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

Pull infiniband update from Roland Dreier:
 "Main batch of InfiniBand/RDMA changes for 3.9:

   - SRP error handling fixes from Bart Van Assche

   - Implementation of memory windows for mlx4 from Shani Michaeli

   - Lots of cxgb4 HW driver fixes from Vipul Pandya

   - Make iSER work for virtual functions, other fixes from Or Gerlitz

   - Fix for bug in qib HW driver from Mike Marciniszyn

   - IPoIB fixes from me, Itai Garbi, Shlomo Pongratz, Yan Burman

   - Various cleanups and warning fixes from Julia Lawall, Paul Bolle,
     Wei Yongjun"

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (41 commits)
  IB/mlx4: Advertise MW support
  IB/mlx4: Support memory window binding
  mlx4: Implement memory windows allocation and deallocation
  mlx4_core: Enable memory windows in {INIT, QUERY}_HCA
  mlx4_core: Disable memory windows for virtual functions
  IPoIB: Free ipoib neigh on path record failure so path rec queries are retried
  IB/srp: Fail I/O requests if the transport is offline
  IB/srp: Avoid endless SCSI error handling loop
  IB/srp: Avoid sending a task management function needlessly
  IB/srp: Track connection state properly
  IB/mlx4: Remove redundant NULL check before kfree
  IB/mlx4: Fix compiler warning about uninitialized 'vlan' variable
  IB/mlx4: Convert is_xxx variables in build_mlx_header() to bool
  IB/iser: Enable iser when FMRs are not supported
  IB/iser: Avoid error prints on EAGAIN registration failures
  IB/iser: Use proper define for the commands per LUN value advertised to SCSI ML
  IB/uverbs: Implement memory windows support in uverbs
  IB/core: Add "type 2" memory windows support
  mlx4_core: Propagate MR deregistration failures to caller
  mlx4_core: Rename MPT-related functions to have mpt_ prefix
  ...

4 files changed, 111 insertions, 19 deletions

diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 6d48fce06b4a..811f91cf5e8c 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -171,6 +171,7 @@ enum {
 #define MLX4_ATTR_EXTENDED_PORT_INFO    cpu_to_be16(0xff90)
 
 enum {
+        MLX4_BMME_FLAG_WIN_TYPE_2B      = 1 <<  1,
         MLX4_BMME_FLAG_LOCAL_INV        = 1 <<  6,
         MLX4_BMME_FLAG_REMOTE_INV       = 1 <<  7,
         MLX4_BMME_FLAG_TYPE_2_WIN       = 1 <<  9,
@@ -238,7 +239,8 @@ enum {
         MLX4_PERM_LOCAL_WRITE   = 1 << 11,
         MLX4_PERM_REMOTE_READ   = 1 << 12,
         MLX4_PERM_REMOTE_WRITE  = 1 << 13,
-        MLX4_PERM_ATOMIC        = 1 << 14
+        MLX4_PERM_ATOMIC        = 1 << 14,
+        MLX4_PERM_BIND_MW       = 1 << 15,
 };
 
 enum {
@@ -504,6 +506,18 @@ struct mlx4_mr {
         int                     enabled;
 };
 
+enum mlx4_mw_type {
+        MLX4_MW_TYPE_1 = 1,
+        MLX4_MW_TYPE_2 = 2,
+};
+
+struct mlx4_mw {
+        u32                     key;
+        u32                     pd;
+        enum mlx4_mw_type       type;
+        int                     enabled;
+};
+
 struct mlx4_fmr {
         struct mlx4_mr          mr;
         struct mlx4_mpt_entry  *mpt;
@@ -802,8 +816,12 @@ u64 mlx4_mtt_addr(struct mlx4_dev *dev, struct mlx4_mtt *mtt);
 
 int mlx4_mr_alloc(struct mlx4_dev *dev, u32 pd, u64 iova, u64 size, u32 access,
                   int npages, int page_shift, struct mlx4_mr *mr);
-void mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr);
+int mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr);
 int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr);
+int mlx4_mw_alloc(struct mlx4_dev *dev, u32 pd, enum mlx4_mw_type type,
+                  struct mlx4_mw *mw);
+void mlx4_mw_free(struct mlx4_dev *dev, struct mlx4_mw *mw);
+int mlx4_mw_enable(struct mlx4_dev *dev, struct mlx4_mw *mw);
 int mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
                    int start_index, int npages, u64 *page_list);
 int mlx4_buf_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
diff --git a/include/linux/mlx4/qp.h b/include/linux/mlx4/qp.h
index 4b4ad6ffef92..67f46ad6920a 100644
--- a/include/linux/mlx4/qp.h
+++ b/include/linux/mlx4/qp.h
@@ -265,6 +265,11 @@ struct mlx4_wqe_lso_seg {
         __be32                  header[0];
 };
 
