add basic support for android targets

README.md

@@ -214,6 +214,9 @@ degree documented below):
   supported on Windows. We also test `i686-pc-windows-msvc`, with the same
   reduced feature set. We might ship Miri with a nightly even when some features
   on these targets regress.
+- `aarch64-linux-android` has basic support and should work for simple targets.
+  Some concurrency and filesystem operations are not supported and will fail at
+  runtime.
 
 ### Common Problems
 

ci.sh

@@ -21,6 +21,7 @@ function run_tests {
   fi
 
   ./miri test --locked
+
   if [ -z "${MIRI_TEST_TARGET+exists}" ]; then
     # Only for host architecture: tests with optimizations (`-O` is what cargo passes, but crank MIR
     # optimizations up all the way).
@@ -50,6 +51,8 @@ case $HOST_TARGET in
     MIRI_TEST_TARGET=i686-unknown-linux-gnu run_tests
     MIRI_TEST_TARGET=aarch64-apple-darwin run_tests
     MIRI_TEST_TARGET=i686-pc-windows-msvc run_tests
+    # TODO: Enable once https://github.com/rust-lang/rust/pull/94813 lands
+    # MIRI_TEST_TARGET=aarch64-linux-android run_tests
     ;;
   x86_64-apple-darwin)
     MIRI_TEST_TARGET=mips64-unknown-linux-gnuabi64 run_tests # big-endian architecture

src/helpers.rs

@@ -493,6 +493,28 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
         )
     }
 
+    /// Helper function used inside the shims of foreign functions to assert that the target OS
+    /// is based on the Linux kernel. It panics showing a message with the `name` of the foreign function
+    /// if this is not the case.
+    fn assert_target_os_is_linux_based(&self, name: &str) {
+        assert!(
+            matches!(self.eval_context_ref().tcx.sess.target.os.as_ref(), "linux" | "android"),
+            "`{}` is only available for supported Linux-based targets",
+            name,
+        )
+    }
+
+    /// Helper function used inside the shims of foreign functions to assert that the target OS
+    /// is part of the UNIX family. It panics showing a message with the `name` of the foreign function
+    /// if this is not the case.
+    fn assert_target_os_is_unix(&self, name: &str) {
+        assert!(
+            target_os_is_unix(self.eval_context_ref().tcx.sess.target.os.as_ref()),
+            "`{}` is only available for supported UNIX family targets",
+            name,
+        );
+    }
+
     /// Get last error variable as a place, lazily allocating thread-local storage for it if
     /// necessary.
     fn last_error_place(&mut self) -> InterpResult<'tcx, MPlaceTy<'tcx, Tag>> {
@@ -883,3 +905,9 @@ impl std::fmt::Display for HexRange {
         write!(f, "[{:#x}..{:#x}]", self.0.start.bytes(), self.0.end().bytes())
     }
 }
+
+/// Helper function used inside the shims of foreign functions to check that the `target_os` is
+/// part of the UNIX family.
+pub fn target_os_is_unix(target_os: &str) -> bool {
+    matches!(target_os, "linux" | "macos" | "android")
+}

src/machine.rs

@@ -430,6 +430,41 @@ impl<'mir, 'tcx> Evaluator<'mir, 'tcx> {
                 this.write_scalar(Scalar::from_u8(0), &place.into())?;
                 Self::add_extern_static(this, "_tls_used", place.ptr);
             }
+            "android" => {
+                // "environ"
+                Self::add_extern_static(
+                    this,
+                    "environ",
+                    this.machine.env_vars.environ.unwrap().ptr,
+                );
+                // A couple zero-initialized pointer-sized extern statics.
+                // Most of them are for weak symbols, which we all set to null (indicating that the
+                // symbol is not supported, and triggering fallback code which ends up calling a
+                // syscall that we do support).
+                for name in &["__cxa_thread_atexit_impl", "getrandom", "statx"] {
+                    let layout = this.machine.layouts.usize;
+                    let place = this.allocate(layout, MiriMemoryKind::ExternStatic.into())?;
+                    this.write_scalar(Scalar::from_machine_usize(0, this), &place.into())?;
+                    Self::add_extern_static(this, name, place.ptr);
+                }
+                for symbol_name in &["signal", "bsd_signal"] {
+                    let layout = this.machine.layouts.mut_raw_ptr;
+                    let dlsym = Dlsym::from_str(symbol_name.as_bytes(), &this.tcx.sess.target.os)?
+                        .unwrap_or_else(|| {
+                            panic!(
+                                "hardcoded `extern static` symbol {} has no dlsym handler",
+                                symbol_name
+                            )
+                        });
+                    let ecx = this.eval_context_mut();
+                    let ptr = ecx.create_fn_alloc_ptr(FnVal::Other(dlsym));
+
+                    let place = this.allocate(layout, MiriMemoryKind::ExternStatic.into())?;
+                    this.write_pointer(ptr, &place.into())?;
+
+                    Self::add_extern_static(this, symbol_name, place.ptr);
+                }
+            }
             _ => {} // No "extern statics" supported on this target
         }
         Ok(())

