Harden gdb.base/coredump-filter.exp

  This testcase seems to assume the target is running Linux, so bare 
metal, simulators and other debugging stubs running different OS' will 
have a hard time executing some of the commands the testcase issues.

Even restricting the testcase to Linux systems (which the patch below 
does), there are still problems with, say, QEMU not providing PID 
information when "info inferior" is issued. As a consequence, the 
subsequent tests will either fail or will not make much sense.

The attached patch checks if PID information is available. If not, it 
just bails out and avoids running into a number of failures.

No regressions on x86-64. For QEMU, i see only a couple PASSes before it 
is done with the testcase.

OK?
  

Luis Gustavo <luis_gustavo@mentor.com> writes:

> diff --git a/gdb/testsuite/gdb.base/coredump-filter.exp b/gdb/testsuite/gdb.base/coredump-filter.exp
> index f3203be..4c6c6ed 100644
> --- a/gdb/testsuite/gdb.base/coredump-filter.exp
> +++ b/gdb/testsuite/gdb.base/coredump-filter.exp
> @@ -15,6 +15,12 @@
>  
>  standard_testfile
>  
> +# This test is Linux-only.
> +if ![istarget *-*-linux*] then {
> +    unsupported "coredump-filter.exp"
> +    return -1
> +}
> +
>  if { [prepare_for_testing "failed to prepare" $testfile $srcfile debug] } {
>      untested "could not compile test program"
>      return -1
> @@ -146,6 +152,11 @@ gdb_test_multiple "info inferiors" "getting inferior pid" {
>      -re "process \($decimal\).*\r\n$gdb_prompt $" {
>  	set infpid $expect_out(1,string)
>      }
> +    -re "Remote target.*$gdb_prompt $" {
> +	# If the target does not provide PID information (like QEMU), just bail

FAOD, QEMU here is QEMU user mode, isn't?

> +	# out as the rest of the test may rely on it, giving spurious failures.
> +	return -1
> +    }
>  }

We can just bail out if infpid is still "" after this gdb_test_multiple.
  

> +# This test is Linux-only.
> +if ![istarget *-*-linux*] then {
> +    unsupported "coredump-filter.exp"
> +    return -1
> +}

s/unsupported/untested/, but it's the same:

 https://sourceware.org/gdb/wiki/GDBTestcaseCookbook#A.22untested.22_calls

Thanks,
Pedro Alves