that patch was moved because `gbp pq` would delete it otherwise, but it
is auto-applied as part of the build on porter boxes.
Signed-off-by: Fabian Grünbichler <debian@fabian.gruenbichler.email>
armel and armhf actually have ARC_CPU arm, but more importantly, x32 has
ARCH_CPU amd64, so can't be differentiated that way from actual amd64.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
s390x seems to choke on this right now with:
dh_dwz --no-dwz-multifile
dwz: debian/libstd-rust-1.74/usr/lib/s390x-linux-gnu/librustc_driver-3d6b215bbf0742fb.so: Too many DIEs, not optimizing
dh_dwz: error: dwz -- debian/libstd-rust-1.74/usr/lib/s390x-linux-gnu/librustc_driver-3d6b215bbf0742fb.so debian/libstd-rust-1.74/usr/lib/s390x-linux-gnu/libstd-ea06fb4ce966fcaa.so debian/libstd-rust-1.74/usr/lib/s390x-linux-gnu/libtest-4b3dbc8d1d2f22fa.so returned exit code 1
dh_dwz: error: Aborting due to earlier error
let's hope twice the default allows it to finish..
Signed-off-by: Fabian Grünbichler <git@fabian.gruenbichler.email>
dh_makeshlibs only accepts sonames starting with digits, which is not the case
for the rustc libraries in libstd-rust-$ver, since those are using a hash.
exclude that package altogether, and do both the shlibs generation *and* the
ldconfig trigger manually.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
give loong64 the same status as other ports, now that is might build. promote
riscv64 to the next higher tier.
Closes: #1071707
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
and allow easily extending the list of architectures where it doesn't
work (yet), if needed.
Closes: #1061125
Signed-off-by: Fabian Grünbichler <debian@fabian.gruenbichler.email>
Without this, the build/ directory will stick around after having run
make_orig-stage0_tarball.sh. When you then run the suggested command
'dpkg-source -b .', you'll get tons of these types of errors:
dpkg-source: error: cannot represent change to build/x86_64-unknown-linux-gnu/stage0/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-89bc084783fdc439.rlib: binary file contents changed
to work in the absense of clang-16 being pulled in, as is the case for split
any/all builds like on buildds.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
the typo causes a fallback to the default value of 8 allowed failures,
which is obviously even worse than the 24 we had before..
Signed-off-by: Fabian Grünbichler <debian@fabian.gruenbichler.email>
they might have vendored slightly different versions than the main
workspace, but we pruned those at tarball generation.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
since `asm!` got stabilized in 1.59, more tests started using it in
1.60. mips64el doesn't support this feature (yet).
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
starting with 1.60, they are no longer able to build rustdoc because the
rustc_driver.so file is too large (>1G) and cannot be mmap-ed:
[...]
error[E0786]: found invalid metadata files for crate `rustc_driver`
--> src/librustdoc/lib.rs:43:1
|
43 | extern crate rustc_driver;
| ^^^^^^^^^^^^^^^^^^^^^^^^^^
|
= note: failed to mmap file '/<<PKGBUILDDIR>>/build/x86_64-unknown-linux-gnux32/stage1/lib/rustlib/x86_64-unknown-linux-gnux32/lib/librustc_driver-55ecdc7db120b4be.so': Cannot allocate memory (os error 12)
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
