Date: Thu, 20 Jun 2024 18:09:45 +1200
From: Barry Song <21cnbao@...il.com>
To: "Huang, Ying" <ying.huang@...el.com>
Cc: akpm@...ux-foundation.org, shuah@...nel.org, linux-mm@...ck.org, 
	ryan.roberts@....com, chrisl@...nel.org, david@...hat.com, hughd@...gle.com, 
	kaleshsingh@...gle.com, kasong@...cent.com, linux-kernel@...r.kernel.org, 
	linux-kselftest@...r.kernel.org, Barry Song <v-songbaohua@...o.com>
Subject: Re: [PATCH] selftests/mm: Introduce a test program to assess swap
 entry allocation for thp_swapout

On Thu, Jun 20, 2024 at 5:22 PM Huang, Ying <ying.huang@...el.com> wrote:
>
> Barry Song <21cnbao@...il.com> writes:
>
> > On Thu, Jun 20, 2024 at 1:55 PM Huang, Ying <ying.huang@...el.com> wrote:
> >>
> >> Barry Song <21cnbao@...il.com> writes:
> >>
> >> > From: Barry Song <v-songbaohua@...o.com>
> >> >
> >> > Both Ryan and Chris have been utilizing the small test program to aid
> >> > in debugging and identifying issues with swap entry allocation. While
> >> > a real or intricate workload might be more suitable for assessing the
> >> > correctness and effectiveness of the swap allocation policy, a small
> >> > test program presents a simpler means of understanding the problem and
> >> > initially verifying the improvements being made.
> >> >
> >> > Let's endeavor to integrate it into the self-test suite. Although it
> >> > presently only accommodates 64KB and 4KB, I'm optimistic that we can
> >> > expand its capabilities to support multiple sizes and simulate more
> >> > complex systems in the future as required.
> >>
> >> IIUC, this is a performance test program instead of functionality test
> >> program.  Does it match the purpose of the kernel selftest?
> >
> > I have a differing perspective. I maintain that the functionality is
> > not functioning
> > as expected. Despite having all the necessary resources for allocation, failure
> > persists, indicating a lack of functionality.
>
> Is there any user visual functionality issue?

Definitely not. If a plane can't take off, taking a train and pretending
there's no functionality issue isn't a solution.

I have never assigned blame for any mistakes here. On the contrary,
I have 100% appreciation for Ryan's work in at least initiating mTHP
swapout w/o being split.

It took countless experiments for humans to make airplanes commercially
viable, but the person who created the first flying airplane remains the
greatest. Similarly, Ryan's efforts, combined with your review of his patch,
have enabled us to achieve a better goal here. Without your work, we can't
get here at all.

However, this is never a reason to refuse to acknowledge that this feature
is not actually working.

