kernel/drivers/target/target_core_tpg.c, branch linux-4.16.y

target: Fix node_acl demo-mode + uncached dynamic shutdown regression

2017-08-10T03:55:19Z

This patch fixes a generate_node_acls = 1 + cache_dynamic_acls = 0 regression, that was introduced by commit 01d4d673558985d9a118e1e05026633c3e2ade9b Author: Nicholas Bellinger Date: Wed Dec 7 12:55:54 2016 -0800 which originally had the proper list_del_init() usage, but was dropped during list review as it was thought unnecessary by HCH. However, list_del_init() usage is required during the special generate_node_acls = 1 + cache_dynamic_acls = 0 case when transport_free_session() does a list_del(&se_nacl->acl_list), followed by target_complete_nacl() doing the same thing. This was manifesting as a general protection fault as reported by Justin: kernel: general protection fault: 0000 [#1] SMP kernel: Modules linked in: kernel: CPU: 0 PID: 11047 Comm: iscsi_ttx Not tainted 4.13.0-rc2.x86_64.1+ #20 kernel: Hardware name: Intel Corporation S5500BC/S5500BC, BIOS S5500.86B.01.00.0064.050520141428 05/05/2014 kernel: task: ffff88026939e800 task.stack: ffffc90007884000 kernel: RIP: 0010:target_put_nacl+0x49/0xb0 kernel: RSP: 0018:ffffc90007887d70 EFLAGS: 00010246 kernel: RAX: dead000000000200 RBX: ffff8802556ca000 RCX: 0000000000000000 kernel: RDX: dead000000000100 RSI: 0000000000000246 RDI: ffff8802556ce028 kernel: RBP: ffffc90007887d88 R08: 0000000000000001 R09: 0000000000000000 kernel: R10: ffffc90007887df8 R11: ffffea0009986900 R12: ffff8802556ce020 kernel: R13: ffff8802556ce028 R14: ffff8802556ce028 R15: ffffffff88d85540 kernel: FS: 0000000000000000(0000) GS:ffff88027fc00000(0000) knlGS:0000000000000000 kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 kernel: CR2: 00007fffe36f5f94 CR3: 0000000009209000 CR4: 00000000003406f0 kernel: DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 kernel: DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 kernel: Call Trace: kernel: transport_free_session+0x67/0x140 kernel: transport_deregister_session+0x7a/0xc0 kernel: iscsit_close_session+0x92/0x210 kernel: iscsit_close_connection+0x5f9/0x840 kernel: iscsit_take_action_for_connection_exit+0xfe/0x110 kernel: iscsi_target_tx_thread+0x140/0x1e0 kernel: ? wait_woken+0x90/0x90 kernel: kthread+0x124/0x160 kernel: ? iscsit_thread_get_cpumask+0x90/0x90 kernel: ? kthread_create_on_node+0x40/0x40 kernel: ret_from_fork+0x22/0x30 kernel: Code: 00 48 89 fb 4c 8b a7 48 01 00 00 74 68 4d 8d 6c 24 08 4c 89 ef e8 e8 28 43 00 48 8b 93 20 04 00 00 48 8b 83 28 04 00 00 4c 89 ef <48> 89 42 08 48 89 10 48 b8 00 01 00 00 00 00 ad de 48 89 83 20 kernel: RIP: target_put_nacl+0x49/0xb0 RSP: ffffc90007887d70 kernel: ---[ end trace f12821adbfd46fed ]--- To address this, go ahead and use proper list_del_list() for all cases of se_nacl->acl_list deletion. Reported-by: Justin Maggard Tested-by: Justin Maggard Cc: Justin Maggard Cc: stable@vger.kernel.org # 4.1+ Signed-off-by: Nicholas Bellinger

target/configfs: Kill se_lun->lun_link_magic

2017-06-09T06:26:38Z

Instead of using a hardcoded magic value in se_lun when verifying a target config_item symlink source during target_fabric_mappedlun_link(), go ahead and use target_fabric_port_item_ops directly instead. Reviewed-by: Christoph Hellwig Cc: Mike Christie Cc: Hannes Reinecke Signed-off-by: Nicholas Bellinger

