The problem of generic/465 in xfs filesystem

Hi all,

On RHEL7.5, running generic/465 with xfs should fail because the fix patch[1] is not merged. But on my environment[3], since the patch set[2] designed to fix generic/450 is ported into RHEL7.5, generic/465 turns into pass on my environment.

I discover that various combinations of cpu and memory can lead to different results of generic/465. On environment[4], generic/465 fails.
On environment[5], generic/465 passes.
On other environment[6], generic/465 always fails.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=ee70daaba82d70766d0723b743d9fdeb3b06102a
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=74cedf9b6c603f2278a05bc91b140b32b434d0b5

    https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d4594acbf6d8f75a27f3578476b6a27d8b13ebb 
[3] OS Version: RHEL7.5(including Alpha, Beta, GA)
    CPU Info: Intel(R) Core(TM)2 Duo CPU E7500  @ 2.93GHz
    Number of Processor: 2
    Memory Info: 2048M/4096M
[4] OS Version: RHEL7.4GA
    CPU Info: Intel(R) Core(TM)2 Duo CPU E7500  @ 2.93GHz
    Number of Processor: 2
    Memory Info: 4096M
[5] OS Version: RHEL7.4GA
    CPU Info: Intel(R) Core(TM)2 Duo CPU E7500  @ 2.93GHz
    Number of Processor: 2
    Memory Info: 2048M
[6] OS Version: RHEL7.5(including Alpha, Beta, GA)
    CPU Info: Intel(R) Core(TM) i3-2120 CPU @ 3.30GHz/Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz
    Number of Processor: 2
    Memory Info: 2048M/4096M

It seems that eithor various combinations of cpu and memory or the fix patch set [2] can lead to different results of generic/465, I am not sure which is the real cause of this issue.

The issue was reported by Yang Xu ([email protected])

Thanks,
Xiao Yang