clock_nanosleep02 from ubuntu_ltp_syscalls failed on some bare metal nodes with X-AWS-4.4 / X-AWS-4.15 / B-AWS-4.15 / B-oem-4.15

Bug #1872388 reported by Po-Hsu Lin
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
ubuntu-kernel-tests
New
Undecided
Unassigned

Bug Description

Issue found on i3.metal and c5.metal amd64 instances with aws 4.15.0-1066.70~16.04.1, passed on r5.metal and other instances

Test failed with:
 tst_timer_test.c:310: FAIL: clock_nanosleep() slept for too long

 startup='Sun Apr 12 19:33:14 2020'
 tst_test.c:1241: INFO: Timeout per run is 0h 05m 00s
 tst_timer_test.c:348: INFO: CLOCK_MONOTONIC resolution 1ns
 tst_timer_test.c:360: INFO: prctl(PR_GET_TIMERSLACK) = 50us
 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 1000us 500 iterations, threshold 450.01us
 tst_timer_test.c:307: INFO: min 1063us, max 1084us, median 1067us, trunc mean 1066.81us (discarded 25)
 tst_timer_test.c:322: PASS: Measured times are within thresholds
 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 2000us 500 iterations, threshold 450.01us
 tst_timer_test.c:307: INFO: min 2071us, max 2078us, median 2077us, trunc mean 2076.76us (discarded 25)
 tst_timer_test.c:322: PASS: Measured times are within thresholds
 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 5000us 300 iterations, threshold 450.04us
 tst_timer_test.c:307: INFO: min 5066us, max 5108us, median 5106us, trunc mean 5103.24us (discarded 15)
 tst_timer_test.c:322: PASS: Measured times are within thresholds
 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 10000us 100 iterations, threshold 450.33us
 tst_timer_test.c:307: INFO: min 10132us, max 10147us, median 10135us, trunc mean 10138.91us (discarded 5)
 tst_timer_test.c:322: PASS: Measured times are within thresholds
 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 25000us 50 iterations, threshold 451.29us
 tst_timer_test.c:307: INFO: min 25221us, max 25258us, median 25233us, trunc mean 25239.48us (discarded 2)
 tst_timer_test.c:322: PASS: Measured times are within thresholds
 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 100000us 10 iterations, threshold 537.00us
 tst_timer_test.c:307: INFO: min 100215us, max 100773us, median 100684us, trunc mean 100638.44us (discarded 1)
 tst_timer_test.c:310: FAIL: clock_nanosleep() slept for too long

 Time: us | Frequency
 --------------------------------------------------------------------------------
 100215 | ***********-
 100245 |
 100275 |
 100305 |
 100335 |
 100365 |
 100395 |
 100425 |
 100455 |
 100485 |
 100515 |
 100545 |
 100575 |
 100605 |
 100635 |
 100665 | ********************************************************************
 100695 | ***********-
 100725 | ***********-
 100755 | ***********-
 --------------------------------------------------------------------------------
 30us | 1 sample = 11.33333 '*', 22.66667 '+', 45.33333 '-', non-zero '.'

 tst_timer_test.c:264: INFO: clock_nanosleep() sleeping for 1000000us 2 iterations, threshold 4400.00us
 tst_timer_test.c:307: INFO: min 1003376us, max 1005030us, median 1003376us, trunc mean 1003376.00us (discarded 1)
 tst_timer_test.c:322: PASS: Measured times are within thresholds

 Summary:
 passed 6
 failed 1
 skipped 0
 warnings 0

Po-Hsu Lin (cypressyew)
tags: added: 4.15 aws sru-20200406 ubuntu-ltp-syscalls xenial
description: updated
Po-Hsu Lin (cypressyew)
summary: clock_nanosleep02 from ubuntu_ltp_syscalls failed on some bare metal
- nodes with X-AWS-4.15
+ nodes with X-AWS-4.15 / B-AWS-4.15
tags: added: bionic
Revision history for this message
Po-Hsu Lin (cypressyew) wrote : Re: clock_nanosleep02 from ubuntu_ltp_syscalls failed on some bare metal nodes with X-AWS-4.15 / B-AWS-4.15

This issue can be found on Bionic 4.15 AWS as well.

I tried to reproduce this manually, however with instance c5.large (which is failing with this case on B-AWS-4.15), I can't reproduce this with 4.15.0-1065-aws / 4.15.0-1066-aws by just running this clock_nanosleep02 test case.

Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

Can be found on Xenial AWS 4.4.0-1108 (i3.metal, c5.metal, pass on r5.metal)

tags: added: sru-20200518
Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

Found on B-oem 4.15.0-1094.104

summary: clock_nanosleep02 from ubuntu_ltp_syscalls failed on some bare metal
- nodes with X-AWS-4.15 / B-AWS-4.15
+ nodes with X-AWS-4.15 / B-AWS-4.15 / B-oem-4.15
Revision history for this message
Po-Hsu Lin (cypressyew) wrote : Re: clock_nanosleep02 from ubuntu_ltp_syscalls failed on some bare metal nodes with X-AWS-4.15 / B-AWS-4.15 / B-oem-4.15

Didn't see this on X-AWS-4.15 this cycle.

tags: added: sru-20200921
tags: added: 5.4 focal
Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

Still visible in Xenial 4.4.0-1118.132 AWS c5.metal

tags: added: 4.4 sru-20201109
summary: clock_nanosleep02 from ubuntu_ltp_syscalls failed on some bare metal
- nodes with X-AWS-4.15 / B-AWS-4.15 / B-oem-4.15
+ nodes with X-AWS-4.4 / X-AWS-4.15 / B-AWS-4.15 / B-oem-4.15
Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

Found on 4.15.0-1059.65 - oracle
BM.Standard2.52 only

Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

Spotted on B-4.15.0-1049.56 fips

tags: added: sru-20201130
Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

BTW on B-4.15 it's because the kernel config does not match its requirement.

Revision history for this message
Po-Hsu Lin (cypressyew) wrote :

Can see this on F-aws 5.4 but only on arm64 instance a1.2xlarge

Revision history for this message
Francis Ginther (fginther) wrote :

Seen with linux-aws 4.15.0-1094.101~16.04.1.

tags: added: sru-20210125
Revision history for this message
Kelsey Steele (kelsey-steele) wrote :

Found on bionic/fips 4.15.0-1053.61

tags: added: fips
Revision history for this message
Guilherme G. Piccoli (gpiccoli) wrote :

Observed in B/oracle (kernel 4.15), cycle sru-20210621.

tags: added: oracle sru-20210621
Po-Hsu Lin (cypressyew)
tags: added: sru-20230320
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.