miri no longer builds after rust-lang/rust#62335

[rust-highfive]

Hello, this is your friendly neighborhood mergebot.
After merging PR #62335, I observed that the tool miri has failing tests.
A follow-up PR to the repository https://github.com/rust-lang/miri is needed to fix the fallout.

cc @Mark-Simulacrum, do you think you would have time to do the follow-up work?
If so, that would be great!

cc @Mark-Simulacrum, the PR reviewer, and @rust-lang/compiler -- nominating for prioritization.

[Mark-Simulacrum]

2019-07-03T17:22:14.8438802Z normalized stderr:
2019-07-03T17:22:15.0846046Z error[E0080]: Miri evaluation error: type validation failed: encountered 65543, but expected a pointer
2019-07-03T17:22:15.0936029Z   --> D:/a/1/s/src/libstd/sys/windows/compat.rs:68:13
2019-07-03T17:22:15.0936200Z    |
2019-07-03T17:22:15.0936318Z 68 |             mem::transmute::<usize, F>(addr)($($argname),*)
2019-07-03T17:22:15.0936484Z    |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Miri evaluation error: type validation failed: encountered 65543, but expected a pointer
2019-07-03T17:22:15.0936579Z    |
2019-07-03T17:22:15.0936705Z    = note: inside call to `std::sys::windows::c::SetThreadStackGuarantee` at D:/a/1/s/src/libstd/sys/windows/stack_overflow.rs:12:12
2019-07-03T17:22:15.0937192Z    = note: inside call to `std::sys::windows::stack_overflow::Handler::new` at D:/a/1/s/src/libstd/sys/windows/stack_overflow.rs:39:14
2019-07-03T17:22:15.0937351Z    = note: inside call to `std::sys::windows::stack_overflow::init` at D:/a/1/s/src/libstd/rt.rs:34:9
2019-07-03T17:22:15.0937457Z    = note: inside call to `std::rt::lang_start_internal` at D:/a/1/s/src/libstd/rt.rs:64:5
2019-07-03T17:22:15.0937584Z    = note: inside call to `std::rt::lang_start::<()>`
2019-07-03T17:22:15.0937824Z 
2019-07-03T17:22:15.0937917Z error: aborting due to previous error
2019-07-03T17:22:15.0937963Z 
2019-07-03T17:22:15.0938062Z For more information about this error, try `rustc --explain E0080`.
2019-07-03T17:22:15.0938125Z 
2019-07-03T17:22:15.0938158Z 
2019-07-03T17:22:15.0938248Z expected stderr:
2019-07-03T17:22:15.0938290Z 
2019-07-03T17:22:15.0938324Z 
2019-07-03T17:22:15.0938402Z diff of stderr:
2019-07-03T17:22:15.0938445Z 
2019-07-03T17:22:15.0938528Z +error[E0080]: Miri evaluation error: type validation failed: encountered 65543, but expected a pointer
2019-07-03T17:22:15.0938660Z +  --> D:/a/1/s/src/libstd/sys/windows/compat.rs:68:13
2019-07-03T17:22:15.0938764Z +   |
2019-07-03T17:22:15.0938835Z +68 |             mem::transmute::<usize, F>(addr)($($argname),*)
2019-07-03T17:22:15.0938977Z +   |             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Miri evaluation error: type validation failed: encountered 65543, but expected a pointer
2019-07-03T17:22:15.0939067Z +   |
2019-07-03T17:22:15.0939195Z +   = note: inside call to `std::sys::windows::c::SetThreadStackGuarantee` at D:/a/1/s/src/libstd/sys/windows/stack_overflow.rs:12:12
2019-07-03T17:22:15.0939426Z +   = note: inside call to `std::sys::windows::stack_overflow::Handler::new` at D:/a/1/s/src/libstd/sys/windows/stack_overflow.rs:39:14
2019-07-03T17:22:15.0939573Z +   = note: inside call to `std::sys::windows::stack_overflow::init` at D:/a/1/s/src/libstd/rt.rs:34:9
2019-07-03T17:22:15.0939677Z +   = note: inside call to `std::rt::lang_start_internal` at D:/a/1/s/src/libstd/rt.rs:64:5
2019-07-03T17:22:15.0939846Z +   = note: inside call to `std::rt::lang_start::<()>`
2019-07-03T17:22:15.0939916Z +
2019-07-03T17:22:15.0940003Z +error: aborting due to previous error
2019-07-03T17:22:15.0940069Z +
2019-07-03T17:22:15.0940177Z +For more information about this error, try `rustc --explain E0080`.
2019-07-03T17:22:15.0940250Z +

from https://dev.azure.com/rust-lang/e71b0ddf-dd27-435a-873c-e30f86eea377/_apis/build/builds/2278/logs/929

[RalfJung]

Looks like a bunch of tests are affected the same way:

2019-07-03T17:22:44.0216742Z failures:
2019-07-03T17:22:44.0216890Z     [ui] run-pass\hashmap.rs
2019-07-03T17:22:44.0217073Z     [ui] run-pass\heap_allocator.rs
2019-07-03T17:22:44.0217242Z     [ui] run-pass\intptrcast.rs
2019-07-03T17:22:44.0217414Z     [ui] run-pass\intptrcast_format.rs

[RalfJung]

Cc @christianpoveda this is likely caused by your change, somehow.

[RalfJung]

Likely caused by a missing force_ptr in function pointer validation. Should be fixed by #62245.

[pvdrz]

I merged @RalfJung's branches with my branches in both the rust and miri repos and all the tests passed :D

[pnkfelix]

triage: P-medium, removing nomination, assigning already assigned to @RalfJung

[RalfJung]

Miri upstream CI passes now, but only because I disabled a bunch of tests. I felt that was the better alternative, compared to not being able to land any update in Miri until a fix happens on the rustc side.

But this means we shouldn't update the in-tree Miri tool yet.

[RalfJung]

PR fixing things for Miri: rust-lang/miri#830. Blocked on #62441.

[RalfJung]

@lzutao Miri master got broken again, CI is failing. Do you want to try to fix that?

[tesuji]

I believe #62403 broke miri. Hopefully I get time to fix it?

[RalfJung]

Should be an easy fix by making Miri not use that trait explicitly. It's yours!

[RalfJung]

Hooray, #62441 landed! Looks like this is blocked on a serde update regression now, judging from https://travis-ci.com/rust-lang/miri/jobs/214779163.

EDIT: blocked on #62562, actually.