target: Don't force session reset if queue_depth does not change

2017-05-05T03:01:40Z

Keeping in the idempotent nature of target_core_fabric_configfs.c, if a queue_depth value is set and it's the same as the existing value, don't attempt to force session reinstatement. Reported-by: Raghu Krishnamurthy Cc: Raghu Krishnamurthy Tested-by: Gary Guo Cc: Gary Guo Signed-off-by: Nicholas Bellinger

target: Avoid mappedlun symlink creation during lun shutdown

2017-03-30T08:36:52Z

This patch closes a race between se_lun deletion during configfs unlink in target_fabric_port_unlink() -> core_dev_del_lun() -> core_tpg_remove_lun(), when transport_clear_lun_ref() blocks waiting for percpu_ref RCU grace period to finish, but a new NodeACL mappedlun is added before the RCU grace period has completed. This can happen in target_fabric_mappedlun_link() because it only checks for se_lun->lun_se_dev, which is not cleared until after transport_clear_lun_ref() percpu_ref RCU grace period finishes. This bug originally manifested as NULL pointer dereference OOPsen in target_stat_scsi_att_intr_port_show_attr_dev() on v4.1.y code, because it dereferences lun->lun_se_dev without a explicit NULL pointer check. In post v4.1 code with target-core RCU conversion, the code in target_stat_scsi_att_intr_port_show_attr_dev() no longer uses se_lun->lun_se_dev, but the same race still exists. To address the bug, go ahead and set se_lun>lun_shutdown as early as possible in core_tpg_remove_lun(), and ensure new NodeACL mappedlun creation in target_fabric_mappedlun_link() fails during se_lun shutdown. Reported-by: James Shen Cc: James Shen Tested-by: James Shen Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Nicholas Bellinger

target: allow ALUA setup for some passthrough backends

2017-03-18T21:47:25Z

This patch allows passthrough backends to use the core/base LIO ALUA setup and state checks, but still handle the execution of commands. This will allow the target_core_user module to execute STPG and RTPG in userspace, and not have to duplicate the ALUA state checks, path information (needed so we can check if command is executable on specific paths) and setup (rtslib sets/updates the configfs ALUA interface like it does for iblock or file). For STPG, the target_core_user userspace daemon, tcmu-runner will still execute the STPG, and to update the core/base LIO state it will use the existing configfs interface. For RTPG, tcmu-runner will loop over configfs and/or cache the state. Signed-off-by: Mike Christie Signed-off-by: Nicholas Bellinger

Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending

2017-03-02T22:52:05Z

Pull SCSI target updates from Nicholas Bellinger: "The highlights this round include: - enable dual mode (initiator + target) qla2xxx operation. (Quinn + Himanshu) - add a framework for qla2xxx async fabric discovery. (Quinn + Himanshu) - enable iscsi PDU DDP completion offload in cxgbit/T6 NICs. (Varun) - fix target-core handling of aborted failed commands. (Bart) - fix a long standing target-core issue NULL pointer dereference with active I/O LUN shutdown. (Rob Millner + Bryant + nab)" * 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (44 commits) target: Add counters for ABORT_TASK success + failure iscsi-target: Fix early login failure statistics misses target: Fix NULL dereference during LUN lookup + active I/O shutdown target: Delete tmr from list before processing target: Fix handling of aborted failed commands uapi: fix linux/target_core_user.h userspace compilation errors target: export protocol identifier qla2xxx: Fix a warning reported by the "smatch" static checker target/iscsi: Fix unsolicited data seq_end_offset calculation target/cxgbit: add T6 iSCSI DDP completion feature target/cxgbit: Enable DDP for T6 only if data sequence and pdu are in order target/cxgbit: Use T6 specific macros to get ETH/IP hdr len target/cxgbit: use cxgb4_tp_smt_idx() to get smt idx target/iscsi: split iscsit_check_dataout_hdr() target: Remove command flag CMD_T_DEV_ACTIVE target: Remove command flag CMD_T_BUSY target: Move session check from target_put_sess_cmd() into target_release_cmd_kref() target: Inline transport_cmd_check_stop() target: Remove an overly chatty debug message target: Stop execution if CMD_T_STOP has been set ...