>
> >>
> >> > Signed-off-by: Barry Song <v-songbaohua@...o.com>
> >> > ---
> >> >  tools/testing/selftests/mm/Makefile           |   1 +
> >> >  .../selftests/mm/thp_swap_allocator_test.c    | 192 ++++++++++++++++++
> >> >  2 files changed, 193 insertions(+)
> >> >  create mode 100644 tools/testing/selftests/mm/thp_swap_allocator_test.c
> >> >
> >> > diff --git a/tools/testing/selftests/mm/Makefile b/tools/testing/selftests/mm/Makefile
> >> > index e1aa09ddaa3d..64164ad66835 100644
> >> > --- a/tools/testing/selftests/mm/Makefile
> >> > +++ b/tools/testing/selftests/mm/Makefile
> >> > @@ -65,6 +65,7 @@ TEST_GEN_FILES += mseal_test
> >> >  TEST_GEN_FILES += seal_elf
> >> >  TEST_GEN_FILES += on-fault-limit
> >> >  TEST_GEN_FILES += pagemap_ioctl
> >> > +TEST_GEN_FILES += thp_swap_allocator_test
> >> >  TEST_GEN_FILES += thuge-gen
> >> >  TEST_GEN_FILES += transhuge-stress
> >> >  TEST_GEN_FILES += uffd-stress
> >> > diff --git a/tools/testing/selftests/mm/thp_swap_allocator_test.c b/tools/testing/selftests/mm/thp_swap_allocator_test.c
> >> > new file mode 100644
> >> > index 000000000000..4443a906d0f8
> >> > --- /dev/null
> >> > +++ b/tools/testing/selftests/mm/thp_swap_allocator_test.c
> >> > @@ -0,0 +1,192 @@
> >> > +// SPDX-License-Identifier: GPL-2.0-or-later
> >> > +/*
> >> > + * thp_swap_allocator_test
> >> > + *
> >> > + * The purpose of this test program is helping check if THP swpout
> >> > + * can correctly get swap slots to swap out as a whole instead of
> >> > + * being split. It randomly releases swap entries through madvise
> >> > + * DONTNEED and do swapout on two memory areas: a memory area for
> >> > + * 64KB THP and the other area for small folios. The second memory
> >> > + * can be enabled by "-s".
> >> > + * Before running the program, we need to setup a zRAM or similar
> >> > + * swap device by:
> >> > + *  echo lzo > /sys/block/zram0/comp_algorithm
> >> > + *  echo 64M > /sys/block/zram0/disksize
> >> > + *  echo never > /sys/kernel/mm/transparent_hugepage/hugepages-2048kB/enabled
> >> > + *  echo always > /sys/kernel/mm/transparent_hugepage/hugepages-64kB/enabled
> >> > + *  mkswap /dev/zram0
> >> > + *  swapon /dev/zram0
> >> > + * The expected result should be 0% anon swpout fallback ratio w/ or
> >> > + * w/o "-s".
> >> > + *
> >> > + * Author(s): Barry Song <v-songbaohua@...o.com>
> >> > + */
> >> > +
> >> > +#define _GNU_SOURCE
> >> > +#include <stdio.h>
> >> > +#include <stdlib.h>
> >> > +#include <unistd.h>
> >> > +#include <string.h>
> >> > +#include <sys/mman.h>
> >> > +#include <errno.h>
> >> > +#include <time.h>
> >> > +
> >> > +#define MEMSIZE_MTHP (60 * 1024 * 1024)
> >> > +#define MEMSIZE_SMALLFOLIO (1 * 1024 * 1024)
> >> > +#define ALIGNMENT_MTHP (64 * 1024)
> >> > +#define ALIGNMENT_SMALLFOLIO (4 * 1024)
> >> > +#define TOTAL_DONTNEED_MTHP (16 * 1024 * 1024)
> >> > +#define TOTAL_DONTNEED_SMALLFOLIO (768 * 1024)
> >> > +#define MTHP_FOLIO_SIZE (64 * 1024)
> >> > +
> >> > +#define SWPOUT_PATH \
> >> > +     "/sys/kernel/mm/transparent_hugepage/hugepages-64kB/stats/swpout"
> >> > +#define SWPOUT_FALLBACK_PATH \
> >> > +     "/sys/kernel/mm/transparent_hugepage/hugepages-64kB/stats/swpout_fallback"
> >> > +
> >> > +static void *aligned_alloc_mem(size_t size, size_t alignment)
> >> > +{
> >> > +     void *mem = NULL;
> >> > +
> >> > +     if (posix_memalign(&mem, alignment, size) != 0) {
> >> > +             perror("posix_memalign");
> >> > +             return NULL;
> >> > +     }
> >> > +     return mem;
> >> > +}
> >> > +
> >> > +static void random_madvise_dontneed(void *mem, size_t mem_size,
> >> > +             size_t align_size, size_t total_dontneed_size)
> >> > +{
> >> > +     size_t num_pages = total_dontneed_size / align_size;
> >> > +     size_t i;
> >> > +     size_t offset;
> >> > +     void *addr;
> >> > +
> >> > +     for (i = 0; i < num_pages; ++i) {
> >> > +             offset = (rand() % (mem_size / align_size)) * align_size;
> >> > +             addr = (char *)mem + offset;
> >> > +             if (madvise(addr, align_size, MADV_DONTNEED) != 0)
> >> > +                     perror("madvise dontneed");
> >>
> >> IIUC, this simulates align_size (generally 64KB) swap-in.  That is, it
> >> simulate the effect of large size swap-in when it's not available in
> >> kernel.  If we have large size swap-in in kernel in the future, this
> >> becomes unnecessary.
> >>
> >> Additionally, we have not reached the consensus that we should always
> >> swap-in with swapped-out size.  So, I suspect that this test may not
> >> reflect real situation in the future.  Although it doesn't reflect
> >> current situation too.
> >
> > Disagree again. releasing the whole mTHP swaps is the best case. Even in
> > the best-case scenario, if we fail, it raises concerns for handling potentially
> > more challenging situations.
>
> Repeating sequential anonymous pages writing is the best case.

I define the best case as the scenario with the least chance of creating
fragments within swapfiles for mTHP to swap out. There is no real
difference whether this is done through swapin or madv_dontneed.

>
> > I don't find it hard to incorporate additional features into this test
> > program to simulate more intricate scenarios.
>
> IMHO, we don't really need this special purpose test.  We can have some
> more general basic tests, for example, sequential anonymous pages
> writing/reading, random anonymous pages writing/reading, and combination
> of them.

I understand that not all things will be loved by all people. However, before
I sent this patch, Chris mentioned that it has been very helpful for him and
strongly suggested that I contribute it to the self-test suite.

By the way, adding sequential and random anonymous pages for
read/write operations is definitely in my plan. The absence of this feature
isn't a convincing reason to disregard it.

>
> --
> Best Regards,
> Huang, Ying
>
> >>
> >> > +
> >> > +             memset(addr, 0x11, align_size);
> >> > +     }
> >> > +}
> >> > +
> >>
> >> [snip]
> >>
> >> --
> >> Best Regards,
> >> Huang, Ying

Thanks
Barry

Powered by blists - more mailing lists