Commit 9f651379 authored by Yixing Liu's avatar Yixing Liu Committed by Juan Zhou
Browse files

RDMA/hns: Kernel notify usr space to stop ring db 

driver inclusion
category: bugfix
bugzilla: https://gitee.com/openeuler/kernel/issues/I98HQV



--------------------------------------------------------------------------

In the reset scenario, if the kernel receives the reset signal,
it needs to notify the user space to stop ring doorbell.

Signed-off-by: default avatarYixing Liu <liuyixing1@huawei.com>
Signed-off-by: default avatarChengchang Tang <tangchengchang@huawei.com>
parent 9e5bcc70

drivers/infiniband/hw/hns/hns_roce_device.h

+5 −0
Original line number Diff line number Diff line
@@ -61,6 +61,7 @@ 


#define HNS_ROCE_CEQ				0

#define HNS_ROCE_AEQ				1

#define HNS_ROCE_IS_RESETTING			1



#define HNS_ROCE_CEQE_SIZE 0x4

#define HNS_ROCE_AEQE_SIZE 0x10
@@ -197,6 +198,7 @@ struct hns_roce_uar { 
enum hns_roce_mmap_type {

	HNS_ROCE_MMAP_TYPE_DB = 1,

	HNS_ROCE_MMAP_TYPE_DWQE,

	HNS_ROCE_MMAP_TYPE_RESET = 4,

};



struct hns_user_mmap_entry {
@@ -211,6 +213,7 @@ struct hns_roce_ucontext { 
	struct list_head	page_list;

	struct mutex		page_mutex;

	struct hns_user_mmap_entry *db_mmap_entry;

	struct hns_user_mmap_entry *reset_mmap_entry;

	u32			config;

};
@@ -1041,6 +1044,8 @@ struct hns_roce_dev { 
	int			loop_idc;

	u32			sdb_offset;

	u32			odb_offset;

	struct page		*reset_page; /* store reset state */

	void			*reset_kaddr; /* addr of reset page */

	const struct hns_roce_hw *hw;

	void			*priv;

	struct workqueue_struct *irq_workq;

drivers/infiniband/hw/hns/hns_roce_hw_v2.c

+53 −2
Original line number Diff line number Diff line
@@ -2607,6 +2607,31 @@ static void free_dip_list(struct hns_roce_dev *hr_dev) 
	spin_unlock_irqrestore(&hr_dev->dip_list_lock, flags);

}



static int hns_roce_v2_get_reset_page(struct hns_roce_dev *hr_dev)

{

	hr_dev->reset_page = alloc_page(GFP_KERNEL | __GFP_ZERO);

	if (!hr_dev->reset_page)

		return -ENOMEM;



	hr_dev->reset_kaddr = vmap(&hr_dev->reset_page, 1, VM_MAP, PAGE_KERNEL);

	if (!hr_dev->reset_kaddr)

		goto err_with_vmap;



	return 0;



err_with_vmap:

	put_page(hr_dev->reset_page);

	return -ENOMEM;

}



static void hns_roce_v2_put_reset_page(struct hns_roce_dev *hr_dev)

{

	vunmap(hr_dev->reset_kaddr);

	hr_dev->reset_kaddr = NULL;

	put_page(hr_dev->reset_page);

	hr_dev->reset_page = NULL;

}



static struct ib_pd *free_mr_init_pd(struct hns_roce_dev *hr_dev)

{

	struct hns_roce_v2_priv *priv = hr_dev->priv;
@@ -2971,14 +2996,21 @@ static int hns_roce_v2_init(struct hns_roce_dev *hr_dev) 
{

	int ret;



	ret = hns_roce_v2_get_reset_page(hr_dev);

	if (ret) {

		dev_err(hr_dev->dev,

			"reset state init failed, ret = %d.\n", ret);

		return ret;

	}



	/* The hns ROCEE requires the extdb info to be cleared before using */

	ret = hns_roce_clear_extdb_list_info(hr_dev);

	if (ret)

		return ret;

		goto err_clear_extdb_failed;



	ret = get_hem_table(hr_dev);

	if (ret)

		return ret;

		goto err_clear_extdb_failed;



	if (hr_dev->is_vf)

		return 0;
@@ -2993,6 +3025,8 @@ static int hns_roce_v2_init(struct hns_roce_dev *hr_dev) 


err_llm_init_failed:

	put_hem_table(hr_dev);

err_clear_extdb_failed:

	hns_roce_v2_put_reset_page(hr_dev);



	return ret;

}
@@ -3004,6 +3038,8 @@ static void hns_roce_v2_exit(struct hns_roce_dev *hr_dev) 
	if (!hr_dev->is_vf)

		hns_roce_free_link_table(hr_dev);



	hns_roce_v2_put_reset_page(hr_dev);



	if (hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP09)

		free_dip_list(hr_dev);

}
@@ -7165,6 +7201,18 @@ int hns_roce_bond_uninit_client(struct hns_roce_bond_group *bond_grp, 


	return 0;

}



static void hns_roce_v2_reset_notify_user(struct hns_roce_dev *hr_dev)

{

	struct hns_roce_v2_reset_state *state;



	state = (struct hns_roce_v2_reset_state *)hr_dev->reset_kaddr;



	state->reset_state = HNS_ROCE_IS_RESETTING;

	/* Ensure reset state was flushed in memory */

	wmb();

}



static int hns_roce_hw_v2_reset_notify_down(struct hnae3_handle *handle)

{

	struct hns_roce_dev *hr_dev;
@@ -7183,6 +7231,9 @@ static int hns_roce_hw_v2_reset_notify_down(struct hnae3_handle *handle) 


	hr_dev->active = false;

	hr_dev->dis_db = true;



	hns_roce_v2_reset_notify_user(hr_dev);



	hr_dev->state = HNS_ROCE_DEVICE_STATE_RST_DOWN;



	return 0;

drivers/infiniband/hw/hns/hns_roce_hw_v2.h

+4 −0
Original line number Diff line number Diff line
@@ -1313,6 +1313,10 @@ struct hns_roce_link_table { 
#define HNS_ROCE_EXT_LLM_ENTRY(addr, id) (((id) << (64 - 12)) | ((addr) >> 12))

#define HNS_ROCE_EXT_LLM_MIN_PAGES(que_num) ((que_num) * 4 + 2)



struct hns_roce_v2_reset_state {

	u32 reset_state; /* stored to use in user space */

};



struct hns_roce_v2_free_mr {

	struct hns_roce_qp *rsv_qp[HNS_ROCE_FREE_MR_USED_QP_NUM];

	struct hns_roce_cq *rsv_cq;

drivers/infiniband/hw/hns/hns_roce_main.c

+44 −0
Original line number Diff line number Diff line
@@ -392,6 +392,7 @@ hns_roce_user_mmap_entry_insert(struct ib_ucontext *ucontext, u64 address, 
				ucontext, &entry->rdma_entry, length, 0);

		break;

	case HNS_ROCE_MMAP_TYPE_DWQE:

	case HNS_ROCE_MMAP_TYPE_RESET:

		ret = rdma_user_mmap_entry_insert_range(

				ucontext, &entry->rdma_entry, length, 1,

				U32_MAX);
@@ -430,6 +431,26 @@ static int hns_roce_alloc_uar_entry(struct ib_ucontext *uctx) 
	return 0;

}



static void hns_roce_dealloc_reset_entry(struct hns_roce_ucontext *context)

{

	if (context->reset_mmap_entry)

		rdma_user_mmap_entry_remove(&context->reset_mmap_entry->rdma_entry);

}



static int hns_roce_alloc_reset_entry(struct ib_ucontext *uctx)

{

	struct hns_roce_ucontext *context = to_hr_ucontext(uctx);

	struct hns_roce_dev *hr_dev = to_hr_dev(uctx->device);



	context->reset_mmap_entry = hns_roce_user_mmap_entry_insert(uctx,

		(u64)hr_dev->reset_kaddr, PAGE_SIZE, HNS_ROCE_MMAP_TYPE_RESET);



	if (!context->reset_mmap_entry)

		return -ENOMEM;



	return 0;

}



static int hns_roce_alloc_ucontext(struct ib_ucontext *uctx,

				   struct ib_udata *udata)

{
@@ -437,6 +458,7 @@ static int hns_roce_alloc_ucontext(struct ib_ucontext *uctx, 
	struct hns_roce_dev *hr_dev = to_hr_dev(uctx->device);

	struct hns_roce_ib_alloc_ucontext_resp resp = {};

	struct hns_roce_ib_alloc_ucontext ucmd = {};

	struct rdma_user_mmap_entry *rdma_entry;

	int ret = -EAGAIN;



	if (!hr_dev->active)
@@ -492,6 +514,15 @@ static int hns_roce_alloc_ucontext(struct ib_ucontext *uctx, 
		mutex_init(&context->page_mutex);

	}



	ret = hns_roce_alloc_reset_entry(uctx);

	if (ret)

		goto error_fail_reset_entry;



	if (context->reset_mmap_entry) {

		rdma_entry = &context->reset_mmap_entry->rdma_entry;

		resp.reset_mmap_key = rdma_user_mmap_get_offset(rdma_entry);

	}



	resp.cqe_size = hr_dev->caps.cqe_sz;



	ret = ib_copy_to_udata(udata, &resp,
@@ -502,6 +533,9 @@ static int hns_roce_alloc_ucontext(struct ib_ucontext *uctx, 
	return 0;



error_fail_copy_to_udata:

	hns_roce_dealloc_reset_entry(context);



error_fail_reset_entry:

	hns_roce_dealloc_uar_entry(context);



error_fail_uar_entry:
@@ -519,6 +553,7 @@ static void hns_roce_dealloc_ucontext(struct ib_ucontext *ibcontext) 
	struct hns_roce_dev *hr_dev = to_hr_dev(ibcontext->device);



	hns_roce_dealloc_uar_entry(context);

	hns_roce_dealloc_reset_entry(context);



	ida_free(&hr_dev->uar_ida.ida, (int)context->uar.logic_idx);

}
@@ -546,6 +581,15 @@ static int hns_roce_mmap(struct ib_ucontext *uctx, struct vm_area_struct *vma) 
	case HNS_ROCE_MMAP_TYPE_DWQE:

		prot = pgprot_device(vma->vm_page_prot);

		break;

	case HNS_ROCE_MMAP_TYPE_RESET:

		if (vma->vm_flags & (VM_WRITE | VM_EXEC)) {

			ret = -EINVAL;

			goto out;

		}



		prot = vma->vm_page_prot;

		pfn = page_to_pfn(hr_dev->reset_page);

		break;

	default:

		ret = -EINVAL;

		goto out;

include/uapi/rdma/hns-abi.h

+2 −0
Original line number Diff line number Diff line
@@ -139,6 +139,8 @@ struct hns_roce_ib_alloc_ucontext_resp { 
	__u32	max_inline_data;

	__u8	congest_type;

	__u8	reserved0[7];

	__aligned_u64 rsv_for_dca[2];

	__aligned_u64 reset_mmap_key;

};



struct hns_roce_ib_alloc_ucontext {