For systems that use CPU isolation (via nohz_full), creating or destroying
a socket with  timestamping (SOF_TIMESTAMPING_OPT_TX_SWHW) might cause a
static key to be enabled/disabled. This in turn causes undesired
IPIs to isolated CPUs.

So enable the static key unconditionally, if CPU isolation is enabled,
thus avoiding the IPIs.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

diff --git a/net/core/dev.c b/net/core/dev.c
index c588808be77f..15a32f5900e6 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -155,6 +155,7 @@
 #include <net/netdev_rx_queue.h>
 #include <net/page_pool/types.h>
 #include <net/page_pool/helpers.h>
+#include <linux/sched/isolation.h>
 
 #include "dev.h"
 #include "net-sysfs.h"
@@ -11851,3 +11852,14 @@ static int __init net_dev_init(void)
 }
 
 subsys_initcall(net_dev_init);
+
+static int __init net_dev_late_init(void)
+{
+	/* avoid static key IPIs to isolated CPUs */
+	if (housekeeping_enabled(HK_TYPE_MISC))
+		net_enable_timestamp();
+
+	return 0;
+}
+
+late_initcall(net_dev_late_init);

Hi,

On Thu, 2024-02-22 at 15:27 -0300, Marcelo Tosatti wrote:
> For systems that use CPU isolation (via nohz_full), creating or destroying
> a socket with  timestamping (SOF_TIMESTAMPING_OPT_TX_SWHW) might cause a
> static key to be enabled/disabled. This in turn causes undesired
> IPIs to isolated CPUs.

I believe Willem is correct, the relevant flag is
SOF_TIMESTAMPING_RX_SOFTWARE, see:

https://elixir.bootlin.com/linux/v6.8-rc5/source/net/core/sock.c#L938

the example you used is a sort of API misuse, placing flag where a bool
is expected. 

Please send a v2 with the updated commit message, you can mention both
SO_TIMESTAMPING (SOF_TIMESTAMPING_RX_SOFTWARE) and SO_TIMESTAMP.

Please also include Willem in recipients list, since he gave feedback
on previous version.

Thanks!

Paolo