Greetings kernel wizards, To summarize from the commit messages, the attached patches are the result of our recent debugging sessions, where we were hunting sudden TCP network throughput drops, where the throughput dropped from tens of Gbps to a few Mbps for an undeterministic amount of time. The first patch adds a new counter for netstat, so users can notice that the sockets are marked as under pressure. The second patch adds a new tracepoint so it is easier to pinpoint which exact cgroup is having difficulties. Individual commit messages should contain more detailed reasoning. Thanks! Daniel Changes: v1 -> v2: Add tracepoint Link: https://lore.kernel.org/netdev/20250707105205.222558-1-daniel.sedlak@xxxxxxxxx/ Daniel Sedlak (1): tcp: account for memory pressure signaled by cgroup Matyas Hurtik (1): mm/vmpressure: add tracepoint for socket pressure detection .../networking/net_cachelines/snmp.rst | 1 + include/net/tcp.h | 14 ++++++----- include/trace/events/memcg.h | 25 +++++++++++++++++++ include/uapi/linux/snmp.h | 1 + mm/vmpressure.c | 3 +++ net/ipv4/proc.c | 1 + 6 files changed, 39 insertions(+), 6 deletions(-) cc: linux-trace-kernel@xxxxxxxxxxxxxxx cc: linux-doc@xxxxxxxxxxxxxxx cc: Steven Rostedt <rostedt@xxxxxxxxxxx> cc: Masami Hiramatsu <mhiramat@xxxxxxxxxx> cc: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx> base-commit: e96ee511c906c59b7c4e6efd9d9b33917730e000 -- 2.39.5
