To enforce uniformity for all PRs, the CI checks if the code
is formatted rigth using `cargo fmt` tool.
Signed-off-by: Costin-Robert Sin <sin.costinrobert@gmail.com>
The existing AIO implementation has some problems:
1) The in_progress field is checked at runtime, not compile time.
2) The mutable field is checked at runtime, not compile time.
3) A downstream lio_listio user must store extra state to track whether
the whole operation is partially, completely, or not at all
submitted.
4) Nix does heap allocation itself, rather than allowing the caller to
choose it. This can result in double (or triple, or quadruple)
boxing.
5) There's no easy way to use lio_listio to submit multiple operations with
a single syscall, but poll each individually.
6) The lio_listio usage is far from transparent and zero-cost.
7) No aio_readv or aio_writev support.
8) priority has type c_int; should be i32
9) aio_return should return a usize instead of an isize, since it only
uses negative values to indicate errors, which Rust represents via
the Result type.
This rewrite solves several problems:
1) Unsolved. I don't think it can be solved without something like
C++'s guaranteed type elision. It might require changing the
signature of Future::poll too.
2) Solved.
3) Solved, by the new in_progress method and by removing the complicated
lio_listio resubmit code.
4) Solved.
5) Solved.
6) Solved, by removing the lio_listo resubmit code. It can be
reimplemented downstream if necessary. Or even in Nix, but it
doesn't fit Nix's theme of zero-cost abstractions.
7) Solved.
8) Solved.
9) Solved.
The rewrite includes functions that don't work on FreeBSD, so add CI
testing for FreeBSD 14 too.
By default only enable tests that will pass on FreeBSD 12.3. But run a
CI job on FreeBSD 14 and set a flag that will enable such tests.
* Install cross the easy way, via cargo
* Don't test in release mode. Nix contains no release-dependent paths,
and release mode testing has to my knowledge never revealed a bug in
Nix.
* Add Linux powerpc back to CI, fixed by the latest cross.
* Check the tests even on platforms that can't run them.
* DRY for the Illumos and Redox sections
* Cross-check iOS from a Linux VM instead of OSX
* Revert the workaround for rust-lang/rustup issue 2774
* I never removed 32-bit OSX and iOS targets that were removed from
Cirrus in PR #1492 .
* I never added Fuchsia (PR #1285) .
* I never added illumos (PR #1394) .
* never added Linux x32 (PR #1366) .
Travis has been super-slow lately (> 6 hours per build). Cirrus is much
faster: about 20 minutes. Cirrus also has slightly better test
coverage, mainly because it doesn't use SECCOMP.
Also,
* Fix the Redox CI build. The old Travis configuration didn't actually
build for Redox, so we never noticed that Redox can't be built with a
stable compiler. Thanks to @coolreader18 for finding this.
* Disable the udp_offload tests on cross-tested platforms. These tests
are failing with ENOPROTOOPT in Cirrus-CI. I suspect it's due to a
lack of support in QEMU. These tests were skipped on Travis because
its kernel was too old.
* Fix require_kernel_version on Cirrus-CI. Cirrus reports the Linux
kernel version as 4.19.112+, which the semver crate can't handle.
* Fix test_setfsuid on Cirrus. When run on Cirrus, it seems like the
file in /tmp gets deleted as soon as it's closed. Probably an
overzealous temporary file cleaner. Use /var/tmp, because no
temporary file cleaner should run in there.
* Skip mount tests on Cirrus. They fail for an unknown reason.
Issue #1351
* Skip the AF_ALG tests on Cirrus-CI
Issue #1352
This change does the following:
1) Adds a CI build on cirrus-ci.com
2) Switches FreeBSD's CI from a jail to a full VM
3) Switches FreeBSD i386's CI from a 32-bit jail to simply using a
cross-compiled binary on a 64-bit VM.
4) Switches FreeBSD i386's CI from using stable rust to 1.24.1 (back
when I added buildbot, rustup didn't support i686-unknown-freebsd)
5) Switches bors to gate on cirrus-ci rather than buildbot
This change does _not_ disable buildbot. That must happen outside of
git.
Fixes#996
bors's timeout should always be twice a long as the test suite takes.
This is to allow Travis to fast-fail a test; if one of the builders
immediately reports a failure, then bors will move on to the next batch,
leaving the slower builders to work through the already-doomed run and
the next one.
At the time this was written, nix's test suite took about an hour to run.
The timeout was raised to four hours, instead of two, to give nix room
to grow and time for delays on Travis's end.
