Re: [PATCH 8/8] Allow pasta to take a command to execute

On Mon, Aug 29, 2022 at 09:16:58PM +0200, Stefano Brivio wrote:

On Fri, 26 Aug 2022 14:58:39 +1000 David Gibson wrote:

...

When not given an existing PID or network namspace to attach to, pasta spawns a shell. Most commands which can spawn a shell in an altered environment can also run other commands in that same environment, which can be useful in automation.

Allow pasta to do the same thing; it can be given an arbitrary command to run in the network and user namespace which pasta creates. If neither a command nor an existing PID or netns to attach to is given, continue to spawn a default shell, as before.

Signed-off-by: David Gibson --- conf.c | 27 ++++++++++++++++++--------- passt.1 | 14 +++++++++----- pasta.c | 33 +++++++++++++++++++++++---------- pasta.h | 2 +- 4 files changed, 51 insertions(+), 25 deletions(-)

diff --git a/conf.c b/conf.c index 2a18124..162c2dd 100644 --- a/conf.c +++ b/conf.c @@ -550,7 +550,8 @@ static int conf_ns_pid(char *userns, char *netns, const char *arg) return 0; }

- return -EINVAL; + /* Not a PID, later code will treat as a command */ + return 0; }

/** @@ -1480,14 +1481,18 @@ void conf(struct ctx *c, int argc, char **argv)

check_root(c);

- if (c->mode == MODE_PASTA && optind + 1 == argc) { - ret = conf_ns_pid(userns, netns, argv[optind]); - if (ret < 0) + if (c->mode == MODE_PASTA) { + if (*netns && optind != argc) { + err("Both --netns and PID or command given"); usage(argv[0]); - } else if (c->mode == MODE_PASTA && *userns - && !*netns && optind == argc) { - err("--userns requires --netns or PID"); - usage(argv[0]); + } else if (optind + 1 == argc) { + ret = conf_ns_pid(userns, netns, argv[optind]); + if (ret < 0) + usage(argv[0]); + } else if (*userns && !*netns && optind == argc) { + err("--userns requires --netns or PID"); + usage(argv[0]); + } } else if (optind != argc) { usage(argv[0]); }

[...]

I haven't really looked into this yet, but I guess we should now handle getopts return codes a bit differently, because this works:

$ ./pasta --config-net -- sh -c 'sleep 1; ip li sh' 1: lo: mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 2: enp9s0: mtu 65520 qdisc pfifo_fast state UNKNOWN mode DEFAULT group default qlen 1000 link/ether aa:3e:39:5f:c6:15 brd ff:ff:ff:ff:ff:ff

while this doesn't:

$ ./pasta --config-net sh -c 'sleep 1; ip li sh' ./pasta: invalid option -- 'c' [...]

despite the fact that there's no ambiguity.

You mean because pasta itself doesn't have a -c option? Attempting to account for that sounds like a bad idea. Requiring -- when the command given has options of its own that aren't supposed to go to the wrapper is pretty common for these sorts of tools. Basically the trade-off is that you either need to require that, or you have to require that all non-option arguments of the wrapper come last (old style POSIXish command line parsing, as opposed to GNUish conventions). The latter is usually more awkward than the former. -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson