Add generic sandbox, better DWARF handling, add DOOM

.github/workflows/rust.yml

@@ -18,16 +18,22 @@ jobs:
       run: sudo apt-get install -y lld
     - name: Install target -- i686-unknown-linux-musl
       run: rustup target add i686-unknown-linux-musl
+    - name: Install target -- x86_64-unknown-freebsd
+      run: rustup target add x86_64-unknown-freebsd
     - name: Build and test (generic)
       run: ./ci/jobs/build-and-test.sh
     - name: Build and test (Linux-only)
       run: ./ci/jobs/build-and-test-linux.sh
+    - name: Check (FreeBSD)
+      run: ./ci/jobs/check-freebsd.sh
   build-and-test-macos:
     runs-on: macos-latest
     steps:
     - uses: actions/checkout@v4
     - name: Build and test
       run: ./ci/jobs/build-and-test.sh
+    - name: Build and test (macOS-only)
+      run: ./ci/jobs/build-and-test-macos.sh
   build-and-test-windows:
     runs-on: windows-latest
     steps:

Cargo.toml

@@ -6,13 +6,13 @@ members = [
     "crates/polkavm-assembler",
     "crates/polkavm-common",
     "crates/polkavm-linux-raw",
-    "crates/polkavm-linux-sandbox",
     "crates/polkavm",
 
     "tools/polkatool",
     "tools/polkavm-linux-raw-generate",
 
     "examples/hosts/hello-world",
+    "examples/hosts/doom",
 ]
 
 [workspace.package]
@@ -31,13 +31,13 @@ polkavm-derive          = { version = "0.2.0", path = "crates/polkavm-derive" }
 polkavm-derive-impl     = { version = "0.2.0", path = "crates/polkavm-derive-impl" }
 polkavm-linker          = { version = "0.2.0", path = "crates/polkavm-linker" }
 polkavm-linux-raw       = { version = "0.2.0", path = "crates/polkavm-linux-raw" }
-polkavm-linux-sandbox   = { version = "0.2.0", path = "crates/polkavm-linux-sandbox" }
 
 clap = "4.4.1"
 env_logger = { version = "0.10.0", default-features = false }
 gimli = { version = "0.28.0", default-features = false }
 hashbrown = { version = "0.14.0", default-features = false }
 iced-x86 = "1.19.0"
+libc = "0.2.146"
 log = "0.4.20"
 object = { version = "0.32.0", default-features = false }
 proc-macro2 = "1.0.63"

ci/jobs/build-and-test-linux.sh

@@ -9,5 +9,14 @@ cd crates/polkavm-zygote
 cargo build --release
 cd ../..
 
-echo ">> cargo run (examples, musl)"
-POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 cargo run --target=i686-unknown-linux-musl -p hello-world-host
+echo ">> cargo run (examples, interpreter, i686-unknown-linux-musl)"
+POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 POLKAVM_BACKEND=interpreter cargo run --target=i686-unknown-linux-musl -p hello-world-host
+
+echo ">> cargo run (examples, interpreter, x86_64-unknown-linux-gnu)"
+POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 POLKAVM_BACKEND=interpreter cargo run --target=x86_64-unknown-linux-gnu -p hello-world-host
+
+echo ">> cargo run (examples, compiler, linux, x86_64-unknown-linux-gnu)"
+POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 POLKAVM_BACKEND=compiler POLKAVM_SANDBOX=linux cargo run --target=x86_64-unknown-linux-gnu -p hello-world-host
+
+echo ">> cargo run (examples, compiler, generic, x86_64-unknown-linux-gnu)"
+POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 POLKAVM_BACKEND=compiler POLKAVM_SANDBOX=generic cargo run --target=x86_64-unknown-linux-gnu -p hello-world-host

ci/jobs/build-and-test-macos.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+
+set -euo pipefail
+cd -- "$(dirname -- "${BASH_SOURCE[0]}")"
+cd ../..
+
+echo ">> cargo run (examples, interpreter, x86_64-apple-darwin)"
+POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 POLKAVM_BACKEND=interpreter cargo run --target=x86_64-apple-darwin -p hello-world-host
+
+echo ">> cargo run (examples, compiler, generic, x86_64-apple-darwin)"
+POLKAVM_TRACE_EXECUTION=1 POLKAVM_ALLOW_INSECURE=1 POLKAVM_BACKEND=compiler POLKAVM_SANDBOX=generic cargo run --target=x86_64-apple-darwin -p hello-world-host

ci/jobs/check-freebsd.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+set -euo pipefail
+cd -- "$(dirname -- "${BASH_SOURCE[0]}")"
+cd ../..
+
+echo ">> cargo check (freebsd)"
+cd crates/polkavm
+cargo check --target=x86_64-unknown-freebsd
+cd ../..

ci/run-all-tests.sh

@@ -14,7 +14,14 @@ fi
 case "$OSTYPE" in
   linux*)
     ./ci/jobs/build-and-test-linux.sh
+  ;;
+  darwin*)
+    ./ci/jobs/build-and-test-macos.sh
+  ;;
 esac
+
+./ci/jobs/check-freebsd.sh
+
 ./ci/jobs/clippy.sh
 ./ci/jobs/rustfmt.sh
 

crates/polkavm-assembler/src/amd64.rs

@@ -1197,6 +1197,30 @@ pub mod inst {
             Some((self.1, 2, 4))
         }
     }
+
+    #[derive(Copy, Clone, PartialEq, Eq, Debug)]
+    pub struct lea_rip_label(pub Reg, pub Label);
+    impl lea_rip_label {
+        const fn encode_const(self) -> EncInst {
+            lea_rip(RegSize::R64, self.0, 0).encode_const()
+        }
+    }
+
+    impl core::fmt::Display for lea_rip_label {
+        fn fmt(&self, fmt: &mut core::fmt::Formatter) -> core::fmt::Result {
+            fmt.write_fmt(core::format_args!("lea {}, [{}]", self.0, self.1))
+        }
+    }
+
+    impl crate::Instruction for lea_rip_label {
+        fn encode(self) -> EncInst {
+            self.encode_const()
+        }
+
+        fn target_fixup(self) -> Option<(Label, u8, u8)> {
+            Some((self.1, 3, 4))
+        }
+    }
 }
 
 #[derive(Copy, Clone, PartialEq, Eq, Debug)]
@@ -1693,4 +1717,25 @@ mod tests {
             );
         });
     }
+
+    #[test]
+    fn lea_rip_label_infinite_loop() {
+        use super::inst::*;
+        let mut asm = crate::Assembler::new();
+        let label = asm.forward_declare_label();
+        asm.push_with_label(label, lea_rip_label(super::Reg::rax, label));
+        let disassembly = disassemble(asm.finalize());
+        assert_eq!(disassembly, "00000000 488d05f9ffffff lea rax, [rip-0x7]");
+    }
+
+    #[test]
+    fn lea_rip_label_next_instruction() {
+        use super::inst::*;
+        let mut asm = crate::Assembler::new();
+        let label = asm.forward_declare_label();
+        asm.push(lea_rip_label(super::Reg::rax, label));
+        asm.push_with_label(label, nop());
+        let disassembly = disassemble(asm.finalize());
+        assert_eq!(disassembly, "00000000 488d0500000000 lea rax, [rip]\n00000007 90 nop");
+    }
 }

crates/polkavm-assembler/src/assembler.rs

@@ -10,7 +10,7 @@ struct Fixup {
 pub struct Assembler {
     origin: u64,
     code: Vec<u8>,
-    labels: Vec<usize>,
+    labels: Vec<isize>,
     fixups: Vec<Fixup>,
 }
 
@@ -41,19 +41,19 @@ impl Assembler {
 
     pub fn forward_declare_label(&mut self) -> Label {
         let label = self.labels.len();
-        self.labels.push(usize::MAX);
+        self.labels.push(isize::MAX);
         Label(label)
     }
 
     pub fn create_label(&mut self) -> Label {
         let label = self.labels.len();
-        self.labels.push(self.code.len());
+        self.labels.push(self.code.len() as isize);
         Label(label)
     }
 
     pub fn define_label(&mut self, label: Label) -> &mut Self {
-        assert_eq!(self.labels[label.0], usize::MAX, "tried to redefine an already defined label");
-        self.labels[label.0] = self.code.len();
+        assert_eq!(self.labels[label.0], isize::MAX, "tried to redefine an already defined label");
+        self.labels[label.0] = self.code.len() as isize;
         self
     }
 
@@ -62,12 +62,16 @@ impl Assembler {
         self.push(inst)
     }
 
-    pub fn get_label_offset(&self, label: Label) -> usize {
+    pub fn get_label_offset(&self, label: Label) -> isize {
         let offset = self.labels[label.0];
-        assert_ne!(offset, usize::MAX, "tried to fetch a label offset for a label that was not defined");
+        assert_ne!(offset, isize::MAX, "tried to fetch a label offset for a label that was not defined");
         offset
     }
 
+    pub fn set_label_offset(&mut self, label: Label, offset: isize) {
+        self.labels[label.0] = offset;
+    }
+
     fn add_fixup_if_necessary(&mut self, bytes: &[u8], inst: impl Instruction) {
         let (target_label, fixup_offset, fixup_length) = match inst.target_fixup() {
             Some(fixup) => fixup,
@@ -106,8 +110,8 @@ impl Assembler {
         for fixup in self.fixups.drain(..) {
             let origin = fixup.instruction_offset + fixup.instruction_length as usize;
             let target_absolute = self.labels[fixup.target_label.0];
-            assert_ne!(target_absolute, usize::MAX);
-            let offset = target_absolute as isize - origin as isize;
+            assert_ne!(target_absolute, isize::MAX);
+            let offset = target_absolute - origin as isize;
             let p = fixup.instruction_offset + fixup.fixup_offset as usize;
             if fixup.fixup_length == 1 {
                 if offset > i8::MAX as isize || offset < i8::MIN as isize {
@@ -134,6 +138,10 @@ impl Assembler {
         self.code.len()
     }
 
+    pub fn resize(&mut self, size: usize, fill_with: u8) {
+        self.code.resize(size, fill_with)
+    }
+
     pub fn clear(&mut self) {
         self.origin = 0;
         self.code.clear();

crates/polkavm-common/src/abi.rs

@@ -143,19 +143,21 @@ impl GuestMemoryConfig {
 
     /// The address at where the program memory starts inside of the VM.
     #[inline]
-    pub const fn user_memory_address(self) -> u32 {
+    pub const fn user_memory_region_address(self) -> u32 {
         VM_ADDR_USER_MEMORY
     }
 
-    /// The size of the program memory inside of the VM, excluding the stack.
+    /// The size of the region in which the program memory resides inside of the VM, excluding the stack.
+    ///
+    /// This also includes the guard page between the read-only data and read-write data.
     #[inline]
-    pub const fn user_memory_size(self) -> u32 {
-        self.ro_data_size + self.rw_data_size + self.bss_size
+    pub const fn user_memory_region_size(self) -> u32 {
+        (self.bss_address() + self.bss_size()) - self.user_memory_region_address()
     }
 
     /// Resets the size of the program memory to zero, excluding the stack.
     #[inline]
-    pub fn clear_user_memory_size(&mut self) {
+    pub fn clear_user_memory_sizes(&mut self) {
         self.ro_data_size = 0;
         self.rw_data_size = 0;
         self.bss_size = 0;
@@ -164,7 +166,7 @@ impl GuestMemoryConfig {
     /// The address at where the program's read-only data starts inside of the VM.
     #[inline]
     pub const fn ro_data_address(self) -> u32 {
-        self.user_memory_address()
+        self.user_memory_region_address()
     }
 
     /// The size of the program's read-only data.
@@ -199,7 +201,7 @@ impl GuestMemoryConfig {
     #[inline]
     pub const fn rw_data_address(self) -> u32 {
         if self.ro_data_size == 0 {
-            self.user_memory_address()
+            self.user_memory_region_address()
         } else {
             self.ro_data_address() + self.ro_data_size + VM_PAGE_SIZE
         }

crates/polkavm-common/src/program.rs

@@ -1625,7 +1625,7 @@ impl<'a> LineProgram<'a> {
         }
 
         // Put an upper limit to how many instructions we'll process.
-        const INSTRUCTION_LIMIT_PER_REGION: usize = 128;
+        const INSTRUCTION_LIMIT_PER_REGION: usize = 256;
 
         let mark_as_finished_on_drop = SetTrueOnDrop(&mut self.is_finished);
         for _ in 0..INSTRUCTION_LIMIT_PER_REGION {