Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1390093 > unrolled thread
| Started by | Lijun Ou <oulijun@huawei.com> |
|---|---|
| First post | 2016-04-28 14:10 +0200 |
| Last post | 2016-05-04 13:40 +0200 |
| Articles | 10 — 3 participants |
Back to article view | Back to linux.kernel
[PATCH v6 00/21] Add HiSilicon RoCE driver Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 20/21] IB/hns: Kconfig and Makefile for RoCE module Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 09/21] IB/hns: Add hca support Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 16/21] IB/hns: Add ah operations support Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 14/21] IB/hns: Add operations support for IB device and port Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 02/21] devicetree: bindings: IB: Add binding document for HiSilicon RoCE Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 08/21] IB/hns: Add icm support Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:10 +0200
[PATCH v6 06/21] IB/hns: Add initial cmd operation Lijun Ou <oulijun@huawei.com> - 2016-04-28 14:20 +0200
Re: [PATCH v6 00/21] Add HiSilicon RoCE driver Leon Romanovsky <leon@kernel.org> - 2016-05-03 20:50 +0200
Re: [PATCH v6 00/21] Add HiSilicon RoCE driver oulijun <oulijun@huawei.com> - 2016-05-04 13:40 +0200
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 00/21] Add HiSilicon RoCE driver |
| Message-ID | <rsSVA-2Qt-5@gated-at.bofh.it> |
The HiSilicon Network Substem is a long term evolution IP which is supposed to be used in HiSilicon ICT SoCs. HNS (HiSilicon Network Sybsystem) also has a hardware support of performing RDMA with RoCEE. The driver for HiSilicon RoCEE(RoCE Engine) is a platform driver and will support mulitple versions of SOCs in future. This version of driver is meant to support Hip06 SoC(which confirms to RoCEEv1 hardware specifications). Changes v5 -> v6: 1. modify the type of obj for unsigned long according the reviews, and modify the same questions in RoCE module. 2. fix the spelling error. 3. fix the Signed-off-by signatures. Changes v4 -> v5: 1. redesign the patchset for RoCE modules in order to split the huge patch into small patches. 2. fix the directory path for RoCE module. Delete the hisilicon level. 3. modify the name of roce_v1_hw into roce_hw_v1. Changes v3 -> v4: 1. modify roce.o into hns-roce.o in Makefile and Kconfig file. Changes v2 -> v3: 1. modify the formats of RoCE driver code base v2 by the experts reviewing. also, it used kmalloc_array instead of kmalloc, kcalloc instead of kzalloc, when refer to memory allocation for array 2. remove some functions without use and unconnected macros 3. modify the binding document with RoCE DT base v2 which added interrupt-names 4. redesign the port_map and si_map in hns_dsaf_roce_reset 5. add HiSilicon RoCE driver maintainers introduction in MAINTAINERS document Changes v1 -> v2: 1. modify the formats of roce driver code by the experts reviewing 2. modify the bindings file with roce dts. add the attribute named interrput-names. 3. modify the way of defining port mode in hns_dsaf_main.c 4. move the Kconfig file into the hns directory and send it with roce Lijun Ou (21): net: hns: Add reset function support for RoCE driver devicetree: bindings: IB: Add binding document for HiSilicon RoCE IB/hns: Add initial main frame driver and get cfg info IB/hns: Add RoCE engine reset function IB/hns: Add initial profile resource IB/hns: Add initial cmd operation IB/hns: Add event queue support IB/hns: Add icm support IB/hns: Add hca support IB/hns: Add process flow to init RoCE engine IB/hns: Add IB device registration IB/hns: Set mtu and gid support IB/hns: Add interface of the protocol stack registration IB/hns: Add operations support for IB device and port IB/hns: Add PD operations support IB/hns: Add ah operations support IB/hns: Add QP operations support IB/hns: Add CQ operations support IB/hns: Add memory region operations support IB/hns: Kconfig and Makefile for RoCE module MAINTAINERS: Add maintainers for HiSilicon RoCE driver .../bindings/infiniband/hisilicon-hns-roce.txt | 107 + MAINTAINERS | 8 + drivers/infiniband/Kconfig | 1 + drivers/infiniband/hw/Makefile | 1 + drivers/infiniband/hw/hns/Kconfig | 10 + drivers/infiniband/hw/hns/Makefile | 9 + drivers/infiniband/hw/hns/hns_roce_ah.c | 109 + drivers/infiniband/hw/hns/hns_roce_alloc.c | 238 ++ drivers/infiniband/hw/hns/hns_roce_cmd.c | 324 +++ drivers/infiniband/hw/hns/hns_roce_cmd.h | 80 + drivers/infiniband/hw/hns/hns_roce_common.h | 302 +++ drivers/infiniband/hw/hns/hns_roce_cq.c | 437 +++ drivers/infiniband/hw/hns/hns_roce_device.h | 732 +++++ drivers/infiniband/hw/hns/hns_roce_eq.c | 756 ++++++ drivers/infiniband/hw/hns/hns_roce_eq.h | 95 + drivers/infiniband/hw/hns/hns_roce_hw_v1.c | 2812 ++++++++++++++++++++ drivers/infiniband/hw/hns/hns_roce_hw_v1.h | 966 +++++++ drivers/infiniband/hw/hns/hns_roce_icm.c | 582 ++++ drivers/infiniband/hw/hns/hns_roce_icm.h | 112 + drivers/infiniband/hw/hns/hns_roce_main.c | 1072 ++++++++ drivers/infiniband/hw/hns/hns_roce_mr.c | 599 +++++ drivers/infiniband/hw/hns/hns_roce_pd.c | 127 + drivers/infiniband/hw/hns/hns_roce_qp.c | 835 ++++++ drivers/infiniband/hw/hns/hns_roce_user.h | 27 + drivers/net/ethernet/hisilicon/hns/hns_dsaf_main.c | 84 + drivers/net/ethernet/hisilicon/hns/hns_dsaf_main.h | 30 + drivers/net/ethernet/hisilicon/hns/hns_dsaf_misc.c | 62 +- drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h | 16 +- 28 files changed, 10522 insertions(+), 11 deletions(-) create mode 100644 Documentation/devicetree/bindings/infiniband/hisilicon-hns-roce.txt create mode 100644 drivers/infiniband/hw/hns/Kconfig create mode 100644 drivers/infiniband/hw/hns/Makefile create mode 100644 drivers/infiniband/hw/hns/hns_roce_ah.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_alloc.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_cmd.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_cmd.h create mode 100644 drivers/infiniband/hw/hns/hns_roce_common.h create mode 100644 drivers/infiniband/hw/hns/hns_roce_cq.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_device.h create mode 100644 drivers/infiniband/hw/hns/hns_roce_eq.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_eq.h create mode 100644 drivers/infiniband/hw/hns/hns_roce_hw_v1.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_hw_v1.h create mode 100644 drivers/infiniband/hw/hns/hns_roce_icm.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_icm.h create mode 100644 drivers/infiniband/hw/hns/hns_roce_main.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_mr.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_pd.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_qp.c create mode 100644 drivers/infiniband/hw/hns/hns_roce_user.h -- 1.9.1
[toc] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 20/21] IB/hns: Kconfig and Makefile for RoCE module |
| Message-ID | <rsSVC-2Qt-49@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch added Kconfig and Makefile for building RoCE module. Signed-off-by: Wei Hu <xavier.huwei@huawei.com> Signed-off-by: Nenglong Zhao <zhaonenglong@hisilicon.com> Signed-off-by: Lijun Ou <oulijun@huawei.com> --- drivers/infiniband/Kconfig | 1 + drivers/infiniband/hw/Makefile | 1 + drivers/infiniband/hw/hns/Kconfig | 10 ++++++++++ drivers/infiniband/hw/hns/Makefile | 9 +++++++++ 4 files changed, 21 insertions(+) create mode 100644 drivers/infiniband/hw/hns/Kconfig create mode 100644 drivers/infiniband/hw/hns/Makefile diff --git a/drivers/infiniband/Kconfig b/drivers/infiniband/Kconfig index 6425c0e..726a4ca 100644 --- a/drivers/infiniband/Kconfig +++ b/drivers/infiniband/Kconfig @@ -74,6 +74,7 @@ source "drivers/infiniband/hw/mlx5/Kconfig" source "drivers/infiniband/hw/nes/Kconfig" source "drivers/infiniband/hw/ocrdma/Kconfig" source "drivers/infiniband/hw/usnic/Kconfig" +source "drivers/infiniband/hw/hns/Kconfig" source "drivers/infiniband/ulp/ipoib/Kconfig" diff --git a/drivers/infiniband/hw/Makefile b/drivers/infiniband/hw/Makefile index c7ad0a4..223eb78 100644 --- a/drivers/infiniband/hw/Makefile +++ b/drivers/infiniband/hw/Makefile @@ -8,3 +8,4 @@ obj-$(CONFIG_MLX5_INFINIBAND) += mlx5/ obj-$(CONFIG_INFINIBAND_NES) += nes/ obj-$(CONFIG_INFINIBAND_OCRDMA) += ocrdma/ obj-$(CONFIG_INFINIBAND_USNIC) += usnic/ +obj-$(CONFIG_INFINIBAND_HISILICON_HNS) += hns/ diff --git a/drivers/infiniband/hw/hns/Kconfig b/drivers/infiniband/hw/hns/Kconfig new file mode 100644 index 0000000..c47c168 --- /dev/null +++ b/drivers/infiniband/hw/hns/Kconfig @@ -0,0 +1,10 @@ +config INFINIBAND_HISILICON_HNS + tristate "Hisilicon Hns ROCE Driver" + depends on NET_VENDOR_HISILICON + depends on ARM64 && HNS && HNS_DSAF && HNS_ENET + ---help--- + This is a ROCE/RDMA driver for the Hisilicon RoCE engine. The engine + is used in Hisilicon Hi1610 and more further ICT SoC. + + To compile this driver as a module, choose M here: the module + will be called hns-roce. diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile new file mode 100644 index 0000000..404a700 --- /dev/null +++ b/drivers/infiniband/hw/hns/Makefile @@ -0,0 +1,9 @@ +# +# Makefile for the HISILICON RoCE drivers. +# + +obj-$(CONFIG_INFINIBAND_HISILICON_HNS) += hns-roce.o +hns-roce-objs := hns_roce_main.o hns_roce_cmd.o hns_roce_eq.o hns_roce_pd.o \ + hns_roce_ah.o hns_roce_icm.o hns_roce_mr.o hns_roce_qp.o \ + hns_roce_cq.o hns_roce_alloc.o hns_roce_hw_v1.o + -- 1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 09/21] IB/hns: Add hca support |
| Message-ID | <rsSVC-2Qt-57@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch mainly setup hca for RoCE. It will do a series of
initial works, as follows:
1. init uar table, allocate uar resource
2. init pd table
3. init cq table
4. init mr table
5. init qp table
Signed-off-by: Wei Hu <xavier.huwei@huawei.com>
Signed-off-by: Nenglong Zhao <zhaonenglong@hisilicon.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
---
drivers/infiniband/hw/hns/hns_roce_alloc.c | 104 ++++++++++++++++
drivers/infiniband/hw/hns/hns_roce_cq.c | 25 ++++
drivers/infiniband/hw/hns/hns_roce_device.h | 73 ++++++++++-
drivers/infiniband/hw/hns/hns_roce_icm.c | 89 +++++++++++++
drivers/infiniband/hw/hns/hns_roce_icm.h | 7 ++
drivers/infiniband/hw/hns/hns_roce_main.c | 79 ++++++++++++
drivers/infiniband/hw/hns/hns_roce_mr.c | 187 ++++++++++++++++++++++++++++
drivers/infiniband/hw/hns/hns_roce_pd.c | 65 ++++++++++
drivers/infiniband/hw/hns/hns_roce_qp.c | 30 +++++
9 files changed, 657 insertions(+), 2 deletions(-)
create mode 100644 drivers/infiniband/hw/hns/hns_roce_alloc.c
