On Tue, 28 Jan 2025 16:04:35 +0100 Eric Dumazet <edumazet(a)google.com> wrote:This so-called 'deadlock' only occurs if a remote TCP stack is unable to send win0 probes. In this case, sending some ACK will not help reliably if these ACK get lost. I find the description tries very hard to hide a bug in another stack, for some reason.Side note: that was fixed meanwhile. Back then, at a first analysis, we thought it was a workaround, but it's the actual fix as we *must* send zero-window probes in that situation, and this commit triggers them: https://passt.top/passt/commit/?id=a740e16fd1b9bdca8d259aa6d37f942a3874425c -- Stefano