| Message ID | 20201217094330.17400-1-alex.bennee@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, 17 Dec 2020 at 09:43, Alex Bennée <alex.bennee@linaro.org> wrote: > > The following changes since commit af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f: > > Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging (2020-12-15 21:24:31 +0000) > > are available in the Git repository at: > > https://github.com/stsquad/qemu.git tags/pull-testing-171220-2 > > for you to fetch changes up to a1c04860cec798914f8cd8ef77fce195f360d36e: > > tests: update for rename of CentOS8 PowerTools repo (2020-12-17 09:38:51 +0000) > > ---------------------------------------------------------------- > Testing and configure updates: > > - add moxie-softmmu to deprecated_targets_list > - improve cross-build KVM coverage > - new --without-default-features configure flag > - add __repr__ for ConsoleSocket for debugging > - build tcg tests with -Werror > - test 32 bit builds with fedora > - remove last traces of debian9 > - hotfix for centos8 powertools repo This seems to fail for x86 builds on my local Linux system with errors like this: make: Entering directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))} QTEST_QEMU_IMG=./qemu-img G_TEST_DBUS_DAEMON=/home/petmay01/linaro/qemu-for-merges/tests/dbus-vmstate-daemon.sh QTEST_QEMU_BINARY=./qemu-system-moxie tests/qtest/qmp-cmd-test --tap -k sh: 1: exec: ./qemu-system-moxie: not found socket_accept failed: Resource temporarily unavailable ** ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) ../../tests/qtest/libqtest.c:172: kill_qemu() tried to terminate QEMU process but encountered exit status 127 (expected 0) ERROR qtest-moxie/qmp-cmd-test - Bail out! ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) Makefile.mtest:2417: recipe for target 'run-test-300' failed make: *** [run-test-300] Error 1 make: Leaving directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' Something in the Moxie deprecation patch not right? If we didn't build the executable we shouldn't try to run tests for it... thanks -- PMM
Peter Maydell <peter.maydell@linaro.org> writes: > On Thu, 17 Dec 2020 at 09:43, Alex Bennée <alex.bennee@linaro.org> wrote: >> >> The following changes since commit af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f: >> >> Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging (2020-12-15 21:24:31 +0000) >> >> are available in the Git repository at: >> >> https://github.com/stsquad/qemu.git tags/pull-testing-171220-2 >> >> for you to fetch changes up to a1c04860cec798914f8cd8ef77fce195f360d36e: >> >> tests: update for rename of CentOS8 PowerTools repo (2020-12-17 09:38:51 +0000) >> >> ---------------------------------------------------------------- >> Testing and configure updates: >> >> - add moxie-softmmu to deprecated_targets_list >> - improve cross-build KVM coverage >> - new --without-default-features configure flag >> - add __repr__ for ConsoleSocket for debugging >> - build tcg tests with -Werror >> - test 32 bit builds with fedora >> - remove last traces of debian9 >> - hotfix for centos8 powertools repo > > This seems to fail for x86 builds on my local Linux system with > errors like this: > make: Entering directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' > MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))} > QTEST_QEMU_IMG=./qemu-img > G_TEST_DBUS_DAEMON=/home/petmay01/linaro/qemu-for-merges/tests/dbus-vmstate-daemon.sh > QTEST_QEMU_BINARY=./qemu-system-moxie tests/qtest/qmp-cmd-test --tap > -k > sh: 1: exec: ./qemu-system-moxie: not found > socket_accept failed: Resource temporarily unavailable > ** > ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: > assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) > ../../tests/qtest/libqtest.c:172: kill_qemu() tried to terminate QEMU > process but encountered exit status 127 (expected 0) > ERROR qtest-moxie/qmp-cmd-test - Bail out! > ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: > assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) > Makefile.mtest:2417: recipe for target 'run-test-300' failed > make: *** [run-test-300] Error 1 > make: Leaving directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' > > Something in the Moxie deprecation patch not right? > If we didn't build the executable we shouldn't try to > run tests for it... We shouldn't but I can't replicate the failure here. I'm guessing it goes away on a "make clean"? Maybe meson is leaving something behind from the previous iteration? > > thanks > -- PMM
On Thu, 17 Dec 2020 at 23:43, Alex Bennée <alex.bennee@linaro.org> wrote: > > > Peter Maydell <peter.maydell@linaro.org> writes: > > > On Thu, 17 Dec 2020 at 09:43, Alex Bennée <alex.bennee@linaro.org> wrote: > >> > >> The following changes since commit af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f: > >> > >> Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging (2020-12-15 21:24:31 +0000) > >> > >> are available in the Git repository at: > >> > >> https://github.com/stsquad/qemu.git tags/pull-testing-171220-2 > >> > >> for you to fetch changes up to a1c04860cec798914f8cd8ef77fce195f360d36e: > >> > >> tests: update for rename of CentOS8 PowerTools repo (2020-12-17 09:38:51 +0000) > >> > >> ---------------------------------------------------------------- > >> Testing and configure updates: > >> > >> - add moxie-softmmu to deprecated_targets_list > >> - improve cross-build KVM coverage > >> - new --without-default-features configure flag > >> - add __repr__ for ConsoleSocket for debugging > >> - build tcg tests with -Werror > >> - test 32 bit builds with fedora > >> - remove last traces of debian9 > >> - hotfix for centos8 powertools repo > > > > This seems to fail for x86 builds on my local Linux system with > > errors like this: > > make: Entering directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' > > MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))} > > QTEST_QEMU_IMG=./qemu-img > > G_TEST_DBUS_DAEMON=/home/petmay01/linaro/qemu-for-merges/tests/dbus-vmstate-daemon.sh > > QTEST_QEMU_BINARY=./qemu-system-moxie tests/qtest/qmp-cmd-test --tap > > -k > > sh: 1: exec: ./qemu-system-moxie: not found > > socket_accept failed: Resource temporarily unavailable > > ** > > ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: > > assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) > > ../../tests/qtest/libqtest.c:172: kill_qemu() tried to terminate QEMU > > process but encountered exit status 127 (expected 0) > > ERROR qtest-moxie/qmp-cmd-test - Bail out! > > ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: > > assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) > > Makefile.mtest:2417: recipe for target 'run-test-300' failed > > make: *** [run-test-300] Error 1 > > make: Leaving directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' > > > > Something in the Moxie deprecation patch not right? > > If we didn't build the executable we shouldn't try to > > run tests for it... > > We shouldn't but I can't replicate the failure here. I'm guessing it > goes away on a "make clean"? No, the 'alldbg' build is the one that does do a 'make clean'; the build-not-from-clean build worked (presumably because the old executable was still left in place from the previous build). thanks -- PMM
Peter Maydell <peter.maydell@linaro.org> writes: > On Thu, 17 Dec 2020 at 23:43, Alex Bennée <alex.bennee@linaro.org> wrote: >> >> >> Peter Maydell <peter.maydell@linaro.org> writes: >> >> > On Thu, 17 Dec 2020 at 09:43, Alex Bennée <alex.bennee@linaro.org> wrote: >> >> >> >> The following changes since commit af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f: >> >> >> >> Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging (2020-12-15 21:24:31 +0000) >> >> >> >> are available in the Git repository at: >> >> >> >> https://github.com/stsquad/qemu.git tags/pull-testing-171220-2 >> >> >> >> for you to fetch changes up to a1c04860cec798914f8cd8ef77fce195f360d36e: >> >> >> >> tests: update for rename of CentOS8 PowerTools repo (2020-12-17 09:38:51 +0000) >> >> >> >> ---------------------------------------------------------------- >> >> Testing and configure updates: >> >> >> >> - add moxie-softmmu to deprecated_targets_list >> >> - improve cross-build KVM coverage >> >> - new --without-default-features configure flag >> >> - add __repr__ for ConsoleSocket for debugging >> >> - build tcg tests with -Werror >> >> - test 32 bit builds with fedora >> >> - remove last traces of debian9 >> >> - hotfix for centos8 powertools repo >> > >> > This seems to fail for x86 builds on my local Linux system with >> > errors like this: >> > make: Entering directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' >> > MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))} >> > QTEST_QEMU_IMG=./qemu-img >> > G_TEST_DBUS_DAEMON=/home/petmay01/linaro/qemu-for-merges/tests/dbus-vmstate-daemon.sh >> > QTEST_QEMU_BINARY=./qemu-system-moxie tests/qtest/qmp-cmd-test --tap >> > -k >> > sh: 1: exec: ./qemu-system-moxie: not found >> > socket_accept failed: Resource temporarily unavailable >> > ** >> > ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: >> > assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) >> > ../../tests/qtest/libqtest.c:172: kill_qemu() tried to terminate QEMU >> > process but encountered exit status 127 (expected 0) >> > ERROR qtest-moxie/qmp-cmd-test - Bail out! >> > ERROR:../../tests/qtest/libqtest.c:308:qtest_init_without_qmp_handshake: >> > assertion failed: (s->fd >= 0 && s->qmp_fd >= 0) >> > Makefile.mtest:2417: recipe for target 'run-test-300' failed >> > make: *** [run-test-300] Error 1 >> > make: Leaving directory '/home/petmay01/linaro/qemu-for-merges/build/alldbg' >> > >> > Something in the Moxie deprecation patch not right? >> > If we didn't build the executable we shouldn't try to >> > run tests for it... >> >> We shouldn't but I can't replicate the failure here. I'm guessing it >> goes away on a "make clean"? > > No, the 'alldbg' build is the one that does do a 'make clean'; > the build-not-from-clean build worked (presumably because the > old executable was still left in place from the previous build). Hmm I wonder if it's related to the configure options then. What's your configure setup for that build? > > thanks > -- PMM
On Fri, 18 Dec 2020 at 11:54, Alex Bennée <alex.bennee@linaro.org> wrote: > > > Peter Maydell <peter.maydell@linaro.org> writes: > > No, the 'alldbg' build is the one that does do a 'make clean'; > > the build-not-from-clean build worked (presumably because the > > old executable was still left in place from the previous build). > > Hmm I wonder if it's related to the configure options then. What's your > configure setup for that build? '../../configure' '--cc=ccache gcc' '--enable-debug' '--python=python3' -- PMM
Peter Maydell <peter.maydell@linaro.org> writes: > On Fri, 18 Dec 2020 at 11:54, Alex Bennée <alex.bennee@linaro.org> wrote: >> >> >> Peter Maydell <peter.maydell@linaro.org> writes: >> > No, the 'alldbg' build is the one that does do a 'make clean'; >> > the build-not-from-clean build worked (presumably because the >> > old executable was still left in place from the previous build). >> >> Hmm I wonder if it's related to the configure options then. What's your >> configure setup for that build? > > '../../configure' '--cc=ccache gcc' '--enable-debug' > '--python=python3' So I did the following: git checkout af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f cd build/bisect rm -rf * '../../configure' '--cc=ccache gcc' '--enable-debug' '--python=python3' make -j30 make check-qtest All ok, then git checkout pull-testing-171220-2 cd build/bisect make clean ag -r "moxie" | wc -l And sure enough there is a lot of detritus left (21 lines by ag's counting). However: make -j30 make check-qtest still passes and doesn't attempt to do anything with moxie. So I'm kind of stuck as to what to do now? Maybe we can compare detritus and see what is left in your tree that causes the test system to get confused and make "clean" really mean it. $ ag -l -r "moxie" Makefile.ninja moxie-softmmu-config-target.h qapi/qapi-visit-misc-target.c qapi/qapi-events-misc-target.h qapi/qapi-types-machine.h qapi/qapi-introspect.c qapi/qapi-types-misc-target.h qapi/qapi-visit-machine.c qapi/qapi-visit-misc-target.h qapi/qapi-events-misc-target.c qapi/qapi-types-machine.c meson-info/intro-buildsystem_files.json docs/qemu-qmp-ref.7 docs/interop/qemu-qmp-ref.html docs/system/deprecated.html docs/system/index.html moxie-softmmu-config-devices.mak moxie-softmmu-config-devices.h build.ninja I'm surprised about build.ninja and Makefile.ninja? Don't they get regenerated by the configure? > > -- PMM
Alex Bennée <alex.bennee@linaro.org> writes: > Peter Maydell <peter.maydell@linaro.org> writes: > >> On Fri, 18 Dec 2020 at 11:54, Alex Bennée <alex.bennee@linaro.org> wrote: >>> >>> >>> Peter Maydell <peter.maydell@linaro.org> writes: >>> > No, the 'alldbg' build is the one that does do a 'make clean'; >>> > the build-not-from-clean build worked (presumably because the >>> > old executable was still left in place from the previous build). >>> >>> Hmm I wonder if it's related to the configure options then. What's your >>> configure setup for that build? >> >> '../../configure' '--cc=ccache gcc' '--enable-debug' >> '--python=python3' > > So I did the following: > > git checkout af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f > cd build/bisect > rm -rf * > '../../configure' '--cc=ccache gcc' '--enable-debug' '--python=python3' > make -j30 > make check-qtest > > All ok, then > > git checkout pull-testing-171220-2 > cd build/bisect > make clean > ag -r "moxie" | wc -l > > And sure enough there is a lot of detritus left (21 lines by ag's > counting). However: > > make -j30 > make check-qtest > > still passes and doesn't attempt to do anything with moxie. So I'm kind > of stuck as to what to do now? Maybe we can compare detritus and see > what is left in your tree that causes the test system to get confused > and make "clean" really mean it. > > $ ag -l -r "moxie" > Makefile.ninja > moxie-softmmu-config-target.h > qapi/qapi-visit-misc-target.c > qapi/qapi-events-misc-target.h > qapi/qapi-types-machine.h > qapi/qapi-introspect.c > qapi/qapi-types-misc-target.h > qapi/qapi-visit-machine.c > qapi/qapi-visit-misc-target.h > qapi/qapi-events-misc-target.c > qapi/qapi-types-machine.c > meson-info/intro-buildsystem_files.json > docs/qemu-qmp-ref.7 > docs/interop/qemu-qmp-ref.html > docs/system/deprecated.html > docs/system/index.html > moxie-softmmu-config-devices.mak > moxie-softmmu-config-devices.h > build.ninja > > I'm surprised about build.ninja and Makefile.ninja? Don't they get > regenerated by the configure? Even weirder, re-run configure and the list grows! $ ag -l -r "moxie" Makefile.ninja moxie-softmmu-config-target.h qapi/qapi-visit-misc-target.c qapi/qapi-events-misc-target.h qapi/qapi-types-machine.h qapi/qapi-introspect.c qapi/qapi-types-misc-target.h qapi/qapi-visit-machine.c qapi/qapi-visit-misc-target.h qapi/qapi-events-misc-target.c qapi/qapi-types-machine.c meson-info/intro-buildsystem_files.json docs/qemu-qmp-ref.7 docs/interop/qemu-qmp-ref.html docs/system/deprecated.html docs/system/index.html moxie-softmmu-config-devices.mak meson-private/microblazeel-softmmu-config-devices.mak.d meson-private/sh4eb-softmmu-config-devices.mak.d meson-private/xtensa-softmmu-config-devices.mak.d meson-private/mips64-softmmu-config-devices.mak.d meson-private/nios2-softmmu-config-devices.mak.d meson-private/rx-softmmu-config-devices.mak.d meson-private/tricore-softmmu-config-devices.mak.d meson-private/s390x-softmmu-config-devices.mak.d meson-private/microblaze-softmmu-config-devices.mak.d meson-private/riscv64-softmmu-config-devices.mak.d meson-private/riscv32-softmmu-config-devices.mak.d meson-private/sparc64-softmmu-config-devices.mak.d meson-private/arm-softmmu-config-devices.mak.d meson-private/xtensaeb-softmmu-config-devices.mak.d meson-private/sh4-softmmu-config-devices.mak.d meson-private/mips-softmmu-config-devices.mak.d meson-private/or1k-softmmu-config-devices.mak.d meson-private/sparc-softmmu-config-devices.mak.d meson-private/x86_64-softmmu-config-devices.mak.d meson-private/avr-softmmu-config-devices.mak.d meson-private/mips64el-softmmu-config-devices.mak.d meson-private/mipsel-softmmu-config-devices.mak.d meson-private/ppc-softmmu-config-devices.mak.d meson-private/aarch64-softmmu-config-devices.mak.d meson-private/i386-softmmu-config-devices.mak.d meson-private/alpha-softmmu-config-devices.mak.d meson-private/ppc64-softmmu-config-devices.mak.d meson-private/hppa-softmmu-config-devices.mak.d meson-private/m68k-softmmu-config-devices.mak.d meson-private/cris-softmmu-config-devices.mak.d moxie-softmmu-config-devices.h build.ninja
