cve-2018-1000204 failed in cve from ubuntu_ltp on Focal/Groovy/Hirsute/Impish

[Kelsey Steele]

65432. 10/09 23:16:14 DEBUG| utils:0153| [stdout] tst_test.c:1248: TINFO: Timeout per run is 0h 05m 00s
65433. 10/09 23:16:14 DEBUG| utils:0153| [stdout] ioctl_sg01.c:81: TINFO: Found SCSI device /dev/sg2
65434. 10/09 23:16:14 DEBUG| utils:0153| [stdout] tst_test.c:1294: TINFO: If you are running on slow machine, try exporting LTP_TIMEOUT_MUL > 1
65435. 10/09 23:16:14 DEBUG| utils:0153| [stdout] tst_test.c:1296: TBROK: Test killed! (timeout?)
65436. 10/09 23:16:14 DEBUG| utils:0153| [stdout]
65437. 10/09 23:16:14 DEBUG| utils:0153| [stdout] Summary:
65438. 10/09 23:16:14 DEBUG| utils:0153| [stdout] passed 0
65439. 10/09 23:16:14 DEBUG| utils:0153| [stdout] failed 0
65440. 10/09 23:16:14 DEBUG| utils:0153| [stdout] skipped 0
65441. 10/09 23:16:14 DEBUG| utils:0153| [stdout] warnings 0
65442. 10/09 23:16:14 DEBUG| utils:0153| [stdout] tag=cve-2018-1000204 stime=1602284878 dur=47 exit=exited stat=2 core=no cu=789 cs=3863

[Po-Hsu Lin (cypressyew)]

Spotted on 5.4.0-1030.32~18.04.1 with VM.DenseIO2.8

As this test is calling the ioctl_sg01, I think we can port the fix for ubuntu_ltp_syscalls in bug 1895281 to ubuntu_ltp

[Po-Hsu Lin (cypressyew)]

BTW for comment #1 it's oracle kernel.

https://kernel.ubuntu.com/git/ubuntu/autotest-client-tests.git/commit/?id=4c7df183775f8d23493e1a844ed47adf522c2e1f

Re-testing Azure / Oracle instances that failed with this.

[Po-Hsu Lin (cypressyew)]

For Azure Focal kernel, I can see this failing on:
5.4.0-1032.33-azure Standard_DS5_v2

Re-testing this.

[Po-Hsu Lin (cypressyew)]

With 3x timeout (15 mins) this test is still failing on 5.4.0-1030.32~18.04.1 with VM.DenseIO2.8 and 5.4.0-1032.33-azure Standard_DS5_v2.

This will need more tests.

[Po-Hsu Lin (cypressyew)]

Found on F-azure 5.4.0-1035.36

[Kelsey Steele (kelsey-steele)]

Spotted on Groovy 5.8.0-51.57 host wright-kernel

[Krzysztof Kozlowski (krzk)]

Found on focal/azure 5.8.0-1034.36~20.04.1-azure

[Krzysztof Kozlowski (krzk)]

Found on focal/oracle 5.8.0-1032.33~20.04.1

[Krzysztof Kozlowski (krzk)]

Found on groovy/oracle 5.8.0-1032.33

[Kleber Sacilotto de Souza (kleber-souza)]

Observed with focal/linux: 5.4.0-76.85.

[Krzysztof Kozlowski (krzk)]

The test is killed not because of timeout (LTP_TIMEOUT_MUL does not matter) but out of memory:

[78071.145199] LTP: starting cve-2018-1000204 (ioctl_sg01)
[78171.386374] ioctl_sg01 invoked oom-killer: gfp_mask=0x100dca(GFP_HIGHUSER_MOVABLE|__GFP_ZERO), order=0, oom_score_adj=0
...
[78171.386673] [ 446891] 0 446891 48988960 48987056 392708096 0 0 ioctl_sg01
...
[78171.386702] Out of memory: Killed process 446891 (ioctl_sg01) total-vm:195955840kB, anon-rss:195946880kB, file-rss:1344kB, shmem-rss:0kB, UID:0 pgtables:383504kB oom_score_adj:0

[Krzysztof Kozlowski (krzk)]

Sent patch to LTP:
https://lists.linux.it/pipermail/ltp/2021-June/023446.html

[Krzysztof Kozlowski (krzk)]

Previously discussed in LTP also here: https://patchwork<email address hidden>/

[Krzysztof Kozlowski (krzk)]

Committed upstream.

[Po-Hsu Lin (cypressyew)]

Note that even with krzk's patch applied, this test still took 11 min to run on H-oracle-5.11 BM.Standard2.52

tag=cve-2018-1000204 stime=1633698520 dur=668 exit=exited stat=0 core=no cu=18189 cs=48615

So the commit to bump timeout is still essential:
https://kernel.ubuntu.com/git/ubuntu/autotest-client-tests.git/commit/?id=4c7df183775f8d23493e1a844ed47adf522c2e1f