Simplify malloc initialization

Singificantly reduce the code needed at malloc initialization.  In
the process getpagesize is simplified by always initializing
GLRO(dl_pagesize).

8 files changed, 65 insertions, 98 deletions

diff --git a/ChangeLog b/ChangeLog
index 9b2b304..05a6b61 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,9 +1,25 @@
 2011-09-10  Ulrich Drepper  <drepper@gmail.com>
 
+	* malloc/arena.c (ptmalloc_init_minimal): Removed.  Initialize all
+	variables statically.
+	(narenas): Initialize.
+	(list_lock): Initialize.
+	(ptmalloc_init): Don't call ptmalloc_init_minimal.  Remove initializtion
+	of main_arena and list_lock.  Small cleanups.
+	Replace all uses of malloc_getpagesize with GLRO(dl_pagesize).
+	* malloc/malloc.c: Remove malloc_getpagesize.  Include <ldsodefs.h>.
+	Add initializers to main_arena and mp_.
+	(malloc_state): Remove pagesize member.  Change all users to use
+	GLRO(dl_pagesize).
+
+	* elf/rtld.c (rtld_global_ro): Initialize _dl_pagesize.
+	* sysdeps/unix/sysv/linux/getpagesize.c: Simplify.  GLRO(dl_pagesize)
+	is always initialized.
+
 	* malloc/malloc.c: Removed unused configurations and dead code.
 	* malloc/arena.c: Likewise.
 	* malloc/hooks.c: Likewise.
-	* malloc/Makefile (-CPPFLAGS-malloc.c): Don't add -DATOMIC_FASTBINS.
+	* malloc/Makefile (CPPFLAGS-malloc.c): Don't add -DATOMIC_FASTBINS.
 
 	* include/tls.h: Removed.  USE___THREAD must always be defined.
 	* bits/libc-tsd.h: Don't handle !USE___THREAD.
diff --git a/elf/rtld.c b/elf/rtld.c
index 4f6bbfd..b3959a3 100644
--- a/elf/rtld.c
+++ b/elf/rtld.c
@@ -161,6 +161,7 @@ struct rtld_global_ro _rtld_global_ro attribute_relro =
     ._dl_lazy = 1,
     ._dl_fpu_control = _FPU_DEFAULT,
     ._dl_pointer_guard = 1,
+    ._dl_pagesize = EXEC_PAGESIZE,
 
     /* Function pointers.  */
     ._dl_debug_printf = _dl_debug_printf,
diff --git a/malloc/arena.c b/malloc/arena.c
index 6d5114b..9114fd2 100644
--- a/malloc/arena.c
+++ b/malloc/arena.c
@@ -76,9 +76,9 @@ extern int sanity_check_heap_info_alignment[(sizeof (heap_info)
 /* Thread specific data */
 
 static tsd_key_t arena_key;
-static mutex_t list_lock;
+static mutex_t list_lock = MUTEX_INITIALIZER;
 #ifdef PER_THREAD
-static size_t narenas;
+static size_t narenas = 1;
 static mstate free_list;
 #endif
 
@@ -353,24 +353,6 @@ next_env_entry (char ***position)
   return result;
 }
 
-/* Set up basic state so that _int_malloc et al can work.  */
-static void
-ptmalloc_init_minimal (void)
-{
-#if DEFAULT_TOP_PAD != 0
-  mp_.top_pad        = DEFAULT_TOP_PAD;
-#endif
-  mp_.n_mmaps_max    = DEFAULT_MMAP_MAX;
-  mp_.mmap_threshold = DEFAULT_MMAP_THRESHOLD;
-  mp_.trim_threshold = DEFAULT_TRIM_THRESHOLD;
-  mp_.pagesize       = malloc_getpagesize;
-#ifdef PER_THREAD
-# define NARENAS_FROM_NCORES(n) ((n) * (sizeof(long) == 4 ? 2 : 8))
-  mp_.arena_test     = NARENAS_FROM_NCORES (1);
-  narenas = 1;
-#endif
-}
-
 
 #ifdef SHARED
 static void *
@@ -386,17 +368,9 @@ libc_hidden_proto (_dl_open_hook);
 static void
 ptmalloc_init (void)
 {
-  const char* s;
-  int secure = 0;
-
   if(__malloc_initialized >= 0) return;
   __malloc_initialized = 0;
 
-  ptmalloc_init_minimal();
-
-  mutex_init(&main_arena.mutex);
-  main_arena.next = &main_arena;
-
 #ifdef SHARED
   /* In case this libc copy is in a non-default namespace, never use brk.
      Likewise if dlopened from statically linked program.  */
@@ -409,12 +383,10 @@ ptmalloc_init (void)
     __morecore = __failing_morecore;
 #endif
 
-  mutex_init(&list_lock);
   tsd_key_create(&arena_key, NULL);
   tsd_setspecific(arena_key, (void *)&main_arena);
   thread_atfork(ptmalloc_lock_all, ptmalloc_unlock_all, ptmalloc_unlock_all2);
-  secure = __libc_enable_secure;
-  s = NULL;
+  const char *s = NULL;
   if (__builtin_expect (_environ != NULL, 1))
     {
       char **runp = _environ;
@@ -438,7 +410,7 @@ ptmalloc_init (void)
 		s = &envline[7];
 	      break;
 	    case 8:
-	      if (! secure)
+	      if (! __builtin_expect (__libc_enable_secure, 0))
 		{
 		  if (memcmp (envline, "TOP_PAD_", 8) == 0)
 		    mALLOPt(M_TOP_PAD, atoi(&envline[9]));
@@ -447,7 +419,7 @@ ptmalloc_init (void)
 		}
 	      break;
 	    case 9:
-	      if (! secure)
+	      if (! __builtin_expect (__libc_enable_secure, 0))
 		{
 		  if (memcmp (envline, "MMAP_MAX_", 9) == 0)
 		    mALLOPt(M_MMAP_MAX, atoi(&envline[10]));
@@ -459,7 +431,7 @@ ptmalloc_init (void)
 	      break;
 #ifdef PER_THREAD
 	    case 10:
-	      if (! secure)
+	      if (! __builtin_expect (__libc_enable_secure, 0))
 		{
 		  if (memcmp (envline, "ARENA_TEST", 10) == 0)
 		    mALLOPt(M_ARENA_TEST, atoi(&envline[11]));
@@ -467,7 +439,7 @@ ptmalloc_init (void)
 	      break;
 #endif
 	    case 15:
-	      if (! secure)
+	      if (! __builtin_expect (__libc_enable_secure, 0))
 		{
 		  if (memcmp (envline, "TRIM_THRESHOLD_", 15) == 0)
 		    mALLOPt(M_TRIM_THRESHOLD, atoi(&envline[16]));
@@ -549,7 +521,7 @@ static heap_info *
 internal_function
 new_heap(size_t size, size_t top_pad)
 {
-  size_t page_mask = malloc_getpagesize - 1;
+  size_t page_mask = GLRO(dl_pagesize) - 1;
   char *p1, *p2;
   unsigned long ul;
   heap_info *h;
@@ -619,7 +591,7 @@ new_heap(size_t size, size_t top_pad)
 static int
 grow_heap(heap_info *h, long diff)
 {
-  size_t page_mask = malloc_getpagesize - 1;
+  size_t page_mask = GLRO(dl_pagesize) - 1;
   long new_size;
 
   diff = (diff + page_mask) & ~page_mask;
@@ -679,7 +651,7 @@ internal_function
 heap_trim(heap_info *heap, size_t pad)
 {
   mstate ar_ptr = heap->ar_ptr;
-  unsigned long pagesz = mp_.pagesize;
+  unsigned long pagesz = GLRO(dl_pagesize);
   mchunkptr top_chunk = top(ar_ptr), p, bck, fwd;
   heap_info *prev_heap;
   long new_size, top_size, extra;
diff --git a/malloc/hooks.c b/malloc/hooks.c
index a0793a6..7817ae9 100644
--- a/malloc/hooks.c
+++ b/malloc/hooks.c
@@ -147,7 +147,7 @@ mem2chunk_check(void* mem, unsigned char **magic_p)
       if(c<=0 || sz<(c+2*SIZE_SZ)) return NULL;
     }
   } else {
-    unsigned long offset, page_mask = malloc_getpagesize-1;
+    unsigned long offset, page_mask = GLRO(dl_pagesize)-1;
 
     /* mmap()ed chunks have MALLOC_ALIGNMENT or higher power-of-two
        alignment relative to the beginning of a page.  Check this
@@ -182,7 +182,7 @@ top_check(void)
   mchunkptr t = top(&main_arena);
   char* brk, * new_brk;
   INTERNAL_SIZE_T front_misalign, sbrk_size;
-  unsigned long pagesz = malloc_getpagesize;
+  unsigned long pagesz = GLRO(dl_pagesize);
 
   if (t == initial_top(&main_arena) ||
       (!chunk_is_mmapped(t) &&
diff --git a/malloc/malloc.c b/malloc/malloc.c
index 0683eee..3cff6d8 100644
--- a/malloc/malloc.c
+++ b/malloc/malloc.c
@@ -170,7 +170,6 @@
     Compilation Environment options:
 
     HAVE_MREMAP                0 unless linux defined
-    malloc_getpagesize         derived from system #includes, or 4096 if not
 
     Changing default word sizes:
 
@@ -226,6 +225,8 @@
 #include <bits/wordsize.h>
 #include <sys/sysinfo.h>
 
+#include <ldsodefs.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -454,7 +455,6 @@ __malloc_assert (const char *assertion, const char *file, unsigned int line,
 #define public_iCOMALLOc __libc_independent_comalloc
 #define public_gET_STATe __malloc_get_state
 #define public_sET_STATe __malloc_set_state
-#define malloc_getpagesize __getpagesize()
 #define open             __open
 #define mmap             __mmap
 #define munmap           __munmap
@@ -589,24 +589,6 @@ void *(*__morecore)(ptrdiff_t) = __default_morecore;
 
 
 /*
-  The system page size. To the extent possible, this malloc manages
-  memory from the system in page-size units.  Note that this value is
-  cached during initialization into a field of malloc_state. So even
-  if malloc_getpagesize is a function, it is only called once.
-
-  The following mechanics for getpagesize were adapted from bsd/gnu
-  getpagesize.h. If none of the system-probes here apply, a value of
-  4096 is used, which should be OK: If they don't apply, then using
-  the actual value probably doesn't impact performance.
-*/
-
-
-#ifndef malloc_getpagesize
-# include <unistd.h>
-# define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
-#endif
-
-/*
   This version of malloc supports the standard SVID/XPG mallinfo
   routine that returns a struct containing usage properties and
   statistics. It should work on any SVID/XPG compliant system that has
@@ -1878,9 +1860,6 @@ struct malloc_par {
      dynamic behavior. */
   int              no_dyn_threshold;
 
-  /* Cache malloc_getpagesize */
-  unsigned int     pagesize;
-
   /* Statistics */
   INTERNAL_SIZE_T  mmapped_mem;
   /*INTERNAL_SIZE_T  sbrked_mem;*/
@@ -1898,11 +1877,25 @@ struct malloc_par {
    before using. This malloc relies on the property that malloc_state
    is initialized to all zeroes (as is true of C statics).  */
 
-static struct malloc_state main_arena;
+static struct malloc_state main_arena =
+  {
+    .mutex = MUTEX_INITIALIZER,
+    .next = &main_arena
+  };
 
 /* There is only one instance of the malloc parameters.  */
 
-static struct malloc_par mp_;
+static struct malloc_par mp_ =
+  {
+    .top_pad        = DEFAULT_TOP_PAD,
+    .n_mmaps_max    = DEFAULT_MMAP_MAX,
+    .mmap_threshold = DEFAULT_MMAP_THRESHOLD,
+    .trim_threshold = DEFAULT_TRIM_THRESHOLD,
+#ifdef PER_THREAD
+# define NARENAS_FROM_NCORES(n) ((n) * (sizeof(long) == 4 ? 2 : 8))
+    .arena_test     = NARENAS_FROM_NCORES (1)
+#endif
+  };
 
