14 files changed, 266 insertions, 263 deletions

diff --git a/drivers/infiniband/hw/mthca/mthca_av.c b/drivers/infiniband/hw/mthca/mthca_av.c
index d58dcbe6648..e596210f11b 100644
--- a/drivers/infiniband/hw/mthca/mthca_av.c
+++ b/drivers/infiniband/hw/mthca/mthca_av.c
@@ -41,16 +41,16 @@
 #include "mthca_dev.h"
 
 struct mthca_av {
-        u32 port_pd;
-        u8  reserved1;
-        u8  g_slid;
-        u16 dlid;
-        u8  reserved2;
-        u8  gid_index;
-        u8  msg_sr;
-        u8  hop_limit;
-        u32 sl_tclass_flowlabel;
-        u32 dgid[4];
+        __be32 port_pd;
+        u8     reserved1;
+        u8     g_slid;
+        __be16 dlid;
+        u8     reserved2;
+        u8     gid_index;
+        u8     msg_sr;
+        u8     hop_limit;
+        __be32 sl_tclass_flowlabel;
+        __be32 dgid[4];
 };
 
 int mthca_create_ah(struct mthca_dev *dev,
@@ -128,7 +128,7 @@ on_hca_fail:
                           av, (unsigned long) ah->avdma);
                 for (j = 0; j < 8; ++j)
                         printk(KERN_DEBUG "  [%2x] %08x\n",
-                               j * 4, be32_to_cpu(((u32 *) av)[j]));
+                               j * 4, be32_to_cpu(((__be32 *) av)[j]));
         }
 
         if (ah->type == MTHCA_AH_ON_HCA) {
@@ -169,7 +169,7 @@ int mthca_read_ah(struct mthca_dev *dev, struct mthca_ah *ah,
 
         header->lrh.service_level   = be32_to_cpu(ah->av->sl_tclass_flowlabel) >> 28;
         header->lrh.destination_lid = ah->av->dlid;
-        header->lrh.source_lid      = ah->av->g_slid & 0x7f;
+        header->lrh.source_lid      = cpu_to_be16(ah->av->g_slid & 0x7f);
         if (ah->av->g_slid & 0x80) {
                 header->grh_present = 1;
                 header->grh.traffic_class =
diff --git a/drivers/infiniband/hw/mthca/mthca_cmd.c b/drivers/infiniband/hw/mthca/mthca_cmd.c
index 0ff5900e093..1e60487ecd7 100644
--- a/drivers/infiniband/hw/mthca/mthca_cmd.c
+++ b/drivers/infiniband/hw/mthca/mthca_cmd.c
@@ -220,20 +220,20 @@ static int mthca_cmd_post(struct mthca_dev *dev,
          * (and some architectures such as ia64 implement memcpy_toio
          * in terms of writeb).
          */
-        __raw_writel(cpu_to_be32(in_param >> 32),           dev->hcr + 0 * 4);
-        __raw_writel(cpu_to_be32(in_param & 0xfffffffful),  dev->hcr + 1 * 4);
-        __raw_writel(cpu_to_be32(in_modifier),              dev->hcr + 2 * 4);
-        __raw_writel(cpu_to_be32(out_param >> 32),          dev->hcr + 3 * 4);
-        __raw_writel(cpu_to_be32(out_param & 0xfffffffful), dev->hcr + 4 * 4);
-        __raw_writel(cpu_to_be32(token << 16),              dev->hcr + 5 * 4);
+        __raw_writel((__force u32) cpu_to_be32(in_param >> 32),           dev->hcr + 0 * 4);
+        __raw_writel((__force u32) cpu_to_be32(in_param & 0xfffffffful),  dev->hcr + 1 * 4);
+        __raw_writel((__force u32) cpu_to_be32(in_modifier),              dev->hcr + 2 * 4);
+        __raw_writel((__force u32) cpu_to_be32(out_param >> 32),          dev->hcr + 3 * 4);
+        __raw_writel((__force u32) cpu_to_be32(out_param & 0xfffffffful), dev->hcr + 4 * 4);
+        __raw_writel((__force u32) cpu_to_be32(token << 16),              dev->hcr + 5 * 4);
 
         /* __raw_writel may not order writes. */
         wmb();
 
-        __raw_writel(cpu_to_be32((1 << HCR_GO_BIT)                |
-                                 (event ? (1 << HCA_E_BIT) : 0)   |
-                                 (op_modifier << HCR_OPMOD_SHIFT) |
-                                 op),                       dev->hcr + 6 * 4);
+        __raw_writel((__force u32) cpu_to_be32((1 << HCR_GO_BIT)                |
+                                               (event ? (1 << HCA_E_BIT) : 0)   |
+                                               (op_modifier << HCR_OPMOD_SHIFT) |
+                                               op),                       dev->hcr + 6 * 4);
 
 out:
         up(&dev->cmd.hcr_sem);
@@ -274,12 +274,14 @@ static int mthca_cmd_poll(struct mthca_dev *dev,
                 goto out;
         }
 
-        if (out_is_imm) {
-                memcpy_fromio(out_param, dev->hcr + HCR_OUT_PARAM_OFFSET, sizeof (u64));
-                be64_to_cpus(out_param);
-        }
+        if (out_is_imm)
+                *out_param = 
+                        (u64) be32_to_cpu((__force __be32)
+                                          __raw_readl(dev->hcr + HCR_OUT_PARAM_OFFSET)) << 32 |
+                        (u64) be32_to_cpu((__force __be32)
+                                          __raw_readl(dev->hcr + HCR_OUT_PARAM_OFFSET + 4));
 
-        *status = be32_to_cpu(__raw_readl(dev->hcr + HCR_STATUS_OFFSET)) >> 24;
+        *status = be32_to_cpu((__force __be32) __raw_readl(dev->hcr + HCR_STATUS_OFFSET)) >> 24;
 
 out:
         up(&dev->cmd.poll_sem);
@@ -1122,7 +1124,7 @@ int mthca_INIT_HCA(struct mthca_dev *dev,
                    u8 *status)
 {
         struct mthca_mailbox *mailbox;
-        u32 *inbox;
+        __be32 *inbox;
         int err;
 
