Resubmitting it again(Probably last one had some format issues).
This patch fixes an unbalanced refcount issue.
Elevating the lock for both kref_put and also for controller node deletion.
Previously, controller deletion was protected but the not the kref_put. This was causing
the other thread to pick up the controller structure which was already kref'd zero.
This was causing the following BUG_ON and also sometimes panic.
WARNING: at lib/kref.c:43 kref_get+0x2d/0x30() (Not tainted)
Hardware name: IBM System x3655 -[7985AC1]-
Modules linked in: fuse scsi_dh_rdac autofs4 nfs lockd fscache nfs_acl
auth_rpcgss sunrpc 8021q garp stp llc ipv6 ib_srp(U) scsi_transport_srp
scsi_tgt ib_cm(U) ib_sa(U) ib_uverbs(U) ib_umad(U) mlx4_ib(U) mlx4_core(U)
ib_mthca(U) ib_mad(U) ib_core(U) dm_mirror dm_region_hash dm_log dm_round_robin
dm_multipath uinput bnx2 ses enclosure sg ibmpex ibmaem ipmi_msghandler
serio_raw k8temp hwmon amd64_edac_mod edac_core edac_mce_amd shpchp i2c_piix4
ext4 mbcache jbd2 sr_mod cdrom sd_mod crc_t10dif sata_svw pata_acpi ata_generic
pata_serverworks aacraid radeon ttm drm_kms_helper drm i2c_algo_bit i2c_core
dm_mod [last unloaded: freq_table]
Pid: 13735, comm: srp_daemon Not tainted 2.6.32-71.el6.x86_64 #1
Call Trace:
[<ffffffff8106b857>] warn_slowpath_common+0x87/0xc0
[<ffffffff8106b8aa>] warn_slowpath_null+0x1a/0x20
[<ffffffff8125c39d>] kref_get+0x2d/0x30
[<ffffffffa01b4029>] rdac_bus_attach+0x459/0x580 [scsi_dh_rdac]
[<ffffffff8135232a>] scsi_dh_handler_attach+0x2a/0x80
[<ffffffff81352c7b>] scsi_dh_notifier+0x9b/0xa0
[<ffffffff814cd7a5>] notifier_call_chain+0x55/0x80
[<ffffffff8109711a>] __blocking_notifier_call_chain+0x5a/0x80
[<ffffffff81097156>] blocking_notifier_call_chain+0x16/0x20
[<ffffffff8132bec5>] device_add+0x515/0x640
[<ffffffff813329e4>] ? attribute_container_device_trigger+0xc4/0xe0
[<ffffffff8134f659>] scsi_sysfs_add_sdev+0x89/0x2c0
[<ffffffff8134d096>] scsi_probe_and_add_lun+0xea6/0xed0
[<ffffffff8134beb2>] ? scsi_alloc_target+0x292/0x2d0
[<ffffffff8134d1e1>] __scsi_scan_target+0x121/0x750
[<ffffffff811df806>] ? sysfs_create_file+0x26/0x30
[<ffffffff8132b759>] ? device_create_file+0x19/0x20
[<ffffffff81332838>] ? attribute_container_add_attrs+0x78/0x90
[<ffffffff814b008c>] ? klist_next+0x4c/0xf0
[<ffffffff81332e30>] ? transport_configure+0x0/0x20
[<ffffffff813329e4>] ? attribute_container_device_trigger+0xc4/0xe0
[<ffffffff8134df40>] scsi_scan_target+0xd0/0xe0
[<ffffffffa02f053a>] srp_create_target+0x75a/0x890 [ib_srp]
[<ffffffff8132a130>] dev_attr_store+0x20/0x30
[<ffffffff811df145>] sysfs_write_file+0xe5/0x170
[<ffffffff8116c818>] vfs_write+0xb8/0x1a0
[<ffffffff810d40a2>] ? audit_syscall_entry+0x272/0x2a0
[<ffffffff8116d251>] sys_write+0x51/0x90
[<ffffffff81013172>] system_call_fastpath+0x16/0x1b
Signed-off-by: Babu Moger <[email protected]>
Acked-by: Mike Snitzer <[email protected]>
---
--- linux-3.3-rc1-new/drivers/scsi/device_handler/scsi_dh_rdac.c.orig 2012-01-19 17:04:48.000000000 -0600
