Re: [patch] Run tests in gdb.reverse

[Pedro Alves <pedro at codesourcery dot com>]

On Monday 07 November 2011 14:54:28, Tom Tromey wrote:
> >>>>> "Yao" == Yao Qi <yao@codesourcery.com> writes:
> 
> Yao> During reviewing "arm reverse debugging" patch, we realize that tests in
> Yao> gdb.reverse are not run in common regression test unless we set some
> Yao> variables in board file.
> 
> Yao> This patch is to convert variable checking in each reverse test case to
> Yao> a proc return value checking, so that gdb.reverse tests will be run on
> Yao> targets that support reverse debugging.
> 
> Thanks for doing this.
> 
> Yao> +proc support_process_record {} {
> Yao> +
> Yao> +    if { [istarget "x86_64-*-linux*"] || [istarget "i\[34567\]86-*-linux*"] } {
> 
> It seems to me that both of the new procs should include the old check
> using target_info as well, just in case.  This will avoid breaking
> anybody's existing setup.

Yes, we need to be able to run the reverse tests against
remote targets that support reverse debugging themselves.
E.g., simulators.  Those can do reverse debugging on their own,
target record is not involved.  Can we make predicates supportS_FOO
(add an S)? support_FOO reads as an action to me.

-- 
Pedro Alves