[LU-18751] Segfault in lnetctl fault command

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.17.0
• Severity: 3
• Assignee: Frank Sehr
• Reporter: Frank Sehr

Description

Lnet ctl cors with the command "lnetctl fault reset \0"

Comments

[Peter Jones]

Tim's patch to disable it is in master-next which has been running for several days so this situation should improve imminently...

[Andreas Dilger]

Failed 51x in the past week.

[Andreas Dilger]

This failed 42x on master this week.

[Gerrit Updater]

"Timothy Day <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/58897
Subject: LU-18751 lnet: disable sanity-lnet/410 test case
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 7e6510604635869952a9b75e76ac1b29c303fa8e

[Jian Yu]

Seems simply failing. It failed 58 times in the last week.

[Frank Sehr]

Is there a core file or is the test simply failing?

Thanks

Frank

[Jian Yu]

sanity-lnet test 410 still failed regularly on master branch:

== sanity-lnet test 410: No segfault in lnetctl fault command ========================================================== 14:37:53 (1745246273)
/usr/sbin/lnetctl lnet unconfigure
/usr/sbin/lnetctl lnet configure
/usr/sbin/lnetctl fault reset
 sanity-lnet test_410: @@@@@@ FAIL: Command should fail

https://testing.whamcloud.com/test_sets/e9492945-ca81-46dc-8c74-53dc1360fd51

[Gerrit Updater]

"Oleg Drokin <[e-mail]>" merged in patch https://review.whamcloud.com/c/fs/lustre-release/+/58623/
Subject: LU-18751 lnet: test lnetctl fault reset directly
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 66b196a022349f05f365744ecec7fafe8833fe2d

[Gerrit Updater]

"Shaun Tancheff <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/58658
Subject: LU-18751 lnet: test lnetctl fault reset directly
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: d879825d0e0a1f7091a33ee6fbad4843d6b8f539

[Gerrit Updater]

"Shaun Tancheff <[e-mail]>" uploaded a new patch: https://review.whamcloud.com/c/fs/lustre-release/+/58623
Subject: LU-18751 lnet: fix sanity-lnet/410 test case
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 0afede9ba0b160d5a3a4b493d410c2195d42ac8d

[Andreas Dilger]

This generated a new Coverity warning in the latest run:

CID 458333: Null pointer dereferences (REVERSE_INULL)
/lnet/utils/lnetconfig/liblnetconfig.c: 955 in infra_ping_nid():

949                 id.pid = LNET_PID_ANY;
950                 /* if no net is specified, libcfs_str2nid() will assume tcp */
951                 id.nid = libcfs_str2nid(token);
952                 if (id.nid == LNET_NID_ANY) {
953                     snprintf(err_str, sizeof(err_str),
954                          "\"cannot parse NID '%s'\"",

CID 458333: Null pointer dereferences (REVERSE_INULL)
Null-checking "token" suggests that it may be null, but it has already been dereferenced on all paths leading to the check.

955                          token ? token : "NULL");
956                     rc = LUSTRE_CFG_RC_BAD_PARAM;
957                     cYAML_build_error(rc, seq_no, MANAGE_CMD,
958                               oper, err_str, err_rc);
959                     continue;
960                 }