Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma fixes from Jason Gunthorpe:

 - syzkaller found a WARN_ON in rxe due to poor lifecycle management of
   resources linked to skbs

 - Missing error path handling in erdma qp creation

 - Initialize the qp number for the GSI QP in erdma

 - Mismatching of DIP, SCC and QP numbers in hns

 - SRQ bug fixes in bnxt_re

 - Memory leak and possibly uninited memory in bnxt_re

 - Remove retired irdma maintainer

 - Fix kfree() for kvalloc() in ODP

 - Fix memory leak in hns

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  RDMA/hns: Fix dip entries leak on devices newer than hip09
  RDMA/core: Free pfn_list with appropriate kvfree call
  MAINTAINERS: Remove bouncing irdma maintainer
  RDMA/bnxt_re: Fix to initialize the PBL array
  RDMA/bnxt_re: Fix a possible memory leak in the driver
  RDMA/bnxt_re: Fix to remove workload check in SRQ limit path
  RDMA/bnxt_re: Fix to do SRQ armena by default
  RDMA/hns: Fix querying wrong SCC context for DIP algorithm
  RDMA/erdma: Fix unset QPN of GSI QP
  RDMA/erdma: Fix ignored return value of init_kernel_qp
  RDMA/rxe: Flush delayed SKBs while releasing RXE resources
This commit is contained in:
Linus Torvalds
2025-08-22 09:13:24 -04:00
12 changed files with 55 additions and 67 deletions

View File

@@ -12281,7 +12281,6 @@ F: include/linux/avf/virtchnl.h
F: include/linux/net/intel/*/
INTEL ETHERNET PROTOCOL DRIVER FOR RDMA
M: Mustafa Ismail <mustafa.ismail@intel.com>
M: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
L: linux-rdma@vger.kernel.org
S: Supported

View File

@@ -115,7 +115,7 @@ static int ib_init_umem_odp(struct ib_umem_odp *umem_odp,
out_free_map:
if (ib_uses_virt_dma(dev))
kfree(map->pfn_list);
kvfree(map->pfn_list);
else
hmm_dma_map_free(dev->dma_device, map);
return ret;
@@ -287,7 +287,7 @@ static void ib_umem_odp_free(struct ib_umem_odp *umem_odp)
mutex_unlock(&umem_odp->umem_mutex);
mmu_interval_notifier_remove(&umem_odp->notifier);
if (ib_uses_virt_dma(dev))
kfree(umem_odp->map.pfn_list);
kvfree(umem_odp->map.pfn_list);
else
hmm_dma_map_free(dev->dma_device, &umem_odp->map);
}

View File

@@ -1921,7 +1921,6 @@ int bnxt_re_modify_srq(struct ib_srq *ib_srq, struct ib_srq_attr *srq_attr,
struct bnxt_re_srq *srq = container_of(ib_srq, struct bnxt_re_srq,
ib_srq);
struct bnxt_re_dev *rdev = srq->rdev;
int rc;
switch (srq_attr_mask) {
case IB_SRQ_MAX_WR:
@@ -1933,11 +1932,8 @@ int bnxt_re_modify_srq(struct ib_srq *ib_srq, struct ib_srq_attr *srq_attr,
return -EINVAL;
srq->qplib_srq.threshold = srq_attr->srq_limit;
rc = bnxt_qplib_modify_srq(&rdev->qplib_res, &srq->qplib_srq);
if (rc) {
ibdev_err(&rdev->ibdev, "Modify HW SRQ failed!");
return rc;
}
bnxt_qplib_srq_arm_db(&srq->qplib_srq.dbinfo, srq->qplib_srq.threshold);
/* On success, update the shadow */
srq->srq_limit = srq_attr->srq_limit;
/* No need to Build and send response back to udata */

View File

@@ -2017,6 +2017,28 @@ static void bnxt_re_free_nqr_mem(struct bnxt_re_dev *rdev)
rdev->nqr = NULL;
}
/* When DEL_GID fails, driver is not freeing GID ctx memory.
* To avoid the memory leak, free the memory during unload
*/
static void bnxt_re_free_gid_ctx(struct bnxt_re_dev *rdev)
{
struct bnxt_qplib_sgid_tbl *sgid_tbl = &rdev->qplib_res.sgid_tbl;
struct bnxt_re_gid_ctx *ctx, **ctx_tbl;
int i;
if (!sgid_tbl->active)
return;
ctx_tbl = sgid_tbl->ctx;
for (i = 0; i < sgid_tbl->max; i++) {
if (sgid_tbl->hw_id[i] == 0xFFFF)
continue;
ctx = ctx_tbl[i];
kfree(ctx);
}
}
static void bnxt_re_dev_uninit(struct bnxt_re_dev *rdev, u8 op_type)
{
u8 type;
@@ -2030,6 +2052,7 @@ static void bnxt_re_dev_uninit(struct bnxt_re_dev *rdev, u8 op_type)
if (test_and_clear_bit(BNXT_RE_FLAG_QOS_WORK_REG, &rdev->flags))
cancel_delayed_work_sync(&rdev->worker);
bnxt_re_free_gid_ctx(rdev);
if (test_and_clear_bit(BNXT_RE_FLAG_RESOURCES_INITIALIZED,
&rdev->flags))
bnxt_re_cleanup_res(rdev);

View File

@@ -705,9 +705,7 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res,
srq->dbinfo.db = srq->dpi->dbr;
srq->dbinfo.max_slot = 1;
srq->dbinfo.priv_db = res->dpi_tbl.priv_db;
if (srq->threshold)
bnxt_qplib_armen_db(&srq->dbinfo, DBC_DBC_TYPE_SRQ_ARMENA);
srq->arm_req = false;
bnxt_qplib_armen_db(&srq->dbinfo, DBC_DBC_TYPE_SRQ_ARMENA);
return 0;
fail:
@@ -717,24 +715,6 @@ int bnxt_qplib_create_srq(struct bnxt_qplib_res *res,
return rc;
}
int bnxt_qplib_modify_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq)
{
struct bnxt_qplib_hwq *srq_hwq = &srq->hwq;
u32 count;
count = __bnxt_qplib_get_avail(srq_hwq);
if (count > srq->threshold) {
srq->arm_req = false;
bnxt_qplib_srq_arm_db(&srq->dbinfo, srq->threshold);
} else {
/* Deferred arming */
srq->arm_req = true;
}
return 0;
}
int bnxt_qplib_query_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq)
{
@@ -776,7 +756,6 @@ int bnxt_qplib_post_srq_recv(struct bnxt_qplib_srq *srq,
struct bnxt_qplib_hwq *srq_hwq = &srq->hwq;
struct rq_wqe *srqe;
struct sq_sge *hw_sge;
u32 count = 0;
int i, next;
spin_lock(&srq_hwq->lock);
@@ -808,15 +787,8 @@ int bnxt_qplib_post_srq_recv(struct bnxt_qplib_srq *srq,
bnxt_qplib_hwq_incr_prod(&srq->dbinfo, srq_hwq, srq->dbinfo.max_slot);
spin_lock(&srq_hwq->lock);
count = __bnxt_qplib_get_avail(srq_hwq);
spin_unlock(&srq_hwq->lock);
/* Ring DB */
bnxt_qplib_ring_prod_db(&srq->dbinfo, DBC_DBC_TYPE_SRQ);
if (srq->arm_req == true && count > srq->threshold) {
srq->arm_req = false;
bnxt_qplib_srq_arm_db(&srq->dbinfo, srq->threshold);
}
return 0;
}

View File

@@ -546,8 +546,6 @@ int bnxt_qplib_enable_nq(struct pci_dev *pdev, struct bnxt_qplib_nq *nq,
srqn_handler_t srq_handler);
int bnxt_qplib_create_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq);
int bnxt_qplib_modify_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq);
int bnxt_qplib_query_srq(struct bnxt_qplib_res *res,
struct bnxt_qplib_srq *srq);
void bnxt_qplib_destroy_srq(struct bnxt_qplib_res *res,

View File

@@ -121,6 +121,7 @@ static int __alloc_pbl(struct bnxt_qplib_res *res,
pbl->pg_arr = vmalloc_array(pages, sizeof(void *));
if (!pbl->pg_arr)
return -ENOMEM;
memset(pbl->pg_arr, 0, pages * sizeof(void *));
pbl->pg_map_arr = vmalloc_array(pages, sizeof(dma_addr_t));
if (!pbl->pg_map_arr) {
@@ -128,6 +129,7 @@ static int __alloc_pbl(struct bnxt_qplib_res *res,
pbl->pg_arr = NULL;
return -ENOMEM;
}
memset(pbl->pg_map_arr, 0, pages * sizeof(dma_addr_t));
pbl->pg_count = 0;
pbl->pg_size = sginfo->pgsize;

View File

@@ -994,6 +994,8 @@ int erdma_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attrs,
old_entry = xa_store(&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),
@@ -1031,7 +1033,9 @@ int erdma_create_qp(struct ib_qp *ibqp, struct ib_qp_init_attr *attrs,
if (ret)
goto err_out_cmd;
} else {
init_kernel_qp(dev, qp, attrs);
ret = init_kernel_qp(dev, qp, attrs);
if (ret)
goto err_out_xa;
}
qp->attrs.max_send_sge = attrs->cap.max_send_sge;

View File

@@ -3043,7 +3043,7 @@ static void hns_roce_v2_exit(struct hns_roce_dev *hr_dev)
if (!hr_dev->is_vf)
hns_roce_free_link_table(hr_dev);
if (hr_dev->pci_dev->revision == PCI_REVISION_ID_HIP09)
if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09)
free_dip_entry(hr_dev);
}
@@ -5476,7 +5476,7 @@ static int hns_roce_v2_query_srqc(struct hns_roce_dev *hr_dev, u32 srqn,
return ret;
}
static int hns_roce_v2_query_sccc(struct hns_roce_dev *hr_dev, u32 qpn,
static int hns_roce_v2_query_sccc(struct hns_roce_dev *hr_dev, u32 sccn,
void *buffer)
{
struct hns_roce_v2_scc_context *context;
@@ -5488,7 +5488,7 @@ static int hns_roce_v2_query_sccc(struct hns_roce_dev *hr_dev, u32 qpn,
return PTR_ERR(mailbox);
ret = hns_roce_cmd_mbox(hr_dev, 0, mailbox->dma, HNS_ROCE_CMD_QUERY_SCCC,
qpn);
sccn);
if (ret)
goto out;

View File

@@ -100,6 +100,7 @@ int hns_roce_fill_res_qp_entry_raw(struct sk_buff *msg, struct ib_qp *ib_qp)
struct hns_roce_v2_qp_context qpc;
struct hns_roce_v2_scc_context sccc;
} context = {};
u32 sccn = hr_qp->qpn;
int ret;
if (!hr_dev->hw->query_qpc)
@@ -116,7 +117,13 @@ int hns_roce_fill_res_qp_entry_raw(struct sk_buff *msg, struct ib_qp *ib_qp)
!hr_dev->hw->query_sccc)
goto out;
ret = hr_dev->hw->query_sccc(hr_dev, hr_qp->qpn, &context.sccc);
if (hr_qp->cong_type == CONG_TYPE_DIP) {
if (!hr_qp->dip)
goto out;
sccn = hr_qp->dip->dip_idx;
}
ret = hr_dev->hw->query_sccc(hr_dev, sccn, &context.sccc);
if (ret)
ibdev_warn_ratelimited(&hr_dev->ib_dev,
"failed to query SCCC, ret = %d.\n",

View File

@@ -345,33 +345,15 @@ int rxe_prepare(struct rxe_av *av, struct rxe_pkt_info *pkt,
static void rxe_skb_tx_dtor(struct sk_buff *skb)
{
struct net_device *ndev = skb->dev;
struct rxe_dev *rxe;
unsigned int qp_index;
struct rxe_qp *qp;
struct rxe_qp *qp = skb->sk->sk_user_data;
int skb_out;
rxe = rxe_get_dev_from_net(ndev);
if (!rxe && is_vlan_dev(ndev))
rxe = rxe_get_dev_from_net(vlan_dev_real_dev(ndev));
if (WARN_ON(!rxe))
return;
qp_index = (int)(uintptr_t)skb->sk->sk_user_data;
if (!qp_index)
return;
qp = rxe_pool_get_index(&rxe->qp_pool, qp_index);
if (!qp)
goto put_dev;
skb_out = atomic_dec_return(&qp->skb_out);
if (qp->need_req_skb && skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW)
if (unlikely(qp->need_req_skb &&
skb_out < RXE_INFLIGHT_SKBS_PER_QP_LOW))
rxe_sched_task(&qp->send_task);
rxe_put(qp);
put_dev:
ib_device_put(&rxe->ib_dev);
sock_put(skb->sk);
}
@@ -383,6 +365,7 @@ static int rxe_send(struct sk_buff *skb, struct rxe_pkt_info *pkt)
sock_hold(sk);
skb->sk = sk;
skb->destructor = rxe_skb_tx_dtor;
rxe_get(pkt->qp);
atomic_inc(&pkt->qp->skb_out);
if (skb->protocol == htons(ETH_P_IP))
@@ -405,6 +388,7 @@ static int rxe_loopback(struct sk_buff *skb, struct rxe_pkt_info *pkt)
sock_hold(sk);
skb->sk = sk;
skb->destructor = rxe_skb_tx_dtor;
rxe_get(pkt->qp);
atomic_inc(&pkt->qp->skb_out);
if (skb->protocol == htons(ETH_P_IP))
@@ -497,6 +481,9 @@ struct sk_buff *rxe_init_packet(struct rxe_dev *rxe, struct rxe_av *av,
goto out;
}
/* Add time stamp to skb. */
skb->tstamp = ktime_get();
skb_reserve(skb, hdr_len + LL_RESERVED_SPACE(ndev));
/* FIXME: hold reference to this netdev until life of this skb. */

View File

@@ -244,7 +244,7 @@ static int rxe_qp_init_req(struct rxe_dev *rxe, struct rxe_qp *qp,
err = sock_create_kern(&init_net, AF_INET, SOCK_DGRAM, 0, &qp->sk);
if (err < 0)
return err;
qp->sk->sk->sk_user_data = (void *)(uintptr_t)qp->elem.index;
qp->sk->sk->sk_user_data = qp;
/* pick a source UDP port number for this QP based on
* the source QPN. this spreads traffic for different QPs