RDMA v6.10 merge window

Normal set of driver updates and small fixes:
 
 - Small improvements and fixes for erdma, efa, hfi1, bnxt_re
 
 - Fix a UAF crash after module unload on leaking restrack entry
 
 - Continue adding full RDMA support in mana with support for EQs, GID's
   and CQs
 
 - Improvements to the mkey cache in mlx5
 
 - DSCP traffic class support in hns and several bug fixes
 
 - Cap the maximum number of MADs in the receive queue to avoid OOM
 
 - Another batch of rxe bug fixes from large scale testing
 
 - __iowrite64_copy() optimizations for write combining MMIO memory
 
 - Remove NULL checks before dev_put/hold()
 
 - EFA support for receive with immediate
 
 - Fix a recent memleaking regression in a cma error path
 -----BEGIN PGP SIGNATURE-----
 
 iHUEABYIAB0WIQRRRCHOFoQz/8F5bUaFwuHvBreFYQUCZkeo2gAKCRCFwuHvBreF
 YbuNAQChzGmS4F0JAn5Wj0CDvkZghELqtvzEb92SzqcgdyQafAD/fC7f23LJ4OsO
 1ZIaQEZu7j9DVg5PKFZ7WfdXjGTKqwA=
 =QRXg
 -----END PGP SIGNATURE-----

Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma updates from Jason Gunthorpe:
 "Aside from the usual things this has an arch update for
  __iowrite64_copy() used by the RDMA drivers.

  This API was intended to generate large 64 byte MemWr TLPs on PCI.
  These days most processors had done this by just repeating writel() in
  a loop. S390 and some new ARM64 designs require a special helper to
  get this to generate.

   - Small improvements and fixes for erdma, efa, hfi1, bnxt_re

   - Fix a UAF crash after module unload on leaking restrack entry

   - Continue adding full RDMA support in mana with support for EQs,
     GID's and CQs

   - Improvements to the mkey cache in mlx5

   - DSCP traffic class support in hns and several bug fixes

   - Cap the maximum number of MADs in the receive queue to avoid OOM

   - Another batch of rxe bug fixes from large scale testing

   - __iowrite64_copy() optimizations for write combining MMIO memory

   - Remove NULL checks before dev_put/hold()

   - EFA support for receive with immediate

   - Fix a recent memleaking regression in a cma error path"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (70 commits)
  RDMA/cma: Fix kmemleak in rdma_core observed during blktests nvme/rdma use siw
  RDMA/IPoIB: Fix format truncation compilation errors
  bnxt_re: avoid shift undefined behavior in bnxt_qplib_alloc_init_hwq
  RDMA/efa: Support QP with unsolicited write w/ imm. receive
  IB/hfi1: Remove generic .ndo_get_stats64
  IB/hfi1: Do not use custom stat allocator
  RDMA/hfi1: Use RMW accessors for changing LNKCTL2
  RDMA/mana_ib: implement uapi for creation of rnic cq
  RDMA/mana_ib: boundary check before installing cq callbacks
  RDMA/mana_ib: introduce a helper to remove cq callbacks
  RDMA/mana_ib: create and destroy RNIC cqs
  RDMA/mana_ib: create EQs for RNIC CQs
  RDMA/core: Remove NULL check before dev_{put, hold}
  RDMA/ipoib: Remove NULL check before dev_{put, hold}
  RDMA/mlx5: Remove NULL check before dev_{put, hold}
  RDMA/mlx5: Track DCT, DCI and REG_UMR QPs as diver_detail resources.
  RDMA/core: Add an option to display driver-specific QPs in the rdmatool
  RDMA/efa: Add shutdown notifier
  RDMA/mana_ib: Fix missing ret value
  IB/mlx5: Use __iowrite64_copy() for write combining stores
  ...

arch/arm64/include/asm/io.h

@@ -139,6 +139,138 @@ extern void __memset_io(volatile void __iomem *, int, size_t);
 #define memcpy_fromio(a,c,l)	__memcpy_fromio((a),(c),(l))
 #define memcpy_toio(c,a,l)	__memcpy_toio((c),(a),(l))
 
+/*
+ * The ARM64 iowrite implementation is intended to support drivers that want to
+ * use write combining. For instance PCI drivers using write combining with a 64
+ * byte __iowrite64_copy() expect to get a 64 byte MemWr TLP on the PCIe bus.
+ *
+ * Newer ARM core have sensitive write combining buffers, it is important that
+ * the stores be contiguous blocks of store instructions. Normal memcpy
+ * approaches have a very low chance to generate write combining.
+ *
+ * Since this is the only API on ARM64 that should be used with write combining
+ * it also integrates the DGH hint which is supposed to lower the latency to
+ * emit the large TLP from the CPU.
+ */
+
+static inline void __const_memcpy_toio_aligned32(volatile u32 __iomem *to,
+						 const u32 *from, size_t count)
+{
+	switch (count) {
+	case 8:
+		asm volatile("str %w0, [%8, #4 * 0]\n"
+			     "str %w1, [%8, #4 * 1]\n"
+			     "str %w2, [%8, #4 * 2]\n"
+			     "str %w3, [%8, #4 * 3]\n"
+			     "str %w4, [%8, #4 * 4]\n"
+			     "str %w5, [%8, #4 * 5]\n"
+			     "str %w6, [%8, #4 * 6]\n"
+			     "str %w7, [%8, #4 * 7]\n"
+			     :
+			     : "rZ"(from[0]), "rZ"(from[1]), "rZ"(from[2]),
+			       "rZ"(from[3]), "rZ"(from[4]), "rZ"(from[5]),
+			       "rZ"(from[6]), "rZ"(from[7]), "r"(to));
+		break;
+	case 4:
+		asm volatile("str %w0, [%4, #4 * 0]\n"
+			     "str %w1, [%4, #4 * 1]\n"
+			     "str %w2, [%4, #4 * 2]\n"
+			     "str %w3, [%4, #4 * 3]\n"
+			     :
+			     : "rZ"(from[0]), "rZ"(from[1]), "rZ"(from[2]),
+			       "rZ"(from[3]), "r"(to));
+		break;
+	case 2:
+		asm volatile("str %w0, [%2, #4 * 0]\n"
+			     "str %w1, [%2, #4 * 1]\n"
+			     :
+			     : "rZ"(from[0]), "rZ"(from[1]), "r"(to));
+		break;
+	case 1:
+		__raw_writel(*from, to);
+		break;
+	default:
+		BUILD_BUG();
+	}
+}
+
+void __iowrite32_copy_full(void __iomem *to, const void *from, size_t count);
+
+static inline void __const_iowrite32_copy(void __iomem *to, const void *from,
+					  size_t count)
+{
+	if (count == 8 || count == 4 || count == 2 || count == 1) {
+		__const_memcpy_toio_aligned32(to, from, count);
+		dgh();
+	} else {
+		__iowrite32_copy_full(to, from, count);
+	}
+}
+
+#define __iowrite32_copy(to, from, count)                  \
+	(__builtin_constant_p(count) ?                     \
+		 __const_iowrite32_copy(to, from, count) : \
+		 __iowrite32_copy_full(to, from, count))
+
+static inline void __const_memcpy_toio_aligned64(volatile u64 __iomem *to,
+						 const u64 *from, size_t count)
+{
+	switch (count) {
+	case 8:
+		asm volatile("str %x0, [%8, #8 * 0]\n"
+			     "str %x1, [%8, #8 * 1]\n"
+			     "str %x2, [%8, #8 * 2]\n"
+			     "str %x3, [%8, #8 * 3]\n"
+			     "str %x4, [%8, #8 * 4]\n"
+			     "str %x5, [%8, #8 * 5]\n"
+			     "str %x6, [%8, #8 * 6]\n"
+			     "str %x7, [%8, #8 * 7]\n"
+			     :
+			     : "rZ"(from[0]), "rZ"(from[1]), "rZ"(from[2]),
+			       "rZ"(from[3]), "rZ"(from[4]), "rZ"(from[5]),
+			       "rZ"(from[6]), "rZ"(from[7]), "r"(to));
+		break;
+	case 4:
+		asm volatile("str %x0, [%4, #8 * 0]\n"
+			     "str %x1, [%4, #8 * 1]\n"
+			     "str %x2, [%4, #8 * 2]\n"
+			     "str %x3, [%4, #8 * 3]\n"
+			     :
+			     : "rZ"(from[0]), "rZ"(from[1]), "rZ"(from[2]),
+			       "rZ"(from[3]), "r"(to));
+		break;
+	case 2:
+		asm volatile("str %x0, [%2, #8 * 0]\n"
+			     "str %x1, [%2, #8 * 1]\n"
+			     :
+			     : "rZ"(from[0]), "rZ"(from[1]), "r"(to));
+		break;
+	case 1:
+		__raw_writeq(*from, to);
+		break;
+	default:
+		BUILD_BUG();
+	}
+}
+
+void __iowrite64_copy_full(void __iomem *to, const void *from, size_t count);
+
+static inline void __const_iowrite64_copy(void __iomem *to, const void *from,
+					  size_t count)
+{
+	if (count == 8 || count == 4 || count == 2 || count == 1) {
+		__const_memcpy_toio_aligned64(to, from, count);
+		dgh();
+	} else {
+		__iowrite64_copy_full(to, from, count);
+	}
+}
+
+#define __iowrite64_copy(to, from, count)                  \
+	(__builtin_constant_p(count) ?                     \
+		 __const_iowrite64_copy(to, from, count) : \
+		 __iowrite64_copy_full(to, from, count))
+
 /*
  * I/O memory mapping functions.
  */

arch/arm64/kernel/io.c

@@ -37,6 +37,48 @@ void __memcpy_fromio(void *to, const volatile void __iomem *from, size_t count)
 }
 EXPORT_SYMBOL(__memcpy_fromio);
 
+/*
+ * This generates a memcpy that works on a from/to address which is aligned to
+ * bits. Count is in terms of the number of bits sized quantities to copy. It
+ * optimizes to use the STR groupings when possible so that it is WC friendly.
+ */
+#define memcpy_toio_aligned(to, from, count, bits)                        \
+	({                                                                \
+		volatile u##bits __iomem *_to = to;                       \
+		const u##bits *_from = from;                              \
+		size_t _count = count;                                    \
+		const u##bits *_end_from = _from + ALIGN_DOWN(_count, 8); \
+                                                                          \
+		for (; _from < _end_from; _from += 8, _to += 8)           \
+			__const_memcpy_toio_aligned##bits(_to, _from, 8); \
+		if ((_count % 8) >= 4) {                                  \
+			__const_memcpy_toio_aligned##bits(_to, _from, 4); \
+			_from += 4;                                       \
+			_to += 4;                                         \
+		}                                                         \
+		if ((_count % 4) >= 2) {                                  \
+			__const_memcpy_toio_aligned##bits(_to, _from, 2); \
+			_from += 2;                                       \
+			_to += 2;                                         \
+		}                                                         \
+		if (_count % 2)                                           \
+			__const_memcpy_toio_aligned##bits(_to, _from, 1); \
+	})
+
+void __iowrite64_copy_full(void __iomem *to, const void *from, size_t count)
+{
+	memcpy_toio_aligned(to, from, count, 64);
+	dgh();
+}
+EXPORT_SYMBOL(__iowrite64_copy_full);
+
+void __iowrite32_copy_full(void __iomem *to, const void *from, size_t count)
+{
+	memcpy_toio_aligned(to, from, count, 32);
+	dgh();
+}
+EXPORT_SYMBOL(__iowrite32_copy_full);
+
 /*
  * Copy data from "real" memory space to IO memory space.
  */

arch/s390/include/asm/io.h

@@ -73,6 +73,21 @@ static inline void ioport_unmap(void __iomem *p)
 #define __raw_writel	zpci_write_u32
 #define __raw_writeq	zpci_write_u64
 
+/* combine single writes by using store-block insn */
+static inline void __iowrite32_copy(void __iomem *to, const void *from,
+				    size_t count)
+{
+	zpci_memcpy_toio(to, from, count * 4);
+}
+#define __iowrite32_copy __iowrite32_copy
+
+static inline void __iowrite64_copy(void __iomem *to, const void *from,
+				    size_t count)
+{
+	zpci_memcpy_toio(to, from, count * 8);
+}
+#define __iowrite64_copy __iowrite64_copy
+
 #endif /* CONFIG_PCI */
 
 #include <asm-generic/io.h>

arch/s390/pci/pci.c

@@ -250,12 +250,6 @@ resource_size_t pcibios_align_resource(void *data, const struct resource *res,
 	return 0;
 }
 
-/* combine single writes by using store-block insn */
-void __iowrite64_copy(void __iomem *to, const void *from, size_t count)
-{
-	zpci_memcpy_toio(to, from, count * 8);
-}
-
 void __iomem *ioremap_prot(phys_addr_t phys_addr, size_t size,
 			   unsigned long prot)
 {

arch/x86/include/asm/io.h

@@ -209,6 +209,23 @@ void memset_io(volatile void __iomem *, int, size_t);
 #define memcpy_toio memcpy_toio
 #define memset_io memset_io
 
+#ifdef CONFIG_X86_64
+/*
+ * Commit 0f07496144c2 ("[PATCH] Add faster __iowrite32_copy routine for
+ * x86_64") says that circa 2006 rep movsl is noticeably faster than a copy
+ * loop.
+ */
+static inline void __iowrite32_copy(void __iomem *to, const void *from,
+				    size_t count)
+{
+	asm volatile("rep ; movsl"
+		     : "=&c"(count), "=&D"(to), "=&S"(from)
+		     : "0"(count), "1"(to), "2"(from)
+		     : "memory");
+}
+#define __iowrite32_copy __iowrite32_copy
+#endif
+
 /*
  * ISA space is 'always mapped' on a typical x86 system, no need to
  * explicitly ioremap() it. The fact that the ISA IO space is mapped

arch/x86/lib/Makefile

@@ -53,7 +53,6 @@ ifneq ($(CONFIG_X86_CMPXCHG64),y)
         lib-y += atomic64_386_32.o
 endif
 else
-        obj-y += iomap_copy_64.o
 ifneq ($(CONFIG_GENERIC_CSUM),y)
         lib-y += csum-partial_64.o csum-copy_64.o csum-wrappers_64.o
 endif

arch/x86/lib/iomap_copy_64.S

@@ -1,15 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0-only */
-/*
- * Copyright 2006 PathScale, Inc.  All Rights Reserved.
- */
-
-#include <linux/linkage.h>
-
-/*
- * override generic version in lib/iomap_copy.c
- */
-SYM_FUNC_START(__iowrite32_copy)
-	movl %edx,%ecx
-	rep movsl
-	RET
-SYM_FUNC_END(__iowrite32_copy)

drivers/infiniband/core/cma.c

@@ -715,8 +715,10 @@ cma_validate_port(struct ib_device *device, u32 port,
 		rcu_read_lock();
 		ndev = rcu_dereference(sgid_attr->ndev);
 		if (!net_eq(dev_net(ndev), dev_addr->net) ||
-		    ndev->ifindex != bound_if_index)
+		    ndev->ifindex != bound_if_index) {
+			rdma_put_gid_attr(sgid_attr);
 			sgid_attr = ERR_PTR(-ENODEV);
+		}
 		rcu_read_unlock();
 		goto out;
 	}

drivers/infiniband/core/device.c

@@ -2174,8 +2174,7 @@ int ib_device_set_netdev(struct ib_device *ib_dev, struct net_device *ndev,
 	spin_unlock_irqrestore(&pdata->netdev_lock, flags);
 
 	add_ndev_hash(pdata);
-	if (old_ndev)
-		__dev_put(old_ndev);
+	__dev_put(old_ndev);
 
 	return 0;
 }
@@ -2235,8 +2234,7 @@ struct net_device *ib_device_get_netdev(struct ib_device *ib_dev,
 		spin_lock(&pdata->netdev_lock);
 		res = rcu_dereference_protected(
 			pdata->netdev, lockdep_is_held(&pdata->netdev_lock));
-		if (res)
-			dev_hold(res);
+		dev_hold(res);
 		spin_unlock(&pdata->netdev_lock);
 	}
 
@@ -2311,9 +2309,7 @@ void ib_enum_roce_netdev(struct ib_device *ib_dev,
 
 			if (filter(ib_dev, port, idev, filter_cookie))
 				cb(ib_dev, port, idev, cookie);
-
-			if (idev)
-				dev_put(idev);
+			dev_put(idev);
 		}
 }
 

drivers/infiniband/core/lag.c

@@ -93,8 +93,7 @@ static struct net_device *rdma_get_xmit_slave_udp(struct ib_device *device,
 	slave = netdev_get_xmit_slave(master, skb,
 				      !!(device->lag_flags &
 					 RDMA_LAG_FLAGS_HASH_ALL_SLAVES));
-	if (slave)
-		dev_hold(slave);
+	dev_hold(slave);
 	rcu_read_unlock();
 	kfree_skb(skb);
 	return slave;

drivers/infiniband/core/nldev.c

@@ -137,6 +137,8 @@ static const struct nla_policy nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
 	[RDMA_NLDEV_ATTR_RES_SUMMARY_ENTRY_NAME]= { .type = NLA_NUL_STRING,
 					.len = RDMA_NLDEV_ATTR_EMPTY_STRING },
 	[RDMA_NLDEV_ATTR_RES_TYPE]		= { .type = NLA_U8 },
+	[RDMA_NLDEV_ATTR_RES_SUBTYPE]		= { .type = NLA_NUL_STRING,
+					.len = RDMA_NLDEV_ATTR_EMPTY_STRING },
 	[RDMA_NLDEV_ATTR_RES_UNSAFE_GLOBAL_RKEY]= { .type = NLA_U32 },
 	[RDMA_NLDEV_ATTR_RES_USECNT]		= { .type = NLA_U64 },
 	[RDMA_NLDEV_ATTR_RES_SRQ]		= { .type = NLA_NESTED },
@@ -164,6 +166,7 @@ static const struct nla_policy nldev_policy[RDMA_NLDEV_ATTR_MAX] = {
 	[RDMA_NLDEV_ATTR_STAT_HWCOUNTER_INDEX]	= { .type = NLA_U32 },
 	[RDMA_NLDEV_ATTR_STAT_HWCOUNTER_DYNAMIC] = { .type = NLA_U8 },
 	[RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE] = { .type = NLA_U8 },
+	[RDMA_NLDEV_ATTR_DRIVER_DETAILS]	= { .type = NLA_U8 },
 };
 
 static int put_driver_name_print_type(struct sk_buff *msg, const char *name,
@@ -399,7 +402,8 @@ err:
 	return -EMSGSIZE;
 }
 
-static int fill_res_info(struct sk_buff *msg, struct ib_device *device)
+static int fill_res_info(struct sk_buff *msg, struct ib_device *device,
+			 bool show_details)
 {
 	static const char * const names[RDMA_RESTRACK_MAX] = {
 		[RDMA_RESTRACK_PD] = "pd",
@@ -424,7 +428,7 @@ static int fill_res_info(struct sk_buff *msg, struct ib_device *device)
 	for (i = 0; i < RDMA_RESTRACK_MAX; i++) {
 		if (!names[i])
 			continue;
-		curr = rdma_restrack_count(device, i);
+		curr = rdma_restrack_count(device, i, show_details);
 		ret = fill_res_info_entry(msg, names[i], curr);
 		if (ret)
 			goto err;
@@ -1305,6 +1309,7 @@ static int nldev_res_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh,
 			      struct netlink_ext_ack *extack)
 {
 	struct nlattr *tb[RDMA_NLDEV_ATTR_MAX];
+	bool show_details = false;
 	struct ib_device *device;
 	struct sk_buff *msg;
 	u32 index;
@@ -1320,6 +1325,9 @@ static int nldev_res_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh,
 	if (!device)
 		return -EINVAL;
 
+	if (tb[RDMA_NLDEV_ATTR_DRIVER_DETAILS])
+		show_details = nla_get_u8(tb[RDMA_NLDEV_ATTR_DRIVER_DETAILS]);
+
 	msg = nlmsg_new(NLMSG_DEFAULT_SIZE, GFP_KERNEL);
 	if (!msg) {
 		ret = -ENOMEM;
@@ -1334,7 +1342,7 @@ static int nldev_res_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh,
 		goto err_free;
 	}
 
-	ret = fill_res_info(msg, device);
+	ret = fill_res_info(msg, device, show_details);
 	if (ret)
 		goto err_free;
 
@@ -1364,7 +1372,7 @@ static int _nldev_res_get_dumpit(struct ib_device *device,
 			RDMA_NL_GET_TYPE(RDMA_NL_NLDEV, RDMA_NLDEV_CMD_RES_GET),
 			0, NLM_F_MULTI);
 
-	if (!nlh || fill_res_info(skb, device)) {
+	if (!nlh || fill_res_info(skb, device, false)) {
 		nlmsg_cancel(skb, nlh);
 		goto out;
 	}
@@ -1534,6 +1542,7 @@ static int res_get_common_dumpit(struct sk_buff *skb,
 	struct rdma_restrack_entry *res;
 	struct rdma_restrack_root *rt;
 	int err, ret = 0, idx = 0;
+	bool show_details = false;
 	struct nlattr *table_attr;
 	struct nlattr *entry_attr;
 	struct ib_device *device;
@@ -1562,6 +1571,9 @@ static int res_get_common_dumpit(struct sk_buff *skb,
 	if (!device)
 		return -EINVAL;
 
+	if (tb[RDMA_NLDEV_ATTR_DRIVER_DETAILS])
+		show_details = nla_get_u8(tb[RDMA_NLDEV_ATTR_DRIVER_DETAILS]);
+
 	/*
 	 * If no PORT_INDEX is supplied, we will return all QPs from that device
 	 */
@@ -1599,6 +1611,9 @@ static int res_get_common_dumpit(struct sk_buff *skb,
 	 * objects.
 	 */
 	xa_for_each(&rt->xa, id, res) {
+		if (xa_get_mark(&rt->xa, res->id, RESTRACK_DD) && !show_details)
+			goto next;
+
 		if (idx < start || !rdma_restrack_get(res))
 			goto next;
 

drivers/infiniband/core/restrack.c

@@ -37,22 +37,6 @@ int rdma_restrack_init(struct ib_device *dev)
 	return 0;
 }
 
-static const char *type2str(enum rdma_restrack_type type)
-{
-	static const char * const names[RDMA_RESTRACK_MAX] = {
-		[RDMA_RESTRACK_PD] = "PD",
-		[RDMA_RESTRACK_CQ] = "CQ",
-		[RDMA_RESTRACK_QP] = "QP",
-		[RDMA_RESTRACK_CM_ID] = "CM_ID",
-		[RDMA_RESTRACK_MR] = "MR",
-		[RDMA_RESTRACK_CTX] = "CTX",
-		[RDMA_RESTRACK_COUNTER] = "COUNTER",
-		[RDMA_RESTRACK_SRQ] = "SRQ",
-	};
-
-	return names[type];
-};
-
 /**
  * rdma_restrack_clean() - clean resource tracking
  * @dev:  IB device
@@ -60,47 +44,14 @@ static const char *type2str(enum rdma_restrack_type type)
 void rdma_restrack_clean(struct ib_device *dev)
 {
 	struct rdma_restrack_root *rt = dev->res;
-	struct rdma_restrack_entry *e;
-	char buf[TASK_COMM_LEN];
-	bool found = false;
-	const char *owner;
 	int i;
 
 	for (i = 0 ; i < RDMA_RESTRACK_MAX; i++) {
 		struct xarray *xa = &dev->res[i].xa;
 
-		if (!xa_empty(xa)) {
-			unsigned long index;
-
-			if (!found) {
-				pr_err("restrack: %s", CUT_HERE);
-				dev_err(&dev->dev, "BUG: RESTRACK detected leak of resources\n");
-			}
-			xa_for_each(xa, index, e) {
-				if (rdma_is_kernel_res(e)) {
-					owner = e->kern_name;
-				} else {
-					/*
-					 * There is no need to call get_task_struct here,
-					 * because we can be here only if there are more
-					 * get_task_struct() call than put_task_struct().
-					 */
-					get_task_comm(buf, e->task);
-					owner = buf;
-				}
-
-				pr_err("restrack: %s %s object allocated by %s is not freed\n",
-				       rdma_is_kernel_res(e) ? "Kernel" :
-							       "User",
-				       type2str(e->type), owner);
-			}
-			found = true;
-		}
+		WARN_ON(!xa_empty(xa));
 		xa_destroy(xa);
 	}
-	if (found)
-		pr_err("restrack: %s", CUT_HERE);
-
 	kfree(rt);
 }
 
@@ -108,8 +59,10 @@ void rdma_restrack_clean(struct ib_device *dev)
  * rdma_restrack_count() - the current usage of specific object
  * @dev:  IB device
  * @type: actual type of object to operate
+ * @show_details: count driver specific objects
  */
-int rdma_restrack_count(struct ib_device *dev, enum rdma_restrack_type type)
+int rdma_restrack_count(struct ib_device *dev, enum rdma_restrack_type type,
+			bool show_details)
 {
 	struct rdma_restrack_root *rt = &dev->res[type];
 	struct rdma_restrack_entry *e;
@@ -117,8 +70,11 @@ int rdma_restrack_count(struct ib_device *dev, enum rdma_restrack_type type)
 	u32 cnt = 0;
 
 	xa_lock(&rt->xa);
-	xas_for_each(&xas, e, U32_MAX)
+	xas_for_each(&xas, e, U32_MAX) {
+		if (xa_get_mark(&rt->xa, e->id, RESTRACK_DD) && !show_details)
+			continue;
 		cnt++;
+	}
 	xa_unlock(&rt->xa);
 	return cnt;
 }
@@ -247,6 +203,9 @@ void rdma_restrack_add(struct rdma_restrack_entry *res)
 		ret = xa_insert(&rt->xa, res->id, res, GFP_KERNEL);
 		if (ret)
 			res->id = 0;
+
+		if (qp->qp_type >= IB_QPT_DRIVER)
+			xa_set_mark(&rt->xa, res->id, RESTRACK_DD);
 	} else if (res->type == RDMA_RESTRACK_COUNTER) {
 		/* Special case to ensure that cntn points to right counter */
 		struct rdma_counter *counter;

drivers/infiniband/core/roce_gid_mgmt.c

@@ -601,8 +601,7 @@ static void del_netdev_default_ips_join(struct ib_device *ib_dev, u32 port,
 
 	rcu_read_lock();
 	master_ndev = netdev_master_upper_dev_get_rcu(rdma_ndev);
-	if (master_ndev)
-		dev_hold(master_ndev);
+	dev_hold(master_ndev);
 	rcu_read_unlock();
 
 	if (master_ndev) {

drivers/infiniband/core/user_mad.c

@@ -63,6 +63,8 @@ MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand userspace MAD packet access");
 MODULE_LICENSE("Dual BSD/GPL");
 
