Header refactoring: Pull size helpers out of jemalloc module.

28 files changed, 642 insertions, 596 deletions

diff --git a/Makefile.in b/Makefile.in
index 868bf8c..fec1397 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -112,6 +112,7 @@ C_SRCS := $(srcroot)src/jemalloc.c \
 	$(srcroot)src/rtree.c \
 	$(srcroot)src/stats.c \
 	$(srcroot)src/spin.c \
+	$(srcroot)src/sz.c \
 	$(srcroot)src/tcache.c \
 	$(srcroot)src/ticker.c \
 	$(srcroot)src/tsd.c \
diff --git a/include/jemalloc/internal/arena_externs.h b/include/jemalloc/internal/arena_externs.h
index 08a6d17..cfb7c6f 100644
--- a/include/jemalloc/internal/arena_externs.h
+++ b/include/jemalloc/internal/arena_externs.h
@@ -6,14 +6,6 @@
 #include "jemalloc/internal/size_classes.h"
 #include "jemalloc/internal/stats.h"
 
-static const size_t large_pad =
-#ifdef JEMALLOC_CACHE_OBLIVIOUS
-    PAGE
-#else
-    0
-#endif
-    ;
-
 extern ssize_t opt_dirty_decay_ms;
 extern ssize_t opt_muzzy_decay_ms;
 
diff --git a/include/jemalloc/internal/arena_inlines_b.h b/include/jemalloc/internal/arena_inlines_b.h
index 16635c1..003abe1 100644
--- a/include/jemalloc/internal/arena_inlines_b.h
+++ b/include/jemalloc/internal/arena_inlines_b.h
@@ -5,6 +5,7 @@
 #include "jemalloc/internal/mutex.h"
 #include "jemalloc/internal/rtree.h"
 #include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/sz.h"
 #include "jemalloc/internal/ticker.h"
 
 static inline szind_t
@@ -127,7 +128,7 @@ arena_salloc(tsdn_t *tsdn, const void *ptr) {
 	    (uintptr_t)ptr, true);
 	assert(szind != NSIZES);
 
-	return index2size(szind);
+	return sz_index2size(szind);
 }
 
 JEMALLOC_ALWAYS_INLINE size_t
@@ -160,7 +161,7 @@ arena_vsalloc(tsdn_t *tsdn, const void *ptr) {
 
 	assert(szind != NSIZES);
 
-	return index2size(szind);
+	return sz_index2size(szind);
 }
 
 static inline void
@@ -257,7 +258,7 @@ arena_sdalloc_no_tcache(tsdn_t *tsdn, void *ptr, size_t size) {
 		 * There is no risk of being confused by a promoted sampled
 		 * object, so base szind and slab on the given size.
 		 */
-		szind = size2index(size);
+		szind = sz_size2index(size);
 		slab = (szind < NBINS);
 	}
 
@@ -269,7 +270,7 @@ arena_sdalloc_no_tcache(tsdn_t *tsdn, void *ptr, size_t size) {
 		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
 		    (uintptr_t)ptr, true, &szind, &slab);
 
-		assert(szind == size2index(size));
+		assert(szind == sz_size2index(size));
 		assert((config_prof && opt_prof) || slab == (szind < NBINS));
 
 		if (config_debug) {
@@ -313,7 +314,7 @@ arena_sdalloc(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
 			rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
 			    (uintptr_t)ptr, true, &local_ctx.szind,
 			    &local_ctx.slab);
-			assert(local_ctx.szind == size2index(size));
+			assert(local_ctx.szind == sz_size2index(size));
 			alloc_ctx = &local_ctx;
 		}
 		slab = alloc_ctx->slab;
@@ -323,7 +324,7 @@ arena_sdalloc(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
 		 * There is no risk of being confused by a promoted sampled
 		 * object, so base szind and slab on the given size.
 		 */
-		szind = size2index(size);
+		szind = sz_size2index(size);
 		slab = (szind < NBINS);
 	}
 
diff --git a/include/jemalloc/internal/arena_types.h b/include/jemalloc/internal/arena_types.h
index 1374eec..01b9096 100644
--- a/include/jemalloc/internal/arena_types.h
+++ b/include/jemalloc/internal/arena_types.h
@@ -1,8 +1,6 @@
 #ifndef JEMALLOC_INTERNAL_ARENA_TYPES_H
 #define JEMALLOC_INTERNAL_ARENA_TYPES_H
 
-#define LARGE_MINCLASS		(ZU(1) << LG_LARGE_MINCLASS)
-
 /* Maximum number of regions in one slab. */
 #define LG_SLAB_MAXREGS		(LG_PAGE - LG_TINY_MIN)
 #define SLAB_MAXREGS		(1U << LG_SLAB_MAXREGS)
diff --git a/include/jemalloc/internal/extent_inlines.h b/include/jemalloc/internal/extent_inlines.h
index 94c4192..bb2bd69 100644
--- a/include/jemalloc/internal/extent_inlines.h
+++ b/include/jemalloc/internal/extent_inlines.h
@@ -6,6 +6,7 @@
 #include "jemalloc/internal/pages.h"
 #include "jemalloc/internal/prng.h"
 #include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sz.h"
 
 static inline void
 extent_lock(tsdn_t *tsdn, extent_t *extent) {
@@ -65,7 +66,7 @@ extent_szind_get(const extent_t *extent) {
 
 static inline size_t
 extent_usize_get(const extent_t *extent) {
-	return index2size(extent_szind_get(extent));
+	return sz_index2size(extent_szind_get(extent));
 }
 
 static inline size_t
diff --git a/include/jemalloc/internal/extent_structs.h b/include/jemalloc/internal/extent_structs.h
index 457891d..d297950 100644
--- a/include/jemalloc/internal/extent_structs.h
+++ b/include/jemalloc/internal/extent_structs.h
@@ -53,7 +53,7 @@ struct extent_s {
 	 * szind: The szind flag indicates usable size class index for
 	 *        allocations residing in this extent, regardless of whether the
 	 *        extent is a slab.  Extent size and usable size often differ
-	 *        even for non-slabs, either due to large_pad or promotion of
+	 *        even for non-slabs, either due to sz_large_pad or promotion of
 	 *        sampled small regions.
 	 *
 	 * nfree: Number of free regions in slab.
diff --git a/include/jemalloc/internal/jemalloc_internal_externs.h b/include/jemalloc/internal/jemalloc_internal_externs.h
index 11e16ec..e10fb27 100644
--- a/include/jemalloc/internal/jemalloc_internal_externs.h
+++ b/include/jemalloc/internal/jemalloc_internal_externs.h
@@ -31,23 +31,6 @@ extern unsigned narenas_auto;
  */
 extern atomic_p_t arenas[];
 
-/*
- * pind2sz_tab encodes the same information as could be computed by
- * pind2sz_compute().
- */
-extern size_t const pind2sz_tab[NPSIZES+1];
-/*
- * index2size_tab encodes the same information as could be computed (at
- * unacceptable cost in some code paths) by index2size_compute().
- */
-extern size_t const index2size_tab[NSIZES];
-/*
- * size2index_tab is a compact lookup table that rounds request sizes up to
- * size classes.  In order to reduce cache footprint, the table is compressed,
- * and all accesses are via size2index().
- */
-extern uint8_t const size2index_tab[];
-
 void *a0malloc(size_t size);
 void a0dalloc(void *ptr);
 void *bootstrap_malloc(size_t size);
diff --git a/include/jemalloc/internal/jemalloc_internal_inlines_a.h b/include/jemalloc/internal/jemalloc_internal_inlines_a.h
index c8e2629..d0bf2ee 100644
--- a/include/jemalloc/internal/jemalloc_internal_inlines_a.h
+++ b/include/jemalloc/internal/jemalloc_internal_inlines_a.h
@@ -7,273 +7,6 @@
 #include "jemalloc/internal/size_classes.h"
 #include "jemalloc/internal/ticker.h"
 
-JEMALLOC_ALWAYS_INLINE pszind_t
-psz2ind(size_t psz) {
-	if (unlikely(psz > LARGE_MAXCLASS)) {
-		return NPSIZES;
-	}
-	{
-		pszind_t x = lg_floor((psz<<1)-1);
-		pszind_t shift = (x < LG_SIZE_CLASS_GROUP + LG_PAGE) ? 0 : x -
-		    (LG_SIZE_CLASS_GROUP + LG_PAGE);
-		pszind_t grp = shift << LG_SIZE_CLASS_GROUP;
-
-		pszind_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_PAGE + 1) ?
-		    LG_PAGE : x - LG_SIZE_CLASS_GROUP - 1;
-
-		size_t delta_inverse_mask = ZD(-1) << lg_delta;
-		pszind_t mod = ((((psz-1) & delta_inverse_mask) >> lg_delta)) &
-		    ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
-
-		pszind_t ind = grp + mod;
-		return ind;
-	}
-}
-
-static inline size_t
-pind2sz_compute(pszind_t pind) {
-	if (unlikely(pind == NPSIZES)) {
-		return LARGE_MAXCLASS + PAGE;
-	}
-	{
-		size_t grp = pind >> LG_SIZE_CLASS_GROUP;
-		size_t mod = pind & ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
-
-		size_t grp_size_mask = ~((!!grp)-1);
-		size_t grp_size = ((ZU(1) << (LG_PAGE +
-		    (LG_SIZE_CLASS_GROUP-1))) << grp) & grp_size_mask;
-
-		size_t shift = (grp == 0) ? 1 : grp;
-		size_t lg_delta = shift + (LG_PAGE-1);
-		size_t mod_size = (mod+1) << lg_delta;
-
-		size_t sz = grp_size + mod_size;
-		return sz;
-	}
-}
-
-static inline size_t
-pind2sz_lookup(pszind_t pind) {
-	size_t ret = (size_t)pind2sz_tab[pind];
-	assert(ret == pind2sz_compute(pind));
-	return ret;
-}
-
-static inline size_t
-pind2sz(pszind_t pind) {
-	assert(pind < NPSIZES+1);
-	return pind2sz_lookup(pind);
-}
-
-static inline size_t
-psz2u(size_t psz) {
-	if (unlikely(psz > LARGE_MAXCLASS)) {
-		return LARGE_MAXCLASS + PAGE;
-	}
-	{
-		size_t x = lg_floor((psz<<1)-1);
-		size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_PAGE + 1) ?
-		    LG_PAGE : x - LG_SIZE_CLASS_GROUP - 1;
-		size_t delta = ZU(1) << lg_delta;
-		size_t delta_mask = delta - 1;
-		size_t usize = (psz + delta_mask) & ~delta_mask;
-		return usize;
-	}
-}
-
-static inline szind_t
-size2index_compute(size_t size) {
-	if (unlikely(size > LARGE_MAXCLASS)) {
-		return NSIZES;
-	}
-#if (NTBINS != 0)
-	if (size <= (ZU(1) << LG_TINY_MAXCLASS)) {
-		szind_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1;
-		szind_t lg_ceil = lg_floor(pow2_ceil_zu(size));
-		return (lg_ceil < lg_tmin ? 0 : lg_ceil - lg_tmin);
-	}
-#endif
-	{
-		szind_t x = lg_floor((size<<1)-1);
-		szind_t shift = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM) ? 0 :
-		    x - (LG_SIZE_CLASS_GROUP + LG_QUANTUM);
-		szind_t grp = shift << LG_SIZE_CLASS_GROUP;
-
-		szind_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1)
-		    ? LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1;
-
-		size_t delta_inverse_mask = ZD(-1) << lg_delta;
-		szind_t mod = ((((size-1) & delta_inverse_mask) >> lg_delta)) &
-		    ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
-
-		szind_t index = NTBINS + grp + mod;
-		return index;
-	}
-}
-
-JEMALLOC_ALWAYS_INLINE szind_t
-size2index_lookup(size_t size) {
-	assert(size <= LOOKUP_MAXCLASS);
-	{
-		szind_t ret = (size2index_tab[(size-1) >> LG_TINY_MIN]);
-		assert(ret == size2index_compute(size));
-		return ret;
-	}
-}
-
-JEMALLOC_ALWAYS_INLINE szind_t
-size2index(size_t size) {
-	assert(size > 0);
-	if (likely(size <= LOOKUP_MAXCLASS)) {
-		return size2index_lookup(size);
-	}
-	return size2index_compute(size);
-}
-
-static inline size_t
-index2size_compute(szind_t index) {
-#if (NTBINS > 0)
-	if (index < NTBINS) {
-		return (ZU(1) << (LG_TINY_MAXCLASS - NTBINS + 1 + index));
-	}
-#endif
-	{
-		size_t reduced_index = index - NTBINS;
-		size_t grp = reduced_index >> LG_SIZE_CLASS_GROUP;
-		size_t mod = reduced_index & ((ZU(1) << LG_SIZE_CLASS_GROUP) -
-		    1);
-
-		size_t grp_size_mask = ~((!!grp)-1);
-		size_t grp_size = ((ZU(1) << (LG_QUANTUM +
-		    (LG_SIZE_CLASS_GROUP-1))) << grp) & grp_size_mask;
-
-		size_t shift = (grp == 0) ? 1 : grp;
-		size_t lg_delta = shift + (LG_QUANTUM-1);
-		size_t mod_size = (mod+1) << lg_delta;
-
-		size_t usize = grp_size + mod_size;
-		return usize;
-	}
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-index2size_lookup(szind_t index) {
-	size_t ret = (size_t)index2size_tab[index];
-	assert(ret == index2size_compute(index));
-	return ret;
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-index2size(szind_t index) {
-	assert(index < NSIZES);
-	return index2size_lookup(index);
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-s2u_compute(size_t size) {
-	if (unlikely(size > LARGE_MAXCLASS)) {
-		return 0;
-	}
-#if (NTBINS > 0)
-	if (size <= (ZU(1) << LG_TINY_MAXCLASS)) {
-		size_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1;
-		size_t lg_ceil = lg_floor(pow2_ceil_zu(size));
-		return (lg_ceil < lg_tmin ? (ZU(1) << lg_tmin) :
-		    (ZU(1) << lg_ceil));
-	}
-#endif
-	{
-		size_t x = lg_floor((size<<1)-1);
-		size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1)
-		    ?  LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1;
-		size_t delta = ZU(1) << lg_delta;
-		size_t delta_mask = delta - 1;
-		size_t usize = (size + delta_mask) & ~delta_mask;
-		return usize;
-	}
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-s2u_lookup(size_t size) {
-	size_t ret = index2size_lookup(size2index_lookup(size));
-
-	assert(ret == s2u_compute(size));
-	return ret;
-}
-
-/*
- * Compute usable size that would result from allocating an object with the
- * specified size.
- */
-JEMALLOC_ALWAYS_INLINE size_t
-s2u(size_t size) {
-	assert(size > 0);
-	if (likely(size <= LOOKUP_MAXCLASS)) {
-		return s2u_lookup(size);
-	}
-	return s2u_compute(size);
-}
-
-/*
- * Compute usable size that would result from allocating an object with the
- * specified size and alignment.
- */
-JEMALLOC_ALWAYS_INLINE size_t
-sa2u(size_t size, size_t alignment) {
-	size_t usize;
-
-	assert(alignment != 0 && ((alignment - 1) & alignment) == 0);
-
-	/* Try for a small size class. */
-	if (size <= SMALL_MAXCLASS && alignment < PAGE) {
-		/*
-		 * Round size up to the nearest multiple of alignment.
-		 *
-		 * This done, we can take advantage of the fact that for each
-		 * small size class, every object is aligned at the smallest
-		 * power of two that is non-zero in the base two representation
-		 * of the size.  For example:
-		 *
-		 *   Size |   Base 2 | Minimum alignment
-		 *   -----+----------+------------------
-		 *     96 |  1100000 |  32
-		 *    144 | 10100000 |  32
-		 *    192 | 11000000 |  64
-		 */
-		usize = s2u(ALIGNMENT_CEILING(size, alignment));
-		if (usize < LARGE_MINCLASS) {
-			return usize;
-		}
-	}
-
-	/* Large size class.  Beware of overflow. */
-
-	if (unlikely(alignment > LARGE_MAXCLASS)) {
-		return 0;
-	}
-
-	/* Make sure result is a large size class. */
-	if (size <= LARGE_MINCLASS) {
-		usize = LARGE_MINCLASS;
-	} else {
-		usize = s2u(size);
-		if (usize < size) {
-			/* size_t overflow. */
-			return 0;
-		}
-	}
-
-	/*
-	 * Calculate the multi-page mapping that large_palloc() would need in
-	 * order to guarantee the alignment.
-	 */
-	if (usize + large_pad + PAGE_CEILING(alignment) - PAGE < usize) {
-		/* size_t overflow. */
-		return 0;
-	}
-	return usize;
-}
-
 JEMALLOC_ALWAYS_INLINE malloc_cpuid_t
 malloc_getcpu(void) {
 	assert(have_percpu_arena);
diff --git a/include/jemalloc/internal/jemalloc_internal_inlines_c.h b/include/jemalloc/internal/jemalloc_internal_inlines_c.h
index 80dfbef..7ffce6f 100644
--- a/include/jemalloc/internal/jemalloc_internal_inlines_c.h
+++ b/include/jemalloc/internal/jemalloc_internal_inlines_c.h
@@ -2,6 +2,7 @@
 #define JEMALLOC_INTERNAL_INLINES_C_H
 
 #include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sz.h"
 #include "jemalloc/internal/witness.h"
 
 JEMALLOC_ALWAYS_INLINE arena_t *
@@ -48,7 +49,7 @@ ipallocztm(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
 	void *ret;
 
 	assert(usize != 0);
-	assert(usize == sa2u(usize, alignment));
+	assert(usize == sz_sa2u(usize, alignment));
 	assert(!is_internal || tcache == NULL);
 	assert(!is_internal || arena == NULL || arena_is_auto(arena));
 	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
@@ -118,7 +119,7 @@ iralloct_realign(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
 	void *p;
 	size_t usize, copysize;
 
-	usize = sa2u(size + extra, alignment);
+	usize = sz_sa2u(size + extra, alignment);
 	if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 		return NULL;
 	}
@@ -128,7 +129,7 @@ iralloct_realign(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
 			return NULL;
 		}
 		/* Try again, without extra this time. */
-		usize = sa2u(size, alignment);
+		usize = sz_sa2u(size, alignment);
 		if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 			return NULL;
 		}
diff --git a/include/jemalloc/internal/prof_inlines_b.h b/include/jemalloc/internal/prof_inlines_b.h
index fba7b99..d670cb7 100644
--- a/include/jemalloc/internal/prof_inlines_b.h
+++ b/include/jemalloc/internal/prof_inlines_b.h
@@ -1,6 +1,8 @@
 #ifndef JEMALLOC_INTERNAL_PROF_INLINES_B_H
 #define JEMALLOC_INTERNAL_PROF_INLINES_B_H
 
+#include "jemalloc/internal/sz.h"
+
 JEMALLOC_ALWAYS_INLINE bool
 prof_active_get_unlocked(void) {
 	/*
@@ -113,7 +115,7 @@ prof_alloc_prep(tsd_t *tsd, size_t usize, bool prof_active, bool update) {
 	prof_tdata_t *tdata;
 	prof_bt_t bt;
 
-	assert(usize == s2u(usize));
+	assert(usize == sz_s2u(usize));
 
 	if (!prof_active || likely(prof_sample_accum_update(tsd, usize, update,
 	    &tdata))) {
diff --git a/include/jemalloc/internal/size_classes.sh b/include/jemalloc/internal/size_classes.sh
index dd562db..998994d 100755
--- a/include/jemalloc/internal/size_classes.sh
+++ b/include/jemalloc/internal/size_classes.sh
@@ -334,6 +334,7 @@ for lg_z in ${lg_zarr} ; do
         echo "#define LOOKUP_MAXCLASS		${lookup_maxclass}"
         echo "#define SMALL_MAXCLASS		${small_maxclass}"
         echo "#define LG_LARGE_MINCLASS	${lg_large_minclass}"
+        echo "#define LARGE_MINCLASS		(ZU(1) << LG_LARGE_MINCLASS)"
         echo "#define LARGE_MAXCLASS		${large_maxclass}"
         echo "#endif"
         echo
diff --git a/include/jemalloc/internal/sz.h b/include/jemalloc/internal/sz.h
new file mode 100644
index 0000000..7f640d5
--- /dev/null
+++ b/include/jemalloc/internal/sz.h
@@ -0,0 +1,317 @@
+#ifndef JEMALLOC_INTERNAL_SIZE_H
+#define JEMALLOC_INTERNAL_SIZE_H
+
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/util.h"
+
+/*
+ * sz module: Size computations.
+ *
+ * Some abbreviations used here:
+ *   p: Page
+ *   ind: Index
+ *   s, sz: Size
+ *   u: Usable size
+ *   a: Aligned
+ *
+ * These are not always used completely consistently, but should be enough to
+ * interpret function names.  E.g. sz_psz2ind converts page size to page size
+ * index; sz_sa2u converts a (size, alignment) allocation request to the usable
+ * size that would result from such an allocation.
+ */
+
+/*
+ * sz_pind2sz_tab encodes the same information as could be computed by
+ * sz_pind2sz_compute().
+ */
+extern size_t const sz_pind2sz_tab[NPSIZES+1];
+/*
+ * sz_index2size_tab encodes the same information as could be computed (at
+ * unacceptable cost in some code paths) by sz_index2size_compute().
+ */
+extern size_t const sz_index2size_tab[NSIZES];
+/*
+ * sz_size2index_tab is a compact lookup table that rounds request sizes up to
+ * size classes.  In order to reduce cache footprint, the table is compressed,
+ * and all accesses are via sz_size2index().
+ */
+extern uint8_t const sz_size2index_tab[];
+
+static const size_t sz_large_pad =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+    PAGE
+#else
+    0
+#endif
+    ;
+
+JEMALLOC_ALWAYS_INLINE pszind_t
+sz_psz2ind(size_t psz) {
+	if (unlikely(psz > LARGE_MAXCLASS)) {
+		return NPSIZES;
+	}
+	{
+		pszind_t x = lg_floor((psz<<1)-1);
+		pszind_t shift = (x < LG_SIZE_CLASS_GROUP + LG_PAGE) ? 0 : x -
+		    (LG_SIZE_CLASS_GROUP + LG_PAGE);
+		pszind_t grp = shift << LG_SIZE_CLASS_GROUP;
+
+		pszind_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_PAGE + 1) ?
+		    LG_PAGE : x - LG_SIZE_CLASS_GROUP - 1;
+
+		size_t delta_inverse_mask = ZD(-1) << lg_delta;
+		pszind_t mod = ((((psz-1) & delta_inverse_mask) >> lg_delta)) &
+		    ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
+
+		pszind_t ind = grp + mod;
+		return ind;
+	}
+}
+
+static inline size_t
+sz_pind2sz_compute(pszind_t pind) {
+	if (unlikely(pind == NPSIZES)) {
+		return LARGE_MAXCLASS + PAGE;
+	}
+	{
+		size_t grp = pind >> LG_SIZE_CLASS_GROUP;
+		size_t mod = pind & ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
+
+		size_t grp_size_mask = ~((!!grp)-1);
+		size_t grp_size = ((ZU(1) << (LG_PAGE +
+		    (LG_SIZE_CLASS_GROUP-1))) << grp) & grp_size_mask;
+
+		size_t shift = (grp == 0) ? 1 : grp;
+		size_t lg_delta = shift + (LG_PAGE-1);
+		size_t mod_size = (mod+1) << lg_delta;
+
+		size_t sz = grp_size + mod_size;
+		return sz;
+	}
+}
+
+static inline size_t
+sz_pind2sz_lookup(pszind_t pind) {
+	size_t ret = (size_t)sz_pind2sz_tab[pind];
+	assert(ret == sz_pind2sz_compute(pind));
+	return ret;
+}
+
+static inline size_t
+sz_pind2sz(pszind_t pind) {
+	assert(pind < NPSIZES+1);
+	return sz_pind2sz_lookup(pind);
+}
+
+static inline size_t
+sz_psz2u(size_t psz) {
+	if (unlikely(psz > LARGE_MAXCLASS)) {
+		return LARGE_MAXCLASS + PAGE;
+	}
+	{
+		size_t x = lg_floor((psz<<1)-1);
+		size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_PAGE + 1) ?
+		    LG_PAGE : x - LG_SIZE_CLASS_GROUP - 1;
+		size_t delta = ZU(1) << lg_delta;
+		size_t delta_mask = delta - 1;
+		size_t usize = (psz + delta_mask) & ~delta_mask;
+		return usize;
+	}
+}
+
+static inline szind_t
+sz_size2index_compute(size_t size) {
+	if (unlikely(size > LARGE_MAXCLASS)) {
+		return NSIZES;
+	}
+#if (NTBINS != 0)
+	if (size <= (ZU(1) << LG_TINY_MAXCLASS)) {
+		szind_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1;
+		szind_t lg_ceil = lg_floor(pow2_ceil_zu(size));
+		return (lg_ceil < lg_tmin ? 0 : lg_ceil - lg_tmin);
+	}
+#endif
+	{
+		szind_t x = lg_floor((size<<1)-1);
+		szind_t shift = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM) ? 0 :
+		    x - (LG_SIZE_CLASS_GROUP + LG_QUANTUM);
+		szind_t grp = shift << LG_SIZE_CLASS_GROUP;
+
+		szind_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1)
+		    ? LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1;
+
+		size_t delta_inverse_mask = ZD(-1) << lg_delta;
+		szind_t mod = ((((size-1) & delta_inverse_mask) >> lg_delta)) &
+		    ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
+
+		szind_t index = NTBINS + grp + mod;
+		return index;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+sz_size2index_lookup(size_t size) {
+	assert(size <= LOOKUP_MAXCLASS);
+	{
+		szind_t ret = (sz_size2index_tab[(size-1) >> LG_TINY_MIN]);
+		assert(ret == sz_size2index_compute(size));
+		return ret;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+sz_size2index(size_t size) {
+	assert(size > 0);
+	if (likely(size <= LOOKUP_MAXCLASS)) {
+		return sz_size2index_lookup(size);
+	}
+	return sz_size2index_compute(size);
+}
+
+static inline size_t
+sz_index2size_compute(szind_t index) {
+#if (NTBINS > 0)
+	if (index < NTBINS) {
+		return (ZU(1) << (LG_TINY_MAXCLASS - NTBINS + 1 + index));
+	}
+#endif
+	{
+		size_t reduced_index = index - NTBINS;
+		size_t grp = reduced_index >> LG_SIZE_CLASS_GROUP;
+		size_t mod = reduced_index & ((ZU(1) << LG_SIZE_CLASS_GROUP) -
+		    1);
+
+		size_t grp_size_mask = ~((!!grp)-1);
+		size_t grp_size = ((ZU(1) << (LG_QUANTUM +
+		    (LG_SIZE_CLASS_GROUP-1))) << grp) & grp_size_mask;
+
+		size_t shift = (grp == 0) ? 1 : grp;
+		size_t lg_delta = shift + (LG_QUANTUM-1);
+		size_t mod_size = (mod+1) << lg_delta;
+
+		size_t usize = grp_size + mod_size;
+		return usize;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_index2size_lookup(szind_t index) {
+	size_t ret = (size_t)sz_index2size_tab[index];
+	assert(ret == sz_index2size_compute(index));
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_index2size(szind_t index) {
+	assert(index < NSIZES);
+	return sz_index2size_lookup(index);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u_compute(size_t size) {
+	if (unlikely(size > LARGE_MAXCLASS)) {
+		return 0;
+	}
+#if (NTBINS > 0)
+	if (size <= (ZU(1) << LG_TINY_MAXCLASS)) {
+		size_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1;
+		size_t lg_ceil = lg_floor(pow2_ceil_zu(size));
+		return (lg_ceil < lg_tmin ? (ZU(1) << lg_tmin) :
+		    (ZU(1) << lg_ceil));
+	}
+#endif
+	{
+		size_t x = lg_floor((size<<1)-1);
+		size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1)
+		    ?  LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1;
+		size_t delta = ZU(1) << lg_delta;
+		size_t delta_mask = delta - 1;
+		size_t usize = (size + delta_mask) & ~delta_mask;
+		return usize;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u_lookup(size_t size) {
+	size_t ret = sz_index2size_lookup(sz_size2index_lookup(size));
+
+	assert(ret == sz_s2u_compute(size));
+	return ret;
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u(size_t size) {
+	assert(size > 0);
+	if (likely(size <= LOOKUP_MAXCLASS)) {
+		return sz_s2u_lookup(size);
+	}
+	return sz_s2u_compute(size);
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size and alignment.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sz_sa2u(size_t size, size_t alignment) {
+	size_t usize;
+
+	assert(alignment != 0 && ((alignment - 1) & alignment) == 0);
+
+	/* Try for a small size class. */
+	if (size <= SMALL_MAXCLASS && alignment < PAGE) {
+		/*
+		 * Round size up to the nearest multiple of alignment.
+		 *
+		 * This done, we can take advantage of the fact that for each
+		 * small size class, every object is aligned at the smallest
+		 * power of two that is non-zero in the base two representation
+		 * of the size.  For example:
+		 *
+		 *   Size |   Base 2 | Minimum alignment
+		 *   -----+----------+------------------
+		 *     96 |  1100000 |  32
+		 *    144 | 10100000 |  32
+		 *    192 | 11000000 |  64
+		 */
+		usize = sz_s2u(ALIGNMENT_CEILING(size, alignment));
+		if (usize < LARGE_MINCLASS) {
+			return usize;
+		}
+	}
+
+	/* Large size class.  Beware of overflow. */
+
+	if (unlikely(alignment > LARGE_MAXCLASS)) {
+		return 0;
+	}
+
+	/* Make sure result is a large size class. */
+	if (size <= LARGE_MINCLASS) {
+		usize = LARGE_MINCLASS;
+	} else {
+		usize = sz_s2u(size);
+		if (usize < size) {
+			/* size_t overflow. */
+			return 0;
+		}
+	}
+
+	/*
+	 * Calculate the multi-page mapping that large_palloc() would need in
+	 * order to guarantee the alignment.
+	 */
+	if (usize + sz_large_pad + PAGE_CEILING(alignment) - PAGE < usize) {
+		/* size_t overflow. */
+		return 0;
+	}
+	return usize;
+}
+
+#endif /* JEMALLOC_INTERNAL_SIZE_H */
diff --git a/include/jemalloc/internal/tcache_inlines.h b/include/jemalloc/internal/tcache_inlines.h
index 8a65ba2..c55bcd2 100644
--- a/include/jemalloc/internal/tcache_inlines.h
+++ b/include/jemalloc/internal/tcache_inlines.h
@@ -3,6 +3,7 @@
 
 #include "jemalloc/internal/jemalloc_internal_types.h"
 #include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/sz.h"
 #include "jemalloc/internal/ticker.h"
 #include "jemalloc/internal/util.h"
 
@@ -95,7 +96,7 @@ tcache_alloc_small(tsd_t *tsd, arena_t *arena, tcache_t *tcache, size_t size,
 	 * statement are all static.
 	 */
 	if (config_prof || (slow_path && config_fill) || unlikely(zero)) {
-		usize = index2size(binind);
+		usize = sz_index2size(binind);
 		assert(tcache_salloc(tsd_tsdn(tsd), ret) == usize);
 	}
 
@@ -147,7 +148,7 @@ tcache_alloc_large(tsd_t *tsd, arena_t *arena, tcache_t *tcache, size_t size,
 			return NULL;
 		}
 
-		ret = large_malloc(tsd_tsdn(tsd), arena, s2u(size), zero);
+		ret = large_malloc(tsd_tsdn(tsd), arena, sz_s2u(size), zero);
 		if (ret == NULL) {
 			return NULL;
 		}
@@ -157,7 +158,7 @@ tcache_alloc_large(tsd_t *tsd, arena_t *arena, tcache_t *tcache, size_t size,
 		/* Only compute usize on demand */
 		if (config_prof || (slow_path && config_fill) ||
 		    unlikely(zero)) {
-			usize = index2size(binind);
+			usize = sz_index2size(binind);
 			assert(usize <= tcache_maxclass);
 		}
 