src/shims/dlsym.rs

@@ -1,6 +1,7 @@
 use rustc_middle::mir;
 use rustc_target::spec::abi::Abi;
 
+use crate::helpers::target_os_is_unix;
 use crate::*;
 use shims::posix::dlsym as posix;
 use shims::windows::dlsym as windows;
@@ -18,7 +19,8 @@ impl Dlsym {
     pub fn from_str<'tcx>(name: &[u8], target_os: &str) -> InterpResult<'tcx, Option<Dlsym>> {
         let name = &*String::from_utf8_lossy(name);
         Ok(match target_os {
-            "linux" | "macos" => posix::Dlsym::from_str(name, target_os)?.map(Dlsym::Posix),
+            target if target_os_is_unix(target) =>
+                posix::Dlsym::from_str(name, target_os)?.map(Dlsym::Posix),
             "windows" => windows::Dlsym::from_str(name)?.map(Dlsym::Windows),
             os => bug!("dlsym not implemented for target_os {}", os),
         })

src/shims/env.rs

@@ -8,6 +8,7 @@ use rustc_data_structures::fx::FxHashMap;
 use rustc_middle::ty::layout::LayoutOf;
 use rustc_target::abi::Size;
 
+use crate::helpers::target_os_is_unix;
 use crate::*;
 
 /// Check whether an operation that writes to a target buffer was successful.
@@ -55,7 +56,7 @@ impl<'tcx> EnvVars<'tcx> {
                 };
                 if forward {
                     let var_ptr = match target_os {
-                        "linux" | "macos" =>
+                        target if target_os_is_unix(target) =>
                             alloc_env_var_as_c_str(name.as_ref(), value.as_ref(), ecx)?,
                         "windows" => alloc_env_var_as_wide_str(name.as_ref(), value.as_ref(), ecx)?,
                         unsupported =>
@@ -113,11 +114,7 @@ impl<'mir, 'tcx: 'mir> EvalContextExt<'mir, 'tcx> for crate::MiriEvalContext<'mi
 pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx> {
     fn getenv(&mut self, name_op: &OpTy<'tcx, Tag>) -> InterpResult<'tcx, Pointer<Option<Tag>>> {
         let this = self.eval_context_mut();
-        let target_os = &this.tcx.sess.target.os;
-        assert!(
-            target_os == "linux" || target_os == "macos",
-            "`getenv` is only available for the UNIX target family"
-        );
+        this.assert_target_os_is_unix("getenv");
 
         let name_ptr = this.read_pointer(name_op)?;
         let name = this.read_os_str_from_c_str(name_ptr)?;
@@ -211,11 +208,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
         value_op: &OpTy<'tcx, Tag>,
     ) -> InterpResult<'tcx, i32> {
         let this = self.eval_context_mut();
-        let target_os = &this.tcx.sess.target.os;
-        assert!(
-            target_os == "linux" || target_os == "macos",
-            "`setenv` is only available for the UNIX target family"
-        );
+        this.assert_target_os_is_unix("setenv");
 
         let name_ptr = this.read_pointer(name_op)?;
         let value_ptr = this.read_pointer(value_op)?;
@@ -285,11 +278,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
 
     fn unsetenv(&mut self, name_op: &OpTy<'tcx, Tag>) -> InterpResult<'tcx, i32> {
         let this = self.eval_context_mut();
-        let target_os = &this.tcx.sess.target.os;
-        assert!(
-            target_os == "linux" || target_os == "macos",
-            "`unsetenv` is only available for the UNIX target family"
-        );
+        this.assert_target_os_is_unix("unsetenv");
 