+++ linux-3.3-rc1-new/drivers/scsi/device_handler/scsi_dh_rdac.c 2012-02-01 10:50:31.000000000 -0600
@@ -364,10 +364,7 @@ static void release_controller(struct kr
struct rdac_controller *ctlr;
ctlr = container_of(kref, struct rdac_controller, kref);
- flush_workqueue(kmpath_rdacd);
- spin_lock(&list_lock);
list_del(&ctlr->node);
- spin_unlock(&list_lock);
kfree(ctlr);
}
@@ -376,20 +373,17 @@ static struct rdac_controller *get_contr
{
struct rdac_controller *ctlr, *tmp;
- spin_lock(&list_lock);
-
list_for_each_entry(tmp, &ctlr_list, node) {
if ((memcmp(tmp->array_id, array_id, UNIQUE_ID_LEN) == 0) &&
(tmp->index == index) &&
(tmp->host == sdev->host)) {
kref_get(&tmp->kref);
- spin_unlock(&list_lock);
return tmp;
}
}
ctlr = kmalloc(sizeof(*ctlr), GFP_ATOMIC);
if (!ctlr)
- goto done;
+ return NULL;
/* initialize fields of controller */
memcpy(ctlr->array_id, array_id, UNIQUE_ID_LEN);
@@ -405,8 +399,7 @@ static struct rdac_controller *get_contr
INIT_WORK(&ctlr->ms_work, send_mode_select);
INIT_LIST_HEAD(&ctlr->ms_head);
list_add(&ctlr->node, &ctlr_list);
-done:
- spin_unlock(&list_lock);
+
return ctlr;
}
@@ -517,9 +510,12 @@ static int initialize_controller(struct
index = 0;
else
index = 1;
+
+ spin_lock(&list_lock);
h->ctlr = get_controller(index, array_name, array_id, sdev);
if (!h->ctlr)
err = SCSI_DH_RES_TEMP_UNAVAIL;
+ spin_unlock(&list_lock);
}
return err;
}
@@ -906,7 +902,9 @@ static int rdac_bus_attach(struct scsi_d
return 0;
clean_ctlr:
+ spin_lock(&list_lock);
kref_put(&h->ctlr->kref, release_controller);
+ spin_unlock(&list_lock);
failed:
kfree(scsi_dh_data);
@@ -921,14 +919,19 @@ static void rdac_bus_detach( struct scsi
struct rdac_dh_data *h;
unsigned long flags;
- spin_lock_irqsave(sdev->request_queue->queue_lock, flags);
scsi_dh_data = sdev->scsi_dh_data;
+ h = (struct rdac_dh_data *) scsi_dh_data->buf;
+ if (h->ctlr && h->ctlr->ms_queued)
+ flush_workqueue(kmpath_rdacd);
+
+ spin_lock_irqsave(sdev->request_queue->queue_lock, flags);
sdev->scsi_dh_data = NULL;
spin_unlock_irqrestore(sdev->request_queue->queue_lock, flags);
- h = (struct rdac_dh_data *) scsi_dh_data->buf;
+ spin_lock(&list_lock);
if (h->ctlr)
kref_put(&h->ctlr->kref, release_controller);
+ spin_unlock(&list_lock);
kfree(scsi_dh_data);
module_put(THIS_MODULE);
sdev_printk(KERN_NOTICE, sdev, "%s: Detached\n", RDAC_NAME);
Reviewed-by: Chandra Seetharaman <[email protected]>
On Thu, 2012-02-02 at 15:21 +0000, Moger, Babu wrote:
> Resubmitting it again(Probably last one had some format issues).
>
> This patch fixes an unbalanced refcount issue.
>
> Elevating the lock for both kref_put and also for controller node deletion.
> Previously, controller deletion was protected but the not the kref_put. This was causing
> the other thread to pick up the controller structure which was already kref'd zero.
>
> This was causing the following BUG_ON and also sometimes panic.
