This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
remote debug broken?
- From: "Robert Trask" <rtrask at broadcom dot com>
- To: gdb at sources dot redhat dot com
- Date: Thu, 18 Mar 2004 14:13:55 -0800
- Subject: remote debug broken?
Hello,
I want to use the remote debug feature on gdb wherein I have a
[superior] debugger running on a host system (Linux I386 pc) and an
[inferior] debugger running on the target (embedded Linux on MIPS). I
have tried this in gdb v5.3 and v6.0, the same behavior results.
I built the superior debugger in the usual way:
./configure // build for native operation
make
then installed in my bin directory.
I built the inferior debugger in the gdbserver directory:
mkdir mipsobj
cd mipsobj
setenv CC <my mips cross gcc path>
../configure mips-linux-gnu
make
And then installed on my target...
I then start up the inferior debugger (gdbserver) on the target:
gdbserver host:tcp_port <prog-to-debug>
This obligingly waits for the superior debugger to contact it and start
matters.
I start gdb on the host in the usual way:
gdb <prog-to-debug>
then:
(gdb) target remote 10.18.99.108:6001
Remote debugging using 10.18.99.108:6001
Sending packet: $Hc-1#09...Ack
Packet received: OK
Sending packet: $qC#b4...Ack
Packet received:
Sending packet: $qOffsets#4b...Ack
Packet received:
Sending packet: $?#3f...Ack
Packet received: T0525:2aac0e50;1d:7fff7ee0;
Couldn't establish connection to remote target
Reply contains invalid hex digit 59
The hex digit 59 is a semicolon. Looking at the gdbserver sources I can
see it thinks (correctly) it's replying to a resume command and
expressly uses semicolons as field delimiters.
Am I missing an important step somewhere, doing something dumb, or
both? Does remote debugging no longer work? Any ideas or suggestions
would be appreciated.
thanks!
/bob