[ECOS] GDB-stub in Redboot fails to handshake

[ECOS] GDB-stub in Redboot fails to handshake

[Les G. Miklosy]

Hello Gents: 

As you can see in this gdb log file, my Sparc target running Redboot fails to 
recognize ($#00) the simple query commands offered by my Linux host over a 
serial connection. The target application is simply hello-world with a linked 
library build from ECOS version 1.0.7. Redboot is a ROMRAM bootloader also 
built from version 1.0.7. I can hear you smacking your foreheads asking why 
use V 1.0.7 , I am using this version because a board support package was 
built around it.

I tried the execution of Redboot with and without a '$' to start the GDB-stub 
with no improvement. Any application I try to load/continue ends with a 
Signal Hangup.  A diff between ECOS version 2.0 and 1.0.7 shows many changes 
to the code particularly the part for initiating gdb-stub. Is this old 
version suspect? Does the solution demand an upgrade to 2.0 or shall I debug 
1.0.7?

Thanks for any guidance,
Les



w +$Hc-1#09
r -
w $Hc-1#09
r <Timeout: 2 seconds>
w $Hc-1#09
r $#00
w +
r $#00
w +
r $#00
w +
r $#00
w +
r <Timeout: 2 seconds>

... repeats


c load
w $X41000000,0:#73
r +$OK#9a

... load application hello-world


Sending packet: $p42#d6...[+]Ack
[$][0][0][0][0][0][0][2][0][#][8][2]Packet received: 00000020
Sending packet: $p43#d7...[+]Ack
[$][4][0][0][0][0][0][5][0][#][8][9]Packet received: 40000050

...

Sending packet: $C01#a4...[+]Ack
[$][T][0][1][4][4][:][4][0][0][1][8][c][1][c][;][0][e][:][0][0][0][0][0][0][0]
[0][;][#][1][0]Packet received: T0144:40018c1c;0e:00000000;

Program received signal SIGHUP, Hangup.
0x40018c1c in ?? ()
    ...

The program is running.  Exit anyway? (y or n)




-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] GDB-stub in Redboot fails to handshake

[Andrew Lunn]

On Sat, Mar 08, 2008 at 05:09:30AM -0800, Les G. Miklosy wrote:
> Hello Gents: 
> 
> As you can see in this gdb log file, my Sparc target running Redboot fails to 
> recognize ($#00) the simple query commands offered by my Linux host over a 
> serial connection. 

The redboot stub only implements a subset of the gdb protocol. 

> The target application is simply hello-world with a linked 
> library build from ECOS version 1.0.7. Redboot is a ROMRAM bootloader also 
> built from version 1.0.7. I can hear you smacking your foreheads asking why 
> use V 1.0.7 , I am using this version because a board support package was 
> built around it.
> 
> I tried the execution of Redboot with and without a '$' to start the GDB-stub 
> with no improvement. Any application I try to load/continue ends with a 
> Signal Hangup.  A diff between ECOS version 2.0 and 1.0.7 shows many changes 
> to the code particularly the part for initiating gdb-stub. Is this old 
> version suspect? Does the solution demand an upgrade to 2.0 or shall I debug 
> 1.0.7?
> 
> Thanks for any guidance,
> Les
> 
> 
> 
> w +$Hc-1#09
> r -
> w $Hc-1#09
> r <Timeout: 2 seconds>
> w $Hc-1#09
> r $#00
> w +
> r $#00
> w +
> r $#00
> w +
> r $#00
> w +
> r <Timeout: 2 seconds>
> 
> ... repeats
> 
> 
> c load
> w $X41000000,0:#73
> r +$OK#9a
> 
> ... load application hello-world
> 
> 
> Sending packet: $p42#d6...[+]Ack
> [$][0][0][0][0][0][0][2][0][#][8][2]Packet received: 00000020
> Sending packet: $p43#d7...[+]Ack
> [$][4][0][0][0][0][0][5][0][#][8][9]Packet received: 40000050
> 
> ...
> 
> Sending packet: $C01#a4...[+]Ack
> [$][T][0][1][4][4][:][4][0][0][1][8][c][1][c][;][0][e][:][0][0][0][0][0][0][0]
> [0][;][#][1][0]Packet received: T0144:40018c1c;0e:00000000;

That fact that it is downloading suggests the stub is working.  

