[LTP] [PATCH v2] move_pages12: handle errno EBUSY for madvise(..., MADV_SOFT_OFFLINE)

[Li Wang]

On Fri, Jul 26, 2019 at 9:31 PM Cyril Hrubis <chrubis@suse.cz> wrote:

> > So, maybe we have to re-evaluate this patch V2 and to figure out why
> > the retry mmap() hitting SIGBUS fails.
>
> One possibility would be that the numa_move_pages() triggers SIGBUS
> while we do the usleep() before we attempt to retry the mmap(). In that
> case the race was present in the test all the time but couldn't be
> triggered because the window where the memory is unmapped was very
> short. If that is the case we should as well set up a handler to SIGBUS
> and ignore it as well.

No, It's not like the numa_move_pages() triggers SIGBUS because in the
end child print:
    move_pages12.c:114: FAIL: move_pages failed: ESRCH
that ESRCH means the child is still alive and detect ppid is not available.

It's more like to retry mmap() triggers SIGBUS while doing the
numa_move_pages() in background. That is very similar to the kernel
bug which was mentioned by commit 6bc9b56433b76e40d(mm: fix race on
soft-offlining ). A race condition between soft offline and
hugetlb_fault which causes unexpected process SIGBUS killing.

And, I will send an email to linux-mm@ to RFC.

--
Regards,
Li Wang