         let name_ptr = this.read_pointer(name_op)?;
         let mut success = None;
@@ -319,11 +308,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
         size_op: &OpTy<'tcx, Tag>,
     ) -> InterpResult<'tcx, Pointer<Option<Tag>>> {
         let this = self.eval_context_mut();
-        let target_os = &this.tcx.sess.target.os;
-        assert!(
-            target_os == "linux" || target_os == "macos",
-            "`getcwd` is only available for the UNIX target family"
-        );
+        this.assert_target_os_is_unix("getcwd");
 
         let buf = this.read_pointer(buf_op)?;
         let size = this.read_scalar(size_op)?.to_machine_usize(&*this.tcx)?;
@@ -378,11 +363,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
 
     fn chdir(&mut self, path_op: &OpTy<'tcx, Tag>) -> InterpResult<'tcx, i32> {
         let this = self.eval_context_mut();
-        let target_os = &this.tcx.sess.target.os;
-        assert!(
-            target_os == "linux" || target_os == "macos",
-            "`getcwd` is only available for the UNIX target family"
-        );
+        this.assert_target_os_is_unix("chdir");
 
         let path = this.read_path_from_c_str(this.read_pointer(path_op)?)?;
 

src/shims/foreign_items.rs

@@ -23,6 +23,7 @@ use rustc_target::{
 
 use super::backtrace::EvalContextExt as _;
 use crate::helpers::convert::Truncate;
+use crate::helpers::target_os_is_unix;
 use crate::*;
 
 /// Returned by `emulate_foreign_item_by_name`.
@@ -700,9 +701,8 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
                 }
             }
 
-            // Platform-specific shims
             _ => match this.tcx.sess.target.os.as_ref() {
-                "linux" | "macos" => return shims::posix::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret),
+                target if target_os_is_unix(target) => return shims::posix::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret),
                 "windows" => return shims::windows::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret),
                 target => throw_unsup_format!("the target `{}` is not supported", target),
             }

src/shims/posix/android/dlsym.rs

@@ -0,0 +1,53 @@
+use helpers::check_arg_count;
+use log::trace;
+use rustc_middle::mir;
+
+use crate::*;
+
+#[derive(Debug, Copy, Clone)]
+#[allow(non_camel_case_types)]
+pub enum Dlsym {
+    signal,
+}
+
+impl Dlsym {
+    // Returns an error for unsupported symbols, and None if this symbol
+    // should become a NULL pointer (pretend it does not exist).
+    pub fn from_str<'tcx>(name: &str) -> InterpResult<'tcx, Option<Dlsym>> {
+        Ok(match &*name {
+            "__pthread_get_minstack" => None,
+            "getrandom" => None, // std falls back to syscall(SYS_getrandom, ...) when this is NULL.
+            "statx" => None,     // std falls back to syscall(SYS_statx, ...) when this is NULL.
+            "signal" | "bsd_signal" => Some(Dlsym::signal), // these have the same signature/implementation
+            "android_set_abort_message" => None, // std falls back to just not doing anything when this is NULL.
+            _ => throw_unsup_format!("unsupported Android dlsym: {}", name),
+        })
+    }
+}
+
+impl<'mir, 'tcx: 'mir> EvalContextExt<'mir, 'tcx> for crate::MiriEvalContext<'mir, 'tcx> {}
+pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx> {
+    fn call_dlsym(
+        &mut self,
+        dlsym: Dlsym,
+        args: &[OpTy<'tcx, Tag>],
+        dest: &PlaceTy<'tcx, Tag>,
+        ret: Option<mir::BasicBlock>,
+    ) -> InterpResult<'tcx> {
+        let this = self.eval_context_mut();
+        assert!(this.tcx.sess.target.os == "android");
+
+        let ret = ret.expect("we don't support any diverging dlsym");
+
+        match dlsym {
+            Dlsym::signal => {
+                let &[ref _sig, ref _func] = check_arg_count(args)?;
+                this.write_null(dest)?;
+            }
+        }
+
+        trace!("{:?}", this.dump_place(**dest));
+        this.go_to_block(ret);
+        Ok(())
+    }
+}

src/shims/posix/android/foreign_items.rs

@@ -0,0 +1,28 @@
+use rustc_middle::mir;
+use rustc_span::Symbol;
+use rustc_target::spec::abi::Abi;
+
+use crate::*;
+use shims::foreign_items::EmulateByNameResult;
+
+impl<'mir, 'tcx: 'mir> EvalContextExt<'mir, 'tcx> for crate::MiriEvalContext<'mir, 'tcx> {}
+pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx> {
+    fn emulate_foreign_item_by_name(
+        &mut self,
+        link_name: Symbol,
+        abi: Abi,
+        args: &[OpTy<'tcx, Tag>],
+        dest: &PlaceTy<'tcx, Tag>,
+        ret: mir::BasicBlock,
+    ) -> InterpResult<'tcx, EmulateByNameResult<'mir, 'tcx>> {
+        let this = self.eval_context_mut();
+
+        match &*link_name.as_str() {
+            _ => {
+                // By default we piggyback off the items emulated for Linux. For now this functions
+                // serves to make adding android-specific syscalls easy
+                return shims::posix::linux::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret);
+            }
+        }
+    }
+}

