nanogui: Thread: Segmentation Fault

Hi Microwindows users!

I run a application with Microwindows Win API on a x86 framebuffer device. 
When my programm stopps with a segmentation fault, I see on this tty where i 
have started the application the last screen of which was drawed.

To get the hanging tty runing i have now to login with telnet and have to make 
a "chvt" to another terminal an then press a key. Then I can switch back to 
the tty where i have started my application. Then I can see the Segmentation 
fault message.

Is there a possibility to kill or reset the framebuffer. That i have not 
always to do a chvt ? When I start my application with a script maybe it is 
possible to do a small programm after my apllication hase stoped to reset the 
screen.

Thanks, for your help    
-- 
mfg
 Manfred Gruber    

On Fri, 2003-05-09 at 08:44, Manfred Gruber wrote:
> To get the hanging tty runing i have now to login with telnet and have to make 
> a "chvt" to another terminal an then press a key. Then I can switch back to 
> the tty where i have started my application. Then I can see the Segmentation 
> fault message.
> Is there a possibility to kill or reset the framebuffer. That i have not 
> always to do a chvt ? When I start my application with a script maybe it is 
> possible to do a small programm after my apllication hase stoped to reset the 
> screen.

Couldn't you just put the two chvt commands into the script after the
line which starts your app?

-- 
------------ Alex Holden - http://www.linuxhacker.org ------------
If it doesn't work, you're not hitting it with a big enough hammer

Am Freitag, 9. Mai 2003 10:34 schrieb Alex Holden:
> On Fri, 2003-05-09 at 08:44, Manfred Gruber wrote:
> > To get the hanging tty runing i have now to login with telnet and have to
> > make a "chvt" to another terminal an then press a key. Then I can switch
> > back to the tty where i have started my application. Then I can see the
> > Segmentation fault message.
> > Is there a possibility to kill or reset the framebuffer. That i have not
> > always to do a chvt ? When I start my application with a script maybe it
> > is possible to do a small programm after my apllication hase stoped to
> > reset the screen.
>
> Couldn't you just put the two chvt commands into the script after the
> line which starts your app?

i have tried this and it works but the problem is that i have to press a key
after a chvt to update the screen and we have a device without a keyboard
only with touch-screen. so for testing it is possible to plug in keyboard,
but when the board is build in the engine there is no keyboard. so i have to
automaticaly reset the screen. to see what happend, it is also not possible
for me to do a logout after the application hangs because there are some logs
which are displayed and on an error i have to see them.

but thanks for help
--
mfg
 Manfred Gruber

On Fri, 2003-05-09 at 10:09, Manfred Gruber wrote:
> i have tried this and it works but the problem is that i have to press a key
> after a chvt to update the screen and we have a device without a keyboard

Is the problem maybe that the screen blanker has activated? If so
perhaps you could disable that with "setterm -blank 0".

-- 
------------ Alex Holden - http://www.linuxhacker.org ------------
If it doesn't work, you're not hitting it with a big enough hammer

Am Freitag, 9. Mai 2003 11:18 schrieben Sie:
> On Fri, 2003-05-09 at 10:09, Manfred Gruber wrote:
> > i have tried this and it works but the problem is that i have to press a
> > key after a chvt to update the screen and we have a device without a
> > keyboard
>
> Is the problem maybe that the screen blanker has activated? If so
> perhaps you could disable that with "setterm -blank 0".

No sorry, that is it not. 
When my application hangs by a segmentation fault. I make after that a chvt 2, 
then a chvt 1 (there my application runs). then I see the hanging screen from 
microwindows. then i press a key and then i can see the logs an the 
Segmentation Fault message and the normal terminal is there...  

maybe there is a icotl or something which i can do to reset the terminal or 
the framebuffer device...
-- 
mfg
 Manfred Gruber    

> To get the hanging tty runing i have now to login with telnet and have to make
a "chvt" to another terminal an then press a key. Then I can switch back to
the tty where i have started my application. Then I can see the Segmentation
fault message.

I normally remote login and run "kbd_mode -a", which resets
the keyboard.  Then I use the ALT-Fx keys on the console
to switch to the desired VT.

> Is there a possibility to kill or reset the framebuffer. That i have not
always to do a chvt ? When I start my application with a script maybe it is
possible to do a small programm after my apllication hase stoped to reset the
screen.

Look at drivers/scr_fb.c and drivers/vtswitch.c at all the ioctl calls,
 and you will see how linux console can be reset in various ways.

Regards,

Greg

hi
what is segmentation fault in viewml and how to solve
it 
it is giving this error while loading some url
thanks

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

: what is segmentation fault in viewml and how to solve
: it 
: it is giving this error while loading some url

As I mentioned, the best way to debug this is to run viewml
on the desktop (not target), from a remote terminal,
with nano-X running on the console.  Login, run
gdb viewml, and start the process.  Then, when it
segfaults, execute a "bt" stack backtrace and you'll
find what it was doing when it crashed.

Regards,

Greg

the following code is generating Segmentation fault at the nano-x server,
please help.
The Fault occurs at GrReadArea(...)

        size_t nDataSize = 0;

        GrGetWindowInfo(GR_ROOT_WINDOW_ID, &info);

        nDataSize = sizeof(GR_PIXELVAL) * info.width * info.height;

        data = reinterpret_cast<GR_PIXELVAL*>(new char[nDataSize]);

        if (data == NULL)
        {
                cout << "failed to malloc mem for buffer\n";
        }

        // Fault occurs here
        GrReadArea(GR_ROOT_WINDOW_ID, info.x, info.y, info.width,
info.height, data);

-- 
Rajan Singh Raghuvanshi

My best guess would be to set nDataSize to a large number,
and see whether that changes anything.  You might
try tracking down the code in nanox/srvnet.c that returns
the server data back to the client, as well as the nanox/client.c
code on the client side.

Regards,

Greg



The Fault occurs at GrReadArea(...)

        size_t nDataSize = 0;

        GrGetWindowInfo(GR_ROOT_WINDOW_ID, &info);

        nDataSize = sizeof(GR_PIXELVAL) * info.width * info.height;

        data = reinterpret_cast<GR_PIXELVAL*>(new char[nDataSize]);

        if (data == NULL)
        {
                cout << "failed to malloc mem for buffer\n";
        }

        // Fault occurs here
        GrReadArea(GR_ROOT_WINDOW_ID, info.x, info.y, info.width,
info.height, data);

-- 
Rajan Singh Raghuvanshi