Question about rr alternative design

Steve Fink sphink at gmail.com
Wed Feb 21 17:42:53 UTC 2018


I find it easiest to start up separate gdb instances for each process
of interest. Use 'rr ps' to find the process you're interested in, and
rr replay -p <pid>. If you care about multiple, start them up
simultaneously in separate windows. You have to piece together the
order of execution between the two manually. The 'when' and
'when-ticks' commands are useful for this.

On Wed, Feb 21, 2018 at 8:45 AM, Octav Chipara <ochipara at gmail.com> wrote:
> Thanks that works great. It is a bit difficult to stop the recording, but I guess the best way is to send a SIGKILL. I hope that rr does a graceful exit saving the needed data.
>
> On the replay side, how do you trace both the parent and child processes (or at least the child). I have tried to use set follow-fork-mode to child with and without setting follow-exec-mode to both.
>
> — Octav
>
>> On Feb 20, 2018, at 2:23 PM, Kyle Huey <me at kylehuey.com> wrote:
>>
>> Pass `-w` to `rr record` to tell it to wait on the daemonized nginx process.
>>
>> - Kyle
>>
>> On Wed, Feb 21, 2018 at 9:17 AM, Octav Chipara <ochipara at gmail.com> wrote:
>>> Thank you for your replies.
>>>
>>> On an unrelated topic, I am trying to obtain a recording of nginx. However,
>>> if I start it from the command line, rr does not record any data (perhaps it
>>> is not following child processes). If this is not support, is it possible to
>>> attach rr to a running process?
>>>
>>> Thanks,
>>> — Octav
>>>
>>>
>>>
>>> On Feb 20, 2018, at 2:09 PM, Robert O'Callahan <robert at ocallahan.org> wrote:
>>>
>>> On Wed, Feb 21, 2018 at 8:30 AM, Nathan Froyd <nfroyd at mozilla.com> wrote:
>>>>
>>>> I don't know whether this alternative was considered
>>>
>>>
>>> Wrapping libc calls was considered. Section 3.1 of the extended technical
>>> report mentions it. As Nathan and Kyle said, it's simpler and more
>>> comprehensive to intercept system calls directly rather than wrap libc
>>> functions.
>>>
>>> Rob
>>> --
>>> Su ot deraeppa sah dna Rehtaf eht htiw saw hcihw, efil lanrete eht uoy ot
>>> mialcorp ew dna, ti ot yfitset dna ti nees evah ew; deraeppa efil eht. Efil
>>> fo Drow eht gninrecnoc mialcorp ew siht - dehcuot evah sdnah ruo dna ta
>>> dekool evah ew hcihw, seye ruo htiw nees evah ew hcihw, draeh evah ew hcihw,
>>> gninnigeb eht morf saw hcihw taht.
>>>
>>>
>>>
>>> _______________________________________________
>>> rr-dev mailing list
>>> rr-dev at mozilla.org
>>> https://mail.mozilla.org/listinfo/rr-dev
>>>
>
> _______________________________________________
> rr-dev mailing list
> rr-dev at mozilla.org
> https://mail.mozilla.org/listinfo/rr-dev


More information about the rr-dev mailing list