Merge branch 'dev'

Restore the essence of 898960247a, which
sabotages tail call optimization.  This is necessary even when the
mutually recursive functions are in separate compilation units.

ChangeLog

@@ -3,8 +3,26 @@ bug fixes are all mentioned, but internal enhancements are omitted here for
 brevity (even though they are more fun to write about).  Much more detail can be
 found in the git revision history:
 
-    http://www.canonware.com/cgi-bin/gitweb.cgi?p=jemalloc.git
-    git://canonware.com/jemalloc.git
+    https://github.com/jemalloc/jemalloc
+
+* 3.5.1 (February 25, 2014)
+
+  This version primarily addresses minor bugs in test code.
+
+  Bug fixes:
+  - Configure Solaris/Illumos to use MADV_FREE.
+  - Fix junk filling for mremap(2)-based huge reallocation.  This is only
+    relevant if configuring with the --enable-mremap option specified.
+  - Avoid compilation failure if 'restrict' C99 keyword is not supported by the
+    compiler.
+  - Add a configure test for SSE2 rather than assuming it is usable on i686
+    systems.  This fixes test compilation errors, especially on 32-bit Linux
+    systems.
+  - Fix mallctl argument size mismatches (size_t vs. uint64_t) in the stats unit
+    test.
+  - Fix/remove flawed alignment-related overflow tests.
+  - Prevent compiler optimizations that could change backtraces in the
+    prof_accum unit test.
 
 * 3.5.0 (January 22, 2014)
 
@@ -16,7 +34,7 @@ found in the git revision history:
     API.  The *allocx() functions are slightly simpler to use because they have
     fewer parameters, they directly return the results of primary interest, and
     mallocx()/rallocx() avoid the strict aliasing pitfall that
-    allocm()/rallocx() share with posix_memalign().  Note that *allocm() is
+    allocm()/rallocm() share with posix_memalign().  Note that *allocm() is
     slated for removal in the next non-bugfix release.
   - Add support for LinuxThreads.
 

Makefile.in

@@ -128,6 +128,8 @@ TESTS_UNIT := $(srcroot)test/unit/bitmap.c \
 	$(srcroot)test/unit/tsd.c \
 	$(srcroot)test/unit/util.c \
 	$(srcroot)test/unit/zero.c
+TESTS_UNIT_AUX := $(srcroot)test/unit/prof_accum_a.c \
+	$(srcroot)test/unit/prof_accum_b.c
 TESTS_INTEGRATION := $(srcroot)test/integration/aligned_alloc.c \
 	$(srcroot)test/integration/allocated.c \
 	$(srcroot)test/integration/mallocx.c \
@@ -155,9 +157,10 @@ C_TESTLIB_STRESS_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.stress.$(O))
 C_TESTLIB_OBJS := $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(C_TESTLIB_STRESS_OBJS)
 
 TESTS_UNIT_OBJS := $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_UNIT_AUX_OBJS := $(TESTS_UNIT_AUX:$(srcroot)%.c=$(objroot)%.$(O))
 TESTS_INTEGRATION_OBJS := $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%.$(O))
 TESTS_STRESS_OBJS := $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%.$(O))
-TESTS_OBJS := $(TESTS_UNIT_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_STRESS_OBJS)
+TESTS_OBJS := $(TESTS_UNIT_OBJS) $(TESTS_UNIT_AUX_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_STRESS_OBJS)
 
 .PHONY: all dist build_doc_html build_doc_man build_doc
 .PHONY: install_bin install_include install_lib
@@ -206,6 +209,12 @@ $(C_TESTLIB_STRESS_OBJS): $(objroot)test/src/%.stress.$(O): $(srcroot)test/src/%
 $(C_TESTLIB_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST -DJEMALLOC_STRESS_TESTLIB
 $(C_TESTLIB_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
 $(TESTS_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
+$(TESTS_UNIT_AUX_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
+define make-unit-link-dep
+$(1): TESTS_UNIT_LINK_OBJS += $(2)
+$(1): $(2)
+endef
+$(foreach test, $(TESTS_UNIT:$(srcroot)test/unit/%.c=$(objroot)test/unit/%$(EXE)), $(eval $(call make-unit-link-dep,$(test),$(filter $(test:%=%_a.$(O)) $(test:%=%_b.$(O)),$(TESTS_UNIT_AUX_OBJS)))))
 $(TESTS_INTEGRATION_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_TEST
 $(TESTS_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST
 $(TESTS_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.c
@@ -248,7 +257,7 @@ $(STATIC_LIBS):
 	@mkdir -p $(@D)
 	$(AR) $(ARFLAGS)@AROUT@ $+
 
-$(objroot)test/unit/%$(EXE): $(objroot)test/unit/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS)
+$(objroot)test/unit/%$(EXE): $(objroot)test/unit/%.$(O) $(TESTS_UNIT_LINK_OBJS) $(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS)
 	@mkdir -p $(@D)
 	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LDFLAGS) $(filter-out -lm,$(LIBS)) -lm $(EXTRA_LDFLAGS)
 
@@ -359,7 +368,7 @@ coverage: check
 	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src unit $(C_TESTLIB_UNIT_OBJS)
 	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src integration $(C_TESTLIB_INTEGRATION_OBJS)
 	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src stress $(C_TESTLIB_STRESS_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/unit unit $(TESTS_UNIT_OBJS)
+	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/unit unit $(TESTS_UNIT_OBJS) $(TESTS_UNIT_AUX_OBJS)
 	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/integration integration $(TESTS_INTEGRATION_OBJS)
 	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/stress integration $(TESTS_STRESS_OBJS)
 endif

configure.ac

@@ -19,7 +19,9 @@ AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
 ]], [[
     return 0;
 ]])],
+              [je_cv_cflags_appended=$1]
               AC_MSG_RESULT([yes]),
+              [je_cv_cflags_appended=]
               AC_MSG_RESULT([no])
               [CFLAGS="${TCFLAGS}"]
 )
@@ -128,6 +130,9 @@ if test "x$CFLAGS" = "x" ; then
   no_CFLAGS="yes"
   if test "x$GCC" = "xyes" ; then
     JE_CFLAGS_APPEND([-std=gnu99])
+    if test "x$je_cv_cflags_appended" = "x-std=gnu99" ; then
+      AC_DEFINE_UNQUOTED([JEMALLOC_HAS_RESTRICT])
+    fi
     JE_CFLAGS_APPEND([-Wall])
     JE_CFLAGS_APPEND([-pipe])
     JE_CFLAGS_APPEND([-g3])
@@ -193,21 +198,21 @@ CPU_SPINWAIT=""
 case "${host_cpu}" in
   i[[345]]86)
 	;;
-  i686)
-	JE_COMPILABLE([__asm__], [], [[__asm__ volatile("pause"); return 0;]],
-	              [je_cv_asm])
-	if test "x${je_cv_asm}" = "xyes" ; then
+  i686|x86_64)
+	JE_COMPILABLE([pause instruction], [],
+	              [[__asm__ volatile("pause"); return 0;]],
+	              [je_cv_pause])
+	if test "x${je_cv_pause}" = "xyes" ; then
 	    CPU_SPINWAIT='__asm__ volatile("pause")'
 	fi
-	AC_DEFINE_UNQUOTED([HAVE_SSE2], [ ])
-	;;
-  x86_64)
-	JE_COMPILABLE([__asm__ syntax], [],
-	              [[__asm__ volatile("pause"); return 0;]], [je_cv_asm])
-	if test "x${je_cv_asm}" = "xyes" ; then
-	    CPU_SPINWAIT='__asm__ volatile("pause")'
+	dnl emmintrin.h fails to compile unless MMX, SSE, and SSE2 are
+	dnl supported.
+	JE_COMPILABLE([SSE2 intrinsics], [
+#include <emmintrin.h>
+], [], [je_cv_sse2])
+	if test "x${je_cv_sse2}" = "xyes" ; then
+	  AC_DEFINE_UNQUOTED([HAVE_SSE2], [ ])
 	fi
-	AC_DEFINE_UNQUOTED([HAVE_SSE2], [ ])
 	;;
   powerpc)
 	AC_DEFINE_UNQUOTED([HAVE_ALTIVEC], [ ])
@@ -296,6 +301,7 @@ case "${host}" in
   *-*-solaris2*)
 	CFLAGS="$CFLAGS"
 	abi="elf"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
 	RPATH='-Wl,-R,$(1)'
 	dnl Solaris needs this for sigwait().
 	CPPFLAGS="$CPPFLAGS -D_POSIX_PTHREAD_SEMANTICS"

include/jemalloc/internal/jemalloc_internal_defs.h.in

@@ -187,6 +187,9 @@
  */
 #undef JEMALLOC_HAS_ALLOCA_H
 
+/* C99 restrict keyword supported. */
+#undef JEMALLOC_HAS_RESTRICT
+
 /* sizeof(int) == 2^LG_SIZEOF_INT. */
 #undef LG_SIZEOF_INT
 

include/jemalloc/internal/jemalloc_internal_macros.h

@@ -45,3 +45,7 @@
 #ifndef __DECONST
 #  define	__DECONST(type, var)	((type)(uintptr_t)(const void *)(var))
 #endif
+
+#ifndef JEMALLOC_HAS_RESTRICT
+#  define restrict
+#endif

include/jemalloc/internal/prof.h

@@ -8,7 +8,11 @@ typedef struct prof_ctx_s prof_ctx_t;
 typedef struct prof_tdata_s prof_tdata_t;
 
