[LU-11838] Support linux kernel version 4.18

• Type: Improvement
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.13.0, Lustre 2.12.3
• Affects Version/s: Lustre 2.13.0
• Environment: Ubuntu18.10 and the beta version of the most beloved distro :-)
• Assignee: James A Simmons
• Reporter: James A Simmons

Description

Linux distributions are looking to move their platforms to the 4.18 kernel which was originally the LTS for the linux kernel (that has changed to 4.19). This ticket covers the work needed to support this kernel version.

Issue Links

• is related to: LU-11803 sanity test 255c fails with 'Ladvise test 12, bad lock count, returned 1, actual 0'
• is related to: LU-12269 Support RHEL 8.0
• is related to: LU-9679 Prepare lustre for adoption into the linux kernel
• is related to: LU-12270 Support linux kernel version 4.19
• is related to: LU-12331 hardened usercopy whitelisting
• is related to: LU-11893 doesn't handle logical network interface properly.
• is related to: LU-6399 Libcfs socket cleanup
• is related to: LU-11233 gcc8 support

Comments

[Gerrit Updater]

Li Dongyang ([e-mail]) uploaded a new patch: https://review.whamcloud.com/34672
Subject: LU-11838 lnet: getname dropping addrlen argument
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: de014af11e436ae3940464320afdb1a5ddd1548a

[James A Simmons]

Chris I have a patch to fix the ip2net handling as well. I need the above patches to land first. The idea is to replace all the LND ipv4 handling by using lnet_ipaddr_enumerate() in config.c

[Chris Horn]

James, I hit the same issue with those three patches applied

[James A Simmons]

@Dongyang Li    - The patch for LU-11803 handles the s_uuid change in newer kernels. I'm testing it now. See:

https://review.whamcloud.com/#/c/33916/

[James A Simmons]

Chris their are two patches to address the bug. So for proper 4.18 kernel support you need 3 patches for LNet.

https://review.whamcloud.com/#/c/34476/

https://review.whamcloud.com/#/c/34392/

https://review.whamcloud.com/#/c/34234

I can add you as  a reviewer. 

[Chris Horn]

The patch https://review.whamcloud.com/33969 seems to have broken ip2nets parsing. I'm not sure if this is the same issue documented in LU-11893.

sles15c01:~ # ip addr show dev eth0
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:0c:29:88:48:01 brd ff:ff:ff:ff:ff:ff
    inet 192.168.2.22/16 brd 192.168.255.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::20c:29ff:fe88:4801/64 scope link
       valid_lft forever preferred_lft forever
sles15c01:~ # cat /etc/modprobe.d/lnet.conf
options lnet ip2nets="tcp99(eth0) 192.168.*.*; tcp4(eth0) 192.168.*.*"
options lnet routes="tcp 192.168.2.26@tcp4"
options lnet lnet_peer_discovery_disabled=0
#options lnet check_routers_before_use=1
sles15c01:~ # modprobe lnet
sles15c01:~ # lnetctl lnet configure --all
configure:
    - lnet:
          errno: -22
          descr: "LNet configure error: Invalid argument"
sles15c01:~ # dmesg | tail --lines 2
[14219.235056] LNetError: 11a-a: ip2nets does not match any local IP interfaces
[14219.237168] LNetError: 3884:0:(config.c:597:lnet_parse_networks()) networks string is undefined
sles15c01:~ #

If I revert https://review.whamcloud.com/33969 then I'm able to start LNet with the above configuration.

[James A Simmons]

No I haven't started ldiskfs support. Only clibet so far. Please push the work.

[Dongyang Li]

James, are you working on the ldiskfs support for 4.18?

If not I have a ldiskfs patchset for centos8 beta, will send it under this ticket later.

Cheers

DY

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/34236/
Subject: LU-11838 llite: remove assert for acl refcount
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: df7bfbb1c7890deed15fd85e75da70d88be2ef7f

[Gerrit Updater]

James Simmons ([e-mail]) uploaded a new patch: https://review.whamcloud.com/34236
Subject: LU-11838 llite: remove assert for acl refcount
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: a4e4ed5cddef37a12169f6fb638c2814985b0c72

[James A Simmons]

The changes need to be address are:

LNet: kernel_getpeername() and kernel_getsockname() dropped last argument.

The struct address_mapping changed from radix to xarry. All llite changes.

Initialize "struct dentry_operations" .d_init using ll_d_init(). For proper parallel lookup support.

Remove LASSERT(atomic_read(&lli->lli_posix_acl->a_refcount) == 1);

That will do it.