emacs segfault on bootstrap

emacs segfault on bootstrap

[Clemens Schueller]

Hey Folks!


I'm trying to compile the emacs from cvs:


A "./configure --prefix=/opt/emacs22 --disable-largefile" runs without
problems.

But when I type "make bootstrap", I become the following errormessage:


+----
| gcc -c -D_BSD_SOURCE   -Demacs -DHAVE_CONFIG_H   -I. -I/misc/cvs/emacs/src -D_BSD_SOURCE -O9 -funroll-loops -ffast-math -malign-double -march=athlon-xp -fomit-frame-pointer -fno-exceptions vm-limit.c
| gcc -Demacs -DHAVE_CONFIG_H   -I. -I/misc/cvs/emacs/src -D_BSD_SOURCE -O9 -funroll-loops -ffast-math -malign-double -march=athlon-xp -fomit-frame-pointer -fno-exceptions -Wl,-znocombreloc /misc/cvs/emacs/src/prefix-args.c -o prefix-args
| echo "dispnew.o frame.o scroll.o xdisp.o xmenu.o window.o charset.o coding.o category.o ccl.o cm.o term.o xfaces.o   emacs.o keyboard.o macros.o keymap.o sysdep.o buffer.o filelock.o insdel.o marker.o minibuf.o fileio.o dired.o filemode.o cmds.o casetab.o casefiddle.o indent.o search.o regex.o undo.o alloc.o data.o doc.o editfns.o callint.o eval.o floatfns.o fns.o print.o lread.o abbrev.o syntax.o unexelf.o bytecode.o process.o callproc.o region-cache.o sound.o atimer.o doprnt.o strftime.o intervals.o textprop.o composite.o md5.o    terminfo.o lastfile.o   vm-limit.o    " > buildobj.lst
| gcc -nostdlib `./prefix-args -Xlinker  -z nocombreloc` -Wl,-znocombreloc -o temacs pre-crt0.o /usr/lib/crt1.o /usr/lib/crti.o dispnew.o frame.o scroll.o xdisp.o xmenu.o window.o charset.o coding.o category.o ccl.o cm.o term.o xfaces.o   emacs.o keyboard.o macros.o keymap.o sysdep.o buffer.o filelock.o insdel.o marker.o minibuf.o fileio.o dired.o filemode.o cmds.o casetab.o casefiddle.o indent.o search.o regex.o undo.o alloc.o data.o doc.o editfns.o callint.o eval.o floatfns.o fns.o print.o lread.o abbrev.o syntax.o unexelf.o bytecode.o process.o callproc.o region-cache.o sound.o atimer.o doprnt.o strftime.o intervals.o textprop.o composite.o md5.o    terminfo.o lastfile.o   vm-limit.o        -lncurses   -lm -lgcc -lc -lgcc /usr/lib/crtn.o 
| ./temacs --batch --load loadup bootstrap
| Loading loadup.el (source)...
| Using load-path (/misc/cvs/emacs/lisp /misc/cvs/emacs/lisp/emacs-lisp /misc/cvs/emacs/lisp/language /misc/cvs/emacs/lisp/international /misc/cvs/emacs/lisp/textmodes)
| Loading emacs-lisp/byte-run (source)...
| Loading emacs-lisp/backquote (source)...
| Loading subr (source)...
| Wrong type argument: listp, [-134382880 [17224787 #0 make[2]: *** [bootstrap-emacs] Speicherzugriffsfehler
| make[2]: Verlasse Verzeichnis »/misc/cvs/emacs/src«
| make[1]: *** [bootstrap-build] Fehler 2
| make[1]: Verlasse Verzeichnis »/misc/cvs/emacs«
| make: *** [bootstrap] Fehler 2
+----

My locale is set to german: Speicherzugriffsfehler --> segfault
                            Fehler --> error


What goes wrong here - anybody an idea?


