6 files changed, 109 insertions, 48 deletions

diff --git a/net/rds/Makefile b/net/rds/Makefile
index b46eca109688..56d3f6023ced 100644
--- a/net/rds/Makefile
+++ b/net/rds/Makefile
@@ -4,7 +4,7 @@ rds-y :=	af_rds.o bind.o cong.o connection.o info.o message.o   \
                         loop.o page.o rdma.o
 
 obj-$(CONFIG_RDS_RDMA) += rds_rdma.o
-rds_rdma-objs :=        rdma_transport.o \
+rds_rdma-y :=   rdma_transport.o \
                         ib.o ib_cm.o ib_recv.o ib_ring.o ib_send.o ib_stats.o \
                         ib_sysctl.o ib_rdma.o \
                         iw.o iw_cm.o iw_recv.o iw_ring.o iw_send.o iw_stats.o \
@@ -12,10 +12,8 @@ rds_rdma-objs :=	rdma_transport.o \
 
 
 obj-$(CONFIG_RDS_TCP) += rds_tcp.o
-rds_tcp-objs :=         tcp.o tcp_connect.o tcp_listen.o tcp_recv.o \
+rds_tcp-y :=            tcp.o tcp_connect.o tcp_listen.o tcp_recv.o \
                         tcp_send.o tcp_stats.o
 
-ifeq ($(CONFIG_RDS_DEBUG), y)
-EXTRA_CFLAGS += -DDEBUG
-endif
+ccflags-$(CONFIG_RDS_DEBUG)     :=      -DDEBUG
 
diff --git a/net/rds/loop.c b/net/rds/loop.c
index c390156b426f..aeec1d483b17 100644
--- a/net/rds/loop.c
+++ b/net/rds/loop.c
@@ -134,8 +134,12 @@ static int rds_loop_conn_alloc(struct rds_connection *conn, gfp_t gfp)
 static void rds_loop_conn_free(void *arg)
 {
         struct rds_loop_connection *lc = arg;
+        unsigned long flags;
+
         rdsdebug("lc %p\n", lc);
+        spin_lock_irqsave(&loop_conns_lock, flags);
         list_del(&lc->loop_node);
+        spin_unlock_irqrestore(&loop_conns_lock, flags);
         kfree(lc);
 }
 
diff --git a/net/rds/message.c b/net/rds/message.c
index a84545dae370..1fd3d29023d7 100644
--- a/net/rds/message.c
+++ b/net/rds/message.c
@@ -224,6 +224,9 @@ struct scatterlist *rds_message_alloc_sgs(struct rds_message *rm, int nents)
         WARN_ON(rm->m_used_sgs + nents > rm->m_total_sgs);
         WARN_ON(!nents);
 
+        if (rm->m_used_sgs + nents > rm->m_total_sgs)
+                return NULL;
+
         sg_ret = &sg_first[rm->m_used_sgs];
         sg_init_table(sg_ret, nents);
         rm->m_used_sgs += nents;
@@ -246,6 +249,10 @@ struct rds_message *rds_message_map_pages(unsigned long *page_addrs, unsigned in
         rm->m_inc.i_hdr.h_len = cpu_to_be32(total_len);
         rm->data.op_nents = ceil(total_len, PAGE_SIZE);
         rm->data.op_sg = rds_message_alloc_sgs(rm, num_sgs);
+        if (!rm->data.op_sg) {
+                rds_message_put(rm);
+                return ERR_PTR(-ENOMEM);
+        }
 
         for (i = 0; i < rm->data.op_nents; ++i) {
                 sg_set_page(&rm->data.op_sg[i],
diff --git a/net/rds/rdma.c b/net/rds/rdma.c
index 1a41debca1ce..4e37c1cbe8b2 100644
--- a/net/rds/rdma.c
+++ b/net/rds/rdma.c
@@ -479,13 +479,38 @@ void rds_atomic_free_op(struct rm_atomic_op *ao)
 
 
 /*
- * Count the number of pages needed to describe an incoming iovec.
+ * Count the number of pages needed to describe an incoming iovec array.
  */
-static int rds_rdma_pages(struct rds_rdma_args *args)
+static int rds_rdma_pages(struct rds_iovec iov[], int nr_iovecs)
+{
+        int tot_pages = 0;
+        unsigned int nr_pages;
+        unsigned int i;
+
+        /* figure out the number of pages in the vector */
+        for (i = 0; i < nr_iovecs; i++) {
+                nr_pages = rds_pages_in_vec(&iov[i]);
+                if (nr_pages == 0)
+                        return -EINVAL;
+
+                tot_pages += nr_pages;
+
+                /*
+                 * nr_pages for one entry is limited to (UINT_MAX>>PAGE_SHIFT)+1,
+                 * so tot_pages cannot overflow without first going negative.
+                 */
+                if (tot_pages < 0)
+                        return -EINVAL;
+        }
+
+        return tot_pages;
+}
+
+int rds_rdma_extra_size(struct rds_rdma_args *args)
 {
         struct rds_iovec vec;
         struct rds_iovec __user *local_vec;
-        unsigned int tot_pages = 0;
+        int tot_pages = 0;
         unsigned int nr_pages;
         unsigned int i;
 
