userfaultfd: add user-mode only option to unprivileged_userfaultfd sysctl knob

With this change, when the knob is set to 0, it allows unprivileged users
to call userfaultfd, like when it is set to 1, but with the restriction
that page faults from only user-mode can be handled.  In this mode, an
unprivileged user (without SYS_CAP_PTRACE capability) must pass
UFFD_USER_MODE_ONLY to userfaultd or the API will fail with EPERM.

This enables administrators to reduce the likelihood that an attacker with
access to userfaultfd can delay faulting kernel code to widen timing
windows for other exploits.

The default value of this knob is changed to 0.  This is required for
correct functioning of pipe mutex.  However, this will fail postcopy live
migration, which will be unnoticeable to the VM guests.  To avoid this,
set 'vm.userfault = 1' in /sys/sysctl.conf.

The main reason this change is desirable as in the short term is that the
Android userland will behave as with the sysctl set to zero.  So without
this commit, any Linux binary using userfaultfd to manage its memory would
behave differently if run within the Android userland.  For more details,
refer to Andrea's reply [1].

[1] https://lore.kernel.org/lkml/20200904033438.GI9411@redhat.com/

Link: https://lkml.kernel.org/r/20201120030411.2690816-3-lokeshgidra@google.com
Signed-off-by: Lokesh Gidra <lokeshgidra@google.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Stephen Smalley <stephen.smalley.work@gmail.com>
Cc: Eric Biggers <ebiggers@kernel.org>
Cc: Daniel Colascione <dancol@dancol.org>
Cc: "Joel Fernandes (Google)" <joel@joelfernandes.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Jeff Vander Stoep <jeffv@google.com>
Cc: <calin@google.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Shaohua Li <shli@fb.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Nitin Gupta <nigupta@nvidia.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Iurii Zaikin <yzaikin@google.com>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Daniel Colascione <dancol@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

userfaultfd: add UFFD_USER_MODE_ONLY

Patch series "Control over userfaultfd kernel-fault handling", v6.

This patch series is split from [1].  The other series enables SELinux
support for userfaultfd file descriptors so that its creation and movement
can be controlled.

It has been demonstrated on various occasions that suspending kernel code
execution for an arbitrary amount of time at any access to userspace
memory (copy_from_user()/copy_to_user()/...) can be exploited to change
the intended behavior of the kernel.  For instance, handling page faults
in kernel-mode using userfaultfd has been exploited in [2, 3].  Likewise,
FUSE, which is similar to userfaultfd in this respect, has been exploited
in [4, 5] for similar outcome.

This small patch series adds a new flag to userfaultfd(2) that allows
callers to give up the ability to handle kernel-mode faults with the
resulting UFFD file object.  It then adds a 'user-mode only' option to the
unprivileged_userfaultfd sysctl knob to require unprivileged callers to
use this new flag.

The purpose of this new interface is to decrease the chance of an
unprivileged userfaultfd user taking advantage of userfaultfd to enhance
security vulnerabilities by lengthening the race window in kernel code.

[1] https://lore.kernel.org/lkml/20200211225547.235083-1-dancol@google.com/
[2] https://duasynt.com/blog/linux-kernel-heap-spray
[3] https://duasynt.com/blog/cve-2016-6187-heap-off-by-one-exploit
[4] https://googleprojectzero.blogspot.com/2016/06/exploiting-recursion-in-linux-kernel_20.html
[5] https://bugs.chromium.org/p/project-zero/issues/detail?id=808

This patch (of 2):

userfaultfd handles page faults from both user and kernel code.  Add a new
UFFD_USER_MODE_ONLY flag for userfaultfd(2) that makes the resulting
userfaultfd object refuse to handle faults from kernel mode, treating
these faults as if SIGBUS were always raised, causing the kernel code to
fail with EFAULT.

A future patch adds a knob allowing administrators to give some processes
the ability to create userfaultfd file objects only if they pass
UFFD_USER_MODE_ONLY, reducing the likelihood that these processes will
exploit userfaultfd's ability to delay kernel page faults to open timing
windows for future exploits.

Link: https://lkml.kernel.org/r/20201120030411.2690816-1-lokeshgidra@google.com
Link: https://lkml.kernel.org/r/20201120030411.2690816-2-lokeshgidra@google.com
Signed-off-by: Daniel Colascione <dancol@google.com>
Signed-off-by: Lokesh Gidra <lokeshgidra@google.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: <calin@google.com>
Cc: Daniel Colascione <dancol@dancol.org>
Cc: Eric Biggers <ebiggers@kernel.org>
Cc: Iurii Zaikin <yzaikin@google.com>
Cc: Jeff Vander Stoep <jeffv@google.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: "Joel Fernandes (Google)" <joel@joelfernandes.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Nitin Gupta <nigupta@nvidia.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Shaohua Li <shli@fb.com>
Cc: Stephen Smalley <stephen.smalley.work@gmail.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_poison: remove CONFIG_PAGE_POISONING_ZERO

CONFIG_PAGE_POISONING_ZERO uses the zero pattern instead of 0xAA.  It was
introduced by commit 1414c7f4f7d7 ("mm/page_poisoning.c: allow for zero
poisoning"), noting that using zeroes retains the benefit of sanitizing
content of freed pages, with the benefit of not having to zero them again
on alloc, and the downside of making some forms of corruption (stray
writes of NULLs) harder to detect than with the 0xAA pattern.  Together
with CONFIG_PAGE_POISONING_NO_SANITY it made possible to sanitize the
contents on free without checking it back on alloc.

These days we have the init_on_free() option to achieve sanitization with
zeroes and to save clearing on alloc (and without checking on alloc).
Arguably if someone does choose to check the poison for corruption on
alloc, the savings of not clearing the page are secondary, and it makes
sense to always use the 0xAA poison pattern.  Thus, remove the
CONFIG_PAGE_POISONING_ZERO option for being redundant.

Link: https://lkml.kernel.org/r/20201113104033.22907-6-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Laura Abbott <labbott@kernel.org>
Cc: Mateusz Nosek <mateusznosek0@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_poison: remove CONFIG_PAGE_POISONING_NO_SANITY

CONFIG_PAGE_POISONING_NO_SANITY skips the check on page alloc whether the
poison pattern was corrupted, suggesting a use-after-free.  The motivation
to introduce it in commit 8823b1dbc05f ("mm/page_poison.c: enable
PAGE_POISONING as a separate option") was to simply sanitize freed pages,
optimally together with CONFIG_PAGE_POISONING_ZERO.

These days we have an init_on_free=1 boot option, which makes this use
case of page poisoning redundant.  For sanitizing, writing zeroes is
sufficient, there is pretty much no benefit from writing the 0xAA poison
pattern to freed pages, without checking it back on alloc.  Thus, remove
this option and suggest init_on_free instead in the main config's help.

Link: https://lkml.kernel.org/r/20201113104033.22907-5-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Laura Abbott <labbott@kernel.org>
Cc: Mateusz Nosek <mateusznosek0@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/power: allow hibernation with page_poison sanity checking

Page poisoning used to be incompatible with hibernation, as the state of
poisoned pages was lost after resume, thus enabling CONFIG_HIBERNATION
forces CONFIG_PAGE_POISONING_NO_SANITY.  For the same reason, the
poisoning with zeroes variant CONFIG_PAGE_POISONING_ZERO used to disable
hibernation.  The latter restriction was removed by commit 1ad1410f632d
("PM / Hibernate: allow hibernation with PAGE_POISONING_ZERO") and
similarly for init_on_free by commit 18451f9f9e58 ("PM: hibernate: fix
crashes with init_on_free=1") by making sure free pages are cleared after
resume.

