Bug 1861414

Summary:	SCHED_DEADLINE: warning on enqueue_task_dl
Product:	Red Hat Enterprise Linux 8	Reporter:	Daniel Bristot de Oliveira <daolivei>
Component:	kernel-rt	Assignee:	Phil Auld <pauld>
kernel-rt sub component:	Scheduler	QA Contact:	Chunyu Hu <chuhu>
Status:	CLOSED CURRENTRELEASE	Docs Contact:
Severity:	medium
Priority:	unspecified	CC:	bhu, cye, jlelli, lgoncalv, pauld, qzhao
Version:	8.1
Target Milestone:	rc
Target Release:	8.0
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2020-08-03 13:41:18 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Daniel Bristot de Oliveira 2020-07-28 14:53:32 UTC

I saw this warning while working on BZ1855447, while using the C version of the raggiro tool:


[ 5114.489908] WARNING: CPU: 12 PID: 334379 at kernel/sched/deadline.c:628 enqueue_task_dl+0x7ba/0x960
[ 5114.489908] Modules linked in: vfio_pci irqbypass vfio_virqfd vfio_iommu_type1 vfio veth nf_conntrack_netlink xt_recent xt_statistic xt_nat xt_addrtype vxlan ip6_udp_tunnel udp_tunnel xt_conntrack ipt_REJECT openvswitch nf_reject_ipv4 nf_conntrack_ipv6 nf_nat_ipv6 nf_defrag_ipv6 nf_conncount nft_chain_route_ipv4 ipt_MASQUERADE xt_comment nft_counter xt_mark nft_compat nft_chain_nat_ipv4 nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat nf_conntrack nf_tables nfnetlink overlay ib_isert iscsi_target_mod ib_srpt target_core_mod ib_srp scsi_transport_srp vfat fat i40iw ext4 mbcache jbd2 iw_cxgb4 rpcrdma rdma_ucm ib_iser ib_umad intel_rapl rdma_cm ib_ipoib skx_edac iw_cm ib_cm nfit libnvdimm intel_powerclamp coretemp intel_cstate iTCO_wdt iTCO_vendor_support dcdbas joydev intel_uncore intel_rapl_perf
[ 5114.489937]  pcspkr mgag200 ttm drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm mei_me i2c_i801 lpc_ich mei ipmi_ssif ipmi_si ipmi_devintf ipmi_msghandler acpi_cpufreq acpi_power_meter xfs libcrc32c mlx5_ib sd_mod sg ib_uverbs ib_core crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel mlx5_core mlxfw ahci igb libahci i40e i2c_algo_bit megaraid_sas dca libata dm_multipath sunrpc dm_mirror dm_region_hash dm_log dm_mod be2iscsi bnx2i cnic uio cxgb4i cxgb4 libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi
[ 5114.489966] CPU: 12 PID: 334379 Comm: raggiro Tainted: G        W        --------- -  - 4.18.0-147.8.1.rt24.101.el8_1.x86_64 #1
[ 5114.489966] Hardware name: Dell Inc. PowerEdge R740/00WGD1, BIOS 1.3.7 02/08/2018
[ 5114.489967] RIP: 0010:enqueue_task_dl+0x7ba/0x960
[ 5114.489968] Code: 4d fe ff ff 80 3d 2c 1b 3f 01 00 0f 85 30 fe ff ff 48 89 04 24 e8 b6 d2 ff ff 0f b6 8b 4c 03 00 00 48 8b 04 24 e9 17 fe ff ff <0f> 0b e9 03 fe ff ff 80 3d ff 1a 3f 01 00 0f 85 29 fe ff ff 48 89
[ 5114.489969] RSP: 0018:ffffa2755ca17e50 EFLAGS: 00010002
[ 5114.489970] RAX: 0000000000025d00 RBX: ffff8c394d5dddc0 RCX: 0000000000000002
[ 5114.489970] RDX: 0000000000025d00 RSI: ffff8c394d5dddc0 RDI: 0000000000000000
[ 5114.489971] RBP: ffff8c50df8e5d00 R08: 00000000ffffffff R09: 0000000000000000
[ 5114.489971] R10: 0000000000000000 R11: 0000000000000000 R12: ffff8c50df8e5d00
[ 5114.489972] R13: ffff8c394d5de0b8 R14: ffff8c50df8e5d00 R15: 000004a325bd05ab
[ 5114.489973] FS:  00007fa3dccff500(0000) GS:ffff8c38e0d80000(0000) knlGS:0000000000000000
[ 5114.489974] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 5114.489974] CR2: 000000c00026d000 CR3: 000000172a8fc001 CR4: 00000000007606e0
[ 5114.489975] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 5114.489975] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 5114.489975] PKRU: 55555554
[ 5114.489976] Call Trace:
[ 5114.489983]  __sched_setscheduler+0x519/0x840
[ 5114.489986]  __x64_sys_sched_setattr+0x7f/0xc0
[ 5114.489991]  do_syscall_64+0x5b/0x1b0
[ 5114.489997]  entry_SYSCALL_64_after_hwframe+0x65/0xca
[ 5114.489999] RIP: 0033:0x7fa3dc81299d
[ 5114.490000] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d bb 64 2c 00 f7 d8 64 89 01 48
[ 5114.490000] RSP: 002b:00007ffefecbfaa8 EFLAGS: 00000202 ORIG_RAX: 000000000000013a
[ 5114.490001] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fa3dc81299d
[ 5114.490002] RDX: 0000000000000000 RSI: 00007ffefecbfb10 RDI: 0000000000000055
[ 5114.490002] RBP: 00007ffefecbfac0 R08: 000000430000004c R09: 00007ffefecbfb10
[ 5114.490003] R10: 0000000000000007 R11: 0000000000000202 R12: 0000000000401210
[ 5114.490003] R13: 00007ffefedb9cb0 R14: 0000000000000000 R15: 0000000000000000
[ 5114.490004] ---[ end trace 0000000000000025 ]---

Comment 1 Juri Lelli 2020-07-28 16:12:21 UTC

Hi,

8.3 commit 4668c7f5b069 ("[kernel] sched/deadline: Initialize ->dl_boosted")
might fix this.

Daniel, would you be able to check if problem is still reproducible with
scratch available at

http://brew-task-repos.usersys.redhat.com/repos/scratch/jlelli/kernel-rt/4.18.0/147.23.1.rt24.117.el8_1.dlboost.test.cki.kt0/x86_64/

Thanks!

Comment 2 Daniel Bristot de Oliveira 2020-08-03 06:31:51 UTC

Hi Juri,

The problem stopped to reproduce with the kernel you provided in comment #1.

Thanks!
-- Daniel

Comment 3 Phil Auld 2020-08-03 13:41:18 UTC

Thanks Daniel.  Since that patch is in already (bz1854179), I'll close this as CURRENTRELEASE.

Comment 4 Juri Lelli 2020-08-03 16:55:58 UTC

Hi Phil,

(In reply to Phil Auld from comment #3)
> Thanks Daniel.  Since that patch is in already (bz1854179), I'll close this
> as CURRENTRELEASE.

The issue has been however reported on 8.1.z. And my scratch had that single
patch from 8.3 backported to current 8.1.z for testing.

Do you think we could backport the SCHED_DEADLINE patch to 8.2.z and 8.1.z?

Thanks.

Comment 5 Phil Auld 2020-08-03 17:17:25 UTC

Hi Juri,

Right, I didn't actually notice that part. We can probably do this in the zstreams. I wonder about the process for that given this bug did not have the flags and actual patch posted.  We could clone this one. Or we could clone bz1854179 and specify it's for the one patch. Neither feels quite right.  Also, though, this is not a customer issue, just something Daniel hit during other test when he happened to be on 8.1 based kernel, right?  It's a trivial patch but how important is it to zstream? 

Cheers,
Phil