[PATCH v5 3/4] Implement -e status=set option

[Paul Chaignon]

Thanks for testing Burkhard!

On Tue, Jul 02, 2019 at 03:01:19PM +0000, Kohl, Burkhard wrote:
[...]

> For any combination of "strace -d -e status=<list>" I see in the debug output
> that strace_close_memstream() wants to publish a closed memstream when
> a process has exited:
>
>       ./strace: next_event: queued pid 19858
>       ./strace: next_event: dequeued pid 19858
>       16:42:53 exit_group(0)                  = ?
>       ./strace: [wait(0x000000) = 19858] WIFEXITED,exitcode=0
>       ./strace: next_event: queued pid 19858
>       ./strace: next_event: dequeued pid 19858
>       16:42:53 +++ exited with 0 +++
>       ./strace: dropped tcb for pid 19858, 0 remain
>       ./strace: memstream already closed

I'm wondering if this could be because print_event_exit() already closed
the memstream on PTRACE_EVENT_EXIT (you should have such an event just
before the debug logs you included here).  I'll get a gdb backtrace to be
sure.

>
> Could this be an off-by-one-error (NUMBER_OF_STATUSES)?

What makes you think so?

Paul