[ptxdist] Bad ncurses configuration for mconf?

[ptxdist] Bad ncurses configuration for mconf?

[Christian Melki]

Ptxdist seems to assume things about the ncurses local build that 
doesn't always turn up to be usable.

I use a ncurses-6.2 build in Slackware. Pretty normal build.

mconf headers point to the ncurses library, not ncursesw.

mconf gets built with:
     -O2 -D_DEFAULT_SOURCE -D_XOPEN_SOURCE=600
and is linked with:
      linux-vdso.so.1 (0x00007ffcd5f30000)
     libncurses.so.6 => /lib64/libncurses.so.6 (0x00007f6914c68000)
     libtinfo.so.6 => /lib64/libtinfo.so.6 (0x00007f6914c37000)
     libc.so.6 => /lib64/libc.so.6 (0x00007f6914a52000)
     libdl.so.2 => /lib64/libdl.so.2 (0x00007f6914a4d000)
     /lib64/ld-linux-x86-64.so.2 (0x00007f6914cbd000)

And mconf segfaults with:
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7f84b9a in waddch () from /lib64/libncurses.so.6

(gdb) bt
#0  0x00007ffff7f84b9a in waddch () from /lib64/libncurses.so.6
#1  0x0000000000415c78 in attr_clear (win=0xcdf820, height=49, 
width=299, attr=<optimized out>) at lxdialog/util.c:250
#2  0x0000000000415cfc in dialog_clear () at lxdialog/util.c:262
#3  0x0000000000416274 in init_dialog (backtitle=backtitle@entry=0x0) at 
lxdialog/util.c:336
#4  0x000000000040272c in main (ac=2, av=0x7ffffffefd78) at mconf.c:1017

... second loop pass. Height = 0 passes the width loop. Height = 1, 
Width = 0 segfaults.

(gdb) print j
$8 = 298
(gdb) n
250                waddch(win, ' ');
(gdb) n
249            for (j = 0; j < width; j++)
(gdb) n
247        for (i = 0; i < height; i++) {
(gdb) n
248            wmove(win, i, 0);
(gdb) print i
$9 = 1
(gdb) n
249            for (j = 0; j < width; j++)
(gdb) n
250                waddch(win, ' ');
(gdb) n
Program received signal SIGSEGV, Segmentation fault.

I am not sure what to blame here or if I understand things.
Or if it's resonable that ncurses should explode like this.

Seems pkg-config cflags answers:
pkg-config --cflags ncurses -D_DEFAULT_SOURCE -D_XOPEN_SOURCE=600

But if I understand things >= 600 = wide char header?
the pkg-config answers the same for ncursesw.

So the program will compile and link just fine.
If I use xterm and rxvt as $TERM the mconf program crashes, but is fine 
with vt100.

Is this related to ext-colors support somehow?
If I force the program to link with ncursesw, then all is fine.

My best guess is that the terminal is ext-color, program gets built with 
wchar-support but linked with ncurses.
This probably shouldn't explode in ncurses, but I think it's also a 
misconfiguration build.

I would suggest that ptxdist uses ncursesw + headers if found and 
ncurses without wchar if not found.

Kernel seems to prefer something similar?
https://elixir.bootlin.com/linux/v5.10.2/source/scripts/kconfig/mconf-cfg.sh

Merry Christmas!

_______________________________________________
ptxdist mailing list
ptxdist@pengutronix.de
To unsubscribe, send a mail with subject "unsubscribe" to ptxdist-request@pengutronix.de

Re: [ptxdist] Bad ncurses configuration for mconf?

[Michael Olbrich]

On Fri, Dec 25, 2020 at 09:08:14PM +0100, Christian Melki wrote:
> Ptxdist seems to assume things about the ncurses local build that doesn't
> always turn up to be usable.
> 
> I use a ncurses-6.2 build in Slackware. Pretty normal build.
> 
> mconf headers point to the ncurses library, not ncursesw.
> 
> mconf gets built with:
>     -O2 -D_DEFAULT_SOURCE -D_XOPEN_SOURCE=600
> and is linked with:
>      linux-vdso.so.1 (0x00007ffcd5f30000)
>     libncurses.so.6 => /lib64/libncurses.so.6 (0x00007f6914c68000)
>     libtinfo.so.6 => /lib64/libtinfo.so.6 (0x00007f6914c37000)
>     libc.so.6 => /lib64/libc.so.6 (0x00007f6914a52000)
>     libdl.so.2 => /lib64/libdl.so.2 (0x00007f6914a4d000)
>     /lib64/ld-linux-x86-64.so.2 (0x00007f6914cbd000)
> 
> And mconf segfaults with:
> Program received signal SIGSEGV, Segmentation fault.
> 0x00007ffff7f84b9a in waddch () from /lib64/libncurses.so.6
> 
> (gdb) bt
> #0  0x00007ffff7f84b9a in waddch () from /lib64/libncurses.so.6
> #1  0x0000000000415c78 in attr_clear (win=0xcdf820, height=49, width=299,
> attr=<optimized out>) at lxdialog/util.c:250
> #2  0x0000000000415cfc in dialog_clear () at lxdialog/util.c:262
> #3  0x0000000000416274 in init_dialog (backtitle=backtitle@entry=0x0) at
> lxdialog/util.c:336
> #4  0x000000000040272c in main (ac=2, av=0x7ffffffefd78) at mconf.c:1017
> 
> ... second loop pass. Height = 0 passes the width loop. Height = 1, Width =
> 0 segfaults.
> 
> (gdb) print j
> $8 = 298
> (gdb) n
> 250                waddch(win, ' ');
> (gdb) n
> 249            for (j = 0; j < width; j++)
> (gdb) n
> 247        for (i = 0; i < height; i++) {
> (gdb) n
> 248            wmove(win, i, 0);
> (gdb) print i
> $9 = 1
> (gdb) n
> 249            for (j = 0; j < width; j++)
> (gdb) n
> 250                waddch(win, ' ');
> (gdb) n
> Program received signal SIGSEGV, Segmentation fault.
> 
> I am not sure what to blame here or if I understand things.
> Or if it's resonable that ncurses should explode like this.
> 
> Seems pkg-config cflags answers:
> pkg-config --cflags ncurses -D_DEFAULT_SOURCE -D_XOPEN_SOURCE=600
> 
> But if I understand things >= 600 = wide char header?
> the pkg-config answers the same for ncursesw.

I see the same pkg-config results here on Debian.
I have no idea how this is supposed to work.

> So the program will compile and link just fine.
> If I use xterm and rxvt as $TERM the mconf program crashes, but is fine with
> vt100.
> 
> Is this related to ext-colors support somehow?
> If I force the program to link with ncursesw, then all is fine.

Hmmm, what happens if you add -DNCURSES_WIDECHAR=0 to the cflags and link
with ncurses? Does it work correctly now?

> My best guess is that the terminal is ext-color, program gets built with
> wchar-support but linked with ncurses.
> This probably shouldn't explode in ncurses, but I think it's also a
> misconfiguration build.
> 
> I would suggest that ptxdist uses ncursesw + headers if found and ncurses
> without wchar if not found.
> 
> Kernel seems to prefer something similar?
> https://elixir.bootlin.com/linux/v5.10.2/source/scripts/kconfig/mconf-cfg.sh

The whole ncurses handling in configure.ac should probably cleaned up and
simplified.

Michael


_______________________________________________
ptxdist mailing list
ptxdist@pengutronix.de
To unsubscribe, send a mail with subject "unsubscribe" to ptxdist-request@pengutronix.de

Re: [ptxdist] Bad ncurses configuration for mconf?

[Christian Melki]

Hi.

I've since reported the error to the maintainer of Ncurses and to the 
slackware community, but it seems the general understanding of ncurses 
packaging is low when I ask around.

Anyway. Here is an excerpt from parts of the dialog with the maintainer.
Might be useful for someone for reference.

<quote>
I've rebuilt with a 20201219 snapshot.
It does not fix the issue. Problem remains.
But a bit clearer.
Investigated the win->line allocation.
After initscr the value seems ok.

(gdb) print stdscr->_line[0]
$14 = {text = 0xce2a10, firstchar = 0, lastchar = 301, oldindex = 0}
(gdb) print stdscr->_line[1]
$15 = {text = 0xce2ed0, firstchar = 0, lastchar = 301, oldindex = 0}
(gdb) print stdscr->_line[2]
$16 = {text = 0xce3390, firstchar = 0, lastchar = 301, oldindex = 0}
(gdb) print stdscr->_line[3]
$17 = {text = 0xce3850, firstchar = 0, lastchar = 301, oldindex = 0}

Breakpoint 1, main (ac=2, av=0x7ffffffefb28) at mconf.c:1000
1000        signal(SIGINT, sig_handler);
(gdb) c
Continuing.

Breakpoint 3, init_dialog (backtitle=backtitle@entry=0x0) at 
lxdialog/util.c:319
319        initscr();        /* Init curses */
(gdb) c
Continuing.

Breakpoint 4, initscr () at ../ncurses/./base/lib_initscr.c:56
56        START_TRACE();
(gdb) c
Continuing.

Hardware watchpoint 5: stdscr->_line[1]->text

Old value = <unreadable>
New value = (chtype *) 0xce2ed0
_nc_setupscreen_sp (spp=0x7ffffffef938, slines=49, scolumns=302, 
output=0x7ffff7f236a0 <_IO_2_1_stdout_>, filtered=0, slk_format=0) at 
../ncurses/./base/lib_set_term.c:744
744        sp->_prescreen = FALSE;
(gdb) c
Continuing.

Hardware watchpoint 5: stdscr->_line[1]->text

Old value = (chtype *) 0xce2ed0
New value = (chtype *) 0x100ce2ed0
0x0000000000415c3f in attr_clear (win=0xce2690, height=49, 
width=width@entry=302, attr=2097408) at lxdialog/util.c:247
247        for (i = 0; i < height; i++) {
(gdb) n
248            wmove(win, i, 0);
(gdb) n
249            for (j = 0; j < width; j++)
(gdb) n
250                waddch(win, ' ');
(gdb) print i
$28 = 0
(gdb) print j
$29 = 0

And at the segfault...
(gdb) print win->_line[0]
$9 = {text = 0xce2a10, firstchar = 0, lastchar = 301, oldindex = 0}
(gdb) print win->_line[1]
$10 = {text = 0x100ce2ed0, firstchar = 0, lastchar = 301, oldindex = 0}
(gdb) print win->_line[2]
$11 = {text = 0xce3390, firstchar = 0, lastchar = 301, oldindex = 0}
(gdb) print win->_line[3]
$12 = {text = 0xce3850, firstchar = 0, lastchar = 301, oldindex = 0}

Seems the line text pointer is botched by something.
</quote>

So in summary a pointer gets corrupted with the same operand value with 
what seems to be a bitwise "or" macro operation.

On 1/8/21 9:46 AM, Michael Olbrich wrote:
> On Fri, Dec 25, 2020 at 09:08:14PM +0100, Christian Melki wrote:
>> Ptxdist seems to assume things about the ncurses local build that doesn't
>> always turn up to be usable.
>>
>> I use a ncurses-6.2 build in Slackware. Pretty normal build.
>>
>> mconf headers point to the ncurses library, not ncursesw.
>>
>> mconf gets built with:
>>      -O2 -D_DEFAULT_SOURCE -D_XOPEN_SOURCE=600
>> and is linked with:
>>       linux-vdso.so.1 (0x00007ffcd5f30000)
>>      libncurses.so.6 => /lib64/libncurses.so.6 (0x00007f6914c68000)
>>      libtinfo.so.6 => /lib64/libtinfo.so.6 (0x00007f6914c37000)
>>      libc.so.6 => /lib64/libc.so.6 (0x00007f6914a52000)
>>      libdl.so.2 => /lib64/libdl.so.2 (0x00007f6914a4d000)
>>      /lib64/ld-linux-x86-64.so.2 (0x00007f6914cbd000)
>>
>> And mconf segfaults with:
>> Program received signal SIGSEGV, Segmentation fault.
>> 0x00007ffff7f84b9a in waddch () from /lib64/libncurses.so.6
>>
>> (gdb) bt
>> #0  0x00007ffff7f84b9a in waddch () from /lib64/libncurses.so.6
>> #1  0x0000000000415c78 in attr_clear (win=0xcdf820, height=49, width=299,
>> attr=<optimized out>) at lxdialog/util.c:250
>> #2  0x0000000000415cfc in dialog_clear () at lxdialog/util.c:262
>> #3  0x0000000000416274 in init_dialog (backtitle=backtitle@entry=0x0) at
>> lxdialog/util.c:336
>> #4  0x000000000040272c in main (ac=2, av=0x7ffffffefd78) at mconf.c:1017
>>
>> ... second loop pass. Height = 0 passes the width loop. Height = 1, Width =
>> 0 segfaults.
>>
>> (gdb) print j
>> $8 = 298
>> (gdb) n
>> 250                waddch(win, ' ');
>> (gdb) n
>> 249            for (j = 0; j < width; j++)
>> (gdb) n
>> 247        for (i = 0; i < height; i++) {
>> (gdb) n
>> 248            wmove(win, i, 0);
>> (gdb) print i
>> $9 = 1
>> (gdb) n
>> 249            for (j = 0; j < width; j++)
>> (gdb) n
>> 250                waddch(win, ' ');
>> (gdb) n
>> Program received signal SIGSEGV, Segmentation fault.
>>
>> I am not sure what to blame here or if I understand things.
>> Or if it's resonable that ncurses should explode like this.
>>
>> Seems pkg-config cflags answers:
>> pkg-config --cflags ncurses -D_DEFAULT_SOURCE -D_XOPEN_SOURCE=600
>>
>> But if I understand things >= 600 = wide char header?
>> the pkg-config answers the same for ncursesw.
> 
> I see the same pkg-config results here on Debian.
> I have no idea how this is supposed to work.
> 
>> So the program will compile and link just fine.
>> If I use xterm and rxvt as $TERM the mconf program crashes, but is fine with
>> vt100.
>>
>> Is this related to ext-colors support somehow?
>> If I force the program to link with ncursesw, then all is fine.
> 
> Hmmm, what happens if you add -DNCURSES_WIDECHAR=0 to the cflags and link
> with ncurses? Does it work correctly now?

I think this is controlled through the headers with the XOPEN_SOURCE 
level unless specified otherwise (ifndef)?

In curses.h:
#ifndef NCURSES_WIDECHAR
#if defined(_XOPEN_SOURCE_EXTENDED) || (defined(_XOPEN_SOURCE) &&
(_XOPEN_SOURCE - 0 >= 500))
#define NCURSES_WIDECHAR 1
#else
#define NCURSES_WIDECHAR 0
#endif
#endif /* NCURSES_WIDECHAR */

But yes. Your suggestion works.
I have never needed widechar for ptxdist.

Anyway.
This is what you meant?
Could be useful for forcibly disable wide char extensions if you're 
linking with ncurses only.

--- configure.ac~	2020-12-16 16:42:25.000000000 +0100
+++ configure.ac	2021-01-08 16:50:08.440868020 +0100
@@ -68,7 +68,7 @@
  	[AC_MSG_ERROR([ncurses headers not found])])
  AC_SUBST(CURSES_LOC)

-CPPFLAGS="${NCURSES_CFLAGS}"
+CPPFLAGS="${NCURSES_CFLAGS} -DNCURSES_WIDECHAR=0"
  AC_ARG_WITH(ncurses, AS_HELP_STRING([--with-ncurses],[Include path to 
the ncurses headers]),
  	[
  		if test "x$withval" != "xyes"; then

>> My best guess is that the terminal is ext-color, program gets built with
>> wchar-support but linked with ncurses.
>> This probably shouldn't explode in ncurses, but I think it's also a
>> misconfiguration build.
>>
>> I would suggest that ptxdist uses ncursesw + headers if found and ncurses
>> without wchar if not found.
>>
>> Kernel seems to prefer something similar?
>> https://elixir.bootlin.com/linux/v5.10.2/source/scripts/kconfig/mconf-cfg.sh
> 
> The whole ncurses handling in configure.ac should probably cleaned up and
> simplified.
> 
> Michael
> 


_______________________________________________
ptxdist mailing list
ptxdist@pengutronix.de
To unsubscribe, send a mail with subject "unsubscribe" to ptxdist-request@pengutronix.de