selftests: drv-net: Limit the max number of queues in procfs_downup_hammer

For NICs with a large (1024+) number of queues, this test can cause
excessive memory fragmentation. This results in OOM errors, and in the
worst case driver/kernel crashes. We don't need to test with the max number
of queues, just enough to create a high likelihood of races between
reconfiguration and stats getting read.

Signed-off-by: Dimitri Daskalakis <dimitri.daskalakis1@gmail.com>
Link: https://patch.msgid.link/20251111225319.3019542-1-dimitri.daskalakis1@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This commit is contained in:
Dimitri Daskalakis
2025-11-11 14:53:19 -08:00
committed by Jakub Kicinski
parent e949824730
commit f766f8cdde

View File

@@ -263,14 +263,15 @@ def procfs_downup_hammer(cfg) -> None:
Reading stats via procfs only holds the RCU lock, drivers often try
to sleep when reading the stats, or don't protect against races.
"""
# Max out the queues, we'll flip between max and 1
# Set a large number of queues,
# we'll flip between min(max_queues, 64) and 1
channels = ethnl.channels_get({'header': {'dev-index': cfg.ifindex}})
if channels['combined-count'] == 0:
rx_type = 'rx'
else:
rx_type = 'combined'
cur_queue_cnt = channels[f'{rx_type}-count']
max_queue_cnt = channels[f'{rx_type}-max']
max_queue_cnt = min(channels[f'{rx_type}-max'], 64)
cmd(f"ethtool -L {cfg.ifname} {rx_type} {max_queue_cnt}")
defer(cmd, f"ethtool -L {cfg.ifname} {rx_type} {cur_queue_cnt}")