platform/kernel/linux-rpi.git
14 months agomm/sparse: remove redundant judgments from macro for_each_present_section_nr
liuq [Fri, 7 Jul 2023 06:05:01 +0000 (14:05 +0800)]
mm/sparse: remove redundant judgments from macro for_each_present_section_nr

next_present_section_nr() has already ensured that
'section_nr<=__highest_present_section_nr', so this check is removed.

Link: https://lkml.kernel.org/r/20230707060501.29184-1-liuq131@chinatelecom.cn
Signed-off-by: liuq <liuq131@chinatelecom.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: compaction: skip the memory hole rapidly when isolating free pages
Baolin Wang [Fri, 7 Jul 2023 08:51:47 +0000 (16:51 +0800)]
mm: compaction: skip the memory hole rapidly when isolating free pages

Just like commit 9721fd82351d ("mm: compaction: skip memory hole
rapidly when isolating migratable pages"), I can see it will also take
more time to skip the larger memory hole (range: 0x1000000000 -
0x1800000000) when isolating free pages on my machine with below memory
layout.  So like commit 9721fd82351d, adding a new helper to skip the
memory hole rapidly, which can reduce the time consumed from about 70us
to less than 1us.

[    0.000000] Zone ranges:
[    0.000000]   DMA      [mem 0x0000000040000000-0x00000000ffffffff]
[    0.000000]   DMA32    empty
[    0.000000]   Normal   [mem 0x0000000100000000-0x0000001fa7ffffff]
[    0.000000] Movable zone start for each node
[    0.000000] Early memory node ranges
[    0.000000]   node   0: [mem 0x0000000040000000-0x0000000fffffffff]
[    0.000000]   node   0: [mem 0x0000001800000000-0x0000001fa3c7ffff]
[    0.000000]   node   0: [mem 0x0000001fa3c80000-0x0000001fa3ffffff]
[    0.000000]   node   0: [mem 0x0000001fa4000000-0x0000001fa402ffff]
[    0.000000]   node   0: [mem 0x0000001fa4030000-0x0000001fa40effff]
[    0.000000]   node   0: [mem 0x0000001fa40f0000-0x0000001fa73cffff]
[    0.000000]   node   0: [mem 0x0000001fa73d0000-0x0000001fa745ffff]
[    0.000000]   node   0: [mem 0x0000001fa7460000-0x0000001fa746ffff]
[    0.000000]   node   0: [mem 0x0000001fa7470000-0x0000001fa758ffff]
[    0.000000]   node   0: [mem 0x0000001fa7590000-0x0000001fa7ffffff]

[shikemeng@huaweicloud.com: avoid missing last page block in section after skip offline sections]
Link: https://lkml.kernel.org/r/20230804110454.2935878-1-shikemeng@huaweicloud.com
Link: https://lkml.kernel.org/r/20230804110454.2935878-2-shikemeng@huaweicloud.com
Link: https://lkml.kernel.org/r/d2ba7e41ee566309b594311207ffca736375fc16.1688715750.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: compaction: use the correct type of list for free pages
Baolin Wang [Fri, 7 Jul 2023 08:51:46 +0000 (16:51 +0800)]
mm: compaction: use the correct type of list for free pages

Use the page->buddy_list instead of page->lru to clarify the correct type
of list for free pages.

Link: https://lkml.kernel.org/r/b21cd8e2e32b9a1d9bc9e43ebf8acaf35e87f8df.1688715750.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Huang, Ying <ying.huang@intel.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: fix some kernel-doc comments
Yang Li [Fri, 7 Jul 2023 09:00:34 +0000 (17:00 +0800)]
mm: fix some kernel-doc comments

Add description of @mm_wr_locked and @mm.
to silence the warnings:

mm/memory.c:1716: warning: Function parameter or member 'mm_wr_locked' not described in 'unmap_vmas'
mm/memory.c:5110: warning: Function parameter or member 'mm' not described in 'mm_account_fault'

Link: https://lkml.kernel.org/r/20230707090034.125511-1-yang.lee@linux.alibaba.com
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: correct stale comment of function check_pte
Kemeng Shi [Fri, 7 Jul 2023 15:39:53 +0000 (23:39 +0800)]
mm: correct stale comment of function check_pte

Commit 2aff7a4755bed ("mm: Convert page_vma_mapped_walk to work on PFNs")
replaced page with pfns in page_vma_mapped_walk structure and updated
"@pvmw->page" to "@pvmw->pfn" in comment of function page_vma_mapped_walk.

This patch update stale "page" to "pfn" in comment of check_pte.

Link: https://lkml.kernel.org/r/20230707153953.1380615-1-shikemeng@huaweicloud.com
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm, netfs, fscache: stop read optimisation when folio removed from pagecache
David Howells [Wed, 28 Jun 2023 10:48:52 +0000 (11:48 +0100)]
mm, netfs, fscache: stop read optimisation when folio removed from pagecache

Fscache has an optimisation by which reads from the cache are skipped
until we know that (a) there's data there to be read and (b) that data
isn't entirely covered by pages resident in the netfs pagecache.  This is
done with two flags manipulated by fscache_note_page_release():

if (...
    test_bit(FSCACHE_COOKIE_HAVE_DATA, &cookie->flags) &&
    test_bit(FSCACHE_COOKIE_NO_DATA_TO_READ, &cookie->flags))
clear_bit(FSCACHE_COOKIE_NO_DATA_TO_READ, &cookie->flags);

where the NO_DATA_TO_READ flag causes cachefiles_prepare_read() to
indicate that netfslib should download from the server or clear the page
instead.

The fscache_note_page_release() function is intended to be called from
->releasepage() - but that only gets called if PG_private or PG_private_2
is set - and currently the former is at the discretion of the network
filesystem and the latter is only set whilst a page is being written to
the cache, so sometimes we miss clearing the optimisation.

Fix this by following Willy's suggestion[1] and adding an address_space
flag, AS_RELEASE_ALWAYS, that causes filemap_release_folio() to always call
->release_folio() if it's set, even if PG_private or PG_private_2 aren't
set.

Note that this would require folio_test_private() and page_has_private() to
become more complicated.  To avoid that, in the places[*] where these are
used to conditionalise calls to filemap_release_folio() and
try_to_release_page(), the tests are removed the those functions just
jumped to unconditionally and the test is performed there.

[*] There are some exceptions in vmscan.c where the check guards more than
just a call to the releaser.  I've added a function, folio_needs_release()
to wrap all the checks for that.

AS_RELEASE_ALWAYS should be set if a non-NULL cookie is obtained from
fscache and cleared in ->evict_inode() before truncate_inode_pages_final()
is called.

Additionally, the FSCACHE_COOKIE_NO_DATA_TO_READ flag needs to be cleared
and the optimisation cancelled if a cachefiles object already contains data
when we open it.

[dwysocha@redhat.com: call folio_mapping() inside folio_needs_release()]
Link: https://github.com/DaveWysochanskiRH/kernel/commit/902c990e311120179fa5de99d68364b2947b79ec
Link: https://lkml.kernel.org/r/20230628104852.3391651-3-dhowells@redhat.com
Fixes: 1f67e6d0b188 ("fscache: Provide a function to note the release of a page")
Fixes: 047487c947e8 ("cachefiles: Implement the I/O routines")
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Dave Wysochanski <dwysocha@redhat.com>
Reported-by: Rohith Surabattula <rohiths.msft@gmail.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Tested-by: SeongJae Park <sj@kernel.org>
Cc: Daire Byrne <daire.byrne@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Steve French <sfrench@samba.org>
Cc: Shyam Prasad N <nspmangalore@gmail.com>
Cc: Rohith Surabattula <rohiths.msft@gmail.com>
Cc: Dave Wysochanski <dwysocha@redhat.com>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Ilya Dryomov <idryomov@gmail.com>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Jingbo Xu <jefflexu@linux.alibaba.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Xiubo Li <xiubli@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: merge folio_has_private()/filemap_release_folio() call pairs
David Howells [Wed, 28 Jun 2023 10:48:51 +0000 (11:48 +0100)]
mm: merge folio_has_private()/filemap_release_folio() call pairs

Patch series "mm, netfs, fscache: Stop read optimisation when folio
removed from pagecache", v7.

This fixes an optimisation in fscache whereby we don't read from the cache
for a particular file until we know that there's data there that we don't
have in the pagecache.  The problem is that I'm no longer using PG_fscache
(aka PG_private_2) to indicate that the page is cached and so I don't get
a notification when a cached page is dropped from the pagecache.

The first patch merges some folio_has_private() and
filemap_release_folio() pairs and introduces a helper,
folio_needs_release(), to indicate if a release is required.

The second patch is the actual fix.  Following Willy's suggestions[1], it
adds an AS_RELEASE_ALWAYS flag to an address_space that will make
filemap_release_folio() always call ->release_folio(), even if
PG_private/PG_private_2 aren't set.  folio_needs_release() is altered to
add a check for this.

This patch (of 2):

Make filemap_release_folio() check folio_has_private().  Then, in most
cases, where a call to folio_has_private() is immediately followed by a
call to filemap_release_folio(), we can get rid of the test in the pair.

There are a couple of sites in mm/vscan.c that this can't so easily be
done.  In shrink_folio_list(), there are actually three cases (something
different is done for incompletely invalidated buffers), but
filemap_release_folio() elides two of them.

In shrink_active_list(), we don't have have the folio lock yet, so the
check allows us to avoid locking the page unnecessarily.

A wrapper function to check if a folio needs release is provided for those
places that still need to do it in the mm/ directory.  This will acquire
additional parts to the condition in a future patch.

After this, the only remaining caller of folio_has_private() outside of
mm/ is a check in fuse.

Link: https://lkml.kernel.org/r/20230628104852.3391651-1-dhowells@redhat.com
Link: https://lkml.kernel.org/r/20230628104852.3391651-2-dhowells@redhat.com
Reported-by: Rohith Surabattula <rohiths.msft@gmail.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Steve French <sfrench@samba.org>
Cc: Shyam Prasad N <nspmangalore@gmail.com>
Cc: Rohith Surabattula <rohiths.msft@gmail.com>
Cc: Dave Wysochanski <dwysocha@redhat.com>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Ilya Dryomov <idryomov@gmail.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Xiubo Li <xiubli@redhat.com>
Cc: Jingbo Xu <jefflexu@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agormap: pass the folio to __page_check_anon_rmap()
Matthew Wilcox (Oracle) [Thu, 6 Jul 2023 19:52:51 +0000 (20:52 +0100)]
rmap: pass the folio to __page_check_anon_rmap()

The lone caller already has the folio, so pass it in instead of deriving
it from the page again.

Link: https://lkml.kernel.org/r/20230706195251.2707542-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: cma: print cma name as well in cma_alloc debug
Pintu Kumar [Thu, 6 Jul 2023 18:33:34 +0000 (00:03 +0530)]
mm: cma: print cma name as well in cma_alloc debug

CMA allocation can happen either from global cma or from dedicated cma
region.

Thus it is helpful to print cma name as well during initial
debugging to confirm cma regions were getting initialized or not.

Link: https://lkml.kernel.org/r/1688668414-12350-1-git-send-email-quic_pintu@quicinc.com
Signed-off-by: Pintu Kumar <quic_pintu@quicinc.com>
Signed-off-by: Pintu Agarwal <pintu.ping@gmail.com>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: remove obsolete comment above struct per_cpu_pages
Miaohe Lin [Thu, 6 Jul 2023 09:24:41 +0000 (17:24 +0800)]
mm: remove obsolete comment above struct per_cpu_pages

Since commit 01b44456a7aa ("mm/page_alloc: replace local_lock with normal
spinlock"), per_cpu_pages is protected by normal spinlock.  Remove the
obsolete comment as it's not that helpful.

Link: https://lkml.kernel.org/r/20230706092441.1574950-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomemory tier: rename destroy_memory_type() to put_memory_type()
Miaohe Lin [Thu, 6 Jul 2023 06:39:05 +0000 (14:39 +0800)]
memory tier: rename destroy_memory_type() to put_memory_type()

It appears that destroy_memory_type() isn't a very good name because we
usually will not free the memory_type here.  So rename it to a more
appropriate name i.e.  put_memory_type().

Link: https://lkml.kernel.org/r/20230706063905.543800-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Suggested-by: Huang, Ying <ying.huang@intel.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Xiao Yang <yangx.jy@fujitsu.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftests/memfd: sysctl: fix MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED
Jeff Xu [Wed, 5 Jul 2023 06:33:15 +0000 (06:33 +0000)]
selftests/memfd: sysctl: fix MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED

Add selftest for sysctl vm.memfd_noexec is 2
(MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED)

memfd_create(.., MFD_EXEC) should fail in this case.

Link: https://lkml.kernel.org/r/20230705063315.3680666-3-jeffxu@google.com
Reported-by: Dominique Martinet <asmadeus@codewreck.org>
Closes: https://lore.kernel.org/linux-mm/CABi2SkXUX_QqTQ10Yx9bBUGpN1wByOi_=gZU6WEy5a8MaQY3Jw@mail.gmail.com/T/
Signed-off-by: Jeff Xu <jeffxu@google.com>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Jorge Lucangeli Obes <jorgelo@chromium.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: kernel test robot <lkp@intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/memfd: sysctl: fix MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED
Jeff Xu [Wed, 5 Jul 2023 06:33:14 +0000 (06:33 +0000)]
mm/memfd: sysctl: fix MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED

Patch series "mm/memfd: fix sysctl MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED", v2.

When sysctl vm.memfd_noexec is 2 (MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED),
memfd_create(.., MFD_EXEC) should fail.

This complies with how MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED is defined -
"memfd_create() without MFD_NOEXEC_SEAL will be rejected"

Thanks to Dominique Martinet <asmadeus@codewreck.org> who reported the bug.
see [1] for context.

[1] https://lore.kernel.org/linux-mm/CABi2SkXUX_QqTQ10Yx9bBUGpN1wByOi_=gZU6WEy5a8MaQY3Jw@mail.gmail.com/T/

This patch (of 2):

When vm.memfd_noexec is 2 (MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED),
memfd_create(.., MFD_EXEC) should fail.

This complies with how MEMFD_NOEXEC_SCOPE_NOEXEC_ENFORCED is
defined - "memfd_create() without MFD_NOEXEC_SEAL will be rejected"

Link: https://lkml.kernel.org/r/20230705063315.3680666-1-jeffxu@google.com
Link: https://lkml.kernel.org/r/20230705063315.3680666-2-jeffxu@google.com
Fixes: 105ff5339f49 ("mm/memfd: add MFD_NOEXEC_SEAL and MFD_EXEC")
Reported-by: Dominique Martinet <asmadeus@codewreck.org>
Closes: https://lore.kernel.org/linux-mm/CABi2SkXUX_QqTQ10Yx9bBUGpN1wByOi_=gZU6WEy5a8MaQY3Jw@mail.gmail.com/T/
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202306301351.kkbSegQW-lkp@intel.com/
Signed-off-by: Jeff Xu <jeffxu@google.com>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Jorge Lucangeli Obes <jorgelo@chromium.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agofs: drop_caches: draining pages before dropping caches
Andrew Yang [Fri, 30 Jun 2023 09:22:02 +0000 (17:22 +0800)]
fs: drop_caches: draining pages before dropping caches

We expect a file page access after dropping caches should be a major
fault, but sometimes it's still a minor fault.  That's because a file page
can't be dropped if it's in a per-cpu pagevec.  Draining all pages from
per-cpu pagevec to lru list before trying to drop caches.

Link: https://lkml.kernel.org/r/20230630092203.16080-1-andrew.yang@mediatek.com
Signed-off-by: Andrew Yang <andrew.yang@mediatek.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomemcg: drop kmem.limit_in_bytes
Michal Hocko [Tue, 4 Jul 2023 11:52:40 +0000 (13:52 +0200)]
memcg: drop kmem.limit_in_bytes

kmem.limit_in_bytes (v1 way to limit kernel memory usage) has been
deprecated since 58056f77502f ("memcg, kmem: further deprecate
kmem.limit_in_bytes") merged in 5.16.  We haven't heard about any serious
users since then but it seems that the mere presence of the file is
causing more harm thatn good.  We (SUSE) have had several bug reports from
customers where Docker based containers started to fail because a write to
kmem.limit_in_bytes has failed.

This was unexpected because runc code only expects ENOENT (kmem disabled)
or EBUSY (tasks already running within cgroup).  So a new error code was
unexpected and the whole container startup failed.  This has been later
addressed by
https://github.com/opencontainers/runc/commit/52390d68040637dfc77f9fda6bbe70952423d380
so current Docker runtimes do not suffer from the problem anymore.  There
are still older version of Docker in use and likely hard to get rid of
completely.

Address this by wiping out the file completely and effectively get back to
pre 4.5 era and CONFIG_MEMCG_KMEM=n configuration.

I would recommend backporting to stable trees which have picked up
58056f77502f ("memcg, kmem: further deprecate kmem.limit_in_bytes").

[mhocko@suse.com: restore _KMEM switch case]
Link: https://lkml.kernel.org/r/ZKe5wxdbvPi5Cwd7@dhcp22.suse.cz
Link: https://lkml.kernel.org/r/20230704115240.14672-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Tejun Heo <tj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: page_alloc: avoid false page outside zone error info
Miaohe Lin [Tue, 4 Jul 2023 11:18:23 +0000 (19:18 +0800)]
mm: page_alloc: avoid false page outside zone error info

If pfn is outside zone boundaries in the first round, ret will be set to
1.  But if pfn is changed to inside the zone boundaries in zone span
seqretry path, ret is still set to 1 leading to false page outside zone
error info.

This is from code inspection.  The race window should be really small thus
hard to trigger in real world.

[akpm@linux-foundation.org: code simplification, per Matthew]
Link: https://lkml.kernel.org/r/20230704111823.940331-1-linmiaohe@huawei.com
Fixes: bdc8cb984576 ("[PATCH] memory hotplug locking: zone span seqlock")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftest: add a testcase of ksm zero pages
xu xin [Tue, 13 Jun 2023 03:09:47 +0000 (11:09 +0800)]
selftest: add a testcase of ksm zero pages

Add a function test_unmerge_zero_page() to test the functionality on
unsharing and counting ksm-placed zero pages and counting of this patch
series.

test_unmerge_zero_page() actually contains four subjct test objects:
(1) whether the count of ksm zero pages can update correctly after merging;
(2) whether the count of ksm zero pages can update correctly after
    unmerging by madvise(...MADV_UNMERGEABLE);
(3) whether the count of ksm zero pages can update correctly after
unmerging by triggering write fault.
(4) whether ksm zero pages are really unmerged.

Link: https://lkml.kernel.org/r/20230613030947.186089-1-yang.yang29@zte.com.cn
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Xiaokai Ran <ran.xiaokai@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc: Xuexin Jiang <jiang.xuexin@zte.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoksm: consider KSM-placed zeropages when calculating KSM profit
xu xin [Tue, 13 Jun 2023 03:09:42 +0000 (11:09 +0800)]
ksm: consider KSM-placed zeropages when calculating KSM profit

When use_zero_pages is enabled, the calculation of ksm profit is not
correct because ksm zero pages is not counted in.  So update the
calculation of KSM profit including the documentation.

Link: https://lkml.kernel.org/r/20230613030942.186041-1-yang.yang29@zte.com.cn
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Xiaokai Ran <ran.xiaokai@zte.com.cn>
Cc: Yang Yang <yang.yang29@zte.com.cn>
Cc: Jiang Xuexin <jiang.xuexin@zte.com.cn>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoksm: add ksm zero pages for each process
xu xin [Tue, 13 Jun 2023 03:09:38 +0000 (11:09 +0800)]
ksm: add ksm zero pages for each process

As the number of ksm zero pages is not included in ksm_merging_pages per
process when enabling use_zero_pages, it's unclear of how many actual
pages are merged by KSM. To let users accurately estimate their memory
demands when unsharing KSM zero-pages, it's necessary to show KSM zero-
pages per process. In addition, it help users to know the actual KSM
profit because KSM-placed zero pages are also benefit from KSM.

since unsharing zero pages placed by KSM accurately is achieved, then
tracking empty pages merging and unmerging is not a difficult thing any
longer.

Since we already have /proc/<pid>/ksm_stat, just add the information of
'ksm_zero_pages' in it.

Link: https://lkml.kernel.org/r/20230613030938.185993-1-yang.yang29@zte.com.cn
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Xiaokai Ran <ran.xiaokai@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc: Xuexin Jiang <jiang.xuexin@zte.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoksm: count all zero pages placed by KSM
xu xin [Tue, 13 Jun 2023 03:09:34 +0000 (11:09 +0800)]
ksm: count all zero pages placed by KSM

As pages_sharing and pages_shared don't include the number of zero pages
merged by KSM, we cannot know how many pages are zero pages placed by KSM
when enabling use_zero_pages, which leads to KSM not being transparent
with all actual merged pages by KSM.  In the early days of use_zero_pages,
zero-pages was unable to get unshared by the ways like MADV_UNMERGEABLE so
it's hard to count how many times one of those zeropages was then
unmerged.

But now, unsharing KSM-placed zero page accurately has been achieved, so
we can easily count both how many times a page full of zeroes was merged
with zero-page and how many times one of those pages was then unmerged.
and so, it helps to estimate memory demands when each and every shared
page could get unshared.

So we add ksm_zero_pages under /sys/kernel/mm/ksm/ to show the number
of all zero pages placed by KSM. Meanwhile, we update the Documentation.

Link: https://lkml.kernel.org/r/20230613030934.185944-1-yang.yang29@zte.com.cn
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc: Xuexin Jiang <jiang.xuexin@zte.com.cn>
Reviewed-by: Xiaokai Ran <ran.xiaokai@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoksm: support unsharing KSM-placed zero pages
xu xin [Tue, 13 Jun 2023 03:09:28 +0000 (11:09 +0800)]
ksm: support unsharing KSM-placed zero pages

Patch series "ksm: support tracking KSM-placed zero-pages", v10.

The core idea of this patch set is to enable users to perceive the number
of any pages merged by KSM, regardless of whether use_zero_page switch has
been turned on, so that users can know how much free memory increase is
really due to their madvise(MERGEABLE) actions.  But the problem is, when
enabling use_zero_pages, all empty pages will be merged with kernel zero
pages instead of with each other as use_zero_pages is disabled, and then
these zero-pages are no longer monitored by KSM.

The motivations to do this is seen at:
https://lore.kernel.org/lkml/202302100915227721315@zte.com.cn/

In one word, we hope to implement the support for KSM-placed zero pages
tracking without affecting the feature of use_zero_pages, so that app
developer can also benefit from knowing the actual KSM profit by getting
KSM-placed zero pages to optimize applications eventually when
/sys/kernel/mm/ksm/use_zero_pages is enabled.

This patch (of 5):

When use_zero_pages of ksm is enabled, madvise(addr, len,
MADV_UNMERGEABLE) and other ways (like write 2 to /sys/kernel/mm/ksm/run)
to trigger unsharing will *not* actually unshare the shared zeropage as
placed by KSM (which is against the MADV_UNMERGEABLE documentation).  As
these KSM-placed zero pages are out of the control of KSM, the related
counts of ksm pages don't expose how many zero pages are placed by KSM
(these special zero pages are different from those initially mapped zero
pages, because the zero pages mapped to MADV_UNMERGEABLE areas are
expected to be a complete and unshared page).

To not blindly unshare all shared zero_pages in applicable VMAs, the patch
use pte_mkdirty (related with architecture) to mark KSM-placed zero pages.
Thus, MADV_UNMERGEABLE will only unshare those KSM-placed zero pages.

In addition, we'll reuse this mechanism to reliably identify KSM-placed
ZeroPages to properly account for them (e.g., calculating the KSM profit
that includes zeropages) in the latter patches.

The patch will not degrade the performance of use_zero_pages as it doesn't
change the way of merging empty pages in use_zero_pages's feature.

Link: https://lkml.kernel.org/r/202306131104554703428@zte.com.cn
Link: https://lkml.kernel.org/r/20230613030928.185882-1-yang.yang29@zte.com.cn
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc: Xuexin Jiang <jiang.xuexin@zte.com.cn>
Reviewed-by: Xiaokai Ran <ran.xiaokai@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/migrate_device: try to handle swapcache pages
Mika Penttilä [Wed, 7 Jun 2023 17:29:44 +0000 (20:29 +0300)]
mm/migrate_device: try to handle swapcache pages

Migrating file pages and swapcache pages into device memory is not
supported.  Try to get rid of the swap cache, and if successful, go ahead
as with other anonymous pages.

Link: https://lkml.kernel.org/r/20230607172944.11713-1-mpenttil@redhat.com
Signed-off-by: Mika Penttilä <mpenttil@redhat.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Alistair Popple <apopple@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftests: cgroup: add zswap-memcg unwanted writeback test
Domenico Cerasuolo [Wed, 21 Jun 2023 15:35:48 +0000 (17:35 +0200)]
selftests: cgroup: add zswap-memcg unwanted writeback test

Add a test to verify that when a memcg hits its limit in zswap, it doesn't
trigger an unwanted writeback that would result in pages not owned by that
memcg to be sent to disk, even if zswap isn't full.  This was fixed by
commit 0bdf0efa180a("zswap: do not shrink if cgroup may not zswap").

Link: https://lkml.kernel.org/r/20230621153548.428093-4-cerasuolodomenico@gmail.com
Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Zefan Li <lizefan.x@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftests: cgroup: add test_zswap with no kmem bypass test
Domenico Cerasuolo [Wed, 21 Jun 2023 15:35:47 +0000 (17:35 +0200)]
selftests: cgroup: add test_zswap with no kmem bypass test

Add a cgroup selftest that verifies memcg charging in zswap.  The original
issue was that kmem bypass was applied to pages swapped out to zswap by
kswapd, resulting in zswapped memory not being charged.  It was fixed by
commit cd08d80ecdac("mm: correctly charge compressed memory to its
memcg").

Link: https://lkml.kernel.org/r/20230621153548.428093-3-cerasuolodomenico@gmail.com
Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Zefan Li <lizefan.x@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftests: cgroup: add test_zswap program
Domenico Cerasuolo [Wed, 21 Jun 2023 15:35:46 +0000 (17:35 +0200)]
selftests: cgroup: add test_zswap program

Patch series "selftests: cgroup: add zswap test program".

This series adds 2 zswap related selftests that verify known and fixed
issues.  A new dedicated test program (test_zswap) is proposed since the
test cases are specific to zswap and hosts specific helpers.

The first patch adds the (empty) test program, while the other 2 add an
actual test function each.

This patch (of 3):

Add empty cgroup-zswap self test scaffold program, test functions to be
added in the next commits.

Link: https://lkml.kernel.org/r/20230621153548.428093-1-cerasuolodomenico@gmail.com
Link: https://lkml.kernel.org/r/20230621153548.428093-2-cerasuolodomenico@gmail.com
Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Zefan Li <lizefan.x@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/page_alloc: use write_seqlock_irqsave() instead write_seqlock() + local_irq_save().
Sebastian Andrzej Siewior [Fri, 23 Jun 2023 20:15:17 +0000 (22:15 +0200)]
mm/page_alloc: use write_seqlock_irqsave() instead write_seqlock() + local_irq_save().

__build_all_zonelists() acquires zonelist_update_seq by first disabling
interrupts via local_irq_save() and then acquiring the seqlock with
write_seqlock().  This is troublesome and leads to problems on PREEMPT_RT.
The problem is that the inner spinlock_t becomes a sleeping lock on
PREEMPT_RT and must not be acquired with disabled interrupts.

The API provides write_seqlock_irqsave() which does the right thing in one
step.  printk_deferred_enter() has to be invoked in non-migrate-able
context to ensure that deferred printing is enabled and disabled on the
same CPU.  This is the case after zonelist_update_seq has been acquired.

There was discussion on the first submission that the order should be:
local_irq_disable();
printk_deferred_enter();
write_seqlock();

to avoid pitfalls like having an unaccounted printk() coming from
write_seqlock_irqsave() before printk_deferred_enter() is invoked.  The
only origin of such a printk() can be a lockdep splat because the lockdep
annotation happens after the sequence count is incremented.  This is
exceptional and subject to change.

It was also pointed that PREEMPT_RT can be affected by the printk problem
since its write_seqlock_irqsave() does not really disable interrupts.
This isn't the case because PREEMPT_RT's printk implementation differs
from the mainline implementation in two important aspects:

- Printing happens in a dedicated threads and not at during the
  invocation of printk().
- In emergency cases where synchronous printing is used, a different
  driver is used which does not use tty_port::lock.

Acquire zonelist_update_seq with write_seqlock_irqsave() and then defer
printk output.

Link: https://lkml.kernel.org/r/20230623201517.yw286Knb@linutronix.de
Fixes: 1007843a91909 ("mm/page_alloc: fix potential deadlock on zonelist_update_seq seqlock")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: John Ogness <john.ogness@linutronix.de>
Cc: Luis Claudio R. Goncalves <lgoncalv@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Waiman Long <longman@redhat.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agozsmalloc: remove zs_compact_control
Minchan Kim [Sat, 24 Jun 2023 05:12:16 +0000 (14:12 +0900)]
zsmalloc: remove zs_compact_control

__zs_compact always putback src_zspage into class list after
migrate_zspage.  Thus, we don't need to keep last position of src_zspage
any more.  Let's remove it.

Link: https://lkml.kernel.org/r/20230624053120.643409-4-senozhatsky@chromium.org
Signed-off-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Alexey Romanov <avromanov@sberdevices.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agozsmalloc: move migration destination zspage inuse check
Sergey Senozhatsky [Sat, 24 Jun 2023 05:12:15 +0000 (14:12 +0900)]
zsmalloc: move migration destination zspage inuse check

Destination zspage fullness check need to be done after zs_object_copy()
because that's where source and destination zspages fullness change.
Checking destination zspage fullness before zs_object_copy() may cause
migration to loop through source zspage sub-pages scanning for allocate
objects just to find out at the end that the destination zspage is full.

Link: https://lkml.kernel.org/r/20230624053120.643409-3-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Alexey Romanov <avromanov@sberdevices.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agozsmalloc: do not scan for allocated objects in empty zspage
Sergey Senozhatsky [Sat, 24 Jun 2023 05:12:14 +0000 (14:12 +0900)]
zsmalloc: do not scan for allocated objects in empty zspage

Patch series "zsmalloc: small compaction improvements", v2.

A tiny series that can reduce the number of find_alloced_obj() invocations
(which perform a linear scan of sub-page) during compaction.  Inspired by
Alexey Romanov's findings.

This patch (of 3):

zspage migration can terminate as soon as it moves the last allocated
object from the source zspage.  Add a simple helper zspage_empty() that
tests zspage ->inuse on each migration iteration.

Link: https://lkml.kernel.org/r/20230624053120.643409-2-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Suggested-by: Alexey Romanov <AVRomanov@sberdevices.ru>
Reviewed-by: Alexey Romanov <avromanov@sberdevices.ru>
Acked-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/mm_init.c: remove obsolete macro HASH_SMALL
Miaohe Lin [Sun, 25 Jun 2023 02:13:23 +0000 (10:13 +0800)]
mm/mm_init.c: remove obsolete macro HASH_SMALL

HASH_SMALL only works when parameter numentries is 0. But the sole caller
futex_init() never calls alloc_large_system_hash() with numentries set to
0. So HASH_SMALL is obsolete and remove it.

Link: https://lkml.kernel.org/r/20230625021323.849147-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: André Almeida <andrealmeid@igalia.com>
Cc: Darren Hart <dvhart@infradead.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/page_alloc: fix min_free_kbytes calculation regarding ZONE_MOVABLE
liuq [Sun, 25 Jun 2023 03:16:56 +0000 (11:16 +0800)]
mm/page_alloc: fix min_free_kbytes calculation regarding ZONE_MOVABLE

The current calculation of min_free_kbytes only uses ZONE_DMA and
ZONE_NORMAL pages,but the ZONE_MOVABLE zone->_watermark[WMARK_MIN] will
also divide part of min_free_kbytes.This will cause the min watermark of
ZONE_NORMAL to be too small in the presence of ZONE_MOVEABLE.

__GFP_HIGH and PF_MEMALLOC allocations usually don't need movable zone
pages, so just like ZONE_HIGHMEM, cap pages_min to a small value in
__setup_per_zone_wmarks().

On my testing machine with 16GB of memory (transparent hugepage is turned
off by default, and movablecore=12G is configured) The following is a
comparative test data of watermark_min

no patch add patch
ZONE_DMA 1 8
ZONE_DMA32 151 709
ZONE_NORMAL 233 1113
ZONE_MOVABLE 1434 128
min_free_kbytes 7288 7326

Link: https://lkml.kernel.org/r/20230625031656.23941-1-liuq131@chinatelecom.cn
Signed-off-by: liuq <liuq131@chinatelecom.cn>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agofs: convert block_commit_write to return void
Bean Huo [Mon, 26 Jun 2023 05:55:18 +0000 (07:55 +0200)]
fs: convert block_commit_write to return void

block_commit_write() always returns 0, this patch changes it to return
void.

Link: https://lkml.kernel.org/r/20230626055518.842392-3-beanhuo@iokpp.de
Signed-off-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Luís Henriques <ocfs2-devel@oss.oracle.com>
Cc: Mark Fasheh <mark@fasheh.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agofs/buffer: clean up block_commit_write
Bean Huo [Mon, 26 Jun 2023 05:55:17 +0000 (07:55 +0200)]
fs/buffer: clean up block_commit_write

Originally inode is used to get blksize, after commit 45bce8f3e343
("fs/buffer.c: make block-size be per-page and protected by the page
lock"), __block_commit_write no longer uses this parameter inode.

[akpm@linux-foundation.org: remove now-unused local `inode']
Link: https://lkml.kernel.org/r/20230626055518.842392-2-beanhuo@iokpp.de
Signed-off-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Luís Henriques <ocfs2-devel@oss.oracle.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: memory-failure: remove unneeded 'inline' annotation
Miaohe Lin [Mon, 26 Jun 2023 11:43:43 +0000 (19:43 +0800)]
mm: memory-failure: remove unneeded 'inline' annotation

Remove unneeded 'inline' annotation from num_poisoned_pages_inc() and
num_poisoned_pages_sub().  No functional change intended.

Link: https://lkml.kernel.org/r/20230626114343.1846587-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomemory tier: use helper function destroy_memory_type()
Miaohe Lin [Mon, 26 Jun 2023 12:10:53 +0000 (20:10 +0800)]
memory tier: use helper function destroy_memory_type()

Use helper function destroy_memory_type() to release memtype instead of
open code it to help improve code readability a bit.  No functional change
intended.

Link: https://lkml.kernel.org/r/20230626121053.1916447-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Wei Xu <weixugc@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: memory-failure: remove unneeded page state check in shake_page()
Miaohe Lin [Wed, 28 Jun 2023 01:49:29 +0000 (09:49 +0800)]
mm: memory-failure: remove unneeded page state check in shake_page()

Remove unneeded PageLRU(p) and is_free_buddy_page(p) check as slab caches
are not shrunk now.  This check can be added back when a lightweight range
based shrinker is available.

Link: https://lkml.kernel.org/r/20230628014929.3441386-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomaple_tree: add a fast path case in mas_wr_slot_store()
Peng Zhang [Wed, 28 Jun 2023 07:36:57 +0000 (15:36 +0800)]
maple_tree: add a fast path case in mas_wr_slot_store()

When expanding a range in two directions, only partially overwriting the
previous and next ranges, the number of entries will not be increased, so
we can just update the pivots as a fast path. However, it may introduce
potential risks in RCU mode, because it updates two pivots. We only
enable it in non-RCU mode.

Link: https://lkml.kernel.org/r/20230628073657.75314-5-zhangpeng.00@bytedance.com
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomaple_tree: optimize mas_wr_append(), also improve duplicating VMAs
Peng Zhang [Wed, 28 Jun 2023 07:36:56 +0000 (15:36 +0800)]
maple_tree: optimize mas_wr_append(), also improve duplicating VMAs

When the new range can be completely covered by the original last range
without touching the boundaries on both sides, two new entries can be
appended to the end as a fast path. We update the original last pivot at
the end, and the newly appended two entries will not be accessed before
this, so it is also safe in RCU mode.

This is useful for sequential insertion, which is what we do in
dup_mmap(). Enabling BENCH_FORK in test_maple_tree and just running
bench_forking() gives the following time-consuming numbers:

before:               after:
17,874.83 msec        15,738.38 msec

It shows about a 12% performance improvement for duplicating VMAs.

Link: https://lkml.kernel.org/r/20230628073657.75314-4-zhangpeng.00@bytedance.com
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomaple_tree: add test for expanding range in RCU mode
Peng Zhang [Wed, 28 Jun 2023 07:36:55 +0000 (15:36 +0800)]
maple_tree: add test for expanding range in RCU mode

Add test for expanding range in RCU mode. If we use the fast path of the
slot store to expand range in RCU mode, this test will fail.

Link: https://lkml.kernel.org/r/20230628073657.75314-3-zhangpeng.00@bytedance.com
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomaple_tree: add test for mas_wr_modify() fast path
Peng Zhang [Wed, 28 Jun 2023 07:36:54 +0000 (15:36 +0800)]
maple_tree: add test for mas_wr_modify() fast path

Patch series "Optimize the fast path of mas_store()", v4.

Add fast paths for mas_wr_append() and mas_wr_slot_store() respectively.
The newly added fast path of mas_wr_append() is used in fork() and how
much it benefits fork() depends on how many VMAs are duplicated.

Thanks Liam for the review.

This patch (of 4):

Add tests for all cases of mas_wr_append() and mas_wr_slot_store().

Link: https://lkml.kernel.org/r/20230628073657.75314-1-zhangpeng.00@bytedance.com
Link: https://lkml.kernel.org/r/20230628073657.75314-2-zhangpeng.00@bytedance.com
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/filemap.c: fix update prev_pos after one read request done
Haibo Li [Wed, 28 Jun 2023 11:02:20 +0000 (19:02 +0800)]
mm/filemap.c: fix update prev_pos after one read request done

ra->prev_pos tracks the last visited byte in the previous read request.
It is used to check whether it is sequential read in ondemand_readahead
and thus affects the readahead window.

After commit 06c0444290ce ("mm/filemap.c: generic_file_buffered_read() now
uses find_get_pages_contig"), update logic of prev_pos is changed.  It
updates prev_pos after each return from filemap_get_pages().  But the read
request from user may be not fully completed at this point.  The updated
prev_pos impacts the subsequent readahead window.

The real problem is performance drop of fsck_msdos between linux-5.4 and
linux-5.15(also linux-6.4).  Comparing to linux-5.4,It spends about 110%
time and read 140% pages.  The read pattern of fsck_msdos is not fully
sequential.

Simplified read pattern of fsck_msdos likes below:
1.read at page offset 0xa,size 0x1000
2.read at other page offset like 0x20,size 0x1000
3.read at page offset 0xa,size 0x4000
4.read at page offset 0xe,size 0x1000

Here is the read status on linux-6.4:
1.after read at page offset 0xa,size 0x1000
    ->page ofs 0xa go into pagecache
2.after read at page offset 0x20,size 0x1000
    ->page ofs 0x20 go into pagecache
3.read at page offset 0xa,size 0x4000
    ->filemap_get_pages read ofs 0xa from pagecache and returns
    ->prev_pos is updated to 0xb and goto next loop
    ->filemap_get_pages tends to read ofs 0xb,size 0x3000
    ->initial_readahead case in ondemand_readahead since prev_pos is
      the same as request ofs.
    ->read 8 pages while async size is 5 pages
      (PageReadahead flag at page 0xe)
4.read at page offset 0xe,size 0x1000
    ->hit page 0xe with PageReadahead flag set,double the ra_size.
      read 16 pages while async size is 16 pages
Now it reads 24 pages while actually uses 5 pages

on linux-5.4:
1.the same as 6.4
2.the same as 6.4
3.read at page offset 0xa,size 0x4000
    ->read ofs 0xa from pagecache
    ->read ofs 0xb,size 0x3000 using page_cache_sync_readahead
      read 3 pages
    ->prev_pos is updated to 0xd before generic_file_buffered_read
      returns
4.read at page offset 0xe,size 0x1000
    ->initial_readahead case in ondemand_readahead since
      request ofs-prev_pos==1
    ->read 4 pages while async size is 3 pages

Now it reads 7 pages while actually uses 5 pages.

In above demo, the initial_readahead case is triggered by offset of user
request on linux-5.4.  While it may be triggered by update logic of
prev_pos on linux-6.4.

To fix the performance drop, update prev_pos after finishing one read
request.

Link: https://lkml.kernel.org/r/20230628110220.120134-1-haibo.li@mediatek.com
Signed-off-by: Haibo Li <haibo.li@mediatek.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftests/mm: add gup test matrix in run_vmtests.sh
Peter Xu [Wed, 28 Jun 2023 21:53:10 +0000 (17:53 -0400)]
selftests/mm: add gup test matrix in run_vmtests.sh

Add a matrix for testing gup based on the current gup_test.  Only run the
matrix when -a is specified because it's a bit slow.

It covers:

  - Different types of huge pages: thp, hugetlb, or no huge page
  - Permissions: Write / Read-only
  - Fast-gup, with/without
  - Types of the GUP: pin / gup / longterm pins
  - Shared / Private memories
  - GUP size: 1 / 512 / random page sizes

Link: https://lkml.kernel.org/r/20230628215310.73782-9-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoselftests/mm: add -a to run_vmtests.sh
Peter Xu [Wed, 28 Jun 2023 21:53:09 +0000 (17:53 -0400)]
selftests/mm: add -a to run_vmtests.sh

Allows to specify optional tests in run_vmtests.sh, where we can run time
consuming test matrix only when user specified "-a".

Link: https://lkml.kernel.org/r/20230628215310.73782-8-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/gup: retire follow_hugetlb_page()
Peter Xu [Wed, 28 Jun 2023 21:53:08 +0000 (17:53 -0400)]
mm/gup: retire follow_hugetlb_page()

Now __get_user_pages() should be well prepared to handle thp completely,
as long as hugetlb gup requests even without the hugetlb's special path.

Time to retire follow_hugetlb_page().

Tweak misc comments to reflect reality of follow_hugetlb_page()'s removal.

Link: https://lkml.kernel.org/r/20230628215310.73782-7-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/gup: accelerate thp gup even for "pages != NULL"
Peter Xu [Wed, 28 Jun 2023 21:53:07 +0000 (17:53 -0400)]
mm/gup: accelerate thp gup even for "pages != NULL"

The acceleration of THP was done with ctx.page_mask, however it'll be
ignored if **pages is non-NULL.

The old optimization was introduced in 2013 in 240aadeedc4a ("mm:
accelerate mm_populate() treatment of THP pages").  It didn't explain why
we can't optimize the **pages non-NULL case.  It's possible that at that
time the major goal was for mm_populate() which should be enough back
then.

Optimize thp for all cases, by properly looping over each subpage, doing
cache flushes, and boost refcounts / pincounts where needed in one go.

This can be verified using gup_test below:

  # chrt -f 1 ./gup_test -m 512 -t -L -n 1024 -r 10

Before:    13992.50 ( +-8.75%)
After:       378.50 (+-69.62%)

Link: https://lkml.kernel.org/r/20230628215310.73782-6-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/gup: cleanup next_page handling
Peter Xu [Wed, 28 Jun 2023 21:53:06 +0000 (17:53 -0400)]
mm/gup: cleanup next_page handling

The only path that doesn't use generic "**pages" handling is the gate vma.
Make it use the same path, meanwhile tune the next_page label upper to
cover "**pages" handling.  This prepares for THP handling for "**pages".

Link: https://lkml.kernel.org/r/20230628215310.73782-5-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/hugetlb: add page_mask for hugetlb_follow_page_mask()
Peter Xu [Wed, 28 Jun 2023 21:53:05 +0000 (17:53 -0400)]
mm/hugetlb: add page_mask for hugetlb_follow_page_mask()

follow_page() doesn't need it, but we'll start to need it when unifying
gup for hugetlb.

Link: https://lkml.kernel.org/r/20230628215310.73782-4-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/hugetlb: prepare hugetlb_follow_page_mask() for FOLL_PIN
Peter Xu [Wed, 28 Jun 2023 21:53:04 +0000 (17:53 -0400)]
mm/hugetlb: prepare hugetlb_follow_page_mask() for FOLL_PIN

follow_page() doesn't use FOLL_PIN, meanwhile hugetlb seems to not be the
target of FOLL_WRITE either.  However add the checks.

Namely, either the need to CoW due to missing write bit, or proper
unsharing on !AnonExclusive pages over R/O pins to reject the follow page.
That brings this function closer to follow_hugetlb_page().

So we don't care before, and also for now.  But we'll care if we switch
over slow-gup to use hugetlb_follow_page_mask().  We'll also care when to
return -EMLINK properly, as that's the gup internal api to mean "we should
unshare".  Not really needed for follow page path, though.

When at it, switching the try_grab_page() to use WARN_ON_ONCE(), to be
clear that it just should never fail.  When error happens, instead of
setting page==NULL, capture the errno instead.

Link: https://lkml.kernel.org/r/20230628215310.73782-3-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/hugetlb: handle FOLL_DUMP well in follow_page_mask()
Peter Xu [Wed, 28 Jun 2023 21:53:03 +0000 (17:53 -0400)]
mm/hugetlb: handle FOLL_DUMP well in follow_page_mask()

Patch series "mm/gup: Unify hugetlb, speed up thp", v4.

Hugetlb has a special path for slow gup that follow_page_mask() is
actually skipped completely along with faultin_page().  It's not only
confusing, but also duplicating a lot of logics that generic gup already
has, making hugetlb slightly special.

This patchset tries to dedup the logic, by first touching up the slow gup
code to be able to handle hugetlb pages correctly with the current follow
page and faultin routines (where we're mostly there..  due to 10 years ago
we did try to optimize thp, but half way done; more below), then at the
last patch drop the special path, then the hugetlb gup will always go the
generic routine too via faultin_page().

Note that hugetlb is still special for gup, mostly due to the pgtable
walking (hugetlb_walk()) that we rely on which is currently per-arch.  But
this is still one small step forward, and the diffstat might be a proof
too that this might be worthwhile.

Then for the "speed up thp" side: as a side effect, when I'm looking at
the chunk of code, I found that thp support is actually partially done.
It doesn't mean that thp won't work for gup, but as long as **pages
pointer passed over, the optimization will be skipped too.  Patch 6 should
address that, so for thp we now get full speed gup.

For a quick number, "chrt -f 1 ./gup_test -m 512 -t -L -n 1024 -r 10"
gives me 13992.50us -> 378.50us.  Gup_test is an extreme case, but just to
show how it affects thp gups.

This patch (of 8):

Firstly, the no_page_table() is meaningless for hugetlb which is a no-op
there, because a hugetlb page always satisfies:

  - vma_is_anonymous() == false
  - vma->vm_ops->fault != NULL

So we can already safely remove it in hugetlb_follow_page_mask(), alongside
with the page* variable.

Meanwhile, what we do in follow_hugetlb_page() actually makes sense for a
dump: we try to fault in the page only if the page cache is already
allocated.  Let's do the same here for follow_page_mask() on hugetlb.

It should so far has zero effect on real dumps, because that still goes
into follow_hugetlb_page().  But this may start to influence a bit on
follow_page() users who mimics a "dump page" scenario, but hopefully in a
good way.  This also paves way for unifying the hugetlb gup-slow.

Link: https://lkml.kernel.org/r/20230628215310.73782-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20230628215310.73782-2-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: James Houghton <jthoughton@google.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kirill A . Shutemov <kirill@shutemov.name>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoarm64: mte: simplify swap tag restoration logic
Peter Collingbourne [Tue, 23 May 2023 00:43:10 +0000 (17:43 -0700)]
arm64: mte: simplify swap tag restoration logic

As a result of the patches "mm: Call arch_swap_restore() from
do_swap_page()" and "mm: Call arch_swap_restore() from unuse_pte()", there
are no circumstances in which a swapped-in page is installed in a page
table without first having arch_swap_restore() called on it.  Therefore,
we no longer need the logic in set_pte_at() that restores the tags, so
remove it.

Link: https://lkml.kernel.org/r/20230523004312.1807357-4-pcc@google.com
Link: https://linux-review.googlesource.com/id/I8ad54476f3b2d0144ccd8ce0c1d7a2963e5ff6f3
Signed-off-by: Peter Collingbourne <pcc@google.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Alexandru Elisei <alexandru.elisei@arm.com>
Cc: Chinwen Chang <chinwen.chang@mediatek.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Evgenii Stepanov <eugenis@google.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: kasan-dev@googlegroups.com
Cc: kasan-dev <kasan-dev@googlegroups.com>
Cc: "Kuan-Ying Lee (李冠穎)" <Kuan-Ying.Lee@mediatek.com>
Cc: Qun-Wei Lin <qun-wei.lin@mediatek.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: call arch_swap_restore() from unuse_pte()
Peter Collingbourne [Tue, 23 May 2023 00:43:09 +0000 (17:43 -0700)]
mm: call arch_swap_restore() from unuse_pte()

We would like to move away from requiring architectures to restore
metadata from swap in the set_pte_at() implementation, as this is not only
error-prone but adds complexity to the arch-specific code.  This requires
us to call arch_swap_restore() before calling swap_free() whenever pages
are restored from swap.  We are currently doing so everywhere except in
unuse_pte(); do so there as well.

Link: https://lkml.kernel.org/r/20230523004312.1807357-3-pcc@google.com
Link: https://linux-review.googlesource.com/id/I68276653e612d64cde271ce1b5a99ae05d6bbc4f
Signed-off-by: Peter Collingbourne <pcc@google.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Alexandru Elisei <alexandru.elisei@arm.com>
Cc: Chinwen Chang <chinwen.chang@mediatek.com>
Cc: Evgenii Stepanov <eugenis@google.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: kasan-dev <kasan-dev@googlegroups.com>
Cc: "Kuan-Ying Lee (李冠穎)" <Kuan-Ying.Lee@mediatek.com>
Cc: Qun-Wei Lin <qun-wei.lin@mediatek.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: make show_free_areas() static
Kefeng Wang [Fri, 30 Jun 2023 06:22:53 +0000 (14:22 +0800)]
mm: make show_free_areas() static

All callers of show_free_areas() pass 0 and NULL, so we can directly use
show_mem() instead of show_free_areas(0, NULL), which could make
show_free_areas() a static function.

Link: https://lkml.kernel.org/r/20230630062253.189440-2-wangkefeng.wang@huawei.com
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: remove arguments of show_mem()
Kefeng Wang [Fri, 30 Jun 2023 06:22:52 +0000 (14:22 +0800)]
mm: remove arguments of show_mem()

All callers of show_mem() pass 0 and NULL, so we can remove the two
arguments by directly calling __show_mem(0, NULL, MAX_NR_ZONES - 1) in
show_mem().

Link: https://lkml.kernel.org/r/20230630062253.189440-1-wangkefeng.wang@huawei.com
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: make MEMFD_CREATE into a selectable config option
Thomas Weißschuh [Fri, 30 Jun 2023 09:08:53 +0000 (11:08 +0200)]
mm: make MEMFD_CREATE into a selectable config option

The memfd_create() syscall, enabled by CONFIG_MEMFD_CREATE, is useful on
its own even when not required by CONFIG_TMPFS or CONFIG_HUGETLBFS.

Split it into its own proper bool option that can be enabled by users.

Move that option into mm/ where the code itself also lies.  Also add
"select" statements to CONFIG_TMPFS and CONFIG_HUGETLBFS so they
automatically enable CONFIG_MEMFD_CREATE as before.

Link: https://lkml.kernel.org/r/20230630-config-memfd-v1-1-9acc3ae38b5a@weissschuh.net
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Tested-by: Zhangjin Wu <falcon@tinylab.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: remove page_rmapping()
ZhangPeng [Sat, 1 Jul 2023 03:28:53 +0000 (11:28 +0800)]
mm: remove page_rmapping()

After converting the last user to folio_raw_mapping(), we can safely
remove the function.

Link: https://lkml.kernel.org/r/20230701032853.258697-3-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: use a folio in fault_dirty_shared_page()
ZhangPeng [Sat, 1 Jul 2023 03:28:52 +0000 (11:28 +0800)]
mm: use a folio in fault_dirty_shared_page()

We can replace four implicit calls to compound_head() with one by using
folio.

Link: https://lkml.kernel.org/r/20230701032853.258697-2-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoswap: stop add to avail list if swap is full
Ma Wupeng [Tue, 27 Jun 2023 12:08:33 +0000 (20:08 +0800)]
swap: stop add to avail list if swap is full

Our test finds a WARN_ON in add_to_avail_list.  During add_to_avail_list,
avail_lists is already in swap_avail_heads, while leads to this WARN_ON.

Here is the simplified calltrace:

------------[ cut here ]------------
Call trace:
 add_to_avail_list+0xb8/0xc0
 swap_range_free+0x110/0x138
 swapcache_free_entries+0x100/0x1c0
 free_swap_slot+0xbc/0xe0
 put_swap_folio+0x1f0/0x2ec
 delete_from_swap_cache+0x6c/0xd0
 folio_free_swap+0xa4/0xe4
 __try_to_reclaim_swap+0x9c/0x190
 free_swap_and_cache+0x84/0x88
 unmap_page_range+0x31c/0x934
 unmap_single_vma.isra.0+0x48/0x84
 unmap_vmas+0x98/0x10c
 exit_mmap+0xa4/0x210
 mmput+0x88/0x158
 do_exit+0x284/0x970
 do_group_exit+0x34/0x90
 post_copy_siginfo_from_user32+0x0/0x1cc
 do_notify_resume+0x15c/0x470
 el0_svc+0x74/0x84
 el0t_64_sync_handler+0xb8/0xbc
 el0t_64_sync+0x190/0x194

During swapoff, try_to_unuse fails to alloc memory due to memory limit and
this leads to the failure of swapoff and causes re-insertion of swap space
back into swap_list.  During _enable_swap_info, this swap device is added
to avail list even this swap device if full.  At the same time, one entry
in this full swap device in released and we try to add this device into
avail list and find it is already in the avail list.  This causes this
WARN_ON.

To fix this.  Don't add to avail list is swap is full.

[akpm@linux-foundation.org: coding-style cleanups]
Link: https://lkml.kernel.org/r/20230627120833.2230766-3-mawupeng1@huawei.com
Signed-off-by: Ma Wupeng <mawupeng1@huawei.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agoswap: cleanup duplicated WARN_ON in add_to_avail_list
Ma Wupeng [Tue, 27 Jun 2023 12:08:32 +0000 (20:08 +0800)]
swap: cleanup duplicated WARN_ON in add_to_avail_list

Patch series "fix WARN_ON in add_to_avail_list".

Empty check for plist_node is checked in add_to_avail_list and plist_add.
Drop the duplicate one in add_to_avail_list.

Link: https://lkml.kernel.org/r/20230627120833.2230766-1-mawupeng1@huawei.com
Link: https://lkml.kernel.org/r/20230627120833.2230766-2-mawupeng1@huawei.com
Signed-off-by: Ma Wupeng <mawupeng1@huawei.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: increase usage of folio_next_index() helper
Sidhartha Kumar [Tue, 27 Jun 2023 17:43:49 +0000 (10:43 -0700)]
mm: increase usage of folio_next_index() helper

Simplify code pattern of 'folio->index + folio_nr_pages(folio)' by using
the existing helper folio_next_index().

Link: https://lkml.kernel.org/r/20230627174349.491803-1-sidhartha.kumar@oracle.com
Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Suggested-by: Christoph Hellwig <hch@infradead.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm/mm_init.c: update obsolete comment in get_pfn_range_for_nid()
Miaohe Lin [Sun, 25 Jun 2023 03:33:40 +0000 (11:33 +0800)]
mm/mm_init.c: update obsolete comment in get_pfn_range_for_nid()

Since commit 633c0666b5a5 ("Memoryless nodes: drop one memoryless node boot
warning"), the warning for a node with no available memory is removed.
Update the corresponding comment.

Link: https://lkml.kernel.org/r/20230625033340.1054103-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomaple_tree: fix a few documentation issues
Thomas Gleixner [Tue, 23 May 2023 20:51:01 +0000 (22:51 +0200)]
maple_tree: fix a few documentation issues

The documentation of mt_next() claims that it starts the search at the
provided index.  That's incorrect as it starts the search after the
provided index.

The documentation of mt_find() is slightly confusing.  "Handles locking"
is not really helpful as it does not explain how the "locking" works.
Also the documentation of index talks about a range, while in reality the
index is updated on a succesful search to the index of the found entry
plus one.

Fix similar issues for mt_find_after() and mt_prev().

Reword the confusing "Note: Will not return the zero entry." comment on
mt_for_each() and document @__index correctly.

Link: https://lkml.kernel.org/r/87ttw2n556.ffs@tglx
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Shanker Donthineni <sdonthineni@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
14 months agomm: madvise: fix uneven accounting of psi
Charan Teja Kalla [Mon, 3 Jul 2023 14:06:41 +0000 (19:36 +0530)]
mm: madvise: fix uneven accounting of psi

A folio turns into a Workingset during:
1) shrink_active_list() placing the folio from active to inactive list.
2) When a workingset transition is happening during the folio refault.

And when Workingset is set on a folio, PSI for memory can be accounted
during a) That folio is being reclaimed and b) Refault of that folio,
for usual reclaims.

This accounting of PSI for memory is not consistent for reclaim +
refault operation between usual reclaim and madvise(COLD/PAGEOUT) which
deactivate or proactively reclaim a folio:
a) A folio started at inactive and moved to active as part of accesses.
Workingset is absent on the folio thus refault of it when reclaimed
through MADV_PAGEOUT operation doesn't account for PSI.

b) When the same folio transition from inactive->active and then to
inactive through shrink_active_list(). Workingset is set on the folio
thus refault of it when reclaimed through MADV_PAGEOUT operation
accounts for PSI.

c) When the same folio is part of active list directly as a result of
folio refault and this was a workingset folio prior to eviction.
Workingset is set on the folio thus the refault of it when reclaimed
through MADV_PAGEOUT/MADV_COLD operation accounts for PSI.

d) MADV_COLD transfers the folio from active list to inactive
list. Such folios may not have the Workingset thus refault operation on
such folio doesn't account for PSI.

As said above, refault operation caused because of MADV_PAGEOUT on a
folio is accounts for memory PSI in b) and c) but not in a). Refault
caused by the reclaim of a folio on which MADV_COLD is performed
accounts memory PSI in c) but not in d). These behaviours are
inconsistent w.r.t usual reclaim + refault operation. Make this PSI
accounting always consistent by turning a folio into a workingset one
whenever it is leaving the active list. Also, accounting of PSI on a
folio whenever it leaves the active list as part of the
MADV_COLD/PAGEOUT operation helps the users whether they are operating
on proper folios[1].

[1] https://lore.kernel.org/all/20230605180013.GD221380@cmpxchg.org/

Link: https://lkml.kernel.org/r/1688393201-11135-1-git-send-email-quic_charante@quicinc.com
Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com>
Suggested-by: Suren Baghdasaryan <surenb@google.com>
Reported-by: Sai Manobhiram Manapragada <quic_smanapra@quicinc.com>
Reported-by: Pavan Kondeti <quic_pkondeti@quicinc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Pavankumar Kondeti <quic_pkondeti@quicinc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
15 months agoLinux 6.5-rc4
Linus Torvalds [Sun, 30 Jul 2023 20:23:47 +0000 (13:23 -0700)]
Linux 6.5-rc4

15 months agoMerge tag 'spi-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Linus Torvalds [Sun, 30 Jul 2023 19:54:31 +0000 (12:54 -0700)]
Merge tag 'spi-fix-v6.5-rc3' of git://git./linux/kernel/git/broonie/spi

Pull spi fixes from Mark Brown:
 "A bunch of fixes for the Qualcomm QSPI driver, fixing multiple issues
  with the newly added DMA mode - it had a number of issues exposed when
  tested in a wider range of use cases, both race condition style issues
  and issues with different inputs to those that had been used in test"

* tag 'spi-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: spi-qcom-qspi: Add mem_ops to avoid PIO for badly sized reads
  spi: spi-qcom-qspi: Fallback to PIO for xfers that aren't multiples of 4 bytes
  spi: spi-qcom-qspi: Add DMA_CHAIN_DONE to ALL_IRQS
  spi: spi-qcom-qspi: Call dma_wmb() after setting up descriptors
  spi: spi-qcom-qspi: Use GFP_ATOMIC flag while allocating for descriptor
  spi: spi-qcom-qspi: Ignore disabled interrupts' status in isr

15 months agoMerge tag 'regulator-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 30 Jul 2023 19:52:05 +0000 (12:52 -0700)]
Merge tag 'regulator-fix-v6.5-rc3' of git://git./linux/kernel/git/broonie/regulator

Pull regulator fixes from Mark Brown:
 "A couple of small fixes for the the mt6358 driver, fixing error
  reporting and a bootstrapping issue"

* tag 'regulator-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: mt6358: Fix incorrect VCN33 sync error message
  regulator: mt6358: Sync VCN33_* enable status after checking ID

15 months agoMerge tag 'usb-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 30 Jul 2023 18:57:51 +0000 (11:57 -0700)]
Merge tag 'usb-6.5-rc4' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are a set of USB driver fixes for 6.5-rc4. Include in here are:

   - new USB serial device ids

   - dwc3 driver fixes for reported issues

   - typec driver fixes for reported problems

   - gadget driver fixes

   - reverts of some problematic USB changes that went into -rc1

  All of these have been in linux-next with no reported problems"

* tag 'usb-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (24 commits)
  usb: misc: ehset: fix wrong if condition
  usb: dwc3: pci: skip BYT GPIO lookup table for hardwired phy
  usb: cdns3: fix incorrect calculation of ep_buf_size when more than one config
  usb: gadget: call usb_gadget_check_config() to verify UDC capability
  usb: typec: Use sysfs_emit_at when concatenating the string
  usb: typec: Iterate pds array when showing the pd list
  usb: typec: Set port->pd before adding device for typec_port
  usb: typec: qcom: fix return value check in qcom_pmic_typec_probe()
  Revert "usb: gadget: tegra-xudc: Fix error check in tegra_xudc_powerdomain_init()"
  Revert "usb: xhci: tegra: Fix error check"
  USB: gadget: Fix the memory leak in raw_gadget driver
  usb: gadget: core: remove unbalanced mutex_unlock in usb_gadget_activate
  Revert "usb: dwc3: core: Enable AutoRetry feature in the controller"
  Revert "xhci: add quirk for host controllers that don't update endpoint DCS"
  USB: quirks: add quirk for Focusrite Scarlett
  usb: xhci-mtk: set the dma max_seg_size
  MAINTAINERS: drop invalid usb/cdns3 Reviewer e-mail
  usb: dwc3: don't reset device side if dwc3 was configured as host-only
  usb: typec: ucsi: move typec_set_mode(TYPEC_STATE_SAFE) to ucsi_unregister_partner()
  usb: ohci-at91: Fix the unhandle interrupt when resume
  ...

15 months agoMerge tag 'tty-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 30 Jul 2023 18:51:36 +0000 (11:51 -0700)]
Merge tag 'tty-6.5-rc4' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial fixes from Greg KH:
 "Here are some small TTY and serial driver fixes for 6.5-rc4 for some
  reported problems. Included in here is:

   - TIOCSTI fix for braille readers

   - documentation fix for minor numbers

   - MAINTAINERS update for new serial files in -rc1

   - minor serial driver fixes for reported problems

  All of these have been in linux-next with no reported problems"

* tag 'tty-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  serial: 8250_dw: Preserve original value of DLF register
  tty: serial: sh-sci: Fix sleeping in atomic context
  serial: sifive: Fix sifive_serial_console_setup() section
  Documentation: devices.txt: reconcile serial/ucc_uart minor numers
  MAINTAINERS: Update TTY layer for lists and recently added files
  tty: n_gsm: fix UAF in gsm_cleanup_mux
  TIOCSTI: always enable for CAP_SYS_ADMIN

15 months agoMerge tag 'staging-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 30 Jul 2023 18:47:56 +0000 (11:47 -0700)]
Merge tag 'staging-6.5-rc4' of git://git./linux/kernel/git/gregkh/staging

Pull staging driver fixes from Greg KH:
 "Here are three small staging driver fixes for 6.5-rc4 that resolve
  some reported problems. These fixes are:

   - fix for an old bug in the r8712 driver

   - fbtft driver fix for a spi device

   - potential overflow fix in the ks7010 driver

  All of these have been in linux-next with no reported problems"

* tag 'staging-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: ks7010: potential buffer overflow in ks_wlan_set_encode_ext()
  staging: fbtft: ili9341: use macro FBTFT_REGISTER_SPI_DRIVER
  staging: r8712: Fix memory leak in _r8712_init_xmit_priv()

15 months agoMerge tag 'char-misc-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 30 Jul 2023 18:44:00 +0000 (11:44 -0700)]
Merge tag 'char-misc-6.5-rc4' of git://git./linux/kernel/git/gregkh/char-misc

Pull char driver and Documentation fixes from Greg KH:
 "Here is a char driver fix and some documentation updates for 6.5-rc4
  that contain the following changes:

   - sram/genalloc bugfix for reported problem

   - security-bugs.rst update based on recent discussions

   - embargoed-hardware-issues minor cleanups and then partial revert
     for the project/company lists

  All of these have been in linux-next for a while with no reported
  problems, and the documentation updates have all been reviewed by the
  relevant developers"

* tag 'char-misc-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  misc/genalloc: Name subpools by of_node_full_name()
  Documentation: embargoed-hardware-issues.rst: add AMD to the list
  Documentation: embargoed-hardware-issues.rst: clean out empty and unused entries
  Documentation: security-bugs.rst: clarify CVE handling
  Documentation: security-bugs.rst: update preferences when dealing with the linux-distros group

15 months agoMerge tag 'probes-fixes-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 30 Jul 2023 18:27:22 +0000 (11:27 -0700)]
Merge tag 'probes-fixes-v6.5-rc3' of git://git./linux/kernel/git/trace/linux-trace

Pull probe fixes from Masami Hiramatsu:

 - probe-events: add NULL check for some BTF API calls which can return
   error code and NULL.

 - ftrace selftests: check fprobe and kprobe event correctly. This fixes
   a miss condition of the test command.

 - kprobes: do not allow probing functions that start with "__cfi_" or
   "__pfx_" since those are auto generated for kernel CFI and not
   executed.

* tag 'probes-fixes-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  kprobes: Prohibit probing on CFI preamble symbol
  selftests/ftrace: Fix to check fprobe event eneblement
  tracing/probes: Fix to add NULL check for BTF APIs

15 months agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 30 Jul 2023 18:19:08 +0000 (11:19 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:
 "x86:

   - Do not register IRQ bypass consumer if posted interrupts not
     supported

   - Fix missed device interrupt due to non-atomic update of IRR

   - Use GFP_KERNEL_ACCOUNT for pid_table in ipiv

   - Make VMREAD error path play nice with noinstr

   - x86: Acquire SRCU read lock when handling fastpath MSR writes

   - Support linking rseq tests statically against glibc 2.35+

   - Fix reference count for stats file descriptors

   - Detect userspace setting invalid CR0

  Non-KVM:

   - Remove coccinelle script that has caused multiple confusion
     ("debugfs, coccinelle: check for obsolete DEFINE_SIMPLE_ATTRIBUTE()
     usage", acked by Greg)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (21 commits)
  KVM: selftests: Expand x86's sregs test to cover illegal CR0 values
  KVM: VMX: Don't fudge CR0 and CR4 for restricted L2 guest
  KVM: x86: Disallow KVM_SET_SREGS{2} if incoming CR0 is invalid
  Revert "debugfs, coccinelle: check for obsolete DEFINE_SIMPLE_ATTRIBUTE() usage"
  KVM: selftests: Verify stats fd is usable after VM fd has been closed
  KVM: selftests: Verify stats fd can be dup()'d and read
  KVM: selftests: Verify userspace can create "redundant" binary stats files
  KVM: selftests: Explicitly free vcpus array in binary stats test
  KVM: selftests: Clean up stats fd in common stats_test() helper
  KVM: selftests: Use pread() to read binary stats header
  KVM: Grab a reference to KVM for VM and vCPU stats file descriptors
  selftests/rseq: Play nice with binaries statically linked against glibc 2.35+
  Revert "KVM: SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid"
  KVM: x86: Acquire SRCU read lock when handling fastpath MSR writes
  KVM: VMX: Use vmread_error() to report VM-Fail in "goto" path
  KVM: VMX: Make VMREAD error path play nice with noinstr
  KVM: x86/irq: Conditionally register IRQ bypass consumer again
  KVM: X86: Use GFP_KERNEL_ACCOUNT for pid_table in ipiv
  KVM: x86: check the kvm_cpu_get_interrupt result before using it
  KVM: x86: VMX: set irr_pending in kvm_apic_update_irr
  ...

15 months agoMerge tag 'locking_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 30 Jul 2023 18:12:32 +0000 (11:12 -0700)]
Merge tag 'locking_urgent_for_v6.5_rc4' of git://git./linux/kernel/git/tip/tip

Pull locking fix from Borislav Petkov:

 - Fix a rtmutex race condition resulting from sharing of the sort key
   between the lock waiters and the PI chain tree (->pi_waiters) of a
   task by giving each tree their own sort key

* tag 'locking_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/rtmutex: Fix task->pi_waiters integrity

15 months agoMerge tag 'x86_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 30 Jul 2023 18:05:35 +0000 (11:05 -0700)]
Merge tag 'x86_urgent_for_v6.5_rc4' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:

 - AMD's automatic IBRS doesn't enable cross-thread branch target
   injection protection (STIBP) for user processes. Enable STIBP on such
   systems.

 - Do not delete (but put the ref instead) of AMD MCE error thresholding
   sysfs kobjects when destroying them in order not to delete the kernfs
   pointer prematurely

 - Restore annotation in ret_from_fork_asm() in order to fix kthread
   stack unwinding from being marked as unreliable and thus breaking
   livepatching

* tag 'x86_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/cpu: Enable STIBP on AMD if Automatic IBRS is enabled
  x86/MCE/AMD: Decrement threshold_bank refcount when removing threshold blocks
  x86: Fix kthread unwind

15 months agoMerge tag 'irq_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 30 Jul 2023 17:59:19 +0000 (10:59 -0700)]
Merge tag 'irq_urgent_for_v6.5_rc4' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Borislav Petkov:

 - Work around an erratum on GIC700, where a race between a CPU handling
   a wake-up interrupt, a change of affinity, and another CPU going to
   sleep can result in a lack of wake-up event on the next interrupt

 - Fix the locking required on a VPE for GICv4

 - Enable Rockchip 3588001 erratum workaround for RK3588S

 - Fix the irq-bcm6345-l1 assumtions of the boot CPU always be the first
   CPU in the system

* tag 'irq_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3: Workaround for GIC-700 erratum 2941627
  irqchip/gic-v3: Enable Rockchip 3588001 erratum workaround for RK3588S
  irqchip/gic-v4.1: Properly lock VPEs when doing a directLPI invalidation
  irq-bcm6345-l1: Do not assume a fixed block to cpu mapping

15 months agoMerge tag '6.5-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 30 Jul 2023 03:49:13 +0000 (20:49 -0700)]
Merge tag '6.5-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fixes from Steve French:
 "Four small SMB3 client fixes:

   - two reconnect fixes (to address the case where non-default
     iocharset gets incorrectly overridden at reconnect with the
     default charset)

   - fix for NTLMSSP_AUTH request setting a flag incorrectly)

   - Add missing check for invalid tlink (tree connection) in ioctl"

* tag '6.5-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: add missing return value check for cifs_sb_tlink
  smb3: do not set NTLMSSP_VERSION flag for negotiate not auth request
  cifs: fix charset issue in reconnection
  fs/nls: make load_nls() take a const parameter

15 months agoMerge tag 'trace-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Sun, 30 Jul 2023 03:40:43 +0000 (20:40 -0700)]
Merge tag 'trace-v6.5-rc3' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Fix to /sys/kernel/tracing/per_cpu/cpu*/stats read and entries.

   If a resize shrinks the buffer it clears the read count to notify
   readers that they need to reset. But the read count is also used for
   accounting and this causes the numbers to be off. Instead, create a
   separate variable to use to notify readers to reset.

 - Fix the ref counts of the "soft disable" mode. The wrong value was
   used for testing if soft disable mode should be enabled or disable,
   but instead, just change the logic to do the enable and disable in
   place when the SOFT_MODE is set or cleared.

 - Several kernel-doc fixes

 - Removal of unused external declarations

* tag 'trace-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing: Fix warning in trace_buffered_event_disable()
  ftrace: Remove unused extern declarations
  tracing: Fix kernel-doc warnings in trace_seq.c
  tracing: Fix kernel-doc warnings in trace_events_trigger.c
  tracing/synthetic: Fix kernel-doc warnings in trace_events_synth.c
  ring-buffer: Fix kernel-doc warnings in ring_buffer.c
  ring-buffer: Fix wrong stat of cpu_buffer->read

15 months agoarch/*/configs/*defconfig: Replace AUTOFS4_FS by AUTOFS_FS
Sven Joachim [Thu, 27 Jul 2023 20:00:41 +0000 (22:00 +0200)]
arch/*/configs/*defconfig: Replace AUTOFS4_FS by AUTOFS_FS

Commit a2225d931f75 ("autofs: remove left-over autofs4 stubs")
promised the removal of the fs/autofs/Kconfig fragment for AUTOFS4_FS
within a couple of releases, but five years later this still has not
happened yet, and AUTOFS4_FS is still enabled in 63 defconfigs.

Get rid of it mechanically:

   git grep -l CONFIG_AUTOFS4_FS -- '*defconfig' |
       xargs sed -i 's/AUTOFS4_FS/AUTOFS_FS/'

Also just remove the AUTOFS4_FS config option stub.  Anybody who hasn't
regenerated their config file in the last five years will need to just
get the new name right when they do.

Signed-off-by: Sven Joachim <svenjoac@gmx.de>
Acked-by: Ian Kent <raven@themaw.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 months agoMerge tag 'loongarch-fixes-6.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 29 Jul 2023 15:59:25 +0000 (08:59 -0700)]
Merge tag 'loongarch-fixes-6.5-1' of git://git./linux/kernel/git/chenhuacai/linux-loongson

Pull LoongArch fixes from Huacai Chen:
 "Some bug fixes for build system, builtin cmdline handling, bpf and
  {copy, clear}_user, together with a trivial cleanup"

* tag 'loongarch-fixes-6.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
  LoongArch: Cleanup __builtin_constant_p() checking for cpu_has_*
  LoongArch: BPF: Fix check condition to call lu32id in move_imm()
  LoongArch: BPF: Enable bpf_probe_read{, str}() on LoongArch
  LoongArch: Fix return value underflow in exception path
  LoongArch: Fix CMDLINE_EXTEND and CMDLINE_BOOTLOADER handling
  LoongArch: Fix module relocation error with binutils 2.41
  LoongArch: Only fiddle with CHECKFLAGS if `need-compiler'

15 months agoKVM: selftests: Expand x86's sregs test to cover illegal CR0 values
Sean Christopherson [Tue, 13 Jun 2023 20:30:37 +0000 (13:30 -0700)]
KVM: selftests: Expand x86's sregs test to cover illegal CR0 values

Add coverage to x86's set_sregs_test to verify KVM rejects vendor-agnostic
illegal CR0 values, i.e. CR0 values whose legality doesn't depend on the
current VMX mode.  KVM historically has neglected to reject bad CR0s from
userspace, i.e. would happily accept a completely bogus CR0 via
KVM_SET_SREGS{2}.

Punt VMX specific subtests to future work, as they would require quite a
bit more effort, and KVM gets coverage for CR0 checks in general through
other means, e.g. KVM-Unit-Tests.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230613203037.1968489-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: VMX: Don't fudge CR0 and CR4 for restricted L2 guest
Sean Christopherson [Tue, 13 Jun 2023 20:30:36 +0000 (13:30 -0700)]
KVM: VMX: Don't fudge CR0 and CR4 for restricted L2 guest

Stuff CR0 and/or CR4 to be compliant with a restricted guest if and only
if KVM itself is not configured to utilize unrestricted guests, i.e. don't
stuff CR0/CR4 for a restricted L2 that is running as the guest of an
unrestricted L1.  Any attempt to VM-Enter a restricted guest with invalid
CR0/CR4 values should fail, i.e. in a nested scenario, KVM (as L0) should
never observe a restricted L2 with incompatible CR0/CR4, since nested
VM-Enter from L1 should have failed.

And if KVM does observe an active, restricted L2 with incompatible state,
e.g. due to a KVM bug, fudging CR0/CR4 instead of letting VM-Enter fail
does more harm than good, as KVM will often neglect to undo the side
effects, e.g. won't clear rmode.vm86_active on nested VM-Exit, and thus
the damage can easily spill over to L1.  On the other hand, letting
VM-Enter fail due to bad guest state is more likely to contain the damage
to L2 as KVM relies on hardware to perform most guest state consistency
checks, i.e. KVM needs to be able to reflect a failed nested VM-Enter into
L1 irrespective of (un)restricted guest behavior.

Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Fixes: bddd82d19e2e ("KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230613203037.1968489-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: x86: Disallow KVM_SET_SREGS{2} if incoming CR0 is invalid
Sean Christopherson [Tue, 13 Jun 2023 20:30:35 +0000 (13:30 -0700)]
KVM: x86: Disallow KVM_SET_SREGS{2} if incoming CR0 is invalid

Reject KVM_SET_SREGS{2} with -EINVAL if the incoming CR0 is invalid,
e.g. due to setting bits 63:32, illegal combinations, or to a value that
isn't allowed in VMX (non-)root mode.  The VMX checks in particular are
"fun" as failure to disallow Real Mode for an L2 that is configured with
unrestricted guest disabled, when KVM itself has unrestricted guest
enabled, will result in KVM forcing VM86 mode to virtual Real Mode for
L2, but then fail to unwind the related metadata when synthesizing a
nested VM-Exit back to L1 (which has unrestricted guest enabled).

Opportunistically fix a benign typo in the prototype for is_valid_cr4().

Cc: stable@vger.kernel.org
Reported-by: syzbot+5feef0b9ee9c8e9e5689@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/000000000000f316b705fdf6e2b4@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230613203037.1968489-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoRevert "debugfs, coccinelle: check for obsolete DEFINE_SIMPLE_ATTRIBUTE() usage"
Sean Christopherson [Wed, 26 Jul 2023 20:29:20 +0000 (13:29 -0700)]
Revert "debugfs, coccinelle: check for obsolete DEFINE_SIMPLE_ATTRIBUTE() usage"

Remove coccinelle's recommendation to use DEFINE_DEBUGFS_ATTRIBUTE()
instead of DEFINE_SIMPLE_ATTRIBUTE().  Regardless of whether or not the
"significant overhead" incurred by debugfs_create_file() is actually
meaningful, warnings from the script have led to a rash of low-quality
patches that have sowed confusion and consumed maintainer time for little
to no benefit.  There have been no less than four attempts to "fix" KVM,
and a quick search on lore shows that KVM is not alone.

This reverts commit 5103068eaca290f890a30aae70085fac44cecaf6.

Link: https://lore.kernel.org/all/87tu2nbnz3.fsf@mpe.ellerman.id.au
Link: https://lore.kernel.org/all/c0b98151-16b6-6d8f-1765-0f7d46682d60@redhat.com
Link: https://lkml.kernel.org/r/20230706072954.4881-1-duminjie%40vivo.com
Link: https://lore.kernel.org/all/Y2FsbufV00jbyF0B@google.com
Link: https://lore.kernel.org/all/Y2ENJJ1YiSg5oHiy@orome
Link: https://lore.kernel.org/all/7560b350e7b23786ce712118a9a504356ff1cca4.camel@kernel.org
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230726202920.507756-1-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: selftests: Verify stats fd is usable after VM fd has been closed
Sean Christopherson [Tue, 11 Jul 2023 23:01:31 +0000 (16:01 -0700)]
KVM: selftests: Verify stats fd is usable after VM fd has been closed

Verify that VM and vCPU binary stats files are usable even after userspace
has put its last direct reference to the VM.  This is a regression test
for a UAF bug where KVM didn't gift the stats files a reference to the VM.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-8-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: selftests: Verify stats fd can be dup()'d and read
Sean Christopherson [Tue, 11 Jul 2023 23:01:30 +0000 (16:01 -0700)]
KVM: selftests: Verify stats fd can be dup()'d and read

Expand the binary stats test to verify that a stats fd can be dup()'d
and read, to (very) roughly simulate userspace passing around the file.
Adding the dup() test is primarily an intermediate step towards verifying
that userspace can read VM/vCPU stats before _and_ after userspace closes
its copy of the VM fd; the dup() test itself is only mildly interesting.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-7-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: selftests: Verify userspace can create "redundant" binary stats files
Sean Christopherson [Tue, 11 Jul 2023 23:01:29 +0000 (16:01 -0700)]
KVM: selftests: Verify userspace can create "redundant" binary stats files

Verify that KVM doesn't artificially limit KVM_GET_STATS_FD to a single
file per VM/vCPU.  There's no known use case for getting multiple stats
fds, but it should work, and more importantly creating multiple files will
make it easier to test that KVM correct manages VM refcounts for stats
files.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-6-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: selftests: Explicitly free vcpus array in binary stats test
Sean Christopherson [Tue, 11 Jul 2023 23:01:28 +0000 (16:01 -0700)]
KVM: selftests: Explicitly free vcpus array in binary stats test

Explicitly free the all-encompassing vcpus array in the binary stats test
so that the test is consistent with respect to freeing all dynamically
allocated resources (versus letting them be freed on exit).

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-5-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: selftests: Clean up stats fd in common stats_test() helper
Sean Christopherson [Tue, 11 Jul 2023 23:01:27 +0000 (16:01 -0700)]
KVM: selftests: Clean up stats fd in common stats_test() helper

Move the stats fd cleanup code into stats_test() and drop the
superfluous vm_stats_test() and vcpu_stats_test() helpers in order to
decouple creation of the stats file from consuming/testing the file
(deduping code is a bonus).  This will make it easier to test various
edge cases related to stats, e.g. that userspace can dup() a stats fd,
that userspace can have multiple stats files for a singleVM/vCPU, etc.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: selftests: Use pread() to read binary stats header
Sean Christopherson [Tue, 11 Jul 2023 23:01:26 +0000 (16:01 -0700)]
KVM: selftests: Use pread() to read binary stats header

Use pread() with an explicit offset when reading the header and the header
name for a binary stats fd so that the common helper and the binary stats
test don't subtly rely on the file effectively being untouched, e.g. to
allow multiple reads of the header, name, etc.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: Grab a reference to KVM for VM and vCPU stats file descriptors
Sean Christopherson [Tue, 11 Jul 2023 23:01:25 +0000 (16:01 -0700)]
KVM: Grab a reference to KVM for VM and vCPU stats file descriptors

Grab a reference to KVM prior to installing VM and vCPU stats file
descriptors to ensure the underlying VM and vCPU objects are not freed
until the last reference to any and all stats fds are dropped.

Note, the stats paths manually invoke fd_install() and so don't need to
grab a reference before creating the file.

Fixes: ce55c049459c ("KVM: stats: Support binary stats retrieval for a VCPU")
Fixes: fcfe1baeddbf ("KVM: stats: Support binary stats retrieval for a VM")
Reported-by: Zheng Zhang <zheng.zhang@email.ucr.edu>
Closes: https://lore.kernel.org/all/CAC_GQSr3xzZaeZt85k_RCBd5kfiOve8qXo7a81Cq53LuVQ5r=Q@mail.gmail.com
Cc: stable@vger.kernel.org
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Message-Id: <20230711230131.648752-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoselftests/rseq: Play nice with binaries statically linked against glibc 2.35+
Sean Christopherson [Fri, 21 Jul 2023 22:33:52 +0000 (15:33 -0700)]
selftests/rseq: Play nice with binaries statically linked against glibc 2.35+

To allow running rseq and KVM's rseq selftests as statically linked
binaries, initialize the various "trampoline" pointers to point directly
at the expect glibc symbols, and skip the dlysm() lookups if the rseq
size is non-zero, i.e. the binary is statically linked *and* the libc
registered its own rseq.

Define weak versions of the symbols so as not to break linking against
libc versions that don't support rseq in any capacity.

The KVM selftests in particular are often statically linked so that they
can be run on targets with very limited runtime environments, i.e. test
machines.

Fixes: 233e667e1ae3 ("selftests/rseq: Uplift rseq selftests for compatibility with glibc-2.35")
Cc: Aaron Lewis <aaronlewis@google.com>
Cc: kvm@vger.kernel.org
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230721223352.2333911-1-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoRevert "KVM: SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid"
Sean Christopherson [Fri, 21 Jul 2023 22:43:37 +0000 (15:43 -0700)]
Revert "KVM: SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid"

Now that handle_fastpath_set_msr_irqoff() acquires kvm->srcu, i.e. allows
dereferencing memslots during WRMSR emulation, drop the requirement that
"next RIP" is valid.  In hindsight, acquiring kvm->srcu would have been a
better fix than avoiding the pastpath, but at the time it was thought that
accessing SRCU-protected data in the fastpath was a one-off edge case.

This reverts commit 5c30e8101e8d5d020b1d7119117889756a6ed713.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230721224337.2335137-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: x86: Acquire SRCU read lock when handling fastpath MSR writes
Sean Christopherson [Fri, 21 Jul 2023 22:43:36 +0000 (15:43 -0700)]
KVM: x86: Acquire SRCU read lock when handling fastpath MSR writes

Temporarily acquire kvm->srcu for read when potentially emulating WRMSR in
the VM-Exit fastpath handler, as several of the common helpers used during
emulation expect the caller to provide SRCU protection.  E.g. if the guest
is counting instructions retired, KVM will query the PMU event filter when
stepping over the WRMSR.

  dump_stack+0x85/0xdf
  lockdep_rcu_suspicious+0x109/0x120
  pmc_event_is_allowed+0x165/0x170
  kvm_pmu_trigger_event+0xa5/0x190
  handle_fastpath_set_msr_irqoff+0xca/0x1e0
  svm_vcpu_run+0x5c3/0x7b0 [kvm_amd]
  vcpu_enter_guest+0x2108/0x2580

Alternatively, check_pmu_event_filter() could acquire kvm->srcu, but this
isn't the first bug of this nature, e.g. see commit 5c30e8101e8d ("KVM:
SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid").  Providing
protection for the entirety of WRMSR emulation will allow reverting the
aforementioned commit, and will avoid having to play whack-a-mole when new
uses of SRCU-protected structures are inevitably added in common emulation
helpers.

Fixes: dfdeda67ea2d ("KVM: x86/pmu: Prevent the PMU from counting disallowed events")
Reported-by: Greg Thelen <gthelen@google.com>
Reported-by: Aaron Lewis <aaronlewis@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230721224337.2335137-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: VMX: Use vmread_error() to report VM-Fail in "goto" path
Sean Christopherson [Fri, 21 Jul 2023 23:56:37 +0000 (16:56 -0700)]
KVM: VMX: Use vmread_error() to report VM-Fail in "goto" path

Use vmread_error() to report VM-Fail on VMREAD for the "asm goto" case,
now that trampoline case has yet another wrapper around vmread_error() to
play nice with instrumentation.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230721235637.2345403-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: VMX: Make VMREAD error path play nice with noinstr
Sean Christopherson [Fri, 21 Jul 2023 23:56:36 +0000 (16:56 -0700)]
KVM: VMX: Make VMREAD error path play nice with noinstr

Mark vmread_error_trampoline() as noinstr, and add a second trampoline
for the CONFIG_CC_HAS_ASM_GOTO_OUTPUT=n case to enable instrumentation
when handling VM-Fail on VMREAD.  VMREAD is used in various noinstr
flows, e.g. immediately after VM-Exit, and objtool rightly complains that
the call to the error trampoline leaves a no-instrumentation section
without annotating that it's safe to do so.

  vmlinux.o: warning: objtool: vmx_vcpu_enter_exit+0xc9:
  call to vmread_error_trampoline() leaves .noinstr.text section

Note, strictly speaking, enabling instrumentation in the VM-Fail path
isn't exactly safe, but if VMREAD fails the kernel/system is likely hosed
anyways, and logging that there is a fatal error is more important than
*maybe* encountering slightly unsafe instrumentation.

Reported-by: Su Hui <suhui@nfschina.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230721235637.2345403-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: x86/irq: Conditionally register IRQ bypass consumer again
Like Xu [Mon, 24 Jul 2023 11:12:36 +0000 (19:12 +0800)]
KVM: x86/irq: Conditionally register IRQ bypass consumer again

As was attempted commit 14717e203186 ("kvm: Conditionally register IRQ
bypass consumer"): "if we don't support a mechanism for bypassing IRQs,
don't register as a consumer.  Initially this applied to AMD processors,
but when AVIC support was implemented for assigned devices,
kvm_arch_has_irq_bypass() was always returning true.

We can still skip registering the consumer where enable_apicv
or posted-interrupts capability is unsupported or globally disabled.
This eliminates meaningless dev_info()s when the connect fails
between producer and consumer", such as on Linux hosts where enable_apicv
or posted-interrupts capability is unsupported or globally disabled.

Cc: Alex Williamson <alex.williamson@redhat.com>
Reported-by: Yong He <alexyonghe@tencent.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217379
Signed-off-by: Like Xu <likexu@tencent.com>
Message-Id: <20230724111236.76570-1-likexu@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: X86: Use GFP_KERNEL_ACCOUNT for pid_table in ipiv
Peng Hao [Fri, 28 Jul 2023 06:49:48 +0000 (14:49 +0800)]
KVM: X86: Use GFP_KERNEL_ACCOUNT for pid_table in ipiv

The pid_table of ipiv is the persistent memory allocated by
per-vcpu, which should be counted into the memory cgroup.

Signed-off-by: Peng Hao <flyingpeng@tencent.com>
Message-Id: <CAPm50aLxCQ3TQP2Lhc0PX3y00iTRg+mniLBqNDOC=t9CLxMwwA@mail.gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: x86: check the kvm_cpu_get_interrupt result before using it
Maxim Levitsky [Wed, 26 Jul 2023 13:59:45 +0000 (16:59 +0300)]
KVM: x86: check the kvm_cpu_get_interrupt result before using it

The code was blindly assuming that kvm_cpu_get_interrupt never returns -1
when there is a pending interrupt.

While this should be true, a bug in KVM can still cause this.

If -1 is returned, the code before this patch was converting it to 0xFF,
and 0xFF interrupt was injected to the guest, which results in an issue
which was hard to debug.

Add WARN_ON_ONCE to catch this case and skip the injection
if this happens again.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20230726135945.260841-4-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: x86: VMX: set irr_pending in kvm_apic_update_irr
Maxim Levitsky [Wed, 26 Jul 2023 13:59:44 +0000 (16:59 +0300)]
KVM: x86: VMX: set irr_pending in kvm_apic_update_irr

When the APICv is inhibited, the irr_pending optimization is used.

Therefore, when kvm_apic_update_irr sets bits in the IRR,
it must set irr_pending to true as well.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20230726135945.260841-3-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agoKVM: x86: VMX: __kvm_apic_update_irr must update the IRR atomically
Maxim Levitsky [Wed, 26 Jul 2023 13:59:43 +0000 (16:59 +0300)]
KVM: x86: VMX: __kvm_apic_update_irr must update the IRR atomically

If APICv is inhibited, then IPIs from peer vCPUs are done by
atomically setting bits in IRR.

This means, that when __kvm_apic_update_irr copies PIR to IRR,
it has to modify IRR atomically as well.

Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20230726135945.260841-2-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
15 months agokprobes: Prohibit probing on CFI preamble symbol
Masami Hiramatsu (Google) [Tue, 11 Jul 2023 01:50:47 +0000 (10:50 +0900)]
kprobes: Prohibit probing on CFI preamble symbol

Do not allow to probe on "__cfi_" or "__pfx_" started symbol, because those
are used for CFI and not executed. Probing it will break the CFI.

Link: https://lore.kernel.org/all/168904024679.116016.18089228029322008512.stgit@devnote2/
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Steven Rostedt (Google) <rostedt@goodmis.org>