Re: Testing lxc 0.6.5 in Fedora 13

[Matt Helsley <matthltc@us.ibm.com>]

On Fri, Mar 26, 2010 at 01:45:22PM +0100, Grzegorz Nosek wrote:
> On Fri, Mar 26, 2010 at 04:53:57AM -0700, Matt Helsley wrote:
> > Yup. strace would need to be modified to use that. I tried that and it still
> > won't work -- I seem to recall it didn't work because strace uses pid values
> > obtained from the wait syscall too. To make it work we'd need to be able to
> > translate those pids in userspace. That's do-able from userspace if you trace
> > all forks descending from the pidns init task. But it's not do-able for
> > simple attaches. That's why I was thinking Eric's setns() might be able to
> > help if strace used it to enter the tracee's pid namespace whenever we need to.
> > 
> > gdb often doesn't use the same methods but has similar problems with pid
> > namespaces.
> 
> Hmm, is there a good reason why strace does not use the data explicitly
> provided by the kernel but instead second-guesses it from syscall return
> values? I don't know anything about ptrace, really, but I'd expect the

strace isn't linux-only. Checking the syscall return values may be seen
as being more portable. At least that's my guess. That said there are
plenty of #ifdefs in strace and patching it to use GETEVENTMSG is quite
a small patch.

However, as I said, that still doesn't "fix" strace so that it can
be used to trace tasks in child pid namespaces. Especially when the
traced tasks are more than one namepace deeper. :(

> kernel to provide the tracer with out-of-band information otherwise
> taken from clone/waitpid/other syscalls?

I don't think the kernel provides special out-of-band methods for fetching
pids related to traced tasks except during fork and clone. Not wait*(). The
rest of ptrace tends to be focused on reading/writing registers/memory and
managing signal delivery.

Cheers,
	-Matt Helsley