@@ -221,7 +222,7 @@ tcache_dalloc_large(tsd_t *tsd, tcache_t *tcache, void *ptr, szind_t binind,
 	assert(tcache_salloc(tsd_tsdn(tsd), ptr) <= tcache_maxclass);
 
 	if (slow_path && config_fill && unlikely(opt_junk_free)) {
-		large_dalloc_junk(ptr, index2size(binind));
+		large_dalloc_junk(ptr, sz_index2size(binind));
 	}
 
 	tbin = tcache_large_bin_get(tcache, binind);
diff --git a/src/arena.c b/src/arena.c
index bf1d6e6..151aad3 100644
--- a/src/arena.c
+++ b/src/arena.c
@@ -289,7 +289,7 @@ arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
 		size_t curlextents = (size_t)(nmalloc - ndalloc);
 		lstats[i].curlextents += curlextents;
 		arena_stats_accum_zu(&astats->allocated_large,
-		    curlextents * index2size(NBINS + i));
+		    curlextents * sz_index2size(NBINS + i));
 	}
 
 	arena_stats_unlock(tsdn, &arena->stats);
@@ -303,12 +303,12 @@ arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
 		for (; i < NBINS; i++) {
 			tcache_bin_t *tbin = tcache_small_bin_get(tcache, i);
 			arena_stats_accum_zu(&astats->tcache_bytes,
-			    tbin->ncached * index2size(i));
+			    tbin->ncached * sz_index2size(i));
 		}
 		for (; i < nhbins; i++) {
 			tcache_bin_t *tbin = tcache_large_bin_get(tcache, i);
 			arena_stats_accum_zu(&astats->tcache_bytes,
-			    tbin->ncached * index2size(i));
+			    tbin->ncached * sz_index2size(i));
 		}
 	}
 	malloc_mutex_prof_read(tsdn,
@@ -467,7 +467,7 @@ arena_large_malloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t usize) {
 	if (usize < LARGE_MINCLASS) {
 		usize = LARGE_MINCLASS;
 	}
-	index = size2index(usize);
+	index = sz_size2index(usize);
 	hindex = (index >= NBINS) ? index - NBINS : 0;
 
 	arena_stats_add_u64(tsdn, &arena->stats,
@@ -483,7 +483,7 @@ arena_large_dalloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t usize) {
 	if (usize < LARGE_MINCLASS) {
 		usize = LARGE_MINCLASS;
 	}
-	index = size2index(usize);
+	index = sz_size2index(usize);
 	hindex = (index >= NBINS) ? index - NBINS : 0;
 
 	arena_stats_add_u64(tsdn, &arena->stats,
@@ -505,21 +505,22 @@ arena_extent_alloc_large(tsdn_t *tsdn, arena_t *arena, size_t usize,
 	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
 	    WITNESS_RANK_CORE, 0);
 
-	szind_t szind = size2index(usize);
+	szind_t szind = sz_size2index(usize);
 	size_t mapped_add;
 	bool commit = true;
 	extent_t *extent = extents_alloc(tsdn, arena, &extent_hooks,
-	    &arena->extents_dirty, NULL, usize, large_pad, alignment, false,
+	    &arena->extents_dirty, NULL, usize, sz_large_pad, alignment, false,
 	    szind, zero, &commit);
 	if (extent == NULL) {
 		extent = extents_alloc(tsdn, arena, &extent_hooks,
-		    &arena->extents_muzzy, NULL, usize, large_pad, alignment,
+		    &arena->extents_muzzy, NULL, usize, sz_large_pad, alignment,
 		    false, szind, zero, &commit);
 	}
-	size_t size = usize + large_pad;
+	size_t size = usize + sz_large_pad;
 	if (extent == NULL) {
 		extent = extent_alloc_wrapper(tsdn, arena, &extent_hooks, NULL,
-		    usize, large_pad, alignment, false, szind, zero, &commit);
+		    usize, sz_large_pad, alignment, false, szind, zero,
+		    &commit);
 		if (config_stats) {
 			/*
 			 * extent may be NULL on OOM, but in that case
@@ -1146,7 +1147,7 @@ arena_reset(tsd_t *tsd, arena_t *arena) {
 		assert(alloc_ctx.szind != NSIZES);
 
 		if (config_stats || (config_prof && opt_prof)) {
-			usize = index2size(alloc_ctx.szind);
+			usize = sz_index2size(alloc_ctx.szind);
 			assert(usize == isalloc(tsd_tsdn(tsd), ptr));
 		}
 		/* Remove large allocation from prof sample set. */
@@ -1278,7 +1279,7 @@ arena_slab_alloc(tsdn_t *tsdn, arena_t *arena, szind_t binind,
 	    WITNESS_RANK_CORE, 0);
 
 	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
-	szind_t szind = size2index(bin_info->reg_size);
+	szind_t szind = sz_size2index(bin_info->reg_size);
 	bool zero = false;
 	bool commit = true;
 	extent_t *slab = extents_alloc(tsdn, arena, &extent_hooks,
@@ -1484,7 +1485,7 @@ arena_malloc_small(tsdn_t *tsdn, arena_t *arena, szind_t binind, bool zero) {
 
 	assert(binind < NBINS);
 	bin = &arena->bins[binind];
-	usize = index2size(binind);
+	usize = sz_index2size(binind);
 
 	malloc_mutex_lock(tsdn, &bin->lock);
 	if ((slab = bin->slabcur) != NULL && extent_nfree_get(slab) > 0) {
@@ -1544,7 +1545,7 @@ arena_malloc_hard(tsdn_t *tsdn, arena_t *arena, size_t size, szind_t ind,
 	if (likely(size <= SMALL_MAXCLASS)) {
 		return arena_malloc_small(tsdn, arena, ind, zero);
 	}
-	return large_malloc(tsdn, arena, index2size(ind), zero);
+	return large_malloc(tsdn, arena, sz_index2size(ind), zero);
 }
 
 void *
@@ -1555,8 +1556,8 @@ arena_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
 	if (usize <= SMALL_MAXCLASS && (alignment < PAGE || (alignment == PAGE
 	    && (usize & PAGE_MASK) == 0))) {
 		/* Small; alignment doesn't require special slab placement. */
-		ret = arena_malloc(tsdn, arena, usize, size2index(usize), zero,
-		    tcache, true);
+		ret = arena_malloc(tsdn, arena, usize, sz_size2index(usize),
+		    zero, tcache, true);
 	} else {
 		if (likely(alignment <= CACHELINE)) {
 			ret = large_malloc(tsdn, arena, usize, zero);
@@ -1581,7 +1582,7 @@ arena_prof_promote(tsdn_t *tsdn, const void *ptr, size_t usize) {
 	    (uintptr_t)ptr, true);
 	arena_t *arena = extent_arena_get(extent);
 
-	szind_t szind = size2index(usize);
+	szind_t szind = sz_size2index(usize);
 	extent_szind_set(extent, szind);
 	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
 	    szind, false);
@@ -1617,7 +1618,7 @@ arena_dalloc_promoted(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
 	size_t usize = arena_prof_demote(tsdn, extent, ptr);
 	if (usize <= tcache_maxclass) {
 		tcache_dalloc_large(tsdn_tsd(tsdn), tcache, ptr,
-		    size2index(usize), slow_path);
+		    sz_size2index(usize), slow_path);
 	} else {
 		large_dalloc(tsdn, extent);
 	}
@@ -1751,17 +1752,17 @@ arena_ralloc_no_move(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
 	}
 
 	extent_t *extent = iealloc(tsdn, ptr);
-	size_t usize_min = s2u(size);
-	size_t usize_max = s2u(size + extra);
+	size_t usize_min = sz_s2u(size);
+	size_t usize_max = sz_s2u(size + extra);
 	if (likely(oldsize <= SMALL_MAXCLASS && usize_min <= SMALL_MAXCLASS)) {
 		/*
 		 * Avoid moving the allocation if the size class can be left the
 		 * same.
 		 */
-		assert(arena_bin_info[size2index(oldsize)].reg_size ==
+		assert(arena_bin_info[sz_size2index(oldsize)].reg_size ==
 		    oldsize);
-		if ((usize_max > SMALL_MAXCLASS || size2index(usize_max) !=
-		    size2index(oldsize)) && (size > oldsize || usize_max <
+		if ((usize_max > SMALL_MAXCLASS || sz_size2index(usize_max) !=
+		    sz_size2index(oldsize)) && (size > oldsize || usize_max <
 		    oldsize)) {
 			return true;
 		}
@@ -1780,10 +1781,10 @@ static void *
 arena_ralloc_move_helper(tsdn_t *tsdn, arena_t *arena, size_t usize,
     size_t alignment, bool zero, tcache_t *tcache) {
 	if (alignment == 0) {
-		return arena_malloc(tsdn, arena, usize, size2index(usize),
+		return arena_malloc(tsdn, arena, usize, sz_size2index(usize),
 		    zero, tcache, true);
 	}
-	usize = sa2u(usize, alignment);
+	usize = sz_sa2u(usize, alignment);
 	if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 		return NULL;
 	}
@@ -1793,7 +1794,7 @@ arena_ralloc_move_helper(tsdn_t *tsdn, arena_t *arena, size_t usize,
 void *
 arena_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t oldsize,
     size_t size, size_t alignment, bool zero, tcache_t *tcache) {
-	size_t usize = s2u(size);
+	size_t usize = sz_s2u(size);
 	if (unlikely(usize == 0 || size > LARGE_MAXCLASS)) {
 		return NULL;
 	}
@@ -1998,7 +1999,7 @@ arena_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
 		goto label_error;
 	}
 
-	arena->extent_grow_next = psz2ind(HUGEPAGE);
+	arena->extent_grow_next = sz_psz2ind(HUGEPAGE);
 	if (malloc_mutex_init(&arena->extent_grow_mtx, "extent_grow",
 	    WITNESS_RANK_EXTENT_GROW, malloc_mutex_rank_exclusive)) {
 		goto label_error;
diff --git a/src/base.c b/src/base.c
index 498e5c9..8e1544f 100644
--- a/src/base.c
+++ b/src/base.c
@@ -5,6 +5,7 @@
 #include "jemalloc/internal/assert.h"
 #include "jemalloc/internal/extent_mmap.h"
 #include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sz.h"
 
 /******************************************************************************/
 /* Data. */
@@ -121,8 +122,8 @@ base_extent_bump_alloc_post(tsdn_t *tsdn, base_t *base, extent_t *extent,
 		 * Compute the index for the largest size class that does not
 		 * exceed extent's size.
 		 */
-		szind_t index_floor = size2index(extent_bsize_get(extent) + 1) -
-		    1;
+		szind_t index_floor =
+		    sz_size2index(extent_bsize_get(extent) + 1) - 1;
 		extent_heap_insert(&base->avail[index_floor], extent);
 	}
 
@@ -171,11 +172,11 @@ base_block_alloc(extent_hooks_t *extent_hooks, unsigned ind,
 	 * HUGEPAGE), or a size large enough to satisfy the requested size and
 	 * alignment, whichever is larger.
 	 */
-	size_t min_block_size = HUGEPAGE_CEILING(psz2u(header_size + gap_size +
-	    usize));
+	size_t min_block_size = HUGEPAGE_CEILING(sz_psz2u(header_size + gap_size
+	    + usize));
 	pszind_t pind_next = (*pind_last + 1 < NPSIZES) ? *pind_last + 1 :
 	    *pind_last;
-	size_t next_block_size = HUGEPAGE_CEILING(pind2sz(pind_next));
+	size_t next_block_size = HUGEPAGE_CEILING(sz_pind2sz(pind_next));
 	size_t block_size = (min_block_size > next_block_size) ? min_block_size
 	    : next_block_size;
 	base_block_t *block = (base_block_t *)base_map(extent_hooks, ind,
@@ -183,7 +184,7 @@ base_block_alloc(extent_hooks_t *extent_hooks, unsigned ind,
 	if (block == NULL) {
 		return NULL;
 	}
-	*pind_last = psz2ind(block_size);
+	*pind_last = sz_psz2ind(block_size);
 	block->size = block_size;
 	block->next = NULL;
 	assert(block_size >= header_size);
@@ -304,7 +305,7 @@ base_alloc_impl(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment,
 
 	extent_t *extent = NULL;
 	malloc_mutex_lock(tsdn, &base->mtx);
-	for (szind_t i = size2index(asize); i < NSIZES; i++) {
+	for (szind_t i = sz_size2index(asize); i < NSIZES; i++) {
 		extent = extent_heap_remove_first(&base->avail[i]);
 		if (extent != NULL) {
 			/* Use existing space. */
diff --git a/src/ckh.c b/src/ckh.c
index 013b624..e95e0a3 100644
--- a/src/ckh.c
+++ b/src/ckh.c
@@ -274,7 +274,7 @@ ckh_grow(tsd_t *tsd, ckh_t *ckh) {
 		size_t usize;
 
 		lg_curcells++;
-		usize = sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+		usize = sz_sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
 		if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 			ret = true;
 			goto label_return;
@@ -319,7 +319,7 @@ ckh_shrink(tsd_t *tsd, ckh_t *ckh) {
 	 */
 	lg_prevbuckets = ckh->lg_curbuckets;
 	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS - 1;
-	usize = sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+	usize = sz_sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
 	if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 		return;
 	}
@@ -395,7 +395,7 @@ ckh_new(tsd_t *tsd, ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
 	ckh->hash = hash;
 	ckh->keycomp = keycomp;
 
-	usize = sa2u(sizeof(ckhc_t) << lg_mincells, CACHELINE);
+	usize = sz_sa2u(sizeof(ckhc_t) << lg_mincells, CACHELINE);
 	if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 		ret = true;
 		goto label_return;
diff --git a/src/ctl.c b/src/ctl.c
index 33275d7..1520c50 100644
--- a/src/ctl.c
+++ b/src/ctl.c
@@ -701,7 +701,7 @@ ctl_arena_stats_amerge(tsdn_t *tsdn, ctl_arena_t *ctl_arena, arena_t *arena) {
 		for (i = 0; i < NBINS; i++) {
 			ctl_arena->astats->allocated_small +=
 			    ctl_arena->astats->bstats[i].curregs *
-			    index2size(i);
+			    sz_index2size(i);
 			ctl_arena->astats->nmalloc_small +=
 			    ctl_arena->astats->bstats[i].nmalloc;
 			ctl_arena->astats->ndalloc_small +=
@@ -2274,7 +2274,8 @@ arenas_bin_i_index(tsdn_t *tsdn, const size_t *mib, size_t miblen, size_t i) {
 }
 
 CTL_RO_NL_GEN(arenas_nlextents, NSIZES - NBINS, unsigned)
-CTL_RO_NL_GEN(arenas_lextent_i_size, index2size(NBINS+(szind_t)mib[2]), size_t)
+CTL_RO_NL_GEN(arenas_lextent_i_size, sz_index2size(NBINS+(szind_t)mib[2]),
+    size_t)
 static const ctl_named_node_t *
 arenas_lextent_i_index(tsdn_t *tsdn, const size_t *mib, size_t miblen,
     size_t i) {
diff --git a/src/extent.c b/src/extent.c
index c3d9baa..fb7a146 100644
--- a/src/extent.c
+++ b/src/extent.c
@@ -220,7 +220,7 @@ extent_size_quantize_floor(size_t size) {
 	assert(size > 0);
 	assert((size & PAGE_MASK) == 0);
 
-	pind = psz2ind(size - large_pad + 1);
+	pind = sz_psz2ind(size - sz_large_pad + 1);
 	if (pind == 0) {
 		/*
 		 * Avoid underflow.  This short-circuit would also do the right
@@ -230,7 +230,7 @@ extent_size_quantize_floor(size_t size) {
 		 */
 		return size;
 	}
-	ret = pind2sz(pind - 1) + large_pad;
+	ret = sz_pind2sz(pind - 1) + sz_large_pad;
 	assert(ret <= size);
 	return ret;
 }
@@ -243,7 +243,7 @@ extent_size_quantize_ceil(size_t size) {
 	size_t ret;
 
 	assert(size > 0);
-	assert(size - large_pad <= LARGE_MAXCLASS);
+	assert(size - sz_large_pad <= LARGE_MAXCLASS);
 	assert((size & PAGE_MASK) == 0);
 
 	ret = extent_size_quantize_floor(size);
@@ -256,7 +256,8 @@ extent_size_quantize_ceil(size_t size) {
 		 * search would potentially find sufficiently aligned available
 		 * memory somewhere lower.
 		 */
-		ret = pind2sz(psz2ind(ret - large_pad + 1)) + large_pad;
+		ret = sz_pind2sz(sz_psz2ind(ret - sz_large_pad + 1)) +
+		    sz_large_pad;
 	}
 	return ret;
 }
@@ -300,7 +301,7 @@ extents_insert_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent,
 
 	size_t size = extent_size_get(extent);
 	size_t psz = extent_size_quantize_floor(size);
-	pszind_t pind = psz2ind(psz);
+	pszind_t pind = sz_psz2ind(psz);
 	if (extent_heap_empty(&extents->heaps[pind])) {
 		bitmap_unset(extents->bitmap, &extents_bitmap_info,
 		    (size_t)pind);
@@ -329,7 +330,7 @@ extents_remove_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent,
 
 	size_t size = extent_size_get(extent);
 	size_t psz = extent_size_quantize_floor(size);
-	pszind_t pind = psz2ind(psz);
+	pszind_t pind = sz_psz2ind(psz);
 	extent_heap_remove(&extents->heaps[pind], extent);
 	if (extent_heap_empty(&extents->heaps[pind])) {
 		bitmap_set(extents->bitmap, &extents_bitmap_info,
@@ -354,7 +355,7 @@ extents_remove_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent,
 static extent_t *
 extents_best_fit_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
     size_t size) {
-	pszind_t pind = psz2ind(extent_size_quantize_ceil(size));
+	pszind_t pind = sz_psz2ind(extent_size_quantize_ceil(size));
 	pszind_t i = (pszind_t)bitmap_ffu(extents->bitmap, &extents_bitmap_info,
 	    (size_t)pind);
 	if (i < NPSIZES+1) {
@@ -376,7 +377,7 @@ extents_first_fit_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
     size_t size) {
 	extent_t *ret = NULL;
 
-	pszind_t pind = psz2ind(extent_size_quantize_ceil(size));
+	pszind_t pind = sz_psz2ind(extent_size_quantize_ceil(size));
 	for (pszind_t i = (pszind_t)bitmap_ffu(extents->bitmap,
 	    &extents_bitmap_info, (size_t)pind); i < NPSIZES+1; i =
 	    (pszind_t)bitmap_ffu(extents->bitmap, &extents_bitmap_info,
@@ -1040,7 +1041,7 @@ extent_grow_retained(tsdn_t *tsdn, arena_t *arena,
 	 * satisfy this request.
 	 */
 	pszind_t egn_skip = 0;
-	size_t alloc_size = pind2sz(arena->extent_grow_next + egn_skip);
+	size_t alloc_size = sz_pind2sz(arena->extent_grow_next + egn_skip);
 	while (alloc_size < alloc_size_min) {
 		egn_skip++;
 		if (arena->extent_grow_next + egn_skip == NPSIZES) {
@@ -1048,7 +1049,7 @@ extent_grow_retained(tsdn_t *tsdn, arena_t *arena,
 			goto label_err;
 		}
 		assert(arena->extent_grow_next + egn_skip < NPSIZES);
-		alloc_size = pind2sz(arena->extent_grow_next + egn_skip);
+		alloc_size = sz_pind2sz(arena->extent_grow_next + egn_skip);
 	}
 
 	extent_t *extent = extent_alloc(tsdn, arena);
@@ -1369,7 +1370,7 @@ extent_try_coalesce(tsdn_t *tsdn, arena_t *arena,
 			extent_unlock(tsdn, prev);
 
 			if (can_coalesce && !extent_coalesce(tsdn, arena,
-			    r_extent_hooks, extents, extent, prev, false, 
+			    r_extent_hooks, extents, extent, prev, false,
 			    growing_retained)) {
 				extent = prev;
 				if (extents->delay_coalesce) {
diff --git a/src/jemalloc.c b/src/jemalloc.c
index 1befb64..268d19c 100644
--- a/src/jemalloc.c
+++ b/src/jemalloc.c
@@ -13,6 +13,7 @@
 #include "jemalloc/internal/rtree.h"
 #include "jemalloc/internal/size_classes.h"
 #include "jemalloc/internal/spin.h"
+#include "jemalloc/internal/sz.h"
 #include "jemalloc/internal/ticker.h"
 #include "jemalloc/internal/util.h"
 
@@ -107,110 +108,6 @@ enum {
 };
 static uint8_t	malloc_slow_flags;
 
-JEMALLOC_ALIGNED(CACHELINE)
-const size_t	pind2sz_tab[NPSIZES+1] = {
-#define PSZ_yes(lg_grp, ndelta, lg_delta)				\
-	(((ZU(1)<<lg_grp) + (ZU(ndelta)<<lg_delta))),
-#define PSZ_no(lg_grp, ndelta, lg_delta)
-#define SC(index, lg_grp, lg_delta, ndelta, psz, bin, pgs, lg_delta_lookup) \
-	PSZ_##psz(lg_grp, ndelta, lg_delta)
-	SIZE_CLASSES
-#undef PSZ_yes
-#undef PSZ_no
-#undef SC
-	(LARGE_MAXCLASS + PAGE)
-};
-
-JEMALLOC_ALIGNED(CACHELINE)
-const size_t	index2size_tab[NSIZES] = {
-#define SC(index, lg_grp, lg_delta, ndelta, psz, bin, pgs, lg_delta_lookup) \
-	((ZU(1)<<lg_grp) + (ZU(ndelta)<<lg_delta)),
-	SIZE_CLASSES
-#undef SC
-};
-
-JEMALLOC_ALIGNED(CACHELINE)
-const uint8_t	size2index_tab[] = {
-#if LG_TINY_MIN == 0
-#warning "Dangerous LG_TINY_MIN"
-#define S2B_0(i)	i,
-#elif LG_TINY_MIN == 1
-#warning "Dangerous LG_TINY_MIN"
-#define S2B_1(i)	i,
-#elif LG_TINY_MIN == 2
-#warning "Dangerous LG_TINY_MIN"
-#define S2B_2(i)	i,
-#elif LG_TINY_MIN == 3
-#define S2B_3(i)	i,
-#elif LG_TINY_MIN == 4
-#define S2B_4(i)	i,
-#elif LG_TINY_MIN == 5
-#define S2B_5(i)	i,
-#elif LG_TINY_MIN == 6
-#define S2B_6(i)	i,
-#elif LG_TINY_MIN == 7
-#define S2B_7(i)	i,
-#elif LG_TINY_MIN == 8
-#define S2B_8(i)	i,
-#elif LG_TINY_MIN == 9
-#define S2B_9(i)	i,
-#elif LG_TINY_MIN == 10
-#define S2B_10(i)	i,
-#elif LG_TINY_MIN == 11
-#define S2B_11(i)	i,
-#else
-#error "Unsupported LG_TINY_MIN"
-#endif
-#if LG_TINY_MIN < 1
-#define S2B_1(i)	S2B_0(i) S2B_0(i)
-#endif
-#if LG_TINY_MIN < 2
-#define S2B_2(i)	S2B_1(i) S2B_1(i)
-#endif
-#if LG_TINY_MIN < 3
-#define S2B_3(i)	S2B_2(i) S2B_2(i)
-#endif
-#if LG_TINY_MIN < 4
-#define S2B_4(i)	S2B_3(i) S2B_3(i)
-#endif
-#if LG_TINY_MIN < 5
-#define S2B_5(i)	S2B_4(i) S2B_4(i)
-#endif
-#if LG_TINY_MIN < 6
-#define S2B_6(i)	S2B_5(i) S2B_5(i)
-#endif
-#if LG_TINY_MIN < 7
-#define S2B_7(i)	S2B_6(i) S2B_6(i)
-#endif
-#if LG_TINY_MIN < 8
-#define S2B_8(i)	S2B_7(i) S2B_7(i)
-#endif
-#if LG_TINY_MIN < 9
-#define S2B_9(i)	S2B_8(i) S2B_8(i)
-#endif
-#if LG_TINY_MIN < 10
-#define S2B_10(i)	S2B_9(i) S2B_9(i)
-#endif
-#if LG_TINY_MIN < 11
-#define S2B_11(i)	S2B_10(i) S2B_10(i)
-#endif
-#define S2B_no(i)
-#define SC(index, lg_grp, lg_delta, ndelta, psz, bin, pgs, lg_delta_lookup) \
-	S2B_##lg_delta_lookup(index)
-	SIZE_CLASSES
-#undef S2B_3
-#undef S2B_4
-#undef S2B_5
-#undef S2B_6
-#undef S2B_7
-#undef S2B_8
-#undef S2B_9
-#undef S2B_10
-#undef S2B_11
-#undef S2B_no
-#undef SC
-};
-
 #ifdef JEMALLOC_THREADED_INIT
 /* Used to let the initializing thread recursively allocate. */
 #  define NO_INITIALIZER	((unsigned long)0)
@@ -333,7 +230,7 @@ a0ialloc(size_t size, bool zero, bool is_internal) {
 		return NULL;
 	}
 
