| Message ID | 20200203090932.19147-18-alex.bennee@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | testing/next (with build fixes, finally GREEN!) | expand |
On 03/02/2020 10.09, Alex Bennée wrote: > I've theorised that a parallel build-tcg is somehow getting confused > when two fedora-30 based cross compilers attempt to build at the same > time. From one data-point so far this may fix the problem although the > plugins job runs quite close to timeout. > > Signed-off-by: Alex Bennée <alex.bennee@linaro.org> > --- > .travis.yml | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/.travis.yml b/.travis.yml > index c1c6df475a8..3b35b7cf04d 100644 > --- a/.travis.yml > +++ b/.travis.yml > @@ -380,7 +380,7 @@ matrix: > - name: "GCC check-tcg (user)" > env: > - CONFIG="--disable-system --enable-debug-tcg" > - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" > + - TEST_BUILD_CMD="make build-tcg" > - TEST_CMD="make check-tcg" > - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" > > @@ -391,7 +391,7 @@ matrix: > - name: "GCC plugins check-tcg (user)" > env: > - CONFIG="--disable-system --enable-plugins --enable-debug-tcg --target-list-exclude=sparc64-linux-user,cris-linux-user" > - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" > + - TEST_BUILD_CMD="make build-tcg" > - TEST_CMD="make check-tcg" > - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" Squash it into patch 14/17 ? Thomas
On 2/3/20 11:01 AM, Thomas Huth wrote: > On 03/02/2020 10.09, Alex Bennée wrote: >> I've theorised that a parallel build-tcg is somehow getting confused >> when two fedora-30 based cross compilers attempt to build at the same >> time. From one data-point so far this may fix the problem although the >> plugins job runs quite close to timeout. >> >> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> >> --- >> .travis.yml | 4 ++-- >> 1 file changed, 2 insertions(+), 2 deletions(-) >> >> diff --git a/.travis.yml b/.travis.yml >> index c1c6df475a8..3b35b7cf04d 100644 >> --- a/.travis.yml >> +++ b/.travis.yml >> @@ -380,7 +380,7 @@ matrix: >> - name: "GCC check-tcg (user)" >> env: >> - CONFIG="--disable-system --enable-debug-tcg" >> - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" >> + - TEST_BUILD_CMD="make build-tcg" >> - TEST_CMD="make check-tcg" >> - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" >> >> @@ -391,7 +391,7 @@ matrix: >> - name: "GCC plugins check-tcg (user)" >> env: >> - CONFIG="--disable-system --enable-plugins --enable-debug-tcg --target-list-exclude=sparc64-linux-user,cris-linux-user" >> - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" >> + - TEST_BUILD_CMD="make build-tcg" >> - TEST_CMD="make check-tcg" >> - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" > > Squash it into patch 14/17 ? I'd rather not, ideally we will revert this patch once the issue is fixed.
Philippe Mathieu-Daudé <philmd@redhat.com> writes: > On 2/3/20 11:01 AM, Thomas Huth wrote: >> On 03/02/2020 10.09, Alex Bennée wrote: >>> I've theorised that a parallel build-tcg is somehow getting confused >>> when two fedora-30 based cross compilers attempt to build at the same >>> time. From one data-point so far this may fix the problem although the >>> plugins job runs quite close to timeout. >>> >>> Signed-off-by: Alex Bennée <alex.bennee@linaro.org> >>> --- >>> .travis.yml | 4 ++-- >>> 1 file changed, 2 insertions(+), 2 deletions(-) >>> >>> diff --git a/.travis.yml b/.travis.yml >>> index c1c6df475a8..3b35b7cf04d 100644 >>> --- a/.travis.yml >>> +++ b/.travis.yml >>> @@ -380,7 +380,7 @@ matrix: >>> - name: "GCC check-tcg (user)" >>> env: >>> - CONFIG="--disable-system --enable-debug-tcg" >>> - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" >>> + - TEST_BUILD_CMD="make build-tcg" >>> - TEST_CMD="make check-tcg" >>> - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" >>> @@ -391,7 +391,7 @@ matrix: >>> - name: "GCC plugins check-tcg (user)" >>> env: >>> - CONFIG="--disable-system --enable-plugins --enable-debug-tcg --target-list-exclude=sparc64-linux-user,cris-linux-user" >>> - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" >>> + - TEST_BUILD_CMD="make build-tcg" >>> - TEST_CMD="make check-tcg" >>> - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" >> Squash it into patch 14/17 ? > > I'd rather not, ideally we will revert this patch once the issue is > fixed. Yeah - there should be no reason that we can't build all the tcg tests in parallel. For some reason I think the two fedora30 based images have an issue, despite there not being a common base.
diff --git a/.travis.yml b/.travis.yml index c1c6df475a8..3b35b7cf04d 100644 --- a/.travis.yml +++ b/.travis.yml @@ -380,7 +380,7 @@ matrix: - name: "GCC check-tcg (user)" env: - CONFIG="--disable-system --enable-debug-tcg" - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" + - TEST_BUILD_CMD="make build-tcg" - TEST_CMD="make check-tcg" - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg" @@ -391,7 +391,7 @@ matrix: - name: "GCC plugins check-tcg (user)" env: - CONFIG="--disable-system --enable-plugins --enable-debug-tcg --target-list-exclude=sparc64-linux-user,cris-linux-user" - - TEST_BUILD_CMD="make -j${JOBS} build-tcg" + - TEST_BUILD_CMD="make build-tcg" - TEST_CMD="make check-tcg" - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-debug-tcg"
I've theorised that a parallel build-tcg is somehow getting confused when two fedora-30 based cross compilers attempt to build at the same time. From one data-point so far this may fix the problem although the plugins job runs quite close to timeout. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> --- .travis.yml | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)