>
> WARNING: at lib/kref.c:43 kref_get+0x2d/0x30() (Not tainted)
> Hardware name: IBM System x3655 -[7985AC1]-
> Modules linked in: fuse scsi_dh_rdac autofs4 nfs lockd fscache nfs_acl
> auth_rpcgss sunrpc 8021q garp stp llc ipv6 ib_srp(U) scsi_transport_srp
> scsi_tgt ib_cm(U) ib_sa(U) ib_uverbs(U) ib_umad(U) mlx4_ib(U) mlx4_core(U)
> ib_mthca(U) ib_mad(U) ib_core(U) dm_mirror dm_region_hash dm_log dm_round_robin
> dm_multipath uinput bnx2 ses enclosure sg ibmpex ibmaem ipmi_msghandler
> serio_raw k8temp hwmon amd64_edac_mod edac_core edac_mce_amd shpchp i2c_piix4
> ext4 mbcache jbd2 sr_mod cdrom sd_mod crc_t10dif sata_svw pata_acpi ata_generic
> pata_serverworks aacraid radeon ttm drm_kms_helper drm i2c_algo_bit i2c_core
> dm_mod [last unloaded: freq_table]
> Pid: 13735, comm: srp_daemon Not tainted 2.6.32-71.el6.x86_64 #1
> Call Trace:
> [<ffffffff8106b857>] warn_slowpath_common+0x87/0xc0
> [<ffffffff8106b8aa>] warn_slowpath_null+0x1a/0x20
> [<ffffffff8125c39d>] kref_get+0x2d/0x30
> [<ffffffffa01b4029>] rdac_bus_attach+0x459/0x580 [scsi_dh_rdac]
> [<ffffffff8135232a>] scsi_dh_handler_attach+0x2a/0x80
> [<ffffffff81352c7b>] scsi_dh_notifier+0x9b/0xa0
> [<ffffffff814cd7a5>] notifier_call_chain+0x55/0x80
> [<ffffffff8109711a>] __blocking_notifier_call_chain+0x5a/0x80
> [<ffffffff81097156>] blocking_notifier_call_chain+0x16/0x20
> [<ffffffff8132bec5>] device_add+0x515/0x640
> [<ffffffff813329e4>] ? attribute_container_device_trigger+0xc4/0xe0
> [<ffffffff8134f659>] scsi_sysfs_add_sdev+0x89/0x2c0
> [<ffffffff8134d096>] scsi_probe_and_add_lun+0xea6/0xed0
> [<ffffffff8134beb2>] ? scsi_alloc_target+0x292/0x2d0
> [<ffffffff8134d1e1>] __scsi_scan_target+0x121/0x750
> [<ffffffff811df806>] ? sysfs_create_file+0x26/0x30
> [<ffffffff8132b759>] ? device_create_file+0x19/0x20
> [<ffffffff81332838>] ? attribute_container_add_attrs+0x78/0x90
> [<ffffffff814b008c>] ? klist_next+0x4c/0xf0
> [<ffffffff81332e30>] ? transport_configure+0x0/0x20
> [<ffffffff813329e4>] ? attribute_container_device_trigger+0xc4/0xe0
> [<ffffffff8134df40>] scsi_scan_target+0xd0/0xe0
> [<ffffffffa02f053a>] srp_create_target+0x75a/0x890 [ib_srp]
> [<ffffffff8132a130>] dev_attr_store+0x20/0x30
> [<ffffffff811df145>] sysfs_write_file+0xe5/0x170
> [<ffffffff8116c818>] vfs_write+0xb8/0x1a0
> [<ffffffff810d40a2>] ? audit_syscall_entry+0x272/0x2a0
> [<ffffffff8116d251>] sys_write+0x51/0x90
> [<ffffffff81013172>] system_call_fastpath+0x16/0x1b
>
>
> Signed-off-by: Babu Moger <[email protected]>
> Acked-by: Mike Snitzer <[email protected]>
> ---
> --- linux-3.3-rc1-new/drivers/scsi/device_handler/scsi_dh_rdac.c.orig 2012-01-19 17:04:48.000000000 -0600
> +++ linux-3.3-rc1-new/drivers/scsi/device_handler/scsi_dh_rdac.c 2012-02-01 10:50:31.000000000 -0600
> @@ -364,10 +364,7 @@ static void release_controller(struct kr
> struct rdac_controller *ctlr;
> ctlr = container_of(kref, struct rdac_controller, kref);
>
> - flush_workqueue(kmpath_rdacd);