@@ -502,14 +527,16 @@ static int rds_rdma_pages(struct rds_rdma_args *args)
                         return -EINVAL;
 
                 tot_pages += nr_pages;
-        }
 
-        return tot_pages;
-}
+                /*
+                 * nr_pages for one entry is limited to (UINT_MAX>>PAGE_SHIFT)+1,
+                 * so tot_pages cannot overflow without first going negative.
+                 */
+                if (tot_pages < 0)
+                        return -EINVAL;
+        }
 
-int rds_rdma_extra_size(struct rds_rdma_args *args)
-{
-        return rds_rdma_pages(args) * sizeof(struct scatterlist);
+        return tot_pages * sizeof(struct scatterlist);
 }
 
 /*
@@ -520,13 +547,12 @@ int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
                           struct cmsghdr *cmsg)
 {
         struct rds_rdma_args *args;
-        struct rds_iovec vec;
         struct rm_rdma_op *op = &rm->rdma;
         int nr_pages;
         unsigned int nr_bytes;
         struct page **pages = NULL;
-        struct rds_iovec __user *local_vec;
-        unsigned int nr;
+        struct rds_iovec iovstack[UIO_FASTIOV], *iovs = iovstack;
+        int iov_size;
         unsigned int i, j;
         int ret = 0;
 
@@ -541,14 +567,31 @@ int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
                 goto out;
         }
 
-        if (args->nr_local > (u64)UINT_MAX) {
+        if (args->nr_local > UIO_MAXIOV) {
                 ret = -EMSGSIZE;
                 goto out;
         }
 
-        nr_pages = rds_rdma_pages(args);
-        if (nr_pages < 0)
+        /* Check whether to allocate the iovec area */
+        iov_size = args->nr_local * sizeof(struct rds_iovec);
+        if (args->nr_local > UIO_FASTIOV) {
+                iovs = sock_kmalloc(rds_rs_to_sk(rs), iov_size, GFP_KERNEL);
+                if (!iovs) {
+                        ret = -ENOMEM;
+                        goto out;
+                }
+        }
+
+        if (copy_from_user(iovs, (struct rds_iovec __user *)(unsigned long) args->local_vec_addr, iov_size)) {
+                ret = -EFAULT;
+                goto out;
+        }
+
+        nr_pages = rds_rdma_pages(iovs, args->nr_local);
+        if (nr_pages < 0) {
+                ret = -EINVAL;
                 goto out;
+        }
 
         pages = kcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL);
         if (!pages) {
@@ -564,6 +607,10 @@ int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
         op->op_recverr = rs->rs_recverr;
         WARN_ON(!nr_pages);
         op->op_sg = rds_message_alloc_sgs(rm, nr_pages);
+        if (!op->op_sg) {
+                ret = -ENOMEM;
+                goto out;
+        }
 
         if (op->op_notify || op->op_recverr) {
                 /* We allocate an uninitialized notifier here, because
@@ -597,50 +644,40 @@ int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
                (unsigned long long)args->remote_vec.addr,
                op->op_rkey);
 
-        local_vec = (struct rds_iovec __user *)(unsigned long) args->local_vec_addr;
-
         for (i = 0; i < args->nr_local; i++) {
-                if (copy_from_user(&vec, &local_vec[i],
-                                   sizeof(struct rds_iovec))) {
-                        ret = -EFAULT;
-                        goto out;
-                }
-
-                nr = rds_pages_in_vec(&vec);
-                if (nr == 0) {
-                        ret = -EINVAL;
-                        goto out;
-                }
+                struct rds_iovec *iov = &iovs[i];
+                /* don't need to check, rds_rdma_pages() verified nr will be +nonzero */
+                unsigned int nr = rds_pages_in_vec(iov);
 
-                rs->rs_user_addr = vec.addr;
-                rs->rs_user_bytes = vec.bytes;
+                rs->rs_user_addr = iov->addr;
+                rs->rs_user_bytes = iov->bytes;
 