 /* Option defaults. */
-#define	PROF_PREFIX_DEFAULT		"jeprof"
+#ifdef JEMALLOC_PROF
+#  define PROF_PREFIX_DEFAULT		"jeprof"
+#else
+#  define PROF_PREFIX_DEFAULT		""
+#endif
 #define	LG_PROF_SAMPLE_DEFAULT		19
 #define	LG_PROF_INTERVAL_DEFAULT	-1
 

src/arena.c

@@ -2476,7 +2476,6 @@ bin_info_run_size_calc(arena_bin_info_t *bin_info, size_t min_run_size)
 			    bin_info->reg_interval) - pad_size;
 		} while (try_hdr_size > try_redzone0_offset);
 	} while (try_run_size <= arena_maxclass
-	    && try_run_size <= arena_maxclass
 	    && RUN_MAX_OVRHD * (bin_info->reg_interval << 3) >
 	    RUN_MAX_OVRHD_RELAX
 	    && (try_redzone0_offset << RUN_BFP) > RUN_MAX_OVRHD * try_run_size

src/huge.c

@@ -171,6 +171,16 @@ huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra,
 				abort();
 			memcpy(ret, ptr, copysize);
 			chunk_dealloc_mmap(ptr, oldsize);
+		} else if (config_fill && zero == false && opt_junk && oldsize
+		    < newsize) {
+			/*
+			 * mremap(2) clobbers the original mapping, so
+			 * junk/zero filling is not preserved.  There is no
+			 * need to zero fill here, since any trailing
+			 * uninititialized memory is demand-zeroed by the
+			 * kernel, but junk filling must be redone.
+			 */
+			memset(ret + oldsize, 0xa5, newsize - oldsize);
 		}
 	} else
 #endif

test/integration/aligned_alloc.c

@@ -47,10 +47,10 @@ TEST_BEGIN(test_oom_errors)
 
 #if LG_SIZEOF_PTR == 3
 	alignment = UINT64_C(0x4000000000000000);
-	size      = UINT64_C(0x8400000000000001);
+	size      = UINT64_C(0xc000000000000001);
 #else
 	alignment = 0x40000000LU;
-	size      = 0x84000001LU;
+	size      = 0xc0000001LU;
 #endif
 	set_errno(0);
 	p = aligned_alloc(alignment, size);

test/integration/allocm.c

@@ -39,28 +39,6 @@ TEST_BEGIN(test_basic)
 }
 TEST_END
 
-TEST_BEGIN(test_alignment_errors)
-{
-	void *p;
-	size_t nsz, rsz, sz, alignment;
-
-#if LG_SIZEOF_PTR == 3
-	alignment = UINT64_C(0x4000000000000000);
-	sz        = UINT64_C(0x8400000000000001);
-#else
-	alignment = 0x40000000LU;
-	sz        = 0x84000001LU;
-#endif
-	nsz = 0;
-	assert_d_eq(nallocm(&nsz, sz, ALLOCM_ALIGN(alignment)), ALLOCM_SUCCESS,
-	    "Unexpected nallocm() error");
-	rsz = 0;
-	assert_d_ne(allocm(&p, &rsz, sz, ALLOCM_ALIGN(alignment)),
-	    ALLOCM_SUCCESS, "Expected error for allocm(&p, %zu, %#x)",
-	    sz, ALLOCM_ALIGN(alignment));
-}
-TEST_END
-
 TEST_BEGIN(test_alignment_and_size)
 {
 	int r;
@@ -126,6 +104,5 @@ main(void)
 
 	return (test(
 	    test_basic,
-	    test_alignment_errors,
 	    test_alignment_and_size));
 }

