unofficial mirror of bug-gnu-emacs@gnu.org 
 help / color / mirror / code / Atom feed
From: Nicolas Richard <theonewiththeevillook@yahoo.fr>
To: Daniel Colascione <dancol@dancol.org>, Eli Zaretskii <eliz@gnu.org>
Cc: 17184@debbugs.gnu.org
Subject: bug#17184: 24.3.50; crash while bootstrapping current trunk
Date: Tue, 08 Apr 2014 14:54:15 +0200	[thread overview]
Message-ID: <5343F177.6040009@yahoo.fr> (raw)
In-Reply-To: <5343E9E3.7060604@dancol.org>

[-- Attachment #1: Type: text/plain, Size: 2886 bytes --]

Le 08/04/2014 14:21, Daniel Colascione a écrit :
> What happens if you add that Ffset check to emacs-24?

I can bootstrap just fine. Attached is the patch I used (cherry picking
had a small conflict)

> What symbol are we trying to set?
> (Use xsymbol.) What does valid_lisp_object_p (definition) actually
> return when you call it?

Ok, here's a new gdb session. This time I batch compile testcover.el
because I noticed that tcover-ses.el requires it and the problem arises
while requiring testcover.el.

Starting program: /home/youngfrog/sources/emacs-from-git/src/emacs --batch -f batch-byte-compile ../lisp/emacs-lisp/testcover.el
Do you need "set solib-search-path" or "set sysroot"?
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/libthread_db.so.1".

[New Thread 0xb639bb40 (LWP 25884)]

Breakpoint 1, terminate_due_to_signal (sig=6, backtrace_limit=40)
    at emacs.c:355
355	  signal (sig, SIG_DFL);
(gdb) bt
#0  terminate_due_to_signal (sig=6, backtrace_limit=40) at emacs.c:355
#1  0x0816baaf in emacs_abort () at sysdep.c:2130
#2  0x081b5a4b in Ffset (symbol=144208610, definition=137959189) at data.c:733
#3  0x081b5b8a in Fdefalias (symbol=144208610, definition=137959189, 
    docstring=139331522) at data.c:777
#4  0x081ce690 in Ffuncall (nargs=3, args=0xbfffb434) at eval.c:2822
#5  0x08205c55 in exec_byte_code (bytestr=144054457, vector=142916517, 
    maxdepth=12, args_template=139331522, nargs=0, args=0x0) at bytecode.c:919
#6  0x0820535c in Fbyte_code (bytestr=144054457, vector=142916517, maxdepth=12)
    at bytecode.c:482
#7  0x081cd5b1 in eval_sub (form=144189750) at eval.c:2191
#8  0x081f20a1 in readevalloop (readcharfun=139400210, stream=0x893dbe8, 
    sourcename=143932745, printflag=false, unibyte=139331522, 
    readfun=139331522, start=139331522, end=139331522) at lread.c:1934
#9  0x081f108f in Fload (file=137504601, noerror=139331522, 
    nomessage=139331546, nosuffix=139331522, must_suffix=139331546)
    at lread.c:1363

#10 0x081d705a in Frequire (feature=143220250, filename=139331522, 
    noerror=139331522) at fns.c:2671

(gdb) p symbol
No symbol "symbol" in current context.
(gdb) f 2
#2  0x081b5a4b in Ffset (symbol=144208610, definition=137959189) at data.c:733
733	    emacs_abort ();
(gdb) p symbol
$1 = 144208610
(gdb) xsymbol
$2 = (struct Lisp_Symbol *) 0x89872e0
"edebug-original-eval-defun"
(gdb) p valid_lisp_object_p (definition)
$3 = 0
(gdb) f 9
#9  0x081f108f in Fload (file=137504601, noerror=139331522, 
    nomessage=139331546, nosuffix=139331522, must_suffix=139331546)
    at lread.c:1363
1363	    readevalloop (Qget_file_char, stream, hist_file_name,
(gdb) p file
$4 = 137504601
(gdb) xpr
Lisp_String
$5 = (struct Lisp_String *) 0x8322758 <pure+324536>
"edebug"
(gdb) 

> Can you share your binary and core dump?

