Matthieu Baerts

#21480 has been merged, this issue can probably be closed.

I just noticed my syzkaller instance found this issue once on the 22nd of June: ``` SELinux: unrecognized netlink message: protocol=4 nlmsg_type=19 sclass=netlink_tcpdiag_socket pid=3892 comm=syz.2.414 netlink: 'syz.3.423': attribute type 13...

It looks like there was another issue... ``` ------------[ cut here ]------------ WARNING: CPU: 1 PID: 156322 at net/mptcp/pm_kernel.c:881 __mark_subflow_endp_available net/mptcp/pm_kernel.c:881 [inline] WARNING: CPU: 1 PID: 156322 at net/mptcp/pm_kernel.c:881 __mark_subflow_endp_available+0x136/0x140...

Probably linked to #606 I suggest closing it then.

Hi @marco-a-itl, Thank you for having reported this bug. > These spurious events occur at every connection re-open timeout (which triggers a subflow closure event), until the entire MPTCP connection...

@marco-a-itl thank you for the additional info, and having tried on a more recent kernel. I tried to reproduce this in MPTCP selftests, but I didn't manage to. Here is...

> > Did I miss something? > > The first closure event is always normal, the following have the spurious event. I'm not sure, but I can't see in the...

OK, I managed to reproduce it: the initial subflow has to be removed after an error. I still need to send the patches (I don't know when I will be...

I forgot to add: on my side, I was also able to reproduce it quicker by forcing a reset at some points, and only allowing one 1 SYN retransmission before...

Hi @roman-shuhov, Thank you for this bug report! I had a very quick look (I will try to check in more details later on, but it might be in a...