[PATCH 3/3] Fix "Remote 'g' packet reply is too long" problems with multiple inferiors

[Ulrich Weigand]

Pedro Alves wrote:

> 	* target.c (default_thread_architecture): Use the sepecified
> 	inferior's architecture, instead of the current inferior's
> 	architecture (via target_gdbarch).

This causes a number of ICEs in the test suite for me, at the point when
default_thread_architecture is called from linux_child_follow_fork here:

          if (!gdbarch_software_single_step_p (target_thread_architecture
                                               (child_lp->ptid)))

Note that at this point there is no inferior for the child, and I think
there will never be one since the child is to be detached immediately.

Given that this is a fork, I guess a simple fix would be to check
for the parent's thread architecture here instead.

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU/Linux compilers and toolchain
  Ulrich.Weigand@de.ibm.com