 #define INIT_HCA_IN_SIZE                 0x200
@@ -1343,7 +1345,7 @@ int mthca_MAP_ICM(struct mthca_dev *dev, struct mthca_icm *icm, u64 virt, u8 *st
 int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt, u8 *status)
 {
         struct mthca_mailbox *mailbox;
-        u64 *inbox;
+        __be64 *inbox;
         int err;
 
         mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL);
@@ -1514,7 +1516,7 @@ int mthca_MODIFY_QP(struct mthca_dev *dev, int trans, u32 num,
                                 if (i % 8 == 0)
                                         printk("  [%02x] ", i * 4);
                                 printk(" %08x",
-                                       be32_to_cpu(((u32 *) mailbox->buf)[i + 2]));
+                                       be32_to_cpu(((__be32 *) mailbox->buf)[i + 2]));
                                 if ((i + 1) % 8 == 0)
                                         printk("\n");
                         }
@@ -1534,7 +1536,7 @@ int mthca_MODIFY_QP(struct mthca_dev *dev, int trans, u32 num,
                                 if (i % 8 == 0)
                                         printk("[%02x] ", i * 4);
                                 printk(" %08x",
-                                       be32_to_cpu(((u32 *) mailbox->buf)[i + 2]));
+                                       be32_to_cpu(((__be32 *) mailbox->buf)[i + 2]));
                                 if ((i + 1) % 8 == 0)
                                         printk("\n");
                         }
diff --git a/drivers/infiniband/hw/mthca/mthca_cq.c b/drivers/infiniband/hw/mthca/mthca_cq.c
index bd7807cec50..907867d1f2e 100644
--- a/drivers/infiniband/hw/mthca/mthca_cq.c
+++ b/drivers/infiniband/hw/mthca/mthca_cq.c
@@ -57,21 +57,21 @@ enum {
  * Must be packed because start is 64 bits but only aligned to 32 bits.
  */
 struct mthca_cq_context {
-        u32 flags;
-        u64 start;
-        u32 logsize_usrpage;
-        u32 error_eqn;          /* Tavor only */
-        u32 comp_eqn;
-        u32 pd;
-        u32 lkey;
-        u32 last_notified_index;
-        u32 solicit_producer_index;
-        u32 consumer_index;
-        u32 producer_index;
-        u32 cqn;
-        u32 ci_db;              /* Arbel only */
-        u32 state_db;           /* Arbel only */
-        u32 reserved;
+        __be32 flags;
+        __be64 start;
+        __be32 logsize_usrpage;
+        __be32 error_eqn;       /* Tavor only */
+        __be32 comp_eqn;
+        __be32 pd;
+        __be32 lkey;
+        __be32 last_notified_index;
+        __be32 solicit_producer_index;
+        __be32 consumer_index;
+        __be32 producer_index;
+        __be32 cqn;
+        __be32 ci_db;           /* Arbel only */
+        __be32 state_db;        /* Arbel only */
+        u32    reserved;
 } __attribute__((packed));
 
 #define MTHCA_CQ_STATUS_OK          ( 0 << 28)
@@ -110,31 +110,31 @@ enum {
 };
 
 struct mthca_cqe {
-        u32 my_qpn;
-        u32 my_ee;
-        u32 rqpn;
-        u16 sl_g_mlpath;
-        u16 rlid;
-        u32 imm_etype_pkey_eec;
-        u32 byte_cnt;
-        u32 wqe;
-        u8  opcode;
-        u8  is_send;
-        u8  reserved;
-        u8  owner;
+        __be32 my_qpn;
+        __be32 my_ee;
+        __be32 rqpn;
+        __be16 sl_g_mlpath;
+        __be16 rlid;
+        __be32 imm_etype_pkey_eec;
+        __be32 byte_cnt;
+        __be32 wqe;
+        u8     opcode;
+        u8     is_send;
+        u8     reserved;
+        u8     owner;
 };
 
 struct mthca_err_cqe {
-        u32 my_qpn;
-        u32 reserved1[3];
-        u8  syndrome;
-        u8  reserved2;
-        u16 db_cnt;
-        u32 reserved3;
-        u32 wqe;
-        u8  opcode;
-        u8  reserved4[2];
-        u8  owner;
+        __be32 my_qpn;
+        u32    reserved1[3];
+        u8     syndrome;
+        u8     reserved2;
+        __be16 db_cnt;
+        u32    reserved3;
+        __be32 wqe;
+        u8     opcode;
+        u8     reserved4[2];
+        u8     owner;
 };
 
 #define MTHCA_CQ_ENTRY_OWNER_SW      (0 << 7)
@@ -193,7 +193,7 @@ static void dump_cqe(struct mthca_dev *dev, void *cqe_ptr)
 static inline void update_cons_index(struct mthca_dev *dev, struct mthca_cq *cq,
                                      int incr)
 {
-        u32 doorbell[2];
+        __be32 doorbell[2];
 
         if (mthca_is_memfree(dev)) {
                 *cq->set_ci_db = cpu_to_be32(cq->cons_index);
@@ -293,7 +293,7 @@ static int handle_error_cqe(struct mthca_dev *dev, struct mthca_cq *cq,
 {
         int err;
         int dbd;
-        u32 new_wqe;
+        __be32 new_wqe;
 
         if (cqe->syndrome == SYNDROME_LOCAL_QP_OP_ERR) {
                 mthca_dbg(dev, "local QP operation err "
@@ -586,13 +586,13 @@ int mthca_poll_cq(struct ib_cq *ibcq, int num_entries,
 
 int mthca_tavor_arm_cq(struct ib_cq *cq, enum ib_cq_notify notify)
 {
-        u32 doorbell[2];
+        __be32 doorbell[2];
 
         doorbell[0] = cpu_to_be32((notify == IB_CQ_SOLICITED ?
                                    MTHCA_TAVOR_CQ_DB_REQ_NOT_SOL :
                                    MTHCA_TAVOR_CQ_DB_REQ_NOT)      |
                                   to_mcq(cq)->cqn);
-        doorbell[1] = 0xffffffff;
+        doorbell[1] = (__force __be32) 0xffffffff;
 
         mthca_write64(doorbell,
                       to_mdev(cq->device)->kar + MTHCA_CQ_DOORBELL,
@@ -604,9 +604,9 @@ int mthca_tavor_arm_cq(struct ib_cq *cq, enum ib_cq_notify notify)
 int mthca_arbel_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify notify)
 {
         struct mthca_cq *cq = to_mcq(ibcq);
-        u32 doorbell[2];
+        __be32 doorbell[2];
         u32 sn;
-        u32 ci;
+        __be32 ci;
 
         sn = cq->arm_sn & 3;
         ci = cpu_to_be32(cq->cons_index);
@@ -813,7 +813,6 @@ int mthca_init_cq(struct mthca_dev *dev, int nent,
         cq_context->flags           = cpu_to_be32(MTHCA_CQ_STATUS_OK      |
                                                   MTHCA_CQ_STATE_DISARMED |
                                                   MTHCA_CQ_FLAG_TR);
-        cq_context->start           = cpu_to_be64(0);
         cq_context->logsize_usrpage = cpu_to_be32((ffs(nent) - 1) << 24);
         if (ctx)
                 cq_context->logsize_usrpage |= cpu_to_be32(ctx->uar.index);
@@ -906,7 +905,7 @@ void mthca_free_cq(struct mthca_dev *dev,
                 mthca_warn(dev, "HW2SW_CQ returned status 0x%02x\n", status);
 
         if (0) {
-                u32 *ctx = mailbox->buf;
+                __be32 *ctx = mailbox->buf;
                 int j;
 
                 printk(KERN_ERR "context for CQN %x (cons index %x, next sw %d)\n",
diff --git a/drivers/infiniband/hw/mthca/mthca_dev.h b/drivers/infiniband/hw/mthca/mthca_dev.h
index 33162a960c7..3519ca4e086 100644
--- a/drivers/infiniband/hw/mthca/mthca_dev.h
+++ b/drivers/infiniband/hw/mthca/mthca_dev.h
@@ -333,14 +333,13 @@ extern void __buggy_use_of_MTHCA_PUT(void);
 
 #define MTHCA_PUT(dest, source, offset)                               \
         do {                                                          \
-                __typeof__(source) *__p =                             \
-                        (__typeof__(source) *) ((char *) (dest) + (offset)); \
+                void *__d = ((char *) (dest) + (offset));             \
                 switch (sizeof(source)) {                             \
-                        case 1: *__p = (source);            break;    \
-                        case 2: *__p = cpu_to_be16(source); break;    \
-                        case 4: *__p = cpu_to_be32(source); break;    \
-                        case 8: *__p = cpu_to_be64(source); break;    \
-                        default: __buggy_use_of_MTHCA_PUT();          \
+                case 1: *(u8 *) __d = (source);                break; \
+                case 2: *(__be16 *) __d = cpu_to_be16(source); break; \
+                case 4: *(__be32 *) __d = cpu_to_be32(source); break; \
+                case 8: *(__be64 *) __d = cpu_to_be64(source); break; \
+                default: __buggy_use_of_MTHCA_PUT();                  \
                 }                                                     \
         } while (0)
 
@@ -435,7 +434,7 @@ int mthca_arbel_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 int mthca_arbel_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
                              struct ib_recv_wr **bad_wr);
 int mthca_free_err_wqe(struct mthca_dev *dev, struct mthca_qp *qp, int is_send,
-                       int index, int *dbd, u32 *new_wqe);
+                       int index, int *dbd, __be32 *new_wqe);
 int mthca_alloc_qp(struct mthca_dev *dev,
                    struct mthca_pd *pd,
                    struct mthca_cq *send_cq,
diff --git a/drivers/infiniband/hw/mthca/mthca_doorbell.h b/drivers/infiniband/hw/mthca/mthca_doorbell.h
index 3be4a4a606a..dd9a44d170c 100644
--- a/drivers/infiniband/hw/mthca/mthca_doorbell.h
+++ b/drivers/infiniband/hw/mthca/mthca_doorbell.h
@@ -58,13 +58,13 @@ static inline void mthca_write64_raw(__be64 val, void __iomem *dest)
         __raw_writeq((__force u64) val, dest);
 }
 
-static inline void mthca_write64(u32 val[2], void __iomem *dest,
+static inline void mthca_write64(__be32 val[2], void __iomem *dest,
                                  spinlock_t *doorbell_lock)
 {
         __raw_writeq(*(u64 *) val, dest);
 }
 
-static inline void mthca_write_db_rec(u32 val[2], u32 *db)
+static inline void mthca_write_db_rec(__be32 val[2], __be32 *db)
 {
         *(u64 *) db = *(u64 *) val;
 }
@@ -87,18 +87,18 @@ static inline void mthca_write64_raw(__be64 val, void __iomem *dest)
         __raw_writel(((__force u32 *) &val)[1], dest + 4);
 }
 
-static inline void mthca_write64(u32 val[2], void __iomem *dest,
+static inline void mthca_write64(__be32 val[2], void __iomem *dest,
                                  spinlock_t *doorbell_lock)
 {
         unsigned long flags;
 
         spin_lock_irqsave(doorbell_lock, flags);
-        __raw_writel(val[0], dest);
-        __raw_writel(val[1], dest + 4);
+        __raw_writel((__force u32) val[0], dest);
+        __raw_writel((__force u32) val[1], dest + 4);
         spin_unlock_irqrestore(doorbell_lock, flags);
 }
 
-static inline void mthca_write_db_rec(u32 val[2], u32 *db)
+static inline void mthca_write_db_rec(__be32 val[2], __be32 *db)
 {
         db[0] = val[0];
         wmb();
diff --git a/drivers/infiniband/hw/mthca/mthca_eq.c b/drivers/infiniband/hw/mthca/mthca_eq.c
index 54a809adab6..18f0981eb0c 100644
--- a/drivers/infiniband/hw/mthca/mthca_eq.c
+++ b/drivers/infiniband/hw/mthca/mthca_eq.c
@@ -52,18 +52,18 @@ enum {
  * Must be packed because start is 64 bits but only aligned to 32 bits.
  */
 struct mthca_eq_context {
-        u32 flags;
-        u64 start;
-        u32 logsize_usrpage;
-        u32 tavor_pd;           /* reserved for Arbel */
-        u8  reserved1[3];
-        u8  intr;
-        u32 arbel_pd;           /* lost_count for Tavor */
-        u32 lkey;
-        u32 reserved2[2];
-        u32 consumer_index;
-        u32 producer_index;
-        u32 reserved3[4];
+        __be32 flags;
+        __be64 start;
+        __be32 logsize_usrpage;
+        __be32 tavor_pd;        /* reserved for Arbel */
+        u8     reserved1[3];
+        u8     intr;
+        __be32 arbel_pd;        /* lost_count for Tavor */
+        __be32 lkey;
+        u32    reserved2[2];
+        __be32 consumer_index;
+        __be32 producer_index;
+        u32    reserved3[4];
 } __attribute__((packed));
 
 #define MTHCA_EQ_STATUS_OK          ( 0 << 28)
@@ -128,28 +128,28 @@ struct mthca_eqe {
         union {
                 u32 raw[6];
                 struct {
-                        u32 cqn;
+                        __be32 cqn;
                 } __attribute__((packed)) comp;
                 struct {
-                        u16 reserved1;
-                        u16 token;
-                        u32 reserved2;
-                        u8  reserved3[3];
-                        u8  status;
-                        u64 out_param;
+                        u16    reserved1;
+                        __be16 token;
+                        u32    reserved2;
+                        u8     reserved3[3];
+                        u8     status;
+                        __be64 out_param;
                 } __attribute__((packed)) cmd;
                 struct {
-                        u32 qpn;
+                        __be32 qpn;
                 } __attribute__((packed)) qp;
                 struct {
-                        u32 cqn;
-                        u32 reserved1;
-                        u8  reserved2[3];
-                        u8  syndrome;
+                        __be32 cqn;
+                        u32    reserved1;
+                        u8     reserved2[3];
+                        u8     syndrome;
                 } __attribute__((packed)) cq_err;
                 struct {
-                        u32 reserved1[2];
-                        u32 port;
+                        u32    reserved1[2];
+                        __be32 port;
                 } __attribute__((packed)) port_change;
         } event;
         u8 reserved3[3];
@@ -168,7 +168,7 @@ static inline u64 async_mask(struct mthca_dev *dev)
 
 static inline void tavor_set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
 {
-        u32 doorbell[2];
+        __be32 doorbell[2];
 
         doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_SET_CI | eq->eqn);
         doorbell[1] = cpu_to_be32(ci & (eq->nent - 1));
@@ -191,8 +191,8 @@ static inline void arbel_set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u
 {
         /* See comment in tavor_set_eq_ci() above. */
         wmb();
-        __raw_writel(cpu_to_be32(ci), dev->eq_regs.arbel.eq_set_ci_base +
-                     eq->eqn * 8);
+        __raw_writel((__force u32) cpu_to_be32(ci),
+                     dev->eq_regs.arbel.eq_set_ci_base + eq->eqn * 8);
         /* We still want ordering, just not swabbing, so add a barrier */
         mb();
 }
@@ -207,7 +207,7 @@ static inline void set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
 
 static inline void tavor_eq_req_not(struct mthca_dev *dev, int eqn)
 {
-        u32 doorbell[2];
+        __be32 doorbell[2];
 
         doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_REQ_NOT | eqn);
         doorbell[1] = 0;
@@ -225,7 +225,7 @@ static inline void arbel_eq_req_not(struct mthca_dev *dev, u32 eqn_mask)
 static inline void disarm_cq(struct mthca_dev *dev, int eqn, int cqn)
 {
         if (!mthca_is_memfree(dev)) {
-                u32 doorbell[2];
+                __be32 doorbell[2];
 
                 doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_DISARM_CQ | eqn);
                 doorbell[1] = cpu_to_be32(cqn);
diff --git a/drivers/infiniband/hw/mthca/mthca_mad.c b/drivers/infiniband/hw/mthca/mthca_mad.c
index 3c7fae6cb12..64fa78722cf 100644
--- a/drivers/infiniband/hw/mthca/mthca_mad.c
+++ b/drivers/infiniband/hw/mthca/mthca_mad.c
@@ -194,7 +194,7 @@ int mthca_process_mad(struct ib_device *ibdev,
 {
         int err;
         u8 status;
-        u16 slid = in_wc ? in_wc->slid : IB_LID_PERMISSIVE;
+        u16 slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
 
         /* Forward locally generated traps to the SM */
         if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP &&
diff --git a/drivers/infiniband/hw/mthca/mthca_mcg.c b/drivers/infiniband/hw/mthca/mthca_mcg.c
index 5be7d949dbf..a2707605f4c 100644
--- a/drivers/infiniband/hw/mthca/mthca_mcg.c
+++ b/drivers/infiniband/hw/mthca/mthca_mcg.c
@@ -42,10 +42,10 @@ enum {
 };
 
 struct mthca_mgm {
-        u32 next_gid_index;
-        u32 reserved[3];
-        u8  gid[16];
-        u32 qp[MTHCA_QP_PER_MGM];
+        __be32 next_gid_index;
+        u32    reserved[3];
+        u8     gid[16];
+        __be32 qp[MTHCA_QP_PER_MGM];
 };
 
