]> pilppa.org Git - linux-2.6-omap-h63xx.git/blobdiff - drivers/infiniband/hw/ipath/ipath_cq.c
Pull battery into release branch
[linux-2.6-omap-h63xx.git] / drivers / infiniband / hw / ipath / ipath_cq.c
index ea78e6dddc90c62b674eed5c2372feca43e71909..08d8ae148cd030ca99896d885e4d7a38af8804fd 100644 (file)
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2006 QLogic, Inc. All rights reserved.
+ * Copyright (c) 2006, 2007 QLogic Corporation. All rights reserved.
  * Copyright (c) 2005, 2006 PathScale, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
@@ -76,20 +76,25 @@ void ipath_cq_enter(struct ipath_cq *cq, struct ib_wc *entry, int solicited)
                }
                return;
        }
-       wc->queue[head].wr_id = entry->wr_id;
-       wc->queue[head].status = entry->status;
-       wc->queue[head].opcode = entry->opcode;
-       wc->queue[head].vendor_err = entry->vendor_err;
-       wc->queue[head].byte_len = entry->byte_len;
-       wc->queue[head].imm_data = (__u32 __force)entry->imm_data;
-       wc->queue[head].qp_num = entry->qp->qp_num;
-       wc->queue[head].src_qp = entry->src_qp;
-       wc->queue[head].wc_flags = entry->wc_flags;
-       wc->queue[head].pkey_index = entry->pkey_index;
-       wc->queue[head].slid = entry->slid;
-       wc->queue[head].sl = entry->sl;
-       wc->queue[head].dlid_path_bits = entry->dlid_path_bits;
-       wc->queue[head].port_num = entry->port_num;
+       if (cq->ip) {
+               wc->uqueue[head].wr_id = entry->wr_id;
+               wc->uqueue[head].status = entry->status;
+               wc->uqueue[head].opcode = entry->opcode;
+               wc->uqueue[head].vendor_err = entry->vendor_err;
+               wc->uqueue[head].byte_len = entry->byte_len;
+               wc->uqueue[head].imm_data = (__u32 __force)entry->imm_data;
+               wc->uqueue[head].qp_num = entry->qp->qp_num;
+               wc->uqueue[head].src_qp = entry->src_qp;
+               wc->uqueue[head].wc_flags = entry->wc_flags;
+               wc->uqueue[head].pkey_index = entry->pkey_index;
+               wc->uqueue[head].slid = entry->slid;
+               wc->uqueue[head].sl = entry->sl;
+               wc->uqueue[head].dlid_path_bits = entry->dlid_path_bits;
+               wc->uqueue[head].port_num = entry->port_num;
+               /* Make sure entry is written before the head index. */
+               smp_wmb();
+       } else
+               wc->kqueue[head] = *entry;
        wc->head = next;
 
        if (cq->notify == IB_CQ_NEXT_COMP ||
@@ -128,6 +133,12 @@ int ipath_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
        int npolled;
        u32 tail;
 
+       /* The kernel can only poll a kernel completion queue */
+       if (cq->ip) {
+               npolled = -EINVAL;
+               goto bail;
+       }
+
        spin_lock_irqsave(&cq->lock, flags);
 
        wc = cq->queue;
@@ -135,30 +146,10 @@ int ipath_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
        if (tail > (u32) cq->ibcq.cqe)
                tail = (u32) cq->ibcq.cqe;
        for (npolled = 0; npolled < num_entries; ++npolled, ++entry) {
-               struct ipath_qp *qp;
-
                if (tail == wc->head)
                        break;
-
-               qp = ipath_lookup_qpn(&to_idev(cq->ibcq.device)->qp_table,
-                                     wc->queue[tail].qp_num);
-               entry->qp = &qp->ibqp;
-               if (atomic_dec_and_test(&qp->refcount))
-                       wake_up(&qp->wait);
-
-               entry->wr_id = wc->queue[tail].wr_id;
-               entry->status = wc->queue[tail].status;
-               entry->opcode = wc->queue[tail].opcode;
-               entry->vendor_err = wc->queue[tail].vendor_err;
-               entry->byte_len = wc->queue[tail].byte_len;
-               entry->imm_data = wc->queue[tail].imm_data;
-               entry->src_qp = wc->queue[tail].src_qp;
-               entry->wc_flags = wc->queue[tail].wc_flags;
-               entry->pkey_index = wc->queue[tail].pkey_index;
-               entry->slid = wc->queue[tail].slid;
-               entry->sl = wc->queue[tail].sl;
-               entry->dlid_path_bits = wc->queue[tail].dlid_path_bits;
-               entry->port_num = wc->queue[tail].port_num;
+               /* The kernel doesn't need a RMB since it has the lock. */
+               *entry = wc->kqueue[tail];
                if (tail >= cq->ibcq.cqe)
                        tail = 0;
                else
@@ -168,6 +159,7 @@ int ipath_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
 
        spin_unlock_irqrestore(&cq->lock, flags);
 
