Use `miri` inside the target directory used by rustc as Miri's target directory
Resolves#1311.
This PR makes Miri use `miri` inside the rustc target directory as its target directory, by letting `cargo-miri` get the rustc target directory by calling `cargo metadata`, append `miri` to it, and pass it with `--target-dir` to Cargo.
Getting the rustc target directory accurately requires calling `cargo metadata` as far as I know, because the `target-dir` can be set in config files in various places that are hard for `cargo-miri` to find.
I also considered https://doc.rust-lang.org/nightly/cargo/reference/unstable.html#custom-named-profiles, but it looks like that requires adding `cargo-features = ["named-profiles"]` to **`Cargo.toml`**, which would be tricky for `cargo-miri`:
```
$ cargo +nightly-2021-06-23 test --config 'profile.miri.inherits="release"' --profile=miri -Z named-profiles -Z unstable-options
error: config profile `miri` is not valid (defined in `--config cli option`)
Caused by:
feature `named-profiles` is required
consider adding `cargo-features = ["named-profiles"]` to the manifest
```
Filter out items other than non-generic functions and statics in our version of `exported_symbols`
[`#[no_mangle]` on a `use` item](https://docs.rs/brotli-decompressor/2.3.1/src/brotli_decompressor/ffi/mod.rs.html#3-5) can make Miri ICE when compiling a dependency (rust-lang/rust#86261):
```rs
#[no_mangle]
use std::{thread,panic, io, boxed, any, string};
```
<details>
```
error: internal compiler error: compiler/rustc_middle/src/ty/mod.rs:1650:13: item_name: no name for DefPath { data: [DisambiguatedDefPathData { data: Misc, disambiguator: 14 }], krate: crate0 }
thread 'rustc' panicked at 'Box<dyn Any>', compiler/rustc_errors/src/lib.rs:1007:9
stack backtrace:
0: std::panicking::begin_panic
1: std::panic::panic_any
2: rustc_errors::HandlerInner::bug
3: rustc_errors::Handler::bug
4: rustc_middle::ty::context::tls::with_opt
5: rustc_middle::util:🐛:opt_span_bug_fmt
6: rustc_middle::util:🐛:bug_fmt
7: rustc_middle::ty::<impl rustc_middle::ty::context::TyCtxt>::item_name
8: rustc_symbol_mangling::symbol_name_provider
9: rustc_query_impl::<impl rustc_query_system::query::config::QueryAccessors<rustc_query_impl::plumbing::QueryCtxt> for rustc_query_impl::queries::symbol_name>::compute
10: rustc_query_system::query::plumbing::get_query_impl
11: <rustc_query_impl::Queries as rustc_middle::ty::query::QueryEngine>::symbol_name
12: rustc_middle::middle::exported_symbols::ExportedSymbol::symbol_name_for_local_instance
13: rustc_codegen_ssa:🔙:symbol_export::symbol_name_for_instance_in_crate
14: rustc_codegen_ssa:🔙:linker::exported_symbols
15: <core::iter::adapters::map::Map<I,F> as core::iter::traits::iterator::Iterator>::fold
16: rustc_codegen_ssa:🔙:linker::LinkerInfo::new
17: rustc_codegen_ssa:🔙:write::start_async_codegen
18: <rustc_codegen_llvm::LlvmCodegenBackend as rustc_codegen_ssa::traits::backend::CodegenBackend>::codegen_crate
19: rustc_interface::passes::QueryContext::enter
20: rustc_interface::queries::Queries::ongoing_codegen
21: rustc_interface::queries::<impl rustc_interface::interface::Compiler>::enter
22: rustc_span::with_source_map
23: rustc_interface::interface::create_compiler_and_run
24: rustc_span::with_session_globals
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
note: the compiler unexpectedly panicked. this is a bug.
note: we would appreciate a bug report: https://github.com/rust-lang/rust/issues/new?labels=C-bug%2C+I-ICE%2C+T-compiler&template=ice.md
note: rustc 1.54.0-nightly (a50d72158 2021-06-08) running on x86_64-unknown-linux-gnu
note: compiler flags: -C embed-bitcode=no -C debuginfo=1 --crate-type lib
note: some of the compiler flags provided by cargo are hidden
query stack during panic:
#0 [symbol_name] computing the symbol for `{misc#14}`
end of query stack
```
</details>
This might be because in #1776, we override the `exported_symbols` query, and our version of `exported_symbols` can return a `use` item which don't have a name if the `use` item is tagged with `#[no_mangle]`, and then:
- `rustc_codegen_ssa:🔙:symbol_export::symbol_name_for_instance_in_crate` is called for for every `exported_symbols`: fb3ea63d9b/compiler/rustc_codegen_ssa/src/back/linker.rs (L1300-L1304)
- it calls `rustc_middle::middle::exported_symbols::ExportedSymbol::symbol_name_for_local_instance`: fb3ea63d9b/compiler/rustc_codegen_ssa/src/back/symbol_export.rs (L412)
- which calls `rustc_symbol_mangling::symbol_name_provider`: fb3ea63d9b/compiler/rustc_middle/src/middle/exported_symbols.rs (L37-L44)
- which calls `item_name`: fb3ea63d9b/compiler/rustc_symbol_mangling/src/lib.rs (L216), which triggers the ICE
It might also be problematic for d39f0c64b8/src/shims/foreign_items.rs (L165) which also uses `item_name`, but Miri cannot compile the dependency, so that code can't be reached.
Therefore, this PR makes `exported_symbols` filter out all items that are not functions or statics, so all items returned will have a name, which avoids the ICE (I have tested it in the https://github.com/jorgecarleitao/arrow2 repository).
(This PR also includes a commit that fixes a small (unrelated) bug for `#[no_mangle]` on associated functions -- I found that because I notice `#[no_mangle]` is supported on associated functions and they should not be filtered out in `exported_symbols`.)
Fixes (when the submodule is bumped) rust-lang/rust#86261.
Report an error if a `#[no_mangle]`/`#[export_name = ...]` function has the same symbol name as a built-in shim
Implements https://github.com/rust-lang/miri/pull/1776#issuecomment-821322605.
The error looks like this:
```
error: found `malloc` symbol definition that clashes with a built-in shim
--> tests/compile-fail/function_calls/exported_symbol_shim_clashing.rs:12:9
|
12 | malloc(0);
| ^^^^^^^^^ found `malloc` symbol definition that clashes with a built-in shim
|
help: the `malloc` symbol is defined here
--> tests/compile-fail/function_calls/exported_symbol_shim_clashing.rs:2:1
|
2 | / extern "C" fn malloc(_: usize) -> *mut std::ffi::c_void {
3 | | //~^ HELP the `malloc` symbol is defined here
4 | | unreachable!()
5 | | }
| |_^
= note: inside `main` at tests/compile-fail/function_calls/exported_symbol_shim_clashing.rs:12:9
```
This does not implement "better error messages than we do currently for arg/ABI mismatches" in https://github.com/rust-lang/miri/pull/1776#issuecomment-821343175 -- I failed to remove all `check_arg_count()` and `check_abi()` (they are still used in `src/shims/intrinsics.rs` and `call_dlsym()`) and they don't receive the name of the shim.
Fix the wrong `EmulateByNameResult::NotSupported` in `syscall` shim
Without the change, the newly added test will fail with:
```diff
-thread 'main' panicked at 'unsupported Miri functionality: can't execute syscall with ID 0', $DIR/unsupported_syscall.rs:10:9
+thread 'main' panicked at 'unsupported Miri functionality: can't call foreign function: syscall', $DIR/unsupported_syscall.rs:10:9
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
```
cc https://github.com/rust-lang/miri/pull/1818#discussion_r648868937
Add support for panicking in the emulated application when unsupported functionality is encountered
This PR fixes#1807 and allows an optional flag to be specified to panic when an unsupported syscall is encountered. In essence, instead of bubbling up an error in the context of the Miri application Miri will panic within the context of the *emulated* application. This feature is desired to allow CI pipelines to determine if a Miri failure is unsupported functionality or actual UB. Please read [this comment](https://github.com/rust-lang/miri/issues/1807#issuecomment-845425076) for the rationale behind this change.
Note: this change does not cover all cases where unsupported functionality errors may be raised. If you search the repo for `throw_unsup_format!` there are many cases that I think are less likely to occur and may still be problematic for some folks.
TODO:
- [x] README documentation on this new flag
- [x] Add tests
fix: avoid stopping machine upon running env operations in isolation
get and set current dir operations used to halt the machine by
throwing an exception in isolation mode. This change updates them to
return a dummy `NotFound` error instead, and keep the machine running.
I started with a custom error using `ErrorKind::Other`, but since it
can't be mapped to a raw OS error, I dropped it. `NotFound` kind of make
sense for get operations, but not much for set operations. But that's
the only error supported for windows currently.
In user interface, added a new flag `-Zmiri-isolation-error` which
takes one of the four values -- hide, warn, warn-nobacktrace, and
abort. This option can be used to configure Miri to either abort or
return an error code upon executing isolated op. If not aborted, Miri
prints a warning, whose verbosity can be configured using this flag.
In implementation, added a new enum `IsolatedOp` to capture all the
settings related to ops requiring communication with the
host. Old `communicate` flag in both miri configs and machine
stats is replaced with a new helper function `communicate()` which
checks `isolated_op` internally.
Added a new helper function `reject_in_isolation` which can be called
by shims to reject ops according to the reject_with settings. Use miri
specific diagnostics function `report_msg` to print backtrace in the
warning. Update it to take an enum value instead of a bool, indicating
the level of diagnostics.
Updated shims related to current dir to use the new APIs. Added a new
test for current dir ops in isolation without halting machine.