 static const u8 zero_gid[16];   /* automatically initialized to 0 */
@@ -94,10 +94,14 @@ static int find_mgm(struct mthca_dev *dev,
         if (0)
                 mthca_dbg(dev, "Hash for %04x:%04x:%04x:%04x:"
                           "%04x:%04x:%04x:%04x is %04x\n",
-                          be16_to_cpu(((u16 *) gid)[0]), be16_to_cpu(((u16 *) gid)[1]),
-                          be16_to_cpu(((u16 *) gid)[2]), be16_to_cpu(((u16 *) gid)[3]),
-                          be16_to_cpu(((u16 *) gid)[4]), be16_to_cpu(((u16 *) gid)[5]),
-                          be16_to_cpu(((u16 *) gid)[6]), be16_to_cpu(((u16 *) gid)[7]),
+                          be16_to_cpu(((__be16 *) gid)[0]),
+                          be16_to_cpu(((__be16 *) gid)[1]),
+                          be16_to_cpu(((__be16 *) gid)[2]),
+                          be16_to_cpu(((__be16 *) gid)[3]),
+                          be16_to_cpu(((__be16 *) gid)[4]),
+                          be16_to_cpu(((__be16 *) gid)[5]),
+                          be16_to_cpu(((__be16 *) gid)[6]),
+                          be16_to_cpu(((__be16 *) gid)[7]),
                           *hash);
 
         *index = *hash;
@@ -258,14 +262,14 @@ int mthca_multicast_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
         if (index == -1) {
                 mthca_err(dev, "MGID %04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x "
                           "not found\n",
-                          be16_to_cpu(((u16 *) gid->raw)[0]),
-                          be16_to_cpu(((u16 *) gid->raw)[1]),
-                          be16_to_cpu(((u16 *) gid->raw)[2]),
-                          be16_to_cpu(((u16 *) gid->raw)[3]),
-                          be16_to_cpu(((u16 *) gid->raw)[4]),
-                          be16_to_cpu(((u16 *) gid->raw)[5]),
-                          be16_to_cpu(((u16 *) gid->raw)[6]),
-                          be16_to_cpu(((u16 *) gid->raw)[7]));
+                          be16_to_cpu(((__be16 *) gid->raw)[0]),
+                          be16_to_cpu(((__be16 *) gid->raw)[1]),
+                          be16_to_cpu(((__be16 *) gid->raw)[2]),
+                          be16_to_cpu(((__be16 *) gid->raw)[3]),
+                          be16_to_cpu(((__be16 *) gid->raw)[4]),
+                          be16_to_cpu(((__be16 *) gid->raw)[5]),
+                          be16_to_cpu(((__be16 *) gid->raw)[6]),
+                          be16_to_cpu(((__be16 *) gid->raw)[7]));
                 err = -EINVAL;
                 goto out;
         }
diff --git a/drivers/infiniband/hw/mthca/mthca_memfree.c b/drivers/infiniband/hw/mthca/mthca_memfree.c
index 9efb0322c76..fba0a53ba6e 100644
--- a/drivers/infiniband/hw/mthca/mthca_memfree.c
+++ b/drivers/infiniband/hw/mthca/mthca_memfree.c
@@ -482,7 +482,7 @@ void mthca_cleanup_user_db_tab(struct mthca_dev *dev, struct mthca_uar *uar,
         }
 }
 
-int mthca_alloc_db(struct mthca_dev *dev, int type, u32 qn, u32 **db)
+int mthca_alloc_db(struct mthca_dev *dev, int type, u32 qn, __be32 **db)
 {
         int group;
         int start, end, dir;
@@ -565,7 +565,7 @@ found:
 
         page->db_rec[j] = cpu_to_be64((qn << 8) | (type << 5));
 
-        *db = (u32 *) &page->db_rec[j];
+        *db = (__be32 *) &page->db_rec[j];
 
 out:
         up(&dev->db_tab->mutex);
diff --git a/drivers/infiniband/hw/mthca/mthca_memfree.h b/drivers/infiniband/hw/mthca/mthca_memfree.h
index 59c2f555b13..bafa51544aa 100644
--- a/drivers/infiniband/hw/mthca/mthca_memfree.h
+++ b/drivers/infiniband/hw/mthca/mthca_memfree.h
@@ -138,7 +138,7 @@ enum {
 
 struct mthca_db_page {
         DECLARE_BITMAP(used, MTHCA_DB_REC_PER_PAGE);
-        u64       *db_rec;
+        __be64    *db_rec;
         dma_addr_t mapping;
 };
 
@@ -173,7 +173,7 @@ void mthca_cleanup_user_db_tab(struct mthca_dev *dev, struct mthca_uar *uar,
 
 int mthca_init_db_tab(struct mthca_dev *dev);
 void mthca_cleanup_db_tab(struct mthca_dev *dev);
-int mthca_alloc_db(struct mthca_dev *dev, int type, u32 qn, u32 **db);
+int mthca_alloc_db(struct mthca_dev *dev, int type, u32 qn, __be32 **db);
 void mthca_free_db(struct mthca_dev *dev, int type, int db_index);
 
 #endif /* MTHCA_MEMFREE_H */
diff --git a/drivers/infiniband/hw/mthca/mthca_mr.c b/drivers/infiniband/hw/mthca/mthca_mr.c
index 15d9f8f290a..0965e66061b 100644
--- a/drivers/infiniband/hw/mthca/mthca_mr.c
+++ b/drivers/infiniband/hw/mthca/mthca_mr.c
@@ -51,18 +51,18 @@ struct mthca_mtt {
  * Must be packed because mtt_seg is 64 bits but only aligned to 32 bits.
  */
 struct mthca_mpt_entry {
-        u32 flags;
-        u32 page_size;
-        u32 key;
-        u32 pd;
-        u64 start;
-        u64 length;
-        u32 lkey;
-        u32 window_count;
-        u32 window_count_limit;
-        u64 mtt_seg;
-        u32 mtt_sz;             /* Arbel only */
-        u32 reserved[2];
+        __be32 flags;
+        __be32 page_size;
+        __be32 key;
+        __be32 pd;
+        __be64 start;
+        __be64 length;
+        __be32 lkey;
+        __be32 window_count;
+        __be32 window_count_limit;
+        __be64 mtt_seg;
+        __be32 mtt_sz;          /* Arbel only */
+        u32    reserved[2];
 } __attribute__((packed));
 
