[PATCH] IB/rdmavt: Fix concurrency panics in QP post_send and modify to error

[PATCH] IB/rdmavt: Fix concurrency panics in QP post_send and modify to error

[Michael J. Ruhl]

From: Michael J. Ruhl <michael.j.ruhl@intel.com>

commit d757c60eca9b22f4d108929a24401e0fdecda0b1 upstream

The RC/UC code path can go through a software loopback. In this code path
the receive side QP is manipulated.

If two threads are working on the QP receive side (i.e. post_send, and
modify_qp to an error state), QP information can be corrupted.

(post_send via loopback)
  set r_sge
  loop
     update r_sge
(modify_qp)
     take r_lock
     update r_sge <---- r_sge is now incorrect
(post_send)
     update r_sge <---- crash, etc.
     ...

This can lead to one of the two following crashes:

 BUG: unable to handle kernel NULL pointer dereference at (null)
  IP:  hfi1_copy_sge+0xf1/0x2e0 [hfi1]
  PGD 8000001fe6a57067 PUD 1fd9e0c067 PMD 0
 Call Trace:
  ruc_loopback+0x49b/0xbc0 [hfi1]
  hfi1_do_send+0x38e/0x3e0 [hfi1]
  _hfi1_do_send+0x1e/0x20 [hfi1]
  process_one_work+0x17f/0x440
  worker_thread+0x126/0x3c0
  kthread+0xd1/0xe0
  ret_from_fork_nospec_begin+0x21/0x21

or:

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000048
  IP:  rvt_clear_mr_refs+0x45/0x370 [rdmavt]
  PGD 80000006ae5eb067 PUD ef15d0067 PMD 0
 Call Trace:
  rvt_error_qp+0xaa/0x240 [rdmavt]
  rvt_modify_qp+0x47f/0xaa0 [rdmavt]
  ib_security_modify_qp+0x8f/0x400 [ib_core]
  ib_modify_qp_with_udata+0x44/0x70 [ib_core]
  modify_qp.isra.23+0x1eb/0x2b0 [ib_uverbs]
  ib_uverbs_modify_qp+0xaa/0xf0 [ib_uverbs]
  ib_uverbs_write+0x272/0x430 [ib_uverbs]
  vfs_write+0xc0/0x1f0
  SyS_write+0x7f/0xf0
  system_call_fastpath+0x1c/0x21

Fix by using the appropriate locking on the receiving QP.

Fixes: 83693bd14606 ("staging/rdma/hfi1: Use rdmavt version of post_send")
Cc: <stable@vger.kernel.org> #v4.9
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 drivers/infiniband/hw/hfi1/ruc.c |   33 +++++++++++++++++++++++----------
 1 file changed, 23 insertions(+), 10 deletions(-)

diff --git a/drivers/infiniband/hw/hfi1/ruc.c b/drivers/infiniband/hw/hfi1/ruc.c
index d151429..2723284 100644
--- a/drivers/infiniband/hw/hfi1/ruc.c
+++ b/drivers/infiniband/hw/hfi1/ruc.c
@@ -346,6 +346,18 @@ int hfi1_ruc_check_hdr(struct hfi1_ibport *ibp, struct ib_header *hdr,
 	return 1;
 }
 
+static enum ib_wc_status loopback_qp_drop(struct rvt_ibport *rvp,
+					  struct rvt_qp *sqp)
+{
+	rvp->n_pkt_drops++;
+	/*
+	 * For RC, the requester would timeout and retry so
+	 * shortcut the timeouts and just signal too many retries.
+	 */
+	return sqp->ibqp.qp_type == IB_QPT_RC ?
+		IB_WC_RETRY_EXC_ERR : IB_WC_SUCCESS;
+}
+
 /**
  * ruc_loopback - handle UC and RC loopback requests
  * @sqp: the sending QP
@@ -418,17 +430,14 @@ static void ruc_loopback(struct rvt_qp *sqp)
 	}
 	spin_unlock_irqrestore(&sqp->s_lock, flags);
 
-	if (!qp || !(ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) ||
+	if (!qp) {
+		send_status = loopback_qp_drop(&ibp->rvp, sqp);
+		goto serr_no_r_lock;
+	}
+	spin_lock_irqsave(&qp->r_lock, flags);
+	if (!(ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) ||
 	    qp->ibqp.qp_type != sqp->ibqp.qp_type) {
-		ibp->rvp.n_pkt_drops++;
-		/*
-		 * For RC, the requester would timeout and retry so
-		 * shortcut the timeouts and just signal too many retries.
-		 */
-		if (sqp->ibqp.qp_type == IB_QPT_RC)
-			send_status = IB_WC_RETRY_EXC_ERR;
-		else
-			send_status = IB_WC_SUCCESS;
+		send_status = loopback_qp_drop(&ibp->rvp, sqp);
 		goto serr;
 	}
 
@@ -601,6 +610,7 @@ static void ruc_loopback(struct rvt_qp *sqp)
 		     wqe->wr.send_flags & IB_SEND_SOLICITED);
 
 send_comp:
+	spin_unlock_irqrestore(&qp->r_lock, flags);
 	spin_lock_irqsave(&sqp->s_lock, flags);
 	ibp->rvp.n_loop_pkts++;
 flush_send:
@@ -627,6 +637,7 @@ static void ruc_loopback(struct rvt_qp *sqp)
 	}
 	if (sqp->s_rnr_retry_cnt < 7)
 		sqp->s_rnr_retry--;
+	spin_unlock_irqrestore(&qp->r_lock, flags);
 	spin_lock_irqsave(&sqp->s_lock, flags);
 	if (!(ib_rvt_state_ops[sqp->state] & RVT_PROCESS_RECV_OK))
 		goto clr_busy;
@@ -655,6 +666,8 @@ static void ruc_loopback(struct rvt_qp *sqp)
 	hfi1_rc_error(qp, wc.status);
 
 serr:
+	spin_unlock_irqrestore(&qp->r_lock, flags);
+serr_no_r_lock:
 	spin_lock_irqsave(&sqp->s_lock, flags);
 	hfi1_send_complete(sqp, wqe, send_status);
 	if (sqp->ibqp.qp_type == IB_QPT_RC) {