This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: one week to go until GDB 7.11 branch creation...
- From: Pedro Alves <palves at redhat dot com>
- To: Yao Qi <qiyaoltc at gmail dot com>, Joel Brobecker <brobecker at adacore dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Thu, 21 Jan 2016 15:24:25 +0000
- Subject: Re: one week to go until GDB 7.11 branch creation...
- Authentication-results: sourceware.org; auth=none
- References: <20160118034414 dot GG4059 at adacore dot com> <86fuxsrz38 dot fsf at gmail dot com> <20160121100545 dot GB5146 at adacore dot com> <8637trrmss dot fsf at gmail dot com> <56A0F798 dot 5050608 at redhat dot com>
On 01/21/2016 03:22 PM, Pedro Alves wrote:
> On 01/21/2016 03:10 PM, Yao Qi wrote:
>>>>>> - A fail in gdb.threads/interrupted-hand-call.exp
>>>>>>
>>>>>> $ make check
>>>>>> RUNTESTFLAGS='--target_board=remote-gdbserver-on-localhost
>>>>>> interrupted-hand-call.exp'
>>>>>>
>>>>>> (gdb) PASS: gdb.threads/interrupted-hand-call.exp: dummy frame popped
>>>>>> continue^M
>>>>>> Continuing.^M
>>>>>> FAIL: gdb.threads/interrupted-hand-call.exp: continue until exit (timeout)
>>>>>> Remote debugging from host 127.0.0.1^M
>>>>>> ptrace(regsets_fetch_inferior_registers) PID=9710: No such process^M
>>>>>> ptrace(regsets_fetch_inferior_registers) PID=9710: No such process^
>>>>>>
>>>>>> I suspect it is about a GDB PR about disappeared inferior, but I can't
>>>>>> find the PR in bugzilla. I'll look into it.
>>>>
>>>> A little confusing, at the very least, but if the program did terminate,
>>>> I would say this issue is not blocking for the release. WDYT?
>> I am afraid not, the program didn't terminate, at least there is one
>> thread, as far as I can tell. Again, nothing useful to say here without
>> further analysis. PR 19508 is opened to track it.
>
> Several tests have this racy issue with gdbserver. It'll usually manifest when
> running the main thread to exit while there are still other threads running.
> Notice how the test program doesn't gracefully terminate/join all threads
> before exiting. So gdb/gdbserver are processing something for one thread,
> and meanwhile the process exits. This is really the same as PR 18749.
BTW, I don't think this is a new issue, so I don't think it should be a
blocker. In fact, we're much better handling these scenarios nowadays
than in past releases. Not great, but better.
Thanks,
Pedro Alves