Re: [commit 7.6.1 only] [patch gdbserver 7.6.1 only] Fix fd leak regression

[Pedro Alves <palves at redhat dot com>]

On 08/29/2013 03:27 PM, Jan Kratochvil wrote:
> On Thu, 29 Aug 2013 16:17:40 +0200, Pedro Alves wrote:

>>> +set test "system fd behavior is known"
>>> +set status [remote_exec target "[standard_output_file $testfile]"]
>>> +if { [lindex $status 0] == 0 } {
>>> +    pass $test
>>> +} else {
>>> +    fail $test
>>> +}
>>> +remote_exec target "ls -l /proc/self/fd/"
>>
>> Before gdbserver's fix, do we get one extra fd from the dejagnu
>> leak, and another extra from gdbserver's leak?  What if we made
>> $testfile count open fds, and then compare that between running
>> under gdb/gdbserver and just under remote_exec ?
> 
> BTW not sure if it is clear 

It wasn't at first, but I figured it out when I noticed it had
no ChangeLog entry.

> but this gdb/testsuite/ part was sent
> accidentally, there is even written no real test and I have not checked in
> anything as I got stuck on the DejaGNU bug.

Understood.  I was suggesting a possible way to not need to
wait for the dejagnu bug to be fixed.  If what I suggested actually
works, I don't see a downside -- we'd just be checking whether
gdbserver itself introduces any leak, which is all we should
care about?

> 
> I have sent a bugreport to <bug-dejagnu@gnu.org> and it got processed today
> although the mail has not yet appeared in the archive:
> 	http://lists.gnu.org/archive/html/bug-dejagnu/
> 
> I hope DejaGNU gets fixed soon so the testcase can be later written as
> UNSUPPORTED (UNTESTED?) with buggy DejaGNU.

Ack.

-- 
Pedro Alves