TIA


Clemens Schueller
-- 
np: Yves Deruyter - Calling Earth

Re: emacs segfault on bootstrap

[cyd]

> A "./configure --prefix=/opt/emacs22 --disable-largefile" runs without
> problems.
>
> But when I type "make bootstrap", I become the following errormessage:
>
> [bootstrap-emacs] Speicherzugriffsfehler
> | make[2]: Verlasse Verzeichnis »/misc/cvs/emacs/src«
> | make[1]: *** [bootstrap-build] Fehler 2
> | make[1]: Verlasse Verzeichnis »/misc/cvs/emacs«
> | make: *** [bootstrap] Fehler 2
> +----
>
> What goes wrong here - anybody an idea?

I can't reproduce this.

Re: emacs segfault on bootstrap

[David Koppelman]

I have encountered a similar problem. Could you try putting

CFLAGS="-g -O2"

near the top of "configure", perhaps just above
the line:
  # Be Bourne compatible

Then re-run configure, do a "make clean" and then a "make bootstrap".

cyd@stupidchicken.com writes:

>> A "./configure --prefix=/opt/emacs22 --disable-largefile" runs without
>> problems.
>>
>> But when I type "make bootstrap", I become the following errormessage:
>>
>> [bootstrap-emacs] Speicherzugriffsfehler
>> | make[2]: Verlasse Verzeichnis Â»/misc/cvs/emacs/srcÂ«
>> | make[1]: *** [bootstrap-build] Fehler 2
>> | make[1]: Verlasse Verzeichnis Â»/misc/cvs/emacsÂ«
>> | make: *** [bootstrap] Fehler 2
>> +----
>>
>> What goes wrong here - anybody an idea?
>
> I can't reproduce this.

Re: emacs segfault on bootstrap

[Chong Yidong]

OK, now I can reproduce this (I had to reinstall gcc-3.4 -- the bug
doesn't show up for gcc4, or for -O2 and below).

I managed to get a backtrace.  A Lisp_Int was passed to print_object,
but, for some reason, print_object is in the Lisp_Vectorlike case
statement instead.

A cursory inspection did not reveal any bugs in print_object.  Smells
like a compiler bug.

Program received signal SIGSEGV, Segmentation fault.
0x0819c3be in print_object (obj=58, printcharfun=137788457, escapeflag=1)
    at print.c:2009
2009          switch (XMISCTYPE (obj))
(gdb) bt
#0  0x0819c3be in print_object (obj=58, printcharfun=137788457, escapeflag=1)
    at print.c:2009
#1  0x0819c4c5 in print_object (obj=137726760, printcharfun=137788457, 
    escapeflag=1) at print.c:1999
#2  0x0819e989 in Fprin1 (object=137726764, printcharfun=137788457)
    at print.c:1295
#3  0x0819eedd in print_error_message (data=142518805, stream=137788457, 
    context=0x2 <Address 0x2 out of bounds>, caller=142518437) at print.c:1097
#4  0x08112a1a in cmd_error_internal (data=142518805, context=0xbfcfffb0 "")
    at keyboard.c:1263
#5  0x08112bae in cmd_error (data=142518805) at keyboard.c:1200
#6  0x081856fc in internal_condition_case (bfun=0x811c420 <command_loop_1>, 
    handlers=137832193, hfun=0x8112ad0 <cmd_error>) at eval.c:1455
#7  0x0810cda6 in command_loop_2 () at keyboard.c:1323
#8  0x0818564a in internal_catch (tag=2, func=0x810cd80 <command_loop_2>, 
    arg=137788409) at eval.c:1211
#9  0x0810cc3d in recursive_edit_1 () at keyboard.c:1302
#10 0x0810cd3c in Frecursive_edit () at keyboard.c:1056
#11 0x0810bff3 in main (argc=3, argv=0xbfd00574) at emacs.c:1789
(gdb) f 1
#1  0x0819c4c5 in print_object (obj=137726760, printcharfun=137788457, 
    escapeflag=1) at print.c:1999
1999                    print_object (tem, printcharfun, escapeflag);
(gdb) p tem
$3 = 2
(gdb) p size
$4 = 10
(gdb) p obj
$5 = 137726760
(gdb) xtype
Lisp_Int
(gdb) xvector
$6 = (struct Lisp_Vector *) 0x8358b28
0
(gdb) p $6->contents[0]
$11 = 58

Re: emacs segfault on bootstrap

[David Koppelman]

FWIW, I get a "Wrong type argument: listp" under many compile
variations, but only a few are followed by seg faults. (See messages
with the subject "Error with -O3 but not -O2: "Wrong type argument:
listp".)

Chong Yidong <cyd@stupidchicken.com> writes:

> OK, now I can reproduce this (I had to reinstall gcc-3.4 -- the bug
> doesn't show up for gcc4, or for -O2 and below).
>
> I managed to get a backtrace.  A Lisp_Int was passed to print_object,
> but, for some reason, print_object is in the Lisp_Vectorlike case
> statement instead.
>
> A cursory inspection did not reveal any bugs in print_object.  Smells
> like a compiler bug.

Re: emacs segfault on bootstrap

[Eli Zaretskii]

> From: Chong Yidong <cyd@stupidchicken.com>
> Date: Sun, 29 Jan 2006 14:09:03 -0500
> Cc: emacs-devel@gnu.org, Clemens Schueller <c.schueller@gmx.at>
> 
> OK, now I can reproduce this (I had to reinstall gcc-3.4 -- the bug
> doesn't show up for gcc4, or for -O2 and below).
> 
> I managed to get a backtrace.  A Lisp_Int was passed to print_object,
> but, for some reason, print_object is in the Lisp_Vectorlike case
> statement instead.
> 
> A cursory inspection did not reveal any bugs in print_object.  Smells
> like a compiler bug.

Thanks.  I'd suggest to look at the machine code in that function and
see if you can spot any compiler bugs.

Re: emacs segfault on bootstrap

[David Koppelman]

Eli Zaretskii <eliz@gnu.org> writes:

> Thanks.  I'd suggest to look at the machine code in that function and
> see if you can spot any compiler bugs.

Is there a way to get a lisp backtrace in temacs?  It seems like the
problem starts before print_object is called.

Re: emacs segfault on bootstrap

[Eli Zaretskii]

> Cc: Chong Yidong <cyd@stupidchicken.com>, c.schueller@gmx.at,
>    emacs-devel@gnu.org
> From: David Koppelman <koppel@ece.lsu.edu>
> Date: Sun, 29 Jan 2006 13:50:12 -0600
> 
> Is there a way to get a lisp backtrace in temacs?

Sure: start GDB from the src directory, and then you will have the
"xbacktrace" command in GDB to do this.  Is that what you want?

Re: emacs segfault on bootstrap

[Richard M. Stallman]

    A cursory inspection did not reveal any bugs in print_object.  Smells
    like a compiler bug.

    Program received signal SIGSEGV, Segmentation fault.
    0x0819c3be in print_object (obj=58, printcharfun=137788457, escapeflag=1)
	at print.c:2009
    2009          switch (XMISCTYPE (obj))
    (gdb) bt
    #0  0x0819c3be in print_object (obj=58, printcharfun=137788457, escapeflag=1)
	at print.c:2009

    (gdb) p obj
    $5 = 137726760
    (gdb) xtype
    Lisp_Int

Do you think that frame 1 mistakenly treated an int
as a vector?

If so, could you look around some more and try to prove it
conclusively?  Try to determine which line did something wrong.
Perhaps you could then easily step through and prove that the
incorrect computation occurs in a certain line or range of lines.

That's all the info needed to make a useful GCC bug report
so that this will get fixed.

Re: emacs segfault on bootstrap

[Clemens Schueller]

Hello!

* On 29.01.2006, David Koppelman wrote:

> I have encountered a similar problem. Could you try putting
>
> CFLAGS="-g -O2"
>
> near the top of "configure", perhaps just above
> the line:
>   # Be Bourne compatible
>
> Then re-run configure, do a "make clean" and then a "make bootstrap".

Thanks for the hint. This was the correct solution:

My CFLAGS Environment contained a "-O9".



Greetinx and thanks a lot 

Clemens Schueller
-- 
np: Jeanette - Hold the Line

Re: emacs segfault on bootstrap

[Richard M. Stallman]

    My CFLAGS Environment contained a "-O9".

I think that means you have encountered a GCC bug.
We should not just ignore it--GCC needs to be fixed, too.

Is it the same one that we have just been investigating,
or a different one?