version/bazel: fix git SHA1 embedding story. (#843)

This is an alternative to #841, using the technique described in
https://groups.google.com/forum/m/#!topic/bazel-discuss/9kqLmOq_m_U. See also
bazelbuild/bazel#2893.

BUILD

@@ -1,16 +0,0 @@
-load(
-    "//bazel:envoy_build_system.bzl",
-    "envoy_package",
-)
-
-envoy_package()
-
-genrule(
-    name = "envoy_version",
-    srcs = glob([".git/**"]),
-    outs = ["version_generated.cc"],
-    cmd = "touch $@ && $(location //tools:gen_git_sha.sh) " +
-          "$$(dirname $(location //tools:gen_git_sha.sh)) $@",
-    local = 1,
-    tools = ["//tools:gen_git_sha.sh"],
-)

bazel/envoy_build_system.bzl

@@ -60,6 +60,7 @@ def envoy_cc_library(name,
                      external_deps = [],
                      tcmalloc_dep = None,
                      repository = "",
+                     linkstamp = None,
                      deps = []):
     if tcmalloc_dep:
         deps += tcmalloc_external_deps(repository)
@@ -76,25 +77,21 @@ def envoy_cc_library(name,
         include_prefix = envoy_include_prefix(PACKAGE_NAME),
         alwayslink = 1,
         linkstatic = 1,
+        linkstamp = linkstamp,
     )
 
 def _git_stamped_genrule(repository, name):
     # To workaround https://github.com/bazelbuild/bazel/issues/2805, we
     # do binary rewriting to replace the linker produced MD5 hash with the
     # version_generated.cc git SHA1 hash (truncated).
-    version_generated = repository + "//:version_generated.cc"
     rewriter = repository + "//tools:git_sha_rewriter.py"
     native.genrule(
         name = name + "_stamped",
-        srcs = [
-            name,
-            version_generated,
-        ],
+        srcs = [name],
         outs = [name + ".stamped"],
         cmd = "cp $(location " + name + ") $@ && " +
               "chmod u+w $@ && " +
-              "$(location " + rewriter + ") " +
-              "$(location " + version_generated + ") $@",
+              "$(location " + rewriter + ") $@",
         tools = [rewriter],
     )
 
@@ -136,8 +133,9 @@ def envoy_cc_binary(name,
         linkstatic = 1,
         visibility = visibility,
         malloc = tcmalloc_external_dep(repository),
-        # See above comment on MD5 hash.
-        stamp = 0,
+        # See above comment on MD5 hash, this is another "force MD5 stamps" to make sure our
+        # rewriting is robust.
+        stamp = 1,
         deps = deps,
     )
 

bazel/get_workspace_status

@@ -0,0 +1,31 @@
+# This file was imported from https://github.com/bazelbuild/bazel at d6fec93.
+
+#!/bin/bash
+
+# This script will be run bazel when building process starts to
+# generate key-value information that represents the status of the
+# workspace. The output should be like
+#
+# KEY1 VALUE1
+# KEY2 VALUE2
+#
+# If the script exits with non-zero code, it's considered as a failure
+# and the output will be discarded.
+
+# The code below presents an implementation that works for git repository
+git_rev=$(git rev-parse HEAD)
+if [[ $? != 0 ]];
+then
+    exit 1
+fi
+echo "BUILD_SCM_REVISION ${git_rev}"
+
+# Check whether there are any uncommited changes
+git diff-index --quiet HEAD --
+if [[ $? == 0 ]];
+then
+    tree_status="Clean"
+else
+    tree_status="Modified"
+fi
+echo "BUILD_SCM_STATUS ${tree_status}"

ci/build_setup.sh

@@ -68,7 +68,10 @@ fi
 cp -f "${ENVOY_SRCDIR}"/ci/WORKSPACE.consumer "${ENVOY_CONSUMER_SRCDIR}"/WORKSPACE
 mkdir -p "${ENVOY_CONSUMER_SRCDIR}"/tools
 # Hack due to https://github.com/lyft/envoy/issues/838.
-ln -sf "${ENVOY_SRCDIR}"/tools/bazel.rc "${ENVOY_CONSUMER_SRCDIR}"/tools/bazel.rc
+ln -sf "${ENVOY_SRCDIR}"/tools/bazel.rc "${ENVOY_CONSUMER_SRCDIR}"/tools/
+mkdir -p "${ENVOY_CONSUMER_SRCDIR}"/bazel
+ln -sf "${ENVOY_SRCDIR}"/bazel/get_workspace_status "${ENVOY_CONSUMER_SRCDIR}"/bazel/
+
 # This is the hash on https://github.com/htuch/envoy-consumer.git we pin to.
 (cd "${ENVOY_CONSUMER_SRCDIR}" && git checkout 94e11fa753a1e787c82cccaec642eda5e5b61ed8)
 

source/common/common/BUILD

@@ -103,9 +103,9 @@ envoy_cc_library(
 envoy_cc_library(
     name = "version_lib",
     srcs = ["version.cc"],
+    linkstamp = "version_linkstamp.cc",
     deps = [
         ":version_includes",
-        "//source/version_generated",
     ],
 )
 

source/common/common/version.cc

@@ -4,8 +4,21 @@
 
 #include "spdlog/spdlog.h"
 
+extern const char build_scm_revision[];
+extern const char build_scm_status[];
+
+const std::string& VersionInfo::revision() {
+  static std::string* git_sha1 = new std::string(build_scm_revision);
+  return *git_sha1;
+}
+
+const std::string& VersionInfo::revisionStatus() {
+  static std::string* status = new std::string(build_scm_status);
+  return *status;
+}
+
 std::string VersionInfo::version() {
-  return fmt::format("{}/{}", GIT_SHA.substr(0, 6),
+  return fmt::format("{}/{}/{}", revision(), revisionStatus(),
 #ifdef NDEBUG
                      "RELEASE"
 #else

source/common/common/version.h

@@ -4,11 +4,13 @@
 
 /**
  * Wraps compiled in code versioning.
- * NOTE: The GIT_SHA is generated into version_generated.cc and is not present in the source tree.
  */
 class VersionInfo {
 public:
+  // Repository revision (e.g. git SHA1).
+  static const std::string& revision();
+  // Repository status (e.g. clean, modified).
+  static const std::string& revisionStatus();
+  // Repository information and build type.
   static std::string version();
-
-  static const std::string GIT_SHA;
 };

source/common/common/version_linkstamp.cc

@@ -0,0 +1,5 @@
+extern const char build_scm_revision[];
+extern const char build_scm_status[];
+
+const char build_scm_revision[] = BUILD_SCM_REVISION;
+const char build_scm_status[] = BUILD_SCM_STATUS;

source/server/BUILD

@@ -103,7 +103,6 @@ envoy_cc_library(
         "//include/envoy/server:options_interface",
         "//source/common/common:macros",
         "//source/common/common:version_lib",
-        "//source/version_generated",
     ],
 )
 
@@ -141,7 +140,6 @@ envoy_cc_library(
         "//source/common/stats:statsd_lib",
         "//source/common/thread_local:thread_local_lib",
         "//source/server/http:admin_lib",
-        "//source/version_generated",
     ],
 )
 

source/server/http/BUILD

@@ -42,7 +42,6 @@ envoy_cc_library(
         "//source/common/router:config_lib",
         "//source/common/upstream:host_utility_lib",
         "//source/server/config/network:http_connection_manager_lib",
-        "//source/version_generated",
     ],
 )
 

source/server/server.cc

@@ -72,7 +72,7 @@ InstanceImpl::InstanceImpl(Options& options, TestHooks& hooks, HotRestart& resta
   failHealthcheck(false);
 
   uint64_t version_int;
-  if (!StringUtil::atoul(VersionInfo::GIT_SHA.substr(0, 6).c_str(), version_int, 16)) {
+  if (!StringUtil::atoul(VersionInfo::revision().substr(0, 6).c_str(), version_int, 16)) {
     throw EnvoyException("compiled GIT SHA is invalid. Invalid build.");
   }
   server_stats_.version_.set(version_int);

test/BUILD

@@ -17,7 +17,6 @@ envoy_cc_test_library(
     ],
     deps = [
         "//source/common/common:thread_lib",
-        "//source/common/common:version_lib",
         "//source/common/event:libevent_lib",
         "//test/test_common:environment_lib",
         "//test/test_common:printers_lib",

tools/bazel.rc

@@ -1,5 +1,7 @@
 # Envoy specific Bazel build/test options.
 
+build --workspace_status_command=bazel/get_workspace_status
+
 build:asan --copt -fsanitize=address
 build:asan --linkopt -fsanitize=address
 build:asan --linkopt -ldl

tools/git_sha_rewriter.py

@@ -41,21 +41,24 @@
 ]
 # We're expecting an MD5 hash, 16 bytes.
 MD5_HASH_LEN = 16
+SHA1_HASH_LEN = 20
 EXPECTED_BUILD_ID_NOTE_LENGTH = len(EXPECTED_BUILD_ID_NOTE_PREFIX) + MD5_HASH_LEN
 
 
 class RewriterException(Exception):
   pass
 
 
-# Extract MD5 hash hex string from version_generated.cc.
+# Extract git SHA1 hash from "envoy --version" stdout.
 def ExtractGitSha(path):
-  with open(path, 'r') as f:
-    contents = f.read()
-    sr = re.search('GIT_SHA\("(\w+)"', contents, flags=re.MULTILINE)
-    if not sr:
-      raise RewriterException('Bad version_generated.cc: %s' % contents)
-    return sr.group(1)
+  version_output = sp.check_output([path, '--version']).strip()
+  sr = re.search('version: (\w+)/', version_output)
+  if not sr:
+    raise RewriterException('Bad envoy --version: %s' % version_output)
+  sha1_hash = sr.group(1)
+  if len(sha1_hash) != 2 * SHA1_HASH_LEN:
+    raise RewriterException('Bad SHA1 hash in %s: %s' % (path, sha1_hash))
+  return sha1_hash
 
 
 # Scrape the offset of .note.gnu.build-id via readelf from the binary. Also
@@ -101,11 +104,10 @@ def RewriteBinary(path, offset, git5_sha1):
 
 
 if __name__ == '__main__':
-  if len(sys.argv) != 3:
-    print('Usage: %s <path to version_generated.cc <Envoy binary path> ' %
-          sys.argv[0])
+  if len(sys.argv) != 2:
+    print 'Usage: %s <Envoy binary path> ' % sys.argv[0]
     sys.exit(1)
-  version_generated = ExtractGitSha(sys.argv[1])
-  envoy_bin_path = sys.argv[2]
+  envoy_bin_path = sys.argv[1]
+  version_generated = ExtractGitSha(envoy_bin_path)
   build_id_note_offset = ExtractBuildIdNoteOffset(envoy_bin_path)
   RewriteBinary(envoy_bin_path, build_id_note_offset, version_generated)