create mode 100644 drivers/infiniband/hw/hns/hns_roce_mr.c
create mode 100644 drivers/infiniband/hw/hns/hns_roce_pd.c
diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c b/drivers/infiniband/hw/hns/hns_roce_alloc.c
new file mode 100644
index 0000000..972d2ac
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
@@ -0,0 +1,104 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/bitmap.h>
+#include <linux/dma-mapping.h>
+#include <linux/errno.h>
+#include <linux/mm.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include "hns_roce_device.h"
+
+int hns_roce_bitmap_alloc(struct hns_roce_bitmap *bitmap, unsigned long *obj)
+{
+ int ret = 0;
+
+ spin_lock(&bitmap->lock);
+ *obj = find_next_zero_bit(bitmap->table, bitmap->max, bitmap->last);
+ if (*obj >= bitmap->max) {
+ bitmap->top = (bitmap->top + bitmap->max + bitmap->reserved_top)
+ & bitmap->mask;
+ *obj = find_first_zero_bit(bitmap->table, bitmap->max);
+ }
+
+ if (*obj < bitmap->max) {
+ set_bit(*obj, bitmap->table);
+ bitmap->last = (*obj + 1);
+ if (bitmap->last == bitmap->max)
+ bitmap->last = 0;
+ *obj |= bitmap->top;
+ } else {
+ ret = -1;
+ }
+
+ spin_unlock(&bitmap->lock);
+
+ return ret;
+}
+
+void hns_roce_bitmap_free(struct hns_roce_bitmap *bitmap, unsigned long obj)
+{
+ hns_roce_bitmap_free_range(bitmap, obj, 1);
+}
+
+void hns_roce_bitmap_free_range(struct hns_roce_bitmap *bitmap,
+ unsigned long obj, int cnt)
+{
+ int i;
+
+ obj &= bitmap->max + bitmap->reserved_top - 1;
+
+ spin_lock(&bitmap->lock);
+ for (i = 0; i < cnt; i++)
+ clear_bit(obj + i, bitmap->table);
+
+ bitmap->last = min(bitmap->last, obj);
+ bitmap->top = (bitmap->top + bitmap->max + bitmap->reserved_top)
+ & bitmap->mask;
+ spin_unlock(&bitmap->lock);
+}
+
+int hns_roce_bitmap_init(struct hns_roce_bitmap *bitmap, u32 num, u32 mask,
+ u32 reserved_bot, u32 reserved_top)
+{
+ u32 i;
+
+ if (num != roundup_pow_of_two(num))
+ return -EINVAL;
+
+ bitmap->last = 0;
+ bitmap->top = 0;
+ bitmap->max = num - reserved_top;
+ bitmap->mask = mask;
+ bitmap->reserved_top = reserved_top;
+ spin_lock_init(&bitmap->lock);
+ bitmap->table = kcalloc(BITS_TO_LONGS(bitmap->max), sizeof(long),
+ GFP_KERNEL);
+ if (!bitmap->table)
+ return -ENOMEM;
+
+ for (i = 0; i < reserved_bot; ++i)
+ set_bit(i, bitmap->table);
+
+ return 0;
+}
+
+void hns_roce_bitmap_cleanup(struct hns_roce_bitmap *bitmap)
+{
+ kfree(bitmap->table);
+}
+
+void hns_roce_cleanup_bitmap(struct hns_roce_dev *hr_dev)
+{
+ hns_roce_cleanup_qp_table(hr_dev);
+ hns_roce_cleanup_cq_table(hr_dev);
+ hns_roce_cleanup_mr_table(hr_dev);
+ hns_roce_cleanup_pd_table(hr_dev);
+ hns_roce_cleanup_uar_table(hr_dev);
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c
index 1dc8635..f7baf82 100644
--- a/drivers/infiniband/hw/hns/hns_roce_cq.c
+++ b/drivers/infiniband/hw/hns/hns_roce_cq.c
@@ -52,3 +52,28 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type)
if (atomic_dec_and_test(&cq->refcount))
complete(&cq->free);
}
+
+int hns_roce_init_cq_table(struct hns_roce_dev *hr_dev)
+{
+ struct hns_roce_cq_table *cq_table = &hr_dev->cq_table;
+ struct device *dev = &hr_dev->pdev->dev;
+ int ret;
+
+ spin_lock_init(&cq_table->lock);
+ INIT_RADIX_TREE(&cq_table->tree, GFP_ATOMIC);
+
+ ret = hns_roce_bitmap_init(&cq_table->bitmap, hr_dev->caps.num_cqs,
+ hr_dev->caps.num_cqs - 1,
+ hr_dev->caps.reserved_cqs, 0);
+ if (ret) {
+ dev_err(dev, "init_cq_table.Failed to bitmap_init.\n");
+ return ret;
+ }
+
+ return 0;
+}
+
+void hns_roce_cleanup_cq_table(struct hns_roce_dev *hr_dev)
+{
+ hns_roce_bitmap_cleanup(&hr_dev->cq_table.bitmap);
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index eba3bca..a6e9efc 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -91,6 +91,38 @@ enum {
HNS_ROCE_CMD_SUCCESS = 1,
};
+struct hns_roce_uar {
+ u64 pfn;
+ unsigned long index;
+};
+
+struct hns_roce_bitmap {
+ /* Bitmap Traversal last a bit which is 1 */
+ unsigned long last;
+ unsigned long top;
+ unsigned long max;
+ unsigned long reserved_top;
+ unsigned long mask;
+ spinlock_t lock;
+ unsigned long *table;
+};
+
+/* Order bitmap length -- bit num compute formula: 1 << (max_order - order) */
+/* Order = 0: bitmap is biggest, order = max bitmap is least (only a bit) */
+/* Every bit repesent to a partner free/used status in bitmap */
+/*
+* Initial, bits of other bitmap are all 0 except that a bit of max_order is 1
+* Bit = 1 represent to idle and available; bit = 0: not available
+*/
+struct hns_roce_buddy {
+ /* Members point to every order level bitmap */
+ unsigned long **bits;
+ /* Represent to avail bits of the order level bitmap */
+ u32 *num_free;
+ int max_order;
+ spinlock_t lock;
+};
+
struct hns_roce_icm_table {
/* ICM type: 0 = qpc 1 = mtt 2 = cqc 3 = srq 4 = other */
u32 type;
@@ -100,13 +132,15 @@ struct hns_roce_icm_table {
unsigned long num_obj;
/*Single obj size */
unsigned long obj_size;
- int lowmem;
- int coherent;
+ int lowmem;
+ int coherent;
struct mutex mutex;
struct hns_roce_icm **icm;
};
struct hns_roce_mr_table {
+ struct hns_roce_bitmap mtpt_bitmap;
+ struct hns_roce_buddy mtt_buddy;
struct hns_roce_icm_table mtt_table;
struct hns_roce_icm_table mtpt_table;
};
@@ -124,13 +158,19 @@ struct hns_roce_cq {
struct completion free;
};
+struct hns_roce_uar_table {
+ struct hns_roce_bitmap bitmap;
+};
+
struct hns_roce_qp_table {
+ struct hns_roce_bitmap bitmap;
spinlock_t lock;
struct hns_roce_icm_table qp_table;
struct hns_roce_icm_table irrl_table;
};
struct hns_roce_cq_table {
+ struct hns_roce_bitmap bitmap;
spinlock_t lock;
struct radix_tree_root tree;
struct hns_roce_icm_table table;
@@ -262,6 +302,9 @@ struct hns_roce_hw {
struct hns_roce_dev {
struct ib_device ib_dev;
struct platform_device *pdev;
+ struct hns_roce_uar priv_uar;
+ spinlock_t sm_lock;
+ spinlock_t cq_db_lock;
spinlock_t bt_cmd_lock;
struct hns_roce_ib_iboe iboe;
@@ -277,6 +320,8 @@ struct hns_roce_dev {
u32 hw_rev;
struct hns_roce_cmdq cmd;
+ struct hns_roce_bitmap pd_bitmap;
+ struct hns_roce_uar_table uar_table;
struct hns_roce_mr_table mr_table;
struct hns_roce_cq_table cq_table;
struct hns_roce_qp_table qp_table;
@@ -299,6 +344,11 @@ static inline struct hns_roce_qp
qpn & (hr_dev->caps.num_qps - 1));
}
+int hns_roce_init_uar_table(struct hns_roce_dev *dev);
+int hns_roce_uar_alloc(struct hns_roce_dev *dev, struct hns_roce_uar *uar);
+void hns_roce_uar_free(struct hns_roce_dev *dev, struct hns_roce_uar *uar);
+void hns_roce_cleanup_uar_table(struct hns_roce_dev *dev);
+
int hns_roce_cmd_init(struct hns_roce_dev *hr_dev);
void hns_roce_cmd_cleanup(struct hns_roce_dev *hr_dev);
void hns_roce_cmd_event(struct hns_roce_dev *hr_dev, u16 token, u8 status,
@@ -306,9 +356,28 @@ void hns_roce_cmd_event(struct hns_roce_dev *hr_dev, u16 token, u8 status,
int hns_roce_cmd_use_events(struct hns_roce_dev *hr_dev);
void hns_roce_cmd_use_polling(struct hns_roce_dev *hr_dev);
+int hns_roce_init_pd_table(struct hns_roce_dev *hr_dev);
+int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev);
int hns_roce_init_eq_table(struct hns_roce_dev *hr_dev);
+int hns_roce_init_cq_table(struct hns_roce_dev *hr_dev);
+int hns_roce_init_qp_table(struct hns_roce_dev *hr_dev);
+void hns_roce_cleanup_pd_table(struct hns_roce_dev *hr_dev);
+void hns_roce_cleanup_mr_table(struct hns_roce_dev *hr_dev);
void hns_roce_cleanup_eq_table(struct hns_roce_dev *hr_dev);
+void hns_roce_cleanup_cq_table(struct hns_roce_dev *hr_dev);
+void hns_roce_cleanup_qp_table(struct hns_roce_dev *hr_dev);
+
+int hns_roce_bitmap_alloc(struct hns_roce_bitmap *bitmap, unsigned long *obj);
+void hns_roce_bitmap_free(struct hns_roce_bitmap *bitmap, unsigned long obj);
+int hns_roce_bitmap_init(struct hns_roce_bitmap *bitmap, u32 num, u32 mask,
+ u32 reserved_bot, u32 resetrved_top);
+void hns_roce_bitmap_cleanup(struct hns_roce_bitmap *bitmap);
+void hns_roce_cleanup_bitmap(struct hns_roce_dev *hr_dev);
+int hns_roce_bitmap_alloc_range(struct hns_roce_bitmap *bitmap, int cnt,
+ int align, unsigned long *obj);
+void hns_roce_bitmap_free_range(struct hns_roce_bitmap *bitmap,
+ unsigned long obj, int cnt);
void hns_roce_cq_completion(struct hns_roce_dev *hr_dev, u32 cqn);
void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type);
diff --git a/drivers/infiniband/hw/hns/hns_roce_icm.c b/drivers/infiniband/hw/hns/hns_roce_icm.c
index 43c8bad..0e854fc 100644
--- a/drivers/infiniband/hw/hns/hns_roce_icm.c
+++ b/drivers/infiniband/hw/hns/hns_roce_icm.c
@@ -339,6 +339,95 @@ int hns_roce_unmap_icm(struct hns_roce_dev *hr_dev,
return ret;
}
+int hns_roce_table_get(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, unsigned long obj)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+ int ret = 0;
+ unsigned long i;
+
+ i = (obj & (table->num_obj - 1)) / (HNS_ROCE_TABLE_CHUNK_SIZE /
+ table->obj_size);
+
+ mutex_lock(&table->mutex);
+
+ if (table->icm[i]) {
+ ++table->icm[i]->refcount;
+ goto out;
+ }
+
+ table->icm[i] = hns_roce_alloc_icm(hr_dev,
+ HNS_ROCE_TABLE_CHUNK_SIZE >> PAGE_SHIFT,
+ (table->lowmem ? GFP_KERNEL :
+ GFP_HIGHUSER) | __GFP_NOWARN,
+ table->coherent);
+ if (!table->icm[i]) {
+ ret = -ENOMEM;
+ dev_err(dev, "hns_roce_alloc_icm failed\n");
+ goto out;
+ }
+
+ /* Inform icm entry mem pa(128K/page, pa starting address)for hw */
+ if (hns_roce_map_icm(hr_dev, table, obj)) {
+ ret = -ENODEV;
+ dev_err(dev, "map icm table failed.\n");
+ goto out;
+ }
+
+ ++table->icm[i]->refcount;
+out:
+ mutex_unlock(&table->mutex);
+ return ret;
+}
+
+void hns_roce_table_put(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, unsigned long obj)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+ unsigned long i;
+
+ i = (obj & (table->num_obj - 1)) /
+ (HNS_ROCE_TABLE_CHUNK_SIZE / table->obj_size);
+
+ mutex_lock(&table->mutex);
+
+ if (--table->icm[i]->refcount == 0) {
+ /* Clear base address table */
+ if (hns_roce_unmap_icm(hr_dev, table, obj))
+ dev_warn(dev, "unmap icm table failed.\n");
+
+ hns_roce_free_icm(hr_dev, table->icm[i], table->coherent);
+ table->icm[i] = NULL;
+ }
+
+ mutex_unlock(&table->mutex);
+}
+
+int hns_roce_table_get_range(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table,
+ unsigned long start, unsigned long end)
+{
+ unsigned long inc = HNS_ROCE_TABLE_CHUNK_SIZE / table->obj_size;
+ unsigned long i = 0;
+ int ret = 0;
+
+ /* Allocate MTT entry memory according to chunk(128K) */
+ for (i = start; i <= end; i += inc) {
+ ret = hns_roce_table_get(hr_dev, table, i);
+ if (ret)
+ goto fail;
+ }
+
+ return 0;
+
+fail:
+ while (i > start) {
+ i -= inc;
+ hns_roce_table_put(hr_dev, table, i);
+ }
+ return ret;
+}
+
int hns_roce_init_icm_table(struct hns_roce_dev *hr_dev,
