mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Trond Myklebust <Trond.Myklebust@netapp.com>
To: Akinobu Mita <akinobu.mita@gmail.com>
Cc: linux-kernel@vger.kernel.org,
	Andy Adamson <andros@citi.umich.edu>,
	"J. Bruce Fields" <bfields@citi.umich.edu>,
	Olaf Kirch <okir@monad.swb.de>
Subject: Re: [PATCH 2/2] auth_gss: unregister gss_domain when unloading module
Date: Sun, 29 Oct 2006 15:15:59 -0500	[thread overview]
Message-ID: <1162152960.5545.57.camel@lade.trondhjem.org> (raw)
In-Reply-To: <20061029133816.GB10295@localhost>

On Sun, 2006-10-29 at 22:38 +0900, Akinobu Mita wrote:
> Reloading rpcsec_gss_krb5 or rpcsec_gss_spkm3 hit duplicate
> registration in svcauth_gss_register_pseudoflavor().
> (If DEBUG_PAGEALLOC is enabled, oops will happen at
> auth_domain_put() --> hlist_del() with uninitialized hlist_node)
> 
> svcauth_gss_register_pseudoflavor(u32 pseudoflavor, char * name)
> {
> 	...
> 
>         test = auth_domain_lookup(name, &new->h);
>         if (test != &new->h) { /* XXX Duplicate registration? */
>                 auth_domain_put(&new->h);
>                 /* dangling ref-count... */
> 	...
> }
> 
> This patch unregisters gss_domain and free it when unloading
> modules.
> 
> - Define svcauth_gss_unregister_pseudoflavor()
>   (doing the opposite of svcauth_gss_register_pseudoflavor())
> 
> - Call svcauth_gss_unregister_pseudoflavor() in gss_mech_free()
> 
> Cc: Andy Adamson <andros@citi.umich.edu>
> Cc: J. Bruce Fields <bfields@citi.umich.edu>
> Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
> Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
> 
>  net/sunrpc/auth_gss/gss_mech_switch.c |    4 ++++
>  net/sunrpc/auth_gss/svcauth_gss.c     |    6 +++---
>  2 files changed, 7 insertions(+), 3 deletions(-)
> 
> Index: work-fault-inject/net/sunrpc/auth_gss/gss_mech_switch.c
> ===================================================================
> --- work-fault-inject.orig/net/sunrpc/auth_gss/gss_mech_switch.c
> +++ work-fault-inject/net/sunrpc/auth_gss/gss_mech_switch.c
> @@ -60,6 +60,9 @@ gss_mech_free(struct gss_api_mech *gm)
>  
>  	for (i = 0; i < gm->gm_pf_num; i++) {
>  		pf = &gm->gm_pfs[i];
> +		if (!pf->auth_domain_name)
> +			continue;
> +		svcauth_gss_unregister_pseudoflavor(pf->auth_domain_name);
>  		kfree(pf->auth_domain_name);
>  		pf->auth_domain_name = NULL;
>  	}
> @@ -93,8 +96,11 @@ gss_mech_svc_setup(struct gss_api_mech *
>  			goto out;
>  		status = svcauth_gss_register_pseudoflavor(pf->pseudoflavor,
>  							pf->auth_domain_name);
> -		if (status)
> +		if (status) {
> +			kfree(pf->auth_domain_name);
> +			pf->auth_domain_name = NULL;
>  			goto out;
> +		}
>  	}
>  	return 0;
>  out:
> Index: work-fault-inject/net/sunrpc/auth_gss/svcauth_gss.c
> ===================================================================
> --- work-fault-inject.orig/net/sunrpc/auth_gss/svcauth_gss.c
> +++ work-fault-inject/net/sunrpc/auth_gss/svcauth_gss.c
> @@ -765,10 +765,12 @@ svcauth_gss_register_pseudoflavor(u32 ps
>  
>  	test = auth_domain_lookup(name, &new->h);
>  	if (test != &new->h) { /* XXX Duplicate registration? */
> -		auth_domain_put(&new->h);
> -		/* dangling ref-count... */
> -		goto out;
> +		WARN_ON(1);
> +		stat = -EBUSY;
> +		kfree(new->h.name);
> +		goto out_free_dom;
>  	}
> +	auth_domain_put(&new->h);
>  	return 0;
>  
>  out_free_dom:
> @@ -779,6 +781,19 @@ out:
>  
>  EXPORT_SYMBOL(svcauth_gss_register_pseudoflavor);
>  
> +void svcauth_gss_unregister_pseudoflavor(char *name)
> +{
> +	struct auth_domain *dom;
> +
> +	dom = auth_domain_find(name);
> +	if (dom) {
> +		auth_domain_put(dom);
> +		auth_domain_put(dom);
> +	}
> +}

Strictly speaking, if you want to be smp-safe, you probably need
something like the following:

	dom = auth_domain_find(name);
	if (dom) {
		spin_lock(&auth_domain_lock);
		if (!hlist_unhashed(dom->hash)) {
			hlist_del_init(dom->hash);
			spin_unlock(&auth_domain_lock);
			auth_domain_put(dom);
		} else
			spin_unlock(&auth_domain_lock);
		auth_domain_put(dom);
	}

and then add a test for hlist_unhashed into auth_domain_put(). If not,
some other processor could race you inside
svcauth_gss_unregister_pseudoflavor.

However since all this is being done under the BKL, then your
alternative above might be sufficient. The only question then would be
why we need auth_domain_lock?

> +EXPORT_SYMBOL(svcauth_gss_unregister_pseudoflavor);
> +
>  static inline int
>  read_u32_from_xdr_buf(struct xdr_buf *buf, int base, u32 *obj)
>  {
> Index: work-fault-inject/include/linux/sunrpc/svcauth_gss.h
> ===================================================================
> --- work-fault-inject.orig/include/linux/sunrpc/svcauth_gss.h
> +++ work-fault-inject/include/linux/sunrpc/svcauth_gss.h
> @@ -22,6 +22,7 @@
>  int gss_svc_init(void);
>  void gss_svc_shutdown(void);
>  int svcauth_gss_register_pseudoflavor(u32 pseudoflavor, char * name);
> +void svcauth_gss_unregister_pseudoflavor(char *name);
>  
>  #endif /* __KERNEL__ */
>  #endif /* _LINUX_SUNRPC_SVCAUTH_GSS_H */

Cheers,
  Trond

  reply	other threads:[~2006-10-29 20:16 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-10-28 18:55 [PATCH] " Akinobu Mita
2006-10-29 13:35 ` Akinobu Mita
2006-10-29 13:37   ` [PATCH 1/2] sunrpc: add missing spin_unlock Akinobu Mita
2006-10-29 13:38     ` [PATCH 2/2] auth_gss: unregister gss_domain when unloading module Akinobu Mita
2006-10-29 20:15       ` Trond Myklebust [this message]
2006-10-30 14:54         ` Akinobu Mita
2006-10-30 15:17           ` Trond Myklebust
2006-10-31  3:15             ` Akinobu Mita
2006-10-31  4:13               ` Neil Brown
2006-10-29 20:21     ` [PATCH 1/2] sunrpc: add missing spin_unlock Trond Myklebust
2006-10-30  5:43       ` Neil Brown
2006-10-30 14:57         ` [PATCH -mm] sunrpc/auth_gss: auth_domain refcount fix Akinobu Mita
2006-11-05 16:35     ` [PATCH 1/2] sunrpc: add missing spin_unlock Peter Zijlstra
2006-11-05 19:45       ` Andrew Morton
2006-11-05 19:58         ` Peter Zijlstra
2006-11-05 22:04         ` Elimar Riesebieter
2006-10-29 19:46 ` [PATCH] auth_gss: unregister gss_domain when unloading module Trond Myklebust

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1162152960.5545.57.camel@lade.trondhjem.org \
    --to=trond.myklebust@netapp.com \
    --cc=akinobu.mita@gmail.com \
    --cc=andros@citi.umich.edu \
    --cc=bfields@citi.umich.edu \
    --cc=linux-kernel@vger.kernel.org \
    --cc=okir@monad.swb.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®