Re: [PATCH] sunrpc: fix refcount leak for rpc auth modules

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




> On Mar 1, 2021, at 11:28 AM, Bruce Fields <bfields@xxxxxxxxxxxx> wrote:
> 
> On Mon, Mar 01, 2021 at 03:20:24PM +0000, Chuck Lever wrote:
>> 
>>> On Feb 26, 2021, at 6:04 PM, Daniel Kobras <kobras@xxxxxxxxxxxxx> wrote:
>>> 
>>> If an auth module's accept op returns SVC_CLOSE, svc_process_common()
>>> enters a call path that does not call svc_authorise() before leaving the
>>> function, and thus leaks a reference on the auth module's refcount. Hence,
>>> make sure calls to svc_authenticate() and svc_authorise() are paired for
>>> all call paths, to make sure rpc auth modules can be unloaded.
>>> 
>>> Fixes: 4d712ef1db05 ("svcauth_gss: Close connection when dropping an incoming message")
>>> Signed-off-by: Daniel Kobras <kobras@xxxxxxxxxxxxx>
>>> ---
>>> Hi!
>>> 
>>> While debugging NFS on a system with misconfigured krb5 settings, we noticed
>>> a suspiciously high refcount on the auth_rpcgss module, despite all of its
>>> consumers already unloaded. I wasn't able to analyze any further on the live
>>> system, but had a look at the code afterwards, and found a path that seems
>>> to leak references if the mechanism's accept() op shuts down a connection
>>> early. Although I couldn't verify, this seem to be a plausible fix.
>>> 
>>> Kind regards,
>>> 
>>> Daniel
>> 
>> Hi Daniel-
>> 
>> I've provisionally included your patch in my NFSD for-rc topic branch
>> here:
>> 
>> git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux.git
>> 
>> Your bug report seems plausible, but I need to take a closer look at that
>> code and your proposed change. Would very much like to hear from others,
>> too.
> 
> So, the effect of this is to call svc_authorise more often.  I think
> that's always safe, because svc_authorise is a no-op unless rq_authops
> is set, it clears rq_authops itself, and rq_authops being set is a
> guarantee that ->accept() already ran.
> 
> It's harder to know if this solves the problem, as I see a lot of other
> mentions of THIS_MODULE in svcauth_gss.c.

Perhaps a deeper audit is necessary.

A small code change to inject SVC_CLOSE returns at random would enable
a more dynamic analysis.


> Possibly orthogonal to this problem, but: svcauth_gss_release
> unconditionally dereferences rqstp->rq_auth_data.  Isn't that a NULL
> dereference if the kmalloc at the start of svcauth_gss_accept() fails?
> 
> Finally, should we care about module reference leaks?

I would prefer that module reference counting work as expected. When it
doesn't that tends to lead to people (say, me) hunting for bugs that
might actually be serious.


> Does anyone really *need* to unload modules?

Anyone who wants to replace the module with a newer build that fixes a
bug. It avoids a full reboot, and for some that's important.


> And will bad stuff happen when the
> count overflows, or does the module code fail safely somehow in the
> overflow case?  I know, bugs are bugs, I should care about fixing all of
> them, shame on me....
> 
> --b.
> 
>> 
>> 
>>> net/sunrpc/svc.c | 6 ++++--
>>> 1 file changed, 4 insertions(+), 2 deletions(-)
>>> 
>>> diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
>>> index 61fb8a18552c..d76dc9d95d16 100644
>>> --- a/net/sunrpc/svc.c
>>> +++ b/net/sunrpc/svc.c
>>> @@ -1413,7 +1413,7 @@ svc_process_common(struct svc_rqst *rqstp, struct kvec *argv, struct kvec *resv)
>>> 
>>> sendit:
>>> 	if (svc_authorise(rqstp))
>>> -		goto close;
>>> +		goto close_xprt;
>>> 	return 1;		/* Caller can now send it */
>>> 
>>> release_dropit:
>>> @@ -1425,6 +1425,8 @@ svc_process_common(struct svc_rqst *rqstp, struct kvec *argv, struct kvec *resv)
>>> 	return 0;
>>> 
>>> close:
>>> +	svc_authorise(rqstp);
>>> +close_xprt:
>>> 	if (rqstp->rq_xprt && test_bit(XPT_TEMP, &rqstp->rq_xprt->xpt_flags))
>>> 		svc_close_xprt(rqstp->rq_xprt);
>>> 	dprintk("svc: svc_process close\n");
>>> @@ -1433,7 +1435,7 @@ svc_process_common(struct svc_rqst *rqstp, struct kvec *argv, struct kvec *resv)
>>> err_short_len:
>>> 	svc_printk(rqstp, "short len %zd, dropping request\n",
>>> 			argv->iov_len);
>>> -	goto close;
>>> +	goto close_xprt;
>>> 
>>> err_bad_rpc:
>>> 	serv->sv_stats->rpcbadfmt++;
>>> -- 
>>> 2.25.1
>>> 
>>> 
>>> -- 
>>> Puzzle ITC Deutschland GmbH
>>> Sitz der Gesellschaft: Eisenbahnstraße 1, 72072 
>>> Tübingen
>>> 
>>> Eingetragen am Amtsgericht Stuttgart HRB 765802
>>> Geschäftsführer: 
>>> Lukas Kallies, Daniel Kobras, Mark Pröhl
>>> 
>> 
>> --
>> Chuck Lever

--
Chuck Lever







[Index of Archives]     [Linux Filesystem Development]     [Linux USB Development]     [Linux Media Development]     [Video for Linux]     [Linux NILFS]     [Linux Audio Users]     [Yosemite Info]     [Linux SCSI]

  Powered by Linux