[PATCH 0/2] tests/qtest: Fix problems running with make vm-build-openbsd

[PATCH 0/2] tests/qtest: Fix problems running with make vm-build-openbsd

[Peter Maydell]

I noticed that 'make vm-build-openbsd' had bitrotted a bit:
 * two tests for the stm32l4x5 forgot a qtest_quit(), which meant
   they would pass all their subtests but then timeout because the
   underlying QEMU process never exited
 * the ahci-test timeout was a bit aggressive

This patchset fixes these.

thanks
-- PMM

Peter Maydell (2):
  tests/qtest: Add missing qtest_quit() to stm32 tests
  tests/qtest: Bump timeout on ahci-test

 tests/qtest/dm163-test.c           | 2 ++
 tests/qtest/stm32l4x5_usart-test.c | 2 ++
 tests/qtest/meson.build            | 1 +
 3 files changed, 5 insertions(+)

-- 
2.34.1

[PATCH 1/2] tests/qtest: Add missing qtest_quit() to stm32 tests

[Peter Maydell]

In the dm163-test and stm32l4x5_usart-test, a couple of subtests are
missing the qtest_quit() call.  The effect of this is that on hosts
other than Linux and FreeBSD the test will timeout after executing
all the tests:

242/845 qemu:qtest+qtest-arm / qtest-arm/dm163-test                               TIMEOUT         60.04s   3 subtests passed
100/845 qemu:qtest+qtest-arm / qtest-arm/stm32l4x5_usart-test                     TIMEOUT        600.02s   5 subtests passed

This happens because the qemu-system-arm binary which the test
starts does not exit, and because it shares the stdout with the
test binary, the overall meson test harness thinks the test is
still running. On Linux and FreeBSD we have an extra safety net
set up in qtest_spawn_qemu() which kills off any QEMU binary that
ends up without a parent. This is intended for the case where
QEMU crashed and didn't respond to a SIGTERM or polite request
to quit, but it also sidestepped the problem in this case.
However, OpenBSD doesn't have a PDEATHSIG equivalent, so we
see the timeouts when running a 'make vm-build-openbsd' run.

Add the missing qtest_quit() calls.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
Many thanks to Dan Berrangé for diagnosing the cause of this hang...
Dan also suggested on IRC that we should support g_autoptr for
the QTestState, so you don't need to manually qtest_quit(). But
for the immediate fix, I just add the missing calls.
---
 tests/qtest/dm163-test.c           | 2 ++
 tests/qtest/stm32l4x5_usart-test.c | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/tests/qtest/dm163-test.c b/tests/qtest/dm163-test.c
index 3161c9208d8..4c8e654af29 100644
--- a/tests/qtest/dm163-test.c
+++ b/tests/qtest/dm163-test.c
@@ -182,6 +182,8 @@ static void test_dm163_gpio_connection(void)
     g_assert_false(qtest_get_irq(qts, LAT_B));
     g_assert_false(qtest_get_irq(qts, SELBK));
     g_assert_false(qtest_get_irq(qts, RST_B));
+
+    qtest_quit(qts);
 }
 
 int main(int argc, char **argv)
diff --git a/tests/qtest/stm32l4x5_usart-test.c b/tests/qtest/stm32l4x5_usart-test.c
index 89025182331..c175ff3064f 100644
--- a/tests/qtest/stm32l4x5_usart-test.c
+++ b/tests/qtest/stm32l4x5_usart-test.c
@@ -202,6 +202,8 @@ static void test_write_read(void)
     qtest_writel(qts, USART1_BASE_ADDR + A_TDR, 0xFFFFFFFF);
     const uint32_t tdr = qtest_readl(qts, USART1_BASE_ADDR + A_TDR);
     g_assert_cmpuint(tdr, ==, 0x000001FF);
+
+    qtest_quit(qts);
 }
 
 static void test_receive_char(void)
-- 
2.34.1

Re: [PATCH 1/2] tests/qtest: Add missing qtest_quit() to stm32 tests

[Daniel P. Berrangé]