                 /* If it's a WRITE operation, we want to pin the pages for reading.
                  * If it's a READ operation, we need to pin the pages for writing.
                  */
-                ret = rds_pin_pages(vec.addr, nr, pages, !op->op_write);
+                ret = rds_pin_pages(iov->addr, nr, pages, !op->op_write);
                 if (ret < 0)
                         goto out;
 
-                rdsdebug("RDS: nr_bytes %u nr %u vec.bytes %llu vec.addr %llx\n",
-                       nr_bytes, nr, vec.bytes, vec.addr);
+                rdsdebug("RDS: nr_bytes %u nr %u iov->bytes %llu iov->addr %llx\n",
+                         nr_bytes, nr, iov->bytes, iov->addr);
 
-                nr_bytes += vec.bytes;
+                nr_bytes += iov->bytes;
 
                 for (j = 0; j < nr; j++) {
-                        unsigned int offset = vec.addr & ~PAGE_MASK;
+                        unsigned int offset = iov->addr & ~PAGE_MASK;
                         struct scatterlist *sg;
 
                         sg = &op->op_sg[op->op_nents + j];
                         sg_set_page(sg, pages[j],
-                                        min_t(unsigned int, vec.bytes, PAGE_SIZE - offset),
+                                        min_t(unsigned int, iov->bytes, PAGE_SIZE - offset),
                                         offset);
 
-                        rdsdebug("RDS: sg->offset %x sg->len %x vec.addr %llx vec.bytes %llu\n",
-                               sg->offset, sg->length, vec.addr, vec.bytes);
+                        rdsdebug("RDS: sg->offset %x sg->len %x iov->addr %llx iov->bytes %llu\n",
+                               sg->offset, sg->length, iov->addr, iov->bytes);
 
-                        vec.addr += sg->length;
-                        vec.bytes -= sg->length;
+                        iov->addr += sg->length;
+                        iov->bytes -= sg->length;
                 }
 
                 op->op_nents += nr;
@@ -655,13 +692,14 @@ int rds_cmsg_rdma_args(struct rds_sock *rs, struct rds_message *rm,
         }
         op->op_bytes = nr_bytes;
 
-        ret = 0;
 out:
+        if (iovs != iovstack)
+                sock_kfree_s(rds_rs_to_sk(rs), iovs, iov_size);
         kfree(pages);
         if (ret)
                 rds_rdma_free_op(op);
-
-        rds_stats_inc(s_send_rdma);
+        else
+                rds_stats_inc(s_send_rdma);
 
         return ret;
 }
@@ -773,6 +811,10 @@ int rds_cmsg_atomic(struct rds_sock *rs, struct rds_message *rm,
         rm->atomic.op_active = 1;
         rm->atomic.op_recverr = rs->rs_recverr;
         rm->atomic.op_sg = rds_message_alloc_sgs(rm, 1);
+        if (!rm->atomic.op_sg) {
+                ret = -ENOMEM;
+                goto err;
+        }
 
         /* verify 8 byte-aligned */
         if (args->local_addr & 0x7) {
diff --git a/net/rds/send.c b/net/rds/send.c
index 0bc9db17a87d..35b9c2e9caf1 100644
--- a/net/rds/send.c
+++ b/net/rds/send.c
@@ -973,6 +973,10 @@ int rds_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
         /* Attach data to the rm */
         if (payload_len) {
                 rm->data.op_sg = rds_message_alloc_sgs(rm, ceil(payload_len, PAGE_SIZE));
+                if (!rm->data.op_sg) {
+                        ret = -ENOMEM;
+                        goto out;
+                }
                 ret = rds_message_copy_from_user(rm, msg->msg_iov, payload_len);
                 if (ret)
                         goto out;
diff --git a/net/rds/tcp.c b/net/rds/tcp.c
index 08a8c6cf2d10..8e0a32001c90 100644
--- a/net/rds/tcp.c
+++ b/net/rds/tcp.c
@@ -221,7 +221,13 @@ static int rds_tcp_conn_alloc(struct rds_connection *conn, gfp_t gfp)
 static void rds_tcp_conn_free(void *arg)
 {
         struct rds_tcp_connection *tc = arg;
+        unsigned long flags;
         rdsdebug("freeing tc %p\n", tc);
+
+        spin_lock_irqsave(&rds_tcp_conn_lock, flags);
+        list_del(&tc->t_tcp_node);
+        spin_unlock_irqrestore(&rds_tcp_conn_lock, flags);
+
         kmem_cache_free(rds_tcp_conn_slab, tc);
 }