Add end-to-end test for libbpf#875

Add an end-to-end test for the fix provided by pull request libbpf#875. The
test basically normalizes an address in a specially crafted binary and
symbolizes the resulting file offset. It fails without commit
1a4e107 ("Use file size in file offset -> virtual offset
translation"), because then the file offset to virtual offset
translation produces a virtual offset that can't be symbolized to the
expected _start function.

Signed-off-by: Daniel Müller <[email protected]>

data/test-block-augmented.ld

@@ -0,0 +1,10 @@
+/* Linker script meant to augment the default one and insert some
+ * fill bytes at a relatively low address (hopefully before any of the
+ * regular relevant code. */
+
+SECTIONS {
+  .whatevs (0x100000): {
+    FILL(0xdead)
+    . = ABSOLUTE(. + 0x300000);
+  }
+}

data/test-block.c

@@ -0,0 +1,46 @@
+/* A binary basically just blocking waiting for input and exiting. It also write
+ * the address of its `_start` function to stdout (unformatted; just a byte
+ * dump).
+ *
+ * It uses raw system calls to avoid dependency on libc, which pulls in
+ * start up code and various other artifacts that perturb ELF layout in
+ * semi-unforeseeable ways, in an attempt to provide us with maximum
+ * control over the final binary.
+ *
+ * Likely only works on x86_64.
+ */
+
+#include <unistd.h>
+#include <sys/syscall.h>
+
+
+void _start(void) {
+  char buf[2];
+  int rc;
+  void* addr = (void*)&_start;
+  /* Write the address of `_start` to stderr. We use stderr because it's
+     unbuffered, so we spare ourselves from the pains of writing a
+     newline as well... */
+  asm volatile (
+      "syscall"
+      : "=a"(rc)
+      : "a"(SYS_write), "D"(STDERR_FILENO), "S"(&addr), "d"(sizeof(addr))
+      : "rcx", "r11", "memory"
+  );
+  asm volatile (
+      "syscall"
+      : "=a"(rc)
+      : "a"(SYS_read), "D"(STDIN_FILENO), "S"(buf), "d"(sizeof(buf))
+      : "rcx", "r11", "memory"
+  );
+  if (rc > 0) {
+    /* No error, so we can exit successfully. */
+    rc = 0;
+  }
+  asm volatile (
+      "syscall"
+      : "=a"(rc)
+      : "a"(SYS_exit), "D"(rc)
+      : "rcx", "r11", "memory"
+  );
+}

dev/build.rs

@@ -477,6 +477,18 @@ fn prepare_test_files() {
     let src = data_dir.join("test-mnt-ns.c");
     cc(&src, "test-mnt-ns.bin", &[]);
 
+    let src = data_dir.join("test-block.c");
+    let ld_script = data_dir.join("test-block-augmented.ld");
+    // Just passing the linker script as "regular" input file causes it
+    // to "augment" the default linker script.
+    let args = &[
+        "-fno-pic",
+        ld_script.to_str().unwrap(),
+        "-Wl,--build-id=none",
+        "-nostdlib",
+    ];
+    cc(&src, "test-block.bin", args);
+
     cc_stable_addrs(
         "test-stable-addrs.bin",
         &["-gdwarf-4", "-Wl,--build-id=none", "-O0"],

tests/blazesym.rs

@@ -805,6 +805,70 @@ fn symbolize_process_with_custom_dispatch() {
     test(process_no_dispatch);
 }
 
+/// Symbolize a normalized address from a binary with an artificially
+/// inflated ELF segment.
+///
+/// This is a regression test for the case that a program header
+/// with a memory size greater than file size is located before a
+/// program header that would otherwise match the file offset. Refer
+/// to commit 1a4e10740652 ("Use file size in file offset -> virtual
+/// offset translation").
+#[cfg(linux)]
+#[test]
+fn symbolize_normalized_large_memsize() {
+    let test_block = Path::new(&env!("CARGO_MANIFEST_DIR"))
+        .join("data")
+        .join("test-block.bin");
+    let mut child = Command::new(&test_block)
+        .stdin(Stdio::piped())
+        .stdout(Stdio::null())
+        .stderr(Stdio::piped())
+        .spawn()
+        .unwrap();
+    let pid = child.id();
+    defer!({
+        // Best effort only. The child may end up terminating gracefully
+        // if everything goes as planned.
+        // TODO: Ideally this kill would be pid FD based to eliminate
+        //       any possibility of killing the wrong entity.
+        let _rc = unsafe { libc::kill(pid as _, libc::SIGKILL) };
+    });
+
+    let mut buf = [0u8; size_of::<Addr>()];
+    let count = child
+        .stderr
+        .as_mut()
+        .unwrap()
+        .read(&mut buf)
+        .expect("failed to read child output");
+    assert_eq!(count, buf.len());
+    let addr = Addr::from_ne_bytes(buf);
+    let pid = Pid::from(child.id());
+
+    let normalizer = Normalizer::new();
+    let normalized = normalizer
+        .normalize_user_addrs(pid, [addr].as_slice())
+        .unwrap();
+
+    assert_eq!(normalized.outputs.len(), 1);
+    assert_eq!(normalized.meta.len(), 1);
+    let file_offset = normalized.outputs[0].0;
+
+    let elf = symbolize::Elf::new(test_block);
+    let src = symbolize::Source::Elf(elf);
+    let symbolizer = Symbolizer::new();
+    let sym = symbolizer
+        .symbolize_single(&src, symbolize::Input::FileOffset(file_offset))
+        .unwrap()
+        .into_sym()
+        .unwrap();
+    assert_eq!(sym.name, "_start");
+
+    // "Signal" the child to terminate gracefully.
+    let () = child.stdin.as_ref().unwrap().write_all(&[0x04]).unwrap();
+    let _status = child.wait().unwrap();
+}
+
 /// Check that we can normalize addresses in an ELF shared object.
 #[cfg(linux)]
 #[test]