 #define MTHCA_MPT_FLAG_SW_OWNS       (0xfUL << 28)
@@ -248,7 +248,7 @@ int mthca_write_mtt(struct mthca_dev *dev, struct mthca_mtt *mtt,
                     int start_index, u64 *buffer_list, int list_len)
 {
         struct mthca_mailbox *mailbox;
-        u64 *mtt_entry;
+        __be64 *mtt_entry;
         int err = 0;
         u8 status;
         int i;
@@ -390,7 +390,7 @@ int mthca_mr_alloc(struct mthca_dev *dev, u32 pd, int buffer_size_shift,
                 for (i = 0; i < sizeof (struct mthca_mpt_entry) / 4; ++i) {
                         if (i % 4 == 0)
                                 printk("[%02x] ", i * 4);
-                        printk(" %08x", be32_to_cpu(((u32 *) mpt_entry)[i]));
+                        printk(" %08x", be32_to_cpu(((__be32 *) mpt_entry)[i]));
                         if ((i + 1) % 4 == 0)
                                 printk("\n");
                 }
@@ -563,7 +563,7 @@ int mthca_fmr_alloc(struct mthca_dev *dev, u32 pd,
                 for (i = 0; i < sizeof (struct mthca_mpt_entry) / 4; ++i) {
                         if (i % 4 == 0)
                                 printk("[%02x] ", i * 4);
-                        printk(" %08x", be32_to_cpu(((u32 *) mpt_entry)[i]));
+                        printk(" %08x", be32_to_cpu(((__be32 *) mpt_entry)[i]));
                         if ((i + 1) % 4 == 0)
                                 printk("\n");
                 }
@@ -670,7 +670,7 @@ int mthca_tavor_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list,
         mpt_entry.length = cpu_to_be64(list_len * (1ull << fmr->attr.page_size));
         mpt_entry.start  = cpu_to_be64(iova);
 
-        writel(mpt_entry.lkey, &fmr->mem.tavor.mpt->key);
+        __raw_writel((__force u32) mpt_entry.lkey, &fmr->mem.tavor.mpt->key);
         memcpy_toio(&fmr->mem.tavor.mpt->start, &mpt_entry.start,
                     offsetof(struct mthca_mpt_entry, window_count) -
                     offsetof(struct mthca_mpt_entry, start));
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index 34e6b8685ba..e2db5e00186 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -81,10 +81,10 @@ static int mthca_query_device(struct ib_device *ibdev,
         }
 
         props->device_cap_flags    = mdev->device_cap_flags;
-        props->vendor_id           = be32_to_cpup((u32 *) (out_mad->data + 36)) &
+        props->vendor_id           = be32_to_cpup((__be32 *) (out_mad->data + 36)) &
                 0xffffff;
-        props->vendor_part_id      = be16_to_cpup((u16 *) (out_mad->data + 30));
-        props->hw_ver              = be16_to_cpup((u16 *) (out_mad->data + 32));
+        props->vendor_part_id      = be16_to_cpup((__be16 *) (out_mad->data + 30));
+        props->hw_ver              = be16_to_cpup((__be16 *) (out_mad->data + 32));
         memcpy(&props->sys_image_guid, out_mad->data +  4, 8);
         memcpy(&props->node_guid,      out_mad->data + 12, 8);
 
@@ -138,16 +138,16 @@ static int mthca_query_port(struct ib_device *ibdev,
                 goto out;
         }
 
-        props->lid               = be16_to_cpup((u16 *) (out_mad->data + 16));
+        props->lid               = be16_to_cpup((__be16 *) (out_mad->data + 16));
         props->lmc               = out_mad->data[34] & 0x7;
-        props->sm_lid            = be16_to_cpup((u16 *) (out_mad->data + 18));
+        props->sm_lid            = be16_to_cpup((__be16 *) (out_mad->data + 18));
         props->sm_sl             = out_mad->data[36] & 0xf;
         props->state             = out_mad->data[32] & 0xf;
         props->phys_state        = out_mad->data[33] >> 4;
-        props->port_cap_flags    = be32_to_cpup((u32 *) (out_mad->data + 20));
+        props->port_cap_flags    = be32_to_cpup((__be32 *) (out_mad->data + 20));
         props->gid_tbl_len       = to_mdev(ibdev)->limits.gid_table_len;
         props->pkey_tbl_len      = to_mdev(ibdev)->limits.pkey_table_len;
-        props->qkey_viol_cntr    = be16_to_cpup((u16 *) (out_mad->data + 48));
+        props->qkey_viol_cntr    = be16_to_cpup((__be16 *) (out_mad->data + 48));
         props->active_width      = out_mad->data[31] & 0xf;
         props->active_speed      = out_mad->data[35] >> 4;
 
@@ -223,7 +223,7 @@ static int mthca_query_pkey(struct ib_device *ibdev,
                 goto out;
         }
 
-        *pkey = be16_to_cpu(((u16 *) out_mad->data)[index % 32]);
+        *pkey = be16_to_cpu(((__be16 *) out_mad->data)[index % 32]);
 
  out:
         kfree(in_mad);
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.h b/drivers/infiniband/hw/mthca/mthca_provider.h
index 727aad8d4f3..624651edf57 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.h
+++ b/drivers/infiniband/hw/mthca/mthca_provider.h
@@ -182,9 +182,9 @@ struct mthca_cq {
 
         /* Next fields are Arbel only */
         int                    set_ci_db_index;
-        u32                   *set_ci_db;
+        __be32                *set_ci_db;
         int                    arm_db_index;
-        u32                   *arm_db;
+        __be32                *arm_db;
         int                    arm_sn;
 
         union {
@@ -207,7 +207,7 @@ struct mthca_wq {
         int        wqe_shift;
 
         int        db_index;    /* Arbel only */
-        u32       *db;
+        __be32    *db;
 };
 
 struct mthca_qp {
diff --git a/drivers/infiniband/hw/mthca/mthca_qp.c b/drivers/infiniband/hw/mthca/mthca_qp.c
index 2f429815d19..8fbb4f1f539 100644
--- a/drivers/infiniband/hw/mthca/mthca_qp.c
+++ b/drivers/infiniband/hw/mthca/mthca_qp.c
@@ -97,62 +97,62 @@ enum {
 };
 
 struct mthca_qp_path {
-        u32 port_pkey;
-        u8  rnr_retry;
-        u8  g_mylmc;
-        u16 rlid;
-        u8  ackto;
-        u8  mgid_index;
-        u8  static_rate;
-        u8  hop_limit;
-        u32 sl_tclass_flowlabel;
-        u8  rgid[16];
+        __be32 port_pkey;
+        u8     rnr_retry;
+        u8     g_mylmc;
+        __be16 rlid;
+        u8     ackto;
+        u8     mgid_index;
+        u8     static_rate;
+        u8     hop_limit;
+        __be32 sl_tclass_flowlabel;
+        u8     rgid[16];
 } __attribute__((packed));
 
 struct mthca_qp_context {
-        u32 flags;
-        u32 tavor_sched_queue;  /* Reserved on Arbel */
-        u8  mtu_msgmax;
-        u8  rq_size_stride;     /* Reserved on Tavor */
-        u8  sq_size_stride;     /* Reserved on Tavor */
-        u8  rlkey_arbel_sched_queue;    /* Reserved on Tavor */
-        u32 usr_page;
-        u32 local_qpn;
-        u32 remote_qpn;
-        u32 reserved1[2];
+        __be32 flags;
+        __be32 tavor_sched_queue; /* Reserved on Arbel */
+        u8     mtu_msgmax;
+        u8     rq_size_stride;  /* Reserved on Tavor */
+        u8     sq_size_stride;  /* Reserved on Tavor */
+        u8     rlkey_arbel_sched_queue; /* Reserved on Tavor */
+        __be32 usr_page;
+        __be32 local_qpn;
+        __be32 remote_qpn;
+        u32    reserved1[2];
         struct mthca_qp_path pri_path;
         struct mthca_qp_path alt_path;
-        u32 rdd;
-        u32 pd;
-        u32 wqe_base;
-        u32 wqe_lkey;
-        u32 params1;
-        u32 reserved2;
-        u32 next_send_psn;
-        u32 cqn_snd;
-        u32 snd_wqe_base_l;     /* Next send WQE on Tavor */
-        u32 snd_db_index;       /* (debugging only entries) */
-        u32 last_acked_psn;
-        u32 ssn;
-        u32 params2;
-        u32 rnr_nextrecvpsn;
-        u32 ra_buff_indx;
-        u32 cqn_rcv;
-        u32 rcv_wqe_base_l;     /* Next recv WQE on Tavor */
-        u32 rcv_db_index;       /* (debugging only entries) */
-        u32 qkey;
-        u32 srqn;
-        u32 rmsn;
-        u16 rq_wqe_counter;     /* reserved on Tavor */
-        u16 sq_wqe_counter;     /* reserved on Tavor */
-        u32 reserved3[18];
+        __be32 rdd;
+        __be32 pd;
+        __be32 wqe_base;
+        __be32 wqe_lkey;
+        __be32 params1;
+        __be32 reserved2;
+        __be32 next_send_psn;
+        __be32 cqn_snd;
+        __be32 snd_wqe_base_l;  /* Next send WQE on Tavor */
+        __be32 snd_db_index;    /* (debugging only entries) */
+        __be32 last_acked_psn;
+        __be32 ssn;
+        __be32 params2;
+        __be32 rnr_nextrecvpsn;
+        __be32 ra_buff_indx;
+        __be32 cqn_rcv;
+        __be32 rcv_wqe_base_l;  /* Next recv WQE on Tavor */
+        __be32 rcv_db_index;    /* (debugging only entries) */
+        __be32 qkey;
+        __be32 srqn;
+        __be32 rmsn;
+        __be16 rq_wqe_counter;  /* reserved on Tavor */
+        __be16 sq_wqe_counter;  /* reserved on Tavor */
+        u32    reserved3[18];
 } __attribute__((packed));
 
 struct mthca_qp_param {
-        u32 opt_param_mask;
-        u32 reserved1;
+        __be32 opt_param_mask;
+        u32    reserved1;
         struct mthca_qp_context context;
-        u32 reserved2[62];
+        u32    reserved2[62];
 } __attribute__((packed));
 
 enum {
@@ -191,62 +191,62 @@ enum {
 };
 
 struct mthca_next_seg {
-        u32 nda_op;             /* [31:6] next WQE [4:0] next opcode */
-        u32 ee_nds;             /* [31:8] next EE  [7] DBD [6] F [5:0] next WQE size */
-        u32 flags;              /* [3] CQ [2] Event [1] Solicit */
-        u32 imm;                /* immediate data */
+        __be32 nda_op;          /* [31:6] next WQE [4:0] next opcode */
+        __be32 ee_nds;          /* [31:8] next EE  [7] DBD [6] F [5:0] next WQE size */
+        __be32 flags;           /* [3] CQ [2] Event [1] Solicit */
+        __be32 imm;             /* immediate data */
 };
 
 struct mthca_tavor_ud_seg {
-        u32 reserved1;
-        u32 lkey;
-        u64 av_addr;
-        u32 reserved2[4];
-        u32 dqpn;
-        u32 qkey;
-        u32 reserved3[2];
+        u32    reserved1;
+        __be32 lkey;
+        __be64 av_addr;
+        u32    reserved2[4];
+        __be32 dqpn;
+        __be32 qkey;
+        u32    reserved3[2];
 };
 
 struct mthca_arbel_ud_seg {
-        u32 av[8];
-        u32 dqpn;
-        u32 qkey;
-        u32 reserved[2];
+        __be32 av[8];
+        __be32 dqpn;
+        __be32 qkey;
+        u32    reserved[2];
 };
 
 struct mthca_bind_seg {
-        u32 flags;              /* [31] Atomic [30] rem write [29] rem read */
-        u32 reserved;
-        u32 new_rkey;
-        u32 lkey;
-        u64 addr;
-        u64 length;
+        __be32 flags;           /* [31] Atomic [30] rem write [29] rem read */
+        u32    reserved;
+        __be32 new_rkey;
+        __be32 lkey;
+        __be64 addr;
+        __be64 length;
 };
 
 struct mthca_raddr_seg {
-        u64 raddr;
-        u32 rkey;
-        u32 reserved;
+        __be64 raddr;
+        __be32 rkey;
+        u32    reserved;
 };
 
 struct mthca_atomic_seg {
-        u64 swap_add;
-        u64 compare;
+        __be64 swap_add;
+        __be64 compare;
 };
 
 struct mthca_data_seg {
-        u32 byte_count;
-        u32 lkey;
-        u64 addr;
+        __be32 byte_count;
+        __be32 lkey;
+        __be64 addr;
 };
 
 struct mthca_mlx_seg {
-        u32 nda_op;
-        u32 nds;
-        u32 flags;              /* [17] VL15 [16] SLR [14:12] static rate
+        __be32 nda_op;
+        __be32 nds;
+        __be32 flags;           /* [17] VL15 [16] SLR [14:12] static rate
                                    [11:8] SL [3] C [2] E */
-        u16 rlid;
-        u16 vcrc;
+        __be16 rlid;
+        __be16 vcrc;
 };
 
 static const u8 mthca_opcode[] = {
@@ -1459,6 +1459,7 @@ static int build_mlx_header(struct mthca_dev *dev, struct mthca_sqp *sqp,
 {
         int header_size;
         int err;
+        u16 pkey;
 
         ib_ud_header_init(256, /* assume a MAD */
                           sqp->ud_header.grh_present,
@@ -1469,8 +1470,8 @@ static int build_mlx_header(struct mthca_dev *dev, struct mthca_sqp *sqp,
                 return err;
         mlx->flags &= ~cpu_to_be32(MTHCA_NEXT_SOLICIT | 1);
         mlx->flags |= cpu_to_be32((!sqp->qp.ibqp.qp_num ? MTHCA_MLX_VL15 : 0) |
-                                  (sqp->ud_header.lrh.destination_lid == 0xffff ?
-                                   MTHCA_MLX_SLR : 0) |
+                                  (sqp->ud_header.lrh.destination_lid ==
+                                   IB_LID_PERMISSIVE ? MTHCA_MLX_SLR : 0) |
                                   (sqp->ud_header.lrh.service_level << 8));
         mlx->rlid = sqp->ud_header.lrh.destination_lid;
         mlx->vcrc = 0;
@@ -1490,18 +1491,16 @@ static int build_mlx_header(struct mthca_dev *dev, struct mthca_sqp *sqp,
         }
 
         sqp->ud_header.lrh.virtual_lane    = !sqp->qp.ibqp.qp_num ? 15 : 0;
-        if (sqp->ud_header.lrh.destination_lid == 0xffff)
-                sqp->ud_header.lrh.source_lid = 0xffff;
+        if (sqp->ud_header.lrh.destination_lid == IB_LID_PERMISSIVE)
+                sqp->ud_header.lrh.source_lid = IB_LID_PERMISSIVE;
         sqp->ud_header.bth.solicited_event = !!(wr->send_flags & IB_SEND_SOLICITED);
         if (!sqp->qp.ibqp.qp_num)
                 ib_get_cached_pkey(&dev->ib_dev, sqp->port,
-                                   sqp->pkey_index,
-                                   &sqp->ud_header.bth.pkey);
+                                   sqp->pkey_index, &pkey);
         else
                 ib_get_cached_pkey(&dev->ib_dev, sqp->port,
-                                   wr->wr.ud.pkey_index,
-                                   &sqp->ud_header.bth.pkey);
-        cpu_to_be16s(&sqp->ud_header.bth.pkey);
+                                   wr->wr.ud.pkey_index, &pkey);
+        sqp->ud_header.bth.pkey = cpu_to_be16(pkey);
         sqp->ud_header.bth.destination_qpn = cpu_to_be32(wr->wr.ud.remote_qpn);
         sqp->ud_header.bth.psn = cpu_to_be32((sqp->send_psn++) & ((1 << 24) - 1));
         sqp->ud_header.deth.qkey = cpu_to_be32(wr->wr.ud.remote_qkey & 0x80000000 ?
@@ -1744,7 +1743,7 @@ int mthca_tavor_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 
 out:
         if (likely(nreq)) {
-                u32 doorbell[2];
+                __be32 doorbell[2];
 
                 doorbell[0] = cpu_to_be32(((qp->sq.next_ind << qp->sq.wqe_shift) +
                                            qp->send_wqe_offset) | f0 | op0);
@@ -1845,7 +1844,7 @@ int mthca_tavor_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 
 out:
         if (likely(nreq)) {
-                u32 doorbell[2];
+                __be32 doorbell[2];
 
                 doorbell[0] = cpu_to_be32((qp->rq.next_ind << qp->rq.wqe_shift) | size0);
                 doorbell[1] = cpu_to_be32((qp->qpn << 8) | nreq);
@@ -2066,7 +2065,7 @@ int mthca_arbel_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 
 out:
         if (likely(nreq)) {
-                u32 doorbell[2];
+                __be32 doorbell[2];
 
                 doorbell[0] = cpu_to_be32((nreq << 24)                  |
                                           ((qp->sq.head & 0xffff) << 8) |
@@ -2176,7 +2175,7 @@ out:
 }
 
 int mthca_free_err_wqe(struct mthca_dev *dev, struct mthca_qp *qp, int is_send,
-                       int index, int *dbd, u32 *new_wqe)
+                       int index, int *dbd, __be32 *new_wqe)
 {
         struct mthca_next_seg *next;