Re: [lkp] [ksm] 40e318e509: ltp.ksm01.fail

[Andrea Arcangeli]

Hello,

On Mon, Jan 25, 2016 at 11:04:18AM +0800, kernel test robot wrote:
> FYI, we noticed the below changes on
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> commit 40e318e509d2c9f3cdb00ef32d2c14b9868af16b ("ksm: introduce ksm_max_page_sharing per page deduplication limit")
> 
> We found LKP KSM test case failed for your patch.
> 
> =========================================================================================
> compiler/kconfig/rootfs/tbox_group/test/testcase:
>   gcc-5/x86_64-nfsroot/debian-x86_64-2015-02-07.cgz/vm-lkp-wsx03-4G/mm_functional_positive/otc_kernel_qa-ts_ltp_ddt
> 
> commit: 
>   77505a33e539477f1fbe5e6136df5b265f39cccf
>   40e318e509d2c9f3cdb00ef32d2c14b9868af16b
> 
> 77505a33e539477f 40e318e509d2c9f3cdb00ef32d 
> ---------------- -------------------------- 
>        fail:runs  %reproduction    fail:runs
>            |             |             |    
>            :4          100%           4:4     otc_kernel_qa-ts_ltp_ddt.ksm01.fail
>            :4          100%           4:4     otc_kernel_qa-ts_ltp_ddt.ksm01_1.fail

In this case it's one rare case of false positive. The change in the
merging levels is a "feature" not a "bug".

LTP validation math on ksm sysfs levels shall be updated to account
for the runtime value of /sys/kernel/mm/ksm/max_page_sharing which in
turn affects pages_sharing/pages_shared. In fact you could add
validation for the new fields like stable_node_dups and
stable_node_chains on top of the existing verifications on pages_shar*.

For now you should re-run the test after:

    echo 2 >/sys/kernel/mm/ksm/run
    echo 10000000 > /sys/kernel/mm/ksm/max_page_sharing
    echo 1 >/sys/kernel/mm/ksm/run

And then LTP will pass as usual.

Thanks for the greatly valuable regression testing work,
Andrea