gitlab-ci: Check building ppc64 without TCG
PATCH 1/1 gitlab-ci: Check building ppc64 without TCG
Building QEMU for ppc64 hosts with --disable-tcg used to break a couple of times in the past, see e.g. commit a01b64cee7 ("target/ppc: Put do_rfi under a TCG-only block") or commit 049b4ad669 ("target/ppc: Fix build warnings when building with 'disable-tcg'"), so we should test this in our CI to avoid such regressions.
---
Not sure whether we really should add new shared runner jobs ... but I currently also don't see a better option for this: We don't have any custom ppc64 runners in our CI, so the only other way to test this is travis - but the Travis extension for gitlab recently broke, so the results there are currently rather neglected, I think...
.gitlab-ci.d/crossbuilds.yml | 8 ++++++++ 1 file changed, 8 insertions(+)
.gitlab-ci.d/crossbuilds.yml +8 -0
@@ -112,6 +112,14 @@ cross-ppc64el-user: variables: IMAGE: debian-ppc64el-cross +cross-ppc64el-kvm-only: + extends: .cross_accel_build_job + needs: + job: ppc64el-debian-cross-container + variables: + IMAGE: debian-ppc64el-cross + EXTRA_CONFIGURE_OPTS: --disable-tcg --without-default-devices + # The riscv64 cross-builds currently use a 'sid' container to get # compilers and libraries. Until something more stable is found we # allow_failure so as not to block CI.
Signature · 1 line
2.31.1
Quoted 22 lines · Thomas Huth
Building QEMU for ppc64 hosts with --disable-tcg used to break a couple of times in the past, see e.g. commit
a01b64cee7("target/ppc: Putdo_rfiunder a TCG-only block") or commit049b4ad669("target/ppc: Fix build warnings when building with 'disable-tcg'"), so we should test this in our CI to avoid such regressions.Signed-off-by Thomas Huth <[email protected]>---
Not sure whether we really should add new shared runner jobs ... but I currently also don't see a better option for this: We don't have any custom ppc64 runners in our CI, so the only other way to test this is travis - but the Travis extension for gitlab recently broke, so the results there are currently rather neglected, I think....gitlab-ci.d/crossbuilds.yml | 8 ++++++++ 1 file changed, 8 insertions(+).gitlab-ci.d/crossbuilds.yml +0
-0
.gitlab-ci.d/crossbuilds.yml cross-ppc64el-user: · line 121
Quoted 10 lines
.gitlab-ci.d/crossbuilds.yml +7
-0@@ -112,6 +112,14 @@ cross-ppc64el-user: variables: IMAGE: debian-ppc64el-cross +cross-ppc64el-kvm-only: + extends: .cross_accel_build_job + needs: + job: ppc64el-debian-cross-container + variables: + IMAGE: debian-ppc64el-cross + EXTRA_CONFIGURE_OPTS: --disable-tcg --without-default-devices
Maybe also --disable-tools --disable-docs?
Regardless:
Quoted 18 lines · Philippe Mathieu-Daudé
Quoted 17 lines · Thomas Huth
Building QEMU for ppc64 hosts with --disable-tcg used to break a couple of times in the past, see e.g. commit
a01b64cee7("target/ppc: Putdo_rfiunder a TCG-only block") or commit049b4ad669("target/ppc: Fix build warnings when building with 'disable-tcg'"), so we should test this in our CI to avoid such regressions.Signed-off-by Thomas Huth <[email protected]>---
Not sure whether we really should add new shared runner jobs ... but I currently also don't see a better option for this: We don't have any custom ppc64 runners in our CI, so the only other way to test this is travis - but the Travis extension for gitlab recently broke, so the results there are currently rather neglected, I think...
.gitlab-ci.d/crossbuilds.yml| 8 ++++++++1 file changed, 8 insertions(+)
Quoted 16 lines
Quoted 6 lines
.gitlab-ci.d/crossbuilds.yml +0
-0@@ -112,6 +112,14 @@ cross-ppc64el-user:variables:
Quoted 8 lines
IMAGE: debian-ppc64el-cross +cross-ppc64el-kvm-only:
+ extends: .cross_accel_build_job
+ needs:
+ job: ppc64el-debian-cross-container
+ variables:
+ IMAGE: debian-ppc64el-cross
+EXTRA_CONFIGURE_OPTS: --disable-tcg --without-default-devicesMaybe also --disable-tools --disable-docs?
These are already added by the template.
Quoted 3 lines
Regardless:
Reviewed-by Philippe Mathieu-Daudé <[email protected]>
Thanks!
Thomas
Quoted 7 lines · Thomas Huth
Building QEMU for ppc64 hosts with --disable-tcg used to break a couple of times in the past, see e.g. commit
a01b64cee7("target/ppc: Putdo_rfiunder a TCG-only block") or commit049b4ad669("target/ppc: Fix build warnings when building with 'disable-tcg'"), so we should test this in our CI to avoid such regressions.Signed-off-by Thomas Huth <[email protected]>
Queued to testing/next, thanks.
Signature · 2 lines
Alex Bennée Virtualisation Tech Lead @ Linaro