From 8cf3f04f45694db0699f608c0e3fb550c607cc88 Mon Sep 17 00:00:00 2001 From: Ralph Campbell Date: Fri, 3 Feb 2006 14:28:48 -0800 Subject: IB/mad: Handle DR SMPs with a LID routed part Fix handling of directed route SMPs with a beginning or ending LID routed part. Signed-off-by: Ralph Campbell Signed-off-by: Hal Rosenstock Signed-off-by: Roland Dreier --- drivers/infiniband/core/mad.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c index d393b504bf26..c82f47a66e48 100644 --- a/drivers/infiniband/core/mad.c +++ b/drivers/infiniband/core/mad.c @@ -665,7 +665,15 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, struct ib_wc mad_wc; struct ib_send_wr *send_wr = &mad_send_wr->send_wr; - if (!smi_handle_dr_smp_send(smp, device->node_type, port_num)) { + /* + * Directed route handling starts if the initial LID routed part of + * a request or the ending LID routed part of a response is empty. + * If we are at the start of the LID routed part, don't update the + * hop_ptr or hop_cnt. See section 14.2.2, Vol 1 IB spec. + */ + if ((ib_get_smp_direction(smp) ? smp->dr_dlid : smp->dr_slid) == + IB_LID_PERMISSIVE && + !smi_handle_dr_smp_send(smp, device->node_type, port_num)) { ret = -EINVAL; printk(KERN_ERR PFX "Invalid directed route\n"); goto out; -- cgit v1.2.2 From 479a079663bd4c5f3d2714643b1b8c406aaba3e0 Mon Sep 17 00:00:00 2001 From: "Michael S. Tsirkin" Date: Tue, 7 Feb 2006 16:37:08 -0800 Subject: IPoIB: Don't start send-only joins while multicast thread is stopped Fix the following race scenario: - Device is up. - Port event or set mcast list triggers ipoib_mcast_stop_thread, this cancels the query and waits on mcast "done" completion. - Completion is called and "done" is set. - Meanwhile, ipoib_mcast_send arrives and starts a new query, re-initializing "done". Fix this by adding a "multicast started" bit and checking it before starting a send-only join. Signed-off-by: Michael S. Tsirkin Signed-off-by: Roland Dreier --- drivers/infiniband/ulp/ipoib/ipoib.h | 1 + drivers/infiniband/ulp/ipoib/ipoib_multicast.c | 15 +++++++++++++++ 2 files changed, 16 insertions(+) diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index e0a5412b7e68..2f85a9a831b1 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -78,6 +78,7 @@ enum { IPOIB_FLAG_SUBINTERFACE = 4, IPOIB_MCAST_RUN = 5, IPOIB_STOP_REAPER = 6, + IPOIB_MCAST_STARTED = 7, IPOIB_MAX_BACKOFF_SECONDS = 16, diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index ccaa0c387076..1c7148208ce8 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -601,6 +601,10 @@ int ipoib_mcast_start_thread(struct net_device *dev) queue_work(ipoib_workqueue, &priv->mcast_task); mutex_unlock(&mcast_mutex); + spin_lock_irq(&priv->lock); + set_bit(IPOIB_MCAST_STARTED, &priv->flags); + spin_unlock_irq(&priv->lock); + return 0; } @@ -611,6 +615,10 @@ int ipoib_mcast_stop_thread(struct net_device *dev, int flush) ipoib_dbg_mcast(priv, "stopping multicast thread\n"); + spin_lock_irq(&priv->lock); + clear_bit(IPOIB_MCAST_STARTED, &priv->flags); + spin_unlock_irq(&priv->lock); + mutex_lock(&mcast_mutex); clear_bit(IPOIB_MCAST_RUN, &priv->flags); cancel_delayed_work(&priv->mcast_task); @@ -693,6 +701,12 @@ void ipoib_mcast_send(struct net_device *dev, union ib_gid *mgid, */ spin_lock(&priv->lock); + if (!test_bit(IPOIB_MCAST_STARTED, &priv->flags)) { + ++priv->stats.tx_dropped; + dev_kfree_skb_any(skb); + goto unlock; + } + mcast = __ipoib_mcast_find(dev, mgid); if (!mcast) { /* Let's create a new send only group now */ @@ -754,6 +768,7 @@ out: ipoib_send(dev, skb, mcast->ah, IB_MULTICAST_QPN); } +unlock: spin_unlock(&priv->lock); } -- cgit v1.2.2 From 7bcb974ef6a0ae903888272c92c66ea779388c01 