On Wed, 13 Mar 2024 11:23:56 +0100 Paul Holzinger wrote:

Thanks for the quick patch.

On 12/03/2024 23:34, Stefano Brivio wrote:

...

Paul reports that, with commit 15001b39ef1d ("conf: set the log level much earlier"), early messages aren't reported to standard error anymore.

The reason is that, once the log mask is changed from LOG_EARLY, we don't force logging to stderr, and this mechanism was abused to have early errors on stderr. Now that we drop LOG_EARLY earlier on, this doesn't work anymore.

Call __openlog() as soon as we know the mode we're running as, using LOG_PERROR. Then, once we detach, if we're not running from an interactive terminal and logging to standard error is not forced, drop LOG_PERROR from the options.

While at it, make sure we don't print messages to standard error reporting that we couldn't log to the system logger, if we didn't open a connection yet. That's expected.

Reported-by: Paul Holzinger Fixes: 15001b39ef1d ("conf: set the log level much earlier") Signed-off-by: Stefano Brivio

Tested-by: Paul Holzinger

And thanks for the quick test.

...

--- log.c | 2 +- passt.c | 14 +++++++------- 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/log.c b/log.c index eafaca2..bdd31b4 100644 --- a/log.c +++ b/log.c @@ -174,7 +174,7 @@ void passt_vsyslog(int pri, const char *format, va_list ap) if (log_opt & LOG_PERROR) fprintf(stderr, "%s", buf + prefix_len);

- if (send(log_sock, buf, n, 0) != n) + if (log_sock >= 0 && send(log_sock, buf, n, 0) != n) fprintf(stderr, "Failed to send %i bytes to syslog\n", n); }

diff --git a/passt.c b/passt.c index f430648..e4cb11e 100644 --- a/passt.c +++ b/passt.c @@ -225,6 +225,8 @@ int main(int argc, char **argv) strncpy(argv0, argv[0], PATH_MAX - 1); name = basename(argv0); if (strstr(name, "pasta")) { + __openlog(log_name = "pasta", LOG_PERROR, LOG_DAEMON); + sa.sa_handler = pasta_child_handler; if (sigaction(SIGCHLD, &sa, NULL)) { die("Couldn't install signal handlers: %s", @@ -237,18 +239,16 @@ int main(int argc, char **argv) }

c.mode = MODE_PASTA; - log_name = "pasta"; } else if (strstr(name, "passt")) { + __openlog(log_name = "passt", LOG_PERROR, LOG_DAEMON); + c.mode = MODE_PASST; - log_name = "passt"; } else { exit(EXIT_FAILURE); }

madvise(pkt_buf, TAP_BUF_BYTES, MADV_HUGEPAGE);

- __openlog(log_name, 0, LOG_DAEMON); - c.epollfd = epoll_create1(EPOLL_CLOEXEC); if (c.epollfd == -1) { perror("epoll_create1"); @@ -269,9 +269,6 @@ int main(int argc, char **argv) conf(&c, argc, argv); trace_init(c.trace);

- if (c.force_stderr || isatty(fileno(stdout))) - __openlog(log_name, LOG_PERROR, LOG_DAEMON); - pasta_netns_quit_init(&c);

tap_sock_init(&c); @@ -314,6 +311,9 @@ int main(int argc, char **argv) if (isolate_prefork(&c)) die("Failed to sandbox process, exiting");

+ if (!c.force_stderr && !isatty(fileno(stdout)))

I still have the feeling this should check the stderr fd instead of stdout as it is logging to stderr. But I guess it shouldn't make a big difference in real situations so not a big deal.

Now that you mention this, I'm quite undecided: true, we're checking if we want to print to standard error, but what we want to check here is if we're running from an interactive terminal, for which testing if stdout is a terminal is quite idiomatic. On the other hand: $ cat isatty.c #include #include int main() { return isatty(fileno(stdout)); } $ gcc -o isatty isatty.c $ ./isatty >/dev/null; echo $? 0 $ ./isatty 2>/dev/null; echo $? 1 so we would (fail to) print to stderr if it's redirected to /dev/null, but not print to stderr if stdout is redirected. At this point I would have a slight preference toward changing it like you suggested. -- Stefano