gdb on arm-wince-pe

gdb on arm-wince-pe

[Pedro Alves]

Hi all!

I have gdb built for target arm-wince-pe on a cygwin host.
This if for I new project I am developing to bring gnu tools for 
arm-wince-pe up to speed.
Check out cegcc if you're interested. 
(http://developer.berlios.de/projects/cegcc/)
I am now looking into gdb to implement a few missing things and trying 
to remove a few quirks here and there.

I have been looking through the code for a few hours to get a grasp on 
gdb internals.
With some patches the gdb/wince-stub.c is now working ok. Not perfect 
yet, but working.
I now want to implement CTRL_c breaking into the remote app.
It currently does nothing when the target program is being debugged, and 
prints "Quit" when there is no program being debugged.

Where should I start looking for/what is needed to implement it (ctrl_c 
breaking)?
I tried debugging arm-wince-pe-gdb on gdb itself, putting some 
breakpoints in the signal handling,
and issuing kill -INT signals to see what happens there, and I see that 
ultimatly, quit (void) is called.
Is there some target_op that I should implement, or should I provide my 
own signal handling?

Cheers,
Pedro Alves

P.S. Please add me to the cc list, as I am not (yet) on this mailing list.

Re: gdb on arm-wince-pe

[Daniel Jacobowitz]

On Tue, Feb 21, 2006 at 08:41:48PM +0000, Pedro Alves wrote:
> Where should I start looking for/what is needed to implement it (ctrl_c 
> breaking)?
> I tried debugging arm-wince-pe-gdb on gdb itself, putting some 
> breakpoints in the signal handling,
> and issuing kill -INT signals to see what happens there, and I see that 
> ultimatly, quit (void) is called.
> Is there some target_op that I should implement, or should I provide my 
> own signal handling?

Take a look at remote.c, which installs its own signal handler.  You'll
want to do something similar.

Really, if possible, I would recommend using remote.c to talk to your
target over the standard GDB remote protocol instead of the
WinCE-specific protocol that wince-stub.c implements.  That's much
better supported.

-- 
Daniel Jacobowitz
CodeSourcery

Re: gdb on arm-wince-pe

[Christopher Faylor]

On Tue, Feb 21, 2006 at 11:40:09PM -0500, Daniel Jacobowitz wrote:
>On Tue, Feb 21, 2006 at 08:41:48PM +0000, Pedro Alves wrote:
>> Where should I start looking for/what is needed to implement it (ctrl_c 
>> breaking)?
>> I tried debugging arm-wince-pe-gdb on gdb itself, putting some 
>> breakpoints in the signal handling,
>> and issuing kill -INT signals to see what happens there, and I see that 
>> ultimatly, quit (void) is called.
>> Is there some target_op that I should implement, or should I provide my 
>> own signal handling?
>
>Take a look at remote.c, which installs its own signal handler.  You'll
>want to do something similar.
>
>Really, if possible, I would recommend using remote.c to talk to your
>target over the standard GDB remote protocol instead of the
>WinCE-specific protocol that wince-stub.c implements.  That's much
>better supported.

As the author of the WinCE-specific protocol, I agree 100%.  Really, I
think that the WinCE targets should be deprecated since they no longer
have a maintainer.

cgf

Re: gdb on arm-wince-pe

[Pedro Alves]

Christopher Faylor wrote:

>On Tue, Feb 21, 2006 at 11:40:09PM -0500, Daniel Jacobowitz wrote:
>  
>
>>On Tue, Feb 21, 2006 at 08:41:48PM +0000, Pedro Alves wrote:
>>    
>>
>>>Where should I start looking for/what is needed to implement it (ctrl_c 
>>>breaking)?
>>>I tried debugging arm-wince-pe-gdb on gdb itself, putting some 
>>>breakpoints in the signal handling,
>>>and issuing kill -INT signals to see what happens there, and I see that 
>>>ultimatly, quit (void) is called.
>>>Is there some target_op that I should implement, or should I provide my 
>>>own signal handling?
>>>      
>>>
>>Take a look at remote.c, which installs its own signal handler.  You'll
>>want to do something similar.
>>
>>Really, if possible, I would recommend using remote.c to talk to your
>>target over the standard GDB remote protocol instead of the
>>WinCE-specific protocol that wince-stub.c implements.  That's much
>>better supported.
>>    
>>
>
>As the author of the WinCE-specific protocol, I agree 100%.  Really, I
>think that the WinCE targets should be deprecated since they no longer
>have a maintainer.
>
>cgf
>
>  
>
Ok, thanks for the quick replies. I'll see what I can do.

Cheers,
Pedro Alves