Mon Sep 17 00:00:00 2001 From: "Michael S. Tsirkin" Date: Tue, 7 Feb 2006 16:39:26 -0800 Subject: IPoIB: Fix another send-only join race Further, there's an additional issue that I saw in testing: ipoib_mcast_send may get called when priv->broadcast is NULL (e.g. if the device was downed and then upped internally because of a port event). If this happends and the send-only join request gets completed before priv->broadcast is set, we get an oops. Signed-off-by: Michael S. Tsirkin Signed-off-by: Roland Dreier --- drivers/infiniband/ulp/ipoib/ipoib_multicast.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index 1c7148208ce8..932bf139a65d 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -701,7 +701,7 @@ void ipoib_mcast_send(struct net_device *dev, union ib_gid *mgid, */ spin_lock(&priv->lock); - if (!test_bit(IPOIB_MCAST_STARTED, &priv->flags)) { + if (!test_bit(IPOIB_MCAST_STARTED, &priv->flags) || !priv->broadcast) { ++priv->stats.tx_dropped; dev_kfree_skb_any(skb); goto unlock; -- cgit v1.2.2 From f295c79b6766b25fe8c1aad88211c54d1caa7e0b Mon Sep 17 00:00:00 2001 From: Roland Dreier Date: Fri, 10 Feb 2006 18:02:44 -0800 Subject: IB/mthca: Don't print debugging info until we have all values When debugging is enabled, the mthca_QUERY_DEV_LIM() firmware command function prints out some of the device limits that it queries. However the debugging prints happen before all of the fields are extracted from the firmware response, so some of the values that get printed are uninitialized junk. Move the prints to the end of the function to fix this. Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mthca/mthca_cmd.c | 38 ++++++++++++++++----------------- 1 file changed, 19 insertions(+), 19 deletions(-) diff --git a/drivers/infiniband/hw/mthca/mthca_cmd.c b/drivers/infiniband/hw/mthca/mthca_cmd.c index f9b9b93dc501..2825615ce81c 100644 --- a/drivers/infiniband/hw/mthca/mthca_cmd.c +++ b/drivers/infiniband/hw/mthca/mthca_cmd.c @@ -1029,25 +1029,6 @@ int mthca_QUERY_DEV_LIM(struct mthca_dev *dev, MTHCA_GET(size, outbox, QUERY_DEV_LIM_UAR_ENTRY_SZ_OFFSET); dev_lim->uar_scratch_entry_sz = size; - mthca_dbg(dev, "Max QPs: %d, reserved QPs: %d, entry size: %d\n", - dev_lim->max_qps, dev_lim->reserved_qps, dev_lim->qpc_entry_sz); - mthca_dbg(dev, "Max SRQs: %d, reserved SRQs: %d, entry size: %d\n", - dev_lim->max_srqs, dev_lim->reserved_srqs, dev_lim->srq_entry_sz); - mthca_dbg(dev, "Max CQs: %d, reserved CQs: %d, entry size: %d\n", - dev_lim->max_cqs, dev_lim->reserved_cqs, dev_lim->cqc_entry_sz); - mthca_dbg(dev, "Max EQs: %d, reserved EQs: %d, entry size: %d\n", - dev_lim->max_eqs, dev_lim->reserved_eqs, dev_lim->eqc_entry_sz); - mthca_dbg(dev, "reserved MPTs: %d, reserved MTTs: %d\n", - dev_lim->reserved_mrws, dev_lim->reserved_mtts); - mthca_dbg(dev, "Max PDs: %d, reserved PDs: %d, reserved UARs: %d\n", - dev_lim->max_pds, dev_lim->reserved_pds, dev_lim->reserved_uars); - mthca_dbg(dev, "Max QP/MCG: %d, reserved MGMs: %d\n", - dev_lim->max_pds, dev_lim->reserved_mgms); - mthca_dbg(dev, "Max CQEs: %d, max WQEs: %d, max SRQ WQEs: %d\n", - dev_lim->max_cq_sz, dev_lim->max_qp_sz, dev_lim->max_srq_sz); - - mthca_dbg(dev, "Flags: %08x\n", dev_lim->flags); - if (mthca_is_memfree(dev)) { MTHCA_GET(field, outbox, QUERY_DEV_LIM_MAX_SRQ_SZ_OFFSET); dev_lim->max_srq_sz = 1 << field; @@ -1093,6 +1074,25 @@ int mthca_QUERY_DEV_LIM(struct mthca_dev *dev, dev_lim->mpt_entry_sz = MTHCA_MPT_ENTRY_SIZE; } + mthca_dbg(dev, "Max QPs: %d, reserved QPs: %d, entry size: %d\n", + dev_lim->max_qps, dev_lim->reserved_qps, dev_lim->qpc_entry_sz); + mthca_dbg(dev, "Max SRQs: %d, reserved SRQs: %d, entry size: %d\n", + dev_lim->max_srqs, dev_lim->reserved_srqs, dev_lim->srq_entry_sz); + mthca_dbg(dev, "Max CQs: %d, reserved CQs: %d, entry size: %d\n", + dev_lim->max_cqs, dev_lim->reserved_cqs, dev_lim->cqc_entry_sz); + mthca_dbg(dev, "Max EQs: %d, reserved EQs: %d, entry size: %d\n", + dev_lim->max_eqs, dev_lim->reserved_eqs, dev_lim->eqc_entry_sz); + mthca_dbg(dev, "reserved MPTs: %d, reserved MTTs: %d\n", + dev_lim->reserved_mrws, dev_lim->reserved_mtts); + mthca_dbg(dev, "Max PDs: %d, reserved PDs: %d, reserved UARs: %d\n", + dev_lim->max_pds, dev_lim->reserved_pds, dev_lim->reserved_uars); + mthca_dbg(dev, "Max QP/MCG: %d, reserved MGMs: %d\n", + dev_lim->max_pds, dev_lim->reserved_mgms); + mthca_dbg(dev, "Max CQEs: %d, max WQEs: %d, max SRQ WQEs: %d\n", + dev_lim->max_cq_sz, dev_lim->max_qp_sz, dev_lim->max_srq_sz); + + mthca_dbg(dev, "Flags: %08x\n", dev_lim->flags); + out: mthca_free_mailbox(dev, mailbox); return err; -- cgit v1.2.2 From 20b83382d1c5d4d1a73fc5671261db5239d1dbb3 Mon Sep 17 00:00:00 2001 From: Roland Dreier Date: Sat, 11 Feb 2006 12:22:12 -0800 Subject: IPoIB: Yet another fix for send-only joins Even after the last fix, it's still possible for a send-only join to start before the join for the broadcast group has finished. This could cause us to create a multicast group using attributes from the broadcast group that haven't been initialized yet, so we would use garbage for the Q_Key, etc. Fix this by waiting until the broadcast group's attached flag is set before starting send-only joins. Signed-off-by: Roland Dreier --- drivers/infiniband/ulp/ipoib/ipoib_multicast.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index 932bf139a65d..a2408d7ec598 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -533,8 +533,10 @@ void ipoib_mcast_join_task(void *dev_ptr) } if (!priv->broadcast) { - priv->broadcast = ipoib_mcast_alloc(dev, 1); - if (!priv->broadcast) { + struct ipoib_mcast *broadcast; + + broadcast = ipoib_mcast_alloc(dev, 1); + if (!broadcast) { ipoib_warn(priv, "failed to allocate broadcast group\n"); mutex_lock(&mcast_mutex); if (test_bit(IPOIB_MCAST_RUN, &priv->flags)) @@ -544,10 +546,11 @@ void ipoib_mcast_join_task(void *dev_ptr) return; } - memcpy(priv->broadcast->mcmember.mgid.raw, priv->dev->broadcast + 4, + spin_lock_irq(&priv->lock); + memcpy(broadcast->mcmember.mgid.raw, priv->dev->broadcast + 4, sizeof (union ib_gid)); + priv->broadcast = broadcast; - spin_lock_irq(&priv->lock); __ipoib_mcast_add(dev, priv->broadcast); spin_unlock_irq(&priv->lock); } @@ -701,7 +704,9 @@ void ipoib_mcast_send(struct net_device *dev, union ib_gid *mgid, */ spin_lock(&priv->lock); - if (!test_bit(IPOIB_MCAST_STARTED, &priv->flags) || !priv->broadcast) { + if (!test_bit(IPOIB_MCAST_STARTED, &priv->flags) || + !priv->broadcast || + !test_bit(IPOIB_MCAST_FLAG_ATTACHED, &priv->broadcast->flags)) { ++priv->stats.tx_dropped; dev_kfree_skb_any(skb); goto unlock; -- cgit v1.2.2 From 7d2babc4874825027848db04d11784070da4456d Mon Sep 17 00:00:00 2001 From: Roland Dreier Date: Mon, 13 Feb 2006 12:19:44 -0800 Subject: IB/mthca: bump driver version and release date Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mthca/mthca_dev.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/infiniband/hw/mthca/mthca_dev.h b/drivers/infiniband/hw/mthca/mthca_dev.h index 2a165fd06e57..e481037288d6 100644 --- a/drivers/infiniband/hw/mthca/mthca_dev.h +++ b/drivers/infiniband/hw/mthca/mthca_dev.h @@ -53,8 +53,8 @@ #define DRV_NAME "ib_mthca" #define PFX DRV_NAME ": " -#define DRV_VERSION "0.06" -#define DRV_RELDATE "June 23, 2005" +#define DRV_VERSION "0.07" +#define DRV_RELDATE "February 13, 2006" enum { MTHCA_FLAG_DDR_HIDDEN = 1 << 1, -- cgit v1.2.2