+enum mlx4_wqe_bind_seg_flags2 {
+        MLX4_WQE_BIND_ZERO_BASED = (1 << 30),
+        MLX4_WQE_BIND_TYPE_2     = (1 << 31),
+};
+
 struct mlx4_wqe_bind_seg {
         __be32                  flags1;
         __be32                  flags2;
@@ -277,9 +282,9 @@ struct mlx4_wqe_bind_seg {
 enum {
         MLX4_WQE_FMR_PERM_LOCAL_READ    = 1 << 27,
         MLX4_WQE_FMR_PERM_LOCAL_WRITE   = 1 << 28,
-        MLX4_WQE_FMR_PERM_REMOTE_READ   = 1 << 29,
-        MLX4_WQE_FMR_PERM_REMOTE_WRITE  = 1 << 30,
-        MLX4_WQE_FMR_PERM_ATOMIC        = 1 << 31
+        MLX4_WQE_FMR_AND_BIND_PERM_REMOTE_READ  = 1 << 29,
+        MLX4_WQE_FMR_AND_BIND_PERM_REMOTE_WRITE = 1 << 30,
+        MLX4_WQE_FMR_AND_BIND_PERM_ATOMIC       = 1 << 31
 };
 
 struct mlx4_wqe_fmr_seg {
@@ -304,12 +309,10 @@ struct mlx4_wqe_fmr_ext_seg {
 };
 
 struct mlx4_wqe_local_inval_seg {
-        __be32                  flags;
-        u32                     reserved1;
+        u64                     reserved1;
         __be32                  mem_key;
-        u32                     reserved2[2];
-        __be32                  guest_id;
-        __be64                  pa;
+        u32                     reserved2;
+        u64                     reserved3[2];
 };
 
 struct mlx4_wqe_raddr_seg {
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 46bc045bbe15..98cc4b29fc5b 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -115,6 +115,8 @@ enum ib_device_cap_flags {
         IB_DEVICE_XRC                   = (1<<20),
         IB_DEVICE_MEM_MGT_EXTENSIONS    = (1<<21),
         IB_DEVICE_BLOCK_MULTICAST_LOOPBACK = (1<<22),
+        IB_DEVICE_MEM_WINDOW_TYPE_2A    = (1<<23),
+        IB_DEVICE_MEM_WINDOW_TYPE_2B    = (1<<24)
 };
 
 enum ib_atomic_cap {
@@ -715,6 +717,11 @@ enum ib_mig_state {
         IB_MIG_ARMED
 };
 
+enum ib_mw_type {
+        IB_MW_TYPE_1 = 1,
+        IB_MW_TYPE_2 = 2
+};
+
 struct ib_qp_attr {
         enum ib_qp_state        qp_state;
         enum ib_qp_state        cur_qp_state;
@@ -758,6 +765,7 @@ enum ib_wr_opcode {
         IB_WR_FAST_REG_MR,
         IB_WR_MASKED_ATOMIC_CMP_AND_SWP,
         IB_WR_MASKED_ATOMIC_FETCH_AND_ADD,
+        IB_WR_BIND_MW,
 };
 
 enum ib_send_flags {
@@ -780,6 +788,23 @@ struct ib_fast_reg_page_list {
         unsigned int            max_page_list_len;
 };
 
+/**
+ * struct ib_mw_bind_info - Parameters for a memory window bind operation.
+ * @mr: A memory region to bind the memory window to.
+ * @addr: The address where the memory window should begin.
+ * @length: The length of the memory window, in bytes.
+ * @mw_access_flags: Access flags from enum ib_access_flags for the window.
+ *
+ * This struct contains the shared parameters for type 1 and type 2
+ * memory window bind operations.
+ */
+struct ib_mw_bind_info {
+        struct ib_mr   *mr;
+        u64             addr;
+        u64             length;
+        int             mw_access_flags;
+};
+
 struct ib_send_wr {
         struct ib_send_wr      *next;
         u64                     wr_id;
@@ -823,6 +848,12 @@ struct ib_send_wr {
                         int                             access_flags;
                         u32                             rkey;
                 } fast_reg;
+                struct {
+                        struct ib_mw            *mw;
+                        /* The new rkey for the memory window. */
+                        u32                      rkey;
+                        struct ib_mw_bind_info   bind_info;
+                } bind_mw;
         } wr;
         u32                     xrc_remote_srq_num;     /* XRC TGT QPs only */
 };
@@ -839,7 +870,8 @@ enum ib_access_flags {
         IB_ACCESS_REMOTE_WRITE  = (1<<1),
         IB_ACCESS_REMOTE_READ   = (1<<2),
         IB_ACCESS_REMOTE_ATOMIC = (1<<3),
-        IB_ACCESS_MW_BIND       = (1<<4)
+        IB_ACCESS_MW_BIND       = (1<<4),
+        IB_ZERO_BASED           = (1<<5)
 };
 
 struct ib_phys_buf {
@@ -862,13 +894,16 @@ enum ib_mr_rereg_flags {
         IB_MR_REREG_ACCESS      = (1<<2)
 };
 
+/**
+ * struct ib_mw_bind - Parameters for a type 1 memory window bind operation.
+ * @wr_id:      Work request id.
+ * @send_flags: Flags from ib_send_flags enum.
+ * @bind_info:  More parameters of the bind operation.
+ */
 struct ib_mw_bind {
-        struct ib_mr   *mr;
-        u64             wr_id;
-        u64             addr;
-        u32             length;
-        int             send_flags;
-        int             mw_access_flags;
+        u64                    wr_id;
+        int                    send_flags;
+        struct ib_mw_bind_info bind_info;
 };
 
 struct ib_fmr_attr {
@@ -991,6 +1026,7 @@ struct ib_mw {
         struct ib_pd            *pd;
         struct ib_uobject       *uobject;
         u32                     rkey;
+        enum ib_mw_type         type;
 };
 
 struct ib_fmr {
@@ -1202,7 +1238,8 @@ struct ib_device {
                                                     int num_phys_buf,
                                                     int mr_access_flags,
                                                     u64 *iova_start);
-        struct ib_mw *             (*alloc_mw)(struct ib_pd *pd);
+        struct ib_mw *             (*alloc_mw)(struct ib_pd *pd,
+                                               enum ib_mw_type type);
         int                        (*bind_mw)(struct ib_qp *qp,
                                               struct ib_mw *mw,
                                               struct ib_mw_bind *mw_bind);
@@ -2019,6 +2056,8 @@ int ib_query_mr(struct ib_mr *mr, struct ib_mr_attr *mr_attr);
  * ib_dereg_mr - Deregisters a memory region and removes it from the
  *   HCA translation table.
  * @mr: The memory region to deregister.
+ *
+ * This function can fail, if the memory region has memory windows bound to it.
  */
 int ib_dereg_mr(struct ib_mr *mr);
 
@@ -2071,10 +2110,22 @@ static inline void ib_update_fast_reg_key(struct ib_mr *mr, u8 newkey)
 }
 
 /**
+ * ib_inc_rkey - increments the key portion of the given rkey. Can be used
+ * for calculating a new rkey for type 2 memory windows.
+ * @rkey - the rkey to increment.
+ */
+static inline u32 ib_inc_rkey(u32 rkey)
+{
+        const u32 mask = 0x000000ff;
+        return ((rkey + 1) & mask) | (rkey & ~mask);
+}
+
+/**
  * ib_alloc_mw - Allocates a memory window.
  * @pd: The protection domain associated with the memory window.
+ * @type: The type of the memory window (1 or 2).
  */
-struct ib_mw *ib_alloc_mw(struct ib_pd *pd);
+struct ib_mw *ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type);
 
 /**
  * ib_bind_mw - Posts a work request to the send queue of the specified
@@ -2084,6 +2135,10 @@ struct ib_mw *ib_alloc_mw(struct ib_pd *pd);
  * @mw: The memory window to bind.
  * @mw_bind: Specifies information about the memory window, including
  *   its address range, remote access rights, and associated memory region.
+ *
+ * If there is no immediate error, the function will update the rkey member
+ * of the mw parameter to its new value. The bind operation can still fail
+ * asynchronously.
  */
 static inline int ib_bind_mw(struct ib_qp *qp,
                              struct ib_mw *mw,
diff --git a/include/uapi/rdma/ib_user_verbs.h b/include/uapi/rdma/ib_user_verbs.h
index 81aba3a73aa3..805711ea2005 100644
--- a/include/uapi/rdma/ib_user_verbs.h
+++ b/include/uapi/rdma/ib_user_verbs.h
@@ -261,6 +261,22 @@ struct ib_uverbs_dereg_mr {
         __u32 mr_handle;
 };
 
+struct ib_uverbs_alloc_mw {
+        __u64 response;
+        __u32 pd_handle;
+        __u8  mw_type;
+        __u8  reserved[3];
+};
+
+struct ib_uverbs_alloc_mw_resp {
+        __u32 mw_handle;
+        __u32 rkey;
+};
+
+struct ib_uverbs_dealloc_mw {
+        __u32 mw_handle;
+};
+
 struct ib_uverbs_create_comp_channel {
         __u64 response;
 };