The O_NONBLOCK behaviour is unrelated to this bug and a separate regression, it's a 9p specific quirk that can get fixed separately.
We don't need it here as the read should also stop when a read request to the server returned 0, which it will once the end of file is reached regardless of O_NONBLOCK. The O_NONBLOCK quirk can save a round-trip here when there is data to read (the read request will bail out immediately and not after the second zero read), but in practice it makes little difference for such regular files, it's really aimed at synthetic servers which "stream" data through a regular file, so a single read can block until there is data and when that data comes the client can get it immediately... It's not even non-blocking behaviour, really ugly and makes tar very slow; so I'm hoping we can get away not reintroducing that quirk but I'll deal with it when we get there. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2056461 Title: autopkgtest-virt-qemu on noble images sometimes hangs doing copydown To manage notifications about this bug go to: https://bugs.launchpad.net/linux/+bug/2056461/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs