Message-ID: <08703C70-DD6E-446A-9ABC-BC2C8E33B8CD@nvidia.com>
Date: Fri, 01 Mar 2024 07:52:46 -0500
From: Zi Yan <ziy@...dia.com>
To: Ryan Roberts <ryan.roberts@....com>
Cc: Aishwarya TCV <aishwarya.tcv@....com>,
 "\"Pankaj Raghav (Samsung)\"" <kernel@...kajraghav.com>, linux-mm@...ck.org,
 "\"Matthew Wilcox (Oracle)\"" <willy@...radead.org>,
 David Hildenbrand <david@...hat.com>, Yang Shi <shy828301@...il.com>,
 Yu Zhao <yuzhao@...gle.com>,
 "\"Kirill A . Shutemov\"" <kirill.shutemov@...ux.intel.com>,
 "Michal Koutný" <mkoutny@...e.com>,
 Roman Gushchin <roman.gushchin@...ux.dev>,
 "\"Zach O'Keefe\"" <zokeefe@...gle.com>, Hugh Dickins <hughd@...gle.com>,
 Luis Chamberlain <mcgrof@...nel.org>,
 Andrew Morton <akpm@...ux-foundation.org>, linux-kernel@...r.kernel.org,
 cgroups@...r.kernel.org, linux-fsdevel@...r.kernel.org,
 linux-kselftest@...r.kernel.org, Mark Brown <broonie@...nel.org>
Subject: Re: [PATCH v5 8/8] mm: huge_memory: enable debugfs to split huge
 pages to any order.

On 1 Mar 2024, at 5:33, Ryan Roberts wrote:

> On 01/03/2024 09:51, Aishwarya TCV wrote:
>>
>>
>> On 26/02/2024 20:55, Zi Yan wrote:
>>> From: Zi Yan <ziy@...dia.com>
>>>
>>> It is used to test split_huge_page_to_list_to_order for pagecache THPs.
>>> Also add test cases for split_huge_page_to_list_to_order via both
>>> debugfs.
>>>
>>> Signed-off-by: Zi Yan <ziy@...dia.com>
>>> ---
>>>  mm/huge_memory.c                              |  34 ++++--
>>>  .../selftests/mm/split_huge_page_test.c       | 115 +++++++++++++++++-
>>>  2 files changed, 131 insertions(+), 18 deletions(-)
>>>
>>
>> Hi Zi,
>>
>> When booting the kernel against next-master(20240228)with Arm64 on
>> Marvell Thunder X2 (TX2), the kselftest-mm test 'split_huge_page_test'
>> is failing in our CI (with rootfs over NFS). I can send the full logs if
>> required.
>
> Just to add, I took a quick eyeball and I think there a couple of potential issues:
>
>   - In create_pagecache_thp_and_fd() you do *fd = open(testfile, O_CREAT ...);
>     where testfile is /mnt/thp_fs/testfile. So if /mnt/thp_fs doesn't exist,
>     then the open will fail I think? I'm pretty sure that's what's happening on
>     our CI. Suggest the test needs to setup this dir itself. Is thp_fs a mounted
>     fs or just a dir? If the latter can you just mktemp()?

The former. the page cache folio split tests require a file system supporting
large folio and I used XFS.

>   - Later in create_pagecache_thp_and_fd() you fail the test if you don't have a
>     filesystem that supports large folios. Can we turn that into a skip? That
>     would reduce noise on the CI.

I can do that. But is this a new requirement that self tests have to be finish
in CI/CD environment? Can you provide a guideline for it? Since I always assume
selftests are just ran by human who can set up environment. In addition, I do
not think it is realistic to make the test file to set up all the environment,
since everyone's machine is different. It is much easier to make the CI/CD
environment to make the mount.

>
> Thanks,
> Ryan
>
>>
>> A bisect (full log below) identified this patch as introducing the
>> failure. Bisected it on the tag "next-20240228" at repo
>> "https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git".
>>
>> This works fine on  Linux version 6.8.0-rc6
>>
>>
>> Sample log from failure against run on TX2:
>> ------
>> 07:17:34.056125  # # ------------------------------
>> 07:17:34.056543  # # running ./split_huge_page_test
>> 07:17:34.056839  # # ------------------------------
>> 07:17:34.057114  # # TAP version 13
>> 07:17:34.058564  # # 1..12
>> 07:17:34.156822  # # ok 1 Split huge pages successful
>> 07:17:34.214074  # # ok 2 Split PTE-mapped huge pages successful
>> 07:17:34.215630  # # # Please enable pr_debug in
>> split_huge_pages_in_file() for more info.
>> 07:17:34.225503  # # # Please check dmesg for more information
>> 07:17:34.225862  # # ok 3 File-backed THP split test done
>> 07:17:34.236944  # # Bail out! Failed to create a file at /mnt/thp_fs#
>> Planned tests != run tests (12 != 3)
>> 07:17:34.237307  # # # Totals: pass:3 fail:0 xfail:0 xpass:0 skip:0 error:0
>> 07:17:34.237620  # # [FAIL]
>> 07:17:34.246430  # not ok 51 split_huge_page_test # exit=1
>>
>>
>> Bisect log:
>> ------
>> git bisect start
>> # good: [d206a76d7d2726f3b096037f2079ce0bd3ba329b] Linux 6.8-rc6
>> git bisect good d206a76d7d2726f3b096037f2079ce0bd3ba329b
>> # bad: [20af1ca418d2c0b11bc2a1fe8c0c88f67bcc2a7e] Add linux-next
>> specific files for 20240228
>> git bisect bad 20af1ca418d2c0b11bc2a1fe8c0c88f67bcc2a7e
>> # bad: [1322f1801e59dddce10591d602d246c1bf49990c] Merge branch 'main' of
>> git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git
>> git bisect bad 1322f1801e59dddce10591d602d246c1bf49990c
>> # bad: [a82f70041487790b7b09fe4bb45436e1b57021d3] Merge branch 'dev' of
>> git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4.git
>> git bisect bad a82f70041487790b7b09fe4bb45436e1b57021d3
>> # bad: [ce90480b9352ba2bebe8946dad9223e3f24c6e9a] Merge branch
>> 'for-next' of
>> git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap.git
>> git bisect bad ce90480b9352ba2bebe8946dad9223e3f24c6e9a
>> # bad: [5daac92ed3881fd0c656478a301a4e1d124100ee] Merge branch
>> 'mm-everything' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
>> git bisect bad 5daac92ed3881fd0c656478a301a4e1d124100ee
>> # good: [acc2643d9e988c63dd4629a9af380ad9ac69c54a] Merge branch
>> 'mm-stable' into mm-unstable
>> git bisect good acc2643d9e988c63dd4629a9af380ad9ac69c54a
>> # good: [0294de8fe7d7c1a7eddc979cbf4c1886406e36b7] Merge branch 'fixes'
>> of git://git.kernel.org/pub/scm/linux/kernel/git/sre/linux-power-supply.git
>> git bisect good 0294de8fe7d7c1a7eddc979cbf4c1886406e36b7
>> # good: [83e0c8f0e777a1ef0977b2f8189101765703b32d] Merge branch
>> 'mm-nonmm-stable' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
>> git bisect good 83e0c8f0e777a1ef0977b2f8189101765703b32d
>> # good: [a739cbe236e0dd3b6ff26a01fa1d31c73d4fac93] mm: memcg: make memcg
>> huge page split support any order split
>> git bisect good a739cbe236e0dd3b6ff26a01fa1d31c73d4fac93
>> # bad: [efb520aa333b2f11daaaaa13f4a598b5ae4ae823] mm: allow non-hugetlb
>> large folios to be batch processed
>> git bisect bad efb520aa333b2f11daaaaa13f4a598b5ae4ae823
>> # bad: [2258bdebb55e3ad3d30fd3849ddb955ff36825de] mm/zsmalloc: don't
>> hold locks of all pages when free_zspage()
>> git bisect bad 2258bdebb55e3ad3d30fd3849ddb955ff36825de
>> # bad: [7fc0be45acf2878cbacc4dba56923c34c3fd8b1e] mm: remove
>> total_mapcount()
>> git bisect bad 7fc0be45acf2878cbacc4dba56923c34c3fd8b1e
>> # good: [d55fac55da2f87ad5a99178e107df09770bbc411] mm: thp: split huge
>> page to any lower order pages
>> git bisect good d55fac55da2f87ad5a99178e107df09770bbc411
>> # bad: [4050d591c1aaf9336c08511fa5984827186e9ad1] mm/memfd: refactor
>> memfd_tag_pins() and memfd_wait_for_pins()
>> git bisect bad 4050d591c1aaf9336c08511fa5984827186e9ad1
>> # bad: [c0ba89c29ef559c95273feb481b049f622c43c17] mm: huge_memory:
>> enable debugfs to split huge pages to any order
>> git bisect bad c0ba89c29ef559c95273feb481b049f622c43c17
>> # first bad commit: [c0ba89c29ef559c95273feb481b049f622c43c17] mm:
>> huge_memory: enable debugfs to split huge pages to any order
>>
>>
>> Thanks,
>> Aishwarya


--
Best Regards,
Yan, Zi

Download attachment "signature.asc" of type "application/pgp-signature" (855 bytes)

Powered by blists - more mailing lists