Add support for building on s390x platform (#8962)

Summary:
This PR adds support for building on s390x including updating travis CI. It uses the previous work in https://github.com/facebook/rocksdb/pull/6168 and adds some more changes to get all current tests (make check and jni tests) to pass. The tests were run with snappy, lz4, bzip2 and zstd all compiled in.

There are a few pieces still needed to get the travis build working that I don't think I can do. adamretter is this something you could help with?

1. A prebuilt https://rocksdb-deps.s3-us-west-2.amazonaws.com/cmake/cmake-3.14.5-Linux-s390x.deb package
2. A https://hub.docker.com/r/evolvedbinary/rocksjava s390x image

Not sure if there is more required for travis. Happy to help in any way I can.

Pull Request resolved: https://github.com/facebook/rocksdb/pull/8962

Reviewed By: mrambacher

Differential Revision: D31802198

Pulled By: pdillinger

fbshipit-source-id: 683511466fa6b505f85ba5a9964a268c6151f0c2

.travis.yml

@@ -5,6 +5,7 @@ os:
 arch:
   - arm64
   - ppc64le
+  - s390x
 compiler:
   - clang
   - gcc
@@ -59,6 +60,12 @@ matrix:
   - os: linux
     arch: ppc64le
     env: JOB_NAME=make-gcc4.8
+  - os: linux
+    arch: s390x
+    env: JOB_NAME=cmake-mingw
+  - os: linux
+    arch: s390x
+    env: JOB_NAME=make-gcc4.8
   - os: linux
     compiler: clang
   - if: type = pull_request AND commit_message !~ /FULL_CI/
@@ -73,6 +80,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=1
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: TEST_GROUP=1
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -81,6 +92,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=2
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: TEST_GROUP=2
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -89,6 +104,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=3
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: TEST_GROUP=3
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -97,6 +116,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: TEST_GROUP=4
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: TEST_GROUP=4
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -109,6 +132,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=java_test
+  - if: type = pull_request AND commit_message !~ /FULL_CI/ AND commit_message !~ /java/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=java_test
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -117,6 +144,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=lite_build
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=lite_build
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -125,6 +156,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=examples
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=examples
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -133,6 +168,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=cmake-gcc8
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=cmake-gcc8
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -141,6 +180,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=cmake-gcc9
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=cmake-gcc9
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -149,6 +192,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=cmake-gcc9-c++20
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=cmake-gcc9-c++20
   - if: type = pull_request AND commit_message !~ /FULL_CI/
     os : linux
     arch: arm64
@@ -157,6 +204,10 @@ matrix:
     os: linux
     arch: ppc64le
     env: JOB_NAME=status_checked
+  - if: type = pull_request AND commit_message !~ /FULL_CI/
+    os: linux
+    arch: s390x
+    env: JOB_NAME=status_checked
 
 install:
   - if [ "${JOB_NAME}" == cmake-gcc8 ]; then

CMakeLists.txt

@@ -267,6 +267,13 @@ if(CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64|AARCH64")
   endif(HAS_ARMV8_CRC)
 endif(CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64|AARCH64")
 
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "s390x")
+  CHECK_C_COMPILER_FLAG("-march=native" HAS_S390X_MARCH_NATIVE)
+  if(HAS_S390X_MARCH_NATIVE)
+    message(STATUS " HAS_S390X_MARCH_NATIVE yes")
+  endif(HAS_S390X_MARCH_NATIVE)
+endif(CMAKE_SYSTEM_PROCESSOR MATCHES "s390x")
+
 option(PORTABLE "build a portable binary" OFF)
 option(FORCE_SSE42 "force building with SSE4.2, even when PORTABLE=ON" OFF)
 option(FORCE_AVX "force building with AVX, even when PORTABLE=ON" OFF)
@@ -292,12 +299,17 @@ if(PORTABLE)
     if(FORCE_AVX2)
       set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mavx2 -mbmi -mlzcnt")
     endif()
+    if(CMAKE_SYSTEM_PROCESSOR MATCHES "^s390x")
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=z196")
+    endif()
   endif()
 else()
   if(MSVC)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
   else()
-    if(NOT CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64" AND NOT HAS_ARMV8_CRC)
+    if(CMAKE_SYSTEM_PROCESSOR MATCHES "^s390x" AND NOT HAS_S390X_MARCH_NATIVE)
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=z196")
+    elseif(NOT CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64" AND NOT HAS_ARMV8_CRC)
       set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
     endif()
   endif()

Makefile

@@ -1997,7 +1997,7 @@ JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/linux
 ifeq ($(PLATFORM), OS_SOLARIS)
 	ARCH := $(shell isainfo -b)
 else ifeq ($(PLATFORM), OS_OPENBSD)
-	ifneq (,$(filter amd64 ppc64 ppc64le arm64 aarch64 sparc64, $(MACHINE)))
+	ifneq (,$(filter amd64 ppc64 ppc64le s390x arm64 aarch64 sparc64, $(MACHINE)))
 		ARCH := 64
 	else
 		ARCH := 32
@@ -2017,7 +2017,7 @@ ifneq ($(origin JNI_LIBC), undefined)
   JNI_LIBC_POSTFIX = -$(JNI_LIBC)
 endif
 
