[LU-13662] LNet: crash on bad user space parameters

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.14.0
• Severity: 3
• Assignee: Amir Shehata (Inactive)
• Reporter: Amir Shehata (Inactive)

Description

The following YAML causes a crash

 net:
  - net type: tcp
    local NI(s):
      - interfaces:
          0: eth0
        tunables:
          peer_credits: 32
      - interfaces:
          0: eth1
        tunables:
          peer_credits: 32
      - interfaces:
          0: eth2
        tunables:
          peer_credits: 32

peer_credits and peer_tx_credits default to 0, which causes an NI to be removed prematurely.

Missing parameters should be defaulted.

Comments

[Peter Jones]

Landed for 2.14

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/38894/
Subject: LU-13662 lnet: handle undefined parameters
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: d934eb3c4f63858245132b55363002cda722c50b

[Gerrit Updater]

Amir Shehata ([e-mail]) uploaded a new patch: https://review.whamcloud.com/38894
Subject: LU-13662 lnet: handle undefined parameters
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 1e9ffc30d74c254e206a32ac8d31fd1e01cfa0de

[Serguei Smirnov]

The following yaml also causes a problem:

 net:
    - net type: tcp
      local NI(s):
        - nid: 10.122.36.80@tcp
          status: up
          interfaces:
          statistics:
              send_count: 0
              recv_count: 0
              drop_count: 0
          tunables:
              peer_timeout: 180
              peer_credits: 8
              peer_buffer_credits: 0
              credits: 256
          lnd tunables:
          tcp bonding: 0
          dev cpt: -1
numa:
    range: 0

# lnetctl import lnetconf2.txt 
error:
	fatal: out of memory
??S??:
Segmentation fault 

This was reproduced on 7.8 Centos, b2_12 patchless client system (https://build.whamcloud.com/job/lustre-b2_12-patchless/83/)