Linux RDMA and InfiniBand development
 help / color / mirror / Atom feed
From: Cheng Xu <chengyou@linux.alibaba.com>
To: jgg@ziepe.ca, leon@kernel.org
Cc: linux-rdma@vger.kernel.org, KaiShen@linux.alibaba.com
Subject: [PATCH for-rc v2] RDMA/erdma: Use IRQ-safe XArray helpers for QP and CQ tables
Date: Fri, 28 Aug 2026 11:03:44 +0800	[thread overview]
Message-ID: <20260828030344.88021-1-chengyou@linux.alibaba.com> (raw)

Locked QP and CQ lookups from EQ interrupts can deadlock with
create-path XArray updates. If an interrupt arrives while the create
path holds the plain xa_lock, the lookup spins forever trying to
acquire the same lock.

Use IRQ-safe XArray helpers for all QP and CQ create-path updates,
including the GSI QP store and error paths. Initialize both arrays with
XA_FLAGS_LOCK_IRQ so sleeping allocations preserve interrupt state.

Fixes: 98df2aee1459 ("RDMA/erdma: Hold CQ references when processing EQ events")
Fixes: a52eeff32024 ("RDMA/erdma: Hold QP references for AE and CM processing")
Signed-off-by: Cheng Xu <chengyou@linux.alibaba.com>
---
Changes from v1:
- Add XA_FLAGS_LOCK_IRQ to the QP and CQ XArrays to fix an issue identified
  by the Sashiko automated review.

 drivers/infiniband/hw/erdma/erdma_main.c  |  4 ++--
 drivers/infiniband/hw/erdma/erdma_verbs.c | 18 +++++++++---------
 2 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/drivers/infiniband/hw/erdma/erdma_main.c b/drivers/infiniband/hw/erdma/erdma_main.c
index 7e87a815e853..445182c6bc5d 100644
--- a/drivers/infiniband/hw/erdma/erdma_main.c
+++ b/drivers/infiniband/hw/erdma/erdma_main.c
@@ -572,8 +572,8 @@ static int erdma_ib_device_add(struct pci_dev *pdev)
 	INIT_LIST_HEAD(&dev->cep_list);
 
 	spin_lock_init(&dev->lock);
-	xa_init_flags(&dev->qp_xa, XA_FLAGS_ALLOC1);
-	xa_init_flags(&dev->cq_xa, XA_FLAGS_ALLOC1);
+	xa_init_flags(&dev->qp_xa, XA_FLAGS_ALLOC1 | XA_FLAGS_LOCK_IRQ);
+	xa_init_flags(&dev->cq_xa, XA_FLAGS_ALLOC1 | XA_FLAGS_LOCK_IRQ);
 	dev->next_alloc_cqn = 1;
 	dev->next_alloc_qpn = 1;
 
diff --git a/drivers/infiniband/hw/erdma/erdma_verbs.c b/drivers/infiniband/hw/erdma/erdma_verbs.c
index 65b1af1e6623..f18b88bba281 100644
--- a/drivers/infiniband/hw/erdma/erdma_verbs.c
+++ b/drivers/infiniband/hw/erdma/erdma_verbs.c
@@ -1021,15 +1021,15 @@ int erdma_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attrs,
 	init_completion(&qp->safe_free);
 
 	if (qp->ibqp.qp_type == IB_QPT_GSI) {
-		old_entry = xa_store(&dev->qp_xa, 1, qp, GFP_KERNEL);
+		old_entry = xa_store_irq(&dev->qp_xa, 1, qp, GFP_KERNEL);
 		if (xa_is_err(old_entry))
 			ret = xa_err(old_entry);
 		else
 			qp->ibqp.qp_num = 1;
 	} else {
-		ret = xa_alloc_cyclic(&dev->qp_xa, &qp->ibqp.qp_num, qp,
-				      XA_LIMIT(1, dev->attrs.max_qp - 1),
-				      &dev->next_alloc_qpn, GFP_KERNEL);
+		ret = xa_alloc_cyclic_irq(&dev->qp_xa, &qp->ibqp.qp_num, qp,
+					  XA_LIMIT(1, dev->attrs.max_qp - 1),
+					  &dev->next_alloc_qpn, GFP_KERNEL);
 	}
 
 	if (ret < 0) {
@@ -1089,7 +1089,7 @@ int erdma_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attrs,
 	else
 		free_kernel_qp(qp);
 err_out_xa:
-	xa_erase(&dev->qp_xa, QP_ID(qp));
+	xa_erase_irq(&dev->qp_xa, QP_ID(qp));
 err_out:
 	return ret;
 }
@@ -1993,9 +1993,9 @@ int erdma_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 	refcount_set(&cq->refcount, 1);
 	init_completion(&cq->free);
 
-	ret = xa_alloc_cyclic(&dev->cq_xa, &cq->cqn, cq,
-			      XA_LIMIT(1, dev->attrs.max_cq - 1),
-			      &dev->next_alloc_cqn, GFP_KERNEL);
+	ret = xa_alloc_cyclic_irq(&dev->cq_xa, &cq->cqn, cq,
+				  XA_LIMIT(1, dev->attrs.max_cq - 1),
+				  &dev->next_alloc_cqn, GFP_KERNEL);
 	if (ret < 0)
 		return ret;
 
@@ -2041,7 +2041,7 @@ int erdma_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
 	}
 
 err_out_xa:
-	xa_erase(&dev->cq_xa, cq->cqn);
+	xa_erase_irq(&dev->cq_xa, cq->cqn);
 
 	return ret;
 }
-- 
2.31.1

             reply	other threads:[~2026-08-28  3:03 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-08-28  3:03 Cheng Xu [this message]
2026-09-02 16:23 ` [PATCH for-rc v2] RDMA/erdma: Use IRQ-safe XArray helpers for QP and CQ tables Leon Romanovsky

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260828030344.88021-1-chengyou@linux.alibaba.com \
    --to=chengyou@linux.alibaba.com \
    --cc=KaiShen@linux.alibaba.com \
    --cc=jgg@ziepe.ca \
    --cc=leon@kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox