Merge remote-tracking branch 'savannah/master' into HEAD

8 files changed, 109 insertions, 39 deletions

diff --git a/src/alloc.c b/src/alloc.c
index b892022125e..dc92d67f163 100644
--- a/src/alloc.c
+++ b/src/alloc.c
@@ -104,6 +104,46 @@ along with GNU Emacs.  If not, see <https://www.gnu.org/licenses/>.  */
 #include "w32heap.h"    /* for sbrk */
 #endif
 
+/* A type with alignment at least as large as any object that Emacs
+   allocates.  This is not max_align_t because some platforms (e.g.,
+   mingw) have buggy malloc implementations that do not align for
+   max_align_t.  This union contains types of all GCALIGNED_STRUCT
+   components visible here.  */
+union emacs_align_type
+{
+  struct frame frame;
+  struct Lisp_Bignum Lisp_Bignum;
+  struct Lisp_Bool_Vector Lisp_Bool_Vector;
+  struct Lisp_Char_Table Lisp_Char_Table;
+  struct Lisp_CondVar Lisp_CondVar;
+  struct Lisp_Finalizer Lisp_Finalizer;
+  struct Lisp_Float Lisp_Float;
+  struct Lisp_Hash_Table Lisp_Hash_Table;
+  struct Lisp_Marker Lisp_Marker;
+  struct Lisp_Misc_Ptr Lisp_Misc_Ptr;
+  struct Lisp_Mutex Lisp_Mutex;
+  struct Lisp_Overlay Lisp_Overlay;
+  struct Lisp_Sub_Char_Table Lisp_Sub_Char_Table;
+  struct Lisp_Subr Lisp_Subr;
+  struct Lisp_User_Ptr Lisp_User_Ptr;
+  struct Lisp_Vector Lisp_Vector;
+  struct terminal terminal;
+  struct thread_state thread_state;
+  struct window window;
+
+  /* Omit the following since they would require including process.h
+     etc.  In practice their alignments never exceed that of the
+     structs already listed.  */
+#if 0
+  struct Lisp_Module_Function Lisp_Module_Function;
+  struct Lisp_Process Lisp_Process;
+  struct save_window_data save_window_data;
+  struct scroll_bar scroll_bar;
+  struct xwidget_view xwidget_view;
+  struct xwidget xwidget;
+#endif
+};
+
 /* MALLOC_SIZE_NEAR (N) is a good number to pass to malloc when
    allocating a block of memory with size close to N bytes.
    For best results N should be a power of 2.
@@ -112,9 +152,9 @@ along with GNU Emacs.  If not, see <https://www.gnu.org/licenses/>.  */
    adds sizeof (size_t) to SIZE for internal overhead, and then rounds
    up to a multiple of MALLOC_ALIGNMENT.  Emacs can improve
    performance a bit on GNU platforms by arranging for the resulting
-   size to be a power of two.  This heuristic is good for glibc 2.0
-   (1997) through at least glibc 2.31 (2020), and does not affect
-   correctness on other platforms.  */
+   size to be a power of two.  This heuristic is good for glibc 2.26
+   (2017) and later, and does not affect correctness on other
+   platforms.  */
 
 #define MALLOC_SIZE_NEAR(n) \
   (ROUNDUP (max (n, sizeof (size_t)), MALLOC_ALIGNMENT) - sizeof (size_t))
@@ -655,25 +695,19 @@ buffer_memory_full (ptrdiff_t nbytes)
 #define COMMON_MULTIPLE(a, b) \
   ((a) % (b) == 0 ? (a) : (b) % (a) == 0 ? (b) : (a) * (b))
 