We can use the same mechanism to instead poison free pages with
PAGE_POISON after resume.  This covers both zero and 0xAA patterns.  Thus
we can remove the Kconfig restriction that disables page poison sanity
checking when hibernation is enabled.

Link: https://lkml.kernel.org/r/20201113104033.22907-4-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> [hibernation]
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Laura Abbott <labbott@kernel.org>
Cc: Mateusz Nosek <mateusznosek0@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_poison: use static key more efficiently

Commit 11c9c7edae06 ("mm/page_poison.c: replace bool variable with static
key") changed page_poisoning_enabled() to a static key check.  However,
the function is not inlined, so each check still involves a function call
with overhead not eliminated when page poisoning is disabled.

Analogically to how debug_pagealloc is handled, this patch converts
page_poisoning_enabled() back to boolean check, and introduces
page_poisoning_enabled_static() for fast paths.  Both functions are
inlined.

The function kernel_poison_pages() is also called unconditionally and does
the static key check inside.  Remove it from there and put it to callers.
Also split it to two functions kernel_poison_pages() and
kernel_unpoison_pages() instead of the confusing bool parameter.

Also optimize the check that enables page poisoning instead of
debug_pagealloc for architectures without proper debug_pagealloc support.
Move the check to init_mem_debugging_and_hardening() to enable a single
static key instead of having two static branches in
page_poisoning_enabled_static().

Link: https://lkml.kernel.org/r/20201113104033.22907-3-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Laura Abbott <labbott@kernel.org>
Cc: Mateusz Nosek <mateusznosek0@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: do not rely on the order of page_poison and init_on_alloc/free parameters

Patch series "cleanup page poisoning", v3.

I have identified a number of issues and opportunities for cleanup with
CONFIG_PAGE_POISON and friends:

 - interaction with init_on_alloc and init_on_free parameters depends on
   the order of parameters (Patch 1)

 - the boot time enabling uses static key, but inefficienty (Patch 2)

 - sanity checking is incompatible with hibernation (Patch 3)

 - CONFIG_PAGE_POISONING_NO_SANITY can be removed now that we have
   init_on_free (Patch 4)

 - CONFIG_PAGE_POISONING_ZERO can be most likely removed now that we
   have init_on_free (Patch 5)

This patch (of 5):

Enabling page_poison=1 together with init_on_alloc=1 or init_on_free=1
produces a warning in dmesg that page_poison takes precedence.  However,
as these warnings are printed in early_param handlers for
init_on_alloc/free, they are not printed if page_poison is enabled later
on the command line (handlers are called in the order of their
parameters), or when init_on_alloc/free is always enabled by the
respective config option - before the page_poison early param handler is
called, it is not considered to be enabled.  This is inconsistent.

We can remove the dependency on order by making the init_on_* parameters
only set a boolean variable, and postponing the evaluation after all early
params have been processed.  Introduce a new
init_mem_debugging_and_hardening() function for that, and move the related
debug_pagealloc processing there as well.

As a result init_mem_debugging_and_hardening() knows always accurately if
init_on_* and/or page_poison options were enabled.  Thus we can also
optimize want_init_on_alloc() and want_init_on_free().  We don't need to
check page_poisoning_enabled() there, we can instead not enable the
init_on_* static keys at all, if page poisoning is enabled.  This results
in a simpler and more effective code.

Link: https://lkml.kernel.org/r/20201113104033.22907-1-vbabka@suse.cz
Link: https://lkml.kernel.org/r/20201113104033.22907-2-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Mateusz Nosek <mateusznosek0@gmail.com>
Cc: Laura Abbott <labbott@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: cma: improve pr_debug log in cma_release()

It is required to print 'count' of pages, along with the pages, passed to
cma_release to debug the cases of mismatched count value passed between
cma_alloc() and cma_release() from a code path.

As an example, consider the below scenario:

1) CMA pool size is 4MB and

2) User doing the erroneous step of allocating 2 pages but freeing 1
   page in a loop from this CMA pool.  The step 2 causes cma_alloc() to
   return NULL at one point of time because of -ENOMEM condition.

And the current pr_debug logs is not giving the info about these types of
allocation patterns because of count value not being printed in
cma_release().

We are printing the count value in the trace logs, just extend the same to
pr_debug logs too.

[akpm@linux-foundation.org: fix printk warning]

Link: https://lkml.kernel.org/r/1606318341-29521-1-git-send-email-charante@codeaurora.org
Signed-off-by: Charan Teja Reddy <charante@codeaurora.org>
Reviewed-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/cma.c: remove redundant cma_mutex lock

The cma_mutex which protects alloc_contig_range() was first appeared in
commit 7ee793a62fa8c ("cma: Remove potential deadlock situation"), at that
time, there is no guarantee the behavior of concurrency inside
alloc_contig_range().

After commit 2c7452a075d4db2dc ("mm/page_isolation.c: make
start_isolate_page_range() fail if already isolated")

  > However, two subsystems (CMA and gigantic
  > huge pages for example) could attempt operations on the same range.  If
  > this happens, one thread may 'undo' the work another thread is doing.
  > This can result in pageblocks being incorrectly left marked as
  > MIGRATE_ISOLATE and therefore not available for page allocation.

The concurrency inside alloc_contig_range() was clarified.

Now we can find that hugepage and virtio call alloc_contig_range() without
any lock, thus cma_mutex is "redundant" in cma_alloc() now.

Link: https://lkml.kernel.org/r/20201020102241.3729-1-lecopzer.chen@mediatek.com
Signed-off-by: Lecopzer Chen <lecopzer.chen@mediatek.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: YJ Chiang <yj.chiang@mediatek.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: migrate: remove unused parameter in migrate_vma_insert_page()

"dst" parameter to migrate_vma_insert_page() is not used anymore.

Link: https://lkml.kernel.org/r/CANubcdUwCAMuUyamG2dkWP=cqSR9MAS=tHLDc95kQkqU-rEnAg@mail.gmail.com
Signed-off-by: Stephen Zhang <starzhangzsd@gmail.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: migrate: return -ENOSYS if THP migration is unsupported

In the current implementation unmap_and_move() would return -ENOMEM if THP
migration is unsupported, then the THP will be split.  If split is failed
just exit without trying to migrate other pages.  It doesn't make too much
sense since there may be enough free memory to migrate other pages and
there may be a lot base pages on the list.

Return -ENOSYS to make consistent with hugetlb.  And if THP split is
failed just skip and try other pages on the list.

Just skip the whole list and exit when free memory is really low.

Link: https://lkml.kernel.org/r/20201113205359.556831-6-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: migrate: clean up migrate_prep{_local}

The migrate_prep{_local} never fails, so it is pointless to have return
value and check the return value.

Link: https://lkml.kernel.org/r/20201113205359.556831-5-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Zi Yan <ziy@nvidia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Song Liu <songliubraving@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: migrate: skip shared exec THP for NUMA balancing

The NUMA balancing skip shared exec base page.  Since
CONFIG_READ_ONLY_THP_FOR_FS was introduced, there are probably shared exec
THP, so skip such THPs for NUMA balancing as well.

And Willy's regular filesystem THP support patches could create shared
exec THP wven without that config.

In addition, the page_is_file_lru() is used to tell if the page is file
cache or not, but it filters out shmem page.  It sounds like a typical
usecase by putting executables in shmem to achieve performance gain via
using shmem-THP, so it sounds worth skipping migration for such case too.

Link: https://lkml.kernel.org/r/20201113205359.556831-4-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: migrate: simplify the logic for handling permanent failure

When unmap_and_move{_huge_page}() returns !-EAGAIN and
!MIGRATEPAGE_SUCCESS, the page would be put back to LRU or proper list if
it is non-LRU movable page.  But, the callers always call
putback_movable_pages() to put the failed pages back later on, so it seems
not very efficient to put every single page back immediately, and the code
looks convoluted.

