Rustup to rustc 1.28.0-nightly (952f344cd 2018-05-18)

[bjorn3]

No description provided.

[oli-obk]

run-pass/closures is failing:

librustc_mir/interpret/eval_context.rs:1111: value_to_primval can't work with fat pointers

[bjorn3]

Backtrace:

[...]
  13: rustc::session::bug_fmt
  14: <rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::value_to_primval
  15: rustc_mir::interpret::operator::<impl rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::intrinsic_with_overflow
  16: <rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::eval_rvalue_into_place
  17: rustc_mir::interpret::step::<impl rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::step
  18: miri::eval_main
  19: <miri::MiriCompilerCalls as rustc_driver::CompilerCalls<'a>>::build_controller::{{closure}}
[...]

Doesn't seem to be related to the changes in this PR

[oli-obk]

Indeed. I'll see what I can do on the rustc side

[oli-obk]

Maybe related : rust-lang/rust#50706

[eddyb]

This argument should be renamed codegen_backend.

[oli-obk]

There's a bunch of new failures: Size::from_bytes: 18446744073709551614 bytes in bits doesn't fit in u64

[bjorn3]

Those failures are inside rustc:

   7: rustc_target::abi::Size::from_bytes
   8: rustc::mir::interpret::value::Pointer::wrapping_signed_offset
   9: <rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, miri::Evaluator<'tcx>> as miri::helpers::EvalContextExt<'tcx>>::wrapping_pointer_offset
  10: <rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, miri::Evaluator<'tcx>> as miri::intrinsic::EvalContextExt<'tcx>>::call_intrinsic
  11: rustc_mir::interpret::terminator::<impl rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::eval_fn_call
  12: rustc_mir::interpret::terminator::<impl rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::eval_terminator
  13: rustc_mir::interpret::step::<impl rustc_mir::interpret::eval_context::EvalContext<'a, 'mir, 'tcx, M>>::step
  14: miri::eval_main

[bjorn3]

[bjorn3]

Submitted rust-lang/rust#50917

[oli-obk]

We need to allow pointers one past the end of an allocation:

pointer computed at offset 10, outside bounds of allocation 462 which has size 9

[RalfJung]

Nope, if the size is 9 then 8 is the last dereferencable offset and 10 is two past the end.

[bjorn3]

rust-lang/rust#50967 required a huge amount of renaming and I still haven't got the majority of the tests passing after several hours :(

[bjorn3]

Blocked on: rust-lang/rust#51086

[bjorn3]

Closing in favour of #379