struct hns_roce_icm_table *table, u32 type,
unsigned long obj_size, unsigned long nobj,
diff --git a/drivers/infiniband/hw/hns/hns_roce_icm.h b/drivers/infiniband/hw/hns/hns_roce_icm.h
index 2acabe7..e1cf60b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_icm.h
+++ b/drivers/infiniband/hw/hns/hns_roce_icm.h
@@ -50,6 +50,13 @@ struct hns_roce_icm_iter {
void hns_roce_free_icm(struct hns_roce_dev *hr_dev,
struct hns_roce_icm *icm, int coherent);
+int hns_roce_table_get(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, unsigned long obj);
+void hns_roce_table_put(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, unsigned long obj);
+int hns_roce_table_get_range(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table,
+ unsigned long start, unsigned long end);
int hns_roce_init_icm_table(struct hns_roce_dev *hr_dev,
struct hns_roce_icm_table *table, u32 type,
unsigned long obj_size, unsigned long nobj,
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 3f9d686..bde7aa0 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -184,6 +184,75 @@ err_unmap_mtt:
}
/**
+* hns_roce_setup_hca - setup host channel adapter
+* @hr_dev: pointer to hns roce device
+* Return : int
+*/
+int hns_roce_setup_hca(struct hns_roce_dev *hr_dev)
+{
+ int ret;
+ struct device *dev = &hr_dev->pdev->dev;
+
+ spin_lock_init(&hr_dev->sm_lock);
+ spin_lock_init(&hr_dev->cq_db_lock);
+ spin_lock_init(&hr_dev->bt_cmd_lock);
+
+ ret = hns_roce_init_uar_table(hr_dev);
+ if (ret) {
+ dev_err(dev, "Failed to initialize uar table. aborting\n");
+ return ret;
+ }
+
+ ret = hns_roce_uar_alloc(hr_dev, &hr_dev->priv_uar);
+ if (ret) {
+ dev_err(dev, "Failed to allocate priv_uar.\n");
+ goto err_uar_table_free;
+ }
+
+ ret = hns_roce_init_pd_table(hr_dev);
+ if (ret) {
+ dev_err(dev, "Failed to init protected domain table.\n");
+ goto err_uar_alloc_free;
+ }
+
+ ret = hns_roce_init_mr_table(hr_dev);
+ if (ret) {
+ dev_err(dev, "Failed to init memory region table.\n");
+ goto err_pd_table_free;
+ }
+
+ ret = hns_roce_init_cq_table(hr_dev);
+ if (ret) {
+ dev_err(dev, "Failed to init completion queue table.\n");
+ goto err_mr_table_free;
+ }
+
+ ret = hns_roce_init_qp_table(hr_dev);
+ if (ret) {
+ dev_err(dev, "Failed to init queue pair table.\n");
+ goto err_cq_table_free;
+ }
+
+ return 0;
+
+err_cq_table_free:
+ hns_roce_cleanup_cq_table(hr_dev);
+
+err_mr_table_free:
+ hns_roce_cleanup_mr_table(hr_dev);
+
+err_pd_table_free:
+ hns_roce_cleanup_pd_table(hr_dev);
+
+err_uar_alloc_free:
+ hns_roce_uar_free(hr_dev, &hr_dev->priv_uar);
+
+err_uar_table_free:
+ hns_roce_cleanup_uar_table(hr_dev);
+ return ret;
+}
+
+/**
* hns_roce_probe - RoCE driver entrance
* @pdev: pointer to platform device
* Return : int
@@ -257,6 +326,15 @@ static int hns_roce_probe(struct platform_device *pdev)
goto error_failed_init_icm;
}
+ ret = hns_roce_setup_hca(hr_dev);
+ if (ret) {
+ dev_err(dev, "setup hca fail!\n");
+ goto error_failed_setup_hca;
+ }
+
+error_failed_setup_hca:
+ hns_roce_cleanup_icm(hr_dev);
+
error_failed_init_icm:
if (hr_dev->cmd_mod)
hns_roce_cmd_use_polling(hr_dev);
@@ -286,6 +364,7 @@ static int hns_roce_remove(struct platform_device *pdev)
{
struct hns_roce_dev *hr_dev = platform_get_drvdata(pdev);
+ hns_roce_cleanup_bitmap(hr_dev);
hns_roce_cleanup_icm(hr_dev);
if (hr_dev->cmd_mod)
diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c
new file mode 100644
index 0000000..3e0fc1b
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_mr.c
@@ -0,0 +1,187 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/platform_device.h>
+#include "hns_roce_device.h"
+
+static int hns_roce_buddy_alloc(struct hns_roce_buddy *buddy, int order,
+ unsigned long *seg)
+{
+ int o;
+ u32 m;
+
+ spin_lock(&buddy->lock);
+
+ for (o = order; o <= buddy->max_order; ++o) {
+ if (buddy->num_free[o]) {
+ m = 1 << (buddy->max_order - o);
+ *seg = find_first_bit(buddy->bits[o], m);
+ if (*seg < m)
+ goto found;
+ }
+ }
+ spin_unlock(&buddy->lock);
+ return -1;
+
+ found:
+ clear_bit(*seg, buddy->bits[o]);
+ --buddy->num_free[o];
+
+ while (o > order) {
+ --o;
+ *seg <<= 1;
+ set_bit(*seg ^ 1, buddy->bits[o]);
+ ++buddy->num_free[o];
+ }
+
+ spin_unlock(&buddy->lock);
+
+ *seg <<= order;
+ return 0;
+}
+
+static void hns_roce_buddy_free(struct hns_roce_buddy *buddy, unsigned long seg,
+ int order)
+{
+ seg >>= order;
+
+ spin_lock(&buddy->lock);
+
+ while (test_bit(seg ^ 1, buddy->bits[order])) {
+ clear_bit(seg ^ 1, buddy->bits[order]);
+ --buddy->num_free[order];
+ seg >>= 1;
+ ++order;
+ }
+
+ set_bit(seg, buddy->bits[order]);
+ ++buddy->num_free[order];
+
+ spin_unlock(&buddy->lock);
+}
+
+static int hns_roce_buddy_init(struct hns_roce_buddy *buddy, int max_order)
+{
+ int i, s;
+
+ buddy->max_order = max_order;
+ spin_lock_init(&buddy->lock);
+
+ buddy->bits = kzalloc((buddy->max_order + 1) * sizeof(long *),
+ GFP_KERNEL);
+ buddy->num_free = kzalloc((buddy->max_order + 1) * sizeof(int *),
+ GFP_KERNEL);
+ if (!buddy->bits || !buddy->num_free)
+ goto err_out;
+
+ for (i = 0; i <= buddy->max_order; ++i) {
+ s = BITS_TO_LONGS(1 << (buddy->max_order - i));
+ buddy->bits[i] = kmalloc_array(s, sizeof(long), GFP_KERNEL);
+ if (!buddy->bits[i])
+ goto err_out_free;
+
+ bitmap_zero(buddy->bits[i], 1 << (buddy->max_order - i));
+ }
+
+ set_bit(0, buddy->bits[buddy->max_order]);
+ buddy->num_free[buddy->max_order] = 1;
+
+ return 0;
+
+err_out_free:
+ for (i = 0; i <= buddy->max_order; ++i)
+ kfree(buddy->bits[i]);
+
+err_out:
+ kfree(buddy->bits);
+ kfree(buddy->num_free);
+ return -ENOMEM;
+}
+
+static void hns_roce_buddy_cleanup(struct hns_roce_buddy *buddy)
+{
+ int i;
+
+ for (i = 0; i <= buddy->max_order; ++i)
+ kfree(buddy->bits[i]);
+
+ kfree(buddy->bits);
+ kfree(buddy->num_free);
+}
+
+static int hns_roce_alloc_mtt_range(struct hns_roce_dev *hr_dev, int order,
+ unsigned long *seg)
+{
+ struct hns_roce_mr_table *mr_table = &hr_dev->mr_table;
+ int ret = 0;
+
+ ret = hns_roce_buddy_alloc(&mr_table->mtt_buddy, order, seg);
+ if (ret == -1)
+ return -1;
+
+ if (hns_roce_table_get_range(hr_dev, &mr_table->mtt_table, *seg,
+ *seg + (1 << order) - 1)) {
+ hns_roce_buddy_free(&mr_table->mtt_buddy, *seg, order);
+ return -1;
+ }
+
+ return 0;
+}
+
+int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev)
+{
+ struct hns_roce_mr_table *mr_table = &hr_dev->mr_table;
+ struct device *dev = &hr_dev->pdev->dev;
+ unsigned long first_seg;
+ int ret = 0;
+
+ ret = hns_roce_bitmap_init(&mr_table->mtpt_bitmap,
+ hr_dev->caps.num_mtpts,
+ hr_dev->caps.num_mtpts - 1,
+ hr_dev->caps.reserved_mrws, 0);
+ if (ret)
+ return ret;
+
+ ret = hns_roce_buddy_init(&mr_table->mtt_buddy,
+ ilog2(hr_dev->caps.num_mtt_segs));
+ if (ret)
+ goto err_buddy;
+
+ if (hr_dev->caps.reserved_mtts) {
+ if (hns_roce_alloc_mtt_range(hr_dev,
+ fls(hr_dev->caps.reserved_mtts - 1),
+ &first_seg) == -1) {
+ dev_err(dev, "MTT table of order %d is too small.\n",
+ mr_table->mtt_buddy.max_order);
+ ret = -ENOMEM;
+ goto err_reserve_mtts;
+ }
+ }
+
+ return 0;
+
+err_reserve_mtts:
+ hns_roce_buddy_cleanup(&mr_table->mtt_buddy);
+
+err_buddy:
+ hns_roce_bitmap_cleanup(&mr_table->mtpt_bitmap);
+ return ret;
+}
+
+void hns_roce_cleanup_mr_table(struct hns_roce_dev *hr_dev)
+{
+ struct hns_roce_mr_table *mr_table = &hr_dev->mr_table;
+
+ hns_roce_buddy_cleanup(&mr_table->mtt_buddy);
+ hns_roce_bitmap_cleanup(&mr_table->mtpt_bitmap);
+}
+
diff --git a/drivers/infiniband/hw/hns/hns_roce_pd.c b/drivers/infiniband/hw/hns/hns_roce_pd.c
new file mode 100644
index 0000000..90b0f33
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_pd.c
@@ -0,0 +1,65 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <asm/page.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <rdma/ib_smi.h>
+#include <rdma/ib_umem.h>
+#include <rdma/ib_user_verbs.h>
+#include "hns_roce_common.h"
+#include "hns_roce_device.h"
+
+int hns_roce_init_pd_table(struct hns_roce_dev *hr_dev)
+{
+ return hns_roce_bitmap_init(&hr_dev->pd_bitmap, hr_dev->caps.num_pds,
+ hr_dev->caps.num_pds - 1,
+ hr_dev->caps.reserved_pds, 0);
+}
+
+void hns_roce_cleanup_pd_table(struct hns_roce_dev *hr_dev)
+{
+ hns_roce_bitmap_cleanup(&hr_dev->pd_bitmap);
+}
+
+int hns_roce_uar_alloc(struct hns_roce_dev *hr_dev, struct hns_roce_uar *uar)
+{
+ struct resource *res;
+ int ret = 0;
+ /* Using bitmap to manager UAR index */
+ ret = hns_roce_bitmap_alloc(&hr_dev->uar_table.bitmap, &uar->index);
+ if (ret == -1)
+ return -ENOMEM;
+
+ uar->index = (uar->index - 1) % hr_dev->caps.phy_num_uars + 1;
+
+ res = platform_get_resource(hr_dev->pdev, IORESOURCE_MEM, 0);
+ uar->pfn = ((res->start) >> PAGE_SHIFT) + uar->index;
+
+ return 0;
+}
+
+void hns_roce_uar_free(struct hns_roce_dev *hr_dev, struct hns_roce_uar *uar)
+{
+ hns_roce_bitmap_free(&hr_dev->uar_table.bitmap, uar->index);
+}
+
+int hns_roce_init_uar_table(struct hns_roce_dev *hr_dev)
+{
+ return hns_roce_bitmap_init(&hr_dev->uar_table.bitmap,
+ hr_dev->caps.num_uars,
+ hr_dev->caps.num_uars - 1,
+ hr_dev->caps.reserved_uars, 0);
+}
+
+void hns_roce_cleanup_uar_table(struct hns_roce_dev *hr_dev)
+{
+ hns_roce_bitmap_cleanup(&hr_dev->uar_table.bitmap);
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c
index e0e41ca..e8d396b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_qp.c
+++ b/drivers/infiniband/hw/hns/hns_roce_qp.c
@@ -13,6 +13,8 @@
#include <rdma/ib_pack.h>
#include "hns_roce_device.h"
+#define SQP_NUM 12
+
void hns_roce_qp_event(struct hns_roce_dev *hr_dev, u32 qpn, int event_type)
{
struct hns_roce_qp_table *qp_table = &hr_dev->qp_table;
@@ -37,3 +39,31 @@ void hns_roce_qp_event(struct hns_roce_dev *hr_dev, u32 qpn, int event_type)
if (atomic_dec_and_test(&qp->refcount))
complete(&qp->free);
}
+
+int hns_roce_init_qp_table(struct hns_roce_dev *hr_dev)
+{
+ struct hns_roce_qp_table *qp_table = &hr_dev->qp_table;
+ int reserved_from_top = 0;
+ int ret;
+
+ spin_lock_init(&qp_table->lock);
+ INIT_RADIX_TREE(&hr_dev->qp_table_tree, GFP_ATOMIC);
+
+ /* A port include two SQP, six port total 12 */
+ ret = hns_roce_bitmap_init(&qp_table->bitmap, hr_dev->caps.num_qps,
+ hr_dev->caps.num_qps - 1,
+ hr_dev->caps.sqp_start + SQP_NUM,
+ reserved_from_top);
+ if (ret) {
+ dev_err(&hr_dev->pdev->dev, "qp bitmap init failed!error=%d\n",
+ ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+void hns_roce_cleanup_qp_table(struct hns_roce_dev *hr_dev)
+{
+ hns_roce_bitmap_cleanup(&hr_dev->qp_table.bitmap);
+}
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 16/21] IB/hns: Add ah operations support |
| Message-ID | <rsSVC-2Qt-59@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch was for implementing of address handle operations.
It includes three verbs that create ah, query ah and destroy
ah. They is completed independently by RoCE driver.
Signed-off-by: Wei Hu <xavier.huwei@huawei.com>
Signed-off-by: Nenglong Zhao <zhaonenglong@hisilicon.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
---
drivers/infiniband/hw/hns/hns_roce_ah.c | 109 ++++++++++++++++++++++++++++
drivers/infiniband/hw/hns/hns_roce_device.h | 31 +++++++-
drivers/infiniband/hw/hns/hns_roce_main.c | 5 ++
3 files changed, 144 insertions(+), 1 deletion(-)
create mode 100644 drivers/infiniband/hw/hns/hns_roce_ah.c
diff --git a/drivers/infiniband/hw/hns/hns_roce_ah.c b/drivers/infiniband/hw/hns/hns_roce_ah.c
new file mode 100644
index 0000000..9d0eb61
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_ah.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/inet.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <rdma/ib_addr.h>
+#include <rdma/ib_cache.h>
+#include "hns_roce_common.h"
+#include "hns_roce_device.h"
+
+#define HNS_ROCE_PORT_NUM_SHIFT 24
+#define HNS_ROCE_VLAN_SL_BIT_MASK 7
+#define HNS_ROCE_VLAN_SL_SHIFT 13
+
+struct ib_ah *hns_roce_create_ah(struct ib_pd *ibpd, struct ib_ah_attr *ah_attr)
+{
+ struct hns_roce_dev *hr_dev = to_hr_dev(ibpd->device);
+ struct device *dev = &hr_dev->pdev->dev;
+ struct ib_gid_attr gid_attr;
+ struct hns_roce_ah *ah;
+ u16 vlan_tag = 0xffff;
+ struct in6_addr in6;
+ union ib_gid sgid;
+ int ret;
+
+ ah = kzalloc(sizeof(*ah), GFP_ATOMIC);
+ if (!ah)
+ return ERR_PTR(-ENOMEM);
+
+ /* Get mac address */
+ memcpy(&in6, ah_attr->grh.dgid.raw, sizeof(ah_attr->grh.dgid.raw));
+ if (rdma_is_multicast_addr(&in6))
+ rdma_get_mcast_mac(&in6, ah->av.mac);
+ else
+ memcpy(ah->av.mac, ah_attr->dmac, sizeof(ah_attr->dmac));
+
+ /* Get source gid */
+ ret = ib_get_cached_gid(ibpd->device, ah_attr->port_num,
+ ah_attr->grh.sgid_index, &sgid, &gid_attr);
+ if (ret) {
+ dev_err(dev, "get sgid failed! ret = %d\n", ret);
+ kfree(ah);
+ return ERR_PTR(ret);
+ }
+
+ if (gid_attr.ndev) {
+ if (is_vlan_dev(gid_attr.ndev))
+ vlan_tag = vlan_dev_vlan_id(gid_attr.ndev);
+ dev_put(gid_attr.ndev);
+ }
+
+ if (vlan_tag < 0x1000)
+ vlan_tag |= (ah_attr->sl & HNS_ROCE_VLAN_SL_BIT_MASK) <<
+ HNS_ROCE_VLAN_SL_SHIFT;
+
+ ah->av.port_pd = cpu_to_be32(to_hr_pd(ibpd)->pdn | (ah_attr->port_num <<
+ HNS_ROCE_PORT_NUM_SHIFT));
+ ah->av.gid_index = ah_attr->grh.sgid_index;
+ ah->av.vlan = cpu_to_le16(vlan_tag);
+ dev_dbg(dev, "gid_index = 0x%x,vlan = 0x%x\n", ah->av.gid_index,
+ ah->av.vlan);
+
+ if (ah_attr->static_rate)
+ ah->av.stat_rate = IB_RATE_10_GBPS;
+
+ memcpy(ah->av.dgid, ah_attr->grh.dgid.raw, HNS_ROCE_GID_SIZE);
+ ah->av.sl_tclass_flowlabel = cpu_to_le32(ah_attr->sl <<
+ HNS_ROCE_SL_SHIFT);
+
+ return &ah->ibah;
+}
+
+int hns_roce_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr)
+{
+ struct hns_roce_ah *ah = to_hr_ah(ibah);
+
+ memset(ah_attr, 0, sizeof(*ah_attr));
+
+ ah_attr->sl = le32_to_cpu(ah->av.sl_tclass_flowlabel) >>
+ HNS_ROCE_SL_SHIFT;
+ ah_attr->port_num = le32_to_cpu(ah->av.port_pd) >>
+ HNS_ROCE_PORT_NUM_SHIFT;
+ ah_attr->static_rate = ah->av.stat_rate;
+ ah_attr->ah_flags = IB_AH_GRH;
+ ah_attr->grh.traffic_class = le32_to_cpu(ah->av.sl_tclass_flowlabel) >>
+ HNS_ROCE_TCLASS_SHIFT;
+ ah_attr->grh.flow_label = le32_to_cpu(ah->av.sl_tclass_flowlabel) &
+ HNS_ROCE_FLOW_LABLE_MASK;
+ ah_attr->grh.hop_limit = ah->av.hop_limit;
+ ah_attr->grh.sgid_index = ah->av.gid_index;
+ memcpy(ah_attr->grh.dgid.raw, ah->av.dgid, HNS_ROCE_GID_SIZE);
+
+ return 0;
+}
+
+int hns_roce_destroy_ah(struct ib_ah *ah)
+{
+ kfree(to_hr_ah(ah));
+
+ return 0;
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 4640c34..df2bc00 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -24,7 +24,6 @@
#define MAC_ADDR_OCTET_NUM 6
#define HNS_ROCE_MAX_MSG_LEN 0x80000000
-
#define HNS_ROCE_BA_SIZE (32 * 4096)
#define HNS_ROCE_MAX_IRQ_NUM 34
@@ -35,6 +34,11 @@
#define HNS_ROCE_AEQE_VEC_NUM 1
#define HNS_ROCE_AEQE_OF_VEC_NUM 1
+/* 4G/4K = 1M */
+#define HNS_ROCE_SL_SHIFT 29
+#define HNS_ROCE_TCLASS_SHIFT 20
+#define HNS_ROCE_FLOW_LABLE_MASK 0xfffff
+
#define HNS_ROCE_MAX_PORTS 6
#define HNS_ROCE_MAX_GID_NUM 16
#define HNS_ROCE_GID_SIZE 16
@@ -208,6 +212,22 @@ struct hns_roce_raq_table {
struct hns_roce_buf_list *e_raq_buf;
};
+struct hns_roce_av {
+ __le32 port_pd;
+ u8 gid_index;
+ u8 stat_rate;
+ u8 hop_limit;
+ __le32 sl_tclass_flowlabel;
+ u8 dgid[HNS_ROCE_GID_SIZE];
+ u8 mac[6];
+ __le16 vlan;
+};
+
+struct hns_roce_ah {
+ struct ib_ah ibah;
+ struct hns_roce_av av;
+};
+
struct hns_roce_cmd_context {
struct completion done;
int result;
@@ -394,6 +414,11 @@ static inline struct hns_roce_pd *to_hr_pd(struct ib_pd *ibpd)
return container_of(ibpd, struct hns_roce_pd, ibpd);
}
+static inline struct hns_roce_ah *to_hr_ah(struct ib_ah *ibah)
+{
+ return container_of(ibah, struct hns_roce_ah, ibah);
+}
+
static inline void hns_roce_write64_k(__be32 val[2], void __iomem *dest)
{
__raw_writeq(*(u64 *) val, dest);
@@ -441,6 +466,10 @@ int hns_roce_bitmap_alloc_range(struct hns_roce_bitmap *bitmap, int cnt,
void hns_roce_bitmap_free_range(struct hns_roce_bitmap *bitmap,
unsigned long obj, int cnt);
+struct ib_ah *hns_roce_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr);
+int hns_roce_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr);
+int hns_roce_destroy_ah(struct ib_ah *ah);
+
struct ib_pd *hns_roce_alloc_pd(struct ib_device *ib_dev,
struct ib_ucontext *context,
struct ib_udata *udata);
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 2352120..7eb88bb 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -605,6 +605,11 @@ int hns_roce_register_device(struct hns_roce_dev *hr_dev)
ib_dev->alloc_pd = hns_roce_alloc_pd;
ib_dev->dealloc_pd = hns_roce_dealloc_pd;
+ /* AH */
+ ib_dev->create_ah = hns_roce_create_ah;
+ ib_dev->query_ah = hns_roce_query_ah;
+ ib_dev->destroy_ah = hns_roce_destroy_ah;
+
ret = ib_register_device(ib_dev, NULL);
if (ret) {
dev_err(dev, "ib_register_device failed!\n");
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 14/21] IB/hns: Add operations support for IB device and port |
| Message-ID | <rsSVC-2Qt-63@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch mainly registered some relative verbs for the kernel.
These operation functions will be called by user. For example:
1. modify device
2. query device
3. query_port
4. modify_port
and so on.
Signed-off-by: Wei Hu <xavier.huwei@huawei.com>
Signed-off-by: Nenglong Zhao <zhaonenglong@hisilicon.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
---
drivers/infiniband/hw/hns/hns_roce_common.h | 4 +
drivers/infiniband/hw/hns/hns_roce_device.h | 21 +++
drivers/infiniband/hw/hns/hns_roce_main.c | 228 ++++++++++++++++++++++++++++
drivers/infiniband/hw/hns/hns_roce_user.h | 17 +++
4 files changed, 270 insertions(+)
create mode 100644 drivers/infiniband/hw/hns/hns_roce_user.h
diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h
index 45b5804..135203e 100644
--- a/drivers/infiniband/hw/hns/hns_roce_common.h
+++ b/drivers/infiniband/hw/hns/hns_roce_common.h
@@ -10,6 +10,10 @@
#ifndef _HNS_ROCE_COMMON_H
#define _HNS_ROCE_COMMON_H
+#ifndef assert
+#define assert(cond)
+#endif
+
#define roce_writel(value, addr) writel((value), (addr))
#define roce_readl(addr) readl((addr))
#define roce_raw_write(value, addr) \
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 240a801..a3ad4a9 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -22,6 +22,8 @@
#define DRV_NAME "hns_roce"
#define MAC_ADDR_OCTET_NUM 6
+#define HNS_ROCE_MAX_MSG_LEN 0x80000000
+
#define HNS_ROCE_BA_SIZE (32 * 4096)
@@ -35,7 +37,10 @@
#define HNS_ROCE_MAX_PORTS 6
#define HNS_ROCE_MAX_GID_NUM 16
+#define HNS_ROCE_GID_SIZE 16
+#define PKEY_ID 0xffff
+#define NODE_DESC_SIZE 64
#define ADDR_SHIFT_12 12
#define ADDR_SHIFT_32 32
#define ADDR_SHIFT_44 44
@@ -106,6 +111,11 @@ struct hns_roce_uar {
unsigned long index;
};
+struct hns_roce_ucontext {
+ struct ib_ucontext ibucontext;
+ struct hns_roce_uar uar;
+};
+
struct hns_roce_bitmap {
/* Bitmap Traversal last a bit which is 1 */
unsigned long last;
@@ -363,6 +373,17 @@ struct hns_roce_dev {
struct hns_roce_hw *hw;
};
+static inline struct hns_roce_dev *to_hr_dev(struct ib_device *ib_dev)
+{
+ return container_of(ib_dev, struct hns_roce_dev, ib_dev);
+}
+
+static inline struct hns_roce_ucontext
+ *to_hr_ucontext(struct ib_ucontext *ibucontext)
+{
+ return container_of(ibucontext, struct hns_roce_ucontext, ibucontext);
+}
+
static inline void hns_roce_write64_k(__be32 val[2], void __iomem *dest)
{
__raw_writeq(*(u64 *) val, dest);
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index c115cb3..4a39444 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -41,6 +41,7 @@
#include <rdma/ib_verbs.h>
#include "hns_roce_common.h"
#include "hns_roce_device.h"
+#include "hns_roce_user.h"
#include "hns_roce_icm.h"
/**
@@ -341,6 +342,217 @@ int hns_roce_setup_mtu_gids(struct hns_roce_dev *hr_dev)
return ret;
}
+static int hns_roce_query_device(struct ib_device *ib_dev,
+ struct ib_device_attr *props,
+ struct ib_udata *uhw)
+{
+ struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
+
+ memset(props, 0, sizeof(*props));
+
+ props->fw_ver = hr_dev->fw_ver;
+ props->sys_image_guid = hr_dev->sys_image_guid;
+ props->max_mr_size = (u64)(~(0ULL));
+ props->page_size_cap = hr_dev->caps.page_size_cap;
+ props->vendor_id = hr_dev->vendor_id;
+ props->vendor_part_id = hr_dev->vendor_part_id;
+ props->hw_ver = hr_dev->hw_rev;
+ props->max_qp = hr_dev->caps.num_qps;
+ props->max_qp_wr = hr_dev->caps.max_wqes;
+ props->device_cap_flags = IB_DEVICE_PORT_ACTIVE_EVENT |
+ IB_DEVICE_RC_RNR_NAK_GEN |
+ IB_DEVICE_LOCAL_DMA_LKEY;
+ props->max_sge = hr_dev->caps.max_sq_sg;
+ props->max_sge_rd = 1;
+ props->max_cq = hr_dev->caps.num_cqs;
+ props->max_cqe = hr_dev->caps.max_cqes;
+ props->max_mr = hr_dev->caps.num_mtpts;
+ props->max_pd = hr_dev->caps.num_pds;
+ props->max_qp_rd_atom = hr_dev->caps.max_qp_dest_rdma;
+ props->max_qp_init_rd_atom = hr_dev->caps.max_qp_init_rdma;
+ props->atomic_cap = IB_ATOMIC_NONE;
+ props->max_pkeys = 1;
+ props->local_ca_ack_delay = hr_dev->caps.local_ca_ack_delay;
+
+ return 0;
+}
+
+static int hns_roce_query_port(struct ib_device *ib_dev, u8 port_num,
+ struct ib_port_attr *props)
+{
+ struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
+ struct device *dev = &hr_dev->pdev->dev;
+ struct net_device *net_dev;
+ unsigned long flags;
+ enum ib_mtu mtu;
+ u8 port;
+
+ assert(port_num > 0);
+ port = port_num - 1;
+
+ memset(props, 0, sizeof(*props));
+
+ props->max_mtu = hr_dev->caps.max_mtu;
+ props->gid_tbl_len = hr_dev->caps.gid_table_len[port];
+ props->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_REINIT_SUP |
+ IB_PORT_VENDOR_CLASS_SUP |
+ IB_PORT_BOOT_MGMT_SUP;
+ props->max_msg_sz = HNS_ROCE_MAX_MSG_LEN;
+ props->pkey_tbl_len = 1;
+ props->active_width = IB_WIDTH_4X;
+ props->active_speed = 1;
+
+ spin_lock_irqsave(&hr_dev->iboe.lock, flags);
+
+ net_dev = hr_dev->iboe.netdevs[port];
+ if (!net_dev) {
+ spin_unlock_irqrestore(&hr_dev->iboe.lock, flags);
+ dev_err(dev, "find netdev %d failed!\r\n", port);
+ return -EINVAL;
+ }
+
+ mtu = iboe_get_mtu(net_dev->mtu);
+ props->active_mtu = mtu ? min(props->max_mtu, mtu) : IB_MTU_256;
+ props->state = (netif_running(net_dev) && netif_carrier_ok(net_dev)) ?