src/shims/posix/android/mod.rs

@@ -0,0 +1,2 @@
+pub mod dlsym;
+pub mod foreign_items;

src/shims/posix/dlsym.rs

@@ -2,13 +2,15 @@ use rustc_middle::mir;
 use rustc_target::spec::abi::Abi;
 
 use crate::*;
+use shims::posix::android::dlsym as android;
 use shims::posix::linux::dlsym as linux;
 use shims::posix::macos::dlsym as macos;
 
 #[derive(Debug, Copy, Clone)]
 pub enum Dlsym {
     Linux(linux::Dlsym),
     MacOs(macos::Dlsym),
+    Android(android::Dlsym),
 }
 
 impl Dlsym {
@@ -18,6 +20,7 @@ impl Dlsym {
         Ok(match target_os {
             "linux" => linux::Dlsym::from_str(name)?.map(Dlsym::Linux),
             "macos" => macos::Dlsym::from_str(name)?.map(Dlsym::MacOs),
+            "android" => android::Dlsym::from_str(name)?.map(Dlsym::Android),
             _ => unreachable!(),
         })
     }
@@ -40,6 +43,8 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
         match dlsym {
             Dlsym::Linux(dlsym) => linux::EvalContextExt::call_dlsym(this, dlsym, args, dest, ret),
             Dlsym::MacOs(dlsym) => macos::EvalContextExt::call_dlsym(this, dlsym, args, dest, ret),
+            Dlsym::Android(dlsym) =>
+                android::EvalContextExt::call_dlsym(this, dlsym, args, dest, ret),
         }
     }
 }

src/shims/posix/foreign_items.rs

@@ -168,6 +168,23 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
                 }
                 this.write_null(dest)?;
             }
+            "memalign" => {
+                let &[ref align, ref size] = this.check_shim(abi, Abi::C { unwind: false }, link_name, args)?;
+                let align = this.read_scalar(align)?.to_machine_usize(this)?;
+                let size = this.read_scalar(size)?.to_machine_usize(this)?;
+
+                // Align must be power of 2.
+                if !align.is_power_of_two() {
+                    throw_ub_format!("memalign: alignment must be a power of two, but is {}", align);
+                }
+
+                let ptr = this.allocate_ptr(
+                    Size::from_bytes(size),
+                    Align::from_bytes(align).unwrap(),
+                    MiriMemoryKind::C.into(),
+                )?;
+                this.write_pointer(ptr, dest)?;
+            }
 
             // Dynamic symbol loading
             "dlsym" => {
@@ -479,6 +496,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
                 match this.tcx.sess.target.os.as_ref() {
                     "linux" => return shims::posix::linux::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret),
                     "macos" => return shims::posix::macos::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret),
+                    "android" => return shims::posix::android::foreign_items::EvalContextExt::emulate_foreign_item_by_name(this, link_name, abi, args, dest, ret),
                     _ => unreachable!(),
                 }
             }

src/shims/posix/fs.rs

@@ -976,7 +976,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
     ) -> InterpResult<'tcx, i32> {
         let this = self.eval_context_mut();
 
-        this.assert_target_os("linux", "statx");
+        this.assert_target_os_is_linux_based("statx");
 
         let dirfd = this.read_scalar(dirfd_op)?.to_i32()?;
         let pathname_ptr = this.read_pointer(pathname_op)?;
@@ -1278,7 +1278,7 @@ pub trait EvalContextExt<'mir, 'tcx: 'mir>: crate::MiriEvalContextExt<'mir, 'tcx
     fn linux_readdir64(&mut self, dirp_op: &OpTy<'tcx, Tag>) -> InterpResult<'tcx, Scalar<Tag>> {
         let this = self.eval_context_mut();
 
-        this.assert_target_os("linux", "readdir64");
+        this.assert_target_os_is_linux_based("readdir64");
 
         let dirp = this.read_scalar(dirp_op)?.to_machine_usize(this)?;