); + /* * Disassociation already completed, the VMA should already be zapped. */ @@ -734,10 +743,12 @@ static void rdma_umap_open(struct vm_area_struct *vma) goto out_unlock; rdma_umap_priv_init(priv, vma, opriv->entry); =20 + mutex_unlock(&ufile->disassociation_lock); up_read(&ufile->hw_destroy_rwsem); return; =20 out_unlock: + mutex_unlock(&ufile->disassociation_lock); up_read(&ufile->hw_destroy_rwsem); out_zap: /* @@ -821,7 +832,7 @@ void uverbs_user_mmap_disassociate(struct ib_uverbs_fil= e *ufile) { struct rdma_umap_priv *priv, *next_priv; =20 - lockdep_assert_held(&ufile->hw_destroy_rwsem); + mutex_lock(&ufile->disassociation_lock); =20 while (1) { struct mm_struct *mm =3D NULL; @@ -847,8 +858,10 @@ void uverbs_user_mmap_disassociate(struct ib_uverbs_fi= le *ufile) break; } mutex_unlock(&ufile->umap_lock); - if (!mm) + if (!mm) { + mutex_unlock(&ufile->disassociation_lock); return; + } =20 /* * The umap_lock is nested under mmap_lock since it used within @@ -878,8 +891,32 @@ void uverbs_user_mmap_disassociate(struct ib_uverbs_fi= le *ufile) mmap_read_unlock(mm); mmput(mm); } + + mutex_unlock(&ufile->disassociation_lock); } =20 +/** + * rdma_user_mmap_disassociate() - Revoke mmaps for a device + * @device: device to revoke + * + * This function should be called by drivers that need to disable mmaps fo= r the + * device, for instance because it is going to be reset. + */ +void rdma_user_mmap_disassociate(struct ib_device *device) +{ + struct ib_uverbs_device *uverbs_dev =3D + ib_get_client_data(device, &uverbs_client); + struct ib_uverbs_file *ufile; + + mutex_lock(&uverbs_dev->lists_mutex); + list_for_each_entry(ufile, &uverbs_dev->uverbs_file_list, list) { + if (ufile->ucontext) + uverbs_user_mmap_disassociate(ufile); + } + mutex_unlock(&uverbs_dev->lists_mutex); +} +EXPORT_SYMBOL(rdma_user_mmap_disassociate); + /* * ib_uverbs_open() does not need the BKL: * @@ -949,6 +986,8 @@ static int ib_uverbs_open(struct inode *inode, struct f= ile *filp) mutex_init(&file->umap_lock); INIT_LIST_HEAD(&file->umaps); =20 + mutex_init(&file->disassociation_lock); + filp->private_data =3D file; list_add_tail(&file->list, &dev->uverbs_file_list); mutex_unlock(&dev->lists_mutex); diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index aa8ede439905..9cb8b5fe7eee 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -2948,6 +2948,14 @@ int rdma_user_mmap_entry_insert_range(struct ib_ucon= text *ucontext, size_t length, u32 min_pgoff, u32 max_pgoff); =20 +#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) +void rdma_user_mmap_disassociate(struct ib_device *device); +#else +static inline void rdma_user_mmap_disassociate(struct ib_device *device) +{ +} +#endif + static inline int rdma_user_mmap_entry_insert_exact(struct ib_ucontext *ucontext, struct rdma_user_mmap_entry *entry, --=20 2.33.0 From nobody Thu Sep 19 16:23:05 2024 Received: from szxga07-in.huawei.com (szxga07-in.huawei.com [45.249.212.35]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id D54FA1DA0F7; Fri, 13 Sep 2024 12:35:57 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=45.249.212.35 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1726230961; cv=none; b=laNcSWFzM9Yr6rjqZKVxWhlFGatauQOOnWuK40NGPYwtdlxKljaUv5ZGFaWzn6YJ8inMAXxjqzkXPiRRRm7X5jT9BXcKbLYY5PoSfCJnBW31w7f+6Q0C2zakBgmmq2lByo+0/j9J42iBrXv1wwETSpgs03Sb3umspioUeE47cLQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1726230961; c=relaxed/simple; bh=RKJVRbU/YeLVaz5xn1kS06BHeX6wi6KEvhyLFydcRP0=; h=From:To:CC:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version:Content-Type; b=QrMYlrnlurVa8ynJApUyFBlmkzrN6Xw6+2cjyQ/jcZ9LEnABD+71Zpf92p0RGiFrLPFkaRZNNYq+f0viDPnvZosjwi6rE7RqjvorXUVjF+ENVfkbYDq/wtOlml9RGHaWTqb2yzRpvr+YIRD2F0y9PbUyAgDSkLNuYgYeKUvmtF8= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=quarantine dis=none) header.from=hisilicon.com; spf=pass smtp.mailfrom=hisilicon.com; arc=none smtp.client-ip=45.249.212.35 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=quarantine dis=none) header.from=hisilicon.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=hisilicon.com Received: from mail.maildlp.com (unknown [172.19.88.234]) by szxga07-in.huawei.com (SkyGuard) with ESMTP id 4X4v0P2SYlz1SB0B; Fri, 13 Sep 2024 20:35:17 +0800 (CST) Received: from kwepemf100018.china.huawei.com (unknown [7.202.181.17]) by mail.maildlp.com (Postfix) with ESMTPS id E73B1140360; Fri, 13 Sep 2024 20:35:49 +0800 (CST) Received: from localhost.localdomain (10.90.30.45) by kwepemf100018.china.huawei.com (7.202.181.17) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.2.1544.11; Fri, 13 Sep 2024 20:35:49 +0800 From: Junxian Huang To: , CC: , , , Subject: [PATCH v5 for-next 2/2] RDMA/hns: Disassociate mmap pages for all uctx when HW is being reset Date: Fri, 13 Sep 2024 20:29:55 +0800 Message-ID: <20240913122955.1283597-3-huangjunxian6@hisilicon.com> X-Mailer: git-send-email 2.30.0 In-Reply-To: <20240913122955.1283597-1-huangjunxian6@hisilicon.com> References: <20240913122955.1283597-1-huangjunxian6@hisilicon.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-ClientProxiedBy: dggems702-chm.china.huawei.com (10.3.19.179) To kwepemf100018.china.huawei.com (7.202.181.17) Content-Type: text/plain; charset="utf-8" From: Chengchang Tang When HW is being reset, userspace should not ring doorbell otherwise it may lead to abnormal consequence such as RAS. Disassociate mmap pages for all uctx to prevent userspace from ringing doorbell to HW. Since all resources will be destroyed during HW reset, no new mmap is allowed after HW reset is completed. Fixes: 9a4435375cd1 ("IB/hns: Add driver files for hns RoCE driver") Signed-off-by: Chengchang Tang Signed-off-by: Junxian Huang --- drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 9 +++++++++ drivers/infiniband/hw/hns/hns_roce_main.c | 5 +++++ 2 files changed, 14 insertions(+) diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniban= d/hw/hns/hns_roce_hw_v2.c index 24e906b9d3ae..4e374b2da101 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -7017,6 +7017,12 @@ static void hns_roce_hw_v2_uninit_instance(struct hn= ae3_handle *handle, =20 handle->rinfo.instance_state =3D HNS_ROCE_STATE_NON_INIT; } + +static void hns_roce_v2_reset_notify_user(struct hns_roce_dev *hr_dev) +{ + rdma_user_mmap_disassociate(&hr_dev->ib_dev); +} + static int hns_roce_hw_v2_reset_notify_down(struct hnae3_handle *handle) { struct hns_roce_dev *hr_dev; @@ -7035,6 +7041,9 @@ static int hns_roce_hw_v2_reset_notify_down(struct hn= ae3_handle *handle) =20 hr_dev->active =3D false; hr_dev->dis_db =3D true; + + hns_roce_v2_reset_notify_user(hr_dev); + hr_dev->state =3D HNS_ROCE_DEVICE_STATE_RST_DOWN; =20 return 0; diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband= /hw/hns/hns_roce_main.c index 4cb0af733587..49315f39361d 100644 --- a/drivers/infiniband/hw/hns/hns_roce_main.c +++ b/drivers/infiniband/hw/hns/hns_roce_main.c @@ -466,6 +466,11 @@ static int hns_roce_mmap(struct ib_ucontext *uctx, str= uct vm_area_struct *vma) pgprot_t prot; int ret; =20 + if (hr_dev->dis_db) { + atomic64_inc(&hr_dev->dfx_cnt[HNS_ROCE_DFX_MMAP_ERR_CNT]); + return -EPERM; + } + rdma_entry =3D rdma_user_mmap_entry_get_pgoff(uctx, vma->vm_pgoff); if (!rdma_entry) { atomic64_inc(&hr_dev->dfx_cnt[HNS_ROCE_DFX_MMAP_ERR_CNT]); --=20 2.33.0