diff --git a/src/lj_alloc.c b/src/lj_alloc.c
index 9e2fb1f6c7..f82c9854c8 100644
--- a/src/lj_alloc.c
+++ b/src/lj_alloc.c
@@ -610,7 +610,7 @@ static int has_segment_link(mstate m, msegmentptr ss)
   noncontiguous segments are added.
 */
 #define TOP_FOOT_SIZE\
-  (align_offset(chunk2mem(0))+pad_request(sizeof(struct malloc_segment))+MIN_CHUNK_SIZE)
+  (align_offset(TWO_SIZE_T_SIZES)+pad_request(sizeof(struct malloc_segment))+MIN_CHUNK_SIZE)
 
 /* ---------------------------- Indexing Bins ---------------------------- */
 
diff --git a/src/lj_def.h b/src/lj_def.h
index ba4dcc9d62..a5bca6b03e 100644
--- a/src/lj_def.h
+++ b/src/lj_def.h
@@ -120,7 +120,7 @@ typedef uintptr_t BloomFilter;
 #define bloomset(b, x)	((b) |= bloombit((x)))
 #define bloomtest(b, x)	((b) & bloombit((x)))
 
-#if defined(__GNUC__) || defined(__psp2__)
+#if defined(__GNUC__) || defined(__clang__) || defined(__psp2__)
 
 #define LJ_NORET	__attribute__((noreturn))
 #define LJ_ALIGN(n)	__attribute__((aligned(n)))
@@ -182,7 +182,7 @@ static LJ_AINLINE uint64_t lj_bswap64(uint64_t x)
 {
   return ((uint64_t)lj_bswap((uint32_t)x)<<32) | lj_bswap((uint32_t)(x>>32));
 }
-#elif (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)
+#elif (__GNUC__ > 4) || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3) || __clang__
 static LJ_AINLINE uint32_t lj_bswap(uint32_t x)
 {
   return (uint32_t)__builtin_bswap32((int32_t)x);
diff --git a/src/lj_emit_x86.h b/src/lj_emit_x86.h
index eaef17fc6c..f4990151ef 100644
--- a/src/lj_emit_x86.h
+++ b/src/lj_emit_x86.h
@@ -45,7 +45,7 @@ static LJ_AINLINE MCode *emit_op(x86Op xo, Reg rr, Reg rb, Reg rx,
     *(uint32_t *)(p+delta-5) = (uint32_t)xo;
     return p+delta-5;
   }
-#if defined(__GNUC__)
+#if defined(__GNUC__) || defined(__clang__)
   if (__builtin_constant_p(xo) && n == -2)
     p[delta-2] = (MCode)(xo >> 24);
   else if (__builtin_constant_p(xo) && n == -3)
diff --git a/src/lj_ircall.h b/src/lj_ircall.h
index 9c195918be..7746965a33 100644
--- a/src/lj_ircall.h
+++ b/src/lj_ircall.h
@@ -332,7 +332,7 @@ extern double lj_vm_sfmax(double a, double b);
 #endif
 
 #if LJ_HASFFI && LJ_NEED_FP64 && !(LJ_TARGET_ARM && LJ_SOFTFP)
-#ifdef __GNUC__
+#if defined(__GNUC__) || defined(__clang__)
 #define fp64_l2d __floatdidf
 #define fp64_ul2d __floatundidf
 #define fp64_l2f __floatdisf
diff --git a/src/lj_mcode.c b/src/lj_mcode.c
index a88d16bd5e..33ca2396d2 100644
--- a/src/lj_mcode.c
+++ b/src/lj_mcode.c
@@ -44,7 +44,7 @@ void lj_mcode_sync(void *start, void *end)
   sys_icache_invalidate(start, (char *)end-(char *)start);
 #elif LJ_TARGET_PPC
   lj_vm_cachesync(start, end);
-#elif defined(__GNUC__)
+#elif defined(__GNUC__) || defined(__clang__)
   __clear_cache(start, end);
 #else
 #error "Missing builtin to flush instruction cache"
diff --git a/src/lj_strfmt.h b/src/lj_strfmt.h
index 0e1d894673..d3b3a0ae02 100644
--- a/src/lj_strfmt.h
+++ b/src/lj_strfmt.h
@@ -118,7 +118,7 @@ LJ_FUNC GCstr * LJ_FASTCALL lj_strfmt_obj(lua_State *L, cTValue *o);
 LJ_FUNC const char *lj_strfmt_pushvf(lua_State *L, const char *fmt,
 				     va_list argp);
 LJ_FUNC const char *lj_strfmt_pushf(lua_State *L, const char *fmt, ...)
-#ifdef __GNUC__
+#if defined(__GNUC__) || defined(__clang__)
   __attribute__ ((format (printf, 2, 3)))
 #endif
   ;
diff --git a/src/lj_strscan.c b/src/lj_strscan.c
index 47fab23902..ae8945e1bd 100644
--- a/src/lj_strscan.c
+++ b/src/lj_strscan.c
@@ -79,7 +79,7 @@ static void strscan_double(uint64_t x, TValue *o, int32_t ex2, int32_t neg)
   /* Avoid double rounding for denormals. */
   if (LJ_UNLIKELY(ex2 <= -1075 && x != 0)) {
     /* NYI: all of this generates way too much code on 32 bit CPUs. */
-#if defined(__GNUC__) && LJ_64
+#if (defined(__GNUC__) || defined(__clang__)) && LJ_64
     int32_t b = (int32_t)(__builtin_clzll(x)^63);
 #else
     int32_t b = (x>>32) ? 32+(int32_t)lj_fls((uint32_t)(x>>32)) :