+ IB_PORT_ACTIVE : IB_PORT_DOWN;
+ props->phys_state = (props->state == IB_PORT_ACTIVE) ? 5 : 3;
+
+ spin_unlock_irqrestore(&hr_dev->iboe.lock, flags);
+
+ return 0;
+}
+
+static enum rdma_link_layer hns_roce_get_link_layer(struct ib_device *device,
+ u8 port_num)
+{
+ return IB_LINK_LAYER_ETHERNET;
+}
+
+static int hns_roce_query_gid(struct ib_device *ib_dev, u8 port_num, int index,
+ union ib_gid *gid)
+{
+ struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
+ struct device *dev = &hr_dev->pdev->dev;
+ u8 gid_idx = 0;
+ u8 port;
+
+ if (port_num < 1 || port_num > hr_dev->caps.num_ports ||
+ index >= hr_dev->caps.gid_table_len[port_num - 1]) {
+ dev_err(dev,
+ "port_num %d index %d illegal! correct range: port_num 1~%d index 0~%d!\n",
+ port_num, index, hr_dev->caps.num_ports,
+ hr_dev->caps.gid_table_len[port_num - 1] - 1);
+ return -EINVAL;
+ }
+
+ port = port_num - 1;
+ gid_idx = hns_get_gid_index(hr_dev, port, index);
+ if (gid_idx >= HNS_ROCE_MAX_GID_NUM) {
+ dev_err(dev, "port_num %d index %d illegal! total gid num %d!\n",
+ port_num, index, HNS_ROCE_MAX_GID_NUM);
+ return -EINVAL;
+ }
+
+ memcpy(gid->raw, hr_dev->iboe.gid_table[gid_idx].raw,
+ HNS_ROCE_GID_SIZE);
+
+ return 0;
+}
+
+static int hns_roce_query_pkey(struct ib_device *ib_dev, u8 port, u16 index,
+ u16 *pkey)
+{
+ *pkey = PKEY_ID;
+
+ return 0;
+}
+
+static int hns_roce_modify_device(struct ib_device *ib_dev, int mask,
+ struct ib_device_modify *props)
+{
+ unsigned long flags;
+
+ if (mask & ~IB_DEVICE_MODIFY_NODE_DESC)
+ return -EOPNOTSUPP;
+
+ if (mask & IB_DEVICE_MODIFY_NODE_DESC) {
+ spin_lock_irqsave(&to_hr_dev(ib_dev)->sm_lock, flags);
+ memcpy(ib_dev->node_desc, props->node_desc, NODE_DESC_SIZE);
+ spin_unlock_irqrestore(&to_hr_dev(ib_dev)->sm_lock, flags);
+ }
+
+ return 0;
+}
+
+static int hns_roce_modify_port(struct ib_device *ib_dev, u8 port_num, int mask,
+ struct ib_port_modify *props)
+{
+ return 0;
+}
+
+static struct ib_ucontext *hns_roce_alloc_ucontext(struct ib_device *ib_dev,
+ struct ib_udata *udata)
+{
+ int ret = 0;
+ struct hns_roce_ucontext *context;
+ struct hns_roce_ib_alloc_ucontext_resp resp;
+ struct hns_roce_dev *hr_dev = to_hr_dev(ib_dev);
+
+ resp.qp_tab_size = hr_dev->caps.num_qps;
+
+ context = kmalloc(sizeof(*context), GFP_KERNEL);
+ if (!context)
+ return ERR_PTR(-ENOMEM);
+
+ ret = hns_roce_uar_alloc(hr_dev, &context->uar);
+ if (ret)
+ goto _error_fail_uar_alloc;
+
+ ret = ib_copy_to_udata(udata, &resp, sizeof(resp));
+ if (ret)
+ goto _error_fail_copy_to_udata;
+
+ return &context->ibucontext;
+
+_error_fail_copy_to_udata:
+hns_roce_uar_free(hr_dev, &context->uar);
+
+_error_fail_uar_alloc:
+ kfree(context);
+
+ return ERR_PTR(ret);
+}
+
+static int hns_roce_dealloc_ucontext(struct ib_ucontext *ibcontext)
+{
+ struct hns_roce_ucontext *context = to_hr_ucontext(ibcontext);
+
+ hns_roce_uar_free(to_hr_dev(ibcontext->device), &context->uar);
+ kfree(context);
+
+ return 0;
+}
+
+static int hns_roce_mmap(struct ib_ucontext *context,
+ struct vm_area_struct *vma)
+{
+ if (((vma->vm_end - vma->vm_start) % PAGE_SIZE) != 0)
+ return -EINVAL;
+
+ if (vma->vm_pgoff == 0) {
+ vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
+ if (io_remap_pfn_range(vma, vma->vm_start,
+ to_hr_ucontext(context)->uar.pfn,
+ PAGE_SIZE, vma->vm_page_prot))
+ return -EAGAIN;
+
+ } else {
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
void hns_roce_unregister_device(struct hns_roce_dev *hr_dev)
{
struct hns_roce_ib_iboe *iboe = &hr_dev->iboe;
@@ -370,6 +582,22 @@ int hns_roce_register_device(struct hns_roce_dev *hr_dev)
ib_dev->local_dma_lkey = hr_dev->caps.reserved_lkey;
ib_dev->num_comp_vectors = hr_dev->caps.num_comp_vectors;
ib_dev->uverbs_abi_ver = 1;
+ ib_dev->uverbs_cmd_mask =
+ (1ULL << IB_USER_VERBS_CMD_GET_CONTEXT) |
+ (1ULL << IB_USER_VERBS_CMD_QUERY_DEVICE) |
+ (1ULL << IB_USER_VERBS_CMD_QUERY_PORT);
+
+ /* HCA||device||port */
+ ib_dev->modify_device = hns_roce_modify_device;
+ ib_dev->query_device = hns_roce_query_device;
+ ib_dev->query_port = hns_roce_query_port;
+ ib_dev->modify_port = hns_roce_modify_port;
+ ib_dev->get_link_layer = hns_roce_get_link_layer;
+ ib_dev->query_gid = hns_roce_query_gid;
+ ib_dev->query_pkey = hns_roce_query_pkey;
+ ib_dev->alloc_ucontext = hns_roce_alloc_ucontext;
+ ib_dev->dealloc_ucontext = hns_roce_dealloc_ucontext;
+ ib_dev->mmap = hns_roce_mmap;
ret = ib_register_device(ib_dev, NULL);
if (ret) {
diff --git a/drivers/infiniband/hw/hns/hns_roce_user.h b/drivers/infiniband/hw/hns/hns_roce_user.h
new file mode 100644
index 0000000..457e77a
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_user.h
@@ -0,0 +1,17 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#ifndef _HNS_ROCE_USER_H
+#define _HNS_ROCE_USER_H
+
+struct hns_roce_ib_alloc_ucontext_resp {
+ __u32 qp_tab_size;
+};
+
+#endif /*_HNS_ROCE_USER_H */
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 02/21] devicetree: bindings: IB: Add binding document for HiSilicon RoCE |
| Message-ID | <rsSVC-2Qt-65@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch added DTS binding document for HiSilicon RoCE driver.
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Signed-off-by: Wei Hu <xavier.huwei@huawei.com>
---
.../bindings/infiniband/hisilicon-hns-roce.txt | 107 +++++++++++++++++++++
1 file changed, 107 insertions(+)
create mode 100644 Documentation/devicetree/bindings/infiniband/hisilicon-hns-roce.txt
diff --git a/Documentation/devicetree/bindings/infiniband/hisilicon-hns-roce.txt b/Documentation/devicetree/bindings/infiniband/hisilicon-hns-roce.txt
new file mode 100644
index 0000000..2c59ed9
--- /dev/null
+++ b/Documentation/devicetree/bindings/infiniband/hisilicon-hns-roce.txt
@@ -0,0 +1,107 @@
+HiSilicon RoCE DT description
+
+HiSilicon RoCE engine is a part of network subsystem.
+It works depending on other part of network wubsytem, such as, gmac and
+dsa fabric.
+
+Additional properties are described here:
+
+Required properties:
+- compatible: Should contain "hisilicon,hns-roce-v1".
+- reg: Physical base address of the roce driver and
+length of memory mapped region.
+- eth-handle: phandle, specifies a reference to a node
+representing a ethernet device.
+- dsaf-handle: phandle, specifies a reference to a node
+representing a dsaf device.
+- #address-cells: must be 2
+- #size-cells: must be 2
+Optional properties:
+- dma-coherent: Present if DMA operations are coherent.
+- interrupt-parent: the interrupt parent of this device.
+- interrupts: should contain 32 completion event irq,1 async event irq
+and 1 event overflow irq.
+- interrupt-names:should be one of 34 irqs for roce device
+ - hns-roce-comp-0 ~ hns-roce-comp-31: 32 complete event irq
+ - hns-roce-async: 1 async event irq
+ - hns-roce-common: named common exception warning irq
+Example:
+ infiniband@c4000000 {
+ compatible = "hisilicon,hns-roce-v1";
+ reg = <0x0 0xc4000000 0x0 0x100000>;
+ dma-coherent;
+ eth-handle = <ð2 ð3 ð4 ð5 ð6 ð7>;
+ dsaf-handle = <&soc0_dsa>;
+ #address-cells = <2>;
+ #size-cells = <2>;
+ interrupt-parent = <&mbigen_dsa>;
+ interrupts = <722 1>,
+ <723 1>,
+ <724 1>,
+ <725 1>,
+ <726 1>,
+ <727 1>,
+ <728 1>,
+ <729 1>,
+ <730 1>,
+ <731 1>,
+ <732 1>,
+ <733 1>,
+ <734 1>,
+ <735 1>,
+ <736 1>,
+ <737 1>,
+ <738 1>,
+ <739 1>,
+ <740 1>,
+ <741 1>,
+ <742 1>,
+ <743 1>,
+ <744 1>,
+ <745 1>,
+ <746 1>,
+ <747 1>,
+ <748 1>,
+ <749 1>,
+ <750 1>,
+ <751 1>,
+ <752 1>,
+ <753 1>,
+ <785 1>,
+ <754 4>;
+
+ interrupt-names = "hns-roce-comp-0",
+ "hns-roce-comp-1",
+ "hns-roce-comp-2",
+ "hns-roce-comp-3",
+ "hns-roce-comp-4",
+ "hns-roce-comp-5",
+ "hns-roce-comp-6",
+ "hns-roce-comp-7",
+ "hns-roce-comp-8",
+ "hns-roce-comp-9",
+ "hns-roce-comp-10",
+ "hns-roce-comp-11",
+ "hns-roce-comp-12",
+ "hns-roce-comp-13",
+ "hns-roce-comp-14",
+ "hns-roce-comp-15",
+ "hns-roce-comp-16",
+ "hns-roce-comp-17",
+ "hns-roce-comp-18",
+ "hns-roce-comp-19",
+ "hns-roce-comp-20",
+ "hns-roce-comp-21",
+ "hns-roce-comp-22",
+ "hns-roce-comp-23",
+ "hns-roce-comp-24",
+ "hns-roce-comp-25",
+ "hns-roce-comp-26",
+ "hns-roce-comp-27",
+ "hns-roce-comp-28",
+ "hns-roce-comp-29",
+ "hns-roce-comp-30",
+ "hns-roce-comp-31",
+ "hns-roce-async",
+ "hns-roce-common";
+ };
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:10 +0200 |
| Subject | [PATCH v6 08/21] IB/hns: Add icm support |
| Message-ID | <rsSVD-2Qt-69@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch mainly added icm support for RoCE. It initializes icm
which managers the relative memory blocks for RoCE. The data
structures of RoCE will be located in it. For example, CQ table,
QP table and MTPT table so on.
Signed-off-by: Wei Hu <xavier.huwei@huawei.com>
Signed-off-by: Nenglong Zhao <zhaonenglong@hisilicon.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
---
drivers/infiniband/hw/hns/hns_roce_common.h | 19 ++
drivers/infiniband/hw/hns/hns_roce_device.h | 30 ++
drivers/infiniband/hw/hns/hns_roce_icm.c | 437 ++++++++++++++++++++++++++++
drivers/infiniband/hw/hns/hns_roce_icm.h | 95 ++++++
drivers/infiniband/hw/hns/hns_roce_main.c | 84 ++++++
5 files changed, 665 insertions(+)
create mode 100644 drivers/infiniband/hw/hns/hns_roce_icm.c
create mode 100644 drivers/infiniband/hw/hns/hns_roce_icm.h
diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h
index 257a7e5..fcac5c4 100644
--- a/drivers/infiniband/hw/hns/hns_roce_common.h
+++ b/drivers/infiniband/hw/hns/hns_roce_common.h
@@ -30,6 +30,22 @@
#define roce_set_bit(origin, shift, val) \
roce_set_field((origin), (1ul << (shift)), (shift), (val))
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_S 0
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_M \
+ (((1UL << 19) - 1) << ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_S)
+
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_S 19
+
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S 20
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M \
+ (((1UL << 2) - 1) << ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S)
+
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_S 22
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_M \
+ (((1UL << 5) - 1) << ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_S)
+
+#define ROCEE_BT_CMD_H_ROCEE_BT_CMD_HW_SYNS_S 31
+
#define ROCEE_CAEP_AEQC_AEQE_SHIFT_CAEP_AEQC_STATE_S 0
#define ROCEE_CAEP_AEQC_AEQE_SHIFT_CAEP_AEQC_STATE_M \
(((1UL << 2) - 1) << ROCEE_CAEP_AEQC_AEQE_SHIFT_CAEP_AEQC_STATE_S)
@@ -70,6 +86,8 @@
#define ROCEE_SYS_IMAGE_GUID_L_REG 0xC
#define ROCEE_SYS_IMAGE_GUID_H_REG 0x10
+#define ROCEE_BT_CMD_H_REG 0x204
+
#define ROCEE_CAEP_AEQE_CONS_IDX_REG 0x3AC
#define ROCEE_CAEP_CEQC_CONS_IDX_0_REG 0x3BC
@@ -82,6 +100,7 @@
#define ROCEE_CAEP_CE_INTERVAL_CFG_REG 0x190
#define ROCEE_CAEP_CE_BURST_NUM_CFG_REG 0x194
+#define ROCEE_BT_CMD_L_REG 0x200
#define ROCEE_MB1_REG 0x210
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 5b2378d..eba3bca 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -91,6 +91,26 @@ enum {
HNS_ROCE_CMD_SUCCESS = 1,
};
+struct hns_roce_icm_table {
+ /* ICM type: 0 = qpc 1 = mtt 2 = cqc 3 = srq 4 = other */
+ u32 type;
+ /* ICM array elment num */
+ unsigned long num_icm;
+ /* ICM entry record obj total num */
+ unsigned long num_obj;
+ /*Single obj size */
+ unsigned long obj_size;
+ int lowmem;
+ int coherent;
+ struct mutex mutex;
+ struct hns_roce_icm **icm;
+};
+
+struct hns_roce_mr_table {
+ struct hns_roce_icm_table mtt_table;
+ struct hns_roce_icm_table mtpt_table;
+};
+
struct hns_roce_buf_list {
void *buf;
dma_addr_t map;
@@ -106,11 +126,14 @@ struct hns_roce_cq {
struct hns_roce_qp_table {
spinlock_t lock;
+ struct hns_roce_icm_table qp_table;
+ struct hns_roce_icm_table irrl_table;
};
struct hns_roce_cq_table {
spinlock_t lock;
struct radix_tree_root tree;
+ struct hns_roce_icm_table table;
};
struct hns_roce_cmd_context {
@@ -239,6 +262,7 @@ struct hns_roce_hw {
struct hns_roce_dev {
struct ib_device ib_dev;
struct platform_device *pdev;
+ spinlock_t bt_cmd_lock;
struct hns_roce_ib_iboe iboe;
int irq[HNS_ROCE_MAX_IRQ_NUM];
@@ -253,6 +277,7 @@ struct hns_roce_dev {
u32 hw_rev;
struct hns_roce_cmdq cmd;
+ struct hns_roce_mr_table mr_table;
struct hns_roce_cq_table cq_table;
struct hns_roce_qp_table qp_table;
struct hns_roce_eq_table eq_table;
@@ -262,6 +287,11 @@ struct hns_roce_dev {
struct hns_roce_hw *hw;
};
+static inline void hns_roce_write64_k(__be32 val[2], void __iomem *dest)
+{
+ __raw_writeq(*(u64 *) val, dest);
+}
+
static inline struct hns_roce_qp
*__hns_roce_qp_lookup(struct hns_roce_dev *hr_dev, u32 qpn)
{
diff --git a/drivers/infiniband/hw/hns/hns_roce_icm.c b/drivers/infiniband/hw/hns/hns_roce_icm.c
new file mode 100644
index 0000000..43c8bad
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_icm.c
@@ -0,0 +1,437 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/mm.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/scatterlist.h>
+#include "hns_roce_device.h"
+#include "hns_roce_icm.h"
+#include "hns_roce_common.h"
+
+#define HW_SYNC_TIMEOUT_MSECS 500
+
+#define HNS_ROCE_ICM_ALLOC_SIZE (1 << 17)
+#define HNS_ROCE_TABLE_CHUNK_SIZE (1 << 17)
+
+#define DMA_ADDR_T_SHIFT 12
+#define BT_CMD_SYNC_SHIFT 31
+#define BT_BA_SHIFT 32
+
+static int hns_roce_alloc_icm_pages(struct scatterlist *mem, int order,
+ gfp_t gfp_mask)
+{
+ struct page *page;
+
+ page = alloc_pages(gfp_mask, order);
+ if (!page)
+ return -ENOMEM;
+
+ sg_set_page(mem, page, PAGE_SIZE << order, 0);
+
+ return 0;
+}
+
+static int hns_roce_alloc_icm_coherent(struct device *dev,
+ struct scatterlist *mem, int order,
+ gfp_t gfp_mask)
+{
+ void *buf = dma_alloc_coherent(dev, PAGE_SIZE << order,
+ &sg_dma_address(mem), gfp_mask);
+ if (!buf) {
+ dev_err(dev, "alloc coherent icm pages failed.\n");
+ return -ENOMEM;
+ }
+
+ sg_set_buf(mem, buf, PAGE_SIZE << order);
+ WARN_ON(mem->offset);
+ sg_dma_len(mem) = PAGE_SIZE << order;
+ return 0;
+}
+
+struct hns_roce_icm *hns_roce_alloc_icm(struct hns_roce_dev *hr_dev, int npages,
+ gfp_t gfp_mask, int coherent)
+{
+ struct hns_roce_icm_chunk *chunk = NULL;
+ struct hns_roce_icm *icm;
+ int cur_order;
+ int ret;
+
+ WARN_ON(coherent && (gfp_mask & __GFP_HIGHMEM));
+
+ icm = kmalloc(sizeof(*icm),
+ gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+ if (!icm)
+ return NULL;
+
+ icm->refcount = 0;
+ INIT_LIST_HEAD(&icm->chunk_list);
+
+ cur_order = get_order(HNS_ROCE_ICM_ALLOC_SIZE);
+
+ while (npages > 0) {
+ if (!chunk) {
+ chunk = kmalloc(sizeof(*chunk),
+ gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+ if (!chunk)
+ goto fail;
+
+ sg_init_table(chunk->mem, HNS_ROCE_ICM_CHUNK_LEN);
+ chunk->npages = 0;
+ chunk->nsg = 0;
+ list_add_tail(&chunk->list, &icm->chunk_list);
+ }
+
+ while (1 << cur_order > npages)
+ --cur_order;
+
+ if (coherent)
+ ret = hns_roce_alloc_icm_coherent(&hr_dev->pdev->dev,
+ &chunk->mem[chunk->npages],
+ cur_order, gfp_mask);
+ else
+ ret = hns_roce_alloc_icm_pages(
+ &chunk->mem[chunk->npages],
+ cur_order, gfp_mask);
+ if (!ret) {
+ ++chunk->npages;
+ if (coherent) {
+ ++chunk->nsg;
+ } else if (chunk->npages == HNS_ROCE_ICM_CHUNK_LEN) {
+ chunk->nsg = dma_map_sg(&hr_dev->pdev->dev,
+ chunk->mem, chunk->npages,
+ DMA_BIDIRECTIONAL);
+ if (chunk->nsg <= 0)
+ goto fail;
+
+ chunk = NULL;
+ }
+ npages -= 1 << cur_order;
+ } else {
+ /*
+ * If failed on alloc 128k memory one time,
+ * no alloc small block memory,
+ * directly return fail
+ */
+ dev_err(&hr_dev->pdev->dev, "**alloc 128K phy mem failed**\n");
+ goto fail;
+ }
+ }
+
+ if (!coherent && chunk) {
+ chunk->nsg = dma_map_sg(&hr_dev->pdev->dev, chunk->mem,
+ chunk->npages, DMA_BIDIRECTIONAL);
+ if (chunk->nsg <= 0)
+ goto fail;
+ }
+
+ return icm;
+
+fail:
+ hns_roce_free_icm(hr_dev, icm, coherent);
+ return NULL;
+}
+
+static void hns_roce_free_icm_pages(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_chunk *chunk)
+{
+ int i;
+
+ if (chunk->nsg > 0)
+ dma_unmap_sg(&hr_dev->pdev->dev, chunk->mem, chunk->npages,
+ DMA_BIDIRECTIONAL);
+
+ for (i = 0; i < chunk->npages; ++i)
+ __free_pages(sg_page(&chunk->mem[i]),
+ get_order(chunk->mem[i].length));
+}
+
+static void hns_roce_free_icm_coherent(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_chunk *chunk)
+{
+ int i;
+
+ for (i = 0; i < chunk->npages; ++i)
+ dma_free_coherent(&hr_dev->pdev->dev, chunk->mem[i].length,
+ lowmem_page_address(sg_page(&chunk->mem[i])),
+ sg_dma_address(&chunk->mem[i]));
+}
+
+void hns_roce_free_icm(struct hns_roce_dev *hr_dev, struct hns_roce_icm *icm,
+ int coherent)
+{
+ struct hns_roce_icm_chunk *chunk, *tmp;
+
+ if (!icm)
+ return;
+
+ list_for_each_entry_safe(chunk, tmp, &icm->chunk_list, list) {
+ if (coherent)
+ hns_roce_free_icm_coherent(hr_dev, chunk);
+ else
+ hns_roce_free_icm_pages(hr_dev, chunk);
+
+ kfree(chunk);
+ }
+
+ kfree(icm);
+}
+
+int hns_roce_map_icm(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, unsigned long obj)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+ spinlock_t *lock = &hr_dev->bt_cmd_lock;
+ unsigned long end = 0;
+ unsigned long flags;
+ struct hns_roce_icm_iter iter;
+ void __iomem *bt_cmd;
+ u32 bt_cmd_h_val = 0;
+ u32 bt_cmd_val[2];
+ u32 bt_cmd_l = 0;
+ u64 bt_ba = 0;
+ int ret = 0;
+
+ /* Find the icm entry */
+ unsigned long i = (obj & (table->num_obj - 1)) /
+ (HNS_ROCE_TABLE_CHUNK_SIZE / table->obj_size);
+
+ switch (table->type) {
+ case ICM_TYPE_QPC:
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S, ICM_TYPE_QPC);
+ break;
+ case ICM_TYPE_MTPT:
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S,
+ ICM_TYPE_MTPT);
+ break;
+ case ICM_TYPE_CQC:
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S, ICM_TYPE_CQC);
+ break;
+ case ICM_TYPE_SRQC:
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S,
+ ICM_TYPE_SRQC);
+ break;
+ default:
+ return ret;
+ }
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_S, obj);
+ roce_set_bit(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_S, 0);
+ roce_set_bit(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_HW_SYNS_S, 1);
+
+ /* Currently iter only a chunk */
+ for (hns_roce_icm_first(table->icm[i], &iter);
+ !hns_roce_icm_last(&iter); hns_roce_icm_next(&iter)) {
+ bt_ba = hns_roce_icm_addr(&iter) >> DMA_ADDR_T_SHIFT;
+
+ spin_lock_irqsave(lock, flags);
+
+ bt_cmd = hr_dev->reg_base + ROCEE_BT_CMD_H_REG;
+
+ end = msecs_to_jiffies(HW_SYNC_TIMEOUT_MSECS) + jiffies;
+ while (1) {
+ if (roce_readl(bt_cmd) >> BT_CMD_SYNC_SHIFT) {
+ if (!(time_before(jiffies, end))) {
+ dev_err(dev, "Write bt_cmd err,hw_sync is not zero.\n");
+ spin_unlock_irqrestore(lock, flags);
+ ret = -EBUSY;
+ return ret;
+ }
+ } else {
+ break;
+ }
+ }
+
+ bt_cmd_l = (u32)bt_ba;
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_S,
+ bt_ba >> BT_BA_SHIFT);
+
+ bt_cmd_val[0] = bt_cmd_l;
+ bt_cmd_val[1] = bt_cmd_h_val;
+ hns_roce_write64_k(bt_cmd_val,
+ hr_dev->reg_base + ROCEE_BT_CMD_L_REG);
+ spin_unlock_irqrestore(lock, flags);
+ }
+
+ return ret;
+}
+
+int hns_roce_unmap_icm(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, unsigned long obj)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+ unsigned long end = 0;
+ unsigned long flags;
+ void __iomem *bt_cmd;
+ uint32_t bt_cmd_val[2];
+ u32 bt_cmd_h_val = 0;
+ int ret = 0;
+
+ switch (table->type) {
+ case ICM_TYPE_QPC:
+ dev_dbg(dev, "UNMAP QPC BT :\n");
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S, ICM_TYPE_QPC);
+ break;
+ case ICM_TYPE_MTPT:
+ dev_dbg(dev, "UNMAP MTPT BT :\n");
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S,
+ ICM_TYPE_MTPT);
+ break;
+ case ICM_TYPE_CQC:
+ dev_dbg(dev, "UNMAP CQC BT :\n");
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S, ICM_TYPE_CQC);
+ break;
+ case ICM_TYPE_SRQC:
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_MDF_S,
+ ICM_TYPE_SRQC);
+ break;
+ default:
+ return ret;
+ }
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_IN_MDF_S, obj);
+ roce_set_bit(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_S, 1);
+ roce_set_bit(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_HW_SYNS_S, 1);
+ roce_set_field(bt_cmd_h_val, ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_M,
+ ROCEE_BT_CMD_H_ROCEE_BT_CMD_BA_H_S, 0);
+
+ spin_lock_irqsave(&hr_dev->bt_cmd_lock, flags);
+
+ bt_cmd = hr_dev->reg_base + ROCEE_BT_CMD_H_REG;
+
+ end = msecs_to_jiffies(HW_SYNC_TIMEOUT_MSECS) + jiffies;
+ while (1) {
+ msleep(20);
+ if (roce_readl(bt_cmd) >> BT_CMD_SYNC_SHIFT) {
+ if (!(time_before(jiffies, end))) {
+ dev_err(dev, "Write bt_cmd err,hw_sync is not zero.\n");
+ spin_unlock_irqrestore(&hr_dev->bt_cmd_lock,
+ flags);
+ return -EBUSY;
+ }
+ } else {
+ break;
+ }
+ }
+
+ bt_cmd_val[0] = 0;
+ bt_cmd_val[1] = bt_cmd_h_val;
+ hns_roce_write64_k(bt_cmd_val, hr_dev->reg_base + ROCEE_BT_CMD_L_REG);
+ spin_unlock_irqrestore(&hr_dev->bt_cmd_lock, flags);
+
+ return ret;
+}
+
+int hns_roce_init_icm_table(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, u32 type,
+ unsigned long obj_size, unsigned long nobj,
+ int reserved, int use_lowmem, int use_coherent)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+ u32 chunk_size;
+ unsigned long obj_per_chunk;
+ unsigned long num_icm;
+ unsigned long i;
+
+ obj_per_chunk = HNS_ROCE_TABLE_CHUNK_SIZE / obj_size;
+ num_icm = (nobj + obj_per_chunk - 1) / obj_per_chunk;
+
+ table->icm = kcalloc(num_icm, sizeof(*table->icm), GFP_KERNEL);
+ if (!table->icm)
+ return -ENOMEM;
+
+ table->type = type;
+ table->num_icm = num_icm;
+ table->num_obj = nobj;
+ table->obj_size = obj_size;
+ table->lowmem = use_lowmem;
+ table->coherent = use_coherent;
+ mutex_init(&table->mutex);
+
+ for (i = 0;
+ i * HNS_ROCE_TABLE_CHUNK_SIZE < reserved * obj_size; ++i) {
+ chunk_size = HNS_ROCE_TABLE_CHUNK_SIZE;
+ if ((i + 1) * HNS_ROCE_TABLE_CHUNK_SIZE > nobj * obj_size)
+ chunk_size = PAGE_ALIGN(nobj * obj_size -
+ i * HNS_ROCE_TABLE_CHUNK_SIZE);
+
+ table->icm[i] = hns_roce_alloc_icm(hr_dev,
+ chunk_size >> PAGE_SHIFT, (use_lowmem ?
+ GFP_KERNEL : GFP_HIGHUSER) | __GFP_NOWARN,
+ use_coherent);
+ if (!table->icm[i])
+ goto _error_failed_alloc_icm;
+
+ if (hns_roce_map_icm(hr_dev, table,
+ i * HNS_ROCE_TABLE_CHUNK_SIZE / obj_size)) {
+ dev_err(dev, "map icm table failed.\n");
+ goto _error_failed_map_icm;
+ }
+
+ /*
+ * Add a reference to this ICM chunk so that it never
+ * Gets freed (since it contains reserved firmware objects).
+ */
+ ++table->icm[i]->refcount;
+ }
+
+ return 0;
+
+_error_failed_map_icm:
+_error_failed_alloc_icm:
+ for (i = 0; i < num_icm; ++i)
+ if (table->icm[i]) {
+ if (hns_roce_unmap_icm(hr_dev, table,
+ i * HNS_ROCE_TABLE_CHUNK_SIZE / obj_size))
+ dev_err(dev, "unmap icm table failed.\n");
+
+ hns_roce_free_icm(hr_dev, table->icm[i], use_coherent);
+ }
+
+ return -ENOMEM;
+}
+
+void hns_roce_cleanup_icm_table(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+ unsigned long i;
+
+ for (i = 0; i < table->num_icm; ++i)
+ if (table->icm[i]) {
+ if (hns_roce_unmap_icm(hr_dev, table,
+ i * HNS_ROCE_TABLE_CHUNK_SIZE / table->obj_size))
+ dev_err(dev, "unmap icm table failed.\n");
+
+ hns_roce_free_icm(hr_dev, table->icm[i],
+ table->coherent);
+ }
+
+ kfree(table->icm);
+}
+
+void hns_roce_cleanup_icm(struct hns_roce_dev *hr_dev)
+{
+ hns_roce_cleanup_icm_table(hr_dev, &hr_dev->cq_table.table);
+ hns_roce_cleanup_icm_table(hr_dev, &hr_dev->qp_table.irrl_table);
+ hns_roce_cleanup_icm_table(hr_dev, &hr_dev->qp_table.qp_table);
+ hns_roce_cleanup_icm_table(hr_dev, &hr_dev->mr_table.mtpt_table);
+ hns_roce_cleanup_icm_table(hr_dev, &hr_dev->mr_table.mtt_table);
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_icm.h b/drivers/infiniband/hw/hns/hns_roce_icm.h
new file mode 100644
index 0000000..2acabe7
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_icm.h
@@ -0,0 +1,95 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#ifndef _HNS_ROCE_ICM_H
+#define _HNS_ROCE_ICM_H
+
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/pci.h>
+
+enum {
+ /* MAP ICM */
+ ICM_TYPE_QPC = 0,
+ ICM_TYPE_MTPT,
+ ICM_TYPE_CQC,
+ ICM_TYPE_SRQC,
+
+ /* UNMAP ICM */
+ ICM_TYPE_MTT,
+ ICM_TYPE_IRRL,
+};
+
+#define HNS_ROCE_ICM_CHUNK_LEN \
+ ((256 - sizeof(struct list_head) - 2 * sizeof(int)) / \
+ (sizeof(struct scatterlist)))
+
+struct hns_roce_icm_chunk {
+ struct list_head list;
+ int npages;
+ int nsg;
+ struct scatterlist mem[HNS_ROCE_ICM_CHUNK_LEN];
+};
+
+struct hns_roce_icm {
+ struct list_head chunk_list;
+ int refcount;
+};
+
+struct hns_roce_icm_iter {
+ struct hns_roce_icm *icm;
+ struct hns_roce_icm_chunk *chunk;
+ int page_idx;
+};
+
+void hns_roce_free_icm(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm *icm, int coherent);
+int hns_roce_init_icm_table(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table, u32 type,
+ unsigned long obj_size, unsigned long nobj,
+ int reserved, int use_lowmem, int use_coherent);
+void hns_roce_cleanup_icm_table(struct hns_roce_dev *hr_dev,
+ struct hns_roce_icm_table *table);
+void hns_roce_cleanup_icm(struct hns_roce_dev *hr_dev);
+
+static inline void hns_roce_icm_first(struct hns_roce_icm *icm,
+ struct hns_roce_icm_iter *iter)
+{
+ iter->icm = icm;
+ iter->chunk = list_empty(&icm->chunk_list) ? NULL :
+ list_entry(icm->chunk_list.next,
+ struct hns_roce_icm_chunk, list);
+ iter->page_idx = 0;
+}
+
+static inline int hns_roce_icm_last(struct hns_roce_icm_iter *iter)
+{
+ return !iter->chunk;
+}
+
+static inline void hns_roce_icm_next(struct hns_roce_icm_iter *iter)
+{
+ if (++iter->page_idx >= iter->chunk->nsg) {
+ if (iter->chunk->list.next == &iter->icm->chunk_list) {
+ iter->chunk = NULL;
+ return;
+ }
+
+ iter->chunk = list_entry(iter->chunk->list.next,
+ struct hns_roce_icm_chunk, list);
+ iter->page_idx = 0;
+ }
+}
+
+static inline dma_addr_t hns_roce_icm_addr(struct hns_roce_icm_iter *iter)
+{
+ return sg_dma_address(&iter->chunk->mem[iter->page_idx]);
+}
+
+#endif /*_HNS_ROCE_ICM_H*/
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index e246636..3f9d686 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -41,6 +41,7 @@
#include <rdma/ib_verbs.h>
#include "hns_roce_common.h"
#include "hns_roce_device.h"
+#include "hns_roce_icm.h"
int hns_roce_get_cfg(struct hns_roce_dev *hr_dev)
{
@@ -111,6 +112,77 @@ void hns_roce_profile_init(struct hns_roce_dev *hr_dev)
hr_dev->hw->hw_profile(hr_dev);
}
+int hns_roce_init_icm(struct hns_roce_dev *hr_dev)
+{
+ int ret;
+ struct device *dev = &hr_dev->pdev->dev;
+
+ ret = hns_roce_init_icm_table(hr_dev,
+ (void *)&hr_dev->mr_table.mtt_table,
+ ICM_TYPE_MTT, hr_dev->caps.mtt_entry_sz,
+ hr_dev->caps.num_mtt_segs, 0, 1, 0);
+ if (ret) {
+ dev_err(dev, "Failed to map MTT context memory, aborting.\n");
+ return ret;
+ }
+
+ ret = hns_roce_init_icm_table(hr_dev,
+ (void *)&hr_dev->mr_table.mtpt_table,
+ ICM_TYPE_MTPT, hr_dev->caps.mtpt_entry_sz,
+ hr_dev->caps.num_mtpts, 0, 1, 1);
+ if (ret) {
+ dev_err(dev, "Failed to map dMPT context memory, aborting.\n");
+ goto err_unmap_mtt;
+ }
+
+ ret = hns_roce_init_icm_table(hr_dev,
+ (void *)&hr_dev->qp_table.qp_table,
+ ICM_TYPE_QPC, hr_dev->caps.qpc_entry_sz,
+ hr_dev->caps.num_qps, 0, 1, 0);
+ if (ret) {
+ dev_err(dev, "Failed to map QP context memory, aborting.\n");
+ goto err_unmap_dmpt;
+ }
+
+ ret = hns_roce_init_icm_table(hr_dev,
+ (void *)&hr_dev->qp_table.irrl_table,
+ ICM_TYPE_IRRL,
+ hr_dev->caps.irrl_entry_sz *
+ hr_dev->caps.max_qp_init_rdma,
+ hr_dev->caps.num_qps, 0, 1, 0);
+ if (ret) {
+ dev_err(dev, "Failed to map irrl_table memory, aborting.\n");
+ goto err_unmap_qp;
+ }
+
+ ret = hns_roce_init_icm_table(hr_dev,
+ (void *)&hr_dev->cq_table.table,
+ ICM_TYPE_CQC, hr_dev->caps.cqc_entry_sz,
+ hr_dev->caps.num_cqs, 0, 1, 0);
+ if (ret) {
+ dev_err(dev, "Failed to map CQ context memory, aborting.\n");
+ goto err_unmap_irrl;
+ }
+
+ return 0;
+
+err_unmap_irrl:
+ hns_roce_cleanup_icm_table(hr_dev,
+ (void *)&hr_dev->qp_table.irrl_table);
+
+err_unmap_qp:
+ hns_roce_cleanup_icm_table(hr_dev, (void *)&hr_dev->qp_table.qp_table);
+
+err_unmap_dmpt:
+ hns_roce_cleanup_icm_table(hr_dev,
+ (void *)&hr_dev->mr_table.mtpt_table);
+
+err_unmap_mtt:
+ hns_roce_cleanup_icm_table(hr_dev, (void *)&hr_dev->mr_table.mtt_table);
+
+ return ret;
+}
+
/**
* hns_roce_probe - RoCE driver entrance
* @pdev: pointer to platform device
@@ -179,6 +251,16 @@ static int hns_roce_probe(struct platform_device *pdev)
}
}
+ ret = hns_roce_init_icm(hr_dev);
+ if (ret) {
+ dev_err(dev, "init icm fail!\n");
+ goto error_failed_init_icm;
+ }
+
+error_failed_init_icm:
+ if (hr_dev->cmd_mod)
+ hns_roce_cmd_use_polling(hr_dev);
+
error_failed_use_event:
hns_roce_cleanup_eq_table(hr_dev);
@@ -204,6 +286,8 @@ static int hns_roce_remove(struct platform_device *pdev)
{
struct hns_roce_dev *hr_dev = platform_get_drvdata(pdev);
+ hns_roce_cleanup_icm(hr_dev);
+
if (hr_dev->cmd_mod)
hns_roce_cmd_use_polling(hr_dev);
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Lijun Ou <oulijun@huawei.com> |
|---|---|
| Date | 2016-04-28 14:20 +0200 |
| Subject | [PATCH v6 06/21] IB/hns: Add initial cmd operation |
| Message-ID | <rsT5g-2UV-7@gated-at.bofh.it> |
| In reply to | #1390093 |
This patch added the operation for cmd, and added some functions
for initializing eq table and selecting cmd mode.
