Re-testing gitlab CI shows the ppc64 and s390x boot_linux tests take 100-150 seconds each. The x86-64 TCG tests take a similar ~100s each, and are the longest-running avocado tests in gitlab. >From avocado-system-centos:
boot_linux.py:BootLinuxX8664.test_pc_i440fx_tcg: PASS (112.34 s) boot_linux.py:BootLinuxX8664.test_pc_q35_tcg: PASS (97.05 s) boot_linux.py:BootLinuxPPC64.test_pseries_tcg: PASS (148.86 s) boot_linux.py:BootLinuxS390X.test_s390_ccw_virtio_tcg: PASS (149.83 s) So mark the x86-64 tests as SPEED=slow as well. Signed-off-by: Nicholas Piggin <npig...@gmail.com> The other way we could go is enabling them all since ppc64 and s390s are now much faster than when they were originally disabled; or to only enable q35, giving at least one boot_linux.py test. [https://gitlab.com/npiggin/qemu/-/jobs/5842257510 for results] --- tests/avocado/boot_linux.py | 16 +++++++--------- 1 file changed, 7 insertions(+), 9 deletions(-) diff --git a/tests/avocado/boot_linux.py b/tests/avocado/boot_linux.py index de4c8805f7..7c9cf6ae15 100644 --- a/tests/avocado/boot_linux.py +++ b/tests/avocado/boot_linux.py @@ -14,6 +14,9 @@ from avocado import skipUnless +# We don't run TCG tests in CI, as booting the current Fedora OS in TCG tests +# is very heavyweight (~100s per test). There are lighter weight distros which +# we use in the machine_aarch64_virt.py, tux_baseline.py, etc. class BootLinuxX8664(LinuxTest): """ @@ -21,6 +24,7 @@ class BootLinuxX8664(LinuxTest): """ timeout = 480 + @skipUnless(os.getenv('SPEED') == 'slow', 'runtime limited') def test_pc_i440fx_tcg(self): """ :avocado: tags=machine:pc @@ -39,6 +43,7 @@ def test_pc_i440fx_kvm(self): self.vm.add_args("-accel", "kvm") self.launch_and_wait(set_up_ssh_connection=False) + @skipUnless(os.getenv('SPEED') == 'slow', 'runtime limited') def test_pc_q35_tcg(self): """ :avocado: tags=machine:q35 @@ -58,9 +63,6 @@ def test_pc_q35_kvm(self): self.launch_and_wait(set_up_ssh_connection=False) -# For Aarch64 we only boot KVM tests in CI as booting the current -# Fedora OS in TCG tests is very heavyweight. There are lighter weight -# distros which we use in the machine_aarch64_virt.py tests. class BootLinuxAarch64(LinuxTest): """ :avocado: tags=arch:aarch64 @@ -84,14 +86,11 @@ def test_virt_kvm(self): self.launch_and_wait(set_up_ssh_connection=False) -# See the tux_baseline.py tests for almost the same coverage in a lot -# less time. class BootLinuxPPC64(LinuxTest): """ :avocado: tags=arch:ppc64 """ - - timeout = 360 + timeout = 480 @skipUnless(os.getenv('SPEED') == 'slow', 'runtime limited') def test_pseries_tcg(self): @@ -108,8 +107,7 @@ class BootLinuxS390X(LinuxTest): """ :avocado: tags=arch:s390x """ - - timeout = 240 + timeout = 480 @skipUnless(os.getenv('SPEED') == 'slow', 'runtime limited') def test_s390_ccw_virtio_tcg(self): -- 2.42.0
