OK, PTAL. I'm sure it's not done but it's better.

Now, you do stuff like this
echo strace > /proc/pid/ctl
or
straceme (no inheritance, not implemented not to do that yet)
straceoff

then cat /proc/pid/strace to get the records

comments welcome

On Sun, Jan 31, 2016 at 2:27 PM ron minnich <[email protected]> wrote:

>
> That was an extremely helpful review, thanks.
>
> The reason to qhangup when the cclose() hits 1 is that there are only two
> ways that happens I can think of:
> 1. something is still reading the /proc/pid/strace file, but there's
> nothing more to trace -- in this case, we need a qhangup
> 2. strace command is gone, and there is a proc running out there. In this
> case there's no harm done with the qhangup since the traced process has no
> need to write to it.
>
> Agreed that it sucks and needs to be done better; I just didn't think of
> anything :-)
>
> We need to talk about this, it gets a tad tricky when the tracer forks and
> the child inherits the fd.
>
> But, above all, I want this to work:
>
> cat /proc/pid/strace
> needs to exit correctly when pid and all its kids are gone.  This works
> with the current kludge.
>
> ron
>

-- 
You received this message because you are subscribed to the Google Groups 
"Akaros" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to