+#define MAX_UMAD_RECV_LIST_SIZE 200000
+
 enum {
 	IB_UMAD_MAX_PORTS  = RDMA_MAX_PORTS,
 	IB_UMAD_MAX_AGENTS = 32,
@@ -113,6 +115,7 @@ struct ib_umad_file {
 	struct mutex		mutex;
 	struct ib_umad_port    *port;
 	struct list_head	recv_list;
+	atomic_t		recv_list_size;
 	struct list_head	send_list;
 	struct list_head	port_list;
 	spinlock_t		send_lock;
@@ -180,24 +183,28 @@ static struct ib_mad_agent *__get_agent(struct ib_umad_file *file, int id)
 	return file->agents_dead ? NULL : file->agent[id];
 }
 
-static int queue_packet(struct ib_umad_file *file,
-			struct ib_mad_agent *agent,
-			struct ib_umad_packet *packet)
+static int queue_packet(struct ib_umad_file *file, struct ib_mad_agent *agent,
+			struct ib_umad_packet *packet, bool is_recv_mad)
 {
 	int ret = 1;
 
 	mutex_lock(&file->mutex);
 
+	if (is_recv_mad &&
+	    atomic_read(&file->recv_list_size) > MAX_UMAD_RECV_LIST_SIZE)
+		goto unlock;
+
 	for (packet->mad.hdr.id = 0;
 	     packet->mad.hdr.id < IB_UMAD_MAX_AGENTS;
 	     packet->mad.hdr.id++)
 		if (agent == __get_agent(file, packet->mad.hdr.id)) {
 			list_add_tail(&packet->list, &file->recv_list);
+			atomic_inc(&file->recv_list_size);
 			wake_up_interruptible(&file->recv_wait);
 			ret = 0;
 			break;
 		}
-
+unlock:
 	mutex_unlock(&file->mutex);
 
 	return ret;
@@ -224,7 +231,7 @@ static void send_handler(struct ib_mad_agent *agent,
 	if (send_wc->status == IB_WC_RESP_TIMEOUT_ERR) {
 		packet->length = IB_MGMT_MAD_HDR;
 		packet->mad.hdr.status = ETIMEDOUT;
-		if (!queue_packet(file, agent, packet))
+		if (!queue_packet(file, agent, packet, false))
 			return;
 	}
 	kfree(packet);
@@ -284,7 +291,7 @@ static void recv_handler(struct ib_mad_agent *agent,
 		rdma_destroy_ah_attr(&ah_attr);
 	}
 
-	if (queue_packet(file, agent, packet))
+	if (queue_packet(file, agent, packet, true))
 		goto err2;
 	return;
 
@@ -409,6 +416,7 @@ static ssize_t ib_umad_read(struct file *filp, char __user *buf,
 
 	packet = list_entry(file->recv_list.next, struct ib_umad_packet, list);
 	list_del(&packet->list);
+	atomic_dec(&file->recv_list_size);
 
 	mutex_unlock(&file->mutex);
 
@@ -421,6 +429,7 @@ static ssize_t ib_umad_read(struct file *filp, char __user *buf,
 		/* Requeue packet */
 		mutex_lock(&file->mutex);
 		list_add(&packet->list, &file->recv_list);
+		atomic_inc(&file->recv_list_size);
 		mutex_unlock(&file->mutex);
 	} else {
 		if (packet->recv_wc)

drivers/infiniband/hw/bnxt_re/qplib_fp.c

@@ -1013,7 +1013,8 @@ int bnxt_qplib_create_qp(struct bnxt_qplib_res *res, struct bnxt_qplib_qp *qp)
 	hwq_attr.stride = sizeof(struct sq_sge);
 	hwq_attr.depth = bnxt_qplib_get_depth(sq);
 	hwq_attr.aux_stride = psn_sz;
-	hwq_attr.aux_depth = bnxt_qplib_set_sq_size(sq, qp->wqe_mode);
+	hwq_attr.aux_depth = psn_sz ? bnxt_qplib_set_sq_size(sq, qp->wqe_mode)
+				    : 0;
 	/* Update msn tbl size */
 	if (BNXT_RE_HW_RETX(qp->dev_cap_flags) && psn_sz) {
 		hwq_attr.aux_depth = roundup_pow_of_two(bnxt_qplib_set_sq_size(sq, qp->wqe_mode));

drivers/infiniband/hw/efa/efa_admin_cmds_defs.h

@@ -110,7 +110,10 @@ struct efa_admin_create_qp_cmd {
 	 *    virtual (IOVA returned by MR registration)
 	 * 1 : rq_virt - If set, RQ ring base address is
 	 *    virtual (IOVA returned by MR registration)
-	 * 7:2 : reserved - MBZ
+	 * 2 : unsolicited_write_recv - If set, work requests
+	 *    will not be consumed for incoming RDMA write with
+	 *    immediate
+	 * 7:3 : reserved - MBZ
 	 */
 	u8 flags;
 
@@ -663,7 +666,9 @@ struct efa_admin_feature_device_attr_desc {
 	 *    polling is supported
 	 * 3 : rdma_write - If set, RDMA Write is supported
 	 *    on TX queues
-	 * 31:4 : reserved - MBZ
+	 * 4 : unsolicited_write_recv - If set, unsolicited
+	 *    write with imm. receive is supported
+	 * 31:5 : reserved - MBZ
 	 */
 	u32 device_caps;
 
@@ -1009,6 +1014,7 @@ struct efa_admin_host_info {
 /* create_qp_cmd */
 #define EFA_ADMIN_CREATE_QP_CMD_SQ_VIRT_MASK                BIT(0)
 #define EFA_ADMIN_CREATE_QP_CMD_RQ_VIRT_MASK                BIT(1)
+#define EFA_ADMIN_CREATE_QP_CMD_UNSOLICITED_WRITE_RECV_MASK BIT(2)
 
 /* modify_qp_cmd */
 #define EFA_ADMIN_MODIFY_QP_CMD_QP_STATE_MASK               BIT(0)
@@ -1044,6 +1050,7 @@ struct efa_admin_host_info {
 #define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RNR_RETRY_MASK   BIT(1)
 #define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_DATA_POLLING_128_MASK BIT(2)
 #define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RDMA_WRITE_MASK  BIT(3)
+#define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_UNSOLICITED_WRITE_RECV_MASK BIT(4)
 
 /* create_eq_cmd */
 #define EFA_ADMIN_CREATE_EQ_CMD_ENTRY_SIZE_WORDS_MASK       GENMASK(4, 0)

drivers/infiniband/hw/efa/efa_com_cmd.c

@@ -32,6 +32,9 @@ int efa_com_create_qp(struct efa_com_dev *edev,
 			params->rq_depth;
 	create_qp_cmd.uar = params->uarn;
 
+	if (params->unsolicited_write_recv)
+		EFA_SET(&create_qp_cmd.flags, EFA_ADMIN_CREATE_QP_CMD_UNSOLICITED_WRITE_RECV, 1);
+
 	err = efa_com_cmd_exec(aq,
 			       (struct efa_admin_aq_entry *)&create_qp_cmd,
 			       sizeof(create_qp_cmd),

drivers/infiniband/hw/efa/efa_com_cmd.h

@@ -27,6 +27,7 @@ struct efa_com_create_qp_params {
 	u16 pd;
 	u16 uarn;
 	u8 qp_type;
+	u8 unsolicited_write_recv : 1;
 };
 
 struct efa_com_create_qp_result {

drivers/infiniband/hw/efa/efa_main.c

@@ -671,11 +671,22 @@ static void efa_remove(struct pci_dev *pdev)
 	efa_remove_device(pdev);
 }
 
+static void efa_shutdown(struct pci_dev *pdev)
+{
+	struct efa_dev *dev = pci_get_drvdata(pdev);
+
+	efa_destroy_eqs(dev);
+	efa_com_dev_reset(&dev->edev, EFA_REGS_RESET_SHUTDOWN);
+	efa_free_irq(dev, &dev->admin_irq);
+	efa_disable_msix(dev);
+}
+
 static struct pci_driver efa_pci_driver = {
 	.name           = DRV_MODULE_NAME,
 	.id_table       = efa_pci_tbl,
 	.probe          = efa_probe,
 	.remove         = efa_remove,
+	.shutdown       = efa_shutdown,
 };
 
 module_pci_driver(efa_pci_driver);

drivers/infiniband/hw/efa/efa_verbs.c

@@ -263,6 +263,9 @@ int efa_query_device(struct ib_device *ibdev,
 		if (EFA_DEV_CAP(dev, RDMA_WRITE))
 			resp.device_caps |= EFA_QUERY_DEVICE_CAPS_RDMA_WRITE;
 
+		if (EFA_DEV_CAP(dev, UNSOLICITED_WRITE_RECV))
+			resp.device_caps |= EFA_QUERY_DEVICE_CAPS_UNSOLICITED_WRITE_RECV;
+
 		if (dev->neqs)
 			resp.device_caps |= EFA_QUERY_DEVICE_CAPS_CQ_NOTIFICATIONS;
 
@@ -639,6 +642,7 @@ int efa_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *init_attr,
 	struct efa_ibv_create_qp cmd = {};
 	struct efa_qp *qp = to_eqp(ibqp);
 	struct efa_ucontext *ucontext;
+	u16 supported_efa_flags = 0;
 	int err;
 
 	ucontext = rdma_udata_to_drv_context(udata, struct efa_ucontext,
@@ -676,13 +680,23 @@ int efa_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *init_attr,
 		goto err_out;
 	}
 
-	if (cmd.comp_mask) {
+	if (cmd.comp_mask || !is_reserved_cleared(cmd.reserved_90)) {
 		ibdev_dbg(&dev->ibdev,
 			  "Incompatible ABI params, unknown fields in udata\n");
 		err = -EINVAL;
 		goto err_out;
 	}
 
+	if (EFA_DEV_CAP(dev, UNSOLICITED_WRITE_RECV))
+		supported_efa_flags |= EFA_CREATE_QP_WITH_UNSOLICITED_WRITE_RECV;
+
+	if (cmd.flags & ~supported_efa_flags) {
+		ibdev_dbg(&dev->ibdev, "Unsupported EFA QP create flags[%#x], supported[%#x]\n",
+			  cmd.flags, supported_efa_flags);
+		err = -EOPNOTSUPP;
+		goto err_out;
+	}
+
 	create_qp_params.uarn = ucontext->uarn;
 	create_qp_params.pd = to_epd(ibqp->pd)->pdn;
 
@@ -722,6 +736,9 @@ int efa_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *init_attr,
 		create_qp_params.rq_base_addr = qp->rq_dma_addr;
 	}
 
+	if (cmd.flags & EFA_CREATE_QP_WITH_UNSOLICITED_WRITE_RECV)
+		create_qp_params.unsolicited_write_recv = true;
+
 	err = efa_com_create_qp(&dev->edev, &create_qp_params,
 				&create_qp_resp);
 	if (err)

drivers/infiniband/hw/erdma/erdma.h

@@ -33,7 +33,8 @@ struct erdma_eq {
 	atomic64_t notify_num;
 
 	void __iomem *db;
-	u64 *db_record;
+	u64 *dbrec;
+	dma_addr_t dbrec_dma;
 };
 
 struct erdma_cmdq_sq {
@@ -48,7 +49,8 @@ struct erdma_cmdq_sq {
 
 	u16 wqebb_cnt;
 
-	u64 *db_record;
+	u64 *dbrec;
+	dma_addr_t dbrec_dma;
 };
 
 struct erdma_cmdq_cq {
@@ -61,7 +63,8 @@ struct erdma_cmdq_cq {
 	u32 ci;
 	u32 cmdsn;
 
-	u64 *db_record;
+	u64 *dbrec;
+	dma_addr_t dbrec_dma;
 
 	atomic64_t armed_num;
 };
@@ -177,9 +180,6 @@ enum {
 	ERDMA_RES_CNT = 2,
 };
 
-#define ERDMA_EXTRA_BUFFER_SIZE ERDMA_DB_SIZE
-#define WARPPED_BUFSIZE(size) ((size) + ERDMA_EXTRA_BUFFER_SIZE)
-
 struct erdma_dev {
 	struct ib_device ibdev;
 	struct net_device *netdev;
@@ -213,6 +213,7 @@ struct erdma_dev {
 	atomic_t num_ctx;
 	struct list_head cep_list;
 
+	struct dma_pool *db_pool;
 	struct dma_pool *resp_pool;
 };
 

drivers/infiniband/hw/erdma/erdma_cmdq.c

@@ -14,7 +14,7 @@ static void arm_cmdq_cq(struct erdma_cmdq *cmdq)
 		      FIELD_PREP(ERDMA_CQDB_CMDSN_MASK, cmdq->cq.cmdsn) |
 		      FIELD_PREP(ERDMA_CQDB_IDX_MASK, cmdq->cq.cmdsn);
 
-	*cmdq->cq.db_record = db_data;
+	*cmdq->cq.dbrec = db_data;
 	writeq(db_data, dev->func_bar + ERDMA_CMDQ_CQDB_REG);
 
 	atomic64_inc(&cmdq->cq.armed_num);
@@ -25,7 +25,7 @@ static void kick_cmdq_db(struct erdma_cmdq *cmdq)
 	struct erdma_dev *dev = container_of(cmdq, struct erdma_dev, cmdq);
 	u64 db_data = FIELD_PREP(ERDMA_CMD_HDR_WQEBB_INDEX_MASK, cmdq->sq.pi);
 
-	*cmdq->sq.db_record = db_data;
+	*cmdq->sq.dbrec = db_data;
 	writeq(db_data, dev->func_bar + ERDMA_CMDQ_SQDB_REG);
 }
 
@@ -89,20 +89,18 @@ static int erdma_cmdq_sq_init(struct erdma_dev *dev)
 {
 	struct erdma_cmdq *cmdq = &dev->cmdq;
 	struct erdma_cmdq_sq *sq = &cmdq->sq;
-	u32 buf_size;
 
 	sq->wqebb_cnt = SQEBB_COUNT(ERDMA_CMDQ_SQE_SIZE);
 	sq->depth = cmdq->max_outstandings * sq->wqebb_cnt;
 
-	buf_size = sq->depth << SQEBB_SHIFT;
-
-	sq->qbuf =
-		dma_alloc_coherent(&dev->pdev->dev, WARPPED_BUFSIZE(buf_size),
-				   &sq->qbuf_dma_addr, GFP_KERNEL);
+	sq->qbuf = dma_alloc_coherent(&dev->pdev->dev, sq->depth << SQEBB_SHIFT,
+				      &sq->qbuf_dma_addr, GFP_KERNEL);
 	if (!sq->qbuf)
 		return -ENOMEM;
 
-	sq->db_record = (u64 *)(sq->qbuf + buf_size);
+	sq->dbrec = dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &sq->dbrec_dma);
+	if (!sq->dbrec)
+		goto err_out;
 
 	spin_lock_init(&sq->lock);
 
@@ -111,30 +109,33 @@ static int erdma_cmdq_sq_init(struct erdma_dev *dev)
 	erdma_reg_write32(dev, ERDMA_REGS_CMDQ_SQ_ADDR_L_REG,
 			  lower_32_bits(sq->qbuf_dma_addr));
 	erdma_reg_write32(dev, ERDMA_REGS_CMDQ_DEPTH_REG, sq->depth);
-	erdma_reg_write64(dev, ERDMA_CMDQ_SQ_DB_HOST_ADDR_REG,
-			  sq->qbuf_dma_addr + buf_size);
+	erdma_reg_write64(dev, ERDMA_CMDQ_SQ_DB_HOST_ADDR_REG, sq->dbrec_dma);
 
 	return 0;
+
+err_out:
+	dma_free_coherent(&dev->pdev->dev, sq->depth << SQEBB_SHIFT,
+			  sq->qbuf, sq->qbuf_dma_addr);
+
+	return -ENOMEM;
 }
 
 static int erdma_cmdq_cq_init(struct erdma_dev *dev)
 {
 	struct erdma_cmdq *cmdq = &dev->cmdq;
 	struct erdma_cmdq_cq *cq = &cmdq->cq;
-	u32 buf_size;
 
 	cq->depth = cmdq->sq.depth;
-	buf_size = cq->depth << CQE_SHIFT;
-
-	cq->qbuf =
-		dma_alloc_coherent(&dev->pdev->dev, WARPPED_BUFSIZE(buf_size),
-				   &cq->qbuf_dma_addr, GFP_KERNEL | __GFP_ZERO);
+	cq->qbuf = dma_alloc_coherent(&dev->pdev->dev, cq->depth << CQE_SHIFT,
+				      &cq->qbuf_dma_addr, GFP_KERNEL);
 	if (!cq->qbuf)
 		return -ENOMEM;
 
 	spin_lock_init(&cq->lock);
 
-	cq->db_record = (u64 *)(cq->qbuf + buf_size);
+	cq->dbrec = dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &cq->dbrec_dma);
+	if (!cq->dbrec)
+		goto err_out;
 
 	atomic64_set(&cq->armed_num, 0);
 
@@ -142,24 +143,25 @@ static int erdma_cmdq_cq_init(struct erdma_dev *dev)
 			  upper_32_bits(cq->qbuf_dma_addr));
 	erdma_reg_write32(dev, ERDMA_REGS_CMDQ_CQ_ADDR_L_REG,
 			  lower_32_bits(cq->qbuf_dma_addr));
-	erdma_reg_write64(dev, ERDMA_CMDQ_CQ_DB_HOST_ADDR_REG,
-			  cq->qbuf_dma_addr + buf_size);
+	erdma_reg_write64(dev, ERDMA_CMDQ_CQ_DB_HOST_ADDR_REG, cq->dbrec_dma);
 
 	return 0;
+
+err_out:
+	dma_free_coherent(&dev->pdev->dev, cq->depth << CQE_SHIFT, cq->qbuf,
+			  cq->qbuf_dma_addr);
+
+	return -ENOMEM;
 }
 
 static int erdma_cmdq_eq_init(struct erdma_dev *dev)
 {
 	struct erdma_cmdq *cmdq = &dev->cmdq;
 	struct erdma_eq *eq = &cmdq->eq;
-	u32 buf_size;
 
 	eq->depth = cmdq->max_outstandings;
-	buf_size = eq->depth << EQE_SHIFT;
-
-	eq->qbuf =
-		dma_alloc_coherent(&dev->pdev->dev, WARPPED_BUFSIZE(buf_size),
-				   &eq->qbuf_dma_addr, GFP_KERNEL | __GFP_ZERO);
+	eq->qbuf = dma_alloc_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT,
+				      &eq->qbuf_dma_addr, GFP_KERNEL);
 	if (!eq->qbuf)
 		return -ENOMEM;
 
@@ -167,17 +169,24 @@ static int erdma_cmdq_eq_init(struct erdma_dev *dev)
 	atomic64_set(&eq->event_num, 0);
 
 	eq->db = dev->func_bar + ERDMA_REGS_CEQ_DB_BASE_REG;
-	eq->db_record = (u64 *)(eq->qbuf + buf_size);
+	eq->dbrec = dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &eq->dbrec_dma);
+	if (!eq->dbrec)
+		goto err_out;
 
 	erdma_reg_write32(dev, ERDMA_REGS_CMDQ_EQ_ADDR_H_REG,
 			  upper_32_bits(eq->qbuf_dma_addr));
 	erdma_reg_write32(dev, ERDMA_REGS_CMDQ_EQ_ADDR_L_REG,
 			  lower_32_bits(eq->qbuf_dma_addr));
 	erdma_reg_write32(dev, ERDMA_REGS_CMDQ_EQ_DEPTH_REG, eq->depth);
-	erdma_reg_write64(dev, ERDMA_CMDQ_EQ_DB_HOST_ADDR_REG,
-			  eq->qbuf_dma_addr + buf_size);
+	erdma_reg_write64(dev, ERDMA_CMDQ_EQ_DB_HOST_ADDR_REG, eq->dbrec_dma);
 
 	return 0;
+
+err_out:
+	dma_free_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT, eq->qbuf,
+			  eq->qbuf_dma_addr);
+
+	return -ENOMEM;
 }
 
 int erdma_cmdq_init(struct erdma_dev *dev)
@@ -211,17 +220,17 @@ int erdma_cmdq_init(struct erdma_dev *dev)
 	return 0;
 
 err_destroy_cq:
-	dma_free_coherent(&dev->pdev->dev,
-			  (cmdq->cq.depth << CQE_SHIFT) +
-				  ERDMA_EXTRA_BUFFER_SIZE,
+	dma_free_coherent(&dev->pdev->dev, cmdq->cq.depth << CQE_SHIFT,
 			  cmdq->cq.qbuf, cmdq->cq.qbuf_dma_addr);
 
+	dma_pool_free(dev->db_pool, cmdq->cq.dbrec, cmdq->cq.dbrec_dma);
+
 err_destroy_sq:
-	dma_free_coherent(&dev->pdev->dev,
-			  (cmdq->sq.depth << SQEBB_SHIFT) +
-				  ERDMA_EXTRA_BUFFER_SIZE,
+	dma_free_coherent(&dev->pdev->dev, cmdq->sq.depth << SQEBB_SHIFT,
 			  cmdq->sq.qbuf, cmdq->sq.qbuf_dma_addr);
 
+	dma_pool_free(dev->db_pool, cmdq->sq.dbrec, cmdq->sq.dbrec_dma);
+
 	return err;
 }
 
@@ -238,18 +247,20 @@ void erdma_cmdq_destroy(struct erdma_dev *dev)
 
 	clear_bit(ERDMA_CMDQ_STATE_OK_BIT, &cmdq->state);
 
-	dma_free_coherent(&dev->pdev->dev,
-			  (cmdq->eq.depth << EQE_SHIFT) +
-				  ERDMA_EXTRA_BUFFER_SIZE,
+	dma_free_coherent(&dev->pdev->dev, cmdq->eq.depth << EQE_SHIFT,
 			  cmdq->eq.qbuf, cmdq->eq.qbuf_dma_addr);
-	dma_free_coherent(&dev->pdev->dev,
-			  (cmdq->sq.depth << SQEBB_SHIFT) +
-				  ERDMA_EXTRA_BUFFER_SIZE,
+
+	dma_pool_free(dev->db_pool, cmdq->eq.dbrec, cmdq->eq.dbrec_dma);
+
+	dma_free_coherent(&dev->pdev->dev, cmdq->sq.depth << SQEBB_SHIFT,
 			  cmdq->sq.qbuf, cmdq->sq.qbuf_dma_addr);
-	dma_free_coherent(&dev->pdev->dev,
-			  (cmdq->cq.depth << CQE_SHIFT) +
-				  ERDMA_EXTRA_BUFFER_SIZE,
+
+	dma_pool_free(dev->db_pool, cmdq->sq.dbrec, cmdq->sq.dbrec_dma);
+
+	dma_free_coherent(&dev->pdev->dev, cmdq->cq.depth << CQE_SHIFT,
 			  cmdq->cq.qbuf, cmdq->cq.qbuf_dma_addr);
+
+	dma_pool_free(dev->db_pool, cmdq->cq.dbrec, cmdq->cq.dbrec_dma);
 }
 
 static void *get_next_valid_cmdq_cqe(struct erdma_cmdq *cmdq)

drivers/infiniband/hw/erdma/erdma_cq.c

@@ -26,7 +26,7 @@ static void notify_cq(struct erdma_cq *cq, u8 solcitied)
 		FIELD_PREP(ERDMA_CQDB_CMDSN_MASK, cq->kern_cq.cmdsn) |
 		FIELD_PREP(ERDMA_CQDB_CI_MASK, cq->kern_cq.ci);
 
-	*cq->kern_cq.db_record = db_data;
+	*cq->kern_cq.dbrec = db_data;
 	writeq(db_data, cq->kern_cq.db);
 }
 

drivers/infiniband/hw/erdma/erdma_eq.c

@@ -13,7 +13,7 @@ void notify_eq(struct erdma_eq *eq)
 	u64 db_data = FIELD_PREP(ERDMA_EQDB_CI_MASK, eq->ci) |
 		      FIELD_PREP(ERDMA_EQDB_ARM_MASK, 1);
 
-	*eq->db_record = db_data;
+	*eq->dbrec = db_data;
 	writeq(db_data, eq->db);
 
 	atomic64_inc(&eq->notify_num);
@@ -83,14 +83,11 @@ void erdma_aeq_event_handler(struct erdma_dev *dev)
 int erdma_aeq_init(struct erdma_dev *dev)
 {
 	struct erdma_eq *eq = &dev->aeq;
-	u32 buf_size;
 
 	eq->depth = ERDMA_DEFAULT_EQ_DEPTH;
-	buf_size = eq->depth << EQE_SHIFT;
 
-	eq->qbuf =
-		dma_alloc_coherent(&dev->pdev->dev, WARPPED_BUFSIZE(buf_size),
-				   &eq->qbuf_dma_addr, GFP_KERNEL | __GFP_ZERO);
+	eq->qbuf = dma_alloc_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT,
+				      &eq->qbuf_dma_addr, GFP_KERNEL);
 	if (!eq->qbuf)
 		return -ENOMEM;
 
@@ -99,26 +96,34 @@ int erdma_aeq_init(struct erdma_dev *dev)
 	atomic64_set(&eq->notify_num, 0);
 
 	eq->db = dev->func_bar + ERDMA_REGS_AEQ_DB_REG;
-	eq->db_record = (u64 *)(eq->qbuf + buf_size);
+	eq->dbrec = dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &eq->dbrec_dma);
+	if (!eq->dbrec)
+		goto err_out;
 
 	erdma_reg_write32(dev, ERDMA_REGS_AEQ_ADDR_H_REG,
 			  upper_32_bits(eq->qbuf_dma_addr));
 	erdma_reg_write32(dev, ERDMA_REGS_AEQ_ADDR_L_REG,
 			  lower_32_bits(eq->qbuf_dma_addr));
 	erdma_reg_write32(dev, ERDMA_REGS_AEQ_DEPTH_REG, eq->depth);
-	erdma_reg_write64(dev, ERDMA_AEQ_DB_HOST_ADDR_REG,
-			  eq->qbuf_dma_addr + buf_size);
+	erdma_reg_write64(dev, ERDMA_AEQ_DB_HOST_ADDR_REG, eq->dbrec_dma);
 
 	return 0;
+
+err_out:
+	dma_free_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT, eq->qbuf,
+			  eq->qbuf_dma_addr);
+
+	return -ENOMEM;
 }
 
 void erdma_aeq_destroy(struct erdma_dev *dev)
 {
 	struct erdma_eq *eq = &dev->aeq;
 