Signed-off-by: Wei Hu <xavier.huwei@huawei.com>
Signed-off-by: Nenglong Zhao <zhaonenglong@hisilicon.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
---
drivers/infiniband/hw/hns/hns_roce_cmd.c | 96 +++++++++++++++++++++++++++++
drivers/infiniband/hw/hns/hns_roce_cmd.h | 19 ++++++
drivers/infiniband/hw/hns/hns_roce_common.h | 2 +
drivers/infiniband/hw/hns/hns_roce_device.h | 41 ++++++++++++
drivers/infiniband/hw/hns/hns_roce_main.c | 14 +++++
5 files changed, 172 insertions(+)
create mode 100644 drivers/infiniband/hw/hns/hns_roce_cmd.c
create mode 100644 drivers/infiniband/hw/hns/hns_roce_cmd.h
diff --git a/drivers/infiniband/hw/hns/hns_roce_cmd.c b/drivers/infiniband/hw/hns/hns_roce_cmd.c
new file mode 100644
index 0000000..597c964
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_cmd.c
@@ -0,0 +1,96 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/dmapool.h>
+#include <linux/errno.h>
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/list.h>
+#include <linux/sched.h>
+#include <linux/slab.h>
+#include "hns_roce_common.h"
+#include "hns_roce_device.h"
+#include "hns_roce_cmd.h"
+
+#define CMD_MAX_NUM 32
+
+int hns_roce_cmd_init(struct hns_roce_dev *hr_dev)
+{
+ struct device *dev = &hr_dev->pdev->dev;
+
+ mutex_init(&hr_dev->cmd.hcr_mutex);
+ sema_init(&hr_dev->cmd.poll_sem, 1);
+ hr_dev->cmd.use_events = 0;
+ hr_dev->cmd.toggle = 1;
+ hr_dev->cmd.max_cmds = CMD_MAX_NUM;
+ hr_dev->cmd.hcr = hr_dev->reg_base + ROCEE_MB1_REG;
+ hr_dev->cmd.pool = dma_pool_create("hns_roce_cmd", dev,
+ HNS_ROCE_MAILBOX_SIZE,
+ HNS_ROCE_MAILBOX_SIZE, 0);
+ if (!hr_dev->cmd.pool) {
+ dev_err(dev, "Couldn't create mailbox pool for cmd.\n");
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+void hns_roce_cmd_cleanup(struct hns_roce_dev *hr_dev)
+{
+ dma_pool_destroy(hr_dev->cmd.pool);
+}
+
+int hns_roce_cmd_use_events(struct hns_roce_dev *hr_dev)
+{
+ struct hns_roce_cmdq *hr_cmd = &hr_dev->cmd;
+ int i;
+
+ hr_cmd->context = kmalloc(hr_cmd->max_cmds *
+ sizeof(struct hns_roce_cmd_context),
+ GFP_KERNEL);
+ if (!hr_cmd->context)
+ return -ENOMEM;
+
+ for (i = 0; i < hr_cmd->max_cmds; ++i) {
+ hr_cmd->context[i].token = i;
+ hr_cmd->context[i].next = i + 1;
+ }
+
+ hr_cmd->context[hr_cmd->max_cmds - 1].next = -1;
+ hr_cmd->free_head = 0;
+
+ sema_init(&hr_cmd->event_sem, hr_cmd->max_cmds);
+ spin_lock_init(&hr_cmd->context_lock);
+
+ for (hr_cmd->token_mask = 1; hr_cmd->token_mask < hr_cmd->max_cmds;
+ hr_cmd->token_mask <<= 1)
+ ;
+ --hr_cmd->token_mask;
+ hr_cmd->use_events = 1;
+
+ down(&hr_cmd->poll_sem);
+
+ return 0;
+}
+
+void hns_roce_cmd_use_polling(struct hns_roce_dev *hr_dev)
+{
+ struct hns_roce_cmdq *hr_cmd = &hr_dev->cmd;
+ int i;
+
+ hr_cmd->use_events = 0;
+
+ for (i = 0; i < hr_cmd->max_cmds; ++i)
+ down(&hr_cmd->event_sem);
+
+ kfree(hr_cmd->context);
+ up(&hr_cmd->poll_sem);
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_cmd.h b/drivers/infiniband/hw/hns/hns_roce_cmd.h
new file mode 100644
index 0000000..4e102a4
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_cmd.h
@@ -0,0 +1,19 @@
+/*
+ * Copyright (c) 2016 Hisilicon Limited.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#ifndef _HNS_ROCE_CMD_H
+#define _HNS_ROCE_CMD_H
+
+#include <linux/dma-mapping.h>
+
+enum {
+ HNS_ROCE_MAILBOX_SIZE = 4096
+};
+
+#endif /* _HNS_ROCE_CMD_H */
diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h
index 553e2a8..5486e0b 100644
--- a/drivers/infiniband/hw/hns/hns_roce_common.h
+++ b/drivers/infiniband/hw/hns/hns_roce_common.h
@@ -21,4 +21,6 @@
#define ROCEE_ACK_DELAY_REG 0x14
+#define ROCEE_MB1_REG 0x210
+
#endif /* _HNS_ROCE_COMMON_H */
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 481d4b7..5dc33f0 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -31,6 +31,40 @@
#define ADDR_SHIFT_32 32
+struct hns_roce_cmd_context {
+ int next;
+ u16 token;
+};
+
+struct hns_roce_cmdq {
+ struct dma_pool *pool;
+ u8 __iomem *hcr;
+ struct mutex hcr_mutex;
+ struct semaphore poll_sem;
+ /*
+ * Event mode: cmd register mutex protection,
+ * ensure to not exceed max_cmds and user use limit region
+ */
+ struct semaphore event_sem;
+ int max_cmds;
+ spinlock_t context_lock;
+ int free_head;
+ struct hns_roce_cmd_context *context;
+ /*
+ * Result of get integer part
+ * which max_comds compute according a power of 2
+ */
+ u16 token_mask;
+ /*
+ * Process whether use event mode, init default non-zero
+ * After the event queue of cmd event ready,
+ * can switch into event mode
+ * close device, switch into poll mode(non event mode)
+ */
+ u8 use_events;
+ u8 toggle;
+};
+
struct hns_roce_ib_iboe {
struct net_device *netdevs[HNS_ROCE_MAX_PORTS];
u8 phy_port[HNS_ROCE_MAX_PORTS];
@@ -100,11 +134,18 @@ struct hns_roce_dev {
u32 vendor_part_id;
u32 hw_rev;
+ struct hns_roce_cmdq cmd;
+
int cmd_mod;
int loop_idc;
struct hns_roce_hw *hw;
};
+int hns_roce_cmd_init(struct hns_roce_dev *hr_dev);
+void hns_roce_cmd_cleanup(struct hns_roce_dev *hr_dev);
+int hns_roce_cmd_use_events(struct hns_roce_dev *hr_dev);
+void hns_roce_cmd_use_polling(struct hns_roce_dev *hr_dev);
+
extern struct hns_roce_hw hns_roce_hw_v1;
#endif /* _HNS_ROCE_DEVICE_H */
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 3e03cc4..fef3531 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -39,6 +39,7 @@
#include <rdma/ib_umem.h>
#include <rdma/ib_user_verbs.h>
#include <rdma/ib_verbs.h>
+#include "hns_roce_common.h"
#include "hns_roce_device.h"
int hns_roce_get_cfg(struct hns_roce_dev *hr_dev)
@@ -158,6 +159,17 @@ static int hns_roce_probe(struct platform_device *pdev)
hns_roce_profile_init(hr_dev);
+ ret = hns_roce_cmd_init(hr_dev);
+ if (ret) {
+ dev_err(dev, "cmd init failed!\n");
+ goto error_failed_cmd_init;
+ }
+
+error_failed_cmd_init:
+ ret = hns_roce_engine_reset(hr_dev, 0);
+ if (ret)
+ dev_err(&hr_dev->pdev->dev, "roce_engine reset fail\n");
+
error_failed_get_cfg:
ib_dealloc_device(&hr_dev->ib_dev);
@@ -172,6 +184,8 @@ static int hns_roce_remove(struct platform_device *pdev)
{
struct hns_roce_dev *hr_dev = platform_get_drvdata(pdev);
+ hns_roce_cmd_cleanup(hr_dev);
+
(void)hns_roce_engine_reset(hr_dev, 0);
ib_dealloc_device(&hr_dev->ib_dev);
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Leon Romanovsky <leon@kernel.org> |
|---|---|
| Date | 2016-05-03 20:50 +0200 |
| Message-ID | <ruNyp-4SV-5@gated-at.bofh.it> |
| In reply to | #1390093 |
[Multipart message — attachments visible in raw view] — view raw
On Thu, Apr 28, 2016 at 08:09:35PM +0800, Lijun Ou wrote: > The HiSilicon Network Substem is a long term evolution IP which is > supposed to be used in HiSilicon ICT SoCs. HNS (HiSilicon Network > Sybsystem) also has a hardware support of performing RDMA with > RoCEE. > The driver for HiSilicon RoCEE(RoCE Engine) is a platform driver and > will support mulitple versions of SOCs in future. This version of driver > is meant to support Hip06 SoC(which confirms to RoCEEv1 hardware > specifications). Please read Dave's comment [1], it is valuable for your code as well. * Please use 'bool' and "true/false" [1] http://marc.info/?l=linux-rdma&m=146229367301442&w=2 Thanks
[toc] | [prev] | [next] | [standalone]
| From | oulijun <oulijun@huawei.com> |
|---|---|
| Date | 2016-05-04 13:40 +0200 |
| Message-ID | <rv3jQ-2ZG-5@gated-at.bofh.it> |
| In reply to | #1393744 |
On 2016/5/4 2:48, Leon Romanovsky wrote: > On Thu, Apr 28, 2016 at 08:09:35PM +0800, Lijun Ou wrote: >> The HiSilicon Network Substem is a long term evolution IP which is >> supposed to be used in HiSilicon ICT SoCs. HNS (HiSilicon Network >> Sybsystem) also has a hardware support of performing RDMA with >> RoCEE. >> The driver for HiSilicon RoCEE(RoCE Engine) is a platform driver and >> will support mulitple versions of SOCs in future. This version of driver >> is meant to support Hip06 SoC(which confirms to RoCEEv1 hardware >> specifications). > > Please read Dave's comment [1], it is valuable for your code as > well. > > * Please use 'bool' and "true/false" > Yes, i see. thanks Thanks Lijun Ou > [1] http://marc.info/?l=linux-rdma&m=146229367301442&w=2 > > Thanks >
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web