Sure: http://homepages.ulb.ac.be/~nrichard/emacs+core.tar.bz2

N.

[-- Attachment #2: 0001-Add-GC-bug-investigation-code.patch --]
[-- Type: text/x-patch, Size: 9502 bytes --]

From a8a8ceb973b87b24d2e7317728c3c1358763bab6 Mon Sep 17 00:00:00 2001
From: Daniel Colascione <dancol@dancol.org>
Date: Wed, 2 Apr 2014 17:18:08 -0700
Subject: [PATCH] Add GC bug investigation code Conflicts: 	lisp/ChangeLog
 	lisp/subr.el 	src/ChangeLog

---
 lisp/subr.el |  51 ++++++++++++++++-------------
 src/alloc.c  | 105 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 src/data.c   |   5 +++
 src/lisp.h   |   3 ++
 4 files changed, 141 insertions(+), 23 deletions(-)

diff --git a/lisp/subr.el b/lisp/subr.el
index e4350bf..177e144 100644
--- a/lisp/subr.el
+++ b/lisp/subr.el
@@ -4290,29 +4290,34 @@ lookup sequence then continues."
     ;; Don't use letrec, because equal (in add/remove-hook) would get trapped
     ;; in a cycle.
     (fset clearfun
-          (lambda ()
-            (with-demoted-errors "set-transient-map PCH: %S"
-              (unless (cond
-                       ((not (eq map (cadr overriding-terminal-local-map)))
-                        ;; There's presumably some other transient-map in
-                        ;; effect.  Wait for that one to terminate before we
-                        ;; remove ourselves.
-                        ;; For example, if isearch and C-u both use transient
-                        ;; maps, then the lifetime of the C-u should be nested
-                        ;; within isearch's, so the pre-command-hook of
-                        ;; isearch should be suspended during the C-u one so
-                        ;; we don't exit isearch just because we hit 1 after
-                        ;; C-u and that 1 exits isearch whereas it doesn't
-                        ;; exit C-u.
-                        t)
-                       ((null keep-pred) nil)
-                       ((eq t keep-pred)
-                        (eq this-command
-                            (lookup-key map (this-command-keys-vector))))
-                       (t (funcall keep-pred)))
-                (internal-pop-keymap map 'overriding-terminal-local-map)
-                (remove-hook 'pre-command-hook clearfun)
-                (when on-exit (funcall on-exit))))))
+          (suspicious-object
+           (lambda ()
+             (with-demoted-errors "set-transient-map PCH: %S"
+               (unless (cond
+                         ((not (eq map (cadr overriding-terminal-local-map)))
+                          ;; There's presumably some other transient-map in
+                          ;; effect.  Wait for that one to terminate before we
+                          ;; remove ourselves.
+                          ;; For example, if isearch and C-u both use transient
+                          ;; maps, then the lifetime of the C-u should be nested
+                          ;; within isearch's, so the pre-command-hook of
+                          ;; isearch should be suspended during the C-u one so
+                          ;; we don't exit isearch just because we hit 1 after
+                          ;; C-u and that 1 exits isearch whereas it doesn't
+                          ;; exit C-u.
+                          t)
+                         ((null keep-pred) nil)
+                         ((eq t keep-pred)
+                          (eq this-command
+                              (lookup-key map (this-command-keys-vector))))
+                         (t (funcall keep-pred)))
+                 (internal-pop-keymap map 'overriding-terminal-local-map)
+                 (remove-hook 'pre-command-hook clearfun)
+                 (when on-exit (funcall on-exit))
+                 ;; Comment out the fset if you want to debug the GC bug.
+;;;		(fset clearfun nil)
+;;;             (set clearfun nil)
+                 )))))
     (add-hook 'pre-command-hook clearfun)
     (internal-push-keymap map 'overriding-terminal-local-map)))
 
diff --git a/src/alloc.c b/src/alloc.c
index 62c3bee..5732d16 100644
--- a/src/alloc.c
+++ b/src/alloc.c
@@ -48,6 +48,10 @@ along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.  */
 