-	dma_free_coherent(&dev->pdev->dev,
-			  WARPPED_BUFSIZE(eq->depth << EQE_SHIFT), eq->qbuf,
+	dma_free_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT, eq->qbuf,
 			  eq->qbuf_dma_addr);
+
+	dma_pool_free(dev->db_pool, eq->dbrec, eq->dbrec_dma);
 }
 
 void erdma_ceq_completion_handler(struct erdma_eq_cb *ceq_cb)
@@ -209,7 +214,6 @@ static void erdma_free_ceq_irq(struct erdma_dev *dev, u16 ceqn)
 static int create_eq_cmd(struct erdma_dev *dev, u32 eqn, struct erdma_eq *eq)
 {
 	struct erdma_cmdq_create_eq_req req;
-	dma_addr_t db_info_dma_addr;
 
 	erdma_cmdq_build_reqhdr(&req.hdr, CMDQ_SUBMOD_COMMON,
 				CMDQ_OPCODE_CREATE_EQ);
@@ -219,9 +223,8 @@ static int create_eq_cmd(struct erdma_dev *dev, u32 eqn, struct erdma_eq *eq)
 	req.qtype = ERDMA_EQ_TYPE_CEQ;
 	/* Vector index is the same as EQN. */
 	req.vector_idx = eqn;
-	db_info_dma_addr = eq->qbuf_dma_addr + (eq->depth << EQE_SHIFT);
-	req.db_dma_addr_l = lower_32_bits(db_info_dma_addr);
-	req.db_dma_addr_h = upper_32_bits(db_info_dma_addr);
+	req.db_dma_addr_l = lower_32_bits(eq->dbrec_dma);
+	req.db_dma_addr_h = upper_32_bits(eq->dbrec_dma);
 
 	return erdma_post_cmd_wait(&dev->cmdq, &req, sizeof(req), NULL, NULL);
 }
@@ -229,12 +232,11 @@ static int create_eq_cmd(struct erdma_dev *dev, u32 eqn, struct erdma_eq *eq)
 static int erdma_ceq_init_one(struct erdma_dev *dev, u16 ceqn)
 {
 	struct erdma_eq *eq = &dev->ceqs[ceqn].eq;
-	u32 buf_size = ERDMA_DEFAULT_EQ_DEPTH << EQE_SHIFT;
 	int ret;
 
-	eq->qbuf =
-		dma_alloc_coherent(&dev->pdev->dev, WARPPED_BUFSIZE(buf_size),
-				   &eq->qbuf_dma_addr, GFP_KERNEL | __GFP_ZERO);
+	eq->depth = ERDMA_DEFAULT_EQ_DEPTH;
+	eq->qbuf = dma_alloc_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT,
+				      &eq->qbuf_dma_addr, GFP_KERNEL);
 	if (!eq->qbuf)
 		return -ENOMEM;
 
@@ -242,10 +244,16 @@ static int erdma_ceq_init_one(struct erdma_dev *dev, u16 ceqn)
 	atomic64_set(&eq->event_num, 0);
 	atomic64_set(&eq->notify_num, 0);
 
-	eq->depth = ERDMA_DEFAULT_EQ_DEPTH;
 	eq->db = dev->func_bar + ERDMA_REGS_CEQ_DB_BASE_REG +
 		 (ceqn + 1) * ERDMA_DB_SIZE;
-	eq->db_record = (u64 *)(eq->qbuf + buf_size);
+
+	eq->dbrec = dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &eq->dbrec_dma);
+	if (!eq->dbrec) {
+		dma_free_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT,
+				  eq->qbuf, eq->qbuf_dma_addr);
+		return -ENOMEM;
+	}
+
 	eq->ci = 0;
 	dev->ceqs[ceqn].dev = dev;
 
@@ -259,7 +267,6 @@ static int erdma_ceq_init_one(struct erdma_dev *dev, u16 ceqn)
 static void erdma_ceq_uninit_one(struct erdma_dev *dev, u16 ceqn)
 {
 	struct erdma_eq *eq = &dev->ceqs[ceqn].eq;
-	u32 buf_size = ERDMA_DEFAULT_EQ_DEPTH << EQE_SHIFT;
 	struct erdma_cmdq_destroy_eq_req req;
 	int err;
 
@@ -276,8 +283,9 @@ static void erdma_ceq_uninit_one(struct erdma_dev *dev, u16 ceqn)
 	if (err)
 		return;
 
-	dma_free_coherent(&dev->pdev->dev, WARPPED_BUFSIZE(buf_size), eq->qbuf,
+	dma_free_coherent(&dev->pdev->dev, eq->depth << EQE_SHIFT, eq->qbuf,
 			  eq->qbuf_dma_addr);
+	dma_pool_free(dev->db_pool, eq->dbrec, eq->dbrec_dma);
 }
 
 int erdma_ceqs_init(struct erdma_dev *dev)

drivers/infiniband/hw/erdma/erdma_hw.h

@@ -240,7 +240,7 @@ struct erdma_cmdq_create_cq_req {
 	u32 qbuf_addr_l;
 	u32 qbuf_addr_h;
 	u32 cfg1;
-	u64 cq_db_info_addr;
+	u64 cq_dbrec_dma;
 	u32 first_page_offset;
 	u32 cfg2;
 };
@@ -335,8 +335,8 @@ struct erdma_cmdq_create_qp_req {
 	u64 rq_buf_addr;
 	u32 sq_mtt_cfg;
 	u32 rq_mtt_cfg;
-	u64 sq_db_info_dma_addr;
-	u64 rq_db_info_dma_addr;
+	u64 sq_dbrec_dma;
+	u64 rq_dbrec_dma;
 
 	u64 sq_mtt_entry[3];
 	u64 rq_mtt_entry[3];

drivers/infiniband/hw/erdma/erdma_main.c

@@ -178,16 +178,26 @@ static int erdma_device_init(struct erdma_dev *dev, struct pci_dev *pdev)
 	if (!dev->resp_pool)
 		return -ENOMEM;
 
+	dev->db_pool = dma_pool_create("erdma_db_pool", &pdev->dev,
+				       ERDMA_DB_SIZE, ERDMA_DB_SIZE, 0);
+	if (!dev->db_pool) {
+		ret = -ENOMEM;
+		goto destroy_resp_pool;
+	}
+
 	ret = dma_set_mask_and_coherent(&pdev->dev,
 					DMA_BIT_MASK(ERDMA_PCI_WIDTH));
 	if (ret)
-		goto destroy_pool;
+		goto destroy_db_pool;
 
 	dma_set_max_seg_size(&pdev->dev, UINT_MAX);
 
 	return 0;
 
-destroy_pool:
+destroy_db_pool:
+	dma_pool_destroy(dev->db_pool);
+
+destroy_resp_pool:
 	dma_pool_destroy(dev->resp_pool);
 
 	return ret;
@@ -195,6 +205,7 @@ destroy_pool:
 
 static void erdma_device_uninit(struct erdma_dev *dev)
 {
+	dma_pool_destroy(dev->db_pool);
 	dma_pool_destroy(dev->resp_pool);
 }
 

drivers/infiniband/hw/erdma/erdma_qp.c

@@ -492,7 +492,7 @@ static void kick_sq_db(struct erdma_qp *qp, u16 pi)
 	u64 db_data = FIELD_PREP(ERDMA_SQE_HDR_QPN_MASK, QP_ID(qp)) |
 		      FIELD_PREP(ERDMA_SQE_HDR_WQEBB_INDEX_MASK, pi);
 
-	*(u64 *)qp->kern_qp.sq_db_info = db_data;
+	*(u64 *)qp->kern_qp.sq_dbrec = db_data;
 	writeq(db_data, qp->kern_qp.hw_sq_db);
 }
 
@@ -557,7 +557,7 @@ static int erdma_post_recv_one(struct erdma_qp *qp,
 		return -EINVAL;
 	}
 
-	*(u64 *)qp->kern_qp.rq_db_info = *(u64 *)rqe;
+	*(u64 *)qp->kern_qp.rq_dbrec = *(u64 *)rqe;
 	writeq(*(u64 *)rqe, qp->kern_qp.hw_rq_db);
 
 	qp->kern_qp.rwr_tbl[qp->kern_qp.rq_pi & (qp->attrs.rq_size - 1)] =

drivers/infiniband/hw/erdma/erdma_verbs.c

@@ -76,10 +76,8 @@ static int create_qp_cmd(struct erdma_ucontext *uctx, struct erdma_qp *qp)
 
 		req.rq_buf_addr = qp->kern_qp.rq_buf_dma_addr;
 		req.sq_buf_addr = qp->kern_qp.sq_buf_dma_addr;
-		req.sq_db_info_dma_addr = qp->kern_qp.sq_buf_dma_addr +
-					  (qp->attrs.sq_size << SQEBB_SHIFT);
-		req.rq_db_info_dma_addr = qp->kern_qp.rq_buf_dma_addr +
-					  (qp->attrs.rq_size << RQE_SHIFT);
+		req.sq_dbrec_dma = qp->kern_qp.sq_dbrec_dma;
+		req.rq_dbrec_dma = qp->kern_qp.rq_dbrec_dma;
 	} else {
 		user_qp = &qp->user_qp;
 		req.sq_cqn_mtt_cfg = FIELD_PREP(
@@ -107,8 +105,8 @@ static int create_qp_cmd(struct erdma_ucontext *uctx, struct erdma_qp *qp)
 		assemble_qbuf_mtt_for_cmd(&user_qp->rq_mem, &req.rq_mtt_cfg,
 					  &req.rq_buf_addr, req.rq_mtt_entry);
 
-		req.sq_db_info_dma_addr = user_qp->sq_db_info_dma_addr;
-		req.rq_db_info_dma_addr = user_qp->rq_db_info_dma_addr;
+		req.sq_dbrec_dma = user_qp->sq_dbrec_dma;
+		req.rq_dbrec_dma = user_qp->rq_dbrec_dma;
 
 		if (uctx->ext_db.enable) {
 			req.sq_cqn_mtt_cfg |=
@@ -209,8 +207,7 @@ static int create_cq_cmd(struct erdma_ucontext *uctx, struct erdma_cq *cq)
 				       ERDMA_MR_MTT_0LEVEL);
 
 		req.first_page_offset = 0;
-		req.cq_db_info_addr =
-			cq->kern_cq.qbuf_dma_addr + (cq->depth << CQE_SHIFT);
+		req.cq_dbrec_dma = cq->kern_cq.dbrec_dma;
 	} else {
 		mem = &cq->user_cq.qbuf_mem;
 		req.cfg0 |=
@@ -233,7 +230,7 @@ static int create_cq_cmd(struct erdma_ucontext *uctx, struct erdma_cq *cq)
 				       mem->mtt_nents);
 
 		req.first_page_offset = mem->page_offset;
-		req.cq_db_info_addr = cq->user_cq.db_info_dma_addr;
+		req.cq_dbrec_dma = cq->user_cq.dbrec_dma;
 
 		if (uctx->ext_db.enable) {
 			req.cfg1 |= FIELD_PREP(
@@ -482,16 +479,24 @@ static void free_kernel_qp(struct erdma_qp *qp)
 	vfree(qp->kern_qp.rwr_tbl);
 
 	if (qp->kern_qp.sq_buf)
-		dma_free_coherent(
-			&dev->pdev->dev,
-			WARPPED_BUFSIZE(qp->attrs.sq_size << SQEBB_SHIFT),
-			qp->kern_qp.sq_buf, qp->kern_qp.sq_buf_dma_addr);
+		dma_free_coherent(&dev->pdev->dev,
+				  qp->attrs.sq_size << SQEBB_SHIFT,
+				  qp->kern_qp.sq_buf,
+				  qp->kern_qp.sq_buf_dma_addr);
+
+	if (qp->kern_qp.sq_dbrec)
+		dma_pool_free(dev->db_pool, qp->kern_qp.sq_dbrec,
+			      qp->kern_qp.sq_dbrec_dma);
 
 	if (qp->kern_qp.rq_buf)
-		dma_free_coherent(
-			&dev->pdev->dev,
-			WARPPED_BUFSIZE(qp->attrs.rq_size << RQE_SHIFT),
-			qp->kern_qp.rq_buf, qp->kern_qp.rq_buf_dma_addr);
+		dma_free_coherent(&dev->pdev->dev,
+				  qp->attrs.rq_size << RQE_SHIFT,
+				  qp->kern_qp.rq_buf,
+				  qp->kern_qp.rq_buf_dma_addr);
+
+	if (qp->kern_qp.rq_dbrec)
+		dma_pool_free(dev->db_pool, qp->kern_qp.rq_dbrec,
+			      qp->kern_qp.rq_dbrec_dma);
 }
 
 static int init_kernel_qp(struct erdma_dev *dev, struct erdma_qp *qp,
@@ -516,20 +521,27 @@ static int init_kernel_qp(struct erdma_dev *dev, struct erdma_qp *qp,
 	if (!kqp->swr_tbl || !kqp->rwr_tbl)
 		goto err_out;
 
-	size = (qp->attrs.sq_size << SQEBB_SHIFT) + ERDMA_EXTRA_BUFFER_SIZE;
+	size = qp->attrs.sq_size << SQEBB_SHIFT;
 	kqp->sq_buf = dma_alloc_coherent(&dev->pdev->dev, size,
 					 &kqp->sq_buf_dma_addr, GFP_KERNEL);
 	if (!kqp->sq_buf)
 		goto err_out;
 
-	size = (qp->attrs.rq_size << RQE_SHIFT) + ERDMA_EXTRA_BUFFER_SIZE;
+	kqp->sq_dbrec =
+		dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &kqp->sq_dbrec_dma);
+	if (!kqp->sq_dbrec)
+		goto err_out;
+
+	size = qp->attrs.rq_size << RQE_SHIFT;
 	kqp->rq_buf = dma_alloc_coherent(&dev->pdev->dev, size,
 					 &kqp->rq_buf_dma_addr, GFP_KERNEL);
 	if (!kqp->rq_buf)
 		goto err_out;
 
-	kqp->sq_db_info = kqp->sq_buf + (qp->attrs.sq_size << SQEBB_SHIFT);
-	kqp->rq_db_info = kqp->rq_buf + (qp->attrs.rq_size << RQE_SHIFT);
+	kqp->rq_dbrec =
+		dma_pool_zalloc(dev->db_pool, GFP_KERNEL, &kqp->rq_dbrec_dma);
+	if (!kqp->rq_dbrec)
+		goto err_out;
 
 	return 0;
 
@@ -864,9 +876,9 @@ erdma_unmap_user_dbrecords(struct erdma_ucontext *ctx,
 }
 
 static int init_user_qp(struct erdma_qp *qp, struct erdma_ucontext *uctx,
-			u64 va, u32 len, u64 db_info_va)
+			u64 va, u32 len, u64 dbrec_va)
 {
-	dma_addr_t db_info_dma_addr;
+	dma_addr_t dbrec_dma;
 	u32 rq_offset;
 	int ret;
 
@@ -889,14 +901,14 @@ static int init_user_qp(struct erdma_qp *qp, struct erdma_ucontext *uctx,
 	if (ret)
 		goto put_sq_mtt;
 
-	ret = erdma_map_user_dbrecords(uctx, db_info_va,
+	ret = erdma_map_user_dbrecords(uctx, dbrec_va,
 				       &qp->user_qp.user_dbr_page,
-				       &db_info_dma_addr);
+				       &dbrec_dma);
 	if (ret)
 		goto put_rq_mtt;
 
-	qp->user_qp.sq_db_info_dma_addr = db_info_dma_addr;
-	qp->user_qp.rq_db_info_dma_addr = db_info_dma_addr + ERDMA_DB_SIZE;
+	qp->user_qp.sq_dbrec_dma = dbrec_dma;
+	qp->user_qp.rq_dbrec_dma = dbrec_dma + ERDMA_DB_SIZE;
 
 	return 0;
 
@@ -1237,9 +1249,10 @@ int erdma_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
 		return err;
 
 	if (rdma_is_kernel_res(&cq->ibcq.res)) {
-		dma_free_coherent(&dev->pdev->dev,
-				  WARPPED_BUFSIZE(cq->depth << CQE_SHIFT),
+		dma_free_coherent(&dev->pdev->dev, cq->depth << CQE_SHIFT,
 				  cq->kern_cq.qbuf, cq->kern_cq.qbuf_dma_addr);
+		dma_pool_free(dev->db_pool, cq->kern_cq.dbrec,
+			      cq->kern_cq.dbrec_dma);
 	} else {
 		erdma_unmap_user_dbrecords(ctx, &cq->user_cq.user_dbr_page);
 		put_mtt_entries(dev, &cq->user_cq.qbuf_mem);
@@ -1279,16 +1292,7 @@ int erdma_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata)
 	wait_for_completion(&qp->safe_free);
 
 	if (rdma_is_kernel_res(&qp->ibqp.res)) {
-		vfree(qp->kern_qp.swr_tbl);
-		vfree(qp->kern_qp.rwr_tbl);
-		dma_free_coherent(
-			&dev->pdev->dev,
-			WARPPED_BUFSIZE(qp->attrs.rq_size << RQE_SHIFT),
-			qp->kern_qp.rq_buf, qp->kern_qp.rq_buf_dma_addr);
-		dma_free_coherent(
-			&dev->pdev->dev,
-			WARPPED_BUFSIZE(qp->attrs.sq_size << SQEBB_SHIFT),
-			qp->kern_qp.sq_buf, qp->kern_qp.sq_buf_dma_addr);
+		free_kernel_qp(qp);
 	} else {
 		put_mtt_entries(dev, &qp->user_qp.sq_mem);
 		put_mtt_entries(dev, &qp->user_qp.rq_mem);
@@ -1588,7 +1592,7 @@ static int erdma_init_user_cq(struct erdma_ucontext *ctx, struct erdma_cq *cq,
 
 	ret = erdma_map_user_dbrecords(ctx, ureq->db_record_va,
 				       &cq->user_cq.user_dbr_page,
-				       &cq->user_cq.db_info_dma_addr);
+				       &cq->user_cq.dbrec_dma);
 	if (ret)
 		put_mtt_entries(dev, &cq->user_cq.qbuf_mem);
 
@@ -1600,19 +1604,27 @@ static int erdma_init_kernel_cq(struct erdma_cq *cq)
 	struct erdma_dev *dev = to_edev(cq->ibcq.device);
 
 	cq->kern_cq.qbuf =
-		dma_alloc_coherent(&dev->pdev->dev,
-				   WARPPED_BUFSIZE(cq->depth << CQE_SHIFT),
+		dma_alloc_coherent(&dev->pdev->dev, cq->depth << CQE_SHIFT,
 				   &cq->kern_cq.qbuf_dma_addr, GFP_KERNEL);
 	if (!cq->kern_cq.qbuf)
 		return -ENOMEM;
 
-	cq->kern_cq.db_record =
-		(u64 *)(cq->kern_cq.qbuf + (cq->depth << CQE_SHIFT));
+	cq->kern_cq.dbrec = dma_pool_zalloc(dev->db_pool, GFP_KERNEL,
+					    &cq->kern_cq.dbrec_dma);
+	if (!cq->kern_cq.dbrec)
+		goto err_out;
+
 	spin_lock_init(&cq->kern_cq.lock);
 	/* use default cqdb addr */
 	cq->kern_cq.db = dev->func_bar + ERDMA_BAR_CQDB_SPACE_OFFSET;
 
 	return 0;
+
+err_out:
+	dma_free_coherent(&dev->pdev->dev, cq->depth << CQE_SHIFT,
+			  cq->kern_cq.qbuf, cq->kern_cq.qbuf_dma_addr);
+
+	return -ENOMEM;
 }
 
 int erdma_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
@@ -1676,9 +1688,10 @@ err_free_res:
 		erdma_unmap_user_dbrecords(ctx, &cq->user_cq.user_dbr_page);
 		put_mtt_entries(dev, &cq->user_cq.qbuf_mem);
 	} else {
-		dma_free_coherent(&dev->pdev->dev,
-				  WARPPED_BUFSIZE(depth << CQE_SHIFT),
+		dma_free_coherent(&dev->pdev->dev, depth << CQE_SHIFT,
 				  cq->kern_cq.qbuf, cq->kern_cq.qbuf_dma_addr);
+		dma_pool_free(dev->db_pool, cq->kern_cq.dbrec,
+			      cq->kern_cq.dbrec_dma);
 	}
 
 err_out_xa:

drivers/infiniband/hw/erdma/erdma_verbs.h

@@ -140,8 +140,8 @@ struct erdma_uqp {
 	struct erdma_mem sq_mem;
 	struct erdma_mem rq_mem;
 
-	dma_addr_t sq_db_info_dma_addr;
-	dma_addr_t rq_db_info_dma_addr;
+	dma_addr_t sq_dbrec_dma;
+	dma_addr_t rq_dbrec_dma;
 
 	struct erdma_user_dbrecords_page *user_dbr_page;
 
@@ -167,8 +167,11 @@ struct erdma_kqp {
 	void *rq_buf;
 	dma_addr_t rq_buf_dma_addr;
 
-	void *sq_db_info;
-	void *rq_db_info;
+	void *sq_dbrec;
+	void *rq_dbrec;
+
+	dma_addr_t sq_dbrec_dma;
+	dma_addr_t rq_dbrec_dma;
 
 	u8 sig_all;
 };
@@ -246,13 +249,14 @@ struct erdma_kcq_info {
 
 	spinlock_t lock;
 	u8 __iomem *db;
-	u64 *db_record;
+	u64 *dbrec;
+	dma_addr_t dbrec_dma;
 };
 
 struct erdma_ucq_info {
 	struct erdma_mem qbuf_mem;
 	struct erdma_user_dbrecords_page *user_dbr_page;
-	dma_addr_t db_info_dma_addr;
+	dma_addr_t dbrec_dma;
 };
 
 struct erdma_cq {

drivers/infiniband/hw/hfi1/ipoib_main.c

@@ -21,36 +21,25 @@ static int hfi1_ipoib_dev_init(struct net_device *dev)
 	struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev);
 	int ret;
 
-	dev->tstats = netdev_alloc_pcpu_stats(struct pcpu_sw_netstats);
-	if (!dev->tstats)
-		return -ENOMEM;
-
 	ret = priv->netdev_ops->ndo_init(dev);
 	if (ret)
-		goto out_ret;
+		return ret;
 
 	ret = hfi1_netdev_add_data(priv->dd,
 				   qpn_from_mac(priv->netdev->dev_addr),
 				   dev);
 	if (ret < 0) {
 		priv->netdev_ops->ndo_uninit(dev);
-		goto out_ret;
+		return ret;
 	}
 
 	return 0;
-out_ret:
-	free_percpu(dev->tstats);
-	dev->tstats = NULL;
-	return ret;
 }
 
 static void hfi1_ipoib_dev_uninit(struct net_device *dev)
 {
 	struct hfi1_ipoib_dev_priv *priv = hfi1_ipoib_priv(dev);
 
-	free_percpu(dev->tstats);
-	dev->tstats = NULL;
-
 	hfi1_netdev_remove_data(priv->dd, qpn_from_mac(priv->netdev->dev_addr));
 
 	priv->netdev_ops->ndo_uninit(dev);
@@ -107,7 +96,6 @@ static const struct net_device_ops hfi1_ipoib_netdev_ops = {
 	.ndo_uninit       = hfi1_ipoib_dev_uninit,
 	.ndo_open         = hfi1_ipoib_dev_open,
 	.ndo_stop         = hfi1_ipoib_dev_stop,
-	.ndo_get_stats64  = dev_get_tstats64,
 };
 
 static int hfi1_ipoib_mcast_attach(struct net_device *dev,
@@ -173,9 +161,6 @@ static void hfi1_ipoib_netdev_dtor(struct net_device *dev)
 
 	hfi1_ipoib_txreq_deinit(priv);
 	hfi1_ipoib_rxq_deinit(priv->netdev);
-
-	free_percpu(dev->tstats);
-	dev->tstats = NULL;
 }
 
 static void hfi1_ipoib_set_id(struct net_device *dev, int id)
@@ -234,6 +219,7 @@ static int hfi1_ipoib_setup_rn(struct ib_device *device,
 
 	netdev->priv_destructor = hfi1_ipoib_netdev_dtor;
 	netdev->needs_free_netdev = true;
+	netdev->pcpu_stat_type = NETDEV_PCPU_STAT_TSTATS;
 
 	return 0;
 }

drivers/infiniband/hw/hfi1/pcie.c

@@ -1207,14 +1207,11 @@ retry:
 		    (u32)lnkctl2);
 	/* only write to parent if target is not as high as ours */
 	if ((lnkctl2 & PCI_EXP_LNKCTL2_TLS) < target_vector) {
-		lnkctl2 &= ~PCI_EXP_LNKCTL2_TLS;
-		lnkctl2 |= target_vector;
-		dd_dev_info(dd, "%s: ..new link control2: 0x%x\n", __func__,
-			    (u32)lnkctl2);
-		ret = pcie_capability_write_word(parent,
-						 PCI_EXP_LNKCTL2, lnkctl2);
+		ret = pcie_capability_clear_and_set_word(parent, PCI_EXP_LNKCTL2,
+							 PCI_EXP_LNKCTL2_TLS,
+							 target_vector);
 		if (ret) {
-			dd_dev_err(dd, "Unable to write to PCI config\n");
+			dd_dev_err(dd, "Unable to change parent PCI target speed\n");
 			return_error = 1;
 			goto done;
 		}
@@ -1223,22 +1220,11 @@ retry:
 	}
 
 	dd_dev_info(dd, "%s: setting target link speed\n", __func__);
-	ret = pcie_capability_read_word(dd->pcidev, PCI_EXP_LNKCTL2, &lnkctl2);
+	ret = pcie_capability_clear_and_set_word(dd->pcidev, PCI_EXP_LNKCTL2,
+						 PCI_EXP_LNKCTL2_TLS,
+						 target_vector);
 	if (ret) {
-		dd_dev_err(dd, "Unable to read from PCI config\n");
-		return_error = 1;
-		goto done;
-	}
-
-	dd_dev_info(dd, "%s: ..old link control2: 0x%x\n", __func__,
-		    (u32)lnkctl2);
-	lnkctl2 &= ~PCI_EXP_LNKCTL2_TLS;
-	lnkctl2 |= target_vector;
-	dd_dev_info(dd, "%s: ..new link control2: 0x%x\n", __func__,
-		    (u32)lnkctl2);
-	ret = pcie_capability_write_word(dd->pcidev, PCI_EXP_LNKCTL2, lnkctl2);
-	if (ret) {
-		dd_dev_err(dd, "Unable to write to PCI config\n");
+		dd_dev_err(dd, "Unable to change device PCI target speed\n");
 		return_error = 1;
 		goto done;
 	}

drivers/infiniband/hw/hns/hns_roce_ah.c

@@ -59,8 +59,10 @@ int hns_roce_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr,
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibah->device);
 	struct hns_roce_ib_create_ah_resp resp = {};
 	struct hns_roce_ah *ah = to_hr_ah(ibah);
-	int ret = 0;
-	u32 max_sl;
+	u8 tclass = get_tclass(grh);
+	u8 priority = 0;
+	u8 tc_mode = 0;
+	int ret;
 
 	if (hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP08 && udata)
 		return -EOPNOTSUPP;
@@ -74,16 +76,23 @@ int hns_roce_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr,
 	ah->av.hop_limit = grh->hop_limit;
 	ah->av.flowlabel = grh->flow_label;
 	ah->av.udp_sport = get_ah_udp_sport(ah_attr);
-	ah->av.tclass = get_tclass(grh);
+	ah->av.tclass = tclass;
 
