[ECOS] GDB remote debug problem ??

[ECOS] GDB remote debug problem ??

[林志勝]

Hi everyone:

    I am porting eCos on my custom MPC823 board.
Redboot can boot from flash and work fine on my board. 
I want use GDB to debug hello example step by step.
If I uncheck the set any breakpoint at target settings.. ; 
The console can see"Hello eCos world " output. 
But if I check set breakpoint at main, the console will show
$T05thread:00000002;40:000204ac;01:00031af8;#b4
and GDB on my NT can't control my borad any more.
What problem may I meet?? Is redboot problem, hello,
or GDB problem??

Thank you very much!!

Ken

Re: [ECOS] GDB remote debug problem ??

[Jonathan Larmour]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1059 bytes --]

ªL§Ó³Ó wrote:
> 
> Hi everyone:
> 
>     I am porting eCos on my custom MPC823 board.
> Redboot can boot from flash and work fine on my board.
> I want use GDB to debug hello example step by step.
> If I uncheck the set any breakpoint at target settings.. ;
> The console can see"Hello eCos world " output.
> But if I check set breakpoint at main, the console will show
> $T05thread:00000002;40:000204ac;01:00031af8;#b4
> and GDB on my NT can't control my borad any more.
> What problem may I meet?? Is redboot problem, hello,
> or GDB problem??

That $T packet indicates that you have hit the breakpoint in main. But
unless you have enabled remote protocol debugging within GDB, you shouldn't
have seen that - it's part of the underlying protocol. And then of course
you could be able to continue the program from the breakpoint by typing
"continue" - I assume this isn't happening?

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Re: [ECOS] GDB remote debug problem ??

[???]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2010 bytes --]

----- Original Message -----
From: "Jonathan Larmour" <jlarmour@redhat.com>
To: "ªL§Ó³Ó" <jclin@itri.org.tw>
Cc: <ecos-discuss@sources.redhat.com>
Sent: Saturday, March 31, 2001 12:36 AM
Subject: Re: [ECOS] GDB remote debug problem ??


ªL§Ó³Ó wrote:
>
> Hi everyone:
>
>     I am porting eCos on my custom MPC823 board.
> Redboot can boot from flash and work fine on my board.
> I want use GDB to debug hello example step by step.
> If I uncheck the set any breakpoint at target settings.. ;
> The console can see"Hello eCos world " output.
> But if I check set breakpoint at main, the console will show
> $T05thread:00000002;40:000204ac;01:00031af8;#b4
> and GDB on my NT can't control my borad any more.
> What problem may I meet?? Is redboot problem, hello,
> or GDB problem??

That $T packet indicates that you have hit the breakpoint in main. But
unless you have enabled remote protocol debugging within GDB, you shouldn't
have seen that - it's part of the underlying protocol. And then of course
you could be able to continue the program from the breakpoint by typing
"continue" - I assume this isn't happening?

Jifl
--
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Hi :
    Thank you for your reply!! I describe my steps
1. Reset my borad by pressing the reset buttom.
2. Run powerpc-eabi-gdb
3. press the 'RUN' buttom
4. select the hello.exe I want to debug
5. select target Remote TCP & set breakpoint at main
6. download OK & I see my serial console print $T ...
7. I can't do anything but press the stop buttom on powerpc-eabi-gdb
8. powerpc-eabi-gdb get no response from from by board & dettach from it.

But if step 5. I select Remote serial , then everything will be OK. I can
stop
the hello program at main and run step by step.
What problem may it happen?? Is my ethernet driver's problem?? or something
??
What point may I check??

Thanks very much!!

Ken

Re: [ECOS] GDB remote debug problem ??

[Jonathan Larmour]

??? wrote:
>     Thank you for your reply!! I describe my steps
> 1. Reset my borad by pressing the reset buttom.
> 2. Run powerpc-eabi-gdb
> 3. press the 'RUN' buttom
> 4. select the hello.exe I want to debug
> 5. select target Remote TCP & set breakpoint at main
> 6. download OK & I see my serial console print $T ...
> 7. I can't do anything but press the stop buttom on powerpc-eabi-gdb
> 8. powerpc-eabi-gdb get no response from from by board & dettach from it.
> 
> But if step 5. I select Remote serial , then everything will be OK. I can
> stop
> the hello program at main and run step by step.
> What problem may it happen?? Is my ethernet driver's problem?? or something
> ??
> What point may I check??

Ah I think I understand what you are describing - you are connecting via
TCP to RedBoot. But then when the board hits a breakpoint, this breakpoint
is being reported over the *serial* line, not the TCP line.

In that case, firstly make sure you are using an up-to-date anonymous CVS
repository. Secondly, in your downloaded application's configuration, have
you done anything like enable included GDB stubs, or claimed virtual
vectors?

Lastly in your port, have you done anything in the hal_diag init that would
cause it to claim the debug channel? Perhaps compare your diag init with
that of another known working target, e.g. look at
hal/arm/sa11x0/var/current/src/hal_diag.c

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Re: [ECOS] GDB remote debug problem ??

[???]

 > Ah I think I understand what you are describing - you are connecting via
> TCP to RedBoot. But then when the board hits a breakpoint, this breakpoint
> is being reported over the *serial* line, not the TCP line.
I think so. But how does it change the channel line??

> In that case, firstly make sure you are using an up-to-date anonymous CVS
> repository. Secondly, in your downloaded application's configuration, have
> you done anything like enable included GDB stubs, or claimed virtual
> vectors?
My hello application use default template, no GDB stubs included, and check
claim comms virtual vectors. I think I shouldn't include GDB stubs,but I
don't
know whether should I check the claim comms virtural vectors??

> Lastly in your port, have you done anything in the hal_diag init that
would
> cause it to claim the debug channel? Perhaps compare your diag init with
> that of another known working target, e.g. look at
> hal/arm/sa11x0/var/current/src/hal_diag.c
I copy my target from POWERPC/MBX, and I don't change any code in
hal_diag.c.
But I will check with the hal/arm/sa11x0/var/current/src/hal_diag.c


> Jifl
> --
> Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
> Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Thanks again !!

Ken

Re: [ECOS] GDB remote debug problem ??

[Jonathan Larmour]

??? wrote:
> 
> > In that case, firstly make sure you are using an up-to-date anonymous CVS
> > repository. Secondly, in your downloaded application's configuration, have
> > you done anything like enable included GDB stubs, or claimed virtual
> > vectors?
> My hello application use default template, no GDB stubs included, and check
> claim comms virtual vectors. I think I shouldn't include GDB stubs,but I
> don't
> know whether should I check the claim comms virtural vectors??

I believe it should work if you don't enable that option.

Jifl
-- 
Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

Re: [ECOS] GDB remote debug problem ??

[???]

> >
> > > In that case, firstly make sure you are using an up-to-date anonymous
CVS
> > > repository. Secondly, in your downloaded application's configuration,
have
> > > you done anything like enable included GDB stubs, or claimed virtual
> > > vectors?
> > My hello application use default template, no GDB stubs included, and
check
> > claim comms virtual vectors. I think I shouldn't include GDB stubs,but I
> > don't
> > know whether should I check the claim comms virtural vectors??
>
> I believe it should work if you don't enable that option.
>
> Jifl
> --
> Red Hat, Rustat House, Clifton Road, Cambridge, UK. Tel: +44 (1223) 271062
> Maybe this world is another planet's Hell -Aldous Huxley || Opinions==mine

It work!!!
Thank you thank you very much!!!

Ken