-ifneq (,$(filter ppc% arm64 aarch64 sparc64, $(MACHINE)))
+ifneq (,$(filter ppc% s390x arm64 aarch64 sparc64, $(MACHINE)))
 	ROCKSDBJNILIB = librocksdbjni-linux-$(MACHINE)$(JNI_LIBC_POSTFIX).so
 else
 	ROCKSDBJNILIB = librocksdbjni-linux$(ARCH)$(JNI_LIBC_POSTFIX).so

build_tools/build_detect_platform

@@ -638,13 +638,19 @@ if test "0$PORTABLE" -eq 0; then
     # Tune for this POWER processor, treating '+' models as base models
     POWER=`LD_SHOW_AUXV=1 /bin/true | grep AT_PLATFORM | grep -E -o power[0-9]+`
     COMMON_FLAGS="$COMMON_FLAGS -mcpu=$POWER -mtune=$POWER "
-  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
-    COMMON_FLAGS="$COMMON_FLAGS -march=z10 "
   elif test -n "`echo $TARGET_ARCHITECTURE | grep -e^arm -e^aarch64`"; then
     # TODO: Handle this with approprite options.
     COMMON_FLAGS="$COMMON_FLAGS"
   elif test -n "`echo $TARGET_ARCHITECTURE | grep ^aarch64`"; then
     COMMON_FLAGS="$COMMON_FLAGS"
+  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
+    if echo 'int main() {}' | $CXX $PLATFORM_CXXFLAGS -x c++ \
+      -fsyntax-only -march=native - -o /dev/null 2>/dev/null; then
+      COMMON_FLAGS="$COMMON_FLAGS -march=native "
+    else
+      COMMON_FLAGS="$COMMON_FLAGS -march=z196 "
+    fi
+    COMMON_FLAGS="$COMMON_FLAGS"
   elif [ "$TARGET_OS" == "IOS" ]; then
     COMMON_FLAGS="$COMMON_FLAGS"
   elif [ "$TARGET_OS" == "AIX" ] || [ "$TARGET_OS" == "SunOS" ]; then
@@ -661,6 +667,10 @@ else
     TRY_SSE_ETC="1"
   fi
 
+  if test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
+    COMMON_FLAGS="$COMMON_FLAGS -march=z196 "
+  fi
+
   if [[ "${PLATFORM}" == "OS_MACOSX" ]]; then
     # For portability compile for macOS 10.12 (2016) or newer
     COMMON_FLAGS="$COMMON_FLAGS -mmacosx-version-min=10.12"

db/db_properties_test.cc

@@ -233,7 +233,7 @@ void VerifySimilar(uint64_t a, uint64_t b, double bias) {
 
 void VerifyTableProperties(
     const TableProperties& base_tp, const TableProperties& new_tp,
-    double filter_size_bias = CACHE_LINE_SIZE >= 256 ? 0.15 : 0.1,
+    double filter_size_bias = CACHE_LINE_SIZE >= 256 ? 0.18 : 0.1,
     double index_size_bias = 0.1, double data_size_bias = 0.1,
     double num_data_blocks_bias = 0.05) {
   VerifySimilar(base_tp.data_size, new_tp.data_size, data_size_bias);
@@ -627,7 +627,8 @@ TEST_F(DBPropertiesTest, AggregatedTablePropertiesAtLevel) {
           value_is_delta_encoded);
       // Gives larger bias here as index block size, filter block size,
       // and data block size become much harder to estimate in this test.
-      VerifyTableProperties(expected_tp, tp, 0.5, 0.4, 0.4, 0.25);
+      VerifyTableProperties(expected_tp, tp, CACHE_LINE_SIZE >= 256 ? 0.6 : 0.5,
+                            0.4, 0.4, 0.25);
     }
   }
 }

db/version_set_test.cc

@@ -2083,7 +2083,7 @@ class VersionSetAtomicGroupTest : public VersionSetTestBase,
         });
     SyncPoint::GetInstance()->SetCallBack(
         "VersionEditHandlerBase::Iterate:Finish", [&](void* arg) {
-          num_recovered_edits_ = *reinterpret_cast<int*>(arg);
+          num_recovered_edits_ = *reinterpret_cast<size_t*>(arg);
         });
     SyncPoint::GetInstance()->SetCallBack(
         "AtomicGroupReadBuffer::AddEdit:AtomicGroup",
@@ -2123,7 +2123,7 @@ class VersionSetAtomicGroupTest : public VersionSetTestBase,
   bool first_in_atomic_group_ = false;
   bool last_in_atomic_group_ = false;
   int num_edits_in_atomic_group_ = 0;
-  int num_recovered_edits_ = 0;
+  size_t num_recovered_edits_ = 0;
   VersionEdit corrupted_edit_;
   VersionEdit edit_with_incorrect_group_size_;
   std::unique_ptr<log::Writer> log_writer_;

