[PATCH 0/2] Improvements for the device-crash-test jobs

[PATCH 0/2] Improvements for the device-crash-test jobs

[Thomas Huth]

Improve the runtime of the device-crash-test jobs by avoiding
to run "configure" again and by forcing to test with TCG only
(instead of testing twice, with TCG and KVM).

Thomas Huth (2):
  gitlab-ci: Avoid to re-run "configure" in the device-crash-test jobs
  scripts/device-crash-test: Add a parameter to run with TCG only

 .gitlab-ci.d/buildtest.yml | 6 +++---
 scripts/device-crash-test  | 4 +++-
 2 files changed, 6 insertions(+), 4 deletions(-)

-- 
2.31.1

[PATCH 1/2] gitlab-ci: Avoid to re-run "configure" in the device-crash-test jobs

[Thomas Huth]

After "make check-venv" had been added to these jobs, they started
to re-run "configure" each time since our logic in the makefile
thinks that some files are out of date here. Avoid it with the same
trick that we are using in buildtest-template.yml already by disabling
the up-to-date check via NINJA=":".

Fixes: 1d8cf47e5b ("tests: run 'device-crash-test' from tests/venv")
Signed-off-by: Thomas Huth <thuth@redhat.com>
---
 .gitlab-ci.d/buildtest.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/.gitlab-ci.d/buildtest.yml b/.gitlab-ci.d/buildtest.yml
index ba6f551752..333eea9dd3 100644
--- a/.gitlab-ci.d/buildtest.yml
+++ b/.gitlab-ci.d/buildtest.yml
@@ -102,7 +102,7 @@ crash-test-debian:
     IMAGE: debian-amd64
   script:
     - cd build
-    - make check-venv
+    - make NINJA=":" check-venv
     - tests/venv/bin/python3 scripts/device-crash-test -q ./qemu-system-i386
 
 build-system-fedora:
@@ -145,7 +145,7 @@ crash-test-fedora:
     IMAGE: fedora
   script:
     - cd build
-    - make check-venv
+    - make NINJA=":" check-venv
     - tests/venv/bin/python3 scripts/device-crash-test -q ./qemu-system-ppc
     - tests/venv/bin/python3 scripts/device-crash-test -q ./qemu-system-riscv32
 
-- 
2.31.1

[PATCH 2/2] scripts/device-crash-test: Add a parameter to run with TCG only

[Thomas Huth]

We're currently facing the problem that the device-crash-test script
runs twice as long in the CI when a runner supports KVM - which sometimes
results in a timeout of the CI job. To get a more deterministic runtime
here, add an option to the script that allows to run it with TCG only.

Reported-by: Eldon Stegall <eldon-qemu@eldondev.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
---
 .gitlab-ci.d/buildtest.yml | 2 +-
 scripts/device-crash-test  | 4 +++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/.gitlab-ci.d/buildtest.yml b/.gitlab-ci.d/buildtest.yml
index 333eea9dd3..bb3650a51c 100644
--- a/.gitlab-ci.d/buildtest.yml
+++ b/.gitlab-ci.d/buildtest.yml
@@ -103,7 +103,7 @@ crash-test-debian:
   script:
     - cd build
     - make NINJA=":" check-venv
-    - tests/venv/bin/python3 scripts/device-crash-test -q ./qemu-system-i386
+    - tests/venv/bin/python3 scripts/device-crash-test -q --tcg-only ./qemu-system-i386
 
 build-system-fedora:
   extends:
diff --git a/scripts/device-crash-test b/scripts/device-crash-test
index 73bcb98693..b74d887331 100755
--- a/scripts/device-crash-test
+++ b/scripts/device-crash-test
@@ -397,7 +397,7 @@ def binariesToTest(args, testcase):
 
 
 def accelsToTest(args, testcase):
-    if getBinaryInfo(args, testcase['binary']).kvm_available:
+    if getBinaryInfo(args, testcase['binary']).kvm_available and not args.tcg_only:
         yield 'kvm'
     yield 'tcg'
 
@@ -510,6 +510,8 @@ def main():
                         help="Full mode: test cases that are expected to fail")
     parser.add_argument('--strict', action='store_true', dest='strict',
                         help="Treat all warnings as fatal")
+    parser.add_argument('--tcg-only', action='store_true', dest='tcg_only',
+                        help="Only test with TCG accelerator")
     parser.add_argument('qemu', nargs='*', metavar='QEMU',
                         help='QEMU binary to run')
     args = parser.parse_args()
-- 
2.31.1

Re: [PATCH 0/2] Improvements for the device-crash-test jobs

[Alex Bennée]

Thomas Huth <thuth@redhat.com> writes:

> Improve the runtime of the device-crash-test jobs by avoiding
> to run "configure" again and by forcing to test with TCG only
> (instead of testing twice, with TCG and KVM).

Queued to testing/next, thanks.

-- 
Alex Bennée
Virtualisation Tech Lead @ Linaro