-	ah->av.sl = rdma_ah_get_sl(ah_attr);
-	max_sl = min_t(u32, MAX_SERVICE_LEVEL, hr_dev->caps.sl_num - 1);
-	if (unlikely(ah->av.sl > max_sl)) {
-		ibdev_err_ratelimited(&hr_dev->ib_dev,
-				      "failed to set sl, sl (%u) shouldn't be larger than %u.\n",
-				      ah->av.sl, max_sl);
+	ret = hr_dev->hw->get_dscp(hr_dev, tclass, &tc_mode, &priority);
+	if (ret == -EOPNOTSUPP)
+		ret = 0;
+
+	if (ret && grh->sgid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP)
+		return ret;
+
+	if (tc_mode == HNAE3_TC_MAP_MODE_DSCP &&
+	    grh->sgid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP)
+		ah->av.sl = priority;
+	else
+		ah->av.sl = rdma_ah_get_sl(ah_attr);
+
+	if (!check_sl_valid(hr_dev, ah->av.sl))
 		return -EINVAL;
-	}
 
 	memcpy(ah->av.dgid, grh->dgid.raw, HNS_ROCE_GID_SIZE);
 	memcpy(ah->av.mac, ah_attr->roce.dmac, ETH_ALEN);
@@ -99,6 +108,8 @@ int hns_roce_create_ah(struct ib_ah *ibah, struct rdma_ah_init_attr *init_attr,
 	}
 
 	if (udata) {
+		resp.priority = ah->av.sl;
+		resp.tc_mode = tc_mode;
 		memcpy(resp.dmac, ah_attr->roce.dmac, ETH_ALEN);
 		ret = ib_copy_to_udata(udata, &resp,
 				       min(udata->outlen, sizeof(resp)));

drivers/infiniband/hw/hns/hns_roce_alloc.c

@@ -153,8 +153,7 @@ int hns_roce_get_kmem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs,
 	return total;
 }
 
-int hns_roce_get_umem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs,
-			   int buf_cnt, struct ib_umem *umem,
+int hns_roce_get_umem_bufs(dma_addr_t *bufs, int buf_cnt, struct ib_umem *umem,
 			   unsigned int page_shift)
 {
 	struct ib_block_iter biter;

drivers/infiniband/hw/hns/hns_roce_cq.c

@@ -149,7 +149,7 @@ static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq)
 		return ret;
 	}
 
-	ret = xa_err(xa_store(&cq_table->array, hr_cq->cqn, hr_cq, GFP_KERNEL));
+	ret = xa_err(xa_store_irq(&cq_table->array, hr_cq->cqn, hr_cq, GFP_KERNEL));
 	if (ret) {
 		ibdev_err(ibdev, "failed to xa_store CQ, ret = %d.\n", ret);
 		goto err_put;
@@ -163,7 +163,7 @@ static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq)
 	return 0;
 
 err_xa:
-	xa_erase(&cq_table->array, hr_cq->cqn);
+	xa_erase_irq(&cq_table->array, hr_cq->cqn);
 err_put:
 	hns_roce_table_put(hr_dev, &cq_table->table, hr_cq->cqn);
 
@@ -182,7 +182,7 @@ static void free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq)
 		dev_err(dev, "DESTROY_CQ failed (%d) for CQN %06lx\n", ret,
 			hr_cq->cqn);
 
-	xa_erase(&cq_table->array, hr_cq->cqn);
+	xa_erase_irq(&cq_table->array, hr_cq->cqn);
 
 	/* Waiting interrupt process procedure carried out */
 	synchronize_irq(hr_dev->eq_table.eq[hr_cq->vector].irq);
@@ -476,13 +476,6 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type)
 	struct ib_event event;
 	struct ib_cq *ibcq;
 
-	hr_cq = xa_load(&hr_dev->cq_table.array,
-			cqn & (hr_dev->caps.num_cqs - 1));
-	if (!hr_cq) {
-		dev_warn(dev, "async event for bogus CQ 0x%06x\n", cqn);
-		return;
-	}
-
 	if (event_type != HNS_ROCE_EVENT_TYPE_CQ_ID_INVALID &&
 	    event_type != HNS_ROCE_EVENT_TYPE_CQ_ACCESS_ERROR &&
 	    event_type != HNS_ROCE_EVENT_TYPE_CQ_OVERFLOW) {
@@ -491,7 +484,16 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type)
 		return;
 	}
 
-	refcount_inc(&hr_cq->refcount);
+	xa_lock(&hr_dev->cq_table.array);
+	hr_cq = xa_load(&hr_dev->cq_table.array,
+			cqn & (hr_dev->caps.num_cqs - 1));
+	if (hr_cq)
+		refcount_inc(&hr_cq->refcount);
+	xa_unlock(&hr_dev->cq_table.array);
+	if (!hr_cq) {
+		dev_warn(dev, "async event for bogus CQ 0x%06x\n", cqn);
+		return;
+	}
 
 	ibcq = &hr_cq->ib_cq;
 	if (ibcq->event_handler) {
@@ -534,4 +536,5 @@ void hns_roce_cleanup_cq_table(struct hns_roce_dev *hr_dev)
 
 	for (i = 0; i < HNS_ROCE_CQ_BANK_NUM; i++)
 		ida_destroy(&hr_dev->cq_table.bank[i].ida);
+	mutex_destroy(&hr_dev->cq_table.bank_mutex);
 }

drivers/infiniband/hw/hns/hns_roce_device.h

@@ -100,6 +100,9 @@
 #define CQ_BANKID_SHIFT 2
 #define CQ_BANKID_MASK GENMASK(1, 0)
 
+#define HNS_ROCE_MAX_CQ_COUNT 0xFFFF
+#define HNS_ROCE_MAX_CQ_PERIOD 0xFFFF
+
 enum {
 	SERV_TYPE_RC,
 	SERV_TYPE_UC,
@@ -645,6 +648,8 @@ struct hns_roce_qp {
 	struct hns_user_mmap_entry *dwqe_mmap_entry;
 	u32			config;
 	enum hns_roce_cong_type	cong_type;
+	u8			tc_mode;
+	u8			priority;
 };
 
 struct hns_roce_ib_iboe {
@@ -923,8 +928,7 @@ struct hns_roce_hw {
 	int (*rereg_write_mtpt)(struct hns_roce_dev *hr_dev,
 				struct hns_roce_mr *mr, int flags,
 				void *mb_buf);
-	int (*frmr_write_mtpt)(struct hns_roce_dev *hr_dev, void *mb_buf,
-			       struct hns_roce_mr *mr);
+	int (*frmr_write_mtpt)(void *mb_buf, struct hns_roce_mr *mr);
 	int (*mw_write_mtpt)(void *mb_buf, struct hns_roce_mw *mw);
 	void (*write_cqc)(struct hns_roce_dev *hr_dev,
 			  struct hns_roce_cq *hr_cq, void *mb_buf, u64 *mtts,
@@ -950,6 +954,8 @@ struct hns_roce_hw {
 	int (*query_sccc)(struct hns_roce_dev *hr_dev, u32 qpn, void *buffer);
 	int (*query_hw_counter)(struct hns_roce_dev *hr_dev,
 				u64 *stats, u32 port, int *hw_counters);
+	int (*get_dscp)(struct hns_roce_dev *hr_dev, u8 dscp,
+			u8 *tc_mode, u8 *priority);
 	const struct ib_device_ops *hns_roce_dev_ops;
 	const struct ib_device_ops *hns_roce_dev_srq_ops;
 };
@@ -1228,7 +1234,7 @@ struct hns_roce_buf *hns_roce_buf_alloc(struct hns_roce_dev *hr_dev, u32 size,
 int hns_roce_get_kmem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs,
 			   int buf_cnt, struct hns_roce_buf *buf,
 			   unsigned int page_shift);
-int hns_roce_get_umem_bufs(struct hns_roce_dev *hr_dev, dma_addr_t *bufs,
+int hns_roce_get_umem_bufs(dma_addr_t *bufs,
 			   int buf_cnt, struct ib_umem *umem,
 			   unsigned int page_shift);
 
@@ -1292,4 +1298,6 @@ struct hns_user_mmap_entry *
 hns_roce_user_mmap_entry_insert(struct ib_ucontext *ucontext, u64 address,
 				size_t length,
 				enum hns_roce_mmap_type mmap_type);
+bool check_sl_valid(struct hns_roce_dev *hr_dev, u8 sl);
+
 #endif /* _HNS_ROCE_DEVICE_H */

drivers/infiniband/hw/hns/hns_roce_hem.c

@@ -281,7 +281,7 @@ static struct hns_roce_hem *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
 	return hem;
 
 fail:
-	hns_roce_free_hem(hr_dev, hem);
+	kfree(hem);
 	return NULL;
 }
 
@@ -877,6 +877,7 @@ void hns_roce_cleanup_hem_table(struct hns_roce_dev *hr_dev,
 
 	if (hns_roce_check_whether_mhop(hr_dev, table->type)) {
 		hns_roce_cleanup_mhop_hem_table(hr_dev, table);
+		mutex_destroy(&table->mutex);
 		return;
 	}
 
@@ -891,6 +892,7 @@ void hns_roce_cleanup_hem_table(struct hns_roce_dev *hr_dev,
 			hns_roce_free_hem(hr_dev, table->hem[i]);
 		}
 
+	mutex_destroy(&table->mutex);
 	kfree(table->hem);
 }
 
@@ -986,15 +988,13 @@ static void hem_list_free_all(struct hns_roce_dev *hr_dev,
 	}
 }
 
-static void hem_list_link_bt(struct hns_roce_dev *hr_dev, void *base_addr,
-			     u64 table_addr)
+static void hem_list_link_bt(void *base_addr, u64 table_addr)
 {
 	*(u64 *)(base_addr) = table_addr;
 }
 
 /* assign L0 table address to hem from root bt */
-static void hem_list_assign_bt(struct hns_roce_dev *hr_dev,
-			       struct hns_roce_hem_item *hem, void *cpu_addr,
+static void hem_list_assign_bt(struct hns_roce_hem_item *hem, void *cpu_addr,
 			       u64 phy_addr)
 {
 	hem->addr = cpu_addr;
@@ -1163,8 +1163,7 @@ static int hem_list_alloc_mid_bt(struct hns_roce_dev *hr_dev,
 		if (level > 1) {
 			pre = hem_ptrs[level - 1];
 			step = (cur->start - pre->start) / step * BA_BYTE_LEN;
-			hem_list_link_bt(hr_dev, pre->addr + step,
-					 cur->dma_addr);
+			hem_list_link_bt(pre->addr + step, cur->dma_addr);
 		}
 	}
 
@@ -1222,7 +1221,7 @@ static int alloc_fake_root_bt(struct hns_roce_dev *hr_dev, void *cpu_base,
 	if (!hem)
 		return -ENOMEM;
 
-	hem_list_assign_bt(hr_dev, hem, cpu_base, phy_base);
+	hem_list_assign_bt(hem, cpu_base, phy_base);
 	list_add(&hem->list, branch_head);
 	list_add(&hem->sibling, leaf_head);
 
@@ -1245,7 +1244,7 @@ static int setup_middle_bt(struct hns_roce_dev *hr_dev, void *cpu_base,
 	/* if exist mid bt, link L1 to L0 */
 	list_for_each_entry_safe(hem, temp_hem, branch_head, list) {
 		offset = (hem->start - r->offset) / step * BA_BYTE_LEN;
-		hem_list_link_bt(hr_dev, cpu_base + offset, hem->dma_addr);
+		hem_list_link_bt(cpu_base + offset, hem->dma_addr);
 		total++;
 	}
 

drivers/infiniband/hw/hns/hns_roce_hem.h

@@ -57,16 +57,16 @@ enum {
 };
 
 #define check_whether_bt_num_3(type, hop_num) \
-	(type < HEM_TYPE_MTT && hop_num == 2)
+	((type) < HEM_TYPE_MTT && (hop_num) == 2)
 
 #define check_whether_bt_num_2(type, hop_num) \
-	((type < HEM_TYPE_MTT && hop_num == 1) || \
-	(type >= HEM_TYPE_MTT && hop_num == 2))
+	(((type) < HEM_TYPE_MTT && (hop_num) == 1) || \
+	((type) >= HEM_TYPE_MTT && (hop_num) == 2))
 
 #define check_whether_bt_num_1(type, hop_num) \
-	((type < HEM_TYPE_MTT && hop_num == HNS_ROCE_HOP_NUM_0) || \
-	(type >= HEM_TYPE_MTT && hop_num == 1) || \
-	(type >= HEM_TYPE_MTT && hop_num == HNS_ROCE_HOP_NUM_0))
+	(((type) < HEM_TYPE_MTT && (hop_num) == HNS_ROCE_HOP_NUM_0) || \
+	((type) >= HEM_TYPE_MTT && (hop_num) == 1) || \
+	((type) >= HEM_TYPE_MTT && (hop_num) == HNS_ROCE_HOP_NUM_0))
 
 struct hns_roce_hem {
 	void *buf;

drivers/infiniband/hw/hns/hns_roce_hw_v2.c

@@ -443,10 +443,6 @@ static int fill_ud_av(struct hns_roce_v2_ud_send_wqe *ud_sq_wqe,
 	hr_reg_write(ud_sq_wqe, UD_SEND_WQE_HOPLIMIT, ah->av.hop_limit);
 	hr_reg_write(ud_sq_wqe, UD_SEND_WQE_TCLASS, ah->av.tclass);
 	hr_reg_write(ud_sq_wqe, UD_SEND_WQE_FLOW_LABEL, ah->av.flowlabel);
-
-	if (WARN_ON(ah->av.sl > MAX_SERVICE_LEVEL))
-		return -EINVAL;
-
 	hr_reg_write(ud_sq_wqe, UD_SEND_WQE_SL, ah->av.sl);
 
 	ud_sq_wqe->sgid_index = ah->av.gid_index;
@@ -2105,7 +2101,7 @@ static void apply_func_caps(struct hns_roce_dev *hr_dev)
 					 caps->gmv_bt_num *
 					 (HNS_HW_PAGE_SIZE / caps->gmv_entry_sz));
 
-		caps->gmv_entry_num = caps->gmv_bt_num * (PAGE_SIZE /
+		caps->gmv_entry_num = caps->gmv_bt_num * (HNS_HW_PAGE_SIZE /
 							  caps->gmv_entry_sz);
 	} else {
 		u32 func_num = max_t(u32, 1, hr_dev->func_num);
@@ -2671,6 +2667,8 @@ static void free_mr_exit(struct hns_roce_dev *hr_dev)
 		kfree(free_mr->rsv_pd);
 		free_mr->rsv_pd = NULL;
 	}
+
+	mutex_destroy(&free_mr->mutex);
 }
 
 static int free_mr_alloc_res(struct hns_roce_dev *hr_dev)
@@ -2821,8 +2819,10 @@ static int free_mr_init(struct hns_roce_dev *hr_dev)
 	mutex_init(&free_mr->mutex);
 
 	ret = free_mr_alloc_res(hr_dev);
-	if (ret)
+	if (ret) {
+		mutex_destroy(&free_mr->mutex);
 		return ret;
+	}
 
 	ret = free_mr_modify_qp(hr_dev);
 	if (ret)
@@ -3208,13 +3208,14 @@ static int set_mtpt_pbl(struct hns_roce_dev *hr_dev,
 
 	/* Aligned to the hardware address access unit */
 	for (i = 0; i < ARRAY_SIZE(pages); i++)
-		pages[i] >>= 6;
+		pages[i] >>= MPT_PBL_BUF_ADDR_S;
 
 	pbl_ba = hns_roce_get_mtr_ba(&mr->pbl_mtr);
 
 	mpt_entry->pbl_size = cpu_to_le32(mr->npages);
-	mpt_entry->pbl_ba_l = cpu_to_le32(pbl_ba >> 3);
-	hr_reg_write(mpt_entry, MPT_PBL_BA_H, upper_32_bits(pbl_ba >> 3));
+	mpt_entry->pbl_ba_l = cpu_to_le32(pbl_ba >> MPT_PBL_BA_ADDR_S);
+	hr_reg_write(mpt_entry, MPT_PBL_BA_H,
+		     upper_32_bits(pbl_ba >> MPT_PBL_BA_ADDR_S));
 
 	mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
 	hr_reg_write(mpt_entry, MPT_PA0_H, upper_32_bits(pages[0]));
@@ -3307,8 +3308,7 @@ static int hns_roce_v2_rereg_write_mtpt(struct hns_roce_dev *hr_dev,
 	return ret;
 }
 
-static int hns_roce_v2_frmr_write_mtpt(struct hns_roce_dev *hr_dev,
-				       void *mb_buf, struct hns_roce_mr *mr)
+static int hns_roce_v2_frmr_write_mtpt(void *mb_buf, struct hns_roce_mr *mr)
 {
 	dma_addr_t pbl_ba = hns_roce_get_mtr_ba(&mr->pbl_mtr);
 	struct hns_roce_v2_mpt_entry *mpt_entry;
@@ -3335,8 +3335,10 @@ static int hns_roce_v2_frmr_write_mtpt(struct hns_roce_dev *hr_dev,
 
 	mpt_entry->pbl_size = cpu_to_le32(mr->npages);
 
-	mpt_entry->pbl_ba_l = cpu_to_le32(lower_32_bits(pbl_ba >> 3));
-	hr_reg_write(mpt_entry, MPT_PBL_BA_H, upper_32_bits(pbl_ba >> 3));
+	mpt_entry->pbl_ba_l = cpu_to_le32(lower_32_bits(pbl_ba >>
+							MPT_PBL_BA_ADDR_S));
+	hr_reg_write(mpt_entry, MPT_PBL_BA_H,
+		     upper_32_bits(pbl_ba >> MPT_PBL_BA_ADDR_S));
 
 	return 0;
 }
@@ -3582,14 +3584,14 @@ static void hns_roce_v2_write_cqc(struct hns_roce_dev *hr_dev,
 		     to_hr_hw_page_shift(hr_cq->mtr.hem_cfg.ba_pg_shift));
 	hr_reg_write(cq_context, CQC_CQE_BUF_PG_SZ,
 		     to_hr_hw_page_shift(hr_cq->mtr.hem_cfg.buf_pg_shift));
-	hr_reg_write(cq_context, CQC_CQE_BA_L, dma_handle >> 3);
-	hr_reg_write(cq_context, CQC_CQE_BA_H, (dma_handle >> (32 + 3)));
+	hr_reg_write(cq_context, CQC_CQE_BA_L, dma_handle >> CQC_CQE_BA_L_S);
+	hr_reg_write(cq_context, CQC_CQE_BA_H, dma_handle >> CQC_CQE_BA_H_S);
 	hr_reg_write_bool(cq_context, CQC_DB_RECORD_EN,
 			  hr_cq->flags & HNS_ROCE_CQ_FLAG_RECORD_DB);
 	hr_reg_write(cq_context, CQC_CQE_DB_RECORD_ADDR_L,
 		     ((u32)hr_cq->db.dma) >> 1);
 	hr_reg_write(cq_context, CQC_CQE_DB_RECORD_ADDR_H,
-		     hr_cq->db.dma >> 32);
+		     hr_cq->db.dma >> CQC_CQE_DB_RECORD_ADDR_H_S);
 	hr_reg_write(cq_context, CQC_CQ_MAX_CNT,
 		     HNS_ROCE_V2_CQ_DEFAULT_BURST_NUM);
 	hr_reg_write(cq_context, CQC_CQ_PERIOD,
@@ -3711,8 +3713,9 @@ static void get_cqe_status(struct hns_roce_dev *hr_dev, struct hns_roce_qp *qp,
 		   wc->status == IB_WC_WR_FLUSH_ERR))
 		return;
 
-	ibdev_err(&hr_dev->ib_dev, "error cqe status 0x%x:\n", cqe_status);
-	print_hex_dump(KERN_ERR, "", DUMP_PREFIX_NONE, 16, 4, cqe,
+	ibdev_err_ratelimited(&hr_dev->ib_dev, "error cqe status 0x%x:\n",
+			      cqe_status);
+	print_hex_dump(KERN_DEBUG, "", DUMP_PREFIX_NONE, 16, 4, cqe,
 		       cq->cqe_size, false);
 	wc->vendor_err = hr_reg_read(cqe, CQE_SUB_STATUS);
 
@@ -4217,8 +4220,7 @@ static void set_access_flags(struct hns_roce_qp *hr_qp,
 }
 
 static void set_qpc_wqe_cnt(struct hns_roce_qp *hr_qp,
-			    struct hns_roce_v2_qp_context *context,
-			    struct hns_roce_v2_qp_context *qpc_mask)
+			    struct hns_roce_v2_qp_context *context)
 {
 	hr_reg_write(context, QPC_SGE_SHIFT,
 		     to_hr_hem_entries_shift(hr_qp->sge.sge_cnt,
@@ -4240,7 +4242,6 @@ static inline int get_pdn(struct ib_pd *ib_pd)
 }
 
 static void modify_qp_reset_to_init(struct ib_qp *ibqp,
-				    const struct ib_qp_attr *attr,
 				    struct hns_roce_v2_qp_context *context,
 				    struct hns_roce_v2_qp_context *qpc_mask)
 {
@@ -4259,7 +4260,7 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp,
 
 	hr_reg_write(context, QPC_RQWS, ilog2(hr_qp->rq.max_gs));
 
-	set_qpc_wqe_cnt(hr_qp, context, qpc_mask);
+	set_qpc_wqe_cnt(hr_qp, context);
 
 	/* No VLAN need to set 0xFFF */
 	hr_reg_write(context, QPC_VLAN_ID, 0xfff);
@@ -4300,7 +4301,6 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp,
 }
 
 static void modify_qp_init_to_init(struct ib_qp *ibqp,
-				   const struct ib_qp_attr *attr,
 				   struct hns_roce_v2_qp_context *context,
 				   struct hns_roce_v2_qp_context *qpc_mask)
 {
@@ -4521,16 +4521,16 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 		return -EINVAL;
 	}
 
-	hr_reg_write(context, QPC_TRRL_BA_L, trrl_ba >> 4);
+	hr_reg_write(context, QPC_TRRL_BA_L, trrl_ba >> QPC_TRRL_BA_L_S);
 	hr_reg_clear(qpc_mask, QPC_TRRL_BA_L);
-	context->trrl_ba = cpu_to_le32(trrl_ba >> (16 + 4));
+	context->trrl_ba = cpu_to_le32(trrl_ba >> QPC_TRRL_BA_M_S);
 	qpc_mask->trrl_ba = 0;
-	hr_reg_write(context, QPC_TRRL_BA_H, trrl_ba >> (32 + 16 + 4));
+	hr_reg_write(context, QPC_TRRL_BA_H, trrl_ba >> QPC_TRRL_BA_H_S);
 	hr_reg_clear(qpc_mask, QPC_TRRL_BA_H);
 
-	context->irrl_ba = cpu_to_le32(irrl_ba >> 6);
+	context->irrl_ba = cpu_to_le32(irrl_ba >> QPC_IRRL_BA_L_S);
 	qpc_mask->irrl_ba = 0;
-	hr_reg_write(context, QPC_IRRL_BA_H, irrl_ba >> (32 + 6));
+	hr_reg_write(context, QPC_IRRL_BA_H, irrl_ba >> QPC_IRRL_BA_H_S);
 	hr_reg_clear(qpc_mask, QPC_IRRL_BA_H);
 
 	hr_reg_enable(context, QPC_RMT_E2E);
@@ -4592,8 +4592,9 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	hr_reg_clear(qpc_mask, QPC_TRRL_HEAD_MAX);
 	hr_reg_clear(qpc_mask, QPC_TRRL_TAIL_MAX);
 
+#define MAX_LP_SGEN 3
 	/* rocee send 2^lp_sgen_ini segs every time */
-	hr_reg_write(context, QPC_LP_SGEN_INI, 3);
+	hr_reg_write(context, QPC_LP_SGEN_INI, MAX_LP_SGEN);
 	hr_reg_clear(qpc_mask, QPC_LP_SGEN_INI);
 
 	if (udata && ibqp->qp_type == IB_QPT_RC &&
@@ -4619,8 +4620,7 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	return 0;
 }
 
-static int modify_qp_rtr_to_rts(struct ib_qp *ibqp,
-				const struct ib_qp_attr *attr, int attr_mask,
+static int modify_qp_rtr_to_rts(struct ib_qp *ibqp, int attr_mask,
 				struct hns_roce_v2_qp_context *context,
 				struct hns_roce_v2_qp_context *qpc_mask)
 {
@@ -4685,7 +4685,7 @@ static int get_dip_ctx_idx(struct ib_qp *ibqp, const struct ib_qp_attr *attr,
 	*tail = (*tail == hr_dev->caps.num_qps - 1) ? 0 : (*tail + 1);
 
 	list_for_each_entry(hr_dip, &hr_dev->dip_list, node) {
-		if (!memcmp(grh->dgid.raw, hr_dip->dgid, 16)) {
+		if (!memcmp(grh->dgid.raw, hr_dip->dgid, GID_LEN_V2)) {
 			*dip_idx = hr_dip->dip_idx;
 			goto out;
 		}
@@ -4828,6 +4828,69 @@ static int fill_cong_field(struct ib_qp *ibqp, const struct ib_qp_attr *attr,
 	return 0;
 }
 
+static int hns_roce_hw_v2_get_dscp(struct hns_roce_dev *hr_dev, u8 dscp,
+				   u8 *tc_mode, u8 *priority)
+{
+	struct hns_roce_v2_priv *priv = hr_dev->priv;
+	struct hnae3_handle *handle = priv->handle;
+	const struct hnae3_ae_ops *ops = handle->ae_algo->ops;
+
+	if (!ops->get_dscp_prio)
+		return -EOPNOTSUPP;
+
+	return ops->get_dscp_prio(handle, dscp, tc_mode, priority);
+}
+
+bool check_sl_valid(struct hns_roce_dev *hr_dev, u8 sl)
+{
+	u32 max_sl;
+
+	max_sl = min_t(u32, MAX_SERVICE_LEVEL, hr_dev->caps.sl_num - 1);
+	if (unlikely(sl > max_sl)) {
+		ibdev_err_ratelimited(&hr_dev->ib_dev,
+				      "failed to set SL(%u). Shouldn't be larger than %u.\n",
+				      sl, max_sl);
+		return false;
+	}
+
+	return true;
+}
+
+static int hns_roce_set_sl(struct ib_qp *ibqp,
+			   const struct ib_qp_attr *attr,
+			   struct hns_roce_v2_qp_context *context,
+			   struct hns_roce_v2_qp_context *qpc_mask)
+{
+	const struct ib_global_route *grh = rdma_ah_read_grh(&attr->ah_attr);
+	struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device);
+	struct hns_roce_qp *hr_qp = to_hr_qp(ibqp);
+	struct ib_device *ibdev = &hr_dev->ib_dev;
+	int ret;
+
+	ret = hns_roce_hw_v2_get_dscp(hr_dev, get_tclass(&attr->ah_attr.grh),
+				      &hr_qp->tc_mode, &hr_qp->priority);
+	if (ret && ret != -EOPNOTSUPP &&
+	    grh->sgid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) {
+		ibdev_err_ratelimited(ibdev,
+				      "failed to get dscp, ret = %d.\n", ret);
+		return ret;
+	}
+
+	if (hr_qp->tc_mode == HNAE3_TC_MAP_MODE_DSCP &&
+	    grh->sgid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP)
+		hr_qp->sl = hr_qp->priority;
+	else
+		hr_qp->sl = rdma_ah_get_sl(&attr->ah_attr);
+
+	if (!check_sl_valid(hr_dev, hr_qp->sl))
+		return -EINVAL;
+
+	hr_reg_write(context, QPC_SL, hr_qp->sl);
+	hr_reg_clear(qpc_mask, QPC_SL);
+
+	return 0;
+}
+
 static int hns_roce_v2_set_path(struct ib_qp *ibqp,
 				const struct ib_qp_attr *attr,
 				int attr_mask,
@@ -4843,25 +4906,18 @@ static int hns_roce_v2_set_path(struct ib_qp *ibqp,
 	int is_roce_protocol;
 	u16 vlan_id = 0xffff;
 	bool is_udp = false;
-	u32 max_sl;
 	u8 ib_port;
 	u8 hr_port;
 	int ret;
 
-	max_sl = min_t(u32, MAX_SERVICE_LEVEL, hr_dev->caps.sl_num - 1);
-	if (unlikely(sl > max_sl)) {
-		ibdev_err_ratelimited(ibdev,
-				      "failed to fill QPC, sl (%u) shouldn't be larger than %u.\n",
-				      sl, max_sl);
-		return -EINVAL;
-	}
-
 	/*
 	 * If free_mr_en of qp is set, it means that this qp comes from
 	 * free mr. This qp will perform the loopback operation.
 	 * In the loopback scenario, only sl needs to be set.
 	 */
 	if (hr_qp->free_mr_en) {
+		if (!check_sl_valid(hr_dev, sl))
+			return -EINVAL;
 		hr_reg_write(context, QPC_SL, sl);
 		hr_reg_clear(qpc_mask, QPC_SL);
 		hr_qp->sl = sl;
@@ -4931,11 +4987,7 @@ static int hns_roce_v2_set_path(struct ib_qp *ibqp,
 	memcpy(context->dgid, grh->dgid.raw, sizeof(grh->dgid.raw));
 	memset(qpc_mask->dgid, 0, sizeof(grh->dgid.raw));
 
-	hr_qp->sl = sl;
-	hr_reg_write(context, QPC_SL, hr_qp->sl);
-	hr_reg_clear(qpc_mask, QPC_SL);
-
-	return 0;
+	return  hns_roce_set_sl(ibqp, attr, context, qpc_mask);
 }
 
 static bool check_qp_state(enum ib_qp_state cur_state,
@@ -4982,15 +5034,14 @@ static int hns_roce_v2_set_abs_fields(struct ib_qp *ibqp,
 
 	if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) {
 		memset(qpc_mask, 0, hr_dev->caps.qpc_sz);
-		modify_qp_reset_to_init(ibqp, attr, context, qpc_mask);
+		modify_qp_reset_to_init(ibqp, context, qpc_mask);
 	} else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_INIT) {
-		modify_qp_init_to_init(ibqp, attr, context, qpc_mask);
+		modify_qp_init_to_init(ibqp, context, qpc_mask);
 	} else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) {
 		ret = modify_qp_init_to_rtr(ibqp, attr, attr_mask, context,
 					    qpc_mask, udata);
 	} else if (cur_state == IB_QPS_RTR && new_state == IB_QPS_RTS) {
-		ret = modify_qp_rtr_to_rts(ibqp, attr, attr_mask, context,
-					   qpc_mask);
+		ret = modify_qp_rtr_to_rts(ibqp, attr_mask, context, qpc_mask);
 	}
 
 	return ret;
@@ -5802,7 +5853,7 @@ static int hns_roce_v2_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
 			dev_info(hr_dev->dev,
 				 "cq_period(%u) reached the upper limit, adjusted to 65.\n",
 				 cq_period);
-			cq_period = HNS_ROCE_MAX_CQ_PERIOD;
+			cq_period = HNS_ROCE_MAX_CQ_PERIOD_HIP08;
 		}
 		cq_period *= HNS_ROCE_CLOCK_ADJUST;
 	}
@@ -6735,6 +6786,7 @@ static const struct hns_roce_hw hns_roce_hw_v2 = {
 	.query_srqc = hns_roce_v2_query_srqc,
 	.query_sccc = hns_roce_v2_query_sccc,
 	.query_hw_counter = hns_roce_hw_v2_query_counter,
+	.get_dscp = hns_roce_hw_v2_get_dscp,
 	.hns_roce_dev_ops = &hns_roce_v2_dev_ops,
 	.hns_roce_dev_srq_ops = &hns_roce_v2_dev_srq_ops,
 };

drivers/infiniband/hw/hns/hns_roce_hw_v2.h

@@ -276,6 +276,10 @@ struct hns_roce_v2_cq_context {
 	__le32 byte_64_se_cqe_idx;
 };
 
+#define CQC_CQE_BA_L_S 3
+#define CQC_CQE_BA_H_S (32 + CQC_CQE_BA_L_S)
+#define CQC_CQE_DB_RECORD_ADDR_H_S 32
+
 #define HNS_ROCE_V2_CQ_DEFAULT_BURST_NUM 0x0
 #define HNS_ROCE_V2_CQ_DEFAULT_INTERVAL	0x0
 
@@ -447,6 +451,12 @@ struct hns_roce_v2_qp_context {
 	struct hns_roce_v2_qp_context_ex ext;
 };
 
+#define QPC_TRRL_BA_L_S 4
+#define QPC_TRRL_BA_M_S (16 + QPC_TRRL_BA_L_S)
+#define QPC_TRRL_BA_H_S (32 + QPC_TRRL_BA_M_S)
+#define QPC_IRRL_BA_L_S 6
+#define QPC_IRRL_BA_H_S (32 + QPC_IRRL_BA_L_S)
+
 #define QPC_FIELD_LOC(h, l) FIELD_LOC(struct hns_roce_v2_qp_context, h, l)
 
 #define QPC_TST QPC_FIELD_LOC(2, 0)
@@ -716,6 +726,9 @@ struct hns_roce_v2_mpt_entry {
 	__le32	byte_64_buf_pa1;
 };
 
+#define MPT_PBL_BUF_ADDR_S 6
+#define MPT_PBL_BA_ADDR_S 3
+
 #define MPT_FIELD_LOC(h, l) FIELD_LOC(struct hns_roce_v2_mpt_entry, h, l)
 
 #define MPT_ST MPT_FIELD_LOC(1, 0)
@@ -1334,7 +1347,7 @@ struct fmea_ram_ecc {
 
 /* only for RNR timeout issue of HIP08 */
 #define HNS_ROCE_CLOCK_ADJUST 1000
-#define HNS_ROCE_MAX_CQ_PERIOD 65
+#define HNS_ROCE_MAX_CQ_PERIOD_HIP08 65
 #define HNS_ROCE_MAX_EQ_PERIOD 65
 #define HNS_ROCE_RNR_TIMER_10NS 1
 #define HNS_ROCE_1US_CFG 999

drivers/infiniband/hw/hns/hns_roce_main.c

@@ -37,9 +37,11 @@
 #include <rdma/ib_smi.h>
 #include <rdma/ib_user_verbs.h>
 #include <rdma/ib_cache.h>
+#include "hnae3.h"
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_hem.h"
+#include "hns_roce_hw_v2.h"
 
 static int hns_roce_set_mac(struct hns_roce_dev *hr_dev, u32 port,
 			    const u8 *addr)
@@ -192,6 +194,12 @@ static int hns_roce_query_device(struct ib_device *ib_dev,
 			    IB_ATOMIC_HCA : IB_ATOMIC_NONE;
 	props->max_pkeys = 1;
 	props->local_ca_ack_delay = hr_dev->caps.local_ca_ack_delay;
+	props->max_ah = INT_MAX;
+	props->cq_caps.max_cq_moderation_period = HNS_ROCE_MAX_CQ_PERIOD;
+	props->cq_caps.max_cq_moderation_count = HNS_ROCE_MAX_CQ_COUNT;
+	if (hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP08)
+		props->cq_caps.max_cq_moderation_period = HNS_ROCE_MAX_CQ_PERIOD_HIP08;
+
 	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) {
 		props->max_srq = hr_dev->caps.num_srqs;
 		props->max_srq_wr = hr_dev->caps.max_srq_wrs;
@@ -421,6 +429,9 @@ static int hns_roce_alloc_ucontext(struct ib_ucontext *uctx,
 	return 0;
 
 error_fail_copy_to_udata:
+	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_CQ_RECORD_DB ||
+	    hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_QP_RECORD_DB)
+		mutex_destroy(&context->page_mutex);
 	hns_roce_dealloc_uar_entry(context);
 
 error_fail_uar_entry:
@@ -437,6 +448,10 @@ static void hns_roce_dealloc_ucontext(struct ib_ucontext *ibcontext)
 	struct hns_roce_ucontext *context = to_hr_ucontext(ibcontext);
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibcontext->device);
 
+	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_CQ_RECORD_DB ||
+	    hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_QP_RECORD_DB)
+		mutex_destroy(&context->page_mutex);
+
 	hns_roce_dealloc_uar_entry(context);
 
 	ida_free(&hr_dev->uar_ida.ida, (int)context->uar.logic_idx);
@@ -925,6 +940,15 @@ err_unmap_dmpt:
 	return ret;
 }
 
+static void hns_roce_teardown_hca(struct hns_roce_dev *hr_dev)
+{
+	hns_roce_cleanup_bitmap(hr_dev);
+
+	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_CQ_RECORD_DB ||
+	    hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_QP_RECORD_DB)
+		mutex_destroy(&hr_dev->pgdir_mutex);
+}
+
 /**
  * hns_roce_setup_hca - setup host channel adapter
  * @hr_dev: pointer to hns roce device
@@ -973,6 +997,10 @@ static int hns_roce_setup_hca(struct hns_roce_dev *hr_dev)
 
 err_uar_table_free:
 	ida_destroy(&hr_dev->uar_ida.ida);
+	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_CQ_RECORD_DB ||
+	    hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_QP_RECORD_DB)
+		mutex_destroy(&hr_dev->pgdir_mutex);
+
 	return ret;
 }
 
@@ -1118,7 +1146,7 @@ error_failed_register_device:
 		hr_dev->hw->hw_exit(hr_dev);
 
 error_failed_engine_init:
-	hns_roce_cleanup_bitmap(hr_dev);
+	hns_roce_teardown_hca(hr_dev);
 
 error_failed_setup_hca:
 	hns_roce_cleanup_hem(hr_dev);
@@ -1148,7 +1176,7 @@ void hns_roce_exit(struct hns_roce_dev *hr_dev)
 
 	if (hr_dev->hw->hw_exit)
 		hr_dev->hw->hw_exit(hr_dev);
-	hns_roce_cleanup_bitmap(hr_dev);
+	hns_roce_teardown_hca(hr_dev);
 	hns_roce_cleanup_hem(hr_dev);
 
 	if (hr_dev->cmd_mod)

drivers/infiniband/hw/hns/hns_roce_mr.c

@@ -162,7 +162,7 @@ static int hns_roce_mr_enable(struct hns_roce_dev *hr_dev,
 	if (mr->type != MR_TYPE_FRMR)
 		ret = hr_dev->hw->write_mtpt(hr_dev, mailbox->buf, mr);
 	else
-		ret = hr_dev->hw->frmr_write_mtpt(hr_dev, mailbox->buf, mr);
+		ret = hr_dev->hw->frmr_write_mtpt(mailbox->buf, mr);
 	if (ret) {
 		dev_err(dev, "failed to write mtpt, ret = %d.\n", ret);
 		goto err_page;
@@ -441,18 +441,18 @@ int hns_roce_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents,
 	struct ib_device *ibdev = &hr_dev->ib_dev;
 	struct hns_roce_mr *mr = to_hr_mr(ibmr);
 	struct hns_roce_mtr *mtr = &mr->pbl_mtr;
-	int ret = 0;
+	int ret, sg_num = 0;
 
 	mr->npages = 0;
 	mr->page_list = kvcalloc(mr->pbl_mtr.hem_cfg.buf_pg_count,
 				 sizeof(dma_addr_t), GFP_KERNEL);
 	if (!mr->page_list)
-		return ret;
+		return sg_num;
 
-	ret = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, hns_roce_set_page);
-	if (ret < 1) {
+	sg_num = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, hns_roce_set_page);
+	if (sg_num < 1) {
 		ibdev_err(ibdev, "failed to store sg pages %u %u, cnt = %d.\n",
-			  mr->npages, mr->pbl_mtr.hem_cfg.buf_pg_count, ret);
+			  mr->npages, mr->pbl_mtr.hem_cfg.buf_pg_count, sg_num);
 		goto err_page_list;
 	}
 
@@ -463,17 +463,16 @@ int hns_roce_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents,
 	ret = hns_roce_mtr_map(hr_dev, mtr, mr->page_list, mr->npages);
 	if (ret) {
 		ibdev_err(ibdev, "failed to map sg mtr, ret = %d.\n", ret);
-		ret = 0;
+		sg_num = 0;
 	} else {
 		mr->pbl_mtr.hem_cfg.buf_pg_shift = (u32)ilog2(ibmr->page_size);
-		ret = mr->npages;
 	}
 
 err_page_list:
 	kvfree(mr->page_list);
 	mr->page_list = NULL;
 
-	return ret;
+	return sg_num;
 }
 
 static void hns_roce_mw_free(struct hns_roce_dev *hr_dev,
@@ -756,7 +755,7 @@ static int mtr_map_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr)
 		return -ENOMEM;
 
 	if (mtr->umem)
-		npage = hns_roce_get_umem_bufs(hr_dev, pages, page_count,
+		npage = hns_roce_get_umem_bufs(pages, page_count,
 					       mtr->umem, page_shift);
 	else
 		npage = hns_roce_get_kmem_bufs(hr_dev, pages, page_count,

drivers/infiniband/hw/hns/hns_roce_qp.c

@@ -410,7 +410,8 @@ static void free_qpn(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp)
 
 	bankid = get_qp_bankid(hr_qp->qpn);
 
-	ida_free(&hr_dev->qp_table.bank[bankid].ida, hr_qp->qpn >> 3);
+	ida_free(&hr_dev->qp_table.bank[bankid].ida,
+		 hr_qp->qpn / HNS_ROCE_QP_BANK_NUM);
 
 	mutex_lock(&hr_dev->qp_table.bank_mutex);
 	hr_dev->qp_table.bank[bankid].inuse--;
@@ -1117,7 +1118,6 @@ static int set_qp_param(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
 }
 
 static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
-				     struct ib_pd *ib_pd,
 				     struct ib_qp_init_attr *init_attr,
 				     struct ib_udata *udata,
 				     struct hns_roce_qp *hr_qp)
@@ -1140,7 +1140,7 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
 	ret = set_qp_param(hr_dev, hr_qp, init_attr, udata, &ucmd);
 	if (ret) {
 		ibdev_err(ibdev, "failed to set QP param, ret = %d.\n", ret);
-		return ret;
+		goto err_out;
 	}
 
 	if (!udata) {
@@ -1148,7 +1148,7 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
 		if (ret) {
 			ibdev_err(ibdev, "failed to alloc wrid, ret = %d.\n",
 				  ret);
-			return ret;
+			goto err_out;
 		}
 	}
 
@@ -1219,6 +1219,8 @@ err_qpn:
 	free_qp_buf(hr_dev, hr_qp);
 err_buf:
 	free_kernel_wrid(hr_qp);
+err_out:
+	mutex_destroy(&hr_qp->mutex);
 	return ret;
 }
 
@@ -1234,6 +1236,7 @@ void hns_roce_qp_destroy(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
 	free_qp_buf(hr_dev, hr_qp);
 	free_kernel_wrid(hr_qp);
 	free_qp_db(hr_dev, hr_qp, udata);
+	mutex_destroy(&hr_qp->mutex);
 }
 
 static int check_qp_type(struct hns_roce_dev *hr_dev, enum ib_qp_type type,
@@ -1271,7 +1274,6 @@ int hns_roce_create_qp(struct ib_qp *qp, struct ib_qp_init_attr *init_attr,
 	struct ib_device *ibdev = qp->device;
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibdev);
 	struct hns_roce_qp *hr_qp = to_hr_qp(qp);
-	struct ib_pd *pd = qp->pd;
 	int ret;
 
 	ret = check_qp_type(hr_dev, init_attr->qp_type, !!udata);
@@ -1286,7 +1288,7 @@ int hns_roce_create_qp(struct ib_qp *qp, struct ib_qp_init_attr *init_attr,
 		hr_qp->phy_port = hr_dev->iboe.phy_port[hr_qp->port];
 	}
 
-	ret = hns_roce_create_qp_common(hr_dev, pd, init_attr, udata, hr_qp);
+	ret = hns_roce_create_qp_common(hr_dev, init_attr, udata, hr_qp);
 	if (ret)
 		ibdev_err(ibdev, "create QP type 0x%x failed(%d)\n",
 			  init_attr->qp_type, ret);
@@ -1386,6 +1388,7 @@ int hns_roce_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 		       int attr_mask, struct ib_udata *udata)
 {
 	struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device);
+	struct hns_roce_ib_modify_qp_resp resp = {};
 	struct hns_roce_qp *hr_qp = to_hr_qp(ibqp);
 	enum ib_qp_state cur_state, new_state;
 	int ret = -EINVAL;
@@ -1427,6 +1430,18 @@ int hns_roce_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 
 	ret = hr_dev->hw->modify_qp(ibqp, attr, attr_mask, cur_state,
 				    new_state, udata);
+	if (ret)
+		goto out;
+
+	if (udata && udata->outlen) {
+		resp.tc_mode = hr_qp->tc_mode;
+		resp.priority = hr_qp->sl;
+		ret = ib_copy_to_udata(udata, &resp,
+				       min(udata->outlen, sizeof(resp)));
+		if (ret)
+			ibdev_err_ratelimited(&hr_dev->ib_dev,
+					      "failed to copy modify qp resp.\n");
+	}
 
 out:
 	mutex_unlock(&hr_qp->mutex);
@@ -1561,5 +1576,7 @@ void hns_roce_cleanup_qp_table(struct hns_roce_dev *hr_dev)
 
 	for (i = 0; i < HNS_ROCE_QP_BANK_NUM; i++)
 		ida_destroy(&hr_dev->qp_table.bank[i].ida);
+	mutex_destroy(&hr_dev->qp_table.bank_mutex);
+	mutex_destroy(&hr_dev->qp_table.scc_mutex);
 	kfree(hr_dev->qp_table.idx_table.spare_idx);
 }

drivers/infiniband/hw/hns/hns_roce_srq.c

@@ -123,7 +123,7 @@ static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
 		return ret;
 	}
 
-	ret = xa_err(xa_store(&srq_table->xa, srq->srqn, srq, GFP_KERNEL));
+	ret = xa_err(xa_store_irq(&srq_table->xa, srq->srqn, srq, GFP_KERNEL));
 	if (ret) {
 		ibdev_err(ibdev, "failed to store SRQC, ret = %d.\n", ret);
 		goto err_put;
@@ -136,7 +136,7 @@ static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
 	return 0;
 
 err_xa:
-	xa_erase(&srq_table->xa, srq->srqn);
+	xa_erase_irq(&srq_table->xa, srq->srqn);
 err_put:
 	hns_roce_table_put(hr_dev, &srq_table->table, srq->srqn);
 
@@ -154,7 +154,7 @@ static void free_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
 		dev_err(hr_dev->dev, "DESTROY_SRQ failed (%d) for SRQN %06lx\n",
 			ret, srq->srqn);
 
-	xa_erase(&srq_table->xa, srq->srqn);
+	xa_erase_irq(&srq_table->xa, srq->srqn);
 
 	if (refcount_dec_and_test(&srq->refcount))
 		complete(&srq->free);
@@ -250,7 +250,7 @@ static void free_srq_wqe_buf(struct hns_roce_dev *hr_dev,
 	hns_roce_mtr_destroy(hr_dev, &srq->buf_mtr);
 }
 
-static int alloc_srq_wrid(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq)
+static int alloc_srq_wrid(struct hns_roce_srq *srq)
 {
 	srq->wrid = kvmalloc_array(srq->wqe_cnt, sizeof(u64), GFP_KERNEL);
 	if (!srq->wrid)
@@ -366,7 +366,7 @@ static int alloc_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq,
 		goto err_idx;
 
 	if (!udata) {
-		ret = alloc_srq_wrid(hr_dev, srq);
+		ret = alloc_srq_wrid(srq);
 		if (ret)
 			goto err_wqe_buf;
 	}
@@ -518,6 +518,7 @@ err_srq_db:
 err_srq_buf:
 	free_srq_buf(hr_dev, srq);
 err_out:
+	mutex_destroy(&srq->mutex);
 	atomic64_inc(&hr_dev->dfx_cnt[HNS_ROCE_DFX_SRQ_CREATE_ERR_CNT]);
 
 	return ret;
@@ -532,6 +533,7 @@ int hns_roce_destroy_srq(struct ib_srq *ibsrq, struct ib_udata *udata)
 	free_srqn(hr_dev, srq);
 	free_srq_db(hr_dev, srq, udata);
 	free_srq_buf(hr_dev, srq);
+	mutex_destroy(&srq->mutex);
 	return 0;
 }
 

drivers/infiniband/hw/mana/cq.c

@@ -9,23 +9,23 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 		      struct ib_udata *udata)
 {
 	struct mana_ib_cq *cq = container_of(ibcq, struct mana_ib_cq, ibcq);
+	struct mana_ib_create_cq_resp resp = {};
+	struct mana_ib_ucontext *mana_ucontext;
 	struct ib_device *ibdev = ibcq->device;
 	struct mana_ib_create_cq ucmd = {};
 	struct mana_ib_dev *mdev;
-	struct gdma_context *gc;
+	bool is_rnic_cq;
+	u32 doorbell;
 	int err;
 
 	mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
-	gc = mdev_to_gc(mdev);
 
-	if (udata->inlen < sizeof(ucmd))
+	cq->comp_vector = attr->comp_vector % ibdev->num_comp_vectors;
+	cq->cq_handle = INVALID_MANA_HANDLE;
+
+	if (udata->inlen < offsetof(struct mana_ib_create_cq, flags))
 		return -EINVAL;
 
-	if (attr->comp_vector > gc->max_num_queues)
-		return -EINVAL;
-
-	cq->comp_vector = attr->comp_vector;
-
 	err = ib_copy_from_udata(&ucmd, udata, min(sizeof(ucmd), udata->inlen));
 	if (err) {
 		ibdev_dbg(ibdev,
@@ -33,42 +33,54 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 		return err;
 	}
 
-	if (attr->cqe > mdev->adapter_caps.max_qp_wr) {
+	is_rnic_cq = !!(ucmd.flags & MANA_IB_CREATE_RNIC_CQ);
+
+	if (!is_rnic_cq && attr->cqe > mdev->adapter_caps.max_qp_wr) {
 		ibdev_dbg(ibdev, "CQE %d exceeding limit\n", attr->cqe);
 		return -EINVAL;
 	}
 
 	cq->cqe = attr->cqe;
-	cq->umem = ib_umem_get(ibdev, ucmd.buf_addr, cq->cqe * COMP_ENTRY_SIZE,
-			       IB_ACCESS_LOCAL_WRITE);
-	if (IS_ERR(cq->umem)) {
-		err = PTR_ERR(cq->umem);
-		ibdev_dbg(ibdev, "Failed to get umem for create cq, err %d\n",
-			  err);
+	err = mana_ib_create_queue(mdev, ucmd.buf_addr, cq->cqe * COMP_ENTRY_SIZE, &cq->queue);
+	if (err) {
+		ibdev_dbg(ibdev, "Failed to create queue for create cq, %d\n", err);
 		return err;
 	}
 
-	err = mana_ib_create_zero_offset_dma_region(mdev, cq->umem, &cq->gdma_region);
-	if (err) {
-		ibdev_dbg(ibdev,
-			  "Failed to create dma region for create cq, %d\n",
-			  err);
-		goto err_release_umem;
+	mana_ucontext = rdma_udata_to_drv_context(udata, struct mana_ib_ucontext,
+						  ibucontext);
+	doorbell = mana_ucontext->doorbell;
+
+	if (is_rnic_cq) {
+		err = mana_ib_gd_create_cq(mdev, cq, doorbell);
+		if (err) {
+			ibdev_dbg(ibdev, "Failed to create RNIC cq, %d\n", err);
+			goto err_destroy_queue;
+		}
+
+		err = mana_ib_install_cq_cb(mdev, cq);
+		if (err) {
+			ibdev_dbg(ibdev, "Failed to install cq callback, %d\n", err);
+			goto err_destroy_rnic_cq;
+		}
 	}
 
-	ibdev_dbg(ibdev,
-		  "create_dma_region ret %d gdma_region 0x%llx\n",
-		  err, cq->gdma_region);
-
-	/*
-	 * The CQ ID is not known at this time. The ID is generated at create_qp
-	 */
-	cq->id = INVALID_QUEUE_ID;
+	resp.cqid = cq->queue.id;
+	err = ib_copy_to_udata(udata, &resp, min(sizeof(resp), udata->outlen));
+	if (err) {
+		ibdev_dbg(&mdev->ib_dev, "Failed to copy to udata, %d\n", err);
+		goto err_remove_cq_cb;
+	}
 
 	return 0;
 
-err_release_umem:
-	ib_umem_release(cq->umem);
+err_remove_cq_cb:
+	mana_ib_remove_cq_cb(mdev, cq);
+err_destroy_rnic_cq:
+	mana_ib_gd_destroy_cq(mdev, cq);
+err_destroy_queue:
+	mana_ib_destroy_queue(mdev, &cq->queue);
+
 	return err;
 }
 
@@ -77,25 +89,17 @@ int mana_ib_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
 	struct mana_ib_cq *cq = container_of(ibcq, struct mana_ib_cq, ibcq);
 	struct ib_device *ibdev = ibcq->device;
 	struct mana_ib_dev *mdev;
-	struct gdma_context *gc;
-	int err;
 
 	mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
-	gc = mdev_to_gc(mdev);
 
-	err = mana_ib_gd_destroy_dma_region(mdev, cq->gdma_region);
-	if (err) {
-		ibdev_dbg(ibdev,
-			  "Failed to destroy dma region, %d\n", err);
-		return err;
-	}
+	mana_ib_remove_cq_cb(mdev, cq);
 
-	if (cq->id != INVALID_QUEUE_ID) {
-		kfree(gc->cq_table[cq->id]);
-		gc->cq_table[cq->id] = NULL;
-	}
+	/* Ignore return code as there is not much we can do about it.
+	 * The error message is printed inside.
+	 */
+	mana_ib_gd_destroy_cq(mdev, cq);
 
-	ib_umem_release(cq->umem);
+	mana_ib_destroy_queue(mdev, &cq->queue);
 
 	return 0;
 }
@@ -113,8 +117,10 @@ int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
 	struct gdma_context *gc = mdev_to_gc(mdev);
 	struct gdma_queue *gdma_cq;
 