 #include <verify.h>
 
+#ifdef HAVE_EXECINFO_H
+#include <execinfo.h>           /* For backtrace */
+#endif
+
 #if (defined ENABLE_CHECKING			\
      && defined HAVE_VALGRIND_VALGRIND_H	\
      && !defined USE_VALGRIND)
@@ -192,6 +196,36 @@ static ptrdiff_t pure_bytes_used_non_lisp;
 
 const char *pending_malloc_warning;
 
+#if 0 /* Normally, pointer sanity only on request... */
+#ifdef ENABLE_CHECKING
+#define SUSPICIOUS_OBJECT_CHECKING 1
+#endif
+#endif
+
+/* ... but unconditionally use SUSPICIOUS_OBJECT_CHECKING while the GC
+   bug is unresolved.  */
+#define SUSPICIOUS_OBJECT_CHECKING 1
+
+#ifdef SUSPICIOUS_OBJECT_CHECKING
+struct suspicious_free_record {
+  void* suspicious_object;
+#ifdef HAVE_EXECINFO_H
+  void* backtrace[128];
+#endif
+};
+static void* suspicious_objects[32];
+static int suspicious_object_index;
+struct suspicious_free_record suspicious_free_history[64];
+static int suspicious_free_history_index;
+/* Find the first currently-monitored suspicious pointer in range
+   [begin,end) or NULL if no such pointer exists.  */
+static void* find_suspicious_object_in_range (void* begin, void* end);
+static void detect_suspicious_free (void* ptr);
+#else
+#define find_suspicious_object_in_range(begin, end) NULL
+#define detect_suspicious_free(ptr) (void)
+#endif
+
 /* Maximum amount of C stack to save when a GC happens.  */
 
 #ifndef MAX_SAVE_STACK
@@ -2914,6 +2948,7 @@ vector_nbytes (struct Lisp_Vector *v)
 static void
 cleanup_vector (struct Lisp_Vector *vector)
 {
+  detect_suspicious_free (vector);
   if (PSEUDOVECTOR_TYPEP (&vector->header, PVEC_FONT)
       && ((vector->header.size & PSEUDOVECTOR_SIZE_MASK)
 	  == FONT_OBJECT_MAX))
@@ -3074,6 +3109,9 @@ allocate_vectorlike (ptrdiff_t len)
       mallopt (M_MMAP_MAX, MMAP_MAX_AREAS);
 #endif
 
+      if (find_suspicious_object_in_range (p, (char*)p + nbytes))
+        emacs_abort ();
+
       consing_since_gc += nbytes;
       vector_cells_consed += len;
     }
@@ -3773,6 +3811,7 @@ refill_memory_reserve (void)
     Vmemory_full = Qnil;
 #endif
 }
+
 \f
 /************************************************************************
 			   C Stack Marking
@@ -6763,6 +6802,71 @@ which_symbols (Lisp_Object obj, EMACS_INT find_max)
    return found;
 }
 
