Re: [RFC][PATCH 3/5] Determine if sender is from ancestor ns

[Sukadev Bhattiprolu]

Bastian Blank [bastian@xxxxxxxxxxxx] wrote:
| On Mon, Dec 01, 2008 at 12:15:06PM -0800, Sukadev Bhattiprolu wrote:
| > Bastian Blank [bastian@xxxxxxxxxxxx] wrote:
| > | If I see this correctly this information is already covered in si_code
| > | with SI_USER and SI_TKILL. SI_KERNEL is used for explicit kernel
| > | generated signals.
| > 
| > Yes, but si_code from sys_rt_sigqueueinfo() cannot be trusted.
| 
| sys_rt_sigqueueinfo disallows setting si_code to any value which
| describes kernel signals from userspace. So using SI_FROMUSER should be
| sufficient.

Hmm, unless I am missing something, sys_rt_sigqueuinfo() does this:

        if (info.si_code >= 0)
                return -EPERM;

This does not prevent user from setting si_code to SI_ASYNCIO, which,
from include/asm-generic/siginfo.h is:

#define SI_ASYNCIO      -4              /* sent by AIO completion */

Also,

#define SI_FROMUSER(siptr)      ((siptr)->si_code <= 0)

SI_ASYNCIO qualifies as SI_FROMUSER() even when it originates from
kernel (usb/core/devio.c async_completed())...

| 
| > IOW, we need to find the namespace of the sender only if the sender is
| > a user process. If signal is originating from kernel, safely checking
| > namespace becomes more complex.
| 
| Where does this imply checking sender for kernel generated signals?

... so what I meant is that in send_signal(), it will be harder to
determine in the SI_ASYNCIO case whether the signal is from driver or
rt_sigqueueinfo().

If we know that it came from rt_sigqueueinfo(), we can safely check
the namespace. If it came from driver we should skip the ns check.

| 
| > Yes, current approach is somewhat hacky. We tried other approaches
| > before and they were either intrusive or required non-trivial changes
| > to semantics of signals to global init or both.
| 
| Message-IDs?

Yes, (Eric Biederman, Dec 2007)
	https://lists.linux-foundation.org/pipermail/containers/2007-December/009152.html

Oleg Nesterov, Aug 2007:
	http://marc.info/?l=linux-kernel&m=118753610515859

I had sent out a summary of the above attempts to Containers list recently:
	https://lists.linux-foundation.org/pipermail/containers/2008-November/013991.html



| 
| > | > +static inline int siginfo_from_ancestor_ns(struct task_struct *t,
| > | > +			siginfo_t *info)
| > | > +{
| > | > +	if (!is_si_special(info) && (info->si_signo & SIG_FROM_USER)) {
| > | > +		/* if t can't see us we are from parent ns */
| > | What?
| > I assume your question is about the comment :-)
| 
| Yes.
| 
| > Yes, a process can see all its descendants and processes in descendant
| > namespaces. But it can only see its ancestors upto the most recent
| > CLONE_NEWPID. (kind of like chroot in filesystems). So if receiver
| > can't see sender, sender must be an ancestor.
| 
| Please add a complete comment to the function which describes the
| function. And don't us "it" for not defined entities.

Ah, I see the problem now. The 't' refers to the task parameter - how
about changing comment to:

	/* If receiver can't see us, we are from parent ns */


| 
| Bastian
| 
| -- 
| I have never understood the female capacity to avoid a direct answer to
| any question.
| 		-- Spock, "This Side of Paradise", stardate 3417.3
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/