[LU-7766] LNetEQAlloc() misuses roundup_pow_of_two()

• Type: Bug
• Resolution: Fixed
• Priority: Blocker
• Fix Version/s: Lustre 2.8.0
• Severity: 3
• Assignee: Oleg Drokin
• Reporter: John Hammond

Description

After http://review.whamcloud.com/16914 we see the following when starting LNet:

[690112.179415] LNet: 8882:0:(lib-eq.c:87:LNetEQAlloc()) EQ callback is guaranteed to get every event, do you still want to set eqcount 1 for polling event which will have locking overhead? Please contact with developer to confirm

Here LNetEQAlloc() is being called with a count of 0 and passing that to roundup_pow_of_two(). This is undefined behavior according to http://lxr.free-electrons.com/source/tools/include/linux/log2.h#L161.

Comments

[Peter Jones]

Landed for 2.8

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch http://review.whamcloud.com/18370/
Subject: LU-7766 lnet: Don't call roundup_pow_of_two on zero in LNetEQAlloc
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: e3705effd8d94be11493da754c5678de5523c877

[Bob Glossman (Inactive)]

just noticed the annoying new lnet warning message today. talked about it on Skype. good to know there's already a patch in flight to fix it.

[Oleg Drokin]

I submitted the same patch upstream a few days ago, but did not notice we imported this bug into our tree too.

[Gerrit Updater]

Oleg Drokin ([e-mail]) uploaded a new patch: http://review.whamcloud.com/18370
Subject: LU-7766 lnet: Don't call roundup_pow_of_two on zero in LNetEQAlloc
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: c0cd8205473d7198ca48278449dcdfb09110f2de