+#ifdef SUSPICIOUS_OBJECT_CHECKING
+
+static void*
+find_suspicious_object_in_range (void* begin, void* end)
+{
+  char* begin_a = begin;
+  char* end_a = end;
+  int i;
+
+  for (i = 0; i < EARRAYSIZE (suspicious_objects); ++i) {
+    char* suspicious_object = suspicious_objects[i];
+    if (begin_a <= suspicious_object && suspicious_object < end_a)
+      return suspicious_object;
+  }
+
+  return NULL;
+}
+
+static void
+detect_suspicious_free (void* ptr)
+{
+  int i;
+  struct suspicious_free_record* rec;
+
+  eassert (ptr != NULL);
+
+  for (i = 0; i < EARRAYSIZE (suspicious_objects); ++i)
+    if (suspicious_objects[i] == ptr)
+      {
+        rec = &suspicious_free_history[suspicious_free_history_index++];
+        if (suspicious_free_history_index ==
+            EARRAYSIZE (suspicious_free_history))
+          {
+            suspicious_free_history_index = 0;
+          }
+
+        memset (rec, 0, sizeof (rec));
+        rec->suspicious_object = ptr;
+#ifdef HAVE_EXECINFO_H
+        backtrace (&rec->backtrace[0], EARRAYSIZE (rec->backtrace));
+#endif
+        suspicious_objects[i] = NULL;
+      }
+}
+
+#endif /* SUSPICIOUS_OBJECT_CHECKING */
+
+DEFUN ("suspicious-object", Fsuspicious_object, Ssuspicious_object, 1, 1, 0,
+       doc: /* Return OBJ, maybe marking it for extra scrutiny.
+If Emacs is compiled with suspicous object checking, capture
+a stack trace when OBJ is freed in order to help track down
+garbage collection bugs.  Otherwise, do nothing and return OBJ.   */)
+   (Lisp_Object obj)
+{
+#ifdef SUSPICIOUS_OBJECT_CHECKING
+  /* Right now, we care only about vectors.  */
+  if (VECTORLIKEP (obj)) {
+    suspicious_objects[suspicious_object_index++] = XVECTOR (obj);
+    if (suspicious_object_index == EARRAYSIZE (suspicious_objects))
+      suspicious_object_index = 0;
+  }
+#endif
+  return obj;
+}
+
 #ifdef ENABLE_CHECKING
 
 bool suppress_checking;
@@ -6933,6 +7037,7 @@ The time is in seconds as a floating point value.  */);
   defsubr (&Sgarbage_collect);
   defsubr (&Smemory_limit);
   defsubr (&Smemory_use_counts);
+  defsubr (&Ssuspicious_object);
 
 #if GC_MARK_STACK == GC_USE_GCPROS_CHECK_ZOMBIES
   defsubr (&Sgc_status);
diff --git a/src/data.c b/src/data.c
index 4ef81f2..dd22098 100644
--- a/src/data.c
+++ b/src/data.c
@@ -727,6 +727,11 @@ DEFUN ("fset", Ffset, Sfset, 2, 2, 0,
   if (AUTOLOADP (function))
     Fput (symbol, Qautoload, XCDR (function));
 
+  /* Convert to eassert or remove after GC bug is found.  In the
+     meantime, check unconditionally, at a slight perf hit.  */
+  if (valid_lisp_object_p (definition) < 1)
+    emacs_abort ();
+
   set_symbol_function (symbol, definition);
 
   return definition;
diff --git a/src/lisp.h b/src/lisp.h
index 30f52b9..c501135 100644
--- a/src/lisp.h
+++ b/src/lisp.h
@@ -58,6 +58,9 @@ INLINE_HEADER_BEGIN
 #define max(a, b) ((a) > (b) ? (a) : (b))
 #define min(a, b) ((a) < (b) ? (a) : (b))
 
+/* Find number of elements in array */
+#define EARRAYSIZE(arr) (sizeof (arr) / sizeof ((arr)[0]))
+
 /* EMACS_INT - signed integer wide enough to hold an Emacs value
    EMACS_INT_MAX - maximum value of EMACS_INT; can be used in #if
    pI - printf length modifier for EMACS_INT
-- 
1.8.3.2


  reply	other threads:[~2014-04-08 12:54 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-04-04 15:26 bug#17184: 24.3.50; crash while bootstrapping current trunk Nicolas Richard
2014-04-04 19:29 ` Eli Zaretskii
2014-04-08 11:20   ` Nicolas Richard
2014-04-08 12:21     ` Daniel Colascione
2014-04-08 12:54       ` Nicolas Richard [this message]
2014-04-08 13:24         ` Daniel Colascione
2014-04-08 13:44           ` Nicolas Richard

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.gnu.org/software/emacs/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5343F177.6040009@yahoo.fr \
    --to=theonewiththeevillook@yahoo.fr \
    --cc=17184@debbugs.gnu.org \
    --cc=dancol@dancol.org \
    --cc=eliz@gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/emacs.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).