linux-kernel - Re: [RFC][PATCH 3/5] Container Freezer: Implement freezer cgroup subsystem

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <1209551985.29759.20.camel@localhost.localdomain>
Date:	Wed, 30 Apr 2008 03:39:45 -0700
From:	Matt Helsley <matthltc@...ibm.com>
To:	Paul Menage <menage@...gle.com>
Cc:	Linux-Kernel <linux-kernel@...r.kernel.org>,
	Cedric Le Goater <clg@...ibm.com>,
	Oren Laadan <orenl@...columbia.edu>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Pavel Machek <pavel@....cz>,
	linux-pm@...ts.linux-foundation.org,
	Linux Containers <containers@...ts.linux-foundation.org>
Subject: Re: [RFC][PATCH 3/5] Container Freezer: Implement freezer cgroup
	subsystem


On Thu, 2008-04-24 at 22:51 -0700, Paul Menage wrote:
> >+static const char *freezer_state_strs[] = {
> >+	"RUNNING\n",
> >+	"FREEZING\n" ,
> >+	"FROZEN\n"
> >+};
> 
> I think it might be cleaner to not include the \n characters in this array.

Sure. Though that might produce weird output from
simple_read_from_buffer() -- no newline.

I've switched this and the strcmp() code below.

> >+static inline int cgroup_frozen(struct task_struct *task)
> >+{
> >+	struct cgroup *cgroup = task_cgroup(task, freezer_subsys_id);
> >+	struct freezer *freezer = cgroup_freezer(cgroup);
> >+	enum freezer_state state;
> >+
> >+	spin_lock(&freezer->lock);
> >+	state = freezer->state;
> >+	spin_unlock(&freezer->lock);
> >+
> >+	return (state == STATE_FROZEN);
> >+}
> 
> You need to be in an RCU critical section or else hold task_lock() in
> order to dereference the cgroup returned from task_cgroup()

What are the rules of using subsystem pointers from the cgroup? Suppose
I did:

rcu_read_lock();
cgroup = task_cgroup(task, freezer_subsys_id);
freezer = cgroup_freezer(cgroup);
state = freezer->state;
rcu_read_unlock();

return (state == STATE_FROZEN);

(And guard writes to freezer->state with the freezer->lock)

?

> I'm not sure that you need to take freezer->lock here - you're just
> reading a single word.

Doesn't the safety of that assumption depend on the architecture _and_
compiler?

> >+
> >+	if (!capable(CAP_SYS_ADMIN))
> >+		return ERR_PTR(-EPERM);
> >+
> 
> Why does everyone keep throwing calls to check CAP_SYS_ADMIN into
> their cgroup create callbacks? You have to be root in order to mount a
> cgroups hierarchy in the first place, and filesystem permissions will
> control who can create new cgroups.

Removed.

> >+static int freezer_can_attach(struct cgroup_subsys *ss,
> >+			      struct cgroup *new_cgroup,
> >+			      struct task_struct *task)
> >+{
> >+	struct freezer *freezer = cgroup_freezer(new_cgroup);
> >+	int retval = 0;
> >+
> >+	if (freezer->state == STATE_FROZEN)
> >+		retval = -EBUSY;
> >+
> >+	return retval;
> >+}
> 
> You should comment here that the call to cgroup_lock() in the
> freezer.state write method prevents a write to that file racing
> against an attach, and hence the can_attach() result will remain valid
> until the attach completes.

OK. I used your comment. :)

> >+static ssize_t freezer_write(struct cgroup *cgroup,
> >+			     struct cftype *cft,
> >+			     struct file *file,
> >+			     const char __user *userbuf,
> >+			     size_t nbytes, loff_t *unused_ppos)
> >+{
> >+	char *buffer;
> >+	int retval = 0;
> >+	enum freezer_state goal_state;
> >+
> >+	if (nbytes >= PATH_MAX)
> >+		return -E2BIG;
> >+
> >+	/* +1 for nul-terminator */
> >+	buffer = kmalloc(nbytes + 1, GFP_KERNEL);
> >+	if (buffer == NULL)
> >+		return -ENOMEM;
> 
> Given that you're copying a string whose maximum valid length is
> "FREEZING" you don't really need to use a dynamically-allocated
> buffer.

Yup. Changed to use a fixed buffer.

> But I really ought to provide a write_string() method that handles
> this kind of copying on behalf of cgroup subsystems, the way it
> already does for 64-bit ints.

Seems like a good idea for this cgroup subsystem at least.

> >+	if (strcmp(buffer, "RUNNING") == 0)
> >+		goal_state = STATE_RUNNING;
> >+	else if (strcmp(buffer, "FROZEN") == 0)
> >+		goal_state = STATE_FROZEN;
> 
> Would it make sense to compare against the strings you already have in
> the array earlier in the file?

Done.

Cheers,
	-Matt Helsley

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/