crash with Emacs CVS 2003-05-09

crash with Emacs CVS 2003-05-09

[Werner LEMBERG]

[I'm using this old CVS version because the current one has severe
problems with fontsets.]

While working with mew I got the segfault below.


    Werner


======================================================================


Program received signal SIGSEGV, Segmentation fault.
0x08077370 in fill_glyph_string (s=0xbfffe274, face_id=12, start=0, end=36, 
    overlaps_p=0) at xdisp.c:16554
16554     PREPARE_FACE_FOR_DISPLAY (f, face);
(gdb) bt full
#0  0x08077370 in fill_glyph_string (s=0xbfffe274, face_id=12, start=0, 
    end=36, overlaps_p=0) at xdisp.c:16554
        face = (struct face *) 0x0
        f = (struct frame *) 0x852b778
        glyph = (struct glyph *) 0x8b9d6d0
        char2b = (XChar2b *) 0xbfffe224
        two_byte_p = (int *) 0xbfffe1f8
        two_byte_p = 0
        start = -1073749388
        end = -1073749468
        glyph = (struct glyph *) 0x8b9d6d0
        last = (struct glyph *) 0x8b9d9a0
        voffset = 0
        glyph_not_available_p = 0
#1  0x08078145 in draw_glyphs (w=0x852b9e8, x=13, row=0x8cbbbf0, 
    area=TEXT_AREA, start=0, end=36, hl=DRAW_NORMAL_TEXT, overlaps_p=0)
    at xdisp.c:17306
        head = (struct glyph_string *) 0xbfffe274
        tail = (struct glyph_string *) 0xbfffe274
        s = (struct glyph_string *) 0xbfffe274
        last_x = 973
        area_width = 12
        x_reached = 147569648
        i = 0
        j = 134654231
        f = (struct frame *) 0x852b778
#2  0x0807c37e in x_write_glyphs (start=0x8b9d6d0, len=36) at xdisp.c:18366
        start = (struct glyph *) 0xc
        x = 147569648
#3  0x08053f05 in update_text_area (w=0x852b9e8, vpos=0) at dispnew.c:4266
        current_row = (struct glyph_row *) 0x8d33e88
        desired_row = (struct glyph_row *) 0x8cbbbf0
        changed_p = 0
#4  0x08054346 in update_window_line (w=0x852b9e8, vpos=0, 
    mouse_face_overwritten_p=0xbfffe528) at dispnew.c:4490
        w = (struct window *) 0x852b9e8
        mouse_face_overwritten_p = (int *) 0x0
        current_row = (struct glyph_row *) 0x8d33e88
        desired_row = (struct glyph_row *) 0x8cbbbf0
        changed_p = 0
#5  0x08053c8b in update_window (w=0x852b9e8, force_p=0) at dispnew.c:4146
        vpos = 0
        i = 12
        end = (struct glyph_row *) 0x8cbbe34
        mode_line_row = (struct glyph_row *) 0xbfffe274
        header_line_row = (struct glyph_row *) 0x0
        changed_p = 1
        mouse_face_overwritten_p = 0
        row = (struct glyph_row *) 0x8cbbbf0
        yb = 24
        n_updated = 1
        desired_matrix = (struct glyph_matrix *) 0x8ce7198
        paused_p = 0
        preempt_count = 9
#6  0x080536f3 in update_window_tree (w=0x8b15500, force_p=0) at dispnew.c:3881
        w = (struct window *) 0x852b9e8
        force_p = 0
        paused_p = 0
#7  0x080535f1 in update_frame (f=0x852b778, force_p=0, inhibit_hairy_id_p=0)
    at dispnew.c:3818
        f = (struct frame *) 0x852b778
        inhibit_hairy_id_p = 0
        paused_p = 139681752
        root_window = (struct window *) 0x8b15500
#8  0x0806c6ba in redisplay_internal (preserve_echo_area=0) at xdisp.c:10005
        f = (struct frame *) 0x852b778
        tail = 12
        frame = 12
        i = 1
        updated = (struct frame **) 0xbfffe5b4
        n = 0
        size = 50
        w = (struct window *) 0x8b15500
        f = (struct frame *) 0xbfffe274
        pause = 0
        must_finish = 1
        tlbufpos = {charpos = 4011, bytepos = 4011}
        tlendpos = {charpos = 0, bytepos = 0}
        number_of_visible_frames = 1
        count = 20
        polling_stopped_here = 1
        consider_all_windows_p = 1
#9  0x0806b5d0 in redisplay () at xdisp.c:9403
No locals.
#10 0x080dccc5 in read_char (commandflag=0, nmaps=0, maps=0x0, 
    prev_event=405511988, used_mouse_menu=0x0) at keyboard.c:2483
        c = 405511940
        count = -1073746704
        local_getcjmp = {{__jmpbuf = {134645296, 0, 139638648, 952765028, 
      1213380472, 674167796}, __mask_was_saved = 139638648, __saved_mask = {
      __val = {952765028, 3221220412, 134640282, 1, 3221220596, 3221220412, 
        134640322, 139638648, 952765028, 35, 0, 952765028, 0, 35, 0, 0, 26, 
        3221220476, 134639811, 952765028, 35, 0, 0, 952765028, 946250844, 
        946250844, 35, 148514108, 148514119, 3221220196, 46, 1}}}}
        save_jump = {{__jmpbuf = {1075908504, 1076493764, 1, 0, 134560441, 
      139638648}, __mask_was_saved = 139639272, __saved_mask = {__val = {
        3221220236, 1075813336, 139477312, 400000, 146395792, 1075813297, 
        139639272, 139638648, 3221220252, 134900369, 139638648, 0, 3221220284, 
        134960887, 139477312, 1, 3221220300, 134559609, 139639272, 405511940, 
        0, 134643057, 139639272, 139638648, 3221220348, 134646598, 139638648, 
        20, 3221220348, 134646584, 139639272, 1}}}}
        key_already_recorded = 0
        tem = 946250844
        save = -1073746704
        previous_echo_area_message = 405511940
        also_record = 405511940
        reread = 0
        gcpro1 = {next = 0x0, var = 0x0, nvars = -1073747108}
        gcpro2 = {next = 0x0, var = 0x4029d77c, nvars = 1077986624}
        last_idle_start = {tv_sec = 139478904, tv_usec = 248}
        polling_stopped_here = 0
#11 0x08146952 in read_filtered_event (no_switch_frame=1, ascii_required=0, 
    error_nonascii=0, input_method=0) at lread.c:465
        val = -1073746704
        delayed_switch_frame = 405511940
#12 0x0813d874 in Fy_or_n_p (prompt=946250844) at fns.c:3068
        prompt = 946250844
        obj = -1073746704
        key = 12
        def = -1073746704
        map = 1480701972
        answer = -1073746704
        xprompt = 946250844
        args = {952765044, 946250844}
        gcpro1 = {next = 0x0, var = 0x0, nvars = 19}
        gcpro2 = {next = 0x182d9174, var = 0x85d498c, nvars = 140598120}
        count = 19

...

Re: crash with Emacs CVS 2003-05-09

[Kenichi Handa]

In article <20030625.080324.112616176.wl@gnu.org>, Werner LEMBERG <wl@gnu.org> writes:
> [I'm using this old CVS version because the current one has severe
> problems with fontsets.]

Could you explain more about that fontset problem?

> While working with mew I got the segfault below.

Do you mean that this segfault is because of that fontset
problem?

---
Ken'ichi HANDA
handa@m17n.org

Re: crash with Emacs CVS 2003-05-09

[Werner LEMBERG]

> > [I'm using this old CVS version because the current one has severe
> > problems with fontsets.]
> 
> Could you explain more about that fontset problem?

It has been reported e.g. yesterday: After start-up, selecting any
fontset doesn't work; it always stays at 16-dot.  Jan (or Kim?) is
aware of this problem and promised a patch this week...

> Do you mean that this segfault is because of that fontset
> problem?

I don't think so because the big change was after 2003-05-09.  Maybe
the segfault doesn't make sense today.


    Werner

Re: crash with Emacs CVS 2003-05-09

[Kenichi Handa]

In article <20030625.084156.112301006.wl@gnu.org>, Werner LEMBERG <wl@gnu.org> writes:
>>  > [I'm using this old CVS version because the current one has severe
>>  > problems with fontsets.]
>>  
>>  Could you explain more about that fontset problem?

> It has been reported e.g. yesterday: After start-up, selecting any
> fontset doesn't work; it always stays at 16-dot.  Jan (or Kim?) is
> aware of this problem and promised a patch this week...

Ah!  I see.

---
Ken'ichi HANDA
handa@m17n.org

Re: crash with Emacs CVS 2003-05-09

[Werner LEMBERG]

> Can you say a way to reproduce this?

No, but the Emacs process is still running in gdb, thus I could do
some further inspection if you tell me what I shall look at.


    Werner

Re: crash with Emacs CVS 2003-05-09

[Richard Stallman]

    No, but the Emacs process is still running in gdb, thus I could do
    some further inspection if you tell me what I shall look at.

I don't know this code any better than you do.  The thing to do
is to study the code in the innermost function, fill_glyph_string,
and figure out what data values were used by the code that crashed,
and then keep working backwards as far as you can.