java/src/main/java/org/rocksdb/util/Environment.java

@@ -106,7 +106,7 @@ public class Environment {
       if (isPowerPC() || isAarch64()) {
         return String.format("%sjni-linux-%s%s", name, ARCH, getLibcPostfix());
       } else if (isS390x()) {
-        return String.format("%sjni-linux%s", name, ARCH);
+        return String.format("%sjni-linux-%s", name, ARCH);
       } else {
         return String.format("%sjni-linux%s%s", name, arch, getLibcPostfix());
       }

monitoring/statistics.h

@@ -100,7 +100,9 @@ class StatisticsImpl : public Statistics {
     void operator delete[](void *p) { port::cacheline_aligned_free(p); }
   };
 
+#ifndef TEST_CACHE_LINE_SIZE
   static_assert(sizeof(StatisticsData) % CACHE_LINE_SIZE == 0, "Expected " TOSTRING(CACHE_LINE_SIZE) "-byte aligned");
+#endif
 
   CoreLocalArray<StatisticsData> per_core_stats_;
 

port/port_posix.h

@@ -193,7 +193,11 @@ extern void InitOnce(OnceType* once, void (*initializer)());
 #define ALIGN_AS(n) /*empty*/
 #else
 #if defined(__s390__)
+#if defined(__GNUC__) && __GNUC__ < 6
+#define CACHE_LINE_SIZE 64U
+#else
 #define CACHE_LINE_SIZE 256U
+#endif
 #elif defined(__powerpc__) || defined(__aarch64__)
 #define CACHE_LINE_SIZE 128U
 #else

third-party/folly/folly/Portability.h

@@ -31,6 +31,12 @@
 #define FOLLY_PPC64 0
 #endif
 
+#if defined(__s390x__)
+#define FOLLY_S390X 1
+#else
+#define FOLLY_S390X 0
+#endif
+
 #if defined(__has_builtin)
 #define FOLLY_HAS_BUILTIN(...) __has_builtin(__VA_ARGS__)
 #else
@@ -57,6 +63,7 @@ constexpr bool kIsArchArm = FOLLY_ARM == 1;
 constexpr bool kIsArchAmd64 = FOLLY_X64 == 1;
 constexpr bool kIsArchAArch64 = FOLLY_AARCH64 == 1;
 constexpr bool kIsArchPPC64 = FOLLY_PPC64 == 1;
+constexpr bool kIsArchS390X = FOLLY_S390X == 1;
 } // namespace folly
 
 namespace folly {

third-party/folly/folly/lang/Align.h

@@ -120,7 +120,7 @@ struct alignas(max_align_v) max_align_t {};
 //
 //  mimic: std::hardware_destructive_interference_size, C++17
 constexpr std::size_t hardware_destructive_interference_size =
-    kIsArchArm ? 64 : 128;
+    (kIsArchArm || kIsArchS390X) ? 64 : 128;
 static_assert(hardware_destructive_interference_size >= max_align_v, "math?");
 
 //  Memory locations within the same cache line are subject to constructive

util/bloom_test.cc

@@ -552,7 +552,8 @@ TEST_P(FullBloomTest, OptimizeForMemory) {
     }
     if (FLAGS_bits_per_key == 10) {
       EXPECT_LE(total_fp_rate / double{nfilters}, 0.011);
-      EXPECT_GE(total_fp_rate / double{nfilters}, 0.008);
+      EXPECT_GE(total_fp_rate / double{nfilters},
+                CACHE_LINE_SIZE >= 256 ? 0.007 : 0.008);
     }
 
     int64_t ex_min_total_size = int64_t{FLAGS_bits_per_key} * total_keys / 8;

util/compression.h

@@ -1136,7 +1136,11 @@ inline CacheAllocationPtr LZ4_Uncompress(const UncompressionInfo& info,
     if (input_length < 8) {
       return nullptr;
     }
-    memcpy(&output_len, input_data, sizeof(output_len));
+    if (port::kLittleEndian) {
+      memcpy(&output_len, input_data, sizeof(output_len));
+    } else {
+      memcpy(&output_len, input_data + 4, sizeof(output_len));
+    }
     input_length -= 8;
     input_data += 8;
   }

utilities/transactions/lock/range/range_tree/lib/portability/toku_time.h

@@ -133,6 +133,10 @@ static inline tokutime_t toku_time_now(void) {
   return result;
 #elif defined(__powerpc__)
   return __ppc_get_timebase();
+#elif defined(__s390x__)
+  uint64_t result;
+  asm volatile("stckf %0" : "=Q"(result) : : "cc");
+  return result;
 #else
 #error No timer implementation for this platform
 #endif