test/integration/mallocx.c

@@ -34,26 +34,6 @@ TEST_BEGIN(test_basic)
 }
 TEST_END
 
-TEST_BEGIN(test_alignment_errors)
-{
-	void *p;
-	size_t nsz, sz, alignment;
-
-#if LG_SIZEOF_PTR == 3
-	alignment = UINT64_C(0x4000000000000000);
-	sz        = UINT64_C(0x8400000000000001);
-#else
-	alignment = 0x40000000LU;
-	sz        = 0x84000001LU;
-#endif
-	nsz = nallocx(sz, MALLOCX_ALIGN(alignment));
-	assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
-	p = mallocx(sz, MALLOCX_ALIGN(alignment));
-	assert_ptr_null(p, "Expected error for mallocx(%zu, %#x)", sz,
-	    MALLOCX_ALIGN(alignment));
-}
-TEST_END
-
 TEST_BEGIN(test_alignment_and_size)
 {
 	size_t nsz, rsz, sz, alignment, total;
@@ -114,6 +94,5 @@ main(void)
 
 	return (test(
 	    test_basic,
-	    test_alignment_errors,
 	    test_alignment_and_size));
 }

test/integration/posix_memalign.c

@@ -43,10 +43,10 @@ TEST_BEGIN(test_oom_errors)
 
 #if LG_SIZEOF_PTR == 3
 	alignment = UINT64_C(0x4000000000000000);
-	size      = UINT64_C(0x8400000000000001);
+	size      = UINT64_C(0xc000000000000001);
 #else
 	alignment = 0x40000000LU;
-	size      = 0x84000001LU;
+	size      = 0xc0000001LU;
 #endif
 	assert_d_ne(posix_memalign(&p, alignment, size), 0,
 	    "Expected error for posix_memalign(&p, %zu, %zu)",

test/unit/junk.c

@@ -92,9 +92,12 @@ test_junk(size_t sz_min, size_t sz_max)
 			s = (char *)rallocx(s, sz+1, 0);
 			assert_ptr_not_null((void *)s,
 			    "Unexpected rallocx() failure");
-			assert_ptr_eq(most_recently_junked, junked,
-			    "Expected region of size %zu to be junk-filled",
-			    sz);
+			if (!config_mremap || sz+1 <= arena_maxclass) {
+				assert_ptr_eq(most_recently_junked, junked,
+				    "Expected region of size %zu to be "
+				    "junk-filled",
+				    sz);
+			}
 		}
 	}
 

test/unit/prof_accum.c

@@ -1,9 +1,4 @@
-#include "test/jemalloc_test.h"
-
-#define	NTHREADS		4
-#define	NALLOCS_PER_THREAD	50
-#define	DUMP_INTERVAL		1
-#define	BT_COUNT_CHECK_INTERVAL	5
+#include "prof_accum.h"
 
 #ifdef JEMALLOC_PROF
 const char *malloc_conf =
@@ -21,37 +16,6 @@ prof_dump_open_intercept(bool propagate_err, const char *filename)
 	return (fd);
 }
 
-#define	alloc_n_proto(n)						\
-static void	*alloc_##n(unsigned bits);
-
-#define	alloc_n_gen(n)							\
-static void *								\
-alloc_##n(unsigned bits)						\
-{									\
-	void *p;							\
-									\
-	if (bits == 0)							\
-		p = mallocx(1, 0);					\
-	else {								\
-		switch (bits & 0x1U) {					\
-		case 0:							\
-			p = alloc_0(bits >> 1);				\
-			break;						\
-		case 1:							\
-			p = alloc_1(bits >> 1);				\
-			break;						\
-		default: not_reached();					\
-		}							\
-	}								\
-	/* Intentionally sabotage tail call optimization. */		\
-	assert_ptr_not_null(p, "Unexpected mallocx() failure");		\
-	return (p);							\
-}
-alloc_n_proto(0)
-alloc_n_proto(1)
-alloc_n_gen(0)
-alloc_n_gen(1)
-
 static void *
 alloc_from_permuted_backtrace(unsigned thd_ind, unsigned iteration)
 {

test/unit/prof_accum.h

@@ -0,0 +1,35 @@
+#include "test/jemalloc_test.h"
+
+#define	NTHREADS		4
+#define	NALLOCS_PER_THREAD	50
+#define	DUMP_INTERVAL		1
+#define	BT_COUNT_CHECK_INTERVAL	5
+
+#define	alloc_n_proto(n)						\
+void	*alloc_##n(unsigned bits);
+alloc_n_proto(0)
+alloc_n_proto(1)
+
+#define	alloc_n_gen(n)							\
+void *									\
+alloc_##n(unsigned bits)						\
+{									\
+	void *p;							\
+									\
+	if (bits == 0)							\
+		p = mallocx(1, 0);					\
+	else {								\
+		switch (bits & 0x1U) {					\
+		case 0:							\
+			p = (alloc_0(bits >> 1));			\
+			break;						\
+		case 1:							\
+			p = (alloc_1(bits >> 1));			\
+			break;						\
+		default: not_reached();					\
+		}							\
+	}								\
+	/* Intentionally sabotage tail call optimization. */		\
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");		\
+	return (p);							\
+}

test/unit/prof_accum_a.c

@@ -0,0 +1,3 @@
+#include "prof_accum.h"
+
+alloc_n_gen(0)

test/unit/prof_accum_b.c

@@ -0,0 +1,3 @@
+#include "prof_accum.h"
+
+alloc_n_gen(1)

test/unit/stats.c

@@ -31,21 +31,25 @@ TEST_END
 
 TEST_BEGIN(test_stats_chunks)
 {
-	size_t current, total, high;
-	size_t sz = sizeof(size_t);
+	size_t current, high;
+	uint64_t total;
+	size_t sz;
 	int expected = config_stats ? 0 : ENOENT;
 
+	sz = sizeof(size_t);
 	assert_d_eq(mallctl("stats.chunks.current", &current, &sz, NULL, 0),
 	    expected, "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
 	assert_d_eq(mallctl("stats.chunks.total", &total, &sz, NULL, 0),
 	    expected, "Unexpected mallctl() result");
+	sz = sizeof(size_t);
 	assert_d_eq(mallctl("stats.chunks.high", &high, &sz, NULL, 0), expected,
 	    "Unexpected mallctl() result");
 
 	if (config_stats) {
 		assert_zu_le(current, high,
 		    "current should be no larger than high");
-		assert_zu_le(high, total,
+		assert_u64_le((uint64_t)high, total,
 		    "high should be no larger than total");
 	}
 }
@@ -247,9 +251,9 @@ TEST_BEGIN(test_stats_arenas_bins)
 {
 	unsigned arena;
 	void *p;
-	size_t sz, allocated;
+	size_t sz, allocated, curruns;
 	uint64_t epoch, nmalloc, ndalloc, nrequests, nfills, nflushes;
-	uint64_t nruns, nreruns, curruns;
+	uint64_t nruns, nreruns;
 	int expected = config_stats ? 0 : ENOENT;
 
 	arena = 0;
@@ -287,6 +291,7 @@ TEST_BEGIN(test_stats_arenas_bins)
 	    NULL, 0), expected, "Unexpected mallctl() result");
 	assert_d_eq(mallctl("stats.arenas.0.bins.0.nreruns", &nreruns, &sz,
 	    NULL, 0), expected, "Unexpected mallctl() result");
+	sz = sizeof(size_t);
 	assert_d_eq(mallctl("stats.arenas.0.bins.0.curruns", &curruns, &sz,
 	    NULL, 0), expected, "Unexpected mallctl() result");
 
@@ -307,7 +312,7 @@ TEST_BEGIN(test_stats_arenas_bins)
 		}
 		assert_u64_gt(nruns, 0,
 		    "At least one run should have been allocated");
-		assert_u64_gt(curruns, 0,
+		assert_zu_gt(curruns, 0,
 		    "At least one run should be currently allocated");
 	}
 
@@ -319,8 +324,8 @@ TEST_BEGIN(test_stats_arenas_lruns)
 {
 	unsigned arena;
 	void *p;
-	uint64_t epoch, nmalloc, ndalloc, nrequests, curruns;
-	size_t sz = sizeof(uint64_t);
+	uint64_t epoch, nmalloc, ndalloc, nrequests;
+	size_t curruns, sz;
 	int expected = config_stats ? 0 : ENOENT;
 
 	arena = 0;
@@ -333,12 +338,14 @@ TEST_BEGIN(test_stats_arenas_lruns)
 	assert_d_eq(mallctl("epoch", NULL, NULL, &epoch, sizeof(epoch)), 0,
 	    "Unexpected mallctl() failure");
 
+	sz = sizeof(uint64_t);
 	assert_d_eq(mallctl("stats.arenas.0.lruns.0.nmalloc", &nmalloc, &sz,
 	    NULL, 0), expected, "Unexpected mallctl() result");
 	assert_d_eq(mallctl("stats.arenas.0.lruns.0.ndalloc", &ndalloc, &sz,
 	    NULL, 0), expected, "Unexpected mallctl() result");
 	assert_d_eq(mallctl("stats.arenas.0.lruns.0.nrequests", &nrequests, &sz,
 	    NULL, 0), expected, "Unexpected mallctl() result");
+	sz = sizeof(size_t);
 	assert_d_eq(mallctl("stats.arenas.0.lruns.0.curruns", &curruns, &sz,
 	    NULL, 0), expected, "Unexpected mallctl() result");