-/* LISP_ALIGNMENT is the alignment of Lisp objects.  It must be at
-   least GCALIGNMENT so that pointers can be tagged.  It also must be
-   at least as strict as the alignment of all the C types used to
-   implement Lisp objects; since pseudovectors can contain any C type,
-   this is max_align_t.  On recent GNU/Linux x86 and x86-64 this can
-   often waste up to 8 bytes, since alignof (max_align_t) is 16 but
-   typical vectors need only an alignment of 8.  Although shrinking
-   the alignment to 8 would save memory, it cost a 20% hit to Emacs
-   CPU performance on Fedora 28 x86-64 when compiled with gcc -m32.  */
-enum { LISP_ALIGNMENT = alignof (union { max_align_t x;
+/* Alignment needed for memory blocks that are allocated via malloc
+   and that contain Lisp objects.  On typical hosts malloc already
+   aligns sufficiently, but extra work is needed on oddball hosts
+   where Emacs would crash if malloc returned a non-GCALIGNED pointer.  */
+enum { LISP_ALIGNMENT = alignof (union { union emacs_align_type x;
                                          GCALIGNED_UNION_MEMBER }) };
 verify (LISP_ALIGNMENT % GCALIGNMENT == 0);
 
 /* True if malloc (N) is known to return storage suitably aligned for
    Lisp objects whenever N is a multiple of LISP_ALIGNMENT.  In
    practice this is true whenever alignof (max_align_t) is also a
-   multiple of LISP_ALIGNMENT.  This works even for x86, where some
-   platform combinations (e.g., GCC 7 and later, glibc 2.25 and
-   earlier) have bugs where alignof (max_align_t) is 16 even though
+   multiple of LISP_ALIGNMENT.  This works even for buggy platforms
+   like MinGW circa 2020, where alignof (max_align_t) is 16 even though
    the malloc alignment is only 8, and where Emacs still works because
    it never does anything that requires an alignment of 16.  */
 enum { MALLOC_IS_LISP_ALIGNED = alignof (max_align_t) % LISP_ALIGNMENT == 0 };
@@ -4660,16 +4694,33 @@ mark_maybe_objects (Lisp_Object const *array, ptrdiff_t nelts)
     mark_maybe_object (*array);
 }
 
+/* A lower bound on the alignment of Lisp objects that need marking.
+   Although 1 is safe, higher values speed up mark_maybe_pointer.
+   If USE_LSB_TAG, this value is typically GCALIGNMENT; otherwise,
+   it's determined by the natural alignment of Lisp structs.
+   All vectorlike objects have alignment at least that of union
+   vectorlike_header and it's unlikely they all have alignment greater,
+   so use the union as a safe and likely-accurate standin for
+   vectorlike objects.  */
+
+enum { GC_OBJECT_ALIGNMENT_MINIMUM
+         = max (GCALIGNMENT,
+                min (alignof (union vectorlike_header),
+                     min (min (alignof (struct Lisp_Cons),
+                               alignof (struct Lisp_Float)),
+                          min (alignof (struct Lisp_String),
+                               alignof (struct Lisp_Symbol))))) };
+
 /* Return true if P might point to Lisp data that can be garbage
    collected, and false otherwise (i.e., false if it is easy to see
    that P cannot point to Lisp data that can be garbage collected).
    Symbols are implemented via offsets not pointers, but the offsets
-   are also multiples of LISP_ALIGNMENT.  */
+   are also multiples of GC_OBJECT_ALIGNMENT_MINIMUM.  */
 
 static bool
 maybe_lisp_pointer (void *p)
 {
-  return (uintptr_t) p % LISP_ALIGNMENT == 0;
+  return (uintptr_t) p % GC_OBJECT_ALIGNMENT_MINIMUM == 0;
 }
 
 /* If P points to Lisp data, mark that as live if it isn't already
@@ -4892,9 +4943,10 @@ test_setjmp (void)
    as a stack scan limit.  */
 typedef union
 {
-  /* Align the stack top properly.  Even if !HAVE___BUILTIN_UNWIND_INIT,
-     jmp_buf may not be aligned enough on darwin-ppc64.  */
-  max_align_t o;
+  /* Make sure stack_top and m_stack_bottom are properly aligned as GC
+     expects.  */
+  Lisp_Object o;
+  void *p;
 #ifndef HAVE___BUILTIN_UNWIND_INIT
   sys_jmp_buf j;
   char c;
diff --git a/src/cmds.c b/src/cmds.c
index 9f96f210b9f..90526612b7a 100644
--- a/src/cmds.c
+++ b/src/cmds.c
@@ -194,7 +194,7 @@ to t.  */)
       SET_PT (newpos);
 
       if (PT > newpos
-          && FETCH_CHAR (PT - 1) == '\n')
+          && FETCH_BYTE (PT_BYTE - 1) == '\n')
         {
           /* If we skipped over a newline that follows
              an invisible intangible run,
@@ -205,7 +205,7 @@ to t.  */)
           break;
         }
       else if (PT > newpos && PT < ZV
-               && FETCH_CHAR (PT) != '\n')
+               && FETCH_BYTE (PT_BYTE) != '\n')
         /* If we skipped something intangible
            and now we're not really at eol,
            keep going.  */
diff --git a/src/coding.c b/src/coding.c
index 34f36d5a86a..071124b4ef1 100644
--- a/src/coding.c
+++ b/src/coding.c
@@ -10395,7 +10395,7 @@ representation of the decoded text.
 
 This function sets `last-coding-system-used' to the precise coding system
 used (which may be different from CODING-SYSTEM if CODING-SYSTEM is
-not fully specified.)  */)
+not fully specified.)  The function does not change the match data.  */)
   (Lisp_Object string, Lisp_Object coding_system, Lisp_Object nocopy, Lisp_Object buffer)
 {
   return code_convert_string (string, coding_system, buffer,
@@ -10415,7 +10415,7 @@ case, the return value is the length of the encoded text.
 
 This function sets `last-coding-system-used' to the precise coding system
 used (which may be different from CODING-SYSTEM if CODING-SYSTEM is
-not fully specified.)  */)
+not fully specified.)  The function does not change the match data.  */)
   (Lisp_Object string, Lisp_Object coding_system, Lisp_Object nocopy, Lisp_Object buffer)
 {
   return code_convert_string (string, coding_system, buffer,
diff --git a/src/lisp.h b/src/lisp.h
index 52242791aa5..d39300e5598 100644
--- a/src/lisp.h
+++ b/src/lisp.h
@@ -251,12 +251,6 @@ DEFINE_GDB_SYMBOL_BEGIN (EMACS_INT, VALMASK)
 # define VALMASK (USE_LSB_TAG ? - (1 << GCTYPEBITS) : VAL_MAX)
 DEFINE_GDB_SYMBOL_END (VALMASK)
 
-#if !USE_LSB_TAG && !defined WIDE_EMACS_INT
-# error "USE_LSB_TAG not supported on this platform; please report this." \
-        "Try 'configure --with-wide-int' to work around the problem."
-error !;
-#endif
-
 /* Minimum alignment requirement for Lisp objects, imposed by the
    internal representation of tagged pointers.  It is 2**GCTYPEBITS if
    USE_LSB_TAG, 1 otherwise.  It must be a literal integer constant,
@@ -277,7 +271,8 @@ error !;
    allocation in a containing union that has GCALIGNED_UNION_MEMBER)
    and does not contain a GC-aligned struct or union, putting
    GCALIGNED_STRUCT after its closing '}' can help the compiler
-   generate better code.
+   generate better code.  Also, such structs should be added to the
+   emacs_align_type union in alloc.c.
 
    Although these macros are reasonably portable, they are not
    guaranteed on non-GCC platforms, as C11 does not require support
@@ -2809,8 +2804,10 @@ struct Lisp_Float
     {
       double data;
       struct Lisp_Float *chain;
+      GCALIGNED_UNION_MEMBER
     } u;
-  } GCALIGNED_STRUCT;
+  };
+verify (GCALIGNED (struct Lisp_Float));
 
 INLINE bool
 (FLOATP) (Lisp_Object x)
diff --git a/src/lread.c b/src/lread.c
index 46725d9b0ff..9f849eda423 100644
--- a/src/lread.c
+++ b/src/lread.c
@@ -3908,6 +3908,10 @@ read_vector (Lisp_Object readcharfun, bool bytecodeflag)
   ptrdiff_t size = list_length (tem);
   Lisp_Object vector = make_nil_vector (size);
 
