| Message ID | 1502237504.4017.1.camel@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, 2017-08-08 at 20:11 -0400, Laurence Oberman wrote: > > On Tue, 2017-08-08 at 22:17 +0800, Ming Lei wrote: > > > > Hi Guys, > > > > > > > > Laurence and I see a system lockup issue when running > > concurrent > > > > big buffered write(4M bytes) to IB SRP on v4.13-rc3. > > > > > > > > 1 how to reproduce > > > > > > > > 1) setup IB_SRR & multi path > > > > > > > > #./start_opensm.sh > > > > #./start_srp.sh > > > > > > > > # cat start_opensm.sh > > > > #!/bin/bash > > > > rmmod ib_srpt > > > > opensm -F opensm.1.conf & > > > > opensm -F opensm.2.conf & > > > > > > > > # cat start_srp.sh > > > > run_srp_daemon -f /etc/ddn/srp_daemon.conf -R 30 -T 10 > > -t 7000 > > > > -ance -i mlx5_0 -p 1 1>/root/srp1.log 2>&1 & > > > > run_srp_daemon -f /etc/ddn/srp_daemon.conf -R 30 -T 10 > > -t 7000 > > > > -ance -i mlx5_1 -p 1 1>/root/srp2.log 2>&1 & > > > > > > > > > > > > 2) run the following test script > > > > > > > > #./test-mp.sh > > > > > > > > #cat test-mp.sh > > > > #!/bin/sh > > > > DEVS="360014051186d24cc27b4d049ffff94e3 > > > > 36001405b2b5c6c24c084b6fa4d55da2f > > > > 36001405b26ebe76dcb94a489f6f245f8" > > > > > > > > for i in $DEVS; do > > > > for j in `seq 0 15`; do > > > > ./hammer_write.sh $i 1>/dev/null 2>&1 & > > > > done > > > > done > > > > > > > > #cat hammer_write.sh > > > > #!/bin/bash > > > > while true; do > > > > dd if=/dev/zero of=/dev/mapper/$1 bs=4096k > > count=800 > > > > done > > > > > > > > > > > > 2 lockup log > > > > [root@ibclient ~]# ./test-mp.sh > > > > [root@ibclient ~]# [ 169.095494] perf: interrupt took too long > > > > (2575 > > > > > > 2500), lowering kernel.perf_event_max_sample_rate to 77000 > > > > > > > > > > [ 178.569124] perf: interrupt took too long (3249 > 3218), > > > > lowering > > > > kernel.perf_event_max_sample_rate to 61000 > > > > [ 190.586624] perf: interrupt took too long (4066 > 4061), > > > > lowering > > > > kernel.perf_event_max_sample_rate to 49000 > > > > [ 210.381836] perf: interrupt took too long (5083 > 5082), > > > > lowering > > > > kernel.perf_event_max_sample_rate to 39000 > > > > [ 240.498659] perf: interrupt took too long (6373 > 6353), > > > > lowering > > > > kernel.perf_event_max_sample_rate to 31000 > > > > > > > > [root@ibclient ~]# > > > > [root@ibclient ~]# [ 292.252795] perf: interrupt took too long > > > > (8006 > > > > > > 7966), lowering kernel.perf_event_max_sample_rate to 24000 > > > > > > > > > > [ 410.354443] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.432725] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.499530] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.564952] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.644493] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.698091] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.755175] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.796784] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.837690] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 410.878743] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.476823] multipath_clone_and_map: 32 callbacks suppressed > > > > [ 463.506773] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.586752] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.656880] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.698919] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.723572] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.751820] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.781110] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.828267] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.856997] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 463.885998] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.572060] multipath_clone_and_map: 5201 callbacks > > suppressed > > > > [ 468.602191] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.655761] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.697540] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.738610] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.779597] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.820705] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.862712] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.904662] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.945132] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 468.985128] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.643013] multipath_clone_and_map: 6569 callbacks > > suppressed > > > > [ 473.673466] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.713929] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.754989] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.796069] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.849768] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.891640] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.933321] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 473.973791] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 474.016739] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 474.059858] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 478.732265] multipath_clone_and_map: 6310 callbacks > > suppressed > > > > [ 478.762164] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 478.802327] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 478.844235] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 478.886113] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 478.927164] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 478.968141] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 479.008772] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 479.060225] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 479.102873] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 479.145341] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 483.834884] multipath_clone_and_map: 5703 callbacks > > suppressed > > > > [ 483.865056] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 483.906792] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 483.959492] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.001768] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.044785] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.087289] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.129854] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.171317] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.212314] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.263931] device-mapper: multipath: blk_get_request() > > returned > > > > -11 - requeuing > > > > [ 484.982608] perf: interrupt took too long (10021 > 10007), > > > > lowering kernel.perf_event_max_sample_rate to 19000 > > > > [ 485.362787] watchdog: BUG: soft lockup - CPU#11 stuck for > > 21s! > > > > [ksoftirqd/11:78] > > > > [ 485.400762] Modules linked in: dm_round_robin xt_CHECKSUM > > > > ipt_MASQUERADE nf_nat_masquerade_ipv4 tun ip6t_rpfilter > > ipt_REJECT > > > > nf_reject_ipv4 ip6t_REJECT nf_reject_ipv6 xt_conntrack ip_set > > > > nfnetlink ebtable_nat ebtable_broute bridge stp llc > > ip6table_nat > > > > nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle > > > > ip6table_security ip6table_raw iptable_nat nf_conntrack_ipv4 > > > > nf_defrag_ipv4 nf_nat_ipv4 nf_nat rpcrdma nf_conntrack ib_isert > > > > iscsi_target_mod iptable_mangle iptable_security iptable_raw > > > > ebtable_filter ebtables ib_iser ip6table_filter ip6_tables > > libiscsi > > > > iptable_filter scsi_transport_iscsi target_core_mod ib_srp > > > > scsi_transport_srp ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad > > > > rdma_cm > > > > ib_cm iw_cm mlx5_ib ib_core intel_powerclamp coretemp kvm_intel > > kvm > > > > irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel > > > > [ 485.809222] pcbc joydev ipmi_si aesni_intel sg hpwdt hpilo > > > > crypto_simd iTCO_wdt cryptd iTCO_vendor_support ipmi_devintf > > > > gpio_ich > > > > glue_helper shpchp lpc_ich ipmi_msghandler acpi_power_meter > > pcspkr > > > > i7core_edac pcc_cpufreq nfsd auth_rpcgss nfs_acl lockd grace > > sunrpc > > > > dm_multipath ip_tables xfs libcrc32c sd_mod amdkfd amd_iommu_v2 > > > > radeon mlx5_core i2c_algo_bit drm_kms_helper syscopyarea > > > > sysfillrect > > > > sysimgblt fb_sys_fops ttm mlxfw drm ptp hpsa pps_core > > > > scsi_transport_sas crc32c_intel bnx2 i2c_core serio_raw devlink > > > > dm_mirror dm_region_hash dm_log dm_mod [last unloaded: ib_srpt] > > > > [ 486.096352] CPU: 11 PID: 78 Comm: ksoftirqd/11 Tainted: > > > > G I 4.13.0-rc3+ #23 > > > > [ 486.142775] Hardware name: HP ProLiant DL380 G7, BIOS P67 > > > > 08/16/2015 > > > > [ 486.178816] task: ffff8a3fe453df40 task.stack: > > ffff9a6cc65a8000 > > > > [ 486.212674] RIP: 0010:__blk_recalc_rq_segments+0xb8/0x3d0 > > > > [ 486.242444] RSP: 0018:ffff9a6cc65aba60 EFLAGS: 00000282 > > > > ORIG_RAX: > > > > ffffffffffffff10 > > > > [ 486.284086] RAX: ffff8a4a8e26e440 RBX: 0000000000001000 RCX: > > > > ffff8a4a8e26e440 > > > > [ 486.323445] RDX: 0000000000000000 RSI: 0000000000001000 RDI: > > > > ffff8a4a8e2699c8 > > > > [ 486.363614] RBP: ffff9a6cc65abae0 R08: 0000000000007000 R09: > > > > 0000000000000001 > > > > [ 486.404621] R10: 0000000000001000 R11: 0000000000000000 R12: > > > > 0000000000000000 > > > > [ 486.445127] R13: 0000000000000000 R14: 0000000000000001 R15: > > > > 0000000000000007 > > > > [ 486.486120] FS: 0000000000000000(0000) > > > > GS:ffff8a4b66940000(0000) > > > > knlGS:0000000000000000 > > > > [ 486.532771] CS: 0010 DS: 0000 ES: 0000 CR0: > > 0000000080050033 > > > > [ 486.565772] CR2: 00007f32bfcee000 CR3: 0000000d67609000 CR4: > > > > 00000000000006e0 > > > > [ 486.606325] Call Trace: > > > > [ 486.620269] ? mempool_free+0x2b/0x80 > > > > [ 486.641248] blk_recalc_rq_segments+0x28/0x40 > > > > [ 486.666754] blk_update_request+0x249/0x310 > > > > [ 486.690838] end_clone_bio+0x46/0x70 [dm_mod] > > > > [ 486.715949] bio_endio+0xa1/0x120 > > > > [ 486.734408] blk_update_request+0xb7/0x310 > > > > [ 486.757313] scsi_end_request+0x38/0x1d0 > > > > [ 486.779375] scsi_io_completion+0x13c/0x630 > > > > [ 486.802928] scsi_finish_command+0xd9/0x120 > > > > [ 486.825875] scsi_softirq_done+0x12a/0x150 > > > > [ 486.849029] __blk_mq_complete_request+0x90/0x140 > > > > [ 486.875484] blk_mq_complete_request+0x16/0x20 > > > > [ 486.900631] scsi_mq_done+0x21/0x70 > > > > [ 486.920462] srp_recv_done+0x48f/0x610 [ib_srp] > > > > [ 486.946212] __ib_process_cq+0x5c/0xb0 [ib_core] > > > > [ 486.972576] ib_poll_handler+0x22/0x70 [ib_core] > > > > [ 486.998638] irq_poll_softirq+0xaf/0x110 > > > > [ 487.021727] __do_softirq+0xc9/0x269 > > > > [ 487.042027] run_ksoftirqd+0x29/0x50 > > > > [ 487.062675] smpboot_thread_fn+0x110/0x160 > > > > [ 487.085597] kthread+0x109/0x140 > > > > [ 487.104113] ? sort_range+0x30/0x30 > > > > [ 487.124120] ? kthread_park+0x60/0x60 > > > > [ 487.144624] ret_from_fork+0x25/0x30 > > > > [ 487.164640] Code: c3 83 fa 09 0f 84 8e 01 00 00 48 8b 44 24 > > 20 > > > > 48 > > > > 8b 4c 24 28 45 31 ff 45 31 c0 45 31 c9 0f b6 80 36 06 00 00 88 > > 44 > > > > 24 > > > > 17 8b 59 28 <44> 8b 71 2c 44 8b 61 34 85 db 0f 84 92 01 00 00 > > 45 89 > > > > f5 45 89 > > > > > > > > Message from syslogd@ibclient at Aug 8 04:12:23 ... > > > > kernel:watchdog: BUG: soft lockup - CPU#11 stuck for 21s! > > > > [ksoftirqd/11:78] > > > > > > > > > > > > > > > > Thanks, > > > > Ming > > > > > Hello Bart > > > > I received your message but my work email is not seeing it. > > So I am replying to the original email by Ming so I can properly > > respond to your most recent message pasted below and keep the > thread > > in > > the right flow. > > > > I am purposefully testing buffered I/O because as you know that was > a > > source of pain for us last year. > > So yes my intention always is to run large buffered and un-buffered > > (direct) sequential I/O as its part of my regular tests that have > > been > > very useful in discovering many issues. > > I also have smaller random and sequential I/O tests driven by fio > > that > > were not seeing the issue. > > > > I will go ahead and test your patch below. I am not surprised Ming > > was > > able to make this happen on the stock 4.13-RC3 but he had to run 16 > > parallel jobs. > > > > It was happening on his kernel with only 3 parallel jobs and had > not > > been happening for me with 3 parallel runs on stock upstream but > > perhaps just takes longer on 4.13-RC3 so 16 parallel runs brings it > > out. > > > > Just one other small point, this is a "hard", not soft lockup. > > > > Unfortunately on the 7.3 base I am unable to capture a full vmcore > > when > > running the latest upstream kernel so if I need to get a vmcore I > > will > > have to use Fedora and the latest upstream. > > > > I will do that if your patch is not successful but reading it and > > looking at the stack dumps I have, it makes sense. > > > > back in a bit > > > > Thanks > > Laurence > > > > ---- > > Hello Laurence, > > > > Is your goal perhaps to simulate a DDN workload? In that case I > think > > you > > need oflag=direct to the dd argument list such that the page cache > > writeback > > code does not alter the size of the write requests. Anyway, this > test > > should > > not trigger a lockup. Can you check whether the patch below makes > the > > soft > > lockup complaints disappear (without changing the hammer_write.sh > > test > > script)? > > > > Thanks, > > > > Bart. > > ----------------------------------------------------------------- > ---- > > ------- > > [PATCH] block: Make blk_mq_delay_kick_requeue_list() rerun the > queue > > at > > a > > quiet time > > > > Drivers like dm-mpath requeue requests if no paths are available > and > > if configured to do so. If the queue depth is sufficiently high and > > the queue rerunning delay sufficiently short then .requeue_work can > > be queued so often that other work items queued on the same work > > queue > > do not get executed. Avoid that this happens by only rerunning the > > queue after no blk_mq_delay_kick_requeue_list() calls have occurred > > during @msecs milliseconds. Since the device mapper core is the > only > > user of blk_mq_delay_kick_requeue_list(), modify the implementation > > of this function instead of creating a new function. > > > > Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> > > --- > > block/blk-mq.c | 4 ++-- > > 1 file changed, 2 insertions(+), 2 deletions(-) > > > > diff --git a/block/blk-mq.c b/block/blk-mq.c > > index 041f7b7fa0d6..8bfea36e92f9 100644 > > --- a/block/blk-mq.c > > +++ b/block/blk-mq.c > > @@ -679,8 +679,8 @@ EXPORT_SYMBOL(blk_mq_kick_requeue_list); > > void blk_mq_delay_kick_requeue_list(struct request_queue *q, > > unsigned long msecs) > > { > > - kblockd_schedule_delayed_work(&q->requeue_work, > > - msecs_to_jiffies(msecs)); > > + kblockd_mod_delayed_work_on(WORK_CPU_UNBOUND, &q- > > >requeue_work, > > + msecs_to_jiffies(msecs)); > > } > > EXPORT_SYMBOL(blk_mq_delay_kick_requeue_list); > > ----- Hello Bart So the patch is working to avoid the clone failures messages and requeue -11 messages of my original issue however both the hard lockup and soft lockup persists. It's is a lot harder to get the hard lockup now though with the patch. The soft LOCKUPS due to the spins on the ksoftirqd are almost expected because of how hard we are driving this and are easy to recreate. I did boot the original Ming kernel and reproduced first for baseline. I then patched it with your simple patch. I was not paying enough attention to Ming's message but he was seeing a ksoftirq soft LOCKUP on the stock upstream kernel NOT the hard LOCKUP. I did not check his log message carefully so assumed he meant hard LOCKUP. I guess that's why you said "soft LOCKUP" I did get into a few soft lockups before I hit another hard LOCKUP. So I am now again not sure about the hard LOCKUP on the stock and 7.4 kernel. I will have to retest and baseline myself as Ming is away now on vacation. The soft lockups still happen but they recover unless panic_on_ soft_lockup is set and I almost expect them with how the softirqd is spinning here. We see high spin rates here and I am only running 7 parallel writes. PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 20 root 20 0 0 0 0 R 100.0 0.0 0:27.11 kworker/u6+ 24 root 20 0 0 0 0 R 100.0 0.0 0:21.56 ksoftirqd/2 60 root 20 0 0 0 0 R 100.0 0.0 0:34.61 ksoftirqd/8 We see high SYS time because of the spinning so the soft LOCKUP is not unexpected to be honest and it recovers. # <----CPU[HYPER]-----><-----------Memory-----------><--------- -Disks-----------> #Time cpu sys inter ctxsw Free Buff Cach Inac Slab Map KBRead Reads KBWrit Writes 21:32:11 52 52 11526 45981 65G 21G 541M 22G 7G 209M 45 11 709791 211 21:32:12 49 49 13315 26037 65G 21G 541M 22G 6G 209M 0 0 606356 195 21:32:13 47 47 13375 21456 65G 21G 541M 22G 6G 209M 0 0 591468 250 21:32:14 33 33 12635 978 65G 21G 541M 22G 5G 209M 0 0 560860 365 21:32:15 40 40 14304 734 67G 19G 541M 20G 5G 209M 0 0 429472 97 21:32:16 36 36 16060 6220 70G 17G 541M 17G 4G 209M 0 0 312852 54 21:32:17 31 31 16071 4962 71G 16G 540M 16G 4G 209M 24 6 253228 44 21:32:18 31 31 15855 868 70G 17G 540M 18G 6G 209M 104 26 237696 62 21:32:19 31 31 15983 661 70G 17G 541M 17G 7G 209M 0 0 162340 37 21:32:20 32 32 16245 1413 70G 17G 541M 18G 9G 209M 68 17 155648 19 21:32:21 44 44 14919 21711 68G 19G 540M 19G 10G 209M 0 0 303104 32 21:32:22 59 59 11661 69544 68G 19G 541M 20G 10G 209M 0 0 620919 84 21:32:23 59 59 12951 1058 67G 20G 540M 20G 10G 209M 0 0 596582 126 21:32:24 68 67 12991 2077 66G 21G 540M 21G 10G 209M 0 0 689504 168 21:32:25 78 78 13295 2927 66G 21G 541M 21G 10G 209M 0 0 667673 144 21:32:26 66 66 13498 7670 65G 21G 541M 22G 10G 209M 0 0 626874 98 21:32:27 55 55 12203 2663 65G 21G 541M 22G 9G 209M 0 0 652096 112 21:32:28 47 47 12567 27804 65G 21G 541M 22G 8G 209M 0 0 534789 137 We are driving this system pretty hard here and the target server is seeing queue fulls and I am used to these if I drive it too hard. [ 568.707572] ib_srpt srpt_write_pending: IB send queue full (needed 97) [ 568.745714] ib_srpt srpt_write_pending: IB send queue full (needed 97) [ 568.783735] ib_srpt srpt_write_pending: IB send queue full (needed 97) [ 568.821583] ib_srpt srpt_write_pending: IB send queue full (needed 97) [ 568.858238] ib_srpt srpt_write_pending: IB send queue full (needed 97) [ 568.895975] ib_srpt srpt_queue_response: IB send queue full (needed 0) [ 568.934253] ib_srpt srpt_write_pending: IB send queue full (needed 97) [ 568.971040] ib_srpt srpt_write_pending: IB send queue full (needed 49) So your patch helps to avoid easily getting the hard LOCKUP, does not help the Soft lockup (as expected). [ 329.305507] watchdog: BUG: soft lockup - CPU#12 stuck for 21s! [ksoftirqd/12:84] [ 329.305508] Modules linked in: dm_round_robin mq_deadline xt_CHECKSUM ipt_MASQUERADE nf_nat_masquerade_ipv4 tun ip6t_rpfilter ipt_REJECT nf_reject_ipv4 ip6t_REJECT nf_reject_ipv6 xt_conntrack ip_set nfnetlink ebtable_nat ebtable_broute bridge stp llc ip6table_nat nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security ip6table_raw iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat rpcrdma nf_conntrack ib_isert iptable_mangle iptable_security iscsi_target_mod iptable_raw ebtable_filter ebtables ip6table_filter ib_iser ip6_tables libiscsi iptable_filter scsi_transport_iscsi target_core_mod ib_srp scsi_transport_srp ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad rdma_cm ib_cm iw_cm mlx5_ib ib_core intel_powerclamp coretemp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul [ 329.305543] ghash_clmulni_intel pcbc joydev aesni_intel ipmi_si sg hpilo ipmi_devintf hpwdt crypto_simd iTCO_wdt gpio_ich iTCO_vendor_support cryptd shpchp acpi_power_meter ipmi_msghandler i7core_edac glue_helper pcspkr lpc_ich pcc_cpufreq nfsd auth_rpcgss nfs_acl lockd grace sunrpc dm_multipath ip_tables xfs libcrc32c sd_mod amdkfd amd_iommu_v2 radeon mlx5_core i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm mlxfw ptp drm pps_core hpsa i2c_core devlink crc32c_intel serio_raw bnx2 scsi_transport_sas dm_mirror dm_region_hash dm_log dm_mod [last unloaded: ib_srpt] [ 329.305564] CPU: 12 PID: 84 Comm: ksoftirqd/12 Tainted: G I 4.13.0-rc3.ming.bart+ #1 [ 329.305566] Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 [ 329.305567] task: ffff91a94c070040 task.stack: ffffaa8ec65d8000 [ 329.305573] RIP: 0010:__blk_recalc_rq_segments+0x1a7/0x3d0 [ 329.305574] RSP: 0018:ffffaa8ec65dba60 EFLAGS: 00000202 ORIG_RAX: ffffffffffffff10 [ 329.305575] RAX: 0000000000000001 RBX: 0000000000001000 RCX: ffff91b1149239c0 [ 329.305576] RDX: 0000000000001000 RSI: 0000000000000001 RDI: 0000000000001000 [ 329.305577] RBP: ffffaa8ec65dbae0 R08: 0000000000001000 R09: 0000000000000001 [ 329.305578] R10: 0000000b2a506000 R11: 0000000000000000 R12: 0000000000000000 [ 329.305579] R13: 0000000000000000 R14: 0000000000000000 R15: 000000000000002e [ 329.305580] FS: 0000000000000000(0000) GS:ffff91b3b7b80000(0000) knlGS:0000000000000000 [ 329.305581] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 329.305582] CR2: 00007ff4591f1860 CR3: 0000000985e09000 CR4: 00000000000006e0 [ 329.305583] Call Trace: [ 329.305588] ? mempool_free+0x2b/0x80 [ 329.305590] blk_recalc_rq_segments+0x28/0x40 [ 329.305592] blk_update_request+0x249/0x310 [ 329.305603] end_clone_bio+0x46/0x70 [dm_mod] [ 329.305607] bio_endio+0xa1/0x120 [ 329.305608] blk_update_request+0xb7/0x310 [ 329.305611] scsi_end_request+0x38/0x1d0 [ 329.305613] scsi_io_completion+0x13c/0x630 [ 329.305615] scsi_finish_command+0xd9/0x120 [ 329.305616] scsi_softirq_done+0x12a/0x150 [ 329.305618] __blk_mq_complete_request+0x90/0x140 [ 329.305624] blk_mq_complete_request+0x16/0x20 [ 329.305625] scsi_mq_done+0x21/0x70 [ 329.305628] srp_recv_done+0x48f/0x610 [ib_srp] [ 329.305639] __ib_process_cq+0x5c/0xb0 [ib_core] [ 329.305646] ib_poll_handler+0x22/0x70 [ib_core] [ 329.305649] irq_poll_softirq+0xaf/0x110 [ 329.305652] __do_softirq+0xc9/0x269 [ 329.305655] run_ksoftirqd+0x29/0x50 [ 329.305657] smpboot_thread_fn+0x110/0x160 [ 329.305659] kthread+0x109/0x140 [ 329.305660] ? sort_range+0x30/0x30 [ 329.305661] ? kthread_park+0x60/0x60 [ 329.305663] ret_from_fork+0x25/0x30 [ 329.305664] Code: 8b 74 24 30 48 8b 7c 24 38 41 83 c7 01 41 89 d0 41 b9 01 00 00 00 48 89 74 24 40 48 89 7c 24 48 8b 41 14 0f b6 f0 25 fb 00 00 00 <83> f8 03 0f 84 a0 00 00 00 83 fe 05 0f 84 97 00 00 00 83 fe 09 [ 331.465885] CPU: 4 PID: 36 Comm: ksoftirqd/4 Tainted: G I L 4.13.0-rc3.ming.bart+ #1 [ 331.516413] Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 [ 331.551403] task: ffff91a94c76df40 task.stack: ffffaa8ec6458000 [ 331.584080] RIP: 0010:__blk_recalc_rq_segments+0xdf/0x3d0 [ 331.614515] RSP: 0018:ffffaa8ec645ba60 EFLAGS: 00000286 ORIG_RAX: ffffffffffffff10 [ 331.656098] RAX: 0000000000001000 RBX: 0000000000001000 RCX: ffff91b29faecdc0 [ 331.696577] RDX: ffffe839e5d75300 RSI: ffff91b29faece48 RDI: ffff91b2831f8348 [ 331.736607] RBP: ffffaa8ec645bae0 R08: 0000000000004000 R09: 0000000000000001 [ 331.776129] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 [ 331.817085] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000021 [ 331.857671] FS: 0000000000000000(0000) GS:ffff91b3b7a80000(0000) knlGS:0000000000000000 [ 331.904232] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 331.935249] CR2: 00007f404526e000 CR3: 0000000985e09000 CR4: 00000000000006e0 [ 331.976273] Call Trace: [ 331.990216] ? mempool_free+0x2b/0x80 [ 332.010310] blk_recalc_rq_segments+0x28/0x40 [ 332.034810] blk_update_request+0x249/0x310 [ 332.058259] end_clone_bio+0x46/0x70 [dm_mod] [ 332.082851] bio_endio+0xa1/0x120 [ 332.101145] blk_update_request+0xb7/0x310 [ 332.123965] scsi_end_request+0x38/0x1d0 [ 332.145943] scsi_io_completion+0x13c/0x630 [ 332.169357] scsi_finish_command+0xd9/0x120 [ 332.193505] scsi_softirq_done+0x12a/0x150 [ 332.217044] __blk_mq_complete_request+0x90/0x140 [ 332.244498] blk_mq_complete_request+0x16/0x20 [ 332.269971] scsi_mq_done+0x21/0x70 [ 332.289591] srp_recv_done+0x48f/0x610 [ib_srp] [ 332.315609] __ib_process_cq+0x5c/0xb0 [ib_core] [ 332.342144] ib_poll_handler+0x22/0x70 [ib_core] [ 332.368724] irq_poll_softirq+0xaf/0x110 [ 332.392595] __do_softirq+0xc9/0x269 [ 332.413220] run_ksoftirqd+0x29/0x50 [ 332.433214] smpboot_thread_fn+0x110/0x160 [ 332.456254] kthread+0x109/0x140 [ 332.475263] ? sort_range+0x30/0x30 [ 332.494624] ? kthread_park+0x60/0x60 [ 332.515413] ret_from_fork+0x25/0x30 [ 332.535180] Code: 17 8b 59 28 44 8b 71 2c 44 8b 61 34 85 db 0f 84 92 01 00 00 45 89 f5 45 89 e2 4c 89 ee 48 c1 e6 04 48 03 71 78 8b 46 08 48 8b 16 <44> 29 e0 48 89 54 24 30 39 d8 0f 47 c3 44 03 56 0c 45 84 db 89 Message from syslogd@ibclient at Aug 8 22:06:38 ... kernel:watchdog: BUG: soft lockup - CPU#4 stuck for 23s! [ksoftirqd/4:36] Message from syslogd@ibclient at Aug 8 22:06:38 ... kernel:watchdog: BUG: soft lockup - CPU#12 stuck for 21s! [ksoftirqd/12:84] I am going to leave this running overnight and see if I wake up to another hard lockup. Thanks as always, again the patch has helped in two ways. 1. It avoids the clone failure and requeue -11 failure messages 2. It seems to take a lot longe rnow to get into a hard lockup. Regards Laurence
On 08/08/2017 10:28 PM, Laurence Oberman wrote: > On Tue, 2017-08-08 at 20:11 -0400, Laurence Oberman wrote: >>> On Tue, 2017-08-08 at 22:17 +0800, Ming Lei wrote: >>>>> Hi Guys, >>>>> >>>>> Laurence and I see a system lockup issue when running >>> concurrent >>>>> big buffered write(4M bytes) to IB SRP on v4.13-rc3. >>>>> >>>>> 1 how to reproduce >>>>> >>>>> 1) setup IB_SRR & multi path >>>>> >>>>> #./start_opensm.sh >>>>> #./start_srp.sh >>>>> >>>>> # cat start_opensm.sh >>>>> #!/bin/bash >>>>> rmmod ib_srpt >>>>> opensm -F opensm.1.conf & >>>>> opensm -F opensm.2.conf & >>>>> >>>>> # cat start_srp.sh >>>>> run_srp_daemon -f /etc/ddn/srp_daemon.conf -R 30 -T 10 >>> -t 7000 >>>>> -ance -i mlx5_0 -p 1 1>/root/srp1.log 2>&1 & >>>>> run_srp_daemon -f /etc/ddn/srp_daemon.conf -R 30 -T 10 >>> -t 7000 >>>>> -ance -i mlx5_1 -p 1 1>/root/srp2.log 2>&1 & >>>>> >>>>> >>>>> 2) run the following test script >>>>> >>>>> #./test-mp.sh >>>>> >>>>> #cat test-mp.sh >>>>> #!/bin/sh >>>>> DEVS="360014051186d24cc27b4d049ffff94e3 >>>>> 36001405b2b5c6c24c084b6fa4d55da2f >>>>> 36001405b26ebe76dcb94a489f6f245f8" >>>>> >>>>> for i in $DEVS; do >>>>> for j in `seq 0 15`; do >>>>> ./hammer_write.sh $i 1>/dev/null 2>&1 & >>>>> done >>>>> done >>>>> >>>>> #cat hammer_write.sh >>>>> #!/bin/bash >>>>> while true; do >>>>> dd if=/dev/zero of=/dev/mapper/$1 bs=4096k >>> count=800 >>>>> done >>>>> >>>>> >>>>> 2 lockup log >>>>> [root@ibclient ~]# ./test-mp.sh >>>>> [root@ibclient ~]# [ 169.095494] perf: interrupt took too long >>>>> (2575 >>>>>>> 2500), lowering kernel.perf_event_max_sample_rate to 77000 >>> >>>>> >>>>> [ 178.569124] perf: interrupt took too long (3249 > 3218), >>>>> lowering >>>>> kernel.perf_event_max_sample_rate to 61000 >>>>> [ 190.586624] perf: interrupt took too long (4066 > 4061), >>>>> lowering >>>>> kernel.perf_event_max_sample_rate to 49000 >>>>> [ 210.381836] perf: interrupt took too long (5083 > 5082), >>>>> lowering >>>>> kernel.perf_event_max_sample_rate to 39000 >>>>> [ 240.498659] perf: interrupt took too long (6373 > 6353), >>>>> lowering >>>>> kernel.perf_event_max_sample_rate to 31000 >>>>> >>>>> [root@ibclient ~]# >>>>> [root@ibclient ~]# [ 292.252795] perf: interrupt took too long >>>>> (8006 >>>>>>> 7966), lowering kernel.perf_event_max_sample_rate to 24000 >>> >>>>> >>>>> [ 410.354443] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.432725] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.499530] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.564952] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.644493] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.698091] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.755175] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.796784] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.837690] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 410.878743] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.476823] multipath_clone_and_map: 32 callbacks suppressed >>>>> [ 463.506773] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.586752] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.656880] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.698919] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.723572] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.751820] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.781110] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.828267] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.856997] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 463.885998] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.572060] multipath_clone_and_map: 5201 callbacks >>> suppressed >>>>> [ 468.602191] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.655761] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.697540] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.738610] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.779597] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.820705] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.862712] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.904662] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.945132] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 468.985128] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.643013] multipath_clone_and_map: 6569 callbacks >>> suppressed >>>>> [ 473.673466] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.713929] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.754989] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.796069] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.849768] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.891640] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.933321] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 473.973791] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 474.016739] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 474.059858] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 478.732265] multipath_clone_and_map: 6310 callbacks >>> suppressed >>>>> [ 478.762164] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 478.802327] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 478.844235] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 478.886113] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 478.927164] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 478.968141] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 479.008772] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 479.060225] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 479.102873] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 479.145341] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 483.834884] multipath_clone_and_map: 5703 callbacks >>> suppressed >>>>> [ 483.865056] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 483.906792] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 483.959492] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.001768] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.044785] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.087289] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.129854] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.171317] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.212314] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.263931] device-mapper: multipath: blk_get_request() >>> returned >>>>> -11 - requeuing >>>>> [ 484.982608] perf: interrupt took too long (10021 > 10007), >>>>> lowering kernel.perf_event_max_sample_rate to 19000 >>>>> [ 485.362787] watchdog: BUG: soft lockup - CPU#11 stuck for >>> 21s! >>>>> [ksoftirqd/11:78] >>>>> [ 485.400762] Modules linked in: dm_round_robin xt_CHECKSUM >>>>> ipt_MASQUERADE nf_nat_masquerade_ipv4 tun ip6t_rpfilter >>> ipt_REJECT >>>>> nf_reject_ipv4 ip6t_REJECT nf_reject_ipv6 xt_conntrack ip_set >>>>> nfnetlink ebtable_nat ebtable_broute bridge stp llc >>> ip6table_nat >>>>> nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle >>>>> ip6table_security ip6table_raw iptable_nat nf_conntrack_ipv4 >>>>> nf_defrag_ipv4 nf_nat_ipv4 nf_nat rpcrdma nf_conntrack ib_isert >>>>> iscsi_target_mod iptable_mangle iptable_security iptable_raw >>>>> ebtable_filter ebtables ib_iser ip6table_filter ip6_tables >>> libiscsi >>>>> iptable_filter scsi_transport_iscsi target_core_mod ib_srp >>>>> scsi_transport_srp ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad >>>>> rdma_cm >>>>> ib_cm iw_cm mlx5_ib ib_core intel_powerclamp coretemp kvm_intel >>> kvm >>>>> irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel >>>>> [ 485.809222] pcbc joydev ipmi_si aesni_intel sg hpwdt hpilo >>>>> crypto_simd iTCO_wdt cryptd iTCO_vendor_support ipmi_devintf >>>>> gpio_ich >>>>> glue_helper shpchp lpc_ich ipmi_msghandler acpi_power_meter >>> pcspkr >>>>> i7core_edac pcc_cpufreq nfsd auth_rpcgss nfs_acl lockd grace >>> sunrpc >>>>> dm_multipath ip_tables xfs libcrc32c sd_mod amdkfd amd_iommu_v2 >>>>> radeon mlx5_core i2c_algo_bit drm_kms_helper syscopyarea >>>>> sysfillrect >>>>> sysimgblt fb_sys_fops ttm mlxfw drm ptp hpsa pps_core >>>>> scsi_transport_sas crc32c_intel bnx2 i2c_core serio_raw devlink >>>>> dm_mirror dm_region_hash dm_log dm_mod [last unloaded: ib_srpt] >>>>> [ 486.096352] CPU: 11 PID: 78 Comm: ksoftirqd/11 Tainted: >>>>> G I 4.13.0-rc3+ #23 >>>>> [ 486.142775] Hardware name: HP ProLiant DL380 G7, BIOS P67 >>>>> 08/16/2015 >>>>> [ 486.178816] task: ffff8a3fe453df40 task.stack: >>> ffff9a6cc65a8000 >>>>> [ 486.212674] RIP: 0010:__blk_recalc_rq_segments+0xb8/0x3d0 >>>>> [ 486.242444] RSP: 0018:ffff9a6cc65aba60 EFLAGS: 00000282 >>>>> ORIG_RAX: >>>>> ffffffffffffff10 >>>>> [ 486.284086] RAX: ffff8a4a8e26e440 RBX: 0000000000001000 RCX: >>>>> ffff8a4a8e26e440 >>>>> [ 486.323445] RDX: 0000000000000000 RSI: 0000000000001000 RDI: >>>>> ffff8a4a8e2699c8 >>>>> [ 486.363614] RBP: ffff9a6cc65abae0 R08: 0000000000007000 R09: >>>>> 0000000000000001 >>>>> [ 486.404621] R10: 0000000000001000 R11: 0000000000000000 R12: >>>>> 0000000000000000 >>>>> [ 486.445127] R13: 0000000000000000 R14: 0000000000000001 R15: >>>>> 0000000000000007 >>>>> [ 486.486120] FS: 0000000000000000(0000) >>>>> GS:ffff8a4b66940000(0000) >>>>> knlGS:0000000000000000 >>>>> [ 486.532771] CS: 0010 DS: 0000 ES: 0000 CR0: >>> 0000000080050033 >>>>> [ 486.565772] CR2: 00007f32bfcee000 CR3: 0000000d67609000 CR4: >>>>> 00000000000006e0 >>>>> [ 486.606325] Call Trace: >>>>> [ 486.620269] ? mempool_free+0x2b/0x80 >>>>> [ 486.641248] blk_recalc_rq_segments+0x28/0x40 >>>>> [ 486.666754] blk_update_request+0x249/0x310 >>>>> [ 486.690838] end_clone_bio+0x46/0x70 [dm_mod] >>>>> [ 486.715949] bio_endio+0xa1/0x120 >>>>> [ 486.734408] blk_update_request+0xb7/0x310 >>>>> [ 486.757313] scsi_end_request+0x38/0x1d0 >>>>> [ 486.779375] scsi_io_completion+0x13c/0x630 >>>>> [ 486.802928] scsi_finish_command+0xd9/0x120 >>>>> [ 486.825875] scsi_softirq_done+0x12a/0x150 >>>>> [ 486.849029] __blk_mq_complete_request+0x90/0x140 >>>>> [ 486.875484] blk_mq_complete_request+0x16/0x20 >>>>> [ 486.900631] scsi_mq_done+0x21/0x70 >>>>> [ 486.920462] srp_recv_done+0x48f/0x610 [ib_srp] >>>>> [ 486.946212] __ib_process_cq+0x5c/0xb0 [ib_core] >>>>> [ 486.972576] ib_poll_handler+0x22/0x70 [ib_core] >>>>> [ 486.998638] irq_poll_softirq+0xaf/0x110 >>>>> [ 487.021727] __do_softirq+0xc9/0x269 >>>>> [ 487.042027] run_ksoftirqd+0x29/0x50 >>>>> [ 487.062675] smpboot_thread_fn+0x110/0x160 >>>>> [ 487.085597] kthread+0x109/0x140 >>>>> [ 487.104113] ? sort_range+0x30/0x30 >>>>> [ 487.124120] ? kthread_park+0x60/0x60 >>>>> [ 487.144624] ret_from_fork+0x25/0x30 >>>>> [ 487.164640] Code: c3 83 fa 09 0f 84 8e 01 00 00 48 8b 44 24 >>> 20 >>>>> 48 >>>>> 8b 4c 24 28 45 31 ff 45 31 c0 45 31 c9 0f b6 80 36 06 00 00 88 >>> 44 >>>>> 24 >>>>> 17 8b 59 28 <44> 8b 71 2c 44 8b 61 34 85 db 0f 84 92 01 00 00 >>> 45 89 >>>>> f5 45 89 >>>>> >>>>> Message from syslogd@ibclient at Aug 8 04:12:23 ... >>>>> kernel:watchdog: BUG: soft lockup - CPU#11 stuck for 21s! >>>>> [ksoftirqd/11:78] >>>>> >>>>> >>>>> >>>>> Thanks, >>>>> Ming >> >>> >>> Hello Bart >>> >>> I received your message but my work email is not seeing it. >>> So I am replying to the original email by Ming so I can properly >>> respond to your most recent message pasted below and keep the >> thread >>> in >>> the right flow. >>> >>> I am purposefully testing buffered I/O because as you know that was >> a >>> source of pain for us last year. >>> So yes my intention always is to run large buffered and un-buffered >>> (direct) sequential I/O as its part of my regular tests that have >>> been >>> very useful in discovering many issues. >>> I also have smaller random and sequential I/O tests driven by fio >>> that >>> were not seeing the issue. >>> >>> I will go ahead and test your patch below. I am not surprised Ming >>> was >>> able to make this happen on the stock 4.13-RC3 but he had to run 16 >>> parallel jobs. >>> >>> It was happening on his kernel with only 3 parallel jobs and had >> not >>> been happening for me with 3 parallel runs on stock upstream but >>> perhaps just takes longer on 4.13-RC3 so 16 parallel runs brings it >>> out. >>> >>> Just one other small point, this is a "hard", not soft lockup. >>> >>> Unfortunately on the 7.3 base I am unable to capture a full vmcore >>> when >>> running the latest upstream kernel so if I need to get a vmcore I >>> will >>> have to use Fedora and the latest upstream. >>> >>> I will do that if your patch is not successful but reading it and >>> looking at the stack dumps I have, it makes sense. >>> >>> back in a bit >>> >>> Thanks >>> Laurence >>> >>> ---- >>> Hello Laurence, >>> >>> Is your goal perhaps to simulate a DDN workload? In that case I >> think >>> you >>> need oflag=direct to the dd argument list such that the page cache >>> writeback >>> code does not alter the size of the write requests. Anyway, this >> test >>> should >>> not trigger a lockup. Can you check whether the patch below makes >> the >>> soft >>> lockup complaints disappear (without changing the hammer_write.sh >>> test >>> script)? >>> >>> Thanks, >>> >>> Bart. >>> ----------------------------------------------------------------- >> ---- >>> ------- >>> [PATCH] block: Make blk_mq_delay_kick_requeue_list() rerun the >> queue >>> at >>> a >>> quiet time >>> >>> Drivers like dm-mpath requeue requests if no paths are available >> and >>> if configured to do so. If the queue depth is sufficiently high and >>> the queue rerunning delay sufficiently short then .requeue_work can >>> be queued so often that other work items queued on the same work >>> queue >>> do not get executed. Avoid that this happens by only rerunning the >>> queue after no blk_mq_delay_kick_requeue_list() calls have occurred >>> during @msecs milliseconds. Since the device mapper core is the >> only >>> user of blk_mq_delay_kick_requeue_list(), modify the implementation >>> of this function instead of creating a new function. >>> >>> Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> >>> --- >>> block/blk-mq.c | 4 ++-- >>> 1 file changed, 2 insertions(+), 2 deletions(-) >>> >>> diff --git a/block/blk-mq.c b/block/blk-mq.c >>> index 041f7b7fa0d6..8bfea36e92f9 100644 >>> --- a/block/blk-mq.c >>> +++ b/block/blk-mq.c >>> @@ -679,8 +679,8 @@ EXPORT_SYMBOL(blk_mq_kick_requeue_list); >>> void blk_mq_delay_kick_requeue_list(struct request_queue *q, >>> unsigned long msecs) >>> { >>> - kblockd_schedule_delayed_work(&q->requeue_work, >>> - msecs_to_jiffies(msecs)); >>> + kblockd_mod_delayed_work_on(WORK_CPU_UNBOUND, &q- >>>> requeue_work, >>> + msecs_to_jiffies(msecs)); >>> } >>> EXPORT_SYMBOL(blk_mq_delay_kick_requeue_list); >>> ----- > > Hello Bart > > So the patch is working to avoid the clone failures messages and > requeue -11 messages of my original issue however both the hard lockup > and soft lockup persists. It's is a lot harder to get the hard lockup > now though with the patch. > > The soft LOCKUPS due to the spins on the ksoftirqd are almost expected > because of how hard we are driving this and are easy to recreate. > > I did boot the original Ming kernel and reproduced first for baseline. > I then patched it with your simple patch. > > I was not paying enough attention to Ming's message but he was seeing a > ksoftirq soft LOCKUP on the stock upstream kernel NOT the hard LOCKUP. > I did not check his log message carefully so assumed he meant hard > LOCKUP. > > I guess that's why you said "soft LOCKUP" > > I did get into a few soft lockups before I hit another hard LOCKUP. > > So I am now again not sure about the hard LOCKUP on the stock and 7.4 > kernel. I will have to retest and baseline myself as Ming is away now > on vacation. > > The soft lockups still happen but they recover unless panic_on_ > soft_lockup is set and I almost expect them with how the softirqd is > spinning here. > > We see high spin rates here and I am only running 7 parallel writes. > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ > COMMAND > 20 root 20 0 0 0 0 R 100.0 0.0 0:27.11 > kworker/u6+ > 24 root 20 0 0 0 0 R 100.0 0.0 0:21.56 > ksoftirqd/2 > 60 root 20 0 0 0 0 R 100.0 0.0 0:34.61 > ksoftirqd/8 > > We see high SYS time because of the spinning so the soft LOCKUP is not > unexpected to be honest and it recovers. > > # <----CPU[HYPER]-----><-----------Memory-----------><--------- > -Disks-----------> > #Time cpu sys inter ctxsw Free Buff Cach Inac Slab Map > KBRead Reads KBWrit Writes > 21:32:11 52 52 11526 45981 65G 21G 541M 22G 7G > 209M 45 11 709791 211 > 21:32:12 49 49 13315 26037 65G 21G 541M 22G 6G > 209M 0 0 606356 195 > 21:32:13 47 47 13375 21456 65G 21G 541M 22G 6G > 209M 0 0 591468 250 > 21:32:14 33 33 12635 978 65G 21G 541M 22G 5G > 209M 0 0 560860 365 > 21:32:15 40 40 14304 734 67G 19G 541M 20G 5G > 209M 0 0 429472 97 > 21:32:16 36 36 16060 6220 70G 17G 541M 17G 4G > 209M 0 0 312852 54 > 21:32:17 31 31 16071 4962 71G 16G 540M 16G 4G > 209M 24 6 253228 44 > 21:32:18 31 31 15855 868 70G 17G 540M 18G 6G > 209M 104 26 237696 62 > 21:32:19 31 31 15983 661 70G 17G 541M 17G 7G > 209M 0 0 162340 37 > 21:32:20 32 32 16245 1413 70G 17G 541M 18G 9G > 209M 68 17 155648 19 > 21:32:21 44 44 14919 21711 68G 19G 540M 19G 10G > 209M 0 0 303104 32 > 21:32:22 59 59 11661 69544 68G 19G 541M 20G 10G > 209M 0 0 620919 84 > 21:32:23 59 59 12951 1058 67G 20G 540M 20G 10G > 209M 0 0 596582 126 > 21:32:24 68 67 12991 2077 66G 21G 540M 21G 10G > 209M 0 0 689504 168 > 21:32:25 78 78 13295 2927 66G 21G 541M 21G 10G > 209M 0 0 667673 144 > 21:32:26 66 66 13498 7670 65G 21G 541M 22G 10G > 209M 0 0 626874 98 > 21:32:27 55 55 12203 2663 65G 21G 541M 22G 9G > 209M 0 0 652096 112 > 21:32:28 47 47 12567 27804 65G 21G 541M 22G 8G > 209M 0 0 534789 137 > > We are driving this system pretty hard here and the target server is > seeing queue fulls and I am used to these if I drive it too hard. > > [ 568.707572] ib_srpt srpt_write_pending: IB send queue full (needed > 97) > [ 568.745714] ib_srpt srpt_write_pending: IB send queue full (needed > 97) > [ 568.783735] ib_srpt srpt_write_pending: IB send queue full (needed > 97) > [ 568.821583] ib_srpt srpt_write_pending: IB send queue full (needed > 97) > [ 568.858238] ib_srpt srpt_write_pending: IB send queue full (needed > 97) > [ 568.895975] ib_srpt srpt_queue_response: IB send queue full (needed > 0) > [ 568.934253] ib_srpt srpt_write_pending: IB send queue full (needed > 97) > [ 568.971040] ib_srpt srpt_write_pending: IB send queue full (needed > 49) > > So your patch helps to avoid easily getting the hard LOCKUP, does not > help the Soft lockup (as expected). > > [ 329.305507] watchdog: BUG: soft lockup - CPU#12 stuck for 21s! > [ksoftirqd/12:84] > [ 329.305508] Modules linked in: dm_round_robin mq_deadline > xt_CHECKSUM ipt_MASQUERADE nf_nat_masquerade_ipv4 tun ip6t_rpfilter > ipt_REJECT nf_reject_ipv4 ip6t_REJECT nf_reject_ipv6 xt_conntrack > ip_set nfnetlink ebtable_nat ebtable_broute bridge stp llc ip6table_nat > nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle > ip6table_security ip6table_raw iptable_nat nf_conntrack_ipv4 > nf_defrag_ipv4 nf_nat_ipv4 nf_nat rpcrdma nf_conntrack ib_isert > iptable_mangle iptable_security iscsi_target_mod iptable_raw > ebtable_filter ebtables ip6table_filter ib_iser ip6_tables libiscsi > iptable_filter scsi_transport_iscsi target_core_mod ib_srp > scsi_transport_srp ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad rdma_cm > ib_cm iw_cm mlx5_ib ib_core intel_powerclamp coretemp kvm_intel kvm > irqbypass crct10dif_pclmul crc32_pclmul > [ 329.305543] ghash_clmulni_intel pcbc joydev aesni_intel ipmi_si sg > hpilo ipmi_devintf hpwdt crypto_simd iTCO_wdt gpio_ich > iTCO_vendor_support cryptd shpchp acpi_power_meter ipmi_msghandler > i7core_edac glue_helper pcspkr lpc_ich pcc_cpufreq nfsd auth_rpcgss > nfs_acl lockd grace sunrpc dm_multipath ip_tables xfs libcrc32c sd_mod > amdkfd amd_iommu_v2 radeon mlx5_core i2c_algo_bit drm_kms_helper > syscopyarea sysfillrect sysimgblt fb_sys_fops ttm mlxfw ptp drm > pps_core hpsa i2c_core devlink crc32c_intel serio_raw bnx2 > scsi_transport_sas dm_mirror dm_region_hash dm_log dm_mod [last > unloaded: ib_srpt] > [ 329.305564] CPU: 12 PID: 84 Comm: ksoftirqd/12 Tainted: > G I 4.13.0-rc3.ming.bart+ #1 > [ 329.305566] Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 > [ 329.305567] task: ffff91a94c070040 task.stack: ffffaa8ec65d8000 > [ 329.305573] RIP: 0010:__blk_recalc_rq_segments+0x1a7/0x3d0 > [ 329.305574] RSP: 0018:ffffaa8ec65dba60 EFLAGS: 00000202 ORIG_RAX: > ffffffffffffff10 > [ 329.305575] RAX: 0000000000000001 RBX: 0000000000001000 RCX: > ffff91b1149239c0 > [ 329.305576] RDX: 0000000000001000 RSI: 0000000000000001 RDI: > 0000000000001000 > [ 329.305577] RBP: ffffaa8ec65dbae0 R08: 0000000000001000 R09: > 0000000000000001 > [ 329.305578] R10: 0000000b2a506000 R11: 0000000000000000 R12: > 0000000000000000 > [ 329.305579] R13: 0000000000000000 R14: 0000000000000000 R15: > 000000000000002e > [ 329.305580] FS: 0000000000000000(0000) GS:ffff91b3b7b80000(0000) > knlGS:0000000000000000 > [ 329.305581] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 329.305582] CR2: 00007ff4591f1860 CR3: 0000000985e09000 CR4: > 00000000000006e0 > [ 329.305583] Call Trace: > [ 329.305588] ? mempool_free+0x2b/0x80 > [ 329.305590] blk_recalc_rq_segments+0x28/0x40 > [ 329.305592] blk_update_request+0x249/0x310 > [ 329.305603] end_clone_bio+0x46/0x70 [dm_mod] > [ 329.305607] bio_endio+0xa1/0x120 > [ 329.305608] blk_update_request+0xb7/0x310 > [ 329.305611] scsi_end_request+0x38/0x1d0 > [ 329.305613] scsi_io_completion+0x13c/0x630 > [ 329.305615] scsi_finish_command+0xd9/0x120 > [ 329.305616] scsi_softirq_done+0x12a/0x150 > [ 329.305618] __blk_mq_complete_request+0x90/0x140 > [ 329.305624] blk_mq_complete_request+0x16/0x20 > [ 329.305625] scsi_mq_done+0x21/0x70 > [ 329.305628] srp_recv_done+0x48f/0x610 [ib_srp] > [ 329.305639] __ib_process_cq+0x5c/0xb0 [ib_core] > [ 329.305646] ib_poll_handler+0x22/0x70 [ib_core] > [ 329.305649] irq_poll_softirq+0xaf/0x110 > [ 329.305652] __do_softirq+0xc9/0x269 > [ 329.305655] run_ksoftirqd+0x29/0x50 > [ 329.305657] smpboot_thread_fn+0x110/0x160 > [ 329.305659] kthread+0x109/0x140 > [ 329.305660] ? sort_range+0x30/0x30 > [ 329.305661] ? kthread_park+0x60/0x60 > [ 329.305663] ret_from_fork+0x25/0x30 > [ 329.305664] Code: 8b 74 24 30 48 8b 7c 24 38 41 83 c7 01 41 89 d0 41 > b9 01 00 00 00 48 89 74 24 40 48 89 7c 24 48 8b 41 14 0f b6 f0 25 fb 00 > 00 00 <83> f8 03 0f 84 a0 00 00 00 83 fe 05 0f 84 97 00 00 00 83 fe 09 > [ 331.465885] CPU: 4 PID: 36 Comm: ksoftirqd/4 Tainted: > G I L 4.13.0-rc3.ming.bart+ #1 > [ 331.516413] Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 > [ 331.551403] task: ffff91a94c76df40 task.stack: ffffaa8ec6458000 > [ 331.584080] RIP: 0010:__blk_recalc_rq_segments+0xdf/0x3d0 > [ 331.614515] RSP: 0018:ffffaa8ec645ba60 EFLAGS: 00000286 ORIG_RAX: > ffffffffffffff10 > [ 331.656098] RAX: 0000000000001000 RBX: 0000000000001000 RCX: > ffff91b29faecdc0 > [ 331.696577] RDX: ffffe839e5d75300 RSI: ffff91b29faece48 RDI: > ffff91b2831f8348 > [ 331.736607] RBP: ffffaa8ec645bae0 R08: 0000000000004000 R09: > 0000000000000001 > [ 331.776129] R10: 0000000000000000 R11: 0000000000000000 R12: > 0000000000000000 > [ 331.817085] R13: 0000000000000000 R14: 0000000000000000 R15: > 0000000000000021 > [ 331.857671] FS: 0000000000000000(0000) GS:ffff91b3b7a80000(0000) > knlGS:0000000000000000 > [ 331.904232] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 331.935249] CR2: 00007f404526e000 CR3: 0000000985e09000 CR4: > 00000000000006e0 > [ 331.976273] Call Trace: > [ 331.990216] ? mempool_free+0x2b/0x80 > [ 332.010310] blk_recalc_rq_segments+0x28/0x40 > [ 332.034810] blk_update_request+0x249/0x310 > [ 332.058259] end_clone_bio+0x46/0x70 [dm_mod] > [ 332.082851] bio_endio+0xa1/0x120 > [ 332.101145] blk_update_request+0xb7/0x310 > [ 332.123965] scsi_end_request+0x38/0x1d0 > [ 332.145943] scsi_io_completion+0x13c/0x630 > [ 332.169357] scsi_finish_command+0xd9/0x120 > [ 332.193505] scsi_softirq_done+0x12a/0x150 > [ 332.217044] __blk_mq_complete_request+0x90/0x140 > [ 332.244498] blk_mq_complete_request+0x16/0x20 > [ 332.269971] scsi_mq_done+0x21/0x70 > [ 332.289591] srp_recv_done+0x48f/0x610 [ib_srp] > [ 332.315609] __ib_process_cq+0x5c/0xb0 [ib_core] > [ 332.342144] ib_poll_handler+0x22/0x70 [ib_core] > [ 332.368724] irq_poll_softirq+0xaf/0x110 > [ 332.392595] __do_softirq+0xc9/0x269 > [ 332.413220] run_ksoftirqd+0x29/0x50 > [ 332.433214] smpboot_thread_fn+0x110/0x160 > [ 332.456254] kthread+0x109/0x140 > [ 332.475263] ? sort_range+0x30/0x30 > [ 332.494624] ? kthread_park+0x60/0x60 > [ 332.515413] ret_from_fork+0x25/0x30 > [ 332.535180] Code: 17 8b 59 28 44 8b 71 2c 44 8b 61 34 85 db 0f 84 92 > 01 00 00 45 89 f5 45 89 e2 4c 89 ee 48 c1 e6 04 48 03 71 78 8b 46 08 48 > 8b 16 <44> 29 e0 48 89 54 24 30 39 d8 0f 47 c3 44 03 56 0c 45 84 db 89 > > Message from syslogd@ibclient at Aug 8 22:06:38 ... > kernel:watchdog: BUG: soft lockup - CPU#4 stuck for 23s! > [ksoftirqd/4:36] > > Message from syslogd@ibclient at Aug 8 22:06:38 ... > kernel:watchdog: BUG: soft lockup - CPU#12 stuck for 21s! > [ksoftirqd/12:84] > > I am going to leave this running overnight and see if I wake up to > another hard lockup. > > Thanks as always, again the patch has helped in two ways. > 1. It avoids the clone failure and requeue -11 failure messages > 2. It seems to take a lot longe rnow to get into a hard lockup. > > Regards > Laurence > Hello Bart Your latest patch on stock upstream without Ming's latest patches is behaving for me. As already mentioned, the requeue -11 and clone failure messages are gone and I am not actually seeing any soft lockups or hard lockups. When Ming gets back I will work with him on his patch set and the lockups. Running 10 parallel writes which easily trips into soft lockups on Ming's kernel (even with your patch) has been stable here on 4.13-RC3 with your patch. I will leave it running for a while now but the patch is good. If it survives 4 hours I will add a Tested-by to your latest patch. Thanks Laurence
On Wed, 2017-08-09 at 12:43 -0400, Laurence Oberman wrote: > Your latest patch on stock upstream without Ming's latest patches is > behaving for me. > > As already mentioned, the requeue -11 and clone failure messages are > gone and I am not actually seeing any soft lockups or hard lockups. > > When Ming gets back I will work with him on his patch set and the lockups. > > Running 10 parallel writes which easily trips into soft lockups on > Ming's kernel (even with your patch) has been stable here on 4.13-RC3 > with your patch. > > I will leave it running for a while now but the patch is good. > > If it survives 4 hours I will add a Tested-by to your latest patch. Hello Laurence, I'm working on an additional patch that should reduce unnecessary requeuing even further. I will let you know when it's ready. Additionally, please trim e-mails when replying such that e-mails do not get too long. Thanks, Bart.
On Wed, Aug 09, 2017 at 05:10:01PM +0000, Bart Van Assche wrote: > On Wed, 2017-08-09 at 12:43 -0400, Laurence Oberman wrote: > > Your latest patch on stock upstream without Ming's latest patches is > > behaving for me. > > > > As already mentioned, the requeue -11 and clone failure messages are > > gone and I am not actually seeing any soft lockups or hard lockups. > > > > When Ming gets back I will work with him on his patch set and the lockups. > > > > Running 10 parallel writes which easily trips into soft lockups on > > Ming's kernel (even with your patch) has been stable here on 4.13-RC3 > > with your patch. > > > > I will leave it running for a while now but the patch is good. > > > > If it survives 4 hours I will add a Tested-by to your latest patch. > > Hello Laurence, > > I'm working on an additional patch that should reduce unnecessary requeuing > even further. I will let you know when it's ready. > > Additionally, please trim e-mails when replying such that e-mails do not get > too long. soft lockup still can be observed easily with patch d4acf3650c7c( block: Make blk_mq_delay_kick_requeue_list() rerun the queue at a quiet time), but no hard lockup. With the patchset of 'blk-mq-sched: improve SCSI-MQ performance', hard lockup can be observed following some failure log: [ 269.277653] device-mapper: multipath: blk_get_request() returned -11 - requeuing [ 269.321244] device-mapper: multipath: blk_get_request() returned -11 - requeuing ... [ 273.421688] scsi host2: SRP abort called [ 273.444577] scsi host2: Sending SRP abort for tag 0x6007e [ 273.673871] scsi host2: Null scmnd for RSP w/tag 0x0000000006007e received on ch 6 / QP 0x30 ... [ 274.372110] device-mapper: multipath: blk_get_request() returned -11 - requeuing [ 278.658671] scsi host2: SRP abort called [ 278.690630] scsi host2: SRP abort called [ 278.717634] scsi host2: SRP abort called [ 278.745629] scsi host2: SRP abort called [ 279.083227] multipath_clone_and_map: 1092 callbacks suppressed .... [ 296.210503] scsi host2: SRP reset_device called .... [ 303.784287] NMI watchdog: Watchdog detected hard LOCKUP on cpu 10 The trick thing is that both hard lockup and soft lockup share one same stack trace. Another question, I don't understand why request is allocated with GFP_ATOMIC in multipath_clone_and_map(), looks it shouldn't be necessary. -- Ming
On Wed, Aug 23, 2017 at 07:35:26PM +0800, Ming Lei wrote: > On Wed, Aug 09, 2017 at 05:10:01PM +0000, Bart Van Assche wrote: > > On Wed, 2017-08-09 at 12:43 -0400, Laurence Oberman wrote: > > > Your latest patch on stock upstream without Ming's latest patches is > > > behaving for me. > > > > > > As already mentioned, the requeue -11 and clone failure messages are > > > gone and I am not actually seeing any soft lockups or hard lockups. > > > > > > When Ming gets back I will work with him on his patch set and the lockups. > > > > > > Running 10 parallel writes which easily trips into soft lockups on > > > Ming's kernel (even with your patch) has been stable here on 4.13-RC3 > > > with your patch. > > > > > > I will leave it running for a while now but the patch is good. > > > > > > If it survives 4 hours I will add a Tested-by to your latest patch. > > > > Hello Laurence, > > > > I'm working on an additional patch that should reduce unnecessary requeuing > > even further. I will let you know when it's ready. > > > > Additionally, please trim e-mails when replying such that e-mails do not get > > too long. > > soft lockup still can be observed easily with patch d4acf3650c7c( > block: Make blk_mq_delay_kick_requeue_list() rerun the queue at a quiet time), > but no hard lockup. > > With the patchset of 'blk-mq-sched: improve SCSI-MQ performance', hard > lockup can be observed following some failure log: > > [ 269.277653] device-mapper: multipath: blk_get_request() returned -11 - requeuing > [ 269.321244] device-mapper: multipath: blk_get_request() returned -11 - requeuing > ... > [ 273.421688] scsi host2: SRP abort called > [ 273.444577] scsi host2: Sending SRP abort for tag 0x6007e > [ 273.673871] scsi host2: Null scmnd for RSP w/tag 0x0000000006007e received on ch 6 / QP 0x30 > ... > [ 274.372110] device-mapper: multipath: blk_get_request() returned -11 - requeuing > [ 278.658671] scsi host2: SRP abort called > [ 278.690630] scsi host2: SRP abort called > [ 278.717634] scsi host2: SRP abort called > [ 278.745629] scsi host2: SRP abort called > [ 279.083227] multipath_clone_and_map: 1092 callbacks suppressed > .... > [ 296.210503] scsi host2: SRP reset_device called > .... > [ 303.784287] NMI watchdog: Watchdog detected hard LOCKUP on cpu 10 > > The trick thing is that both hard lockup and soft lockup share > one same stack trace. Actually I can reproduce hard lockup too on the latest linus tree(v4.13-rc6+) by just making the test more aggressive: 1) run hammer_write.sh 32-concurrently. 2) write 8M each time So it isn't the patchset of 'blk-mq-sched: improve SCSI-MQ performance' which causes the lockup. There must be issue somewhere else, and one thing I saw is that dm's req's completion handling isn't very efficient: - when low level driver's req is completed, blk_update_request() is called from scsi_end_request() - inside blk_update_request(), end_clone_bio() is called for each bio - in end_clone_bio(), blk_update_request() is called to do partial update for the dm req. That means if one request has N bios, the same dm req need to do the expensive partial update N times. In theory, the partial update might be avoided, I guess the reason is that the dm rq can't be retrieved via the low level driver's rq. [1] hard lockup log [ 434.068240] NMI watchdog: Watchdog detected hard LOCKUP on cpu 2 [ 434.068243] Modules linked in: dm_round_robin xt_CHECKSUM ipt_MASQUERADE nf_nat_masquerade_ipv4 tun ip6t_rpfilter ipt_REJECT nf_reject_ipv4 ip6t_REJECT nf_reject_ipv6 xt_conntrack ip_set nfnetlink ebtable_nat ebtable_broute bridge stp llc ip6table_nat nf_conntrack_ipv6 nf_defrag_ipv6 nf_nat_ipv6 ip6table_mangle ip6table_security ip6table_raw iptable_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack rpcrdma iptable_mangle ib_isert iscsi_target_mod iptable_security iptable_raw ebtable_filter ebtables ip6table_filter ib_iser libiscsi ip6_tables scsi_transport_iscsi iptable_filter target_core_mod ib_srp scsi_transport_srp ib_ipoib rdma_ucm ib_ucm ib_uverbs ib_umad rdma_cm ib_cm iw_cm mlx5_ib ib_core intel_powerclamp coretemp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel [ 434.068279] pcbc aesni_intel iTCO_wdt crypto_simd gpio_ich ipmi_si iTCO_vendor_support cryptd joydev glue_helper ipmi_devintf hpwdt acpi_power_meter hpilo sg ipmi_msghandler lpc_ich i7core_edac pcspkr shpchp pcc_cpufreq nfsd auth_rpcgss nfs_acl lockd grace sunrpc dm_multipath ip_tables xfs libcrc32c sd_mod amdkfd amd_iommu_v2 radeon mlx5_core i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops ttm mlxfw drm ptp hpsa pps_core crc32c_intel serio_raw bnx2 i2c_core devlink scsi_transport_sas dm_mirror dm_region_hash dm_log dm_mod [last unloaded: ib_srpt] [ 434.068314] CPU: 2 PID: 1039 Comm: dd Tainted: G I 4.13.0-rc6.linus+ #28 [ 434.068314] Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 [ 434.068316] task: ffff9458f4498040 task.stack: ffffbdb48fd10000 [ 434.068323] RIP: 0010:__blk_recalc_rq_segments+0xd9/0x3d0 [ 434.068324] RSP: 0018:ffff945b37a43d08 EFLAGS: 00000086 [ 434.068325] RAX: ffff9458978a16c0 RBX: 0000000000001000 RCX: ffff9458978a16c0 [ 434.068326] RDX: 0000000000000000 RSI: ffff9458978a1748 RDI: ffff9458978a69c8 [ 434.068327] RBP: ffff945b37a43d88 R08: 0000000000014000 R09: 0000000000000001 [ 434.068328] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 [ 434.068329] R13: 0000000000000000 R14: 0000000000000000 R15: 000000000000001d [ 434.068330] FS: 00007f27b13ea740(0000) GS:ffff945b37a40000(0000) knlGS:0000000000000000 [ 434.068331] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 434.068332] CR2: 00007f27aa600000 CR3: 0000000b1e2af000 CR4: 00000000000006e0 [ 434.068333] Call Trace: [ 434.068335] <IRQ> [ 434.068340] ? mempool_free+0x2b/0x80 [ 434.068343] blk_recalc_rq_segments+0x28/0x40 [ 434.068345] blk_update_request+0x249/0x310 [ 434.068363] end_clone_bio+0x46/0x70 [dm_mod] [ 434.068367] bio_endio+0xa1/0x120 [ 434.068370] blk_update_request+0xb7/0x310 [ 434.068374] scsi_end_request+0x38/0x1d0 [ 434.068375] scsi_io_completion+0x13c/0x630 [ 434.068378] scsi_finish_command+0xd9/0x120 [ 434.068381] scsi_softirq_done+0x12a/0x150 [ 434.068383] __blk_mq_complete_request_remote+0x13/0x20 [ 434.068387] flush_smp_call_function_queue+0x52/0x110 [ 434.068388] generic_smp_call_function_single_interrupt+0x13/0x30 [ 434.068391] smp_call_function_single_interrupt+0x27/0x40 [ 434.068393] call_function_single_interrupt+0x93/0xa0 [ 434.068396] RIP: 0010:copy_user_generic_string+0x2c/0x40 [ 434.068397] RSP: 0018:ffffbdb48fd13c78 EFLAGS: 00010246 ORIG_RAX: ffffffffffffff04 [ 434.068398] RAX: 00007f27aa1f3000 RBX: 0000000000001000 RCX: 0000000000000040 [ 434.068399] RDX: 0000000000000000 RSI: 00007f27aa1f2e00 RDI: ffff945a2f7b9e00 [ 434.068400] RBP: ffffbdb48fd13c80 R08: ffff945a2fcf20b0 R09: 0000000000001000 [ 434.068400] R10: 00000000fffff000 R11: 0000000000000000 R12: 0000000000001000 [ 434.068401] R13: ffff945a2f7ba000 R14: ffffbdb48fd13e20 R15: ffff945be4817950 [ 434.068402] </IRQ> [ 434.068405] ? I_BDEV+0x20/0x20 [ 434.068408] ? copyin+0x26/0x30 [ 434.068410] iov_iter_copy_from_user_atomic+0xbd/0x2d0 [ 434.068412] generic_perform_write+0xdd/0x1b0 [ 434.068415] __generic_file_write_iter+0x19b/0x1e0 [ 434.068416] blkdev_write_iter+0x8a/0x100 [ 434.068420] __vfs_write+0xf3/0x170 [ 434.068423] vfs_write+0xb2/0x1b0 [ 434.068427] ? syscall_trace_enter+0x1d0/0x2b0 [ 434.068429] SyS_write+0x55/0xc0 [ 434.068431] do_syscall_64+0x67/0x150 [ 434.068434] entry_SYSCALL64_slow_path+0x25/0x25 [ 434.068435] RIP: 0033:0x7f27b0f0dc90 [ 434.068436] RSP: 002b:00007ffe24d1d178 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 [ 434.068438] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f27b0f0dc90 [ 434.068438] RDX: 0000000000800000 RSI: 00007f27aa0fa000 RDI: 0000000000000001 [ 434.068439] RBP: 0000000000800000 R08: ffffffffffffffff R09: 0000000000802003 [ 434.068440] R10: 00007ffe24d1cf00 R11: 0000000000000246 R12: 0000000000800000 [ 434.068441] R13: 00007f27aa0fa000 R14: 00007f27aa8fa000 R15: 0000000000000000 [ 434.068442] Code: 06 00 00 88 44 24 17 8b 59 28 44 8b 71 2c 44 8b 61 34 85 db 0f 84 92 01 00 00 45 89 f5 45 89 e2 4c 89 ee 48 c1 e6 04 48 03 71 78 <8b> 46 08 48 8b 16 44 29 e0 48 89 54 24 30 39 d8 0f 47 c3 44 03 [ 434.068461] Kernel panic - not syncing: Hard LOCKUP [ 434.068463] CPU: 2 PID: 1039 Comm: dd Tainted: G I 4.13.0-rc6.linus+ #28 [ 434.068463] Hardware name: HP ProLiant DL380 G7, BIOS P67 08/16/2015 [ 434.068463] Call Trace: [ 434.068464] <NMI> [ 434.068467] dump_stack+0x63/0x87 [ 434.068470] panic+0xeb/0x245 [ 434.068472] nmi_panic+0x3f/0x40 [ 434.068477] watchdog_overflow_callback+0x106/0x130 [ 434.068480] __perf_event_overflow+0x54/0xe0 [ 434.068482] perf_event_overflow+0x14/0x20 [ 434.068486] intel_pmu_handle_irq+0x203/0x4b0 [ 434.068490] perf_event_nmi_handler+0x2d/0x50 [ 434.068492] nmi_handle+0x61/0x110 [ 434.068494] default_do_nmi+0xf6/0x120 [ 434.068495] do_nmi+0x113/0x190 [ 434.068497] end_repeat_nmi+0x1a/0x1e [ 434.068499] RIP: 0010:__blk_recalc_rq_segments+0xd9/0x3d0 [ 434.068499] RSP: 0018:ffff945b37a43d08 EFLAGS: 00000086 [ 434.068500] RAX: ffff9458978a16c0 RBX: 0000000000001000 RCX: ffff9458978a16c0 [ 434.068501] RDX: 0000000000000000 RSI: ffff9458978a1748 RDI: ffff9458978a69c8 [ 434.068502] RBP: ffff945b37a43d88 R08: 0000000000014000 R09: 0000000000000001 [ 434.068502] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 [ 434.068503] R13: 0000000000000000 R14: 0000000000000000 R15: 000000000000001d [ 434.068505] ? __blk_recalc_rq_segments+0xd9/0x3d0 [ 434.068508] ? __blk_recalc_rq_segments+0xd9/0x3d0 [ 434.068510] </NMI> [ 434.068510] <IRQ> [ 434.068512] ? mempool_free+0x2b/0x80 [ 434.068514] blk_recalc_rq_segments+0x28/0x40 [ 434.068516] blk_update_request+0x249/0x310 [ 434.068521] end_clone_bio+0x46/0x70 [dm_mod] [ 434.068522] bio_endio+0xa1/0x120 [ 434.068523] blk_update_request+0xb7/0x310 [ 434.068525] scsi_end_request+0x38/0x1d0 [ 434.068527] scsi_io_completion+0x13c/0x630 [ 434.068528] scsi_finish_command+0xd9/0x120 [ 434.068531] scsi_softirq_done+0x12a/0x150 [ 434.068534] __blk_mq_complete_request_remote+0x13/0x20 [ 434.068536] flush_smp_call_function_queue+0x52/0x110 [ 434.068537] generic_smp_call_function_single_interrupt+0x13/0x30 [ 434.068539] smp_call_function_single_interrupt+0x27/0x40 [ 434.068542] call_function_single_interrupt+0x93/0xa0 [ 434.068543] RIP: 0010:copy_user_generic_string+0x2c/0x40 [ 434.068544] RSP: 0018:ffffbdb48fd13c78 EFLAGS: 00010246 ORIG_RAX: ffffffffffffff04 [ 434.068545] RAX: 00007f27aa1f3000 RBX: 0000000000001000 RCX: 0000000000000040 [ 434.068546] RDX: 0000000000000000 RSI: 00007f27aa1f2e00 RDI: ffff945a2f7b9e00 [ 434.068546] RBP: ffffbdb48fd13c80 R08: ffff945a2fcf20b0 R09: 0000000000001000 [ 434.068547] R10: 00000000fffff000 R11: 0000000000000000 R12: 0000000000001000 [ 434.068548] R13: ffff945a2f7ba000 R14: ffffbdb48fd13e20 R15: ffff945be4817950 [ 434.068548] </IRQ> [ 434.068550] ? I_BDEV+0x20/0x20 [ 434.068551] ? copyin+0x26/0x30 [ 434.068553] iov_iter_copy_from_user_atomic+0xbd/0x2d0 [ 434.068554] generic_perform_write+0xdd/0x1b0 [ 434.068557] __generic_file_write_iter+0x19b/0x1e0 [ 434.068560] blkdev_write_iter+0x8a/0x100 [ 434.068562] __vfs_write+0xf3/0x170 [ 434.068565] vfs_write+0xb2/0x1b0 [ 434.068566] ? syscall_trace_enter+0x1d0/0x2b0 [ 434.068568] SyS_write+0x55/0xc0 [ 434.068569] do_syscall_64+0x67/0x150 [ 434.068572] entry_SYSCALL64_slow_path+0x25/0x25 [ 434.068573] RIP: 0033:0x7f27b0f0dc90 [ 434.068574] RSP: 002b:00007ffe24d1d178 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 [ 434.068575] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f27b0f0dc90 [ 434.068576] RDX: 0000000000800000 RSI: 00007f27aa0fa000 RDI: 0000000000000001 [ 434.068577] RBP: 0000000000800000 R08: ffffffffffffffff R09: 0000000000802003 [ 434.068578] R10: 00007ffe24d1cf00 R11: 0000000000000246 R12: 0000000000800000 [ 434.068579] R13: 00007f27aa0fa000 R14: 00007f27aa8fa000 R15: 0000000000000000 [ 434.100416] Kernel Offset: 0x13800000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) [ 438.636598] ---[ end Kernel panic - not syncing: Hard LOCKUP
On Wed, 2017-08-23 at 19:35 +0800, Ming Lei wrote: > soft lockup still can be observed easily with patch d4acf3650c7c( > block: Make blk_mq_delay_kick_requeue_list() rerun the queue at a quiet time), > but no hard lockup. Hello Ming, I never claimed that commit d4acf3650c7c was a full solution to avoid these soft lockups. I assume these results mean that patch series "[PATCH 0/7] Device mapper and dm-mpath fixes" was not applied to the kernel used in your tests? A colleague of you has reported that that patch series is sufficient to avoid these soft lockups. See also https://www.redhat.com/archives/dm-devel/2017-August/msg00122.html. Bart.
On Wed, Aug 23, 2017 at 11:12 PM, Bart Van Assche <Bart.VanAssche@wdc.com> wrote: > On Wed, 2017-08-23 at 19:35 +0800, Ming Lei wrote: >> soft lockup still can be observed easily with patch d4acf3650c7c( >> block: Make blk_mq_delay_kick_requeue_list() rerun the queue at a quiet time), >> but no hard lockup. > > Hello Ming, > > I never claimed that commit d4acf3650c7c was a full solution to avoid these > soft lockups. I assume these results mean that patch series "[PATCH 0/7] > Device mapper and dm-mpath fixes" was not applied to the kernel used in your > tests? A colleague of you has reported that that patch series is sufficient OK, thanks for your clarification. I run v4.13-rc6 with block for-next, so these patches are not applied in my kernel.
On Wed, Aug 23 2017 at 11:12am -0400, Bart Van Assche <Bart.VanAssche@wdc.com> wrote: > On Wed, 2017-08-23 at 19:35 +0800, Ming Lei wrote: > > soft lockup still can be observed easily with patch d4acf3650c7c( > > block: Make blk_mq_delay_kick_requeue_list() rerun the queue at a quiet time), > > but no hard lockup. > > Hello Ming, > > I never claimed that commit d4acf3650c7c was a full solution to avoid these > soft lockups. I assume these results mean that patch series "[PATCH 0/7] > Device mapper and dm-mpath fixes" was not applied to the kernel used in your > tests? A colleague of you has reported that that patch series is sufficient > to avoid these soft lockups. See also > https://www.redhat.com/archives/dm-devel/2017-August/msg00122.html. FYI, I'll try to review/stage that patchset for 4.14 over the next few days. (I'm not yet back to working full-time but will try to handle my upstream responsibilities) Mike
diff --git a/block/blk-mq.c b/block/blk-mq.c index 041f7b7fa0d6..8bfea36e92f9 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -679,8 +679,8 @@ EXPORT_SYMBOL(blk_mq_kick_requeue_list); void blk_mq_delay_kick_requeue_list(struct request_queue *q, unsigned long msecs) { - kblockd_schedule_delayed_work(&q->requeue_work, - msecs_to_jiffies(msecs)); + kblockd_mod_delayed_work_on(WORK_CPU_UNBOUND, &q->requeue_work, + msecs_to_jiffies(msecs)); } EXPORT_SYMBOL(blk_mq_delay_kick_requeue_list); -----