Re: [fluxbox-users] QT applicacion dies on plain fluxbox, but not on fluxbox on top of LXDE

On Wed, Nov 11, 2015 at 4:41 PM, Nable <[email protected]> wrote:
> Ok, let the fun go.
> What version of hpmyroom are you using?
> Does this one look familiar?
> $ wc -c hpmyroom_deb
> 46537605 hpmyroom_deb
> $ md5sum hpmyroom_deb
> 4f2e520894f453a533f81d1163ddfa85  hpmyroom_deb
>
> It was extracted from hpmyroom_v10.4.0.0168_i386.deb, thus the "_deb" suffix.
> It's a 32-bit version without even -fomit-frame-pointer, so debugging
> seems to be easy:
>
>  832e2c6:       8d 83 e8 ad e7 fe       lea    eax,[ebx-0x1185218]
>  832e2cc:       89 11                   mov    DWORD PTR [ecx],edx
>  832e2ce:       c7 41 28 00 00 00 00    mov    DWORD PTR [ecx+0x28],0x0
>  832e2d5:       c7 41 2c 00 00 00 00    mov    DWORD PTR [ecx+0x2c],0x0
>  832e2dc:       89 04 24                mov    DWORD PTR [esp],eax
>  832e2df:       e8 3c 87 d2 ff          call   8056a20 <getenv@plt>
>  832e2e4:       b9 03 00 00 00          mov    ecx,0x3
>  832e2e9:       89 c6                   mov    esi,eax
>  832e2eb:       f3 a6                   repz cmpsb ; <<< here it fails
> in your case
>
> As you can see, this application tries to get the value of some
> environment variable and then compare it with some string (i'm a bit
> in a hurry now to find the exact values by myself). But if this
> environment variable is missing, getenv returns NULL and application
> fails in a bad way. Now you can find exact values of interesting
> strings:
> $ gdb hpmyroom
> tb *0x832e2df
> run
> print (char*)$eax
> print (char*)$edi
>
> First print gives you the variable name, second one shows some string.
> What are these strings?


First, thanks a lot for the help...

I'm using the 64-bit deb package.  And version 10.4.0.0168.  I believe
the binary is the same 32-bit though...

Now to the strings when not using any DE, but just plain fluxbox,
which cause hpmyroom to die:

Temporary breakpoint 1, 0x0832e2df in ?? ()
(gdb) print (char*)$eax
$1 = 0x960dde8 "DESKTOP_SESSION"
(gdb) print (char*)$edi
$2 = 0x960ddf8 "kde

Of course, when using plain fluxbox, DESKTOP_SESSION environment
variable is not defined:

% echo $DESKTOP_SESSION
%

When using fluxbox on top of LXDE, that environment variable is defined:

% echo $DESKTOP_SESSION
LXDE
%

However hpmyroom stills prints "kde" on $edi:

Temporary breakpoint 1, 0x0832e2df in ?? ()
(gdb) print (char*)$eax
$1 = 0x960dde8 "DESKTOP_SESSION"
(gdb) print (char*)$edi
$2 = 0x960ddf8 "kde"

Which surprises me.  I was expecting LXDE there, but I was wrong.  And
with fluxbox on top of LXDE hpmyroom does not die...

I tried then setting DESKTOP_SESSION to NONE, on a non DE plain fluxbox session:

% echo $DESKTOP_SESSION
NONE
%

And then hpmyroom still prints "kde" on $edi (as expected from the LXDE trial):

Temporary breakpoint 1, 0x0832e2df in ?? ()
(gdb) print (char*)$eax
$1 = 0x960dde8 "DESKTOP_SESSION"
(gdb) print (char*)$edi
$2 = 0x960ddf8 "kde"

However now hpmyroom doesn't die.  Or at least it doesn't die where it
was before.

What value would you suggest setting for DESKTOP_SESSION on a non DE
plain fluxbox session?  Is there actually a standard for that?

What would be your suggestions be (besides providing the feedback
about this weird dependency)?  Specially to define a work around
meanwhile (would NONE be OK?).

Thanks,

-- 
Javier

------------------------------------------------------------------------------