-	return iallocztm(TSDN_NULL, size, size2index(size), zero, NULL,
+	return iallocztm(TSDN_NULL, size, sz_size2index(size), zero, NULL,
 	    is_internal, arena_get(TSDN_NULL, 0, true), true);
 }
 
@@ -1687,10 +1584,11 @@ imalloc_sample(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd,
 	size_t bumped_usize = usize;
 
 	if (usize <= SMALL_MAXCLASS) {
-		assert(((dopts->alignment == 0) ? s2u(LARGE_MINCLASS) :
-		    sa2u(LARGE_MINCLASS, dopts->alignment)) == LARGE_MINCLASS);
-		ind_large = size2index(LARGE_MINCLASS);
-		bumped_usize = s2u(LARGE_MINCLASS);
+		assert(((dopts->alignment == 0) ? sz_s2u(LARGE_MINCLASS) :
+		    sz_sa2u(LARGE_MINCLASS, dopts->alignment))
+		    == LARGE_MINCLASS);
+		ind_large = sz_size2index(LARGE_MINCLASS);
+		bumped_usize = sz_s2u(LARGE_MINCLASS);
 		ret = imalloc_no_sample(sopts, dopts, tsd, bumped_usize,
 		    bumped_usize, ind_large);
 		if (unlikely(ret == NULL)) {
@@ -1792,16 +1690,16 @@ imalloc_body(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd) {
 	/* This is the beginning of the "core" algorithm. */
 
 	if (dopts->alignment == 0) {
-		ind = size2index(size);
+		ind = sz_size2index(size);
 		if (unlikely(ind >= NSIZES)) {
 			goto label_oom;
 		}
 		if (config_stats || (config_prof && opt_prof)) {
-			usize = index2size(ind);
+			usize = sz_index2size(ind);
 			assert(usize > 0 && usize <= LARGE_MAXCLASS);
 		}
 	} else {
-		usize = sa2u(size, dopts->alignment);
+		usize = sz_sa2u(size, dopts->alignment);
 		if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 			goto label_oom;
 		}
@@ -2155,10 +2053,10 @@ ifree(tsd_t *tsd, void *ptr, tcache_t *tcache, bool slow_path) {
 
 	size_t usize;
 	if (config_prof && opt_prof) {
-		usize = index2size(alloc_ctx.szind);
+		usize = sz_index2size(alloc_ctx.szind);
 		prof_free(tsd, ptr, usize, &alloc_ctx);
 	} else if (config_stats) {
-		usize = index2size(alloc_ctx.szind);
+		usize = sz_index2size(alloc_ctx.szind);
 	}
 	if (config_stats) {
 		*tsd_thread_deallocatedp_get(tsd) += usize;
@@ -2192,7 +2090,7 @@ isfree(tsd_t *tsd, void *ptr, size_t usize, tcache_t *tcache, bool slow_path) {
 		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
 		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
 		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
-		assert(alloc_ctx.szind == size2index(usize));
+		assert(alloc_ctx.szind == sz_size2index(usize));
 		ctx = &alloc_ctx;
 		prof_free(tsd, ptr, usize, ctx);
 	} else {
@@ -2247,16 +2145,16 @@ je_realloc(void *ptr, size_t size) {
 		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
 		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
 		assert(alloc_ctx.szind != NSIZES);
-		old_usize = index2size(alloc_ctx.szind);
+		old_usize = sz_index2size(alloc_ctx.szind);
 		assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
 		if (config_prof && opt_prof) {
-			usize = s2u(size);
+			usize = sz_s2u(size);
 			ret = unlikely(usize == 0 || usize > LARGE_MAXCLASS) ?
 			    NULL : irealloc_prof(tsd, ptr, old_usize, usize,
 			    &alloc_ctx);
 		} else {
 			if (config_stats) {
-				usize = s2u(size);
+				usize = sz_s2u(size);
 			}
 			ret = iralloc(tsd, ptr, old_usize, size, 0, false);
 		}
@@ -2601,10 +2499,11 @@ je_rallocx(void *ptr, size_t size, int flags) {
 	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
 	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
 	assert(alloc_ctx.szind != NSIZES);
-	old_usize = index2size(alloc_ctx.szind);
+	old_usize = sz_index2size(alloc_ctx.szind);
 	assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
 	if (config_prof && opt_prof) {
-		usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment);
+		usize = (alignment == 0) ?
+		    sz_s2u(size) : sz_sa2u(size, alignment);
 		if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
 			goto label_oom;
 		}
@@ -2685,10 +2584,10 @@ ixallocx_prof(tsd_t *tsd, void *ptr, size_t old_usize, size_t size,
 	 * prof_realloc() will use the actual usize to decide whether to sample.
 	 */
 	if (alignment == 0) {
-		usize_max = s2u(size+extra);
+		usize_max = sz_s2u(size+extra);
 		assert(usize_max > 0 && usize_max <= LARGE_MAXCLASS);
 	} else {
-		usize_max = sa2u(size+extra, alignment);
+		usize_max = sz_sa2u(size+extra, alignment);
 		if (unlikely(usize_max == 0 || usize_max > LARGE_MAXCLASS)) {
 			/*
 			 * usize_max is out of range, and chances are that
@@ -2737,7 +2636,7 @@ je_xallocx(void *ptr, size_t size, size_t extra, int flags) {
 	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
 	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
 	assert(alloc_ctx.szind != NSIZES);
-	old_usize = index2size(alloc_ctx.szind);
+	old_usize = sz_index2size(alloc_ctx.szind);
 	assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
 	/*
 	 * The API explicitly absolves itself of protecting against (size +
@@ -2847,9 +2746,9 @@ inallocx(tsdn_t *tsdn, size_t size, int flags) {
 
 	size_t usize;
 	if (likely((flags & MALLOCX_LG_ALIGN_MASK) == 0)) {
-		usize = s2u(size);
+		usize = sz_s2u(size);
 	} else {
-		usize = sa2u(size, MALLOCX_ALIGN_GET_SPECIFIED(flags));
+		usize = sz_sa2u(size, MALLOCX_ALIGN_GET_SPECIFIED(flags));
 	}
 	witness_assert_lockless(tsdn_witness_tsdp_get(tsdn));
 	return usize;
diff --git a/src/large.c b/src/large.c
index 79d2c9d..27a2c67 100644
--- a/src/large.c
+++ b/src/large.c
@@ -12,7 +12,7 @@
 
 void *
 large_malloc(tsdn_t *tsdn, arena_t *arena, size_t usize, bool zero) {
-	assert(usize == s2u(usize));
+	assert(usize == sz_s2u(usize));
 
 	return large_palloc(tsdn, arena, usize, CACHELINE, zero);
 }
@@ -27,7 +27,7 @@ large_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
 
 	assert(!tsdn_null(tsdn) || arena != NULL);
 
-	ausize = sa2u(usize, alignment);
+	ausize = sz_sa2u(usize, alignment);
 	if (unlikely(ausize == 0 || ausize > LARGE_MAXCLASS)) {
 		return NULL;
 	}
@@ -97,7 +97,7 @@ large_ralloc_no_move_shrink(tsdn_t *tsdn, extent_t *extent, size_t usize) {
 	arena_t *arena = extent_arena_get(extent);
 	size_t oldusize = extent_usize_get(extent);
 	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
-	size_t diff = extent_size_get(extent) - (usize + large_pad);
+	size_t diff = extent_size_get(extent) - (usize + sz_large_pad);
 
 	assert(oldusize > usize);
 
@@ -108,8 +108,8 @@ large_ralloc_no_move_shrink(tsdn_t *tsdn, extent_t *extent, size_t usize) {
 	/* Split excess pages. */
 	if (diff != 0) {
 		extent_t *trail = extent_split_wrapper(tsdn, arena,
-		    &extent_hooks, extent, usize + large_pad, size2index(usize),
-		    false, diff, NSIZES, false);
+		    &extent_hooks, extent, usize + sz_large_pad,
+		    sz_size2index(usize), false, diff, NSIZES, false);
 		if (trail == NULL) {
 			return true;
 		}
@@ -178,7 +178,7 @@ large_ralloc_no_move_expand(tsdn_t *tsdn, extent_t *extent, size_t usize,
 	}
 	rtree_ctx_t rtree_ctx_fallback;
 	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
-	szind_t szind = size2index(usize);
+	szind_t szind = sz_size2index(usize);
 	extent_szind_set(extent, szind);
 	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx,
 	    (uintptr_t)extent_addr_get(extent), szind, false);
diff --git a/src/prof.c b/src/prof.c
index aa67486..61dfa2c 100644
--- a/src/prof.c
+++ b/src/prof.c
@@ -556,7 +556,7 @@ prof_gctx_create(tsdn_t *tsdn, prof_bt_t *bt) {
 	 */
 	size_t size = offsetof(prof_gctx_t, vec) + (bt->len * sizeof(void *));
 	prof_gctx_t *gctx = (prof_gctx_t *)iallocztm(tsdn, size,
-	    size2index(size), false, NULL, true, arena_get(TSDN_NULL, 0, true),
+	    sz_size2index(size), false, NULL, true, arena_get(TSDN_NULL, 0, true),
 	    true);
 	if (gctx == NULL) {
 		return NULL;
@@ -819,7 +819,7 @@ prof_lookup(tsd_t *tsd, prof_bt_t *bt) {
 
