Tests cannot call g_error

[Doug Evans]

I stumbled on this writing a new test, using tests/qtest/e1000e-test.c as a template.

g_error() causes SIGTRAP, not SIGABRT, and thus the abort handler doesn't get run.
This in turn means qemu is not killed, which hangs the test because the tap-driver.pl script hangs waiting for more input.
There are a few tests that call g_error().

The SIGABRT handler explicitly kills qemu, e.g.:

qos-test.c:
    qtest_add_abrt_handler(kill_qemu_hook_func, s);

ref:
https://git.qemu.org/?p=qemu.git;a=blob;f=tests/qtest/libqtest.c;h=e49f3a1e45f4cd96279241fdb2bbe231029ab922;hb=HEAD#l272

But not unexpectedly there's no such handler for SIGTRAP.

Apply this patch to trigger a repro:

diff --git a/tests/qtest/e1000e-test.c b/tests/qtest/e1000e-test.c
index fc226fdfeb..e83ace1b5c 100644
--- a/tests/qtest/e1000e-test.c
+++ b/tests/qtest/e1000e-test.c
@@ -87,6 +87,9 @@ static void e1000e_send_verify(QE1000E *d, int *test_sockets, QGuestAllocator *a
     /* Wait for TX WB interrupt */
     e1000e_wait_isr(d, E1000E_TX0_MSG_ID);

+ g_message("Test g_error hang ...");
+ g_error("Pretend something timed out");
+
     /* Check DD bit */
     g_assert_cmphex(le32_to_cpu(descr.upper.data) & dsta_dd, ==, dsta_dd);

Then:

configure
make
make check-qtest-i386

check-qtest-i386 will take awhile. To repro faster:

$ grep qtest-i386/qos-test Makefile.mtest
.test.name.229 := qtest-i386/qos-test
$ make run-test-229
Running test qtest-i386/qos-test
** Message: 18:40:49.821: Test g_error hang ...

** (tests/qtest/qos-test:3820728): ERROR **: 18:40:49.821: Pretend something timed out
ERROR qtest-i386/qos-test - Bail out! FATAL-ERROR: Pretend something timed out

At this point things are hung because tap-driver.pl is still waiting for input because qemu is still running.

[Doug Evans (dje)]

An alternative is of course to allow g_error to be called.
One might restrict tests to not call it, but it might be impractical to impose that on all code that goes into a test.
One will need to find a way to get glib to not call G_BREAKPOINT for this case.

  if (debugger_present && breakpoint)
    G_BREAKPOINT ();
  else
    g_abort ();

https://gitlab.gnome.org/GNOME/glib/-/blob/master/glib/gmessages.c#L555

[Doug Evans (dje)]

I don't know QEMU that well yet, but the following question arises: Why can't QEMU be driven in a way that allows it to see that its controlling parent has died -> causing QEMU to terminate as well. That way the test doesn't need to care how it dies (e.g., we don't want a segfault to hang testing; and nor do we, I think, want to install signal handlers for every possible signal).

[Thomas Huth (th-huth)]

The QEMU project is currently moving its bug tracking to another system.
For this we need to know which bugs are still valid and which could be
closed already. Thus we are setting the bug state to "Incomplete" now.

If the bug has already been fixed in the latest upstream version of QEMU,
then please close this ticket as "Fix released".

If it is not fixed yet and you think that this bug report here is still
valid, then you have two options:

1) If you already have an account on gitlab.com, please open a new ticket
for this problem in our new tracker here:

    https://gitlab.com/qemu-project/qemu/-/issues

and then close this ticket here on Launchpad (or let it expire auto-
matically after 60 days). Please mention the URL of this bug ticket on
Launchpad in the new ticket on GitLab.

2) If you don't have an account on gitlab.com and don't intend to get
one, but still would like to keep this ticket opened, then please switch
the state back to "New" or "Confirmed" within the next 60 days (other-
wise it will get closed as "Expired"). We will then eventually migrate
the ticket automatically to the new system (but you won't be the reporter
of the bug in the new system and thus you won't get notified on changes
anymore).

Thank you and sorry for the inconvenience.

[Launchpad Janitor (janitor)]

[Expired for QEMU because there has been no activity for 60 days.]