[LU-5597] Kernel Module.symvers take precendene instead of MOFED provided Created: 08/Sep/14  Updated: 09/Oct/15  Resolved: 29/Jul/15

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.6.0
Fix Version/s: Lustre 2.8.0

Type: Bug Priority: Major
Reporter: Aleksey Senin Assignee: Nathaniel Clark
Resolution: Fixed Votes: 0
Labels: None
Environment:

Ubuntu 14.04
Kernel - 3.13.0-32-generic
MLNX_OFED_LINUX-2.3-1.0.0


Issue Links:
Related
is related to LU-5953 lustre[-dkms] needs to automatically ... Resolved
Severity: 3
Rank (Obsolete): 15642

 Description   

I was able to compile ko2iblnd and other module but all symbol versions were taken from /usr/src/linux-headers-3.13.0-32-generic/Module.symvers instead of /usr/src/mlnx-ofed-kernel-2.3/Module.symvers and as result load of ko2iblnd module fails.



 Comments   
Comment by Jodi Levi (Inactive) [ 12/Sep/14 ]

James,
Can you provide more details on this issue?
Thank you!

Comment by Aleksey Senin [ 12/Sep/14 ]

I opened earlier another discussion on HPDD forum, so all details are there. It is probably better to union these two discussion.

Here is the description of existing case.

After installing Mellanox MOFED, I configured Luster in the following way
./configure --disable-server --with-o2ib=/usr/src/mlnx-ofed-kernel-2.3
and was able to compile kernel modules.
But the version of function symbols that were compiled in ko2iblnd
module belongs to original kernel modules and not to the OFED
modules. As results compiled module refuse to load.

For example ib_destroy_cq function

modprobe --dump-modversions
/lib/modules/3.13.0-32-generic/extra/klnds/o2iblnd/ko2iblnd.ko |grep
ib_destroy_cq
0xad52ad2c ib_destroy_cq

And this is the address from the MOFED Module.symvers file
root at kickseed:/usr/src/lustre-release# grep ib_destroy_cq Module.symvers
0x9a1f091e __crc_ib_destroy_cq
/lib/modules/3.13.0-32-generic/updates/dkms/ib_core (unknown)

It looks like an issue in Lustre.

Setup

Ubuntu-14.04
kernel-3.13.0.32-generic
MOFED-2.3

Comment by Aleksey Senin [ 18/Sep/14 ]

Meanwhile, until the issue is fixed I've created my own fix in GitHub repository
https://github.com/ahlabenadam/lustre_fix.git

Comment by Gerrit Updater [ 08/Dec/14 ]

Vladimir Saveliev (vladimir_saveliev@xyratex.com) uploaded a new patch: http://review.whamcloud.com/12980
Subject: LU-5597 mdd: initialize mdd's obd->obd_vars
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: b0859473660bdcc132e60c1f3a561ddb1772a352

Comment by Vladimir V. Saveliev [ 08/Dec/14 ]

I mistakenly submitted http://review.whamcloud.com/12980 with LU-5597 instead of LU-5997. Please, ignore Gerrit Updater's comment.

Comment by James A Simmons [ 23/Apr/15 ]

Can you try the latest master branch from Intel. The patch from LU-5953 should have addressed this issue. BTW I'm running the same setup as you are Aleksey

Comment by Gerrit Updater [ 04/Jul/15 ]

Nathaniel Clark (nathaniel.l.clark@intel.com) uploaded a new patch: http://review.whamcloud.com/15498
Subject: LU-5597 build: Ensure MOFED Module symvers are used
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 07976ad725b846b05dc132587fed4f0541e469c5

Comment by Gerrit Updater [ 29/Jul/15 ]

Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/15498/
Subject: LU-5597 build: Ensure MOFED Module symvers are used
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 6cba4c1958b8bc25e7c675fd7ec6bf2772c7c0f0

Comment by Peter Jones [ 29/Jul/15 ]

Landed for 2.8

Generated at Sat Feb 10 01:52:51 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.