+	if (cq->queue.id >= gc->max_num_cqs)
+		return -EINVAL;
 	/* Create CQ table entry */
-	WARN_ON(gc->cq_table[cq->id]);
+	WARN_ON(gc->cq_table[cq->queue.id]);
 	gdma_cq = kzalloc(sizeof(*gdma_cq), GFP_KERNEL);
 	if (!gdma_cq)
 		return -ENOMEM;
@@ -122,7 +128,18 @@ int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
 	gdma_cq->cq.context = cq;
 	gdma_cq->type = GDMA_CQ;
 	gdma_cq->cq.callback = mana_ib_cq_handler;
-	gdma_cq->id = cq->id;
-	gc->cq_table[cq->id] = gdma_cq;
+	gdma_cq->id = cq->queue.id;
+	gc->cq_table[cq->queue.id] = gdma_cq;
 	return 0;
 }
+
+void mana_ib_remove_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
+{
+	struct gdma_context *gc = mdev_to_gc(mdev);
+
+	if (cq->queue.id >= gc->max_num_cqs || cq->queue.id == INVALID_QUEUE_ID)
+		return;
+
+	kfree(gc->cq_table[cq->queue.id]);
+	gc->cq_table[cq->queue.id] = NULL;
+}

drivers/infiniband/hw/mana/device.c

@@ -15,6 +15,7 @@ static const struct ib_device_ops mana_ib_dev_ops = {
 	.driver_id = RDMA_DRIVER_MANA,
 	.uverbs_abi_ver = MANA_IB_UVERBS_ABI_VERSION,
 
+	.add_gid = mana_ib_gd_add_gid,
 	.alloc_pd = mana_ib_alloc_pd,
 	.alloc_ucontext = mana_ib_alloc_ucontext,
 	.create_cq = mana_ib_create_cq,
@@ -23,18 +24,21 @@ static const struct ib_device_ops mana_ib_dev_ops = {
 	.create_wq = mana_ib_create_wq,
 	.dealloc_pd = mana_ib_dealloc_pd,
 	.dealloc_ucontext = mana_ib_dealloc_ucontext,
+	.del_gid = mana_ib_gd_del_gid,
 	.dereg_mr = mana_ib_dereg_mr,
 	.destroy_cq = mana_ib_destroy_cq,
 	.destroy_qp = mana_ib_destroy_qp,
 	.destroy_rwq_ind_table = mana_ib_destroy_rwq_ind_table,
 	.destroy_wq = mana_ib_destroy_wq,
 	.disassociate_ucontext = mana_ib_disassociate_ucontext,
+	.get_link_layer = mana_ib_get_link_layer,
 	.get_port_immutable = mana_ib_get_port_immutable,
 	.mmap = mana_ib_mmap,
 	.modify_qp = mana_ib_modify_qp,
 	.modify_wq = mana_ib_modify_wq,
 	.query_device = mana_ib_query_device,
 	.query_gid = mana_ib_query_gid,
+	.query_pkey = mana_ib_query_pkey,
 	.query_port = mana_ib_query_port,
 	.reg_user_mr = mana_ib_reg_user_mr,
 
@@ -51,8 +55,10 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 {
 	struct mana_adev *madev = container_of(adev, struct mana_adev, adev);
 	struct gdma_dev *mdev = madev->mdev;
+	struct net_device *upper_ndev;
 	struct mana_context *mc;
 	struct mana_ib_dev *dev;
+	u8 mac_addr[ETH_ALEN];
 	int ret;
 
 	mc = mdev->driver_data;
@@ -74,9 +80,25 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 	 * num_comp_vectors needs to set to the max MSIX index
 	 * when interrupts and event queues are implemented
 	 */
-	dev->ib_dev.num_comp_vectors = 1;
+	dev->ib_dev.num_comp_vectors = mdev->gdma_context->max_num_queues;
 	dev->ib_dev.dev.parent = mdev->gdma_context->dev;
 
+	rcu_read_lock(); /* required to get upper dev */
+	upper_ndev = netdev_master_upper_dev_get_rcu(mc->ports[0]);
+	if (!upper_ndev) {
+		rcu_read_unlock();
+		ret = -ENODEV;
+		ibdev_err(&dev->ib_dev, "Failed to get master netdev");
+		goto free_ib_device;
+	}
+	ether_addr_copy(mac_addr, upper_ndev->dev_addr);
+	ret = ib_device_set_netdev(&dev->ib_dev, upper_ndev, 1);
+	rcu_read_unlock();
+	if (ret) {
+		ibdev_err(&dev->ib_dev, "Failed to set ib netdev, ret %d", ret);
+		goto free_ib_device;
+	}
+
 	ret = mana_gd_register_device(&mdev->gdma_context->mana_ib);
 	if (ret) {
 		ibdev_err(&dev->ib_dev, "Failed to register device, ret %d",
@@ -92,15 +114,36 @@ static int mana_ib_probe(struct auxiliary_device *adev,
 		goto deregister_device;
 	}
 
+	ret = mana_ib_create_eqs(dev);
+	if (ret) {
+		ibdev_err(&dev->ib_dev, "Failed to create EQs, ret %d", ret);
+		goto deregister_device;
+	}
+
+	ret = mana_ib_gd_create_rnic_adapter(dev);
+	if (ret)
+		goto destroy_eqs;
+
+	ret = mana_ib_gd_config_mac(dev, ADDR_OP_ADD, mac_addr);
+	if (ret) {
+		ibdev_err(&dev->ib_dev, "Failed to add Mac address, ret %d",
+			  ret);
+		goto destroy_rnic;
+	}
+
 	ret = ib_register_device(&dev->ib_dev, "mana_%d",
 				 mdev->gdma_context->dev);
 	if (ret)
-		goto deregister_device;
+		goto destroy_rnic;
 
 	dev_set_drvdata(&adev->dev, dev);
 
 	return 0;
 
+destroy_rnic:
+	mana_ib_gd_destroy_rnic_adapter(dev);
+destroy_eqs:
+	mana_ib_destroy_eqs(dev);
 deregister_device:
 	mana_gd_deregister_device(dev->gdma_dev);
 free_ib_device:
@@ -113,9 +156,9 @@ static void mana_ib_remove(struct auxiliary_device *adev)
 	struct mana_ib_dev *dev = dev_get_drvdata(&adev->dev);
 
 	ib_unregister_device(&dev->ib_dev);
-
+	mana_ib_gd_destroy_rnic_adapter(dev);
+	mana_ib_destroy_eqs(dev);
 	mana_gd_deregister_device(dev->gdma_dev);
-
 	ib_dealloc_device(&dev->ib_dev);
 }
 

drivers/infiniband/hw/mana/main.c

@@ -237,6 +237,47 @@ void mana_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
 		ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
 }
 
+int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
+			 struct mana_ib_queue *queue)
+{
+	struct ib_umem *umem;
+	int err;
+
+	queue->umem = NULL;
+	queue->id = INVALID_QUEUE_ID;
+	queue->gdma_region = GDMA_INVALID_DMA_REGION;
+
+	umem = ib_umem_get(&mdev->ib_dev, addr, size, IB_ACCESS_LOCAL_WRITE);
+	if (IS_ERR(umem)) {
+		err = PTR_ERR(umem);
+		ibdev_dbg(&mdev->ib_dev, "Failed to get umem, %d\n", err);
+		return err;
+	}
+
+	err = mana_ib_create_zero_offset_dma_region(mdev, umem, &queue->gdma_region);
+	if (err) {
+		ibdev_dbg(&mdev->ib_dev, "Failed to create dma region, %d\n", err);
+		goto free_umem;
+	}
+	queue->umem = umem;
+
+	ibdev_dbg(&mdev->ib_dev, "created dma region 0x%llx\n", queue->gdma_region);
+
+	return 0;
+free_umem:
+	ib_umem_release(umem);
+	return err;
+}
+
+void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue)
+{
+	/* Ignore return code as there is not much we can do about it.
+	 * The error message is printed inside.
+	 */
+	mana_ib_gd_destroy_dma_region(mdev, queue->gdma_region);
+	ib_umem_release(queue->umem);
+}
+
 static int
 mana_ib_gd_first_dma_region(struct mana_ib_dev *dev,
 			    struct gdma_context *gc,
@@ -484,11 +525,18 @@ int mana_ib_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
 int mana_ib_get_port_immutable(struct ib_device *ibdev, u32 port_num,
 			       struct ib_port_immutable *immutable)
 {
-	/*
-	 * This version only support RAW_PACKET
-	 * other values need to be filled for other types
-	 */
+	struct ib_port_attr attr;
+	int err;
+
+	err = ib_query_port(ibdev, port_num, &attr);
+	if (err)
+		return err;
+
+	immutable->pkey_tbl_len = attr.pkey_tbl_len;
+	immutable->gid_tbl_len = attr.gid_tbl_len;
 	immutable->core_cap_flags = RDMA_CORE_PORT_RAW_PACKET;
+	if (port_num == 1)
+		immutable->core_cap_flags |= RDMA_CORE_PORT_IBA_ROCE_UDP_ENCAP;
 
 	return 0;
 }
@@ -514,7 +562,42 @@ int mana_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props,
 int mana_ib_query_port(struct ib_device *ibdev, u32 port,
 		       struct ib_port_attr *props)
 {
-	/* This version doesn't return port properties */
+	struct net_device *ndev = mana_ib_get_netdev(ibdev, port);
+
+	if (!ndev)
+		return -EINVAL;
+
+	memset(props, 0, sizeof(*props));
+	props->max_mtu = IB_MTU_4096;
+	props->active_mtu = ib_mtu_int_to_enum(ndev->mtu);
+
+	if (netif_carrier_ok(ndev) && netif_running(ndev)) {
+		props->state = IB_PORT_ACTIVE;
+		props->phys_state = IB_PORT_PHYS_STATE_LINK_UP;
+	} else {
+		props->state = IB_PORT_DOWN;
+		props->phys_state = IB_PORT_PHYS_STATE_DISABLED;
+	}
+
+	props->active_width = IB_WIDTH_4X;
+	props->active_speed = IB_SPEED_EDR;
+	props->pkey_tbl_len = 1;
+	if (port == 1)
+		props->gid_tbl_len = 16;
+
+	return 0;
+}
+
+enum rdma_link_layer mana_ib_get_link_layer(struct ib_device *device, u32 port_num)
+{
+	return IB_LINK_LAYER_ETHERNET;
+}
+
+int mana_ib_query_pkey(struct ib_device *ibdev, u32 port, u16 index, u16 *pkey)
+{
+	if (index != 0)
+		return -EINVAL;
+	*pkey = IB_DEFAULT_PKEY_FULL;
 	return 0;
 }
 
@@ -570,3 +653,238 @@ int mana_ib_gd_query_adapter_caps(struct mana_ib_dev *dev)
 
 	return 0;
 }
+
+int mana_ib_create_eqs(struct mana_ib_dev *mdev)
+{
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	struct gdma_queue_spec spec = {};
+	int err, i;
+
+	spec.type = GDMA_EQ;
+	spec.monitor_avl_buf = false;
+	spec.queue_size = EQ_SIZE;
+	spec.eq.callback = NULL;
+	spec.eq.context = mdev;
+	spec.eq.log2_throttle_limit = LOG2_EQ_THROTTLE;
+	spec.eq.msix_index = 0;
+
+	err = mana_gd_create_mana_eq(&gc->mana_ib, &spec, &mdev->fatal_err_eq);
+	if (err)
+		return err;
+
+	mdev->eqs = kcalloc(mdev->ib_dev.num_comp_vectors, sizeof(struct gdma_queue *),
+			    GFP_KERNEL);
+	if (!mdev->eqs) {
+		err = -ENOMEM;
+		goto destroy_fatal_eq;
+	}
+
+	for (i = 0; i < mdev->ib_dev.num_comp_vectors; i++) {
+		spec.eq.msix_index = (i + 1) % gc->num_msix_usable;
+		err = mana_gd_create_mana_eq(mdev->gdma_dev, &spec, &mdev->eqs[i]);
+		if (err)
+			goto destroy_eqs;
+	}
+
+	return 0;
+
+destroy_eqs:
+	while (i-- > 0)
+		mana_gd_destroy_queue(gc, mdev->eqs[i]);
+	kfree(mdev->eqs);
+destroy_fatal_eq:
+	mana_gd_destroy_queue(gc, mdev->fatal_err_eq);
+	return err;
+}
+
+void mana_ib_destroy_eqs(struct mana_ib_dev *mdev)
+{
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	int i;
+
+	mana_gd_destroy_queue(gc, mdev->fatal_err_eq);
+
+	for (i = 0; i < mdev->ib_dev.num_comp_vectors; i++)
+		mana_gd_destroy_queue(gc, mdev->eqs[i]);
+
+	kfree(mdev->eqs);
+}
+
+int mana_ib_gd_create_rnic_adapter(struct mana_ib_dev *mdev)
+{
+	struct mana_rnic_create_adapter_resp resp = {};
+	struct mana_rnic_create_adapter_req req = {};
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	int err;
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_ADAPTER, sizeof(req), sizeof(resp));
+	req.hdr.req.msg_version = GDMA_MESSAGE_V2;
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.notify_eq_id = mdev->fatal_err_eq->id;
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to create RNIC adapter err %d", err);
+		return err;
+	}
+	mdev->adapter_handle = resp.adapter;
+
+	return 0;
+}
+
+int mana_ib_gd_destroy_rnic_adapter(struct mana_ib_dev *mdev)
+{
+	struct mana_rnic_destroy_adapter_resp resp = {};
+	struct mana_rnic_destroy_adapter_req req = {};
+	struct gdma_context *gc;
+	int err;
+
+	gc = mdev_to_gc(mdev);
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_ADAPTER, sizeof(req), sizeof(resp));
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.adapter = mdev->adapter_handle;
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to destroy RNIC adapter err %d", err);
+		return err;
+	}
+
+	return 0;
+}
+
+int mana_ib_gd_add_gid(const struct ib_gid_attr *attr, void **context)
+{
+	struct mana_ib_dev *mdev = container_of(attr->device, struct mana_ib_dev, ib_dev);
+	enum rdma_network_type ntype = rdma_gid_attr_network_type(attr);
+	struct mana_rnic_config_addr_resp resp = {};
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	struct mana_rnic_config_addr_req req = {};
+	int err;
+
+	if (ntype != RDMA_NETWORK_IPV4 && ntype != RDMA_NETWORK_IPV6) {
+		ibdev_dbg(&mdev->ib_dev, "Unsupported rdma network type %d", ntype);
+		return -EINVAL;
+	}
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CONFIG_IP_ADDR, sizeof(req), sizeof(resp));
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.adapter = mdev->adapter_handle;
+	req.op = ADDR_OP_ADD;
+	req.sgid_type = (ntype == RDMA_NETWORK_IPV6) ? SGID_TYPE_IPV6 : SGID_TYPE_IPV4;
+	copy_in_reverse(req.ip_addr, attr->gid.raw, sizeof(union ib_gid));
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to config IP addr err %d\n", err);
+		return err;
+	}
+
+	return 0;
+}
+
+int mana_ib_gd_del_gid(const struct ib_gid_attr *attr, void **context)
+{
+	struct mana_ib_dev *mdev = container_of(attr->device, struct mana_ib_dev, ib_dev);
+	enum rdma_network_type ntype = rdma_gid_attr_network_type(attr);
+	struct mana_rnic_config_addr_resp resp = {};
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	struct mana_rnic_config_addr_req req = {};
+	int err;
+
+	if (ntype != RDMA_NETWORK_IPV4 && ntype != RDMA_NETWORK_IPV6) {
+		ibdev_dbg(&mdev->ib_dev, "Unsupported rdma network type %d", ntype);
+		return -EINVAL;
+	}
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CONFIG_IP_ADDR, sizeof(req), sizeof(resp));
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.adapter = mdev->adapter_handle;
+	req.op = ADDR_OP_REMOVE;
+	req.sgid_type = (ntype == RDMA_NETWORK_IPV6) ? SGID_TYPE_IPV6 : SGID_TYPE_IPV4;
+	copy_in_reverse(req.ip_addr, attr->gid.raw, sizeof(union ib_gid));
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to config IP addr err %d\n", err);
+		return err;
+	}
+
+	return 0;
+}
+
+int mana_ib_gd_config_mac(struct mana_ib_dev *mdev, enum mana_ib_addr_op op, u8 *mac)
+{
+	struct mana_rnic_config_mac_addr_resp resp = {};
+	struct mana_rnic_config_mac_addr_req req = {};
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	int err;
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CONFIG_MAC_ADDR, sizeof(req), sizeof(resp));
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.adapter = mdev->adapter_handle;
+	req.op = op;
+	copy_in_reverse(req.mac_addr, mac, ETH_ALEN);
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to config Mac addr err %d", err);
+		return err;
+	}
+
+	return 0;
+}
+
+int mana_ib_gd_create_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq, u32 doorbell)
+{
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	struct mana_rnic_create_cq_resp resp = {};
+	struct mana_rnic_create_cq_req req = {};
+	int err;
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_CREATE_CQ, sizeof(req), sizeof(resp));
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.adapter = mdev->adapter_handle;
+	req.gdma_region = cq->queue.gdma_region;
+	req.eq_id = mdev->eqs[cq->comp_vector]->id;
+	req.doorbell_page = doorbell;
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to create cq err %d", err);
+		return err;
+	}
+
+	cq->queue.id  = resp.cq_id;
+	cq->cq_handle = resp.cq_handle;
+	/* The GDMA region is now owned by the CQ handle */
+	cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
+
+	return 0;
+}
+
+int mana_ib_gd_destroy_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
+{
+	struct gdma_context *gc = mdev_to_gc(mdev);
+	struct mana_rnic_destroy_cq_resp resp = {};
+	struct mana_rnic_destroy_cq_req req = {};
+	int err;
+
+	if (cq->cq_handle == INVALID_MANA_HANDLE)
+		return 0;
+
+	mana_gd_init_req_hdr(&req.hdr, MANA_IB_DESTROY_CQ, sizeof(req), sizeof(resp));
+	req.hdr.dev_id = gc->mana_ib.dev_id;
+	req.adapter = mdev->adapter_handle;
+	req.cq_handle = cq->cq_handle;
+
+	err = mana_gd_send_request(gc, sizeof(req), &req, sizeof(resp), &resp);
+
+	if (err) {
+		ibdev_err(&mdev->ib_dev, "Failed to destroy cq err %d", err);
+		return err;
+	}
+
+	return 0;
+}

drivers/infiniband/hw/mana/mana_ib.h

@@ -45,19 +45,26 @@ struct mana_ib_adapter_caps {
 	u32 max_inline_data_size;
 };
 
+struct mana_ib_queue {
+	struct ib_umem *umem;
+	u64 gdma_region;
+	u64 id;
+};
+
 struct mana_ib_dev {
 	struct ib_device ib_dev;
 	struct gdma_dev *gdma_dev;
+	mana_handle_t adapter_handle;
+	struct gdma_queue *fatal_err_eq;
+	struct gdma_queue **eqs;
 	struct mana_ib_adapter_caps adapter_caps;
 };
 
 struct mana_ib_wq {
 	struct ib_wq ibwq;
-	struct ib_umem *umem;
+	struct mana_ib_queue queue;
 	int wqe;
 	u32 wq_buf_size;
-	u64 gdma_region;
-	u64 id;
 	mana_handle_t rx_object;
 };
 
@@ -82,22 +89,17 @@ struct mana_ib_mr {
 
 struct mana_ib_cq {
 	struct ib_cq ibcq;
-	struct ib_umem *umem;
+	struct mana_ib_queue queue;
 	int cqe;
-	u64 gdma_region;
-	u64 id;
 	u32 comp_vector;
+	mana_handle_t  cq_handle;
 };
 
 struct mana_ib_qp {
 	struct ib_qp ibqp;
 
-	/* Work queue info */
-	struct ib_umem *sq_umem;
-	int sqe;
-	u64 sq_gdma_region;
-	u64 sq_id;
-	mana_handle_t tx_object;
+	mana_handle_t qp_handle;
+	struct mana_ib_queue raw_sq;
 
 	/* The port on the IB device, starting with 1 */
 	u32 port;
@@ -114,6 +116,12 @@ struct mana_ib_rwq_ind_table {
 
 enum mana_ib_command_code {
 	MANA_IB_GET_ADAPTER_CAP = 0x30001,
+	MANA_IB_CREATE_ADAPTER  = 0x30002,
+	MANA_IB_DESTROY_ADAPTER = 0x30003,
+	MANA_IB_CONFIG_IP_ADDR	= 0x30004,
+	MANA_IB_CONFIG_MAC_ADDR	= 0x30005,
+	MANA_IB_CREATE_CQ       = 0x30008,
+	MANA_IB_DESTROY_CQ      = 0x30009,
 };
 
 struct mana_ib_query_adapter_caps_req {
@@ -142,6 +150,86 @@ struct mana_ib_query_adapter_caps_resp {
 	u32 max_inline_data_size;
 }; /* HW Data */
 
+struct mana_rnic_create_adapter_req {
+	struct gdma_req_hdr hdr;
+	u32 notify_eq_id;
+	u32 reserved;
+	u64 feature_flags;
+}; /*HW Data */
+
+struct mana_rnic_create_adapter_resp {
+	struct gdma_resp_hdr hdr;
+	mana_handle_t adapter;
+}; /* HW Data */
+
+struct mana_rnic_destroy_adapter_req {
+	struct gdma_req_hdr hdr;
+	mana_handle_t adapter;
+}; /*HW Data */
+
+struct mana_rnic_destroy_adapter_resp {
+	struct gdma_resp_hdr hdr;
+}; /* HW Data */
+
+enum mana_ib_addr_op {
+	ADDR_OP_ADD = 1,
+	ADDR_OP_REMOVE = 2,
+};
+
+enum sgid_entry_type {
+	SGID_TYPE_IPV4 = 1,
+	SGID_TYPE_IPV6 = 2,
+};
+
+struct mana_rnic_config_addr_req {
+	struct gdma_req_hdr hdr;
+	mana_handle_t adapter;
+	enum mana_ib_addr_op op;
+	enum sgid_entry_type sgid_type;
+	u8 ip_addr[16];
+}; /* HW Data */
+
+struct mana_rnic_config_addr_resp {
+	struct gdma_resp_hdr hdr;
+}; /* HW Data */
+
+struct mana_rnic_config_mac_addr_req {
+	struct gdma_req_hdr hdr;
+	mana_handle_t adapter;
+	enum mana_ib_addr_op op;
+	u8 mac_addr[ETH_ALEN];
+	u8 reserved[6];
+}; /* HW Data */
+
+struct mana_rnic_config_mac_addr_resp {
+	struct gdma_resp_hdr hdr;
+}; /* HW Data */
+
+struct mana_rnic_create_cq_req {
+	struct gdma_req_hdr hdr;
+	mana_handle_t adapter;
+	u64 gdma_region;
+	u32 eq_id;
+	u32 doorbell_page;
+}; /* HW Data */
+
+struct mana_rnic_create_cq_resp {
+	struct gdma_resp_hdr hdr;
+	mana_handle_t cq_handle;
+	u32 cq_id;
+	u32 reserved;
+}; /* HW Data */
+
+struct mana_rnic_destroy_cq_req {
+	struct gdma_req_hdr hdr;
+	mana_handle_t adapter;
+	mana_handle_t cq_handle;
+}; /* HW Data */
+
+struct mana_rnic_destroy_cq_resp {
+	struct gdma_resp_hdr hdr;
+}; /* HW Data */
+
 static inline struct gdma_context *mdev_to_gc(struct mana_ib_dev *mdev)
 {
 	return mdev->gdma_dev->gdma_context;
@@ -158,7 +246,16 @@ static inline struct net_device *mana_ib_get_netdev(struct ib_device *ibdev, u32
 	return mc->ports[port - 1];
 }
 
+static inline void copy_in_reverse(u8 *dst, const u8 *src, u32 size)
+{
+	u32 i;
+
+	for (i = 0; i < size; i++)
+		dst[size - 1 - i] = src[i];
+}
+
 int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq);
+void mana_ib_remove_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq);
 
 int mana_ib_create_zero_offset_dma_region(struct mana_ib_dev *dev, struct ib_umem *umem,
 					  mana_handle_t *gdma_region);
@@ -169,6 +266,10 @@ int mana_ib_create_dma_region(struct mana_ib_dev *dev, struct ib_umem *umem,
 int mana_ib_gd_destroy_dma_region(struct mana_ib_dev *dev,
 				  mana_handle_t gdma_region);
 
+int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
+			 struct mana_ib_queue *queue);
+void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue);
+
 struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
 				struct ib_wq_init_attr *init_attr,
 				struct ib_udata *udata);
@@ -231,4 +332,26 @@ int mana_ib_query_gid(struct ib_device *ibdev, u32 port, int index,
 void mana_ib_disassociate_ucontext(struct ib_ucontext *ibcontext);
 
 int mana_ib_gd_query_adapter_caps(struct mana_ib_dev *mdev);
+
+int mana_ib_create_eqs(struct mana_ib_dev *mdev);
+
+void mana_ib_destroy_eqs(struct mana_ib_dev *mdev);
+
+int mana_ib_gd_create_rnic_adapter(struct mana_ib_dev *mdev);
+
+int mana_ib_gd_destroy_rnic_adapter(struct mana_ib_dev *mdev);
+
+int mana_ib_query_pkey(struct ib_device *ibdev, u32 port, u16 index, u16 *pkey);
+
+enum rdma_link_layer mana_ib_get_link_layer(struct ib_device *device, u32 port_num);
+
+int mana_ib_gd_add_gid(const struct ib_gid_attr *attr, void **context);
+
+int mana_ib_gd_del_gid(const struct ib_gid_attr *attr, void **context);
+
+int mana_ib_gd_config_mac(struct mana_ib_dev *mdev, enum mana_ib_addr_op op, u8 *mac);
+
+int mana_ib_gd_create_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq, u32 doorbell);
+
+int mana_ib_gd_destroy_cq(struct mana_ib_dev *mdev, struct mana_ib_cq *cq);
 #endif

drivers/infiniband/hw/mana/mr.c

@@ -135,7 +135,7 @@ struct ib_mr *mana_ib_reg_user_mr(struct ib_pd *ibpd, u64 start, u64 length,
 	}
 
 	ibdev_dbg(ibdev,
-		  "create_dma_region ret %d gdma_region %llx\n", err,
+		  "created dma region for user-mr 0x%llx\n",
 		  dma_region_handle);
 
 	mr_params.pd_handle = pd->pd_handle;

drivers/infiniband/hw/mana/qp.c

@@ -95,11 +95,9 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 	struct mana_ib_qp *qp = container_of(ibqp, struct mana_ib_qp, ibqp);
 	struct mana_ib_dev *mdev =
 		container_of(pd->device, struct mana_ib_dev, ib_dev);
-	struct gdma_context *gc = mdev_to_gc(mdev);
 	struct ib_rwq_ind_table *ind_tbl = attr->rwq_ind_tbl;
 	struct mana_ib_create_qp_rss_resp resp = {};
 	struct mana_ib_create_qp_rss ucmd = {};
-	struct gdma_queue **gdma_cq_allocated;
 	mana_handle_t *mana_ind_table;
 	struct mana_port_context *mpc;
 	unsigned int ind_tbl_size;
@@ -173,13 +171,6 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		goto fail;
 	}
 
