From c2a5cb4ce88bd16677127ac4d87ae5fcb4df89a0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Joel=20H=C3=B6ner?= <joel@elastic.co>
Date: Tue, 6 Feb 2024 12:23:46 +0100
Subject: [PATCH] CI: add basic FFI struct validation

---
 .github/workflows/main.yml | 30 +++++++++++++++++++++-
 validate-structs.py        | 52 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 81 insertions(+), 1 deletion(-)
 create mode 100644 validate-structs.py

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index c4334c4..ab0d7bc 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -18,6 +18,7 @@ jobs:
           - name: "Linux (all features)"
             image_name: "ubuntu-22.04"
             extra_args: "--all-features"
+            do_struct_checks: true
           - name: "Linux (minimal)"
             image_name: "ubuntu-22.04"
             extra_args: "--no-default-features"
@@ -35,7 +36,34 @@ jobs:
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
       - name: Build
-        run: cargo build ${{ matrix.extra_args }}
+        run: cargo build --all --examples ${{ matrix.extra_args }}
+      - name: Validate FFI structs
+        if: matrix.do_struct_checks
+        run: |
+          set -eu
+
+          executable=target/debug/examples/pattern
+          if [[ ! -f "${executable}" ]]; then
+            echo "Test executable '${executable}' not built: skipping test"
+            exit 0
+          fi
+
+          sudo apt-get -qq install -y gdb
+
+          gdb -q "${executable}" -batch \
+            -ex 'source validate-structs.py' \
+            1>script-out.txt 2>gdb-stderr.txt
+
+          if [[ $(cat ./script-out.txt) != *"ALL STRUCTS OK"* ]]; then
+            echo >&2 "ERROR: struct validation failed!"
+            echo >&2 "================================"
+            cat  >&2 ./gdb-stderr.txt
+            echo >&2 "================================"
+            cat  >&2 ./script-out.txt
+            exit 1
+          fi
+
+          echo PASSED.
       - name: Test
         run: cargo test ${{ matrix.extra_args }}
 
diff --git a/validate-structs.py b/validate-structs.py
new file mode 100644
index 0000000..6e2f6c5
--- /dev/null
+++ b/validate-structs.py
@@ -0,0 +1,52 @@
+# NOTE: this isn't a standalone script -- meant to be run within gdb!
+
+# gdb's rust mode doesn't parse the `<` and `>` in the types correctly
+gdb.execute('set language c++')
+
+
+checked_types = [
+    # decoder
+    ('zydis::decoder::Decoder', 'ZydisDecoder'),
+    ('zydis::ffi::decoder::AccessedFlags<zydis::enums::CpuFlag>', 'ZydisAccessedFlags'),
+    ('zydis::ffi::decoder::AccessedFlags<zydis::enums::FpuFlag>', 'ZydisAccessedFlags'),
+    ('zydis::ffi::decoder::AvxInfo', 'ZydisDecodedInstructionAvx'),
+    ('zydis::ffi::decoder::MetaInfo', 'ZydisDecodedInstructionMeta'),
+    ('zydis::ffi::decoder::RawInfo', 'ZydisDecodedInstructionRaw'),
+    ('zydis::ffi::decoder::MemoryInfo', 'ZydisDecodedOperandMem'),
+    ('zydis::ffi::decoder::PointerInfo', 'ZydisDecodedOperandPtr'),
+    ('zydis::enums::generated::Register', 'ZydisDecodedOperandReg'),
+    ('zydis::ffi::decoder::ImmediateInfo', 'ZydisDecodedOperandImm'),
+    ('zydis::ffi::decoder::DecodedOperand', 'ZydisDecodedOperand'),
+    ('zydis::ffi::decoder::DecoderContext', 'ZydisDecoderContext'),
+
+    # encoder
+    ('zydis::ffi::encoder::OperandRegister', '((ZydisEncoderOperand*)(0))->reg'),
+    ('zydis::ffi::encoder::OperandPointer', '((ZydisEncoderOperand*)(0))->ptr'),
+    ('zydis::ffi::encoder::OperandMemory', '((ZydisEncoderOperand*)(0))->mem'),
+    ('zydis::ffi::encoder::EncoderOperand', 'ZydisEncoderOperand'),
+    ('zydis::ffi::encoder::EncoderRequest', 'ZydisEncoderRequest'),
+
+    # formatter
+    ('zydis::ffi::formatter::Formatter', 'ZydisFormatter'),
+    ('zydis::ffi::formatter::FormatterBuffer', 'ZydisFormatterBuffer'),
+
+    # zycore
+    ('zydis::ffi::zycore::ZyanVector', 'ZyanVector'),
+    ('zydis::ffi::zycore::ZyanString', 'ZyanString'),
+
+    # TODO: incomplete, add more ..
+]
+
+
+def sizeof(ty: str) -> int:
+    return gdb.parse_and_eval(f'sizeof({ty})')
+
+
+for bind_ty, c_ty in checked_types:
+    bind_ty_sz = sizeof(bind_ty)
+    c_ty_sz = sizeof(c_ty)
+    
+    assert bind_ty_sz == c_ty_sz, \
+        f'binding type {bind_ty} is {bind_ty_sz} bytes, but expected {c_ty_sz}'
+
+print("ALL STRUCTS OK")