+  /* Avoid accessing past the end of a vector if the vector is too
+     small to be valid for bytecode.  */
+  bytecodeflag &= COMPILED_STACK_DEPTH < size;
+
   Lisp_Object *ptr = XVECTOR (vector)->contents;
   for (ptrdiff_t i = 0; i < size; i++)
     {
diff --git a/src/pdumper.c b/src/pdumper.c
index 19dbacca896..29e3560ee5a 100644
--- a/src/pdumper.c
+++ b/src/pdumper.c
@@ -2620,7 +2620,7 @@ dump_vectorlike_generic (struct dump_context *ctx,
       Lisp_Object out;
       const Lisp_Object *vslot = &v->contents[i];
       /* In the wide case, we're always misaligned.  */
-#ifndef WIDE_EMACS_INT
+#if INTPTR_MAX == EMACS_INT_MAX
       eassert (ctx->offset % sizeof (out) == 0);
 #endif
       dump_object_start (ctx, &out, sizeof (out));
diff --git a/src/thread.c b/src/thread.c
index df1a7053826..b638dd77f8b 100644
--- a/src/thread.c
+++ b/src/thread.c
@@ -717,12 +717,17 @@ run_thread (void *state)
 {
   /* Make sure stack_top and m_stack_bottom are properly aligned as GC
      expects.  */
-  max_align_t stack_pos;
+  union
+  {
+    Lisp_Object o;
+    void *p;
+    char c;
+  } stack_pos;
 
   struct thread_state *self = state;
   struct thread_state **iter;
 
-  self->m_stack_bottom = self->stack_top = (char *) &stack_pos;
+  self->m_stack_bottom = self->stack_top = &stack_pos.c;
   self->thread_id = sys_thread_self ();
 
   if (self->thread_name)
diff --git a/src/xdisp.c b/src/xdisp.c
index cf15f579b58..db0ec683159 100644
--- a/src/xdisp.c
+++ b/src/xdisp.c
@@ -8290,8 +8290,8 @@ get_visually_first_element (struct it *it)
     }
   else if (it->bidi_it.charpos == bob
            || (!string_p
-               && (FETCH_CHAR (it->bidi_it.bytepos - 1) == '\n'
-                   || FETCH_CHAR (it->bidi_it.bytepos) == '\n')))
+               && (FETCH_BYTE (it->bidi_it.bytepos - 1) == '\n'
+                   || FETCH_BYTE (it->bidi_it.bytepos) == '\n')))
     {
       /* If we are at the beginning of a line/string, we can produce
          the next element right away.  */
@@ -15566,6 +15566,12 @@ redisplay_internal (void)
           if (it.current_x != this_line_start_x)
             goto cancel;
 
+          /* Give up on this optimization if the line starts with a
+             string with display property that draws on the fringes,
+             as that might interfere with line-prefix display.  */
+          if (it.sp > 1
+              && it.method == GET_FROM_IMAGE && it.image_id == -1)
+            goto cancel;
           redisplay_trace ("trying display optimization 1\n");
           w->cursor.vpos = -1;
           overlay_arrow_seen = false;
@@ -20338,6 +20344,12 @@ try_window_id (struct window *w)
 
       if (! init_to_row_end (&it, w, last_unchanged_at_beg_row))
         GIVE_UP (18);
+      /* Give up if the row starts with a display property that draws
+         on the fringes, since that could prevent correct display of
+         line-prefix and wrap-prefix.  */
+      if (it.sp > 1
+          && it.method == GET_FROM_IMAGE && it.image_id == -1)
+        GIVE_UP (26);
       start_pos = it.current.pos;
 
       /* Start displaying new lines in the desired matrix at the same
@@ -24182,7 +24194,7 @@ the `bidi-class' property of a character.  */)
           itb.charpos = BEGV;
           itb.bytepos = BEGV_BYTE;
         }
-      else if (FETCH_CHAR (from_bpos - 1) == '\n')
+      else if (FETCH_BYTE (from_bpos - 1) == '\n')
         {
           itb.charpos = from_pos;
           itb.bytepos = from_bpos;