target: Fix NULL dereference during LUN lookup + active I/O shutdown

2017-02-27T00:08:44Z

When transport_clear_lun_ref() is shutting down a se_lun via configfs with new I/O in-flight, it's possible to trigger a NULL pointer dereference in transport_lookup_cmd_lun() due to the fact percpu_ref_get() doesn't do any __PERCPU_REF_DEAD checking before incrementing lun->lun_ref.count after lun->lun_ref has switched to atomic_t mode. This results in a NULL pointer dereference as LUN shutdown code in core_tpg_remove_lun() continues running after the existing ->release() -> core_tpg_lun_ref_release() callback completes, and clears the RCU protected se_lun->lun_se_dev pointer. During the OOPs, the state of lun->lun_ref in the process which triggered the NULL pointer dereference looks like the following on v4.1.y stable code: struct se_lun { lun_link_magic = 4294932337, lun_status = TRANSPORT_LUN_STATUS_FREE, ..... lun_se_dev = 0x0, lun_sep = 0x0, ..... lun_ref = { count = { counter = 1 }, percpu_count_ptr = 3, release = 0xffffffffa02fa1e0 , confirm_switch = 0x0, force_atomic = false, rcu = { next = 0xffff88154fa1a5d0, func = 0xffffffff8137c4c0 } } } To address this bug, use percpu_ref_tryget_live() to ensure once __PERCPU_REF_DEAD is visable on all CPUs and ->lun_ref has switched to atomic_t, all new I/Os will fail to obtain a new lun->lun_ref reference. Also use an explicit percpu_ref_kill_and_confirm() callback to block on ->lun_ref_comp to allow the first stage and associated RCU grace period to complete, and then block on ->lun_ref_shutdown waiting for the final percpu_ref_put() to drop the last reference via transport_lun_remove_cmd() before continuing with core_tpg_remove_lun() shutdown. Reported-by: Rob Millner Tested-by: Rob Millner Cc: Rob Millner Tested-by: Vaibhav Tandon Cc: Vaibhav Tandon Tested-by: Bryant G. Ly Cc: # v3.14+ Signed-off-by: Nicholas Bellinger

IB/srpt: Accept GUIDs as port names

2017-01-27T19:31:50Z

Port and ACL information must be configured before an initiator logs in. Make it possible to configure this information before a subnet prefix has been assigned to a port by not only accepting GIDs as target port and initiator port names but by also accepting port GUIDs. Add a 'priv' member to struct se_wwn to allow target drivers to associate their own data with struct se_wwn. Reported-by: Doug Ledford References: http://www.spinics.net/lists/linux-rdma/msg39505.html Signed-off-by: Bart Van Assche Cc: Christoph Hellwig Cc: Nicholas Bellinger Signed-off-by: Doug Ledford

target: remove sess_kref and ->shutdown_session

2016-05-10T08:19:49Z

Both of them are unused now that drivers handle any delayed session shutdown internally. Signed-off-by: Christoph Hellwig Signed-off-by: Nicholas Bellinger

target: make ->shutdown_session optional

2016-05-10T08:19:18Z

Turns out the template and thus many drivers got the return value wrong: 0 means the fabrics driver needs to put a session reference, which no driver except for the iSCSI target drivers did. Fortunately none of these drivers supports explicit Node ACLs, so the bug was harmless. Even without that only qla2xxx and iscsi every did real work in shutdown_session, so get rid of the boilerplate code in all other drivers. Signed-off-by: Christoph Hellwig Signed-off-by: Nicholas Bellinger