 		/* Link a prof_tctx_t into gctx for this thread. */
 		ret.v = iallocztm(tsd_tsdn(tsd), sizeof(prof_tctx_t),
-		    size2index(sizeof(prof_tctx_t)), false, NULL, true,
+		    sz_size2index(sizeof(prof_tctx_t)), false, NULL, true,
 		    arena_ichoose(tsd, NULL), true);
 		if (ret.p == NULL) {
 			if (new_gctx) {
@@ -1899,7 +1899,7 @@ prof_tdata_init_impl(tsd_t *tsd, uint64_t thr_uid, uint64_t thr_discrim,
 
 	/* Initialize an empty cache for this thread. */
 	tdata = (prof_tdata_t *)iallocztm(tsd_tsdn(tsd), sizeof(prof_tdata_t),
-	    size2index(sizeof(prof_tdata_t)), false, NULL, true,
+	    sz_size2index(sizeof(prof_tdata_t)), false, NULL, true,
 	    arena_get(TSDN_NULL, 0, true), true);
 	if (tdata == NULL) {
 		return NULL;
@@ -2135,7 +2135,7 @@ prof_thread_name_alloc(tsdn_t *tsdn, const char *thread_name) {
 		return "";
 	}
 
-	ret = iallocztm(tsdn, size, size2index(size), false, NULL, true,
+	ret = iallocztm(tsdn, size, sz_size2index(size), false, NULL, true,
 	    arena_get(TSDN_NULL, 0, true), true);
 	if (ret == NULL) {
 		return NULL;
diff --git a/src/sz.c b/src/sz.c
new file mode 100644
index 0000000..0986615
--- /dev/null
+++ b/src/sz.c
@@ -0,0 +1,106 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/sz.h"
+
+JEMALLOC_ALIGNED(CACHELINE)
+const size_t sz_pind2sz_tab[NPSIZES+1] = {
+#define PSZ_yes(lg_grp, ndelta, lg_delta)				\
+	(((ZU(1)<<lg_grp) + (ZU(ndelta)<<lg_delta))),
+#define PSZ_no(lg_grp, ndelta, lg_delta)
+#define SC(index, lg_grp, lg_delta, ndelta, psz, bin, pgs, lg_delta_lookup) \
+	PSZ_##psz(lg_grp, ndelta, lg_delta)
+	SIZE_CLASSES
+#undef PSZ_yes
+#undef PSZ_no
+#undef SC
+	(LARGE_MAXCLASS + PAGE)
+};
+
+JEMALLOC_ALIGNED(CACHELINE)
+const size_t sz_index2size_tab[NSIZES] = {
+#define SC(index, lg_grp, lg_delta, ndelta, psz, bin, pgs, lg_delta_lookup) \
+	((ZU(1)<<lg_grp) + (ZU(ndelta)<<lg_delta)),
+	SIZE_CLASSES
+#undef SC
+};
+
+JEMALLOC_ALIGNED(CACHELINE)
+const uint8_t sz_size2index_tab[] = {
+#if LG_TINY_MIN == 0
+#warning "Dangerous LG_TINY_MIN"
+#define S2B_0(i)	i,
+#elif LG_TINY_MIN == 1
+#warning "Dangerous LG_TINY_MIN"
+#define S2B_1(i)	i,
+#elif LG_TINY_MIN == 2
+#warning "Dangerous LG_TINY_MIN"
+#define S2B_2(i)	i,
+#elif LG_TINY_MIN == 3
+#define S2B_3(i)	i,
+#elif LG_TINY_MIN == 4
+#define S2B_4(i)	i,
+#elif LG_TINY_MIN == 5
+#define S2B_5(i)	i,
+#elif LG_TINY_MIN == 6
+#define S2B_6(i)	i,
+#elif LG_TINY_MIN == 7
+#define S2B_7(i)	i,
+#elif LG_TINY_MIN == 8
+#define S2B_8(i)	i,
+#elif LG_TINY_MIN == 9
+#define S2B_9(i)	i,
+#elif LG_TINY_MIN == 10
+#define S2B_10(i)	i,
+#elif LG_TINY_MIN == 11
+#define S2B_11(i)	i,
+#else
+#error "Unsupported LG_TINY_MIN"
+#endif
+#if LG_TINY_MIN < 1
+#define S2B_1(i)	S2B_0(i) S2B_0(i)
+#endif
+#if LG_TINY_MIN < 2
+#define S2B_2(i)	S2B_1(i) S2B_1(i)
+#endif
+#if LG_TINY_MIN < 3
+#define S2B_3(i)	S2B_2(i) S2B_2(i)
+#endif
+#if LG_TINY_MIN < 4
+#define S2B_4(i)	S2B_3(i) S2B_3(i)
+#endif
+#if LG_TINY_MIN < 5
+#define S2B_5(i)	S2B_4(i) S2B_4(i)
+#endif
+#if LG_TINY_MIN < 6
+#define S2B_6(i)	S2B_5(i) S2B_5(i)
+#endif
+#if LG_TINY_MIN < 7
+#define S2B_7(i)	S2B_6(i) S2B_6(i)
+#endif
+#if LG_TINY_MIN < 8
+#define S2B_8(i)	S2B_7(i) S2B_7(i)
+#endif
+#if LG_TINY_MIN < 9
+#define S2B_9(i)	S2B_8(i) S2B_8(i)
+#endif
+#if LG_TINY_MIN < 10
+#define S2B_10(i)	S2B_9(i) S2B_9(i)
+#endif
+#if LG_TINY_MIN < 11
+#define S2B_11(i)	S2B_10(i) S2B_10(i)
+#endif
+#define S2B_no(i)
+#define SC(index, lg_grp, lg_delta, ndelta, psz, bin, pgs, lg_delta_lookup) \
+	S2B_##lg_delta_lookup(index)
+	SIZE_CLASSES
+#undef S2B_3
+#undef S2B_4
+#undef S2B_5
+#undef S2B_6
+#undef S2B_7
+#undef S2B_8
+#undef S2B_9
+#undef S2B_10
+#undef S2B_11
+#undef S2B_no
+#undef SC
+};
diff --git a/src/tcache.c b/src/tcache.c
index 96ebe67..6355805 100644
--- a/src/tcache.c
+++ b/src/tcache.c
@@ -383,7 +383,7 @@ tsd_tcache_data_init(tsd_t *tsd) {
 	assert(tcache_small_bin_get(tcache, 0)->avail == NULL);
 	size_t size = stack_nelms * sizeof(void *);
 	/* Avoid false cacheline sharing. */
-	size = sa2u(size, CACHELINE);
+	size = sz_sa2u(size, CACHELINE);
 
 	void *avail_array = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true,
 	    NULL, true, arena_get(TSDN_NULL, 0, true));
@@ -430,7 +430,7 @@ tcache_create_explicit(tsd_t *tsd) {
 	stack_offset = size;
 	size += stack_nelms * sizeof(void *);
 	/* Avoid false cacheline sharing. */
-	size = sa2u(size, CACHELINE);
+	size = sz_sa2u(size, CACHELINE);
 
 	tcache = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true, NULL, true,
 	    arena_get(TSDN_NULL, 0, true));
@@ -655,7 +655,7 @@ tcache_boot(tsdn_t *tsdn) {
 		return true;
 	}
 
-	nhbins = size2index(tcache_maxclass) + 1;
+	nhbins = sz_size2index(tcache_maxclass) + 1;
 
 	/* Initialize tcache_bin_info. */
 	tcache_bin_info = (tcache_bin_info_t *)base_alloc(tsdn, b0get(), nhbins
diff --git a/src/zone.c b/src/zone.c
index 37bc8da..9d3b7b4 100644
--- a/src/zone.c
+++ b/src/zone.c
@@ -244,7 +244,7 @@ zone_good_size(malloc_zone_t *zone, size_t size) {
 	if (size == 0) {
 		size = 1;
 	}
-	return s2u(size);
+	return sz_s2u(size);
 }
 
 static kern_return_t
diff --git a/test/unit/arena_reset.c b/test/unit/arena_reset.c
index 958453d..6409a92 100644
--- a/test/unit/arena_reset.c
+++ b/test/unit/arena_reset.c
@@ -81,7 +81,7 @@ vsalloc(tsdn_t *tsdn, const void *ptr) {
 		return 0;
 	}
 
-	return index2size(szind);
+	return sz_index2size(szind);
 }
 
 static unsigned
diff --git a/test/unit/retained.c b/test/unit/retained.c
index 883bf4a..d51a598 100644
--- a/test/unit/retained.c
+++ b/test/unit/retained.c
@@ -1,5 +1,7 @@
 #include "test/jemalloc_test.h"
 
+#include "jemalloc/internal/spin.h"
+
 static unsigned		arena_ind;
 static size_t		sz;
 static size_t		esz;
@@ -100,7 +102,7 @@ TEST_BEGIN(test_retained) {
 
 	arena_ind = do_arena_create(NULL);
 	sz = nallocx(HUGEPAGE, 0);
-	esz = sz + large_pad;
+	esz = sz + sz_large_pad;
 
 	atomic_store_u(&epoch, 0, ATOMIC_RELAXED);
 
@@ -136,9 +138,9 @@ TEST_BEGIN(test_retained) {
 		arena_t *arena = arena_get(tsdn_fetch(), arena_ind, false);
 		size_t usable = 0;
 		size_t fragmented = 0;
-		for (pszind_t pind = psz2ind(HUGEPAGE); pind <
+		for (pszind_t pind = sz_psz2ind(HUGEPAGE); pind <
 		    arena->extent_grow_next; pind++) {
-			size_t psz = pind2sz(pind);
+			size_t psz = sz_pind2sz(pind);
 			size_t psz_fragmented = psz % esz;
 			size_t psz_usable = psz - psz_fragmented;
 			/*
diff --git a/test/unit/rtree.c b/test/unit/rtree.c
index 752dde9..814837b 100644
--- a/test/unit/rtree.c
+++ b/test/unit/rtree.c
@@ -86,7 +86,8 @@ TEST_END
 TEST_BEGIN(test_rtree_extrema) {
 	extent_t extent_a, extent_b;
 	extent_init(&extent_a, NULL, NULL, LARGE_MINCLASS, false,
-	    size2index(LARGE_MINCLASS), 0, extent_state_active, false, false);
+	    sz_size2index(LARGE_MINCLASS), 0, extent_state_active, false,
+	    false);
 	extent_init(&extent_b, NULL, NULL, 0, false, NSIZES, 0,
 	    extent_state_active, false, false);
 
diff --git a/test/unit/size_classes.c b/test/unit/size_classes.c
index 70a86ad..bcff560 100644
--- a/test/unit/size_classes.c
+++ b/test/unit/size_classes.c
@@ -27,11 +27,11 @@ TEST_BEGIN(test_size_classes) {
 	szind_t index, max_index;
 
 	max_size_class = get_max_size_class();
-	max_index = size2index(max_size_class);
+	max_index = sz_size2index(max_size_class);
 
-	for (index = 0, size_class = index2size(index); index < max_index ||
+	for (index = 0, size_class = sz_index2size(index); index < max_index ||
 	    size_class < max_size_class; index++, size_class =
-	    index2size(index)) {
+	    sz_index2size(index)) {
 		assert_true(index < max_index,
 		    "Loop conditionals should be equivalent; index=%u, "
 		    "size_class=%zu (%#zx)", index, size_class, size_class);
@@ -39,42 +39,44 @@ TEST_BEGIN(test_size_classes) {
 		    "Loop conditionals should be equivalent; index=%u, "
 		    "size_class=%zu (%#zx)", index, size_class, size_class);
 
