Re: qla2xxx cause BUG on kernel-4.17-rc6

[Laurence Oberman <loberman@redhat.com>]

On Wed, 2018-06-06 at 18:31 +0000, Madhani, Himanshu wrote:
> Hi Li, 
> 
> > On Jun 6, 2018, at 11:05 AM, Laurence Oberman <loberman@redhat.com>
> > wrote:
> > 
> > On Wed, 2018-06-06 at 16:01 +0000, Madhani, Himanshu wrote:
> > > > On Jun 6, 2018, at 8:56 AM, Martin K. Petersen <martin.petersen
> > > > @ora
> > > > cle.com> wrote:
> > > > 
> > > > 
> > > > Himanshu,
> > > > 
> > > > Ping?
> > > > 
> > > 
> > > Will look at this one. Sorry, somehow fell thru cracks. 
> > > 
> > > 
> > > > > Hi scsi experts,
> > > > > 
> > > > > Not sure who is the right person to ask, I just hit this bug
> > > > > on
> > > > > my HP
> > > > > DL385 platform, can any one of you take a look?
> > > > > 
> > > > > system config:
> > > > > -----------------
> > > > > HP ProLiant DL385 G7
> > > > > AMD Opteron(TM) Processor 6234
> > > > > 16384 MB memory, 369 GB disk space
> > > > > 
> > > > > 
> > > > > [   24.539274] qla2xxx [0000:0c:00.7]-500a:5: LOOP UP
> > > > > detected
> > > > > (10 Gbps).
> > > > > [   24.577259] BUG: unable to handle kernel NULL pointer
> > > > > dereference
> > > > > at 0000000000000102
> > > > > [   24.623133] PGD 0 P4D 0
> > > > > [   24.636760] Oops: 0000 [#1] SMP NOPTI
> > > > > [   24.656942] Modules linked in: i2c_algo_bit drm_kms_helper
> > > > > sr_mod(+) syscopyarea sysfillrect sysimgblt cdrom fb_sys_fops
> > > > > ata_generic ttm pata_acpi sd_mod ahci pata_atiixp sfc(+)
> > > > > qla2xxx(+)
> > > > > libahci drm qla4xxx(+) nvme_fc hpsa mdio libiscsi qlcnic(+)
> > > > > nvme_fabrics scsi_transport_sas serio_raw mtd crc32c_intel
> > > > > libata
> > > > > nvme_core i2c_core scsi_transport_iscsi tg3 scsi_transport_fc
> > > > > bnx2
> > > > > iscsi_boot_sysfs dm_multipath dm_mirror dm_region_hash dm_log
> > > > > dm_mod
> > > > > [   24.887449] CPU: 0 PID: 177 Comm: kworker/0:3 Not tainted
> > > > > 4.17.0-rc6 #1
> > > > > [   24.925119] Hardware name: HP ProLiant DL385 G7, BIOS A18
> > > > > 08/15/2012
> > > > > [   24.962106] Workqueue: events work_for_cpu_fn
> > > > > [   24.987098] RIP: 0010:__queue_work+0x1f/0x3a0
> > > > > [   25.011672] RSP: 0018:ffff992642ceba10 EFLAGS: 00010082
> > > > > [   25.042116] RAX: 0000000000000082 RBX: 0000000000000082
> > > > > RCX:
> > > > > 0000000000000000
> > > > > [   25.083293] RDX: ffff8cf9abc6d7d0 RSI: 0000000000000000
> > > > > RDI:
> > > > > 0000000000002000
> > > > > [   25.123094] RBP: 0000000000000000 R08: 0000000000025a40
> > > > > R09:
> > > > > ffff8cf9aade2880
> > > > > [   25.164087] R10: 0000000000000000 R11: ffff992642ceb6f0
> > > > > R12:
> > > > > ffff8cf9abc6d7d0
> > > > > [   25.202280] R13: 0000000000002000 R14: ffff8cf9abc6d7b8
> > > > > R15:
> > > > > 0000000000002000
> > > > > [   25.242050] FS:  0000000000000000(0000) f9b5c00000(0000)
> > > > > knlGS:0000000000000000
> > > > > [   25.977565] CS:  0010 DS: 0000 ES: 0000 CR0:
> > > > > 0000000080050033
> > > > > [   26.010457] CR2: 0000000000000102 CR3: 000000030760a000
> > > > > CR4:
> > > > > 00000000000406f0
> > > > > [   26.051048] Call Trace:
> > > > > [   26.063572]  ? __switch_to_asm+0x34/0x70
> > > > > [   26.086079]  queue_work_on+0x24/0x40
> > > > > [   26.107090]  qla2x00_post_work+0x81/0xb0 [qla2xxx]
> > > > > [   26.133356]  qla2x00_async_event+0x1ad/0x1a20 [qla2xxx]
> > > > > [   26.164075]  ? lock_timer_base+0x67/0x80
> > > > > [   26.186420]  ? try_to_del_timer_sync+0x4d/0x80
> > > > > [   26.212284]  ? del_timer_sync+0x35/0x40
> > > > > [   26.234080]  ? schedule_timeout+0x165/0x2f0
> > > > > [   26.259575]  qla82xx_poll+0x13e/0x180 [qla2xxx]
> > > > > [   26.285740]  qla2x00_mailbox_command+0x74b/0xf50 [qla2xxx]
> > > > > [   26.319040]  qla82xx_set_driver_version+0x13b/0x1c0
> > > > > [qla2xxx]
> > > > > [   26.352108]  ? qla2x00_init_rings+0x206/0x3f0 [qla2xxx]
> > > > > [   26.381733]  qla2x00_initialize_adapter+0x35c/0x7f0
> > > > > [qla2xxx]
> > > > > [   26.413240]  qla2x00_probe_one+0x1479/0x2390 [qla2xxx]
> > > > > [   26.442055]  local_pci_probe+0x3f/0xa0
> > > > > [   26.463108]  work_for_cpu_fn+0x10/0x20
> > > > > [   26.483295]  process_one_work+0x152/0x350
> > > > > [   26.505730]  worker_thread+0x1cf/0x3e0
> > > > > [   26.527090]  kthread+0xf5/0x130
> > > > > [   26.545085]  ? max_active_store+0x80/0x80
> > > > > [   26.568085]  ? kthread_bind+0x10/0x10
> > > > > [   26.589533]  ret_from_fork+0x22/0x40
> > > > > [   26.610192] Code: 00 00 66 2e 0f 1f 84 00 00 00 00 00 0f
> > > > > 1f 44
> > > > > 00
> > > > > 00 41 57 41 89 ff 41 56 41 55 41 89 fd 41 54 49 89 d4 55 48
> > > > > 89 f5
> > > > > 53
> > > > > 48 83 ec 0 86 02 01 00 00 01 0f 85 80 02 00 00 49 c7 c6 c0 ec
> > > > > 01
> > > > > 00 41
> > > > > [   27.308540] RIP: __queue_work+0x1f/0x3a0 RSP:
> > > > > ffff992642ceba10
> > > > > [   27.341591] CR2: 0000000000000102
> > > > > [   27.360208] ---[ end trace 01b7b7ae2c005cf3 ]---
> > > > 
> > > > -- 
> > > > Martin K. Petersen	Oracle Linux Engineering
> > > 
> > > Thanks,
> > > - Himanshu
> > > 
> > 
> > I can't find the original message for this that Martin reminded us
> > of.
> > 
> > To the person who logged this:
> > How many times has this happened and was it after a kernel update.
> > What is the history, what is the exact Qlogic card, etc.
> > Do you have the rest of the log log leading to the invalid pointer
> > fault
> > 
> > Thanks
> > Laurence
> 
> From the Snippet of Log provided looks like the crash is with 10G
> FCoE adapter. 
> 
> Can you try this untested diff to see if it resolves issue. 
> 
> Basically we are initializing adapter so driver will start receiving
> AEN notification
> but we have not yet allocated work queue for it. 
> 
> 
> ————— <snip> ————
> 
> diff --git a/drivers/scsi/qla2xxx/qla_os.c
> b/drivers/scsi/qla2xxx/qla_os.c
> index 30bf4b9..462d825 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -3229,6 +3229,8 @@ qla2x00_probe_one(struct pci_dev *pdev, const
> struct pci_device_id *id)
>             "req->req_q_in=%p req->req_q_out=%p rsp->rsp_q_in=%p rsp-
> >rsp_q_out=%p.\n",
>             req->req_q_in, req->req_q_out, rsp->rsp_q_in, rsp-
> >rsp_q_out);
> +       ha->wq = alloc_workqueue("qla2xxx_wq", 0, 0);
> +
>         if (ha->isp_ops->initialize_adapter(base_vha)) {
>                 ql_log(ql_log_fatal, base_vha, 0x00d6,
>                     "Failed to initialize adapter - Adapter flags
> %x.\n",
> @@ -3270,7 +3272,7 @@ qla2x00_probe_one(struct pci_dev *pdev, const
> struct pci_device_id *id)
>             host->can_queue, base_vha->req,
>             base_vha->mgmt_svr_loop_id, host->sg_tablesize);
>         INIT_WORK(&base_vha->iocb_work, qla2x00_iocb_work_fn);
> -       ha->wq = alloc_workqueue("qla2xxx_wq", 0, 0);
> +
>         if (ha->mqenable) {
>                 bool mq = false;
> 
> ————— </snip> ————
> 
> Thanks,
> - Himanshu
> 

Makes sense, but how did they escape this happening before ?
I cannot find the one that we looked at together about this but mine
was not @10G