mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Nicholas A. Bellinger" <nab@linux-iscsi.org>
To: Christoph Hellwig <hch@lst.de>
Cc: "Nicholas A. Bellinger" <nab@daterainc.com>,
	target-devel <target-devel@vger.kernel.org>,
	linux-scsi <linux-scsi@vger.kernel.org>,
	lkml <linux-kernel@vger.kernel.org>,
	Sagi Grimberg <sagig@mellanox.com>,
	Hannes Reinecke <hare@suse.de>, Andy Grover <agrover@redhat.com>,
	Vasu Dev <vasu.dev@linux.intel.com>, Vu Pham <vu@mellanox.com>
Subject: Re: [PATCH-v2 4/4] target: Obtain se_node_acl->acl_kref during get_initiator_node_acl
Date: Wed, 13 Jan 2016 00:00:28 -0800	[thread overview]
Message-ID: <1452672028.27508.40.camel@haakon3.risingtidesystems.com> (raw)
In-Reply-To: <20160112150941.GC2058@lst.de>

On Tue, 2016-01-12 at 16:09 +0100, Christoph Hellwig wrote:
> > +++ b/drivers/target/target_core_tpg.c
> > @@ -75,9 +75,16 @@ struct se_node_acl *core_tpg_get_initiator_node_acl(
> >  	unsigned char *initiatorname)
> >  {
> >  	struct se_node_acl *acl;
> > -
> > +	/*
> > +	 * Obtain the acl_kref now, which will be dropped upon the
> > +	 * release of se_sess memory within transport_free_session().
> > +	 */
> 
> As said before this is incorrect, or rather shold be incorrect.  I
> think it should be removed, but if you want to keep it should be
> in core_tpg_check_initiator_node_acl and not here.
> 

Done.

Here's a incremental patch being folded in for -v3:

diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index 52ae8ba..62f02ea 100644
--- a/drivers/target/target_core_tpg.c
+++ b/drivers/target/target_core_tpg.c
@@ -75,16 +75,9 @@ struct se_node_acl *core_tpg_get_initiator_node_acl(
        unsigned char *initiatorname)
 {
        struct se_node_acl *acl;
-       /*
-        * Obtain the acl_kref now, which will be dropped upon the
-        * release of se_sess memory within transport_free_session().
-        */
+
        mutex_lock(&tpg->acl_node_mutex);
        acl = __core_tpg_get_initiator_node_acl(tpg, initiatorname);
-       if (acl) {
-               if (!kref_get_unless_zero(&acl->acl_kref))
-                       acl = NULL;
-       }
        mutex_unlock(&tpg->acl_node_mutex);
 
        return acl;
@@ -236,11 +229,24 @@ struct se_node_acl *core_tpg_check_initiator_node_acl(
        unsigned char *initiatorname)
 {
        struct se_node_acl *acl;
+       /*
+        * Obtain the acl_kref now, which will be dropped upon the
+        * release of se_sess memory within transport_free_session().
+        */
+       mutex_lock(&tpg->acl_node_mutex);
+       acl = __core_tpg_get_initiator_node_acl(tpg, initiatorname);
+       if (acl) {
+               if (!kref_get_unless_zero(&acl->acl_kref))
+                       acl = NULL;
+               mutex_unlock(&tpg->acl_node_mutex);
+               if (!acl)
+                       goto check_demo_mode;
 
-       acl = core_tpg_get_initiator_node_acl(tpg, initiatorname);
-       if (acl)
                return acl;
+       }
+       mutex_unlock(&tpg->acl_node_mutex);
 
+check_demo_mode:
        if (!tpg->se_tpg_tfo->tpg_check_demo_mode(tpg))
                return NULL;
> 
> 
> 
> The call of core_tpg_get_initiator_node_acl in
> iscsit_build_sendtargets_response needs a put to balance out the
> get of this kref while we're at it.

How about the following..?

diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index a81c0e5..4125520 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -3383,6 +3383,7 @@ iscsit_build_sendtargets_response(struct iscsi_cmd *cmd,
        struct iscsi_portal_group *tpg;
        struct iscsi_tiqn *tiqn;
        struct iscsi_tpg_np *tpg_np;
+       struct se_node_acl *se_acl;
        int buffer_len, end_of_buf = 0, len = 0, payload_len = 0;
        int target_name_printed;
        unsigned char buf[ISCSI_IQN_LEN+12]; /* iqn + "TargetName=" + \0 */
@@ -3435,10 +3436,14 @@ iscsit_build_sendtargets_response(struct iscsi_cmd *cmd,
 
                        if ((tpg->tpg_attrib.generate_node_acls == 0) &&
                            (tpg->tpg_attrib.demo_mode_discovery == 0) &&
-                           (!core_tpg_get_initiator_node_acl(&tpg->tpg_se_tpg,
-                               cmd->conn->sess->sess_ops->InitiatorName))) {
+                           (!(se_acl = core_tpg_get_initiator_node_acl(&tpg->tpg_se_tpg,
+                               cmd->conn->sess->sess_ops->InitiatorName)))) {
                                continue;
                        }
+                       if (se_acl) {
+                               target_put_nacl(se_acl);
+                               se_acl = NULL;
+                       }
 
                        spin_lock(&tpg->tpg_state_lock);
                        active = (tpg->tpg_state == TPG_STATE_ACTIVE);
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index c5035b9..a7c1bb5 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -431,6 +431,7 @@ void target_put_nacl(struct se_node_acl *nacl)
 {
        kref_put(&nacl->acl_kref, target_complete_nacl);
 }
+EXPORT_SYMBOL(target_put_nacl);
 
 void transport_deregister_session_configfs(struct se_session *se_sess)
 {

  reply	other threads:[~2016-01-13  8:00 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-01-10 20:28 [PATCH-v2 0/4] target: Close se_node_acl lookup race Nicholas A. Bellinger
2016-01-10 20:28 ` [PATCH-v2 1/4] tcm_fc: Convert acl lookup to modern get_initiator_node_acl usage Nicholas A. Bellinger
2016-01-11 21:05   ` Bart Van Assche
2016-01-12 15:02   ` Christoph Hellwig
2016-01-13  7:19     ` Nicholas A. Bellinger
2016-01-10 20:28 ` [PATCH-v2 2/4] ib_srpt: " Nicholas A. Bellinger
2016-01-11 21:34   ` Bart Van Assche
2016-01-13 16:39     ` Sagi Grimberg
2016-01-10 20:28 ` [PATCH-v2 3/4] target: Fix change depth se_session reference usage Nicholas A. Bellinger
2016-01-12 15:07   ` Christoph Hellwig
2016-01-13  7:29     ` Nicholas A. Bellinger
2016-01-13  8:24       ` Christoph Hellwig
2016-01-13  8:56         ` Nicholas A. Bellinger
2016-01-13  8:27   ` Christoph Hellwig
2016-01-13  8:58     ` Nicholas A. Bellinger
2016-01-13 16:45   ` Sagi Grimberg
2016-01-10 20:28 ` [PATCH-v2 4/4] target: Obtain se_node_acl->acl_kref during get_initiator_node_acl Nicholas A. Bellinger
2016-01-12 15:09   ` Christoph Hellwig
2016-01-13  8:00     ` Nicholas A. Bellinger [this message]
2016-01-13  8:15       ` Christoph Hellwig
2016-01-13  8:38         ` Nicholas A. Bellinger
2016-01-13  8:49           ` Christoph Hellwig

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=1452672028.27508.40.camel@haakon3.risingtidesystems.com \
    --to=nab@linux-iscsi.org \
    --cc=agrover@redhat.com \
    --cc=hare@suse.de \
    --cc=hch@lst.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-scsi@vger.kernel.org \
    --cc=nab@daterainc.com \
    --cc=sagig@mellanox.com \
    --cc=target-devel@vger.kernel.org \
    --cc=vasu.dev@linux.intel.com \
    --cc=vu@mellanox.com \
    /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®