Add vectorscan (same abi as hyperscan) for arm 64 (#29881)

Fix #29276 Use Vectorscan for Arm 64 CPUs Risk Level: Low Testing: Covered by original commit Platform Specific Features: aarch64 Signed-off-by: Samuel Bercovici <samuelb@radware.com>
envoyproxy · Nov 28, 2023 · dcf8850 · dcf8850
1 parent bc2fdb7
commit dcf8850
Show file tree

Hide file tree

Showing 8 changed files with 152 additions and 16 deletions.
diff --git a/bazel/foreign_cc/BUILD b/bazel/foreign_cc/BUILD
@@ -119,7 +119,7 @@ configure_make(
     # Workaround for the issue with statically linked libstdc++
     # using -l:libstdc++.a.
     env = {
-        "CXXFLAGS": "-lstdc++ -Wno-unused-command-line-argument",
+        "CXXFLAGS": "--static -lstdc++ -Wno-unused-command-line-argument",
     },
     lib_source = "@net_colm_open_source_colm//:all",
     out_binaries = ["colm"],
@@ -141,7 +141,7 @@ configure_make(
     # Workaround for the issue with statically linked libstdc++
     # using -l:libstdc++.a.
     env = {
-        "CXXFLAGS": "-lstdc++ -Wno-unused-command-line-argument",
+        "CXXFLAGS": "--static -lstdc++ -Wno-unused-command-line-argument",
     },
     lib_source = "@net_colm_open_source_ragel//:all",
     out_binaries = ["ragel"],

diff --git a/bazel/foreign_cc/vectorscan.patch b/bazel/foreign_cc/vectorscan.patch
@@ -0,0 +1,33 @@
+diff --git a/CMakeLists.txt b/CMakeLists.txt
+index 1db128b..ee3b4a9 100644
+--- a/CMakeLists.txt
++++ b/CMakeLists.txt
+@@ -27,7 +27,7 @@ include (${CMAKE_MODULE_PATH}/platform.cmake)
+ include (${CMAKE_MODULE_PATH}/boost.cmake)
+ include (${CMAKE_MODULE_PATH}/ragel.cmake)
+
+-find_package(PkgConfig REQUIRED)
++find_package(PkgConfig QUIET)
+
+ find_program(RAGEL ragel)
+
+diff --git a/cmake/sqlite3.cmake b/cmake/sqlite3.cmake
+index 92b18ce..5291726 100644
+--- a/cmake/sqlite3.cmake
++++ b/cmake/sqlite3.cmake
+@@ -2,6 +2,8 @@
+ # a lot of noise to find sqlite
+ #
+
++if(NOT SQLITE_SKIP_CHECK)
++
+ option(SQLITE_PREFER_STATIC "Build sqlite3 statically instead of using an installed lib" OFF)
+
+ if(NOT SQLITE_PREFER_STATIC)
+@@ -33,4 +35,6 @@ else()
+     endif()
+ endif()
+
++endif()
++
+ # that's enough about sqlite
diff --git a/bazel/repositories.bzl b/bazel/repositories.bzl
@@ -1,8 +1,8 @@
-load(":dev_binding.bzl", "envoy_dev_binding")
+load("@com_google_googleapis//:repository_rules.bzl", "switched_rules_by_language")
 load("@envoy_api//bazel:envoy_http_archive.bzl", "envoy_http_archive")
 load("@envoy_api//bazel:external_deps.bzl", "load_repository_locations")
+load(":dev_binding.bzl", "envoy_dev_binding")
 load(":repository_locations.bzl", "PROTOC_VERSIONS", "REPOSITORY_LOCATIONS_SPEC")
-load("@com_google_googleapis//:repository_rules.bzl", "switched_rules_by_language")
 
 PPC_SKIP_TARGETS = ["envoy.filters.http.lua"]
 
@@ -311,6 +311,7 @@ def envoy_dependencies(skip_targets = []):
     _com_github_google_quiche()
     _com_googlesource_googleurl()
     _io_hyperscan()
+    _io_vectorscan()
     _io_opentracing_cpp()
     _net_colm_open_source_colm()
     _net_colm_open_source_ragel()
@@ -710,6 +711,15 @@ def _io_hyperscan():
         patches = ["@envoy//bazel/foreign_cc:hyperscan.patch"],
     )
 
+def _io_vectorscan():
+    external_http_archive(
+        name = "io_vectorscan",
+        build_file_content = BUILD_ALL_CONTENT,
+        type = "tar.gz",
+        patch_args = ["-p1"],
+        patches = ["@envoy//bazel/foreign_cc:vectorscan.patch"],
+    )
+
 def _io_opentracing_cpp():
     external_http_archive(
         name = "io_opentracing_cpp",

diff --git a/bazel/repository_locations.bzl b/bazel/repository_locations.bzl
@@ -499,6 +499,24 @@ REPOSITORY_LOCATIONS_SPEC = dict(
         license = "BSD-3-Clause",
         license_url = "https://github.com/intel/hyperscan/blob/v{version}/LICENSE",
     ),
+    io_vectorscan = dict(
+        project_name = "Vectorscan",
+        project_desc = "Hyperscan port for additional CPU architectures",
+        project_url = "https://www.vectorcamp.gr/vectorscan/",
+        version = "5.4.11",
+        sha256 = "905f76ad1fa9e4ae0eb28232cac98afdb96c479666202c5a4c27871fb30a2711",
+        strip_prefix = "vectorscan-vectorscan-{version}",
+        urls = ["https://codeload.github.com/VectorCamp/vectorscan/tar.gz/refs/tags/vectorscan/{version}"],
+        use_category = ["dataplane_ext"],
+        extensions = [
+            "envoy.matching.input_matchers.hyperscan",
+            "envoy.regex_engines.hyperscan",
+        ],
+        release_date = "2023-11-20",
+        cpe = "N/A",
+        license = "BSD-3-Clause",
+        license_url = "https://github.com/VectorCamp/vectorscan/blob/vectorscan/{version}/LICENSE",
+    ),
     io_opentracing_cpp = dict(
         project_name = "OpenTracing",
         project_desc = "Vendor-neutral APIs and instrumentation for distributed tracing",

diff --git a/contrib/all_contrib_extensions.bzl b/contrib/all_contrib_extensions.bzl
@@ -7,12 +7,16 @@ def envoy_contrib_linux_x86_64_constraints():
         "@platforms//cpu:x86_64",
     ]
 
+def envoy_contrib_linux_aarch64_constraints():
+    return [
+        "@platforms//os:linux",
+        "@platforms//cpu:aarch64",
+    ]
+
 ARM64_SKIP_CONTRIB_TARGETS = [
     "envoy.tls.key_providers.cryptomb",
     "envoy.tls.key_providers.qat",
-    "envoy.matching.input_matchers.hyperscan",
     "envoy.network.connection_balance.dlb",
-    "envoy.regex_engines.hyperscan",
 ]
 PPC_SKIP_CONTRIB_TARGETS = [
     "envoy.tls.key_providers.cryptomb",

diff --git a/contrib/hyperscan/matching/input_matchers/source/BUILD b/contrib/hyperscan/matching/input_matchers/source/BUILD
@@ -11,6 +11,7 @@ load(
 )
 load(
     "//contrib:all_contrib_extensions.bzl",
+    "envoy_contrib_linux_aarch64_constraints",
     "envoy_contrib_linux_x86_64_constraints",
 )
 
@@ -41,8 +42,34 @@ envoy_cmake(
     ],
 )
 
+envoy_cmake(
+    name = "vectorscan",
+    build_data = ["@org_boost//:header"],
+    cache_entries = {
+        "BOOST_ROOT": "$EXT_BUILD_ROOT/external/org_boost",
+        #        "BUILD_SVE2_BITPERM": "on",
+        #        "BUILD_SVE2": "on",
+        #        "BUILD_SVE": "on",
+        "BUILD_EXAMPLES": "off",
+        "BUILD_SHARED_LIBS": "off",
+        "CMAKE_BUILD_TYPE": "Release",
+        "CMAKE_INSTALL_LIBDIR": "lib",
+        "FAT_RUNTIME": "off",
+        "SQLITE_SKIP_CHECK": "on",
+        "RAGEL": "$EXT_BUILD_DEPS/ragel/bin/ragel",
+    },
+    default_cache_entries = {},
+    lib_source = "@io_vectorscan//:all",
+    out_static_libs = ["libhs.a"],
+    tags = ["skip_on_windows"],
+    target_compatible_with = envoy_contrib_linux_aarch64_constraints(),
+    deps = [
+        envoy_external_dep_path("ragel"),
+    ],
+)
+
 envoy_cc_library(
-    name = "matcher_lib",
+    name = "hyperscan_matcher_lib",
     srcs = ["matcher.cc"],
     hdrs = ["matcher.h"],
     deps = [
@@ -53,12 +80,25 @@ envoy_cc_library(
     ],
 )
 
+envoy_cc_library(
+    name = "vectorscan_matcher_lib",
+    srcs = ["matcher.cc"],
+    hdrs = ["matcher.h"],
+    deps = [
+        ":vectorscan",
+        "//envoy/common:regex_interface",
+        "//envoy/matcher:matcher_interface",
+        "//envoy/thread_local:thread_local_interface",
+    ],
+)
+
 envoy_cc_contrib_extension(
     name = "config",
     srcs = ["config.cc"],
     hdrs = ["config.h"],
     defines = select({
         "//bazel:linux_x86_64": [],
+        "//bazel:linux_aarch64": [],
         "//conditions:default": [
             "HYPERSCAN_DISABLED=1",
         ],
@@ -71,7 +111,10 @@ envoy_cc_contrib_extension(
         "@envoy_api//contrib/envoy/extensions/matching/input_matchers/hyperscan/v3alpha:pkg_cc_proto",
     ] + select({
         "//bazel:linux_x86_64": [
-            ":matcher_lib",
+            ":hyperscan_matcher_lib",
+        ],
+        "//bazel:linux_aarch64": [
+            ":vectorscan_matcher_lib",
         ],
         "//conditions:default": [
         ],

diff --git a/contrib/hyperscan/matching/input_matchers/test/BUILD b/contrib/hyperscan/matching/input_matchers/test/BUILD
@@ -22,36 +22,54 @@ envoy_cc_test(
     name = "matcher_test",
     srcs = ["matcher_test.cc"],
     deps = [
-        "//contrib/hyperscan/matching/input_matchers/source:matcher_lib",
         "//source/common/thread_local:thread_local_lib",
         "//test/mocks/event:event_mocks",
         "//test/mocks/thread_local:thread_local_mocks",
         "//test/test_common:utility_lib",
-    ],
+    ] + select({
+        "//bazel:linux_x86_64": [
+            "//contrib/hyperscan/matching/input_matchers/source:hyperscan_matcher_lib",
+        ],
+        "//bazel:linux_aarch64": [
+            "//contrib/hyperscan/matching/input_matchers/source:vectorscan_matcher_lib",
+        ],
+    }),
 )
 
 envoy_cc_benchmark_binary(
     name = "hyperscan_speed_test",
     srcs = ["hyperscan_speed_test.cc"],
     external_deps = ["benchmark"],
     deps = [
-        "//contrib/hyperscan/matching/input_matchers/source:hyperscan",
         "//source/common/common:assert_lib",
         "//source/common/common:utility_lib",
         "@com_googlesource_code_re2//:re2",
-    ],
+    ] + select({
+        "//bazel:linux_x86_64": [
+            "//contrib/hyperscan/matching/input_matchers/source:hyperscan",
+        ],
+        "//bazel:linux_aarch64": [
+            "//contrib/hyperscan/matching/input_matchers/source:vectorscan",
+        ],
+    }),
 )
 
 envoy_cc_benchmark_binary(
     name = "matcher_speed_test",
     srcs = ["matcher_speed_test.cc"],
     external_deps = ["benchmark"],
     deps = [
-        "//contrib/hyperscan/matching/input_matchers/source:matcher_lib",
         "//source/common/common:assert_lib",
         "//source/common/common:regex_lib",
         "//source/common/common:utility_lib",
         "//source/common/thread_local:thread_local_lib",
         "//test/mocks/event:event_mocks",
-    ],
+    ] + select({
+        "//bazel:linux_x86_64": [
+            "//contrib/hyperscan/matching/input_matchers/source:hyperscan_matcher_lib",
+        ],
+        "//bazel:linux_aarch64": [
+            "//contrib/hyperscan/matching/input_matchers/source:vectorscan_matcher_lib",
+        ],
+    }),
 )
diff --git a/contrib/hyperscan/regex_engines/source/BUILD b/contrib/hyperscan/regex_engines/source/BUILD
@@ -14,9 +14,15 @@ envoy_cc_library(
     srcs = ["regex.cc"],
     hdrs = ["regex.h"],
     deps = [
-        "//contrib/hyperscan/matching/input_matchers/source:matcher_lib",
         "//envoy/common:regex_interface",
-    ],
+    ] + select({
+        "//bazel:linux_x86_64": [
+            "//contrib/hyperscan/matching/input_matchers/source:hyperscan_matcher_lib",
+        ],
+        "//bazel:linux_aarch64": [
+            "//contrib/hyperscan/matching/input_matchers/source:vectorscan_matcher_lib",
+        ],
+    }),
 )
 
 envoy_cc_contrib_extension(
@@ -25,6 +31,7 @@ envoy_cc_contrib_extension(
     hdrs = ["config.h"],
     defines = select({
         "//bazel:linux_x86_64": [],
+        "//bazel:linux_aarch64": [],
         "//conditions:default": [
             "HYPERSCAN_DISABLED=1",
         ],
@@ -36,6 +43,9 @@ envoy_cc_contrib_extension(
         "//bazel:linux_x86_64": [
             ":regex_lib",
         ],
+        "//bazel:linux_aarch64": [
+            ":regex_lib",
+        ],
         "//conditions:default": [
         ],
     }),