 
 #ifdef PER_THREAD
@@ -2070,7 +2063,7 @@ static void do_check_chunk(mstate av, mchunkptr p)
       assert(((char*)p) < min_address || ((char*)p) >= max_address);
     }
     /* chunk is page-aligned */
-    assert(((p->prev_size + sz) & (mp_.pagesize-1)) == 0);
+    assert(((p->prev_size + sz) & (GLRO(dl_pagesize)-1)) == 0);
     /* mem is aligned */
     assert(aligned_OK(chunk2mem(p)));
   }
@@ -2231,7 +2224,7 @@ static void do_check_malloc_state(mstate av)
     return;
 
   /* pagesize is a power of 2 */
-  assert((mp_.pagesize & (mp_.pagesize-1)) == 0);
+  assert((GLRO(dl_pagesize) & (GLRO(dl_pagesize)-1)) == 0);
 
   /* A contiguous main_arena is consistent with sbrk_base.  */
   if (av == &main_arena && contiguous(av))
@@ -2389,7 +2382,7 @@ static void* sYSMALLOc(INTERNAL_SIZE_T nb, mstate av)
 
   unsigned long   sum;            /* for updating stats */
 
-  size_t          pagemask  = mp_.pagesize - 1;
+  size_t          pagemask  = GLRO(dl_pagesize) - 1;
   bool            tried_mmap = false;
 
 
@@ -2804,7 +2797,7 @@ static int sYSTRIm(size_t pad, mstate av)
   char* new_brk;         /* address returned by post-check sbrk call */
   size_t pagesz;
 
-  pagesz = mp_.pagesize;
+  pagesz = GLRO(dl_pagesize);
   top_size = chunksize(av->top);
 
   /* Release in pagesize units, keeping at least one page */
@@ -2867,7 +2860,7 @@ munmap_chunk(mchunkptr p)
      page size.  But gcc does not recognize the optimization possibility
      (in the moment at least) so we combine the two values into one before
      the bit test.  */
-  if (__builtin_expect (((block | total_size) & (mp_.pagesize - 1)) != 0, 0))
+  if (__builtin_expect (((block | total_size) & (GLRO(dl_pagesize) - 1)) != 0, 0))
     {
       malloc_printerr (check_action, "munmap_chunk(): invalid pointer",
 		       chunk2mem (p));
@@ -2889,13 +2882,13 @@ static mchunkptr
 internal_function
 mremap_chunk(mchunkptr p, size_t new_size)
 {
-  size_t page_mask = mp_.pagesize - 1;
+  size_t page_mask = GLRO(dl_pagesize) - 1;
   INTERNAL_SIZE_T offset = p->prev_size;
   INTERNAL_SIZE_T size = chunksize(p);
   char *cp;
 
   assert (chunk_is_mmapped(p));
-  assert(((size + offset) & (mp_.pagesize-1)) == 0);
+  assert(((size + offset) & (GLRO(dl_pagesize)-1)) == 0);
 
   /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
   new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask;
@@ -3157,7 +3150,7 @@ public_vALLOc(size_t bytes)
   if(__malloc_initialized < 0)
     ptmalloc_init ();
 
-  size_t pagesz = mp_.pagesize;
+  size_t pagesz = GLRO(dl_pagesize);
 
   __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, size_t,
 					__const __malloc_ptr_t)) =
@@ -3201,8 +3194,8 @@ public_pVALLOc(size_t bytes)
   if(__malloc_initialized < 0)
     ptmalloc_init ();
 
-  size_t pagesz = mp_.pagesize;
-  size_t page_mask = mp_.pagesize - 1;
+  size_t pagesz = GLRO(dl_pagesize);
+  size_t page_mask = GLRO(dl_pagesize) - 1;
   size_t rounded_bytes = (bytes + page_mask) & ~(page_mask);
 
   __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, size_t,
@@ -4582,7 +4575,7 @@ _int_valloc(mstate av, size_t bytes)
 {
   /* Ensure initialization/consolidation */
   if (have_fastchunks(av)) malloc_consolidate(av);
-  return _int_memalign(av, mp_.pagesize, bytes);
+  return _int_memalign(av, GLRO(dl_pagesize), bytes);
 }
 