On Thu, Sep 05, 2024 at 05:55:53PM +0100, Peter Maydell wrote:
> In the dm163-test and stm32l4x5_usart-test, a couple of subtests are
> missing the qtest_quit() call.  The effect of this is that on hosts
> other than Linux and FreeBSD the test will timeout after executing
> all the tests:
> 
> 242/845 qemu:qtest+qtest-arm / qtest-arm/dm163-test                               TIMEOUT         60.04s   3 subtests passed
> 100/845 qemu:qtest+qtest-arm / qtest-arm/stm32l4x5_usart-test                     TIMEOUT        600.02s   5 subtests passed
> 
> This happens because the qemu-system-arm binary which the test
> starts does not exit, and because it shares the stdout with the
> test binary, the overall meson test harness thinks the test is
> still running. On Linux and FreeBSD we have an extra safety net
> set up in qtest_spawn_qemu() which kills off any QEMU binary that
> ends up without a parent. This is intended for the case where
> QEMU crashed and didn't respond to a SIGTERM or polite request
> to quit, but it also sidestepped the problem in this case.
> However, OpenBSD doesn't have a PDEATHSIG equivalent, so we
> see the timeouts when running a 'make vm-build-openbsd' run.
> 
> Add the missing qtest_quit() calls.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> Many thanks to Dan Berrangé for diagnosing the cause of this hang...
> Dan also suggested on IRC that we should support g_autoptr for
> the QTestState, so you don't need to manually qtest_quit(). But
> for the immediate fix, I just add the missing calls.
> ---
>  tests/qtest/dm163-test.c           | 2 ++
>  tests/qtest/stm32l4x5_usart-test.c | 2 ++
>  2 files changed, 4 insertions(+)

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>


With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|

[PATCH 2/2] tests/qtest: Bump timeout on ahci-test

[Peter Maydell]

On my OpenBSD VM test system, the ahci-test sometimes hits its 60 second
timeout. It has 75 subtests and allowing at least two seconds per
subtest seems reasonable. Bump it to 150s.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 tests/qtest/meson.build | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/qtest/meson.build b/tests/qtest/meson.build
index 2f0d3ef0809..fc852f3d8ba 100644
--- a/tests/qtest/meson.build
+++ b/tests/qtest/meson.build
@@ -1,4 +1,5 @@
 slow_qtests = {
+  'ahci-test': 150,
   'aspeed_smc-test': 360,
   'bios-tables-test' : 910,
   'cdrom-test' : 610,
-- 
2.34.1

Re: [PATCH 2/2] tests/qtest: Bump timeout on ahci-test

[Daniel P. Berrangé]

On Thu, Sep 05, 2024 at 05:55:54PM +0100, Peter Maydell wrote:
> On my OpenBSD VM test system, the ahci-test sometimes hits its 60 second
> timeout. It has 75 subtests and allowing at least two seconds per
> subtest seems reasonable. Bump it to 150s.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>  tests/qtest/meson.build | 1 +
>  1 file changed, 1 insertion(+)

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>


With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|

Re: [PATCH 0/2] tests/qtest: Fix problems running with make vm-build-openbsd

[Peter Maydell]

On Thu, 5 Sept 2024 at 17:55, Peter Maydell <peter.maydell@linaro.org> wrote:
>
> I noticed that 'make vm-build-openbsd' had bitrotted a bit:
>  * two tests for the stm32l4x5 forgot a qtest_quit(), which meant
>    they would pass all their subtests but then timeout because the
>    underlying QEMU process never exited
>  * the ahci-test timeout was a bit aggressive
>
> This patchset fixes these.
>
> thanks
> -- PMM
>
> Peter Maydell (2):
>   tests/qtest: Add missing qtest_quit() to stm32 tests
>   tests/qtest: Bump timeout on ahci-test

I've applied these directly to master to fix the
test failures when I apply a pullreq.

thanks
-- PMM