[cdt-patch] Proposed CDT 1.2 patches

[Ashish Karkare <ashish.karkare@xxxxxxxxxxx>]

Hello!

I am a contributor to the TimeSys Eclipse/CDT based tools  effort.

In the process of testing our tools with large applications running on a
variety of different embedded systems, we have come across a few small
quirks in the interaction between CDT and gdb/MI.

I would like to propose a set of patches for three issues that we have seen:

1. If the stack-info-depth command on one of the threads returns an error,
   CDT does not display the stack trace for that thread.  This behavior is
   different from command line gdb, where gdb displays all the frames that it
   can, and then prints an error message.

   The proposed patch is to retry the stack-info-depth command- which
   generally succeeds because gdb patches the offending stack frame- and then
   display the results.

2. If gdb takes too long to load the application, the MISession can time out,
   leaving behind orphan session resources, including possibly a running
   gdb process. This can lead to failure of subsequent launches, even after
   the user increases timeout values.

   The proposed patch is to catch the exception and terminate the session, and
   then display the session time out message.

3. Different versions of gdb seem to vary greatly in their ability at retrieving
   the full target register set. This is a gdb bug: gdb/MI will list a
   register as available, but then  subsequently report an error while
   querying the value of the register.  When this happens, the register
   pane will be empty.

   The proposed patch is to catch the exception, and display the partial
   partial register set.


All of the above patches can be applied to the CDT-1.2 branch.

I will mail each of the proposed patches to the list separately.

Regards,

Ashish