This line T0144:40018c1c;0e:00000000; indicates it has thrown an
exception. My guess is you are downloading your hello world image over
the top of redboot. What RAM addresses does your application image
use? What RAM does redboot use?

Newer versions of redboot would probably give you a warning you are
about to stomp of it, but something as old as 1.0.7 probably does not
have the check.

     Andrew

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] GDB-stub in Redboot fails to handshake

[Gary Thomas]

Andrew Lunn wrote:
> On Sat, Mar 08, 2008 at 05:09:30AM -0800, Les G. Miklosy wrote:
>> Hello Gents: 
>>
>> As you can see in this gdb log file, my Sparc target running Redboot fails to 
>> recognize ($#00) the simple query commands offered by my Linux host over a 
>> serial connection. 
> 
> The redboot stub only implements a subset of the gdb protocol. 
> 
>> The target application is simply hello-world with a linked 
>> library build from ECOS version 1.0.7. Redboot is a ROMRAM bootloader also 
>> built from version 1.0.7. I can hear you smacking your foreheads asking why 
>> use V 1.0.7 , I am using this version because a board support package was 
>> built around it.
>>
>> I tried the execution of Redboot with and without a '$' to start the GDB-stub 
>> with no improvement. Any application I try to load/continue ends with a 
>> Signal Hangup.  A diff between ECOS version 2.0 and 1.0.7 shows many changes 
>> to the code particularly the part for initiating gdb-stub. Is this old 
>> version suspect? Does the solution demand an upgrade to 2.0 or shall I debug 
>> 1.0.7?
>>
>> Thanks for any guidance,
>> Les
>>
>>
>>
>> w +$Hc-1#09
>> r -
>> w $Hc-1#09
>> r <Timeout: 2 seconds>
>> w $Hc-1#09
>> r $#00
>> w +
>> r $#00
>> w +
>> r $#00
>> w +
>> r $#00
>> w +
>> r <Timeout: 2 seconds>
>>
>> ... repeats
>>
>>
>> c load
>> w $X41000000,0:#73
>> r +$OK#9a
>>
>> ... load application hello-world
>>
>>
>> Sending packet: $p42#d6...[+]Ack
>> [$][0][0][0][0][0][0][2][0][#][8][2]Packet received: 00000020
>> Sending packet: $p43#d7...[+]Ack
>> [$][4][0][0][0][0][0][5][0][#][8][9]Packet received: 40000050
>>
>> ...
>>
>> Sending packet: $C01#a4...[+]Ack
>> [$][T][0][1][4][4][:][4][0][0][1][8][c][1][c][;][0][e][:][0][0][0][0][0][0][0]
>> [0][;][#][1][0]Packet received: T0144:40018c1c;0e:00000000;
> 
> That fact that it is downloading suggests the stub is working.  
> 
> This line T0144:40018c1c;0e:00000000; indicates it has thrown an
> exception. My guess is you are downloading your hello world image over
> the top of redboot. What RAM addresses does your application image
> use? What RAM does redboot use?
> 
> Newer versions of redboot would probably give you a warning you are
> about to stomp of it, but something as old as 1.0.7 probably does not
> have the check.

Sadly, no.  Once the GDB protocol is entered, all of the checks that
RedBoot might be able to make (e.g. during the 'load' command), are
off since RedBoot is really out of the picture, only GDB is running.

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] GDB-stub in Redboot fails to handshake

[Les G. Miklosy]

Thankyou Gary and Andrew for that reply.

Here is the memory usage I have setup for my builds:

Redboot ROMRAM Start Address = 0x40_000_000
Virtual Vector Table = 0x40_100_000
Linked Library Address = 0x41_000_000
Application Start Address = 0x41_000_000

The Redboot load size is roughly 550,000 bytes or 0x86470, so I thought with 
this build there is plenty of room for the application to fit in. I can load 
and run all four Sample applications from the Redboot side with this setup 
OK, all load at 0x41000000. I just can't make the handshake from the gdb side 
and the subsequent run.

Les 

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss