Re: [Libguestfs] [p2v PATCH 09/10] rely on Linux for killing nbdkit, when the parent thread exits

2022-03-31 Thread Richard W.M. Jones
On Thu, Mar 31, 2022 at 03:35:55PM +0200, Laszlo Ersek wrote: > And yes, that's the crux: if we need to keep the PID tracking / killing > / reaping (and I agree that we do!), then what does "--exit-with-parent" > buy us? I don't think it improves anything. So if I can't remove the PID > tracking

Re: [Libguestfs] [p2v PATCH 09/10] rely on Linux for killing nbdkit, when the parent thread exits

2022-03-31 Thread Laszlo Ersek
On 03/31/22 15:11, Richard W.M. Jones wrote: > So this patch unfortunately isn't correct for a few reasons: > > (1) Non-Linux/BSD OSes don't support a concept like > --exit-with-parent. nbdkit will exit with an error if you try to use > this feature on such OSes. You can test if

Re: [Libguestfs] [p2v PATCH 09/10] rely on Linux for killing nbdkit, when the parent thread exits

2022-03-31 Thread Richard W.M. Jones
So this patch unfortunately isn't correct for a few reasons: (1) Non-Linux/BSD OSes don't support a concept like --exit-with-parent. nbdkit will exit with an error if you try to use this feature on such OSes. You can test if --exit-with-parent is supported before trying to use it by: nbdkit

[Libguestfs] [p2v PATCH 09/10] rely on Linux for killing nbdkit, when the parent thread exits

2022-03-31 Thread Laszlo Ersek
We currently track the PIDs of the NBD servers (nbdkit only, at this point) so that we can kill and reap them in cleanup_data_conns(). cleanup_data_conns() is called from three kinds of contexts: (1) in start_conversion(), followed immediately by exit (EXIT_FAILURE); (2) at the end of