[Lldb-commits] [lldb] r228306 - In fd leak test report all open fds, not only first failure

[Ed Maste]

On 5 February 2015 at 11:19, Vince Harron <vharron at google.com> wrote:
> Hi Ed,
>
> What is the motivation for this change?

Hi Vince,

I applied Pavel's ProcessMonitor leak fix to FreeBSD, and then found
there's still another fd leak coming from within python.  This change
is just to make it easier to identify all leaked fds at once rather
than running the test, fixing one leak, running again, fixing the
next, etc.  When there are no leaks this test fstat()s fds 3 through
256, so this change shouldn't introduce any extra syscall overhead in
the passing case.

As an aside, it's unfortunate that there's so much duplicated code in
Process/FreeBSD and Process/Linux.  It's been like that as long as
I've been involved with LLDB though, and I've never been able to find
time to try cleaning it up.