Closed tiif closed 3 weeks ago
Initially there is a test, but it took way too long for ./miri test
to finish, so I removed it. But surprisingly, it completed fairly fast with ./miri run --dep ./tests/pass-dep/concurrency/linux-futex.rs
.
Test:
// Reproduce https://github.com/rust-lang/miri/issues/3647. This should not ICE.
fn large_timeout() {
let futex: i32 = 123;
unsafe {
libc::syscall(
libc::SYS_futex,
addr_of!(futex),
libc::FUTEX_WAIT,
123,
&libc::timespec { tv_sec: 184467440839020, tv_nsec: 117558982 },
);
}
}
The linux-futex
test became very slow after this new commit?
@rustbot author
@rustbot ready
Looks good, thanks! @bors r+
:pushpin: Commit f2c5071b09aef74906c202c7617b20111b563c0b has been approved by RalfJung
It is now in the queue for this repository.
:hourglass: Testing commit f2c5071b09aef74906c202c7617b20111b563c0b with merge e09827881fb388b8012a6c3c021d357ead61507d...
Thanks, but no squashing? The commit messages are not very neat.
:sunny: Test successful - checks-actions Approved by: RalfJung Pushing e09827881fb388b8012a6c3c021d357ead61507d to master...
Ah yeah, I forgot about squashing oops
Hmm how the unmerge works? Should I open a new pr with the squashed commits?
No, it's too late. Once this landed it can never be un-done.
Fixes #3647.
This PR changed the type of
nanoseconds
fromu64
tou128
. Induration_since
, nanoseconds is manually converted to second by dividing it with 1e9. But overflow is still possible.