 /*
@@ -4597,7 +4590,7 @@ _int_pvalloc(mstate av, size_t bytes)
 
   /* Ensure initialization/consolidation */
   if (have_fastchunks(av)) malloc_consolidate(av);
-  pagesz = mp_.pagesize;
+  pagesz = GLRO(dl_pagesize);
   return _int_memalign(av, pagesz, (bytes + pagesz - 1) & ~(pagesz - 1));
 }
 
@@ -4611,7 +4604,7 @@ static int mTRIm(mstate av, size_t pad)
   /* Ensure initialization/consolidation */
   malloc_consolidate (av);
 
-  const size_t ps = mp_.pagesize;
+  const size_t ps = GLRO(dl_pagesize);
   int psindex = bin_index (ps);
   const size_t psm1 = ps - 1;
 
diff --git a/nptl/ChangeLog b/nptl/ChangeLog
index 56f4d1d..b343939 100644
--- a/nptl/ChangeLog
+++ b/nptl/ChangeLog
@@ -1,5 +1,7 @@
 2011-09-10  Ulrich Drepper  <drepper@gmail.com>
 
+	* sysdeps/pthread/malloc-machine.h: Define MUTEX_INITIALIZER.
+
 	* sysdeps/unix/sysv/linux/i386/i486/sem_post.S: Don't handle
 	!USE___THREAD.
 	* sysdeps/unix/sysv/linux/i386/i486/sem_timedwait.S: Likewise.
diff --git a/nptl/sysdeps/pthread/malloc-machine.h b/nptl/sysdeps/pthread/malloc-machine.h
index e99aaa7..73503ae 100644
--- a/nptl/sysdeps/pthread/malloc-machine.h
+++ b/nptl/sysdeps/pthread/malloc-machine.h
@@ -1,6 +1,6 @@
 /* Basic platform-independent macro definitions for mutexes,
    thread-specific data and parameters for malloc.
-   Copyright (C) 2003, 2007, 2008 Free Software Foundation, Inc.
+   Copyright (C) 2003, 2007, 2008, 2011 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
 
    The GNU C Library is free software; you can redistribute it and/or
@@ -32,6 +32,7 @@ __libc_lock_define (typedef, mutex_t)
 #define mutex_lock(m)		__libc_lock_lock (*(m))
 #define mutex_trylock(m)	__libc_lock_trylock (*(m))
 #define mutex_unlock(m)		__libc_lock_unlock (*(m))
+#define MUTEX_INITIALIZER	LLL_LOCK_INITIALIZER
 
 /* This is defined by newer gcc version unique for each module.  */
 extern void *__dso_handle __attribute__ ((__weak__));
diff --git a/sysdeps/unix/sysv/linux/getpagesize.c b/sysdeps/unix/sysv/linux/getpagesize.c
index 0866079..0de52ea 100644
--- a/sysdeps/unix/sysv/linux/getpagesize.c
+++ b/sysdeps/unix/sysv/linux/getpagesize.c
@@ -28,26 +28,8 @@
 int
 __getpagesize ()
 {
-#ifdef __ASSUME_AT_PAGESIZE
   assert (GLRO(dl_pagesize) != 0);
   return GLRO(dl_pagesize);
-#else
-  if (GLRO(dl_pagesize) != 0)
-    return GLRO(dl_pagesize);
-
-# ifdef	EXEC_PAGESIZE
-  return EXEC_PAGESIZE;
-# else	/* No EXEC_PAGESIZE.  */
-#  ifdef NBPG
-#   ifndef CLSIZE
-#    define CLSIZE	1
-#   endif	/* No CLSIZE.  */
-  return NBPG * CLSIZE;
-#  else	/* No NBPG.  */
-  return NBPC;
-#  endif	/* NBPG.  */
-# endif	/* EXEC_PAGESIZE.  */
-#endif
 }
 libc_hidden_def (__getpagesize)
 weak_alias (__getpagesize, getpagesize)