Re: RFD: Non-Disruptive Core Dump Infrastructure

[KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>]

On 9/12/2013 12:45 AM, Suzuki K. Poulose wrote:
> On 09/12/2013 12:57 AM, KOSAKI Motohiro wrote:
>> (9/3/13 4:39 AM), Janani Venkataraman wrote:
>>> Hello,
>>>
>>> We are working on an infrastructure to create a system core file of a
>>> specific
>>> process at run-time, non-disruptively. It can also be extended to a
>>> case where
>>> a process is able to take a self-core dump.
>>>
>>> gcore, an existing utility creates a core image of the specified
>>> process. It
>>> attaches to the process using gdb and runs the gdb gcore command and then
>>> detaches. In gcore the dump cannot be issued from a signal handler
>>> context as
>>> fork() is not signal safe and moreover it is disruptive in nature as
>>> the gdb
>>> attaches using ptrace which sends a SIGSTOP signal. Hence the gcore
>>> method
>>> cannot be used if the process wants to initiate a self dump.
>>
>> Maybe I'm missing something. But why gcore uses c-level fork()? gcore
>> need to
>> call pthread-at-fork handler? No. gcore need to flush stdio buffer? No.
>>
> Let me clarify. If an application wants to dump itself, it has to do a
> fork() and then exec the gcore with the pid of the appication to
> generate the dump.

Oh, I did think the fork() is used for no application stop dump. But it is
incorrect.

Hmm. However, if an application _itself_ want to dump itself. They can avoid
to use signal handler properly. I'm missing the point of this discussion
completely.

So, I'd keep silence while.

> 
> So, if the application wants to initiate the dump from a signal handler
> context, it may lead to trouble.