added trivial malloc fast-path benchmark

While this is not good representation of real-world production malloc
behavior, it is representative of length (instruction-wise and well as
cycle-wise) of fast-path. So this is better than nothing.

3 files changed, 111 insertions, 0 deletions

diff --git a/.gitignore b/.gitignore
index 3e74f99..055a09d 100644
--- a/.gitignore
+++ b/.gitignore
@@ -19,6 +19,8 @@
 /atomicops_unittest
 /atomicops_unittest.exe
 /autom4te.cache/
+/benchmark/.deps
+/benchmark/.dirstamp
 /compile
 /config.guess
 /config.log
@@ -59,6 +61,10 @@
 /m4/ltsugar.m4
 /m4/ltversion.m4
 /m4/lt~obsolete.m4
+/malloc_bench
+/malloc_bench.exe
+/malloc_bench_shared
+/malloc_bench_shared.exe
 /malloc_extension_c_test
 /malloc_extension_debug_test
 /malloc_extension_test
diff --git a/Makefile.am b/Makefile.am
index a5a9af8..bae605a 100755
--- a/Makefile.am
+++ b/Makefile.am
@@ -847,6 +847,18 @@ endif WITH_STACK_TRACE
 
 endif WITH_DEBUGALLOC
 
+noinst_PROGRAMS += malloc_bench malloc_bench_shared
+
+malloc_bench_SOURCES = benchmark/malloc_bench.cc
+malloc_bench_CXXFLAGS = $(PTHREAD_CFLAGS) $(AM_CXXFLAGS) $(NO_BUILTIN_CXXFLAGS)
+malloc_bench_LDFLAGS = $(PTHREAD_CFLAGS) $(TCMALLOC_FLAGS) -static
+malloc_bench_LDADD = libtcmalloc_minimal.la $(PTHREAD_LIBS)
+
+malloc_bench_shared_SOURCES = benchmark/malloc_bench.cc
+malloc_bench_shared_CXXFLAGS = $(PTHREAD_CFLAGS) $(AM_CXXFLAGS) $(NO_BUILTIN_CXXFLAGS)
+malloc_bench_shared_LDFLAGS = $(PTHREAD_CFLAGS) $(TCMALLOC_FLAGS)
+malloc_bench_shared_LDADD = libtcmalloc_minimal.la $(PTHREAD_LIBS)
+
 
 ### ------- tcmalloc (thread-caching malloc + heap profiler + heap checker)
 
diff --git a/benchmark/malloc_bench.cc b/benchmark/malloc_bench.cc
new file mode 100644
index 0000000..e5e0d38
--- /dev/null
+++ b/benchmark/malloc_bench.cc
@@ -0,0 +1,93 @@
+#include <stdlib.h>
+#include <stdio.h>
+
+int main(void)
+{
+	long long i = 1LL<<(28-4);
+	size_t sz = 32;
+	printf("i = %lld\n", i);
+	for (;i>0;i--) {
+		void *p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		p = malloc(sz);
+		if (!p) {
+			abort();
+		}
+		free(p);
+		sz = ((sz | reinterpret_cast<size_t>(p)) & 511) + 16;
+	}
+	return 0;
+}