[DPDK/ethdev Bug 1829] net/mlx5:mbuf leak when start device failed
bugzilla at dpdk.org
bugzilla at dpdk.org
Fri Nov 14 04:48:28 CET 2025
http://bugs.dpdk.org/show_bug.cgi?id=1829
Bug ID: 1829
Summary: net/mlx5:mbuf leak when start device failed
Product: DPDK
Version: 24.11
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
Priority: Normal
Component: ethdev
Assignee: dev at dpdk.org
Reporter: zhang_zx0130 at 163.com
Target Milestone: ---
An error 'CQ creation failure' occurred when starting the device to setup rxq,
and causing mbuf leakage.
(1)We suspect that the CQ creation failure error is due to a compatibility
issue between the firmware and the driver.
(2)but the mbuf leak problem does exist, in the mlx5_rxq_start function,
priv->obj_ops.rxq_obj_new(rxq) returns an error because of the CQ error. and
goto release rxq(mlx5_rxq_release), since the values if rxq->rq_ci and
rxq->rq_pi have not been properly initialized, it causes an inability to
release the mbufs(rxq_free_elts_sprq: used=1024) .
log:
mlx5_net: port 1 starting device
mlx5_net: port 1 Rx queues number update: 1 -> 1
mlx5_net: port 1 Tx queue 0 allocated and configured 1024 WRs
mlx5_net: Port 1 txq 0 updated with 0x50000c338fe8.
mlx5_net: Port 1 max work queue size is 32768.
mlx5_net: Port 1 dev_cap.max_sge is 30.
mlx5_net: port 1 SPRQ queue 0 allocated and configured 1024 segments (max 1024
packets)
mlx5_net: Port 1 rxq 0 updated with 0x50000c342cd8.
mlx5_net: Port 1 Port 1 Rx queue 0 CQ creation failure.
mlx5_net: Port 1 Port 1 Rx queue 0 freeing 1024 WRs
mlx5_net: Port 1 Port 1 Rx queue allocation failed: Cannot allocate memory
mlx5_net: Port 1 Port 1 Rx queue 0 freeing WRs
DPDK: 24.11.2
driver in kernel: 24.10-2.1.8
firmware: 14.25.1020 (MT_2420110034)
#ethtool -i eth0
driver: mlx5_core
version: 24.10-2.1.8
firmware-version: 14.25.1020 (MT_2420110034)
expansion-rom-version:
bus-info: 0000:02:00.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: no
supports-register-dump: no
supports-priv-flags: yes
--
You are receiving this mail because:
You are the assignee for the bug.
More information about the dev
mailing list