-	gdma_cq_allocated = kcalloc(ind_tbl_size, sizeof(*gdma_cq_allocated),
-				    GFP_KERNEL);
-	if (!gdma_cq_allocated) {
-		ret = -ENOMEM;
-		goto fail;
-	}
-
 	qp->port = port;
 
 	for (i = 0; i < ind_tbl_size; i++) {
@@ -192,13 +183,13 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		ibcq = ibwq->cq;
 		cq = container_of(ibcq, struct mana_ib_cq, ibcq);
 
-		wq_spec.gdma_region = wq->gdma_region;
+		wq_spec.gdma_region = wq->queue.gdma_region;
 		wq_spec.queue_size = wq->wq_buf_size;
 
-		cq_spec.gdma_region = cq->gdma_region;
+		cq_spec.gdma_region = cq->queue.gdma_region;
 		cq_spec.queue_size = cq->cqe * COMP_ENTRY_SIZE;
 		cq_spec.modr_ctx_id = 0;
-		eq = &mpc->ac->eqs[cq->comp_vector % gc->max_num_queues];
+		eq = &mpc->ac->eqs[cq->comp_vector];
 		cq_spec.attached_eq = eq->eq->id;
 
 		ret = mana_create_wq_obj(mpc, mpc->port_handle, GDMA_RQ,
@@ -210,18 +201,18 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		}
 
 		/* The GDMA regions are now owned by the WQ object */
-		wq->gdma_region = GDMA_INVALID_DMA_REGION;
-		cq->gdma_region = GDMA_INVALID_DMA_REGION;
+		wq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
+		cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
 
-		wq->id = wq_spec.queue_index;
-		cq->id = cq_spec.queue_index;
+		wq->queue.id = wq_spec.queue_index;
+		cq->queue.id = cq_spec.queue_index;
 
 		ibdev_dbg(&mdev->ib_dev,
-			  "ret %d rx_object 0x%llx wq id %llu cq id %llu\n",
-			  ret, wq->rx_object, wq->id, cq->id);
+			  "rx_object 0x%llx wq id %llu cq id %llu\n",
+			  wq->rx_object, wq->queue.id, cq->queue.id);
 
-		resp.entries[i].cqid = cq->id;
-		resp.entries[i].wqid = wq->id;
+		resp.entries[i].cqid = cq->queue.id;
+		resp.entries[i].wqid = wq->queue.id;
 
 		mana_ind_table[i] = wq->rx_object;
 
@@ -229,8 +220,6 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		ret = mana_ib_install_cq_cb(mdev, cq);
 		if (ret)
 			goto fail;
-
-		gdma_cq_allocated[i] = gc->cq_table[cq->id];
 	}
 	resp.num_entries = i;
 
@@ -250,7 +239,6 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
 		goto fail;
 	}
 
-	kfree(gdma_cq_allocated);
 	kfree(mana_ind_table);
 
 	return 0;
@@ -262,13 +250,10 @@ fail:
 		wq = container_of(ibwq, struct mana_ib_wq, ibwq);
 		cq = container_of(ibcq, struct mana_ib_cq, ibcq);
 
-		gc->cq_table[cq->id] = NULL;
-		kfree(gdma_cq_allocated[i]);
-
+		mana_ib_remove_cq_cb(mdev, cq);
 		mana_destroy_wq_obj(mpc, GDMA_RQ, wq->rx_object);
 	}
 
-	kfree(gdma_cq_allocated);
 	kfree(mana_ind_table);
 
 	return ret;
@@ -287,15 +272,12 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 	struct mana_ib_ucontext *mana_ucontext =
 		rdma_udata_to_drv_context(udata, struct mana_ib_ucontext,
 					  ibucontext);
-	struct gdma_context *gc = mdev_to_gc(mdev);
 	struct mana_ib_create_qp_resp resp = {};
 	struct mana_ib_create_qp ucmd = {};
-	struct gdma_queue *gdma_cq = NULL;
 	struct mana_obj_spec wq_spec = {};
 	struct mana_obj_spec cq_spec = {};
 	struct mana_port_context *mpc;
 	struct net_device *ndev;
-	struct ib_umem *umem;
 	struct mana_eq *eq;
 	int eq_vec;
 	u32 port;
@@ -344,56 +326,39 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 	ibdev_dbg(&mdev->ib_dev, "ucmd sq_buf_addr 0x%llx port %u\n",
 		  ucmd.sq_buf_addr, ucmd.port);
 
-	umem = ib_umem_get(ibpd->device, ucmd.sq_buf_addr, ucmd.sq_buf_size,
-			   IB_ACCESS_LOCAL_WRITE);
-	if (IS_ERR(umem)) {
-		err = PTR_ERR(umem);
-		ibdev_dbg(&mdev->ib_dev,
-			  "Failed to get umem for create qp-raw, err %d\n",
-			  err);
-		goto err_free_vport;
-	}
-	qp->sq_umem = umem;
-
-	err = mana_ib_create_zero_offset_dma_region(mdev, qp->sq_umem,
-						    &qp->sq_gdma_region);
+	err = mana_ib_create_queue(mdev, ucmd.sq_buf_addr, ucmd.sq_buf_size, &qp->raw_sq);
 	if (err) {
 		ibdev_dbg(&mdev->ib_dev,
-			  "Failed to create dma region for create qp-raw, %d\n",
-			  err);
-		goto err_release_umem;
+			  "Failed to create queue for create qp-raw, err %d\n", err);
+		goto err_free_vport;
 	}
 
-	ibdev_dbg(&mdev->ib_dev,
-		  "create_dma_region ret %d gdma_region 0x%llx\n",
-		  err, qp->sq_gdma_region);
-
 	/* Create a WQ on the same port handle used by the Ethernet */
-	wq_spec.gdma_region = qp->sq_gdma_region;
+	wq_spec.gdma_region = qp->raw_sq.gdma_region;
 	wq_spec.queue_size = ucmd.sq_buf_size;
 
-	cq_spec.gdma_region = send_cq->gdma_region;
+	cq_spec.gdma_region = send_cq->queue.gdma_region;
 	cq_spec.queue_size = send_cq->cqe * COMP_ENTRY_SIZE;
 	cq_spec.modr_ctx_id = 0;
-	eq_vec = send_cq->comp_vector % gc->max_num_queues;
+	eq_vec = send_cq->comp_vector;
 	eq = &mpc->ac->eqs[eq_vec];
 	cq_spec.attached_eq = eq->eq->id;
 
 	err = mana_create_wq_obj(mpc, mpc->port_handle, GDMA_SQ, &wq_spec,
-				 &cq_spec, &qp->tx_object);
+				 &cq_spec, &qp->qp_handle);
 	if (err) {
 		ibdev_dbg(&mdev->ib_dev,
 			  "Failed to create wq for create raw-qp, err %d\n",
 			  err);
-		goto err_destroy_dma_region;
+		goto err_destroy_queue;
 	}
 
 	/* The GDMA regions are now owned by the WQ object */
-	qp->sq_gdma_region = GDMA_INVALID_DMA_REGION;
-	send_cq->gdma_region = GDMA_INVALID_DMA_REGION;
+	qp->raw_sq.gdma_region = GDMA_INVALID_DMA_REGION;
+	send_cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
 
-	qp->sq_id = wq_spec.queue_index;
-	send_cq->id = cq_spec.queue_index;
+	qp->raw_sq.id = wq_spec.queue_index;
+	send_cq->queue.id = cq_spec.queue_index;
 
 	/* Create CQ table entry */
 	err = mana_ib_install_cq_cb(mdev, send_cq);
@@ -401,11 +366,11 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 		goto err_destroy_wq_obj;
 
 	ibdev_dbg(&mdev->ib_dev,
-		  "ret %d qp->tx_object 0x%llx sq id %llu cq id %llu\n", err,
-		  qp->tx_object, qp->sq_id, send_cq->id);
+		  "qp->qp_handle 0x%llx sq id %llu cq id %llu\n",
+		  qp->qp_handle, qp->raw_sq.id, send_cq->queue.id);
 
-	resp.sqid = qp->sq_id;
-	resp.cqid = send_cq->id;
+	resp.sqid = qp->raw_sq.id;
+	resp.cqid = send_cq->queue.id;
 	resp.tx_vp_offset = pd->tx_vp_offset;
 
 	err = ib_copy_to_udata(udata, &resp, sizeof(resp));
@@ -413,23 +378,19 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
 		ibdev_dbg(&mdev->ib_dev,
 			  "Failed copy udata for create qp-raw, %d\n",
 			  err);
-		goto err_release_gdma_cq;
+		goto err_remove_cq_cb;
 	}
 
 	return 0;
 
-err_release_gdma_cq:
-	kfree(gdma_cq);
-	gc->cq_table[send_cq->id] = NULL;
+err_remove_cq_cb:
+	mana_ib_remove_cq_cb(mdev, send_cq);
 
 err_destroy_wq_obj:
-	mana_destroy_wq_obj(mpc, GDMA_SQ, qp->tx_object);
+	mana_destroy_wq_obj(mpc, GDMA_SQ, qp->qp_handle);
 
-err_destroy_dma_region:
-	mana_ib_gd_destroy_dma_region(mdev, qp->sq_gdma_region);
-
-err_release_umem:
-	ib_umem_release(umem);
+err_destroy_queue:
+	mana_ib_destroy_queue(mdev, &qp->raw_sq);
 
 err_free_vport:
 	mana_ib_uncfg_vport(mdev, pd, port);
@@ -503,12 +464,9 @@ static int mana_ib_destroy_qp_raw(struct mana_ib_qp *qp, struct ib_udata *udata)
 	mpc = netdev_priv(ndev);
 	pd = container_of(ibpd, struct mana_ib_pd, ibpd);
 
-	mana_destroy_wq_obj(mpc, GDMA_SQ, qp->tx_object);
+	mana_destroy_wq_obj(mpc, GDMA_SQ, qp->qp_handle);
 
-	if (qp->sq_umem) {
-		mana_ib_gd_destroy_dma_region(mdev, qp->sq_gdma_region);
-		ib_umem_release(qp->sq_umem);
-	}
+	mana_ib_destroy_queue(mdev, &qp->raw_sq);
 
 	mana_ib_uncfg_vport(mdev, pd, qp->port);
 

drivers/infiniband/hw/mana/wq.c

@@ -13,7 +13,6 @@ struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
 		container_of(pd->device, struct mana_ib_dev, ib_dev);
 	struct mana_ib_create_wq ucmd = {};
 	struct mana_ib_wq *wq;
-	struct ib_umem *umem;
 	int err;
 
 	if (udata->inlen < sizeof(ucmd))
@@ -32,39 +31,18 @@ struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
 
 	ibdev_dbg(&mdev->ib_dev, "ucmd wq_buf_addr 0x%llx\n", ucmd.wq_buf_addr);
 
-	umem = ib_umem_get(pd->device, ucmd.wq_buf_addr, ucmd.wq_buf_size,
-			   IB_ACCESS_LOCAL_WRITE);
-	if (IS_ERR(umem)) {
-		err = PTR_ERR(umem);
+	err = mana_ib_create_queue(mdev, ucmd.wq_buf_addr, ucmd.wq_buf_size, &wq->queue);
+	if (err) {
 		ibdev_dbg(&mdev->ib_dev,
-			  "Failed to get umem for create wq, err %d\n", err);
+			  "Failed to create queue for create wq, %d\n", err);
 		goto err_free_wq;
 	}
 
-	wq->umem = umem;
 	wq->wqe = init_attr->max_wr;
 	wq->wq_buf_size = ucmd.wq_buf_size;
 	wq->rx_object = INVALID_MANA_HANDLE;
-
-	err = mana_ib_create_zero_offset_dma_region(mdev, wq->umem, &wq->gdma_region);
-	if (err) {
-		ibdev_dbg(&mdev->ib_dev,
-			  "Failed to create dma region for create wq, %d\n",
-			  err);
-		goto err_release_umem;
-	}
-
-	ibdev_dbg(&mdev->ib_dev,
-		  "create_dma_region ret %d gdma_region 0x%llx\n",
-		  err, wq->gdma_region);
-
-	/* WQ ID is returned at wq_create time, doesn't know the value yet */
-
 	return &wq->ibwq;
 
-err_release_umem:
-	ib_umem_release(umem);
-
 err_free_wq:
 	kfree(wq);
 
@@ -86,8 +64,7 @@ int mana_ib_destroy_wq(struct ib_wq *ibwq, struct ib_udata *udata)
 
 	mdev = container_of(ib_dev, struct mana_ib_dev, ib_dev);
 
-	mana_ib_gd_destroy_dma_region(mdev, wq->gdma_region);
-	ib_umem_release(wq->umem);
+	mana_ib_destroy_queue(mdev, &wq->queue);
 
 	kfree(wq);
 

drivers/infiniband/hw/mlx5/main.c

@@ -264,8 +264,7 @@ static struct net_device *mlx5_ib_get_netdev(struct ib_device *device,
 	 */
 	read_lock(&ibdev->port[port_num - 1].roce.netdev_lock);
 	ndev = ibdev->port[port_num - 1].roce.netdev;
-	if (ndev)
-		dev_hold(ndev);
+	dev_hold(ndev);
 	read_unlock(&ibdev->port[port_num - 1].roce.netdev_lock);
 
 out:

drivers/infiniband/hw/mlx5/mem.c

@@ -30,6 +30,7 @@
  * SOFTWARE.
  */
 
+#include <linux/io.h>
 #include <rdma/ib_umem_odp.h>
 #include "mlx5_ib.h"
 #include <linux/jiffies.h>
@@ -108,7 +109,6 @@ static int post_send_nop(struct mlx5_ib_dev *dev, struct ib_qp *ibqp, u64 wr_id,
 	__be32 mmio_wqe[16] = {};
 	unsigned long flags;
 	unsigned int idx;
-	int i;
 
 	if (unlikely(dev->mdev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR))
 		return -EIO;
@@ -148,10 +148,8 @@ static int post_send_nop(struct mlx5_ib_dev *dev, struct ib_qp *ibqp, u64 wr_id,
 	 * we hit doorbell
 	 */
 	wmb();
-	for (i = 0; i < 8; i++)
-		mlx5_write64(&mmio_wqe[i * 2],
-			     bf->bfreg->map + bf->offset + i * 8);
-	io_stop_wc();
+	__iowrite64_copy(bf->bfreg->map + bf->offset, mmio_wqe,
+			 sizeof(mmio_wqe) / 8);
 
 	bf->offset ^= bf->buf_size;
 

drivers/infiniband/hw/mlx5/mlx5_ib.h

@@ -643,9 +643,10 @@ struct mlx5_ib_mkey {
 	unsigned int ndescs;
 	struct wait_queue_head wait;
 	refcount_t usecount;
-	/* User Mkey must hold either a rb_key or a cache_ent. */
+	/* Cacheable user Mkey must hold either a rb_key or a cache_ent. */
 	struct mlx5r_cache_rb_key rb_key;
 	struct mlx5_cache_ent *cache_ent;
+	u8 cacheable : 1;
 };
 
 #define MLX5_IB_MTT_PRESENT (MLX5_IB_MTT_READ | MLX5_IB_MTT_WRITE)

drivers/infiniband/hw/mlx5/mr.c

@@ -1158,6 +1158,7 @@ static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd,
 		if (IS_ERR(mr))
 			return mr;
 		mr->mmkey.rb_key = rb_key;
+		mr->mmkey.cacheable = true;
 		return mr;
 	}
 
@@ -1168,6 +1169,7 @@ static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd,
 	mr->ibmr.pd = pd;
 	mr->umem = umem;
 	mr->page_shift = order_base_2(page_size);
+	mr->mmkey.cacheable = true;
 	set_mr_fields(dev, mr, umem->length, access_flags, iova);
 
 	return mr;
@@ -1570,7 +1572,8 @@ static bool can_use_umr_rereg_access(struct mlx5_ib_dev *dev,
 	unsigned int diffs = current_access_flags ^ target_access_flags;
 
 	if (diffs & ~(IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_WRITE |
-		      IB_ACCESS_REMOTE_READ | IB_ACCESS_RELAXED_ORDERING))
+		      IB_ACCESS_REMOTE_READ | IB_ACCESS_RELAXED_ORDERING |
+		      IB_ACCESS_REMOTE_ATOMIC))
 		return false;
 	return mlx5r_umr_can_reconfig(dev, current_access_flags,
 				      target_access_flags);
@@ -1835,6 +1838,23 @@ end:
 	return ret;
 }
 
+static int mlx5_revoke_mr(struct mlx5_ib_mr *mr)
+{
+	struct mlx5_ib_dev *dev = to_mdev(mr->ibmr.device);
+	struct mlx5_cache_ent *ent = mr->mmkey.cache_ent;
+
+	if (mr->mmkey.cacheable && !mlx5r_umr_revoke_mr(mr) && !cache_ent_find_and_store(dev, mr))
+		return 0;
+
+	if (ent) {
+		spin_lock_irq(&ent->mkeys_queue.lock);
+		ent->in_use--;
+		mr->mmkey.cache_ent = NULL;
+		spin_unlock_irq(&ent->mkeys_queue.lock);
+	}
+	return destroy_mkey(dev, mr);
+}
+
 int mlx5_ib_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
 {
 	struct mlx5_ib_mr *mr = to_mmr(ibmr);
@@ -1880,16 +1900,9 @@ int mlx5_ib_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
 	}
 
 	/* Stop DMA */
-	if (mr->umem && mlx5r_umr_can_load_pas(dev, mr->umem->length))
-		if (mlx5r_umr_revoke_mr(mr) ||
-		    cache_ent_find_and_store(dev, mr))
-			mr->mmkey.cache_ent = NULL;
-
-	if (!mr->mmkey.cache_ent) {
-		rc = destroy_mkey(to_mdev(mr->ibmr.device), mr);
-		if (rc)
-			return rc;
-	}
+	rc = mlx5_revoke_mr(mr);
+	if (rc)
+		return rc;
 
 	if (mr->umem) {
 		bool is_odp = is_odp_mr(mr);

drivers/infiniband/hw/mlx5/qp.c

@@ -3097,7 +3097,6 @@ static int create_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
 	switch (qp->type) {
 	case MLX5_IB_QPT_DCT:
 		err = create_dct(dev, pd, qp, params);
-		rdma_restrack_no_track(&qp->ibqp.res);
 		break;
 	case MLX5_IB_QPT_DCI:
 		err = create_dci(dev, pd, qp, params);
@@ -3109,9 +3108,9 @@ static int create_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
 		err = mlx5_ib_create_gsi(pd, qp, params->attr);
 		break;
 	case MLX5_IB_QPT_HW_GSI:
-	case MLX5_IB_QPT_REG_UMR:
 		rdma_restrack_no_track(&qp->ibqp.res);
 		fallthrough;
+	case MLX5_IB_QPT_REG_UMR:
 	default:
 		if (params->udata)
 			err = create_user_qp(dev, pd, qp, params);

drivers/infiniband/hw/mlx5/restrack.c

@@ -156,6 +156,34 @@ static int fill_res_cq_entry_raw(struct sk_buff *msg, struct ib_cq *ibcq)
 	return fill_res_raw(msg, dev, MLX5_SGMT_TYPE_PRM_QUERY_CQ, cq->mcq.cqn);
 }
 
+static int fill_res_qp_entry(struct sk_buff *msg, struct ib_qp *ibqp)
+{
+	struct mlx5_ib_qp *qp = to_mqp(ibqp);
+	int ret;
+
+	if (qp->type < IB_QPT_DRIVER)
+		return 0;
+
+	switch (qp->type) {
+	case MLX5_IB_QPT_REG_UMR:
+		ret = nla_put_string(msg, RDMA_NLDEV_ATTR_RES_SUBTYPE,
+				     "REG_UMR");
+		break;
+	case MLX5_IB_QPT_DCT:
+		ret = nla_put_string(msg, RDMA_NLDEV_ATTR_RES_SUBTYPE, "DCT");
+		break;
+	case MLX5_IB_QPT_DCI:
+		ret = nla_put_string(msg, RDMA_NLDEV_ATTR_RES_SUBTYPE, "DCI");
+		break;
+	default:
+		return 0;
+	}
+	if (ret)
+		return ret;
+
+	return nla_put_u8(msg, RDMA_NLDEV_ATTR_RES_TYPE, IB_QPT_DRIVER);
+}
+
 static int fill_res_qp_entry_raw(struct sk_buff *msg, struct ib_qp *ibqp)
 {
 	struct mlx5_ib_dev *dev = to_mdev(ibqp->device);
@@ -168,6 +196,7 @@ static const struct ib_device_ops restrack_ops = {
 	.fill_res_cq_entry_raw = fill_res_cq_entry_raw,
 	.fill_res_mr_entry = fill_res_mr_entry,
 	.fill_res_mr_entry_raw = fill_res_mr_entry_raw,
+	.fill_res_qp_entry = fill_res_qp_entry,
 	.fill_res_qp_entry_raw = fill_res_qp_entry_raw,
 	.fill_stat_mr_entry = fill_stat_mr_entry,
 };

drivers/infiniband/sw/rxe/rxe_comp.c

@@ -122,25 +122,16 @@ void retransmit_timer(struct timer_list *t)
 	spin_lock_irqsave(&qp->state_lock, flags);
 	if (qp->valid) {
 		qp->comp.timeout = 1;
-		rxe_sched_task(&qp->comp.task);
+		rxe_sched_task(&qp->send_task);
 	}
 	spin_unlock_irqrestore(&qp->state_lock, flags);
 }
 
 void rxe_comp_queue_pkt(struct rxe_qp *qp, struct sk_buff *skb)
 {
-	int must_sched;
-
+	rxe_counter_inc(SKB_TO_PKT(skb)->rxe, RXE_CNT_SENDER_SCHED);
 	skb_queue_tail(&qp->resp_pkts, skb);
-
-	must_sched = skb_queue_len(&qp->resp_pkts) > 1;
-	if (must_sched != 0)
-		rxe_counter_inc(SKB_TO_PKT(skb)->rxe, RXE_CNT_COMPLETER_SCHED);
-
-	if (must_sched)
-		rxe_sched_task(&qp->comp.task);
-	else
-		rxe_run_task(&qp->comp.task);
+	rxe_sched_task(&qp->send_task);
 }
 
 static inline enum comp_state get_wqe(struct rxe_qp *qp,
@@ -325,7 +316,7 @@ static inline enum comp_state check_ack(struct rxe_qp *qp,
 					qp->comp.psn = pkt->psn;
 					if (qp->req.wait_psn) {
 						qp->req.wait_psn = 0;
-						rxe_sched_task(&qp->req.task);
+						qp->req.again = 1;
 					}
 				}
 				return COMPST_ERROR_RETRY;
@@ -476,7 +467,7 @@ static void do_complete(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
 	 */
 	if (qp->req.wait_fence) {
 		qp->req.wait_fence = 0;
-		rxe_sched_task(&qp->req.task);
+		qp->req.again = 1;
 	}
 }
 
@@ -515,7 +506,7 @@ static inline enum comp_state complete_ack(struct rxe_qp *qp,
 		if (qp->req.need_rd_atomic) {
 			qp->comp.timeout_retry = 0;
 			qp->req.need_rd_atomic = 0;
-			rxe_sched_task(&qp->req.task);
+			qp->req.again = 1;
 		}
 	}
 
@@ -541,7 +532,7 @@ static inline enum comp_state complete_wqe(struct rxe_qp *qp,
 
 		if (qp->req.wait_psn) {
 			qp->req.wait_psn = 0;
-			rxe_sched_task(&qp->req.task);
+			qp->req.again = 1;
 		}
 	}
 
@@ -654,6 +645,8 @@ int rxe_completer(struct rxe_qp *qp)
 	int ret;
 	unsigned long flags;
 
+	qp->req.again = 0;
+
 	spin_lock_irqsave(&qp->state_lock, flags);
 	if (!qp->valid || qp_state(qp) == IB_QPS_ERR ||
 			  qp_state(qp) == IB_QPS_RESET) {
@@ -737,7 +730,7 @@ int rxe_completer(struct rxe_qp *qp)
 
 			if (qp->req.wait_psn) {
 				qp->req.wait_psn = 0;
-				rxe_sched_task(&qp->req.task);
+				qp->req.again = 1;
 			}
 
 			state = COMPST_DONE;
@@ -792,7 +785,7 @@ int rxe_completer(struct rxe_qp *qp)
 							RXE_CNT_COMP_RETRY);
 					qp->req.need_retry = 1;
 					qp->comp.started_retry = 1;
-					rxe_sched_task(&qp->req.task);
+					qp->req.again = 1;
 				}
 				goto done;
 
@@ -843,8 +836,9 @@ done:
 	ret = 0;
 	goto out;
 exit:
-	ret = -EAGAIN;
+	ret = (qp->req.again) ? 0 : -EAGAIN;
 out:
+	qp->req.again = 0;
 	if (pkt)
 		free_pkt(pkt);
 	return ret;

drivers/infiniband/sw/rxe/rxe_hw_counters.c

@@ -14,7 +14,7 @@ static const struct rdma_stat_desc rxe_counter_descs[] = {
 	[RXE_CNT_RCV_RNR].name             =  "rcvd_rnr_err",
 	[RXE_CNT_SND_RNR].name             =  "send_rnr_err",
 	[RXE_CNT_RCV_SEQ_ERR].name         =  "rcvd_seq_err",
-	[RXE_CNT_COMPLETER_SCHED].name     =  "ack_deferred",
+	[RXE_CNT_SENDER_SCHED].name        =  "ack_deferred",
 	[RXE_CNT_RETRY_EXCEEDED].name      =  "retry_exceeded_err",
 	[RXE_CNT_RNR_RETRY_EXCEEDED].name  =  "retry_rnr_exceeded_err",
 	[RXE_CNT_COMP_RETRY].name          =  "completer_retry_err",

drivers/infiniband/sw/rxe/rxe_hw_counters.h

@@ -18,7 +18,7 @@ enum rxe_counters {
 	RXE_CNT_RCV_RNR,
 	RXE_CNT_SND_RNR,
 	RXE_CNT_RCV_SEQ_ERR,
-	RXE_CNT_COMPLETER_SCHED,
+	RXE_CNT_SENDER_SCHED,
 	RXE_CNT_RETRY_EXCEEDED,
 	RXE_CNT_RNR_RETRY_EXCEEDED,
 	RXE_CNT_COMP_RETRY,

drivers/infiniband/sw/rxe/rxe_loc.h

@@ -164,7 +164,8 @@ void rxe_dealloc(struct ib_device *ib_dev);
 
 int rxe_completer(struct rxe_qp *qp);
 int rxe_requester(struct rxe_qp *qp);
-int rxe_responder(struct rxe_qp *qp);
+int rxe_sender(struct rxe_qp *qp);
+int rxe_receiver(struct rxe_qp *qp);
 
 /* rxe_icrc.c */
 int rxe_icrc_init(struct rxe_dev *rxe);

drivers/infiniband/sw/rxe/rxe_net.c

@@ -345,46 +345,52 @@ int rxe_prepare(struct rxe_av *av, struct rxe_pkt_info *pkt,
 
 static void rxe_skb_tx_dtor(struct sk_buff *skb)
 {
-	struct sock *sk = skb->sk;
-	struct rxe_qp *qp = sk->sk_user_data;
-	int skb_out = atomic_dec_return(&qp->skb_out);
+	struct net_device *ndev = skb->dev;
+	struct rxe_dev *rxe;
+	unsigned int qp_index;
+	struct rxe_qp *qp;
+	int skb_out;
 
-	if (unlikely(qp->need_req_skb &&
-		     skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW))
-		rxe_sched_task(&qp->req.task);
+	rxe = rxe_get_dev_from_net(ndev);
+	if (!rxe && is_vlan_dev(ndev))
+		rxe = rxe_get_dev_from_net(vlan_dev_real_dev(ndev));
+	if (WARN_ON(!rxe))
+		return;
+
+	qp_index = (int)(uintptr_t)skb->sk->sk_user_data;
+	if (!qp_index)
+		return;
+
+	qp = rxe_pool_get_index(&rxe->qp_pool, qp_index);
+	if (!qp)
+		goto put_dev;
+
+	skb_out = atomic_dec_return(&qp->skb_out);
+	if (qp->need_req_skb && skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW)
+		rxe_sched_task(&qp->send_task);
 
 	rxe_put(qp);
+put_dev:
+	ib_device_put(&rxe->ib_dev);
+	sock_put(skb->sk);
 }
 
 static int rxe_send(struct sk_buff *skb, struct rxe_pkt_info *pkt)
 {
 	int err;
+	struct sock *sk = pkt->qp->sk->sk;
 
+	sock_hold(sk);
+	skb->sk = sk;
 	skb->destructor = rxe_skb_tx_dtor;
-	skb->sk = pkt->qp->sk->sk;
-
-	rxe_get(pkt->qp);
 	atomic_inc(&pkt->qp->skb_out);
 
-	if (skb->protocol == htons(ETH_P_IP)) {
+	if (skb->protocol == htons(ETH_P_IP))
 		err = ip_local_out(dev_net(skb_dst(skb)->dev), skb->sk, skb);
-	} else if (skb->protocol == htons(ETH_P_IPV6)) {
+	else
 		err = ip6_local_out(dev_net(skb_dst(skb)->dev), skb->sk, skb);
-	} else {
-		rxe_dbg_qp(pkt->qp, "Unknown layer 3 protocol: %d\n",
-				skb->protocol);
-		atomic_dec(&pkt->qp->skb_out);
-		rxe_put(pkt->qp);
-		kfree_skb(skb);
-		return -EINVAL;
-	}
 
-	if (unlikely(net_xmit_eval(err))) {
-		rxe_dbg_qp(pkt->qp, "error sending packet: %d\n", err);
-		return -EAGAIN;
-	}
-
-	return 0;
+	return err;
 }
 
 /* fix up a send packet to match the packets
@@ -392,8 +398,15 @@ static int rxe_send(struct sk_buff *skb, struct rxe_pkt_info *pkt)
  */
 static int rxe_loopback(struct sk_buff *skb, struct rxe_pkt_info *pkt)
 {
+	struct sock *sk = pkt->qp->sk->sk;
+
 	memcpy(SKB_TO_PKT(skb), pkt, sizeof(*pkt));
 
+	sock_hold(sk);
+	skb->sk = sk;
+	skb->destructor = rxe_skb_tx_dtor;
+	atomic_inc(&pkt->qp->skb_out);
+
 	if (skb->protocol == htons(ETH_P_IP))
 		skb_pull(skb, sizeof(struct iphdr));
 	else
@@ -440,12 +453,6 @@ int rxe_xmit_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt,
 		return err;
 	}
 
-	if ((qp_type(qp) != IB_QPT_RC) &&
-	    (pkt->mask & RXE_END_MASK)) {
-		pkt->wqe->state = wqe_state_done;
-		rxe_sched_task(&qp->comp.task);
-	}
-
 	rxe_counter_inc(rxe, RXE_CNT_SENT_PKTS);
 	goto done;
 

drivers/infiniband/sw/rxe/rxe_pool.c

@@ -119,7 +119,7 @@ void rxe_pool_cleanup(struct rxe_pool *pool)
 int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem,
 				bool sleepable)
 {
-	int err;
+	int err = -EINVAL;
 	gfp_t gfp_flags;
 
 	if (atomic_inc_return(&pool->num_elem) > pool->max_elem)
@@ -147,7 +147,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem,
 
 err_cnt:
 	atomic_dec(&pool->num_elem);
-	return -EINVAL;
+	return err;
 }
 
 void *rxe_pool_get_index(struct rxe_pool *pool, u32 index)

drivers/infiniband/sw/rxe/rxe_qp.c

@@ -244,7 +244,7 @@ static int rxe_qp_init_req(struct rxe_dev *rxe, struct rxe_qp *qp,
 	err = sock_create_kern(&init_net, AF_INET, SOCK_DGRAM, 0, &qp->sk);
 	if (err < 0)
 		return err;
-	qp->sk->sk->sk_user_data = qp;
+	qp->sk->sk->sk_user_data = (void *)(uintptr_t)qp->elem.index;
 
 	/* pick a source UDP port number for this QP based on
 	 * the source QPN. this spreads traffic for different QPs
@@ -265,8 +265,7 @@ static int rxe_qp_init_req(struct rxe_dev *rxe, struct rxe_qp *qp,
 	qp->req.opcode		= -1;
 	qp->comp.opcode		= -1;
 
-	rxe_init_task(&qp->req.task, qp, rxe_requester);
-	rxe_init_task(&qp->comp.task, qp, rxe_completer);
+	rxe_init_task(&qp->send_task, qp, rxe_sender);
 
 	qp->qp_timeout_jiffies = 0; /* Can't be set for UD/UC in modify_qp */
 	if (init->qp_type == IB_QPT_RC) {
@@ -337,7 +336,7 @@ static int rxe_qp_init_resp(struct rxe_dev *rxe, struct rxe_qp *qp,
 			return err;
 	}
 
-	rxe_init_task(&qp->resp.task, qp, rxe_responder);
+	rxe_init_task(&qp->recv_task, qp, rxe_receiver);
 
 	qp->resp.opcode		= OPCODE_NONE;
 	qp->resp.msn		= 0;
@@ -514,14 +513,12 @@ err1:
 static void rxe_qp_reset(struct rxe_qp *qp)
 {
 	/* stop tasks from running */
-	rxe_disable_task(&qp->resp.task);
-	rxe_disable_task(&qp->comp.task);
-	rxe_disable_task(&qp->req.task);
+	rxe_disable_task(&qp->recv_task);
+	rxe_disable_task(&qp->send_task);
 
 	/* drain work and packet queuesc */
-	rxe_requester(qp);
-	rxe_completer(qp);
-	rxe_responder(qp);
+	rxe_sender(qp);
+	rxe_receiver(qp);
 
 	if (qp->rq.queue)
 		rxe_queue_reset(qp->rq.queue);
@@ -548,9 +545,8 @@ static void rxe_qp_reset(struct rxe_qp *qp)
 	cleanup_rd_atomic_resources(qp);
 
 	/* reenable tasks */
-	rxe_enable_task(&qp->resp.task);
-	rxe_enable_task(&qp->comp.task);
-	rxe_enable_task(&qp->req.task);
+	rxe_enable_task(&qp->recv_task);
+	rxe_enable_task(&qp->send_task);
 }
 
 /* move the qp to the error state */
@@ -562,9 +558,8 @@ void rxe_qp_error(struct rxe_qp *qp)
 	qp->attr.qp_state = IB_QPS_ERR;
 
 	/* drain work and packet queues */
-	rxe_sched_task(&qp->resp.task);
-	rxe_sched_task(&qp->comp.task);
-	rxe_sched_task(&qp->req.task);
+	rxe_sched_task(&qp->recv_task);
+	rxe_sched_task(&qp->send_task);
 	spin_unlock_irqrestore(&qp->state_lock, flags);
 }
 
@@ -575,8 +570,7 @@ static void rxe_qp_sqd(struct rxe_qp *qp, struct ib_qp_attr *attr,
 
 	spin_lock_irqsave(&qp->state_lock, flags);
 	qp->attr.sq_draining = 1;
-	rxe_sched_task(&qp->comp.task);
-	rxe_sched_task(&qp->req.task);
+	rxe_sched_task(&qp->send_task);
 	spin_unlock_irqrestore(&qp->state_lock, flags);
 }
 
@@ -821,19 +815,15 @@ static void rxe_qp_do_cleanup(struct work_struct *work)
 		del_timer_sync(&qp->rnr_nak_timer);
 	}
 
-	if (qp->resp.task.func)
-		rxe_cleanup_task(&qp->resp.task);
+	if (qp->recv_task.func)
+		rxe_cleanup_task(&qp->recv_task);
 
-	if (qp->req.task.func)
-		rxe_cleanup_task(&qp->req.task);
-
-	if (qp->comp.task.func)
-		rxe_cleanup_task(&qp->comp.task);
+	if (qp->send_task.func)
+		rxe_cleanup_task(&qp->send_task);
 
 	/* flush out any receive wr's or pending requests */
-	rxe_requester(qp);
-	rxe_completer(qp);
-	rxe_responder(qp);
+	rxe_sender(qp);
+	rxe_receiver(qp);
 
 	if (qp->sq.queue)
 		rxe_queue_cleanup(qp->sq.queue);

drivers/infiniband/sw/rxe/rxe_req.c

@@ -108,7 +108,7 @@ void rnr_nak_timer(struct timer_list *t)
 		/* request a send queue retry */
 		qp->req.need_retry = 1;
 		qp->req.wait_for_rnr_timer = 0;
-		rxe_sched_task(&qp->req.task);
+		rxe_sched_task(&qp->send_task);
 	}
 	spin_unlock_irqrestore(&qp->state_lock, flags);
 }
@@ -545,6 +545,8 @@ static void update_wqe_state(struct rxe_qp *qp,
 	if (pkt->mask & RXE_END_MASK) {
 		if (qp_type(qp) == IB_QPT_RC)
 			wqe->state = wqe_state_pending;
+		else
+			wqe->state = wqe_state_done;
 	} else {
 		wqe->state = wqe_state_processing;
 	}
@@ -573,30 +575,6 @@ static void update_wqe_psn(struct rxe_qp *qp,
 		qp->req.psn = (qp->req.psn + 1) & BTH_PSN_MASK;
 }
 
-static void save_state(struct rxe_send_wqe *wqe,
-		       struct rxe_qp *qp,
-		       struct rxe_send_wqe *rollback_wqe,
-		       u32 *rollback_psn)
-{
-	rollback_wqe->state = wqe->state;
-	rollback_wqe->first_psn = wqe->first_psn;
-	rollback_wqe->last_psn = wqe->last_psn;
-	rollback_wqe->dma = wqe->dma;
-	*rollback_psn = qp->req.psn;
-}
-
-static void rollback_state(struct rxe_send_wqe *wqe,
-			   struct rxe_qp *qp,
-			   struct rxe_send_wqe *rollback_wqe,
-			   u32 rollback_psn)
-{
-	wqe->state = rollback_wqe->state;
-	wqe->first_psn = rollback_wqe->first_psn;
-	wqe->last_psn = rollback_wqe->last_psn;
-	wqe->dma = rollback_wqe->dma;
-	qp->req.psn = rollback_psn;
-}
-
 static void update_state(struct rxe_qp *qp, struct rxe_pkt_info *pkt)
 {
 	qp->req.opcode = pkt->opcode;
@@ -655,12 +633,6 @@ static int rxe_do_local_ops(struct rxe_qp *qp, struct rxe_send_wqe *wqe)
 	wqe->status = IB_WC_SUCCESS;
 	qp->req.wqe_index = queue_next_index(qp->sq.queue, qp->req.wqe_index);
 
-	/* There is no ack coming for local work requests
-	 * which can lead to a deadlock. So go ahead and complete
-	 * it now.
-	 */
-	rxe_sched_task(&qp->comp.task);
-
 	return 0;
 }
 
@@ -676,8 +648,6 @@ int rxe_requester(struct rxe_qp *qp)
 	int opcode;
 	int err;
 	int ret;
-	struct rxe_send_wqe rollback_wqe;
-	u32 rollback_psn;
 	struct rxe_queue *q = qp->sq.queue;
 	struct rxe_ah *ah;
 	struct rxe_av *av;
@@ -786,7 +756,6 @@ int rxe_requester(struct rxe_qp *qp)
 						       qp->req.wqe_index);
 			wqe->state = wqe_state_done;
 			wqe->status = IB_WC_SUCCESS;
-			rxe_sched_task(&qp->comp.task);
 			goto done;
 		}
 		payload = mtu;
@@ -799,9 +768,6 @@ int rxe_requester(struct rxe_qp *qp)
 	pkt.mask = rxe_opcode[opcode].mask;
 	pkt.wqe = wqe;
 
-	/* save wqe state before we build and send packet */
-	save_state(wqe, qp, &rollback_wqe, &rollback_psn);
-
 	av = rxe_get_av(&pkt, &ah);
 	if (unlikely(!av)) {
 		rxe_dbg_qp(qp, "Failed no address vector\n");
@@ -834,31 +800,14 @@ int rxe_requester(struct rxe_qp *qp)
 	if (ah)
 		rxe_put(ah);
 
-	/* update wqe state as though we had sent it */
-	update_wqe_state(qp, wqe, &pkt);
-	update_wqe_psn(qp, wqe, &pkt, payload);
-
 	err = rxe_xmit_packet(qp, &pkt, skb);
 	if (err) {
-		if (err != -EAGAIN) {
-			wqe->status = IB_WC_LOC_QP_OP_ERR;
-			goto err;
-		}
-
-		/* the packet was dropped so reset wqe to the state
-		 * before we sent it so we can try to resend
-		 */
-		rollback_state(wqe, qp, &rollback_wqe, rollback_psn);
-
-		/* force a delay until the dropped packet is freed and
-		 * the send queue is drained below the low water mark
-		 */
-		qp->need_req_skb = 1;
-
-		rxe_sched_task(&qp->req.task);
-		goto exit;
+		wqe->status = IB_WC_LOC_QP_OP_ERR;
+		goto err;
 	}
 
+	update_wqe_state(qp, wqe, &pkt);
+	update_wqe_psn(qp, wqe, &pkt, payload);
 	update_state(qp, &pkt);
 
 	/* A non-zero return value will cause rxe_do_task to
@@ -878,3 +827,20 @@ exit:
 out:
 	return ret;
 }
+
+int rxe_sender(struct rxe_qp *qp)
+{
+	int req_ret;
+	int comp_ret;
+
+	/* process the send queue */
+	req_ret = rxe_requester(qp);
+
+	/* process the response queue */
+	comp_ret = rxe_completer(qp);
+
+	/* exit the task loop if both requester and completer
+	 * are ready
+	 */
+	return (req_ret && comp_ret) ? -EAGAIN : 0;
+}

drivers/infiniband/sw/rxe/rxe_resp.c

@@ -49,18 +49,8 @@ static char *resp_state_name[] = {
 /* rxe_recv calls here to add a request packet to the input queue */
 void rxe_resp_queue_pkt(struct rxe_qp *qp, struct sk_buff *skb)
 {
-	int must_sched;
-	struct rxe_pkt_info *pkt = SKB_TO_PKT(skb);
-
 	skb_queue_tail(&qp->req_pkts, skb);
-
-	must_sched = (pkt->opcode == IB_OPCODE_RC_RDMA_READ_REQUEST) ||
-			(skb_queue_len(&qp->req_pkts) > 1);
-
-	if (must_sched)
-		rxe_sched_task(&qp->resp.task);
-	else
-		rxe_run_task(&qp->resp.task);
+	rxe_sched_task(&qp->recv_task);
 }
 
 static inline enum resp_states get_req(struct rxe_qp *qp,
@@ -1485,7 +1475,7 @@ static void flush_recv_queue(struct rxe_qp *qp, bool notify)
 	qp->resp.wqe = NULL;
 }
 
-int rxe_responder(struct rxe_qp *qp)
+int rxe_receiver(struct rxe_qp *qp)
 {
 	struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
 	enum resp_states state;

drivers/infiniband/sw/rxe/rxe_verbs.c

@@ -888,6 +888,7 @@ static int rxe_post_send_kernel(struct rxe_qp *qp,
 {
 	int err = 0;
 	unsigned long flags;
+	int good = 0;
 
 	spin_lock_irqsave(&qp->sq.sq_lock, flags);
 	while (ibwr) {
@@ -895,18 +896,16 @@ static int rxe_post_send_kernel(struct rxe_qp *qp,
 		if (err) {
 			*bad_wr = ibwr;
 			break;
+		} else {
+			good++;
 		}
 		ibwr = ibwr->next;
 	}
 	spin_unlock_irqrestore(&qp->sq.sq_lock, flags);
 
-	if (!err)
-		rxe_sched_task(&qp->req.task);
-
-	spin_lock_irqsave(&qp->state_lock, flags);
-	if (qp_state(qp) == IB_QPS_ERR)
-		rxe_sched_task(&qp->comp.task);
-	spin_unlock_irqrestore(&qp->state_lock, flags);
+	/* kickoff processing of any posted wqes */
+	if (good)
+		rxe_sched_task(&qp->send_task);
 
 	return err;
 }
@@ -936,7 +935,7 @@ static int rxe_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr,
 
 	if (qp->is_user) {
 		/* Utilize process context to do protocol processing */
-		rxe_run_task(&qp->req.task);
+		rxe_sched_task(&qp->send_task);
 	} else {
 		err = rxe_post_send_kernel(qp, wr, bad_wr);
 		if (err)
@@ -1046,7 +1045,7 @@ static int rxe_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr,
 
 	spin_lock_irqsave(&qp->state_lock, flags);
 	if (qp_state(qp) == IB_QPS_ERR)
-		rxe_sched_task(&qp->resp.task);
+		rxe_sched_task(&qp->recv_task);
 	spin_unlock_irqrestore(&qp->state_lock, flags);
 
 	return err;

drivers/infiniband/sw/rxe/rxe_verbs.h

@@ -113,7 +113,7 @@ struct rxe_req_info {
 	int			need_retry;
 	int			wait_for_rnr_timer;
 	int			noack_pkts;
-	struct rxe_task		task;
+	int			again;
 };
 
 struct rxe_comp_info {
@@ -124,7 +124,6 @@ struct rxe_comp_info {
 	int			started_retry;
 	u32			retry_cnt;
 	u32			rnr_retry;
-	struct rxe_task		task;
 };
 
 enum rdatm_res_state {
@@ -196,7 +195,6 @@ struct rxe_resp_info {
 	unsigned int		res_head;
 	unsigned int		res_tail;
 	struct resp_res		*res;
-	struct rxe_task		task;
 };
 
 struct rxe_qp {
@@ -229,6 +227,9 @@ struct rxe_qp {
 	struct sk_buff_head	req_pkts;
 	struct sk_buff_head	resp_pkts;
 
+	struct rxe_task		send_task;
+	struct rxe_task		recv_task;
+
 	struct rxe_req_info	req;
 	struct rxe_comp_info	comp;
 	struct rxe_resp_info	resp;

drivers/infiniband/ulp/ipoib/ipoib_main.c

@@ -329,8 +329,7 @@ static struct net_device *ipoib_get_master_net_dev(struct net_device *dev)
 
 	rcu_read_lock();
 	master = netdev_master_upper_dev_get_rcu(dev);
-	if (master)
-		dev_hold(master);
+	dev_hold(master);
 	rcu_read_unlock();
 
 	if (master)

drivers/infiniband/ulp/ipoib/ipoib_vlan.c

@@ -184,8 +184,12 @@ int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey)
 
 	ppriv = ipoib_priv(pdev);
 
-	snprintf(intf_name, sizeof(intf_name), "%s.%04x",
-		 ppriv->dev->name, pkey);
+	/* If you increase IFNAMSIZ, update snprintf below
+	 * to allow longer names.
+	 */
+	BUILD_BUG_ON(IFNAMSIZ != 16);
+	snprintf(intf_name, sizeof(intf_name), "%.10s.%04x", ppriv->dev->name,
+		 pkey);
 
 	ndev = ipoib_intf_alloc(ppriv->ca, ppriv->port, intf_name);
 	if (IS_ERR(ndev)) {

drivers/net/ethernet/hisilicon/hns3/hns3_enet.c

@@ -2068,8 +2068,6 @@ static void hns3_tx_push_bd(struct hns3_enet_ring *ring, int num)
 	__iowrite64_copy(ring->tqp->mem_base, desc,
 			 (sizeof(struct hns3_desc) * HNS3_MAX_PUSH_BD_NUM) /
 			 HNS3_BYTES_PER_64BIT);
-
-	io_stop_wc();
 }
 
 static void hns3_tx_mem_doorbell(struct hns3_enet_ring *ring)
@@ -2088,8 +2086,6 @@ static void hns3_tx_mem_doorbell(struct hns3_enet_ring *ring)
 	u64_stats_update_begin(&ring->syncp);
 	ring->stats.tx_mem_doorbell += ring->pending_buf;
 	u64_stats_update_end(&ring->syncp);
-
-	io_stop_wc();
 }
 
 static void hns3_tx_doorbell(struct hns3_enet_ring *ring, int num,

include/linux/io.h

@@ -16,9 +16,15 @@
 struct device;
 struct resource;
 
-__visible void __iowrite32_copy(void __iomem *to, const void *from, size_t count);
+#ifndef __iowrite32_copy
+void __iowrite32_copy(void __iomem *to, const void *from, size_t count);
+#endif
+
 void __ioread32_copy(void *to, const void __iomem *from, size_t count);
+
+#ifndef __iowrite64_copy
 void __iowrite64_copy(void __iomem *to, const void *from, size_t count);
+#endif
 
 #ifdef CONFIG_MMU
 int ioremap_page_range(unsigned long addr, unsigned long end,

include/rdma/restrack.h

@@ -14,6 +14,9 @@
 #include <uapi/rdma/rdma_netlink.h>
 #include <linux/xarray.h>
 
+/* Mark entry as containing driver specific details, it is used to provide QP subtype for now */
+#define RESTRACK_DD XA_MARK_1
+
 struct ib_device;
 struct sk_buff;
 
@@ -116,8 +119,8 @@ struct rdma_restrack_entry {
 	u32 id;
 };
 
-int rdma_restrack_count(struct ib_device *dev,
-			enum rdma_restrack_type type);
+int rdma_restrack_count(struct ib_device *dev, enum rdma_restrack_type type,
+			bool show_details);
 /**
  * rdma_is_kernel_res() - check the owner of resource
  * @res:  resource entry

include/uapi/rdma/efa-abi.h

@@ -85,11 +85,17 @@ enum {
 	EFA_QP_DRIVER_TYPE_SRD = 0,
 };
 
+enum {
+	EFA_CREATE_QP_WITH_UNSOLICITED_WRITE_RECV = 1 << 0,
+};
+
 struct efa_ibv_create_qp {
 	__u32 comp_mask;
 	__u32 rq_ring_size; /* bytes */
 	__u32 sq_ring_size; /* bytes */
 	__u32 driver_qp_type;
+	__u16 flags;
+	__u8 reserved_90[6];
 };
 
 struct efa_ibv_create_qp_resp {
@@ -123,6 +129,7 @@ enum {
 	EFA_QUERY_DEVICE_CAPS_CQ_WITH_SGID     = 1 << 3,
 	EFA_QUERY_DEVICE_CAPS_DATA_POLLING_128 = 1 << 4,
 	EFA_QUERY_DEVICE_CAPS_RDMA_WRITE = 1 << 5,
+	EFA_QUERY_DEVICE_CAPS_UNSOLICITED_WRITE_RECV = 1 << 6,
 };
 
 struct efa_ibv_ex_query_device_resp {

include/uapi/rdma/hns-abi.h

@@ -109,6 +109,12 @@ struct hns_roce_ib_create_qp_resp {
 	__aligned_u64 dwqe_mmap_key;
 };
 
+struct hns_roce_ib_modify_qp_resp {
+	__u8	tc_mode;
+	__u8	priority;
+	__u8	reserved[6];
+};
+
 enum {
 	HNS_ROCE_EXSGE_FLAGS = 1 << 0,
 	HNS_ROCE_RQ_INLINE_FLAGS = 1 << 1,
@@ -143,7 +149,8 @@ struct hns_roce_ib_alloc_pd_resp {
 
 struct hns_roce_ib_create_ah_resp {
 	__u8 dmac[6];
-	__u8 reserved[2];
+	__u8 priority;
+	__u8 tc_mode;
 };
 
 #endif /* HNS_ABI_USER_H */

include/uapi/rdma/mana-abi.h

@@ -16,8 +16,20 @@
 
 #define MANA_IB_UVERBS_ABI_VERSION 1
 
+enum mana_ib_create_cq_flags {
+	MANA_IB_CREATE_RNIC_CQ	= 1 << 0,
+};
+
 struct mana_ib_create_cq {
 	__aligned_u64 buf_addr;
+	__u16	flags;
+	__u16	reserved0;
+	__u32	reserved1;
+};
+
+struct mana_ib_create_cq_resp {
+	__u32 cqid;
+	__u32 reserved;
 };
 
 struct mana_ib_create_qp {

include/uapi/rdma/rdma_netlink.h

@@ -558,6 +558,12 @@ enum rdma_nldev_attr {
 
 	RDMA_NLDEV_SYS_ATTR_PRIVILEGED_QKEY_MODE, /* u8 */
 
+	RDMA_NLDEV_ATTR_DRIVER_DETAILS,		/* u8 */
+	/*
+	 * QP subtype string, used for driver QPs
+	 */
+	RDMA_NLDEV_ATTR_RES_SUBTYPE,		/* string */
+
 	/*
 	 * Always the end
 	 */

lib/iomap_copy.c

@@ -16,9 +16,8 @@
  * time.  Order of access is not guaranteed, nor is a memory barrier
  * performed afterwards.
  */
-void __attribute__((weak)) __iowrite32_copy(void __iomem *to,
-					    const void *from,
-					    size_t count)
+#ifndef __iowrite32_copy
+void __iowrite32_copy(void __iomem *to, const void *from, size_t count)
 {
 	u32 __iomem *dst = to;
 	const u32 *src = from;
@@ -28,6 +27,7 @@ void __attribute__((weak)) __iowrite32_copy(void __iomem *to,
 		__raw_writel(*src++, dst++);
 }
 EXPORT_SYMBOL_GPL(__iowrite32_copy);
+#endif
 
 /**
  * __ioread32_copy - copy data from MMIO space, in 32-bit units
@@ -60,9 +60,8 @@ EXPORT_SYMBOL_GPL(__ioread32_copy);
  * time.  Order of access is not guaranteed, nor is a memory barrier
  * performed afterwards.
  */
-void __attribute__((weak)) __iowrite64_copy(void __iomem *to,
-					    const void *from,
-					    size_t count)
+#ifndef __iowrite64_copy
+void __iowrite64_copy(void __iomem *to, const void *from, size_t count)
 {
 #ifdef CONFIG_64BIT
 	u64 __iomem *dst = to;
@@ -75,5 +74,5 @@ void __attribute__((weak)) __iowrite64_copy(void __iomem *to,
 	__iowrite32_copy(to, from, count * 2);
 #endif
 }
-
 EXPORT_SYMBOL_GPL(__iowrite64_copy);
+#endif