integer overflow in integration::subkey_backsig test on 32-bit architectures
We are running cargo test --release
during Fedora package builds, and I have now encountered an test failure on 32-bit architectures (i686 and armv7hl):
running 18 tests
[...]
test integration::subkey_backsig ... FAILED
[...]
failures:
---- integration::subkey_backsig stdout ----
Linting sha1-backsig-signing-subkey-pub.pgp
thread 'integration::subkey_backsig' panicked at 'Unexpected return code, failed var == 2
command=`"/builddir/build/BUILD/sequoia-keyring-linter-0.5.0/target/release/sq-keyring-linter" "sha1-backsig-signing-subkey-pub.pgp"`
code=101
stdout=```""```
stderr=```"thread \'main\' panicked at \'overflow when adding duration to instant\', library/std/src/time.rs:539:31\nnote: run with `RUST_BACKTRACE=1` environment variable to display a backtrace\n"```
', /usr/share/cargo/registry/assert_cmd-1.0.3/src/assert.rs:258:13
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
failures:
integration::subkey_backsig
test result: FAILED. 17 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.24s
error: test failed, to rerun pass '--test tests'
Looks like there's an integer overflow in time calculation somewhere - this is handled by a warning and wrapping around the result in debug mode, but will cause a panic in release mode.
Is this overflow of concern, or can I ignore this failure on 32-bit architectures? If it helps, I can try to run tests with RUST_BACKTRACE=1
.