Re: [Qemu-devel] [PULL v3 00/28] ivshmem deprecation, qtests, typedefs and gnu99

[Philippe Mathieu-Daudé]

On Fri, Jan 18, 2019 at 2:15 PM Philippe Mathieu-Daudé
<address@hidden> wrote:
> On 1/18/19 12:47 PM, Peter Maydell wrote:
> > On Thu, 17 Jan 2019 at 13:38, Thomas Huth <address@hidden> wrote:
> >>
> >>  Hi Peter!
> >>
> >> The following changes since commit 
> >> 6f2f34177a25bffd6fd92a05e6e66c8d22d97094:
> >>
> >>   Merge remote-tracking branch 'remotes/thibault/tags/samuel-thibault' 
> >> into staging (2019-01-15 18:32:57 +0000)
> >>
> >> are available in the git repository at:
> >>
> >>   https://gitlab.com/huth/qemu.git tags/pull-request-2019-01-17
> >>
> >> for you to fetch changes up to 3f0832c85380321ef697d64342b389648f8abb1c:
> >>
> >>   tests/hexloader-test: Don't pass -nographic to the QEMU under test 
> >> (2019-01-17 14:25:12 +0100)
> >>
> >> ----------------------------------------------------------------
> >> - Remove deprecated "ivshmem" legacy device
> >> - Bug fix for vhost-user-test
> >> - Use more CONFIG Makefile switches for qtests
> >> - Get rid of global_qtests in some more qtests
> >> - typedef cleanups
> >> - Fixes for compiling with Clang
> >> - Force C standard to gnu99
> >> ----------------------------------------------------------------
> >
> > In the netbsd VM, the ipmi-bt-test is consistently failing:
> >
> > MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))}
> > QTEST_QEMU_BINARY=i386-softmmu
> > /qemu-system-i386 QTEST_QEMU_IMG=qemu-img tests/ipmi-bt-test -m=quick
> > -k --tap < /dev/null | ./s
> > cripts/tap-driver.pl --test-name="ipmi-bt-test"
> > **
> > ERROR:tests/ipmi-bt-test.c:319:test_connect: assertion failed: (rv == 1)
> > ERROR - too few tests run (expected 4, got 0)
> > [1]   Abort trap (core dumped) MALLOC_PERTURB_=... |
> >       Done(1)                 ./scripts/tap-dr...
> > /var/tmp/qemu-test.7opHvo/tests/Makefile.include:856: recipe for
> > target 'check-qtest-i386' faile
> > d
> > gmake: *** [check-qtest-i386] Error 1
> > gmake: *** Waiting for unfinished jobs....
> >
> >
> > This has been kind of intermittent for a while, I've seen
> > it on-and-off, but with this pullreq it's failed four times
> > in a row, so maybe something in here is making it less
> > intermittent.
>
> This file does:
>
>     qtest_add_func("/ipmi/extern/connect", test_connect);
>     qtest_add_func("/ipmi/extern/bt_base", test_bt_base);
>     qtest_add_func("/ipmi/extern/bt_enable_irq", test_enable_irq);
>     qtest_add_func("/ipmi/extern/bt_base_irq", test_bt_base);
>     ret = g_test_run();
>
> The first "test" test_connect() binds emu_lfd, the 2nd test
> test_bt_base() calls bt_cmd() which also use emu_lfd... could it be unbound?

I meant 'emu_fd' (not emu_lfd) in both cases.

> This is racy, since (from g_test_run() doc):
>
> "In general, the tests and sub-suites within each suite are run in the
> order in which they are defined. [...]
> However, you should never make the actual result of a test depend on the
> order that tests are run in. If you need to ensure that some particular
> code runs before or after a given test case, use g_test_add(), which
> lets you specify setup and teardown functions."