Put the failed page on a separate list, then splice the list to migrate
list when all pages are tried.  It is the caller's responsibility to call
putback_movable_pages() to handle failures.  This also makes the code
simpler and more readable.

After the change the rules are:
    * Success: non hugetlb page will be freed, hugetlb page will be put
               back
    * -EAGAIN: stay on the from list
    * -ENOMEM: stay on the from list
    * Other errno: put on ret_pages list then splice to from list

The from list would be empty iff all pages are migrated successfully, it
was not so before.  This has no impact to current existing callsites.

Link: https://lkml.kernel.org/r/20201113205359.556831-3-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Zi Yan <ziy@nvidia.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Song Liu <songliubraving@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: truncate_complete_page() does not exist any more

Patch series "mm: misc migrate cleanup and improvement", v3.

This patch (of 5):

The commit 9f4e41f4717832e ("mm: refactor truncate_complete_page()")
refactored truncate_complete_page(), and it is not existed anymore,
correct the comment in vmscan and migrate to avoid confusion.

Link: https://lkml.kernel.org/r/20201113205359.556831-1-shy828301@gmail.com
Link: https://lkml.kernel.org/r/20201113205359.556831-2-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: support THPs in zero_user_segments

We can only kmap() one subpage of a THP at a time, so loop over all
relevant subpages, skipping ones which don't need to be zeroed.  This is
too large to inline when THPs are enabled and we actually need highmem, so
put it in highmem.c.

[willy@infradead.org: start1 was allowed to be less than start2]

Link: https://lkml.kernel.org/r/20201124041507.28996-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Naresh Kamboju <naresh.kamboju@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/migrate.c: optimize migrate_vma_pages() mmu notifier

When migrating a zero page or pte_none() anonymous page to device private
memory, migrate_vma_setup() will initialize the src[] array with a NULL
PFN.  This lets the device driver allocate device private memory and clear
it instead of DMAing a page of zeros over the device bus.

Since the source page didn't exist at the time, no struct page was locked
nor a migration PTE inserted into the CPU page tables.  The actual PTE
insertion happens in migrate_vma_pages() when it tries to insert the
device private struct page PTE into the CPU page tables.
migrate_vma_pages() has to call the mmu notifiers again since another
device could fault on the same page before the page table locks are
acquired.

Allow device drivers to optimize the invalidation similar to
migrate_vma_setup() by calling mmu_notifier_range_init() which sets struct
mmu_notifier_range event type to MMU_NOTIFY_MIGRATE and the
migrate_pgmap_owner field.

Link: https://lkml.kernel.org/r/20201021191335.10916-1-rcampbell@nvidia.com
Signed-off-by: Ralph Campbell <rcampbell@nvidia.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/migrate.c: fix comment spelling

The word in the comment is misspelled, it should be "include".

Link: https://lkml.kernel.org/r/20201024114144.GA20552@lilong
Signed-off-by: Long Li <lonuxli.64@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/oom_kill: change comment and rename is_dump_unreclaim_slabs()

Change the comment of is_dump_unreclaim_slabs(), it just check whether
nr_unreclaimable slabs amount is greater than user memory, and explain why
we dump unreclaim slabs.

Rename it to should_dump_unreclaim_slab() maybe better.

Link: https://lkml.kernel.org/r/20201030182704.GA53949@rlk
Signed-off-by: Hui Su <sh_def@163.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/compaction: make defer_compaction and compaction_deferred static

defer_compaction() and compaction_deferred() and compaction_restarting()
in mm/compaction.c won't be used in other files, so make them static, and
remove the declaration in the header file.

Take the chance to fix a typo.

Link: https://lkml.kernel.org/r/20201123170801.GA9625@rlk
Signed-off-by: Hui Su <sh_def@163.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Nitin Gupta <nigupta@nvidia.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Mateusz Nosek <mateusznosek0@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/compaction: move compaction_suitable's comment to right place

Since commit 837d026d560c ("mm/compaction: more trace to understand
when/why compaction start/finish"), the comment place is not suitable.

So move compaction_suitable's comment to right place.

Link: https://lkml.kernel.org/r/20201116144121.GA385717@rlk
Signed-off-by: Hui Su <sh_def@163.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/compaction: rename 'start_pfn' to 'iteration_start_pfn' in compact_zone()

There are two 'start_pfn' declared in compact_zone() which have different
meanings.  Rename the second one to 'iteration_start_pfn' to prevent
confusion.

Also, remove an useless semicolon.

Link: https://lkml.kernel.org/r/20201019115044.1571-1-yanfei.xu@windriver.com
Signed-off-by: Yanfei Xu <yanfei.xu@windriver.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

z3fold: remove preempt disabled sections for RT

Replace get_cpu_ptr() with migrate_disable()+this_cpu_ptr() so RT can take
spinlocks that become sleeping locks.

Signed-off-by Mike Galbraith <efault@gmx.de>

Link: https://lkml.kernel.org/r/20201209145151.18994-3-vitaly.wool@konsulko.com
Signed-off-by: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

z3fold: stricter locking and more careful reclaim

Use temporary slots in reclaim function to avoid possible race when
freeing those.

While at it, make sure we check CLAIMED flag under page lock in the
reclaim function to make sure we are not racing with z3fold_alloc().

Link: https://lkml.kernel.org/r/20201209145151.18994-4-vitaly.wool@konsulko.com
Signed-off-by: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: <stable@vger.kernel.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

z3fold: simplify freeing slots

Patch series "z3fold: stability / rt fixes".

Address z3fold stability issues under stress load, primarily in the
reclaim and free aspects.  Besides, it fixes the locking problems that
were only seen in real-time kernel configuration.

This patch (of 3):

There used to be two places in the code where slots could be freed, namely
when freeing the last allocated handle from the slots and when releasing
the z3fold header these slots aree linked to.  The logic to decide on
whether to free certain slots was complicated and error prone in both
functions and it led to failures in RT case.

To fix that, make free_handle() the single point of freeing slots.

Link: https://lkml.kernel.org/r/20201209145151.18994-1-vitaly.wool@konsulko.com
Link: https://lkml.kernel.org/r/20201209145151.18994-2-vitaly.wool@konsulko.com
Signed-off-by: Vitaly Wool <vitaly.wool@konsulko.com>
Tested-by: Mike Galbraith <efault@gmx.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page_isolation: do not isolate the max order page

A max order page has no buddy page and never merges to another order.  So
isolating and then freeing it is pointless.

Link: https://lkml.kernel.org/r/20201202122114.75316-1-songmuchun@bytedance.com
Fixes: 3c605096d315 ("mm/page_alloc: restrict max order of merging on isolated pageblock")
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmscan.c: remove the filename in the top of file comment

No point in having the filename inside the file.

Link: https://lkml.kernel.org/r/20201115141541.3878-1-hymmsx.yu@gmail.com
Signed-off-by: logic.yu <hymmsx.yu@gmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmscan: drop unneeded assignment in kswapd()

The refactoring to kswapd() in commit e716f2eb24de ("mm, vmscan: prevent
kswapd sleeping prematurely due to mismatched classzone_idx") turned an
assignment to reclaim_order into a dead store, as in all further paths,
reclaim_order will be assigned again before it is used.

make clang-analyzer on x86_64 tinyconfig caught my attention with:

  mm/vmscan.c: warning: Although the value stored to 'reclaim_order' is used in the enclosing expression, the value is never actually read from 'reclaim_order' [clang-analyzer-deadcode.DeadStores]

Compilers will detect this unneeded assignment and optimize this anyway.
So, the resulting binary is identical before and after this change.

Simplify the code and remove unneeded assignment to make clang-analyzer
happy.

No functional change. No change in binary code.

Link: https://lkml.kernel.org/r/20201004125827.17679-1-lukas.bulwahn@gmail.com
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: don't wake kswapd prematurely when watermark boosting is disabled

On 2-node NUMA hosts we see bursts of kswapd reclaim and subsequent
pressure spikes and stalls from cache refaults while there is plenty of
free memory in the system.

Usually, kswapd is woken up when all eligible nodes in an allocation are
full.  But the code related to watermark boosting can wake kswapd on one
full node while the other one is mostly empty.  This may be justified to
fight fragmentation, but is currently unconditionally done whether
watermark boosting is occurring or not.

In our case, many of our workloads' throughput scales with available
memory, and pure utilization is a more tangible concern than trends
around longer-term fragmentation.  As a result we generally disable
watermark boosting.

Wake kswapd only woken when watermark boosting is requested.

Link: https://lkml.kernel.org/r/20201020175833.397286-1-hannes@cmpxchg.org
Fixes: 1c30844d2dfe ("mm: reclaim small amounts of memory when an external fragmentation event occurs")
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

hugetlb: fix an error code in hugetlb_reserve_pages()

Preserve the error code from region_add() instead of returning success.

Link: https://lkml.kernel.org/r/X9NGZWnZl5/Mt99R@mwanda
Fixes: 0db9d74ed884 ("hugetlb: disable region_add file_region coalescing")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Mina Almasry <almasrymina@google.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hugetlb: remove unneeded initialization

hugetlb_add_hstate initializes nr_huge_pages and free_huge_pages to 0, but
since hstates[] is a global variable, all its fields are defined to 0
already.

Link: https://lkml.kernel.org/r/20201119112141.6452-1-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: hugetlb: fix type of delta parameter and related local variables in gather_surplus_pages()

On 64-bit machine, delta variable in hugetlb_acct_memory() may be larger
than 0xffffffff, but gather_surplus_pages() can only use the low 32-bit
value now.  So we need to fix type of delta parameter and related local
variables in gather_surplus_pages().

Link: https://lkml.kernel.org/r/1605793733-3573-1-git-send-email-liu.xiang@zlingsmart.com
Reported-by: Ma Chenggong <ma.chenggong@zlingsmart.com>
Signed-off-by: Liu Xiang <liu.xiang@zlingsmart.com>
Signed-off-by: Pan Jiagen <pan.jiagen@zlingsmart.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Liu Xiang <liuxiang_1999@126.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

khugepaged: add parameter explanations for kernel-doc markup

Add missed parameter explanation for some kernel-doc warnings:

  mm/khugepaged.c:102: warning: Function parameter or member 'nr_pte_mapped_thp' not described in 'mm_slot'
  mm/khugepaged.c:102: warning: Function parameter or member 'pte_mapped_thp' not described in 'mm_slot'
  mm/khugepaged.c:1424: warning: Function parameter or member 'mm' not described in 'collapse_pte_mapped_thp'
  mm/khugepaged.c:1424: warning: Function parameter or member 'addr' not described in 'collapse_pte_mapped_thp'
  mm/khugepaged.c:1626: warning: Function parameter or member 'mm' not described in 'collapse_file'
  mm/khugepaged.c:1626: warning: Function parameter or member 'file' not described in 'collapse_file'
  mm/khugepaged.c:1626: warning: Function parameter or member 'start' not described in 'collapse_file'
  mm/khugepaged.c:1626: warning: Function parameter or member 'hpage' not described in 'collapse_file'
  mm/khugepaged.c:1626: warning: Function parameter or member 'node' not described in 'collapse_file'

Link: https://lkml.kernel.org/r/1605597325-25284-1-git-send-email-alex.shi@linux.alibaba.com
Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/huge_mm.h: remove extern keyword

The external function definitions don't need the "extern" keyword.  Remove
them so future changes don't copy the function definition style.

Link: https://lkml.kernel.org/r/20201106235135.32109-1-rcampbell@nvidia.com
Signed-off-by: Ralph Campbell <rcampbell@nvidia.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/hugetlb.c: just use put_page_testzero() instead of page_count()

We test the page reference count is zero or not here, it can be a bug here
if page refercence count is not zero.  So we can just use
put_page_testzero() instead of page_count().

Link: https://lkml.kernel.org/r/20201007170949.GA6416@rlk
Signed-off-by: Hui Su <sh_def@163.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: return -EBUSY when migration fails

Currently, we return -EIO when we fail to migrate the page.

Migrations' failures are rather transient as they can happen due to
several reasons, e.g: high page refcount bump, mapping->migrate_page
failing etc.  All meaning that at that time the page could not be
migrated, but that has nothing to do with an EIO error.

Let us return -EBUSY instead, as we do in case we failed to isolate the
page.

While are it, let us remove the "ret" print as its value does not change.

Link: https://lkml.kernel.org/r/20201209092818.30417-1-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,memory_failure: always pin the page in madvise_inject_error

madvise_inject_error() uses get_user_pages_fast to translate the address
we specified to a page.  After [1], we drop the extra reference count for
memory_failure() path.  That commit says that memory_failure wanted to
keep the pin in order to take the page out of circulation.

The truth is that we need to keep the page pinned, otherwise the page
might be re-used after the put_page() and we can end up messing with
someone else's memory.

E.g:

CPU0
process X CPU1
 madvise_inject_error
  get_user_pages
   put_page
page gets reclaimed
process Y allocates the page
  memory_failure
   // We mess with process Y memory

madvise() is meant to operate on a self address space, so messing with
pages that do not belong to us seems the wrong thing to do.
To avoid that, let us keep the page pinned for memory_failure as well.

Pages for DAX mappings will release this extra refcount in
memory_failure_dev_pagemap.

[1] ("23e7b5c2e271: mm, madvise_inject_error:
      Let memory_failure() optionally take a page reference")

Link: https://lkml.kernel.org/r/20201207094818.8518-1-osalvador@suse.de
Fixes: 23e7b5c2e271 ("mm, madvise_inject_error: Let memory_failure() optionally take a page reference")
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: remove drain_all_pages from shake_page

get_hwpoison_page already drains pcplists, previously disabling them when
trying to grab a refcount.  We do not need shake_page to take care of it
anymore.

Link: https://lkml.kernel.org/r/20201204102558.31607-4-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Qian Cai <qcai@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: disable pcplists before grabbing a refcount

Currently, we have a sort of retry mechanism to make sure pages in
pcp-lists are spilled to the buddy system, so we can handle those.

We can save us this extra checks with the new disable-pcplist mechanism
that is available with [1].

zone_pcplist_disable makes sure to 1) disable pcplists, so any page that
is freed up from that point onwards will end up in the buddy system and 2)
drain pcplists, so those pages that already in pcplists are spilled to
buddy.

With that, we can make a common entry point for grabbing a refcount from
both soft_offline and memory_failure paths that is guarded by
zone_pcplist_disable/zone_pcplist_enable.

[1] https://patchwork.kernel.org/project/linux-mm/cover/20201111092812.11329-1-vbabka@suse.cz/

Link: https://lkml.kernel.org/r/20201204102558.31607-3-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Qian Cai <qcai@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: refactor get_any_page

Patch series "HWPoison: Refactor get page interface", v2.

This patch (of 3):

When we want to grab a refcount via get_any_page, we call __get_any_page
that calls get_hwpoison_page to get the actual refcount.

get_any_page() is only there because we have a sort of retry mechanism in
case the page we met is unknown to us or if we raced with an allocation.

Also __get_any_page() prints some messages about the page type in case the
page was a free page or the page type was unknown, but if anything, we
only need to print a message in case the pagetype was unknown, as that is
reporting an error down the chain.

Let us merge get_any_page() and __get_any_page(), and let the message be
printