linux_old1/drivers/target/target_core_tmr.c

434 lines
12 KiB
C
Raw Normal View History

/*******************************************************************************
* Filename: target_core_tmr.c
*
* This file contains SPC-3 task management infrastructure
*
* (c) Copyright 2009-2013 Datera, Inc.
*
* Nicholas A. Bellinger <nab@kernel.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
******************************************************************************/
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/list.h>
#include <linux/export.h>
#include <target/target_core_base.h>
#include <target/target_core_backend.h>
#include <target/target_core_fabric.h>
#include "target_core_internal.h"
#include "target_core_alua.h"
#include "target_core_pr.h"
int core_tmr_alloc_req(
struct se_cmd *se_cmd,
void *fabric_tmr_ptr,
u8 function,
gfp_t gfp_flags)
{
struct se_tmr_req *tmr;
tmr = kzalloc(sizeof(struct se_tmr_req), gfp_flags);
if (!tmr) {
pr_err("Unable to allocate struct se_tmr_req\n");
return -ENOMEM;
}
se_cmd->se_cmd_flags |= SCF_SCSI_TMR_CDB;
se_cmd->se_tmr_req = tmr;
tmr->task_cmd = se_cmd;
tmr->fabric_tmr_ptr = fabric_tmr_ptr;
tmr->function = function;
INIT_LIST_HEAD(&tmr->tmr_list);
return 0;
}
EXPORT_SYMBOL(core_tmr_alloc_req);
void core_tmr_release_req(struct se_tmr_req *tmr)
{
struct se_device *dev = tmr->tmr_dev;
unsigned long flags;
if (dev) {
spin_lock_irqsave(&dev->se_tmr_lock, flags);
list_del_init(&tmr->tmr_list);
spin_unlock_irqrestore(&dev->se_tmr_lock, flags);
target: Fix transport_get_lun_for_tmr failure cases This patch fixes two possible NULL pointer dereferences in target v4.0 code where se_tmr release path in core_tmr_release_req() can OOPs upon transport_get_lun_for_tmr() failure by attempting to access se_device or se_tmr->tmr_list without a valid member of se_device->tmr_list during transport_free_se_cmd() release. This patch moves the se_tmr->tmr_dev pointer assignment in transport_get_lun_for_tmr() until after possible -ENODEV failures during unpacked_lun lookup. This addresses an OOPs originally reported with LIO v4.1 upstream on .39 code here: TARGET_CORE[qla2xxx]: Detected NON_EXISTENT_LUN Access for 0x00000000 BUG: unable to handle kernel NULL pointer dereference at 0000000000000550 IP: [<ffffffff81035ec4>] __ticket_spin_trylock+0x4/0x20 PGD 0 Oops: 0000 [#1] SMP last sysfs file: /sys/devices/system/cpu/cpu23/cache/index2/shared_cpu_map CPU 1 Modules linked in: netconsole target_core_pscsi target_core_file tcm_qla2xxx target_core_iblock tcm_loop target_core_mod configfs ipmi_devintf ipmi_si ipmi_msghandler serio_raw i7core_edac ioatdma dca edac_core ps_bdrv ses enclosure usbhid usb_storage ahci qla2xxx hid uas e1000e mpt2sas libahci mlx4_core scsi_transport_fc scsi_transport_sas raid_class scsi_tgt [last unloaded: netconsole] Pid: 0, comm: kworker/0:0 Tainted: G W 2.6.39+ #1 Xyratex Storage Server RIP: 0010:[<ffffffff81035ec4>] [<ffffffff81035ec4>]__ticket_spin_trylock+0x4/0x20 RSP: 0018:ffff88063e803c08 EFLAGS: 00010286 RAX: ffff880619ab45e0 RBX: 0000000000000550 RCX: 0000000000000000 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000550 RBP: ffff88063e803c08 R08: 0000000000000002 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000568 R13: 0000000000000001 R14: 0000000000000000 R15: ffff88060cd96a20 FS: 0000000000000000(0000) GS:ffff88063e800000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000550 CR3: 0000000001a03000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process kworker/0:0 (pid: 0, threadinfo ffff880619ab8000, task ffff880619ab45e0) Stack: ffff88063e803c28 ffffffff812cf039 0000000000000550 0000000000000568 ffff88063e803c58 ffffffff8157071e ffffffffa028a1dc ffff88060f7e4600 0000000000000550 ffff880616961480 ffff88063e803c78 ffffffffa028a1dc Call Trace: <IRQ> [<ffffffff812cf039>] do_raw_spin_trylock+0x19/0x50 [<ffffffff8157071e>] _raw_spin_lock+0x3e/0x70 [<ffffffffa028a1dc>] ? core_tmr_release_req+0x2c/0x60 [target_core_mod] [<ffffffffa028a1dc>] core_tmr_release_req+0x2c/0x60 [target_core_mod] [<ffffffffa028d0d2>] transport_free_se_cmd+0x22/0x50 [target_core_mod] [<ffffffffa028d120>] transport_release_cmd_to_pool+0x20/0x40 [target_core_mod] [<ffffffffa028e525>] transport_generic_free_cmd+0xa5/0xb0 [target_core_mod] [<ffffffffa0147cc4>] tcm_qla2xxx_handle_tmr+0xc4/0xd0 [tcm_qla2xxx] [<ffffffffa0191ba3>] __qla24xx_handle_abts+0xd3/0x150 [qla2xxx] [<ffffffffa0197651>] qla_tgt_response_pkt+0x171/0x520 [qla2xxx] [<ffffffffa0197a2d>] qla_tgt_response_pkt_all_vps+0x2d/0x220 [qla2xxx] [<ffffffffa0171dd3>] qla24xx_process_response_queue+0x1a3/0x670 [qla2xxx] [<ffffffffa0196281>] ? qla24xx_atio_pkt+0x81/0x120 [qla2xxx] [<ffffffffa0174025>] ? qla24xx_msix_default+0x45/0x2a0 [qla2xxx] [<ffffffffa0174198>] qla24xx_msix_default+0x1b8/0x2a0 [qla2xxx] [<ffffffff810dadb4>] handle_irq_event_percpu+0x54/0x210 [<ffffffff810dafb8>] handle_irq_event+0x48/0x70 [<ffffffff810dd5ee>] ? handle_edge_irq+0x1e/0x110 [<ffffffff810dd647>] handle_edge_irq+0x77/0x110 [<ffffffff8100d362>] handle_irq+0x22/0x40 [<ffffffff8157b28d>] do_IRQ+0x5d/0xe0 [<ffffffff81571413>] common_interrupt+0x13/0x13 <EOI> [<ffffffff813003f7>] ? intel_idle+0xd7/0x130 [<ffffffff813003f0>] ? intel_idle+0xd0/0x130 [<ffffffff8144832b>] cpuidle_idle_call+0xab/0x1c0 [<ffffffff8100a26b>] cpu_idle+0xab/0xf0 [<ffffffff81566c59>] start_secondary+0x1cb/0x1d2 Reported-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2011-06-24 07:48:32 +08:00
}
kfree(tmr);
}
static int target_check_cdb_and_preempt(struct list_head *list,
struct se_cmd *cmd)
{
struct t10_pr_registration *reg;
if (!list)
return 0;
list_for_each_entry(reg, list, pr_reg_abort_list) {
if (reg->pr_res_key == cmd->pr_res_key)
return 0;
}
return 1;
}
static bool __target_check_io_state(struct se_cmd *se_cmd,
struct se_session *tmr_sess, int tas)
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
{
struct se_session *sess = se_cmd->se_sess;
assert_spin_locked(&sess->sess_cmd_lock);
WARN_ON_ONCE(!irqs_disabled());
/*
* If command already reached CMD_T_COMPLETE state within
* target_complete_cmd() or CMD_T_FABRIC_STOP due to shutdown,
* this se_cmd has been passed to fabric driver and will
* not be aborted.
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
*
* Otherwise, obtain a local se_cmd->cmd_kref now for TMR
* ABORT_TASK + LUN_RESET for CMD_T_ABORTED processing as
* long as se_cmd->cmd_kref is still active unless zero.
*/
spin_lock(&se_cmd->t_state_lock);
if (se_cmd->transport_state & (CMD_T_COMPLETE | CMD_T_FABRIC_STOP)) {
pr_debug("Attempted to abort io tag: %llu already complete or"
" fabric stop, skipping\n", se_cmd->tag);
spin_unlock(&se_cmd->t_state_lock);
return false;
}
if (se_cmd->transport_state & CMD_T_PRE_EXECUTE) {
if (se_cmd->scsi_status) {
pr_debug("Attempted to abort io tag: %llu early failure"
" status: 0x%02x\n", se_cmd->tag,
se_cmd->scsi_status);
spin_unlock(&se_cmd->t_state_lock);
return false;
}
}
scsi: target: Make the session shutdown code also wait for commands that are being aborted Target drivers must call target_sess_cmd_list_set_waiting() and target_wait_for_sess_cmds() before freeing a session. Since freeing a session is only safe after all commands that are associated with a session have finished, make target_wait_for_sess_cmds() also wait for commands that are being aborted. Instead of setting a flag in each pending command from target_sess_cmd_list_set_waiting() and waiting in target_wait_for_sess_cmds() on a per-command completion, only set a per-session flag in the former function and wait on a per-session completion in the latter function. This change is safe because once a SCSI initiator system has submitted a command a target system is always allowed to execute it to completion. See also commit 0f4a943168f3 ("target: Fix remote-port TMR ABORT + se_cmd fabric stop"). This patch is based on the following two patches: * Bart Van Assche, target: Simplify session shutdown code, February 19, 2015 (https://github.com/bvanassche/linux/commit/8df5463d7d7619f2f1b70cfe5172eaef0aa52815). * Christoph Hellwig, target: Rework session shutdown code, December 7, 2015 (http://thread.gmane.org/gmane.linux.scsi.target.devel/10695). Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2018-06-23 05:52:53 +08:00
if (sess->sess_tearing_down) {
pr_debug("Attempted to abort io tag: %llu already shutdown,"
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
" skipping\n", se_cmd->tag);
spin_unlock(&se_cmd->t_state_lock);
return false;
}
se_cmd->transport_state |= CMD_T_ABORTED;
if ((tmr_sess != se_cmd->se_sess) && tas)
se_cmd->transport_state |= CMD_T_TAS;
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
spin_unlock(&se_cmd->t_state_lock);
return kref_get_unless_zero(&se_cmd->cmd_kref);
}
void core_tmr_abort_task(
struct se_device *dev,
struct se_tmr_req *tmr,
struct se_session *se_sess)
{
struct se_cmd *se_cmd;
unsigned long flags;
u64 ref_tag;
spin_lock_irqsave(&se_sess->sess_cmd_lock, flags);
list_for_each_entry(se_cmd, &se_sess->sess_cmd_list, se_cmd_list) {
if (dev != se_cmd->se_dev)
continue;
/* skip task management functions, including tmr->task_cmd */
if (se_cmd->se_cmd_flags & SCF_SCSI_TMR_CDB)
continue;
ref_tag = se_cmd->tag;
if (tmr->ref_task_tag != ref_tag)
continue;
printk("ABORT_TASK: Found referenced %s task_tag: %llu\n",
se_cmd->se_tfo->get_fabric_name(), ref_tag);
if (!__target_check_io_state(se_cmd, se_sess, 0))
continue;
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
cancel_work_sync(&se_cmd->work);
transport_wait_for_tasks(se_cmd);
if (!transport_cmd_finish_abort(se_cmd))
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 11:00:17 +08:00
target_put_sess_cmd(se_cmd);
printk("ABORT_TASK: Sending TMR_FUNCTION_COMPLETE for"
" ref_tag: %llu\n", ref_tag);
tmr->response = TMR_FUNCTION_COMPLETE;
atomic_long_inc(&dev->aborts_complete);
return;
}
spin_unlock_irqrestore(&se_sess->sess_cmd_lock, flags);
printk("ABORT_TASK: Sending TMR_TASK_DOES_NOT_EXIST for ref_tag: %lld\n",
tmr->ref_task_tag);
tmr->response = TMR_TASK_DOES_NOT_EXIST;
atomic_long_inc(&dev->aborts_no_task);
}
static void core_tmr_drain_tmr_list(
struct se_device *dev,
struct se_tmr_req *tmr,
struct list_head *preempt_and_abort_list)
{
LIST_HEAD(drain_tmr_list);
struct se_session *sess;
struct se_tmr_req *tmr_p, *tmr_pp;
struct se_cmd *cmd;
unsigned long flags;
bool rc;
/*
* Release all pending and outgoing TMRs aside from the received
* LUN_RESET tmr..
*/
spin_lock_irqsave(&dev->se_tmr_lock, flags);
if (tmr)
list_del_init(&tmr->tmr_list);
list_for_each_entry_safe(tmr_p, tmr_pp, &dev->dev_tmr_list, tmr_list) {
cmd = tmr_p->task_cmd;
if (!cmd) {
pr_err("Unable to locate struct se_cmd for TMR\n");
continue;
}
/*
* If this function was called with a valid pr_res_key
* parameter (eg: for PROUT PREEMPT_AND_ABORT service action
* skip non registration key matching TMRs.
*/
if (target_check_cdb_and_preempt(preempt_and_abort_list, cmd))
continue;
sess = cmd->se_sess;
if (WARN_ON_ONCE(!sess))
continue;
spin_lock(&sess->sess_cmd_lock);
spin_lock(&cmd->t_state_lock);
if (!(cmd->transport_state & CMD_T_ACTIVE) ||
(cmd->transport_state & CMD_T_FABRIC_STOP)) {
spin_unlock(&cmd->t_state_lock);
spin_unlock(&sess->sess_cmd_lock);
continue;
}
if (cmd->t_state == TRANSPORT_ISTATE_PROCESSING) {
spin_unlock(&cmd->t_state_lock);
spin_unlock(&sess->sess_cmd_lock);
continue;
}
scsi: target: Make the session shutdown code also wait for commands that are being aborted Target drivers must call target_sess_cmd_list_set_waiting() and target_wait_for_sess_cmds() before freeing a session. Since freeing a session is only safe after all commands that are associated with a session have finished, make target_wait_for_sess_cmds() also wait for commands that are being aborted. Instead of setting a flag in each pending command from target_sess_cmd_list_set_waiting() and waiting in target_wait_for_sess_cmds() on a per-command completion, only set a per-session flag in the former function and wait on a per-session completion in the latter function. This change is safe because once a SCSI initiator system has submitted a command a target system is always allowed to execute it to completion. See also commit 0f4a943168f3 ("target: Fix remote-port TMR ABORT + se_cmd fabric stop"). This patch is based on the following two patches: * Bart Van Assche, target: Simplify session shutdown code, February 19, 2015 (https://github.com/bvanassche/linux/commit/8df5463d7d7619f2f1b70cfe5172eaef0aa52815). * Christoph Hellwig, target: Rework session shutdown code, December 7, 2015 (http://thread.gmane.org/gmane.linux.scsi.target.devel/10695). Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2018-06-23 05:52:53 +08:00
if (sess->sess_tearing_down) {
spin_unlock(&cmd->t_state_lock);
spin_unlock(&sess->sess_cmd_lock);
continue;
}
cmd->transport_state |= CMD_T_ABORTED;
spin_unlock(&cmd->t_state_lock);
rc = kref_get_unless_zero(&cmd->cmd_kref);
if (!rc) {
printk("LUN_RESET TMR: non-zero kref_get_unless_zero\n");
spin_unlock(&sess->sess_cmd_lock);
continue;
}
spin_unlock(&sess->sess_cmd_lock);
list_move_tail(&tmr_p->tmr_list, &drain_tmr_list);
}
spin_unlock_irqrestore(&dev->se_tmr_lock, flags);
list_for_each_entry_safe(tmr_p, tmr_pp, &drain_tmr_list, tmr_list) {
list_del_init(&tmr_p->tmr_list);
cmd = tmr_p->task_cmd;
pr_debug("LUN_RESET: %s releasing TMR %p Function: 0x%02x,"
" Response: 0x%02x, t_state: %d\n",
(preempt_and_abort_list) ? "Preempt" : "", tmr_p,
tmr_p->function, tmr_p->response, cmd->t_state);
cancel_work_sync(&cmd->work);
transport_wait_for_tasks(cmd);
if (!transport_cmd_finish_abort(cmd))
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 11:00:17 +08:00
target_put_sess_cmd(cmd);
}
}
static void core_tmr_drain_state_list(
struct se_device *dev,
struct se_cmd *prout_cmd,
struct se_session *tmr_sess,
int tas,
struct list_head *preempt_and_abort_list)
{
LIST_HEAD(drain_task_list);
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
struct se_session *sess;
struct se_cmd *cmd, *next;
unsigned long flags;
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
int rc;
/*
* Complete outstanding commands with TASK_ABORTED SAM status.
*
* This is following sam4r17, section 5.6 Aborting commands, Table 38
* for TMR LUN_RESET:
*
* a) "Yes" indicates that each command that is aborted on an I_T nexus
* other than the one that caused the SCSI device condition is
* completed with TASK ABORTED status, if the TAS bit is set to one in
* the Control mode page (see SPC-4). "No" indicates that no status is
* returned for aborted commands.
*
* d) If the logical unit reset is caused by a particular I_T nexus
* (e.g., by a LOGICAL UNIT RESET task management function), then "yes"
* (TASK_ABORTED status) applies.
*
* Otherwise (e.g., if triggered by a hard reset), "no"
* (no TASK_ABORTED SAM status) applies.
*
* Note that this seems to be independent of TAS (Task Aborted Status)
* in the Control Mode Page.
*/
spin_lock_irqsave(&dev->execute_task_lock, flags);
list_for_each_entry_safe(cmd, next, &dev->state_list, state_list) {
/*
* For PREEMPT_AND_ABORT usage, only process commands
* with a matching reservation key.
*/
if (target_check_cdb_and_preempt(preempt_and_abort_list, cmd))
continue;
/*
* Not aborting PROUT PREEMPT_AND_ABORT CDB..
*/
if (prout_cmd == cmd)
continue;
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
sess = cmd->se_sess;
if (WARN_ON_ONCE(!sess))
continue;
spin_lock(&sess->sess_cmd_lock);
rc = __target_check_io_state(cmd, tmr_sess, tas);
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
spin_unlock(&sess->sess_cmd_lock);
if (!rc)
continue;
list_move_tail(&cmd->state_list, &drain_task_list);
cmd->state_active = false;
}
spin_unlock_irqrestore(&dev->execute_task_lock, flags);
while (!list_empty(&drain_task_list)) {
cmd = list_entry(drain_task_list.next, struct se_cmd, state_list);
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
list_del_init(&cmd->state_list);
target_show_cmd("LUN_RESET: ", cmd);
pr_debug("LUN_RESET: ITT[0x%08llx] - %s pr_res_key: 0x%016Lx\n",
cmd->tag, (preempt_and_abort_list) ? "preempt" : "",
cmd->pr_res_key);
/*
* If the command may be queued onto a workqueue cancel it now.
*
* This is equivalent to removal from the execute queue in the
* loop above, but we do it down here given that
* cancel_work_sync may block.
*/
target: Fix LUN_RESET active I/O handling for ACK_KREF This patch fixes a NULL pointer se_cmd->cmd_kref < 0 refcount bug during TMR LUN_RESET with active se_cmd I/O, that can be triggered during se_cmd descriptor shutdown + release via core_tmr_drain_state_list() code. To address this bug, add common __target_check_io_state() helper for ABORT_TASK + LUN_RESET w/ CMD_T_COMPLETE checking, and set CMD_T_ABORTED + obtain ->cmd_kref for both cases ahead of last target_put_sess_cmd() after TFO->aborted_task() -> transport_cmd_finish_abort() callback has completed. It also introduces SCF_ACK_KREF to determine when transport_cmd_finish_abort() needs to drop the second extra reference, ahead of calling target_put_sess_cmd() for the final kref_put(&se_cmd->cmd_kref). It also updates transport_cmd_check_stop() to avoid holding se_cmd->t_state_lock while dropping se_cmd device state via target_remove_from_state_list(), now that core_tmr_drain_state_list() is holding the se_device lock while checking se_cmd state from within TMR logic. Finally, move transport_put_cmd() release of SGL + TMR + extended CDB memory into target_free_cmd_mem() in order to avoid potential resource leaks in TMR ABORT_TASK + LUN_RESET code-paths. Also update target_release_cmd_kref() accordingly. Reviewed-by: Quinn Tran <quinn.tran@qlogic.com> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.de> Cc: Andy Grover <agrover@redhat.com> Cc: Mike Christie <mchristi@redhat.com> Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2016-01-12 13:31:09 +08:00
cancel_work_sync(&cmd->work);
transport_wait_for_tasks(cmd);
if (!transport_cmd_finish_abort(cmd))
target: Fix kref->refcount underflow in transport_cmd_finish_abort This patch fixes a se_cmd->cmd_kref underflow during CMD_T_ABORTED when a fabric driver drops it's second reference from below the target_core_tmr.c based callers of transport_cmd_finish_abort(). Recently with the conversion of kref to refcount_t, this bug was manifesting itself as: [705519.601034] refcount_t: underflow; use-after-free. [705519.604034] INFO: NMI handler (kgdb_nmi_handler) took too long to run: 20116.512 msecs [705539.719111] ------------[ cut here ]------------ [705539.719117] WARNING: CPU: 3 PID: 26510 at lib/refcount.c:184 refcount_sub_and_test+0x33/0x51 Since the original kref atomic_t based kref_put() didn't check for underflow and only invoked the final callback when zero was reached, this bug did not manifest in practice since all se_cmd memory is using preallocated tags. To address this, go ahead and propigate the existing return from transport_put_cmd() up via transport_cmd_finish_abort(), and change transport_cmd_finish_abort() + core_tmr_handle_tas_abort() callers to only do their local target_put_sess_cmd() if necessary. Reported-by: Bart Van Assche <bart.vanassche@sandisk.com> Tested-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Mike Christie <mchristi@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Cc: Himanshu Madhani <himanshu.madhani@qlogic.com> Cc: Sagi Grimberg <sagig@mellanox.com> Cc: stable@vger.kernel.org # 3.14+ Tested-by: Gary Guo <ghg@datera.io> Tested-by: Chu Yuan Lin <cyl@datera.io> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
2017-06-03 11:00:17 +08:00
target_put_sess_cmd(cmd);
}
}
int core_tmr_lun_reset(
struct se_device *dev,
struct se_tmr_req *tmr,
struct list_head *preempt_and_abort_list,
struct se_cmd *prout_cmd)
{
struct se_node_acl *tmr_nacl = NULL;
struct se_portal_group *tmr_tpg = NULL;
struct se_session *tmr_sess = NULL;
int tas;
/*
* TASK_ABORTED status bit, this is configurable via ConfigFS
* struct se_device attributes. spc4r17 section 7.4.6 Control mode page
*
* A task aborted status (TAS) bit set to zero specifies that aborted
* tasks shall be terminated by the device server without any response
* to the application client. A TAS bit set to one specifies that tasks
* aborted by the actions of an I_T nexus other than the I_T nexus on
* which the command was received shall be completed with TASK ABORTED
* status (see SAM-4).
*/
tas = dev->dev_attrib.emulate_tas;
/*
* Determine if this se_tmr is coming from a $FABRIC_MOD
* or struct se_device passthrough..
*/
if (tmr && tmr->task_cmd && tmr->task_cmd->se_sess) {
tmr_sess = tmr->task_cmd->se_sess;
tmr_nacl = tmr_sess->se_node_acl;
tmr_tpg = tmr_sess->se_tpg;
if (tmr_nacl && tmr_tpg) {
pr_debug("LUN_RESET: TMR caller fabric: %s"
" initiator port %s\n",
tmr_tpg->se_tpg_tfo->get_fabric_name(),
tmr_nacl->initiatorname);
}
}
pr_debug("LUN_RESET: %s starting for [%s], tas: %d\n",
(preempt_and_abort_list) ? "Preempt" : "TMR",
dev->transport->name, tas);
core_tmr_drain_tmr_list(dev, tmr, preempt_and_abort_list);
core_tmr_drain_state_list(dev, prout_cmd, tmr_sess, tas,
preempt_and_abort_list);
/*
* Clear any legacy SPC-2 reservation when called during
* LOGICAL UNIT RESET
*/
if (!preempt_and_abort_list &&
(dev->dev_reservation_flags & DRF_SPC2_RESERVATIONS)) {
spin_lock(&dev->dev_reservation_lock);
dev->dev_reserved_node_acl = NULL;
dev->dev_reservation_flags &= ~DRF_SPC2_RESERVATIONS;
spin_unlock(&dev->dev_reservation_lock);
pr_debug("LUN_RESET: SCSI-2 Released reservation\n");
}
atomic_long_inc(&dev->num_resets);
pr_debug("LUN_RESET: %s for [%s] Complete\n",
(preempt_and_abort_list) ? "Preempt" : "TMR",
2011-07-19 16:55:10 +08:00
dev->transport->name);
return 0;
}