similar to: [libnbd PATCH v3 00/19] pass LISTEN_FDNAMES with systemd socket activation

V4 was here (incorrectly versioned on the mailing list as v3): <http://mid.mail-archive.com/20230323121016.1442655-1-lersek at redhat.com>. See the Notes section on each patch for the v5 updates. Laszlo Ersek (2): socket activation: generalize environment construction socket activation: set LISTEN_FDNAMES Richard W.M. Jones (2): common/include: Copy ascii-ctype functions from nbdkit

v1 was posted here: https://www.redhat.com/archives/libguestfs/2019-September/thread.html#00337 v2: - Drop the first patch. - Hopefully fix the multiple issues with fork-safety and general behaviour on error paths. Note this requires execvpe for which there seems to be no equivalent on FreeBSD, except some kind of tedious path parsing (but can we assign to environ?) Rich.

[attempting to loop in systemd folks; this started in libnbd at https://listman.redhat.com/archives/libguestfs/2023-March/031178.html - although I may have to retry since I'm not a usual subscriber of systemd-devel] On Fri, Mar 24, 2023 at 11:32:26AM +0100, Laszlo Ersek wrote: > >> @@ -245,6 +245,9 @@ CONNECT_SA.START: > >> "LISTEN_PID=",

On 9/30/19 11:32 AM, Richard W.M. Jones wrote: > This adds new APIs for running a local NBD server and connecting to it > using systemd socket activation (instead of stdin/stdout). > > This includes interop tests against nbdkit and qemu-nbd which I > believe are the only NBD servers supporting socket activation. (If we > find others then we can add more interop tests in

On Thu, Mar 23, 2023 at 01:10:16PM +0100, Laszlo Ersek wrote: > When the user calls nbd_set_socket_activation_name before calling > nbd_connect_system_socket_activation, pass the name down to the server > through LISTEN_FDNAMES. This has no effect unless the new API has > been called to set the socket name to a non-empty string. > >

On 10/1/19 9:11 AM, Richard W.M. Jones wrote: > On Tue, Oct 01, 2019 at 08:24:33AM -0500, Eric Blake wrote: >>> +#else /* !HAVE_EXECVPE */ >>> + SET_NEXT_STATE (%.DEAD) >>> + set_error (ENOTSUP, "platform does not support socket activation"); >>> + return 0; >>> +#endif >> >> We probably ought to add a matching

On 3/23/23 20:27, Eric Blake wrote: > On Thu, Mar 23, 2023 at 01:10:16PM +0100, Laszlo Ersek wrote: >> When the user calls nbd_set_socket_activation_name before calling >> nbd_connect_system_socket_activation, pass the name down to the server >> through LISTEN_FDNAMES. This has no effect unless the new API has >> been called to set the socket name to a non-empty string.

Per POSIX, execvp() is not safe to call in a child process forked from a multi-threaded process. We can now replace the execvp() call in the child process with a call to our fork-safe (async-signal-safe) variant. Prepare our internal execvpe context on the parent's construction path, use the context in the child, and release the context in the parent on the way out, regardless of whether the

This is an alternative approach to https://listman.redhat.com/archives/libguestfs/2023-January/030535.html After discussing this with Dan Berrange we came to the conclusion that you really might want to set LISTEN_FDNAMES to arbitrary short strings (or not set it). Especially when talking to qemu-storage-daemon which would allow you to use these names on the command line. Rich.

If the user calls nbd_kill_subprocess, we shouldn't kill the process again when we close the handle (since the process has likely gone and we might be killing a different process). --- lib/handle.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/lib/handle.c b/lib/handle.c index 2af25fe..5ad818e 100644 --- a/lib/handle.c +++ b/lib/handle.c @@ -315,6 +315,8 @@

For commands with long names and/or short descriptors, you can end up going over 72 characters in the first line of the man page (causing podwrapper to complain). Wrap these lines. --- generator/generator | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/generator/generator b/generator/generator index 7d3f656..ad1cb6b 100755 --- a/generator/generator +++ b/generator/generator

systemd allows sockets passed through socket activation to be named with the protocol they require. We only ever pass one socket, name it. This environment variable is currently ignored by qemu-nbd and nbdkit, but might be used by qemu-storage-daemon: https://lists.nongnu.org/archive/html/qemu-devel/2023-01/msg06114.html --- generator/states-connect-socket-activation.c | 41

This adds coalescing of adjacent extents of the same type, as mentioned by Eric Blake in the commit message here: https://github.com/libguestfs/libnbd/commit/46072f6611f80245846a445766da071e457b00cd The patch series is rather long because it detours through adding the <vector.h> library from nbdkit into libnbd and replacing ad hoc uses of realloc, char ** etc in various places. Rich.

Correctly report "asprintf" in the error message when asprintf() fails. Signed-off-by: Laszlo Ersek <lersek at redhat.com> --- generator/states-connect-socket-activation.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/generator/states-connect-socket-activation.c b/generator/states-connect-socket-activation.c index 3b621b8be44f..c46a0bf5c0a3 100644 ---

On 1/28/23 13:47, Richard W.M. Jones wrote: > systemd allows sockets passed through socket activation to be named > with the protocol they require. We only ever pass one socket, name > it. This environment variable is currently ignored by qemu-nbd and > nbdkit, but might be used by qemu-storage-daemon: > > https://lists.nongnu.org/archive/html/qemu-devel/2023-01/msg06114.html

F_SETFD takes an "int", so it stands to reason that FD_CLOEXEC expands to an "int". In turn, it's bad hygiene to manipulate the sign bit of (signed) integers with bit operations: ~FD_CLOEXEC Convert FD_CLOEXEC to "unsigned int" for the bitwise complement operator: ~(unsigned)FD_CLOEXEC The bitwise complement then results in an "unsigned int".

A prior patch highlighted the following leak: when any construction step fails in the parent after fork(), the child process is leaked. We could plug the leak by inserting a new error handling section for killing the child process and reaping it with waitpid(). However, it would raise some new questions (what signal to send, ensure the child not ignore that signal, hope that whatever process

This is version 4 of the following sub-series: [libnbd PATCH v3 06/29] lib/utils: introduce xwrite() as a more robust write() [libnbd PATCH v3 07/29] lib/utils: add async-signal-safe assert() [libnbd PATCH v3 08/29] lib/utils: add unit test for async-signal-safe assert() http://mid.mail-archive.com/20230215141158.2426855-7-lersek at redhat.com

This isn't quite finished because not all of the tests or examples have been updated, but it demonstrates an idea: Should we forget about the concept of having multiple connections managed under a single handle? In this patch there is a single ‘struct nbd_handle *’ which manages a single state machine and connection (and therefore no nbd_connection). To connect to a multi-conn server you must

This is a series of patches to libnbd and nbdkit adding AF_VSOCK support. On the host side it allows you to start an nbdkit instance which listens on a virtio-vsock socket: $ ./nbdkit -fv --vsock memory 1G ... nbdkit: debug: bound to vsock 2:10809 On the guest side you can then use libnbd to connect to the server: $ ./run nbdsh -c 'h.connect_vsock(2, 10809)' -c