On 18/12/20 18:17, Alex Bennée wrote: >> I'm surprised about build.ninja and Makefile.ninja Don't they get >> regenerated by the configure? This just means that meson _looked_ at some moxie file, so the files are a dependency of configure. > Even weirder, re-run configure and the list grows! > > meson-private/microblazeel-softmmu-config-devices.mak.d > meson-private/sh4eb-softmmu-config-devices.mak.d > meson-private/xtensa-softmmu-config-devices.mak.d > meson-private/mips64-softmmu-config-devices.mak.d > meson-private/nios2-softmmu-config-devices.mak.d > meson-private/rx-softmmu-config-devices.mak.d > meson-private/tricore-softmmu-config-devices.mak.d > meson-private/s390x-softmmu-config-devices.mak.d > meson-private/microblaze-softmmu-config-devices.mak.d > meson-private/riscv64-softmmu-config-devices.mak.d > meson-private/riscv32-softmmu-config-devices.mak.d > meson-private/sparc64-softmmu-config-devices.mak.d > meson-private/arm-softmmu-config-devices.mak.d > meson-private/xtensaeb-softmmu-config-devices.mak.d > meson-private/sh4-softmmu-config-devices.mak.d > meson-private/mips-softmmu-config-devices.mak.d > meson-private/or1k-softmmu-config-devices.mak.d > meson-private/sparc-softmmu-config-devices.mak.d > meson-private/x86_64-softmmu-config-devices.mak.d > meson-private/avr-softmmu-config-devices.mak.d > meson-private/mips64el-softmmu-config-devices.mak.d > meson-private/mipsel-softmmu-config-devices.mak.d > meson-private/ppc-softmmu-config-devices.mak.d > meson-private/aarch64-softmmu-config-devices.mak.d > meson-private/i386-softmmu-config-devices.mak.d > meson-private/alpha-softmmu-config-devices.mak.d > meson-private/ppc64-softmmu-config-devices.mak.d > meson-private/hppa-softmmu-config-devices.mak.d > meson-private/m68k-softmmu-config-devices.mak.d > meson-private/cris-softmmu-config-devices.mak.d > moxie-softmmu-config-devices.h > build.ninja This is expected since there's still an unconditional "source hw/moxie/Kconfig". Alex, would you like to just drop the deprecation patch? I can pick it up in my tree and try to debug it. Paolo
The following changes since commit af3f37319cb1e1ca0c42842ecdbd1bcfc64a4b6f: Merge remote-tracking branch 'remotes/bonzini-gitlab/tags/for-upstream' into staging (2020-12-15 21:24:31 +0000) are available in the Git repository at: https://github.com/stsquad/qemu.git tags/pull-testing-171220-2 for you to fetch changes up to a1c04860cec798914f8cd8ef77fce195f360d36e: tests: update for rename of CentOS8 PowerTools repo (2020-12-17 09:38:51 +0000) ---------------------------------------------------------------- Testing and configure updates: - add moxie-softmmu to deprecated_targets_list - improve cross-build KVM coverage - new --without-default-features configure flag - add __repr__ for ConsoleSocket for debugging - build tcg tests with -Werror - test 32 bit builds with fedora - remove last traces of debian9 - hotfix for centos8 powertools repo ---------------------------------------------------------------- Alex Bennée (8): configure: include moxie-softmmu in deprecated_targets_list gitlab: include aarch64-softmmu and ppc64-softmmu cross-system-build configure: move gettext detection to meson.build configure: add --without-default-features python: add __repr__ to ConsoleSocket to aid debugging gitlab: move --without-default-devices build from Travis gitlab: add --without-default-features build tests/tcg: build tests with -Werror Daniel P. Berrangé (1): tests: update for rename of CentOS8 PowerTools repo Thomas Huth (2): gitlab-CI: Test 32-bit builds with the fedora-i386-cross container tests/docker: Remove the remainders of debian9 containers from the Makefile configure | 176 +++++++++++----------- .gitlab-ci.d/crossbuilds.yml | 22 ++- .gitlab-ci.yml | 16 +- .travis.yml | 8 - meson_options.txt | 2 +- po/meson.build | 2 +- python/qemu/console_socket.py | 7 + tests/docker/Makefile.include | 6 +- tests/docker/dockerfiles/centos8.docker | 5 +- tests/docker/dockerfiles/fedora-i386-cross.docker | 18 ++- tests/tcg/Makefile.target | 2 +- 11 files changed, 150 insertions(+), 114 deletions(-)