+bail:
        return npolled;
 }
 
@@ -204,7 +196,7 @@ static void send_complete(unsigned long data)
  *
  * Called by ib_create_cq() in the generic verbs code.
  */
-struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
+struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries, int comp_vector,
                              struct ib_ucontext *context,
                              struct ib_udata *udata)
 {
@@ -212,6 +204,7 @@ struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
        struct ipath_cq *cq;
        struct ipath_cq_wc *wc;
        struct ib_cq *ret;
+       u32 sz;
 
        if (entries < 1 || entries > ib_ipath_max_cqes) {
                ret = ERR_PTR(-EINVAL);
@@ -232,7 +225,12 @@ struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
         * We need to use vmalloc() in order to support mmap and large
         * numbers of entries.
         */
-       wc = vmalloc_user(sizeof(*wc) + sizeof(struct ib_wc) * entries);
+       sz = sizeof(*wc);
+       if (udata && udata->outlen >= sizeof(__u64))
+               sz += sizeof(struct ib_uverbs_wc) * (entries + 1);
+       else
+               sz += sizeof(struct ib_wc) * (entries + 1);
+       wc = vmalloc_user(sz);
        if (!wc) {
                ret = ERR_PTR(-ENOMEM);
                goto bail_cq;
@@ -243,33 +241,20 @@ struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
         * See ipath_mmap() for details.
         */
        if (udata && udata->outlen >= sizeof(__u64)) {
-               struct ipath_mmap_info *ip;
-               __u64 offset = (__u64) wc;
                int err;
 
-               err = ib_copy_to_udata(udata, &offset, sizeof(offset));
-               if (err) {
-                       ret = ERR_PTR(err);
+               cq->ip = ipath_create_mmap_info(dev, sz, context, wc);
+               if (!cq->ip) {
+                       ret = ERR_PTR(-ENOMEM);
                        goto bail_wc;
                }
 
-               /* Allocate info for ipath_mmap(). */
-               ip = kmalloc(sizeof(*ip), GFP_KERNEL);
-               if (!ip) {
-                       ret = ERR_PTR(-ENOMEM);
-                       goto bail_wc;
+               err = ib_copy_to_udata(udata, &cq->ip->offset,
+                                      sizeof(cq->ip->offset));
+               if (err) {
+                       ret = ERR_PTR(err);
+                       goto bail_ip;
                }
-               cq->ip = ip;
-               ip->context = context;
-               ip->obj = wc;
-               kref_init(&ip->ref);
-               ip->mmap_cnt = 0;
-               ip->size = PAGE_ALIGN(sizeof(*wc) +
-                                     sizeof(struct ib_wc) * entries);
-               spin_lock_irq(&dev->pending_lock);
-               ip->next = dev->pending_mmaps;
-               dev->pending_mmaps = ip;
-               spin_unlock_irq(&dev->pending_lock);
        } else
                cq->ip = NULL;
 
@@ -277,12 +262,18 @@ struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
        if (dev->n_cqs_allocated == ib_ipath_max_cqs) {
                spin_unlock(&dev->n_cqs_lock);
                ret = ERR_PTR(-ENOMEM);
-               goto bail_wc;
+               goto bail_ip;
        }
 
        dev->n_cqs_allocated++;
        spin_unlock(&dev->n_cqs_lock);
 
+       if (cq->ip) {
+               spin_lock_irq(&dev->pending_lock);
+               list_add(&cq->ip->pending_mmaps, &dev->pending_mmaps);
+               spin_unlock_irq(&dev->pending_lock);
+       }
+
        /*
         * ib_create_cq() will initialize cq->ibcq except for cq->ibcq.cqe.
         * The number of entries should be >= the number requested or return
@@ -301,12 +292,12 @@ struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
 
        goto done;
 
+bail_ip:
+       kfree(cq->ip);
 bail_wc:
        vfree(wc);
-
 bail_cq:
        kfree(cq);
-
 done:
        return ret;
 }
@@ -340,17 +331,18 @@ int ipath_destroy_cq(struct ib_cq *ibcq)
 /**
  * ipath_req_notify_cq - change the notification type for a completion queue
  * @ibcq: the completion queue
- * @notify: the type of notification to request
+ * @notify_flags: the type of notification to request
  *
  * Returns 0 for success.
  *
  * This may be called from interrupt context.  Also called by
  * ib_req_notify_cq() in the generic verbs code.
  */
-int ipath_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify notify)
+int ipath_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags notify_flags)
 {
        struct ipath_cq *cq = to_icq(ibcq);
        unsigned long flags;
+       int ret = 0;
 
        spin_lock_irqsave(&cq->lock, flags);
        /*
@@ -358,9 +350,15 @@ int ipath_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify notify)
         * any other transitions (see C11-31 and C11-32 in ch. 11.4.2.2).
         */
        if (cq->notify != IB_CQ_NEXT_COMP)
-               cq->notify = notify;
+               cq->notify = notify_flags & IB_CQ_SOLICITED_MASK;
+
+       if ((notify_flags & IB_CQ_REPORT_MISSED_EVENTS) &&
+           cq->queue->head != cq->queue->tail)
+               ret = 1;
+
        spin_unlock_irqrestore(&cq->lock, flags);
-       return 0;
+
+       return ret;
 }
 
 /**
@@ -376,6 +374,7 @@ int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
        struct ipath_cq_wc *wc;
        u32 head, tail, n;
        int ret;
+       u32 sz;
 
        if (cqe < 1 || cqe > ib_ipath_max_cqes) {
                ret = -EINVAL;
@@ -385,7 +384,12 @@ int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
        /*
         * Need to use vmalloc() if we want to support large #s of entries.
         */
-       wc = vmalloc_user(sizeof(*wc) + sizeof(struct ib_wc) * cqe);
+       sz = sizeof(*wc);
+       if (udata && udata->outlen >= sizeof(__u64))
+               sz += sizeof(struct ib_uverbs_wc) * (cqe + 1);
+       else
+               sz += sizeof(struct ib_wc) * (cqe + 1);
+       wc = vmalloc_user(sz);
        if (!wc) {
                ret = -ENOMEM;
                goto bail;
@@ -400,7 +404,7 @@ int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
 
                ret = ib_copy_to_udata(udata, &offset, sizeof(offset));
                if (ret)
-                       goto bail;
+                       goto bail_free;
        }
 
        spin_lock_irq(&cq->lock);
@@ -420,13 +424,14 @@ int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
        else
                n = head - tail;
        if (unlikely((u32)cqe < n)) {
-               spin_unlock_irq(&cq->lock);
-               vfree(wc);
                ret = -EOVERFLOW;
-               goto bail;
+               goto bail_unlock;
        }
        for (n = 0; tail != head; n++) {
-               wc->queue[n] = old_wc->queue[tail];
+               if (cq->ip)
+                       wc->uqueue[n] = old_wc->uqueue[tail];
+               else
+                       wc->kqueue[n] = old_wc->kqueue[tail];
                if (tail == (u32) cq->ibcq.cqe)
                        tail = 0;
                else
@@ -444,17 +449,20 @@ int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
                struct ipath_ibdev *dev = to_idev(ibcq->device);
                struct ipath_mmap_info *ip = cq->ip;
 
-               ip->obj = wc;
-               ip->size = PAGE_ALIGN(sizeof(*wc) +
-                                     sizeof(struct ib_wc) * cqe);
+               ipath_update_mmap_info(dev, ip, sz, wc);
                spin_lock_irq(&dev->pending_lock);
-               ip->next = dev->pending_mmaps;
-               dev->pending_mmaps = ip;
+               if (list_empty(&ip->pending_mmaps))
+                       list_add(&ip->pending_mmaps, &dev->pending_mmaps);
                spin_unlock_irq(&dev->pending_lock);
        }
 
        ret = 0;
+       goto bail;
 
+bail_unlock:
+       spin_unlock_irq(&cq->lock);
+bail_free:
+       vfree(wc);
 bail:
        return ret;
 }