-		assert_u_eq(index, size2index(size_class),
-		    "size2index() does not reverse index2size(): index=%u -->"
-		    " size_class=%zu --> index=%u --> size_class=%zu", index,
-		    size_class, size2index(size_class),
-		    index2size(size2index(size_class)));
-		assert_zu_eq(size_class, index2size(size2index(size_class)),
-		    "index2size() does not reverse size2index(): index=%u -->"
-		    " size_class=%zu --> index=%u --> size_class=%zu", index,
-		    size_class, size2index(size_class),
-		    index2size(size2index(size_class)));
-
-		assert_u_eq(index+1, size2index(size_class+1),
+		assert_u_eq(index, sz_size2index(size_class),
+		    "sz_size2index() does not reverse sz_index2size(): index=%u"
+		    " --> size_class=%zu --> index=%u --> size_class=%zu",
+		    index, size_class, sz_size2index(size_class),
+		    sz_index2size(sz_size2index(size_class)));
+		assert_zu_eq(size_class,
+		    sz_index2size(sz_size2index(size_class)),
+		    "sz_index2size() does not reverse sz_size2index(): index=%u"
+		    " --> size_class=%zu --> index=%u --> size_class=%zu",
+		    index, size_class, sz_size2index(size_class),
+		    sz_index2size(sz_size2index(size_class)));
+
+		assert_u_eq(index+1, sz_size2index(size_class+1),
 		    "Next size_class does not round up properly");
 
 		assert_zu_eq(size_class, (index > 0) ?
-		    s2u(index2size(index-1)+1) : s2u(1),
-		    "s2u() does not round up to size class");
-		assert_zu_eq(size_class, s2u(size_class-1),
-		    "s2u() does not round up to size class");
-		assert_zu_eq(size_class, s2u(size_class),
-		    "s2u() does not compute same size class");
-		assert_zu_eq(s2u(size_class+1), index2size(index+1),
-		    "s2u() does not round up to next size class");
+		    sz_s2u(sz_index2size(index-1)+1) : sz_s2u(1),
+		    "sz_s2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_s2u(size_class-1),
+		    "sz_s2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_s2u(size_class),
+		    "sz_s2u() does not compute same size class");
+		assert_zu_eq(sz_s2u(size_class+1), sz_index2size(index+1),
+		    "sz_s2u() does not round up to next size class");
 	}
 
-	assert_u_eq(index, size2index(index2size(index)),
-	    "size2index() does not reverse index2size()");
-	assert_zu_eq(max_size_class, index2size(size2index(max_size_class)),
-	    "index2size() does not reverse size2index()");
-
-	assert_zu_eq(size_class, s2u(index2size(index-1)+1),
-	    "s2u() does not round up to size class");
-	assert_zu_eq(size_class, s2u(size_class-1),
-	    "s2u() does not round up to size class");
-	assert_zu_eq(size_class, s2u(size_class),
-	    "s2u() does not compute same size class");
+	assert_u_eq(index, sz_size2index(sz_index2size(index)),
+	    "sz_size2index() does not reverse sz_index2size()");
+	assert_zu_eq(max_size_class, sz_index2size(
+	    sz_size2index(max_size_class)),
+	    "sz_index2size() does not reverse sz_size2index()");
+
+	assert_zu_eq(size_class, sz_s2u(sz_index2size(index-1)+1),
+	    "sz_s2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_s2u(size_class-1),
+	    "sz_s2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_s2u(size_class),
+	    "sz_s2u() does not compute same size class");
 }
 TEST_END
 
@@ -83,10 +85,11 @@ TEST_BEGIN(test_psize_classes) {
 	pszind_t pind, max_pind;
 
 	max_psz = get_max_size_class() + PAGE;
-	max_pind = psz2ind(max_psz);
+	max_pind = sz_psz2ind(max_psz);
 
-	for (pind = 0, size_class = pind2sz(pind); pind < max_pind || size_class
-	    < max_psz; pind++, size_class = pind2sz(pind)) {
+	for (pind = 0, size_class = sz_pind2sz(pind);
+	    pind < max_pind || size_class < max_psz;
+	    pind++, size_class = sz_pind2sz(pind)) {
 		assert_true(pind < max_pind,
 		    "Loop conditionals should be equivalent; pind=%u, "
 		    "size_class=%zu (%#zx)", pind, size_class, size_class);
@@ -94,42 +97,42 @@ TEST_BEGIN(test_psize_classes) {
 		    "Loop conditionals should be equivalent; pind=%u, "
 		    "size_class=%zu (%#zx)", pind, size_class, size_class);
 
-		assert_u_eq(pind, psz2ind(size_class),
-		    "psz2ind() does not reverse pind2sz(): pind=%u -->"
+		assert_u_eq(pind, sz_psz2ind(size_class),
+		    "sz_psz2ind() does not reverse sz_pind2sz(): pind=%u -->"
 		    " size_class=%zu --> pind=%u --> size_class=%zu", pind,
-		    size_class, psz2ind(size_class),
-		    pind2sz(psz2ind(size_class)));
-		assert_zu_eq(size_class, pind2sz(psz2ind(size_class)),
-		    "pind2sz() does not reverse psz2ind(): pind=%u -->"
+		    size_class, sz_psz2ind(size_class),
+		    sz_pind2sz(sz_psz2ind(size_class)));
+		assert_zu_eq(size_class, sz_pind2sz(sz_psz2ind(size_class)),
+		    "sz_pind2sz() does not reverse sz_psz2ind(): pind=%u -->"
 		    " size_class=%zu --> pind=%u --> size_class=%zu", pind,
-		    size_class, psz2ind(size_class),
-		    pind2sz(psz2ind(size_class)));
+		    size_class, sz_psz2ind(size_class),
+		    sz_pind2sz(sz_psz2ind(size_class)));
 
-		assert_u_eq(pind+1, psz2ind(size_class+1),
+		assert_u_eq(pind+1, sz_psz2ind(size_class+1),
 		    "Next size_class does not round up properly");
 
 		assert_zu_eq(size_class, (pind > 0) ?
-		    psz2u(pind2sz(pind-1)+1) : psz2u(1),
-		    "psz2u() does not round up to size class");
-		assert_zu_eq(size_class, psz2u(size_class-1),
-		    "psz2u() does not round up to size class");
-		assert_zu_eq(size_class, psz2u(size_class),
-		    "psz2u() does not compute same size class");
-		assert_zu_eq(psz2u(size_class+1), pind2sz(pind+1),
-		    "psz2u() does not round up to next size class");
+		    sz_psz2u(sz_pind2sz(pind-1)+1) : sz_psz2u(1),
+		    "sz_psz2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_psz2u(size_class-1),
+		    "sz_psz2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_psz2u(size_class),
+		    "sz_psz2u() does not compute same size class");
+		assert_zu_eq(sz_psz2u(size_class+1), sz_pind2sz(pind+1),
+		    "sz_psz2u() does not round up to next size class");
 	}
 
-	assert_u_eq(pind, psz2ind(pind2sz(pind)),
-	    "psz2ind() does not reverse pind2sz()");
-	assert_zu_eq(max_psz, pind2sz(psz2ind(max_psz)),
-	    "pind2sz() does not reverse psz2ind()");
-
-	assert_zu_eq(size_class, psz2u(pind2sz(pind-1)+1),
-	    "psz2u() does not round up to size class");
-	assert_zu_eq(size_class, psz2u(size_class-1),
-	    "psz2u() does not round up to size class");
-	assert_zu_eq(size_class, psz2u(size_class),
-	    "psz2u() does not compute same size class");
+	assert_u_eq(pind, sz_psz2ind(sz_pind2sz(pind)),
+	    "sz_psz2ind() does not reverse sz_pind2sz()");
+	assert_zu_eq(max_psz, sz_pind2sz(sz_psz2ind(max_psz)),
+	    "sz_pind2sz() does not reverse sz_psz2ind()");
+
+	assert_zu_eq(size_class, sz_psz2u(sz_pind2sz(pind-1)+1),
+	    "sz_psz2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_psz2u(size_class-1),
+	    "sz_psz2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_psz2u(size_class),
+	    "sz_psz2u() does not compute same size class");
 }
 TEST_END
 
@@ -139,35 +142,35 @@ TEST_BEGIN(test_overflow) {
 	max_size_class = get_max_size_class();
 	max_psz = max_size_class + PAGE;
 
-	assert_u_eq(size2index(max_size_class+1), NSIZES,
-	    "size2index() should return NSIZES on overflow");
-	assert_u_eq(size2index(ZU(PTRDIFF_MAX)+1), NSIZES,
-	    "size2index() should return NSIZES on overflow");
-	assert_u_eq(size2index(SIZE_T_MAX), NSIZES,
-	    "size2index() should return NSIZES on overflow");
-
-	assert_zu_eq(s2u(max_size_class+1), 0,
-	    "s2u() should return 0 for unsupported size");
-	assert_zu_eq(s2u(ZU(PTRDIFF_MAX)+1), 0,
-	    "s2u() should return 0 for unsupported size");
-	assert_zu_eq(s2u(SIZE_T_MAX), 0,
-	    "s2u() should return 0 on overflow");
-
-	assert_u_eq(psz2ind(max_size_class+1), NPSIZES,
-	    "psz2ind() should return NPSIZES on overflow");
-	assert_u_eq(psz2ind(ZU(PTRDIFF_MAX)+1), NPSIZES,
-	    "psz2ind() should return NPSIZES on overflow");
-	assert_u_eq(psz2ind(SIZE_T_MAX), NPSIZES,
-	    "psz2ind() should return NPSIZES on overflow");
-
-	assert_zu_eq(psz2u(max_size_class+1), max_psz,
-	    "psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported"
+	assert_u_eq(sz_size2index(max_size_class+1), NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+	assert_u_eq(sz_size2index(ZU(PTRDIFF_MAX)+1), NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+	assert_u_eq(sz_size2index(SIZE_T_MAX), NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+
+	assert_zu_eq(sz_s2u(max_size_class+1), 0,
+	    "sz_s2u() should return 0 for unsupported size");
+	assert_zu_eq(sz_s2u(ZU(PTRDIFF_MAX)+1), 0,
+	    "sz_s2u() should return 0 for unsupported size");
+	assert_zu_eq(sz_s2u(SIZE_T_MAX), 0,
+	    "sz_s2u() should return 0 on overflow");
+
+	assert_u_eq(sz_psz2ind(max_size_class+1), NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+	assert_u_eq(sz_psz2ind(ZU(PTRDIFF_MAX)+1), NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+	assert_u_eq(sz_psz2ind(SIZE_T_MAX), NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+
+	assert_zu_eq(sz_psz2u(max_size_class+1), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported"
 	    " size");
-	assert_zu_eq(psz2u(ZU(PTRDIFF_MAX)+1), max_psz,
-	    "psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported "
+	assert_zu_eq(sz_psz2u(ZU(PTRDIFF_MAX)+1), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported "
 	    "size");
-	assert_zu_eq(psz2u(SIZE_T_MAX), max_psz,
-	    "psz2u() should return (LARGE_MAXCLASS + PAGE) on overflow");
+	assert_zu_eq(sz_psz2u(SIZE_T_MAX), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) on overflow");
 }
 TEST_END