> - spin_lock(&list_lock);
> list_del(&ctlr->node);
> - spin_unlock(&list_lock);
> kfree(ctlr);
> }
>
> @@ -376,20 +373,17 @@ static struct rdac_controller *get_contr
> {
> struct rdac_controller *ctlr, *tmp;
>
> - spin_lock(&list_lock);
> -
> list_for_each_entry(tmp, &ctlr_list, node) {
> if ((memcmp(tmp->array_id, array_id, UNIQUE_ID_LEN) == 0) &&
> (tmp->index == index) &&
> (tmp->host == sdev->host)) {
> kref_get(&tmp->kref);
> - spin_unlock(&list_lock);
> return tmp;
> }
> }
> ctlr = kmalloc(sizeof(*ctlr), GFP_ATOMIC);
> if (!ctlr)
> - goto done;
> + return NULL;
>
> /* initialize fields of controller */
> memcpy(ctlr->array_id, array_id, UNIQUE_ID_LEN);
> @@ -405,8 +399,7 @@ static struct rdac_controller *get_contr
> INIT_WORK(&ctlr->ms_work, send_mode_select);
> INIT_LIST_HEAD(&ctlr->ms_head);
> list_add(&ctlr->node, &ctlr_list);
> -done:
> - spin_unlock(&list_lock);
> +
> return ctlr;
> }
>
> @@ -517,9 +510,12 @@ static int initialize_controller(struct
> index = 0;
> else
> index = 1;
> +
> + spin_lock(&list_lock);
> h->ctlr = get_controller(index, array_name, array_id, sdev);
> if (!h->ctlr)
> err = SCSI_DH_RES_TEMP_UNAVAIL;
> + spin_unlock(&list_lock);
> }
> return err;
> }
> @@ -906,7 +902,9 @@ static int rdac_bus_attach(struct scsi_d
> return 0;
>
> clean_ctlr:
> + spin_lock(&list_lock);
> kref_put(&h->ctlr->kref, release_controller);
> + spin_unlock(&list_lock);
>
> failed:
> kfree(scsi_dh_data);
> @@ -921,14 +919,19 @@ static void rdac_bus_detach( struct scsi
> struct rdac_dh_data *h;
> unsigned long flags;
>
> - spin_lock_irqsave(sdev->request_queue->queue_lock, flags);
> scsi_dh_data = sdev->scsi_dh_data;
> + h = (struct rdac_dh_data *) scsi_dh_data->buf;
> + if (h->ctlr && h->ctlr->ms_queued)
> + flush_workqueue(kmpath_rdacd);
> +
> + spin_lock_irqsave(sdev->request_queue->queue_lock, flags);
> sdev->scsi_dh_data = NULL;
> spin_unlock_irqrestore(sdev->request_queue->queue_lock, flags);
>
> - h = (struct rdac_dh_data *) scsi_dh_data->buf;
> + spin_lock(&list_lock);
> if (h->ctlr)
> kref_put(&h->ctlr->kref, release_controller);
> + spin_unlock(&list_lock);
> kfree(scsi_dh_data);
> module_put(THIS_MODULE);
> sdev_printk(KERN_NOTICE, sdev, "%s: Detached\n", RDAC_NAME);
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
James, Can you please pick this up for the next update.
> -----Original Message-----
> From: Chandra Seetharaman [mailto:[email protected]]
> Sent: Thursday, February 02, 2012 1:44 PM
> To: Moger, Babu
> Cc: [email protected]; device-mapper development; Mike
> Snitzer
> Subject: Re: [PATCH] scsi_dh_rdac: Fix for unbalanced reference count
>
> Reviewed-by: Chandra Seetharaman <[email protected]>
>
> On Thu, 2012-02-02 at 15:21 +0000, Moger, Babu wrote:
> > Resubmitting it again(Probably last one had some format issues).
> >
> > This patch fixes an unbalanced refcount issue.
> >
> > Elevating the lock for both kref_put and also for controller node
> deletion.
> > Previously, controller deletion was protected but the not the
> kref_put. This was causing
> > the other thread to pick up the controller structure which was
> already kref'd zero.
> >
> > This was causing the following BUG_ON and also sometimes panic.
> >
> > WARNING: at lib/kref.c:43 kref_get+0x2d/0x30() (Not tainted)
> > Hardware name: IBM System x3655 -[7985AC1]-
> > Modules linked in: fuse scsi_dh_rdac autofs4 nfs lockd fscache
> nfs_acl
> > auth_rpcgss sunrpc 8021q garp stp llc ipv6 ib_srp(U)
> scsi_transport_srp
> > scsi_tgt ib_cm(U) ib_sa(U) ib_uverbs(U) ib_umad(U) mlx4_ib(U)
> mlx4_core(U)
> > ib_mthca(U) ib_mad(U) ib_core(U) dm_mirror dm_region_hash dm_log
> dm_round_robin
> > dm_multipath uinput bnx2 ses enclosure sg ibmpex ibmaem
> ipmi_msghandler
> > serio_raw k8temp hwmon amd64_edac_mod edac_core edac_mce_amd shpchp
> i2c_piix4
> > ext4 mbcache jbd2 sr_mod cdrom sd_mod crc_t10dif sata_svw pata_acpi
> ata_generic
> > pata_serverworks aacraid radeon ttm drm_kms_helper drm i2c_algo_bit
> i2c_core
> > dm_mod [last unloaded: freq_table]
> > Pid: 13735, comm: srp_daemon Not tainted 2.6.32-71.el6.x86_64 #1
> > Call Trace:
> > [<ffffffff8106b857>] warn_slowpath_common+0x87/0xc0
> > [<ffffffff8106b8aa>] warn_slowpath_null+0x1a/0x20
> > [<ffffffff8125c39d>] kref_get+0x2d/0x30
> > [<ffffffffa01b4029>] rdac_bus_attach+0x459/0x580 [scsi_dh_rdac]
> > [<ffffffff8135232a>] scsi_dh_handler_attach+0x2a/0x80
> > [<ffffffff81352c7b>] scsi_dh_notifier+0x9b/0xa0
> > [<ffffffff814cd7a5>] notifier_call_chain+0x55/0x80
> > [<ffffffff8109711a>] __blocking_notifier_call_chain+0x5a/0x80
> > [<ffffffff81097156>] blocking_notifier_call_chain+0x16/0x20
> > [<ffffffff8132bec5>] device_add+0x515/0x640
> > [<ffffffff813329e4>] ? attribute_container_device_trigger+0xc4/0xe0
> > [<ffffffff8134f659>] scsi_sysfs_add_sdev+0x89/0x2c0
> > [<ffffffff8134d096>] scsi_probe_and_add_lun+0xea6/0xed0
> > [<ffffffff8134beb2>] ? scsi_alloc_target+0x292/0x2d0
> > [<ffffffff8134d1e1>] __scsi_scan_target+0x121/0x750
> > [<ffffffff811df806>] ? sysfs_create_file+0x26/0x30
> > [<ffffffff8132b759>] ? device_create_file+0x19/0x20
> > [<ffffffff81332838>] ? attribute_container_add_attrs+0x78/0x90
> > [<ffffffff814b008c>] ? klist_next+0x4c/0xf0
> > [<ffffffff81332e30>] ? transport_configure+0x0/0x20
> > [<ffffffff813329e4>] ? attribute_container_device_trigger+0xc4/0xe0
> > [<ffffffff8134df40>] scsi_scan_target+0xd0/0xe0
> > [<ffffffffa02f053a>] srp_create_target+0x75a/0x890 [ib_srp]
> > [<ffffffff8132a130>] dev_attr_store+0x20/0x30
> > [<ffffffff811df145>] sysfs_write_file+0xe5/0x170
> > [<ffffffff8116c818>] vfs_write+0xb8/0x1a0
> > [<ffffffff810d40a2>] ? audit_syscall_entry+0x272/0x2a0
> > [<ffffffff8116d251>] sys_write+0x51/0x90
> > [<ffffffff81013172>] system_call_fastpath+0x16/0x1b
> >
> >
> > Signed-off-by: Babu Moger <[email protected]>
> > Acked-by: Mike Snitzer <[email protected]>
> > ---
> > --- linux-3.3-rc1-new/drivers/scsi/device_handler/scsi_dh_rdac.c.orig
> 2012-01-19 17:04:48.000000000 -0600
> > +++ linux-3.3-rc1-new/drivers/scsi/device_handler/scsi_dh_rdac.c
> 2012-02-01 10:50:31.000000000 -0600
> > @@ -364,10 +364,7 @@ static void release_controller(struct kr
> > struct rdac_controller *ctlr;
> > ctlr = container_of(kref, struct rdac_controller, kref);
> >
> > - flush_workqueue(kmpath_rdacd);
> > - spin_lock(&list_lock);
> > list_del(&ctlr->node);
> > - spin_unlock(&list_lock);
> > kfree(ctlr);
> > }
> >
> > @@ -376,20 +373,17 @@ static struct rdac_controller *get_contr
> > {
> > struct rdac_controller *ctlr, *tmp;
> >
> > - spin_lock(&list_lock);
> > -
> > list_for_each_entry(tmp, &ctlr_list, node) {
> > if ((memcmp(tmp->array_id, array_id, UNIQUE_ID_LEN) == 0)
> &&
> > (tmp->index == index) &&
> > (tmp->host == sdev->host)) {
> > kref_get(&tmp->kref);
> > - spin_unlock(&list_lock);
> > return tmp;
> > }
> > }
> > ctlr = kmalloc(sizeof(*ctlr), GFP_ATOMIC);
> > if (!ctlr)
> > - goto done;
> > + return NULL;
> >
> > /* initialize fields of controller */
> > memcpy(ctlr->array_id, array_id, UNIQUE_ID_LEN);
> > @@ -405,8 +399,7 @@ static struct rdac_controller *get_contr
> > INIT_WORK(&ctlr->ms_work, send_mode_select);
> > INIT_LIST_HEAD(&ctlr->ms_head);
> > list_add(&ctlr->node, &ctlr_list);
> > -done:
> > - spin_unlock(&list_lock);
> > +
> > return ctlr;
> > }
> >
> > @@ -517,9 +510,12 @@ static int initialize_controller(struct
> > index = 0;
> > else
> > index = 1;
> > +
> > + spin_lock(&list_lock);
> > h->ctlr = get_controller(index, array_name, array_id,
> sdev);
> > if (!h->ctlr)
> > err = SCSI_DH_RES_TEMP_UNAVAIL;
> > + spin_unlock(&list_lock);
> > }
> > return err;
> > }
> > @@ -906,7 +902,9 @@ static int rdac_bus_attach(struct scsi_d
> > return 0;
> >
> > clean_ctlr:
> > + spin_lock(&list_lock);
> > kref_put(&h->ctlr->kref, release_controller);
> > + spin_unlock(&list_lock);
> >
> > failed:
> > kfree(scsi_dh_data);
> > @@ -921,14 +919,19 @@ static void rdac_bus_detach( struct scsi
> > struct rdac_dh_data *h;
> > unsigned long flags;
> >
> > - spin_lock_irqsave(sdev->request_queue->queue_lock, flags);
> > scsi_dh_data = sdev->scsi_dh_data;
> > + h = (struct rdac_dh_data *) scsi_dh_data->buf;
> > + if (h->ctlr && h->ctlr->ms_queued)
> > + flush_workqueue(kmpath_rdacd);
> > +
> > + spin_lock_irqsave(sdev->request_queue->queue_lock, flags);
> > sdev->scsi_dh_data = NULL;
> > spin_unlock_irqrestore(sdev->request_queue->queue_lock, flags);
> >
> > - h = (struct rdac_dh_data *) scsi_dh_data->buf;
> > + spin_lock(&list_lock);
> > if (h->ctlr)
> > kref_put(&h->ctlr->kref, release_controller);
> > + spin_unlock(&list_lock);
> > kfree(scsi_dh_data);
> > module_put(THIS_MODULE);
> > sdev_printk(KERN_NOTICE, sdev, "%s: Detached\n", RDAC_NAME);
> >
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-scsi"
> in
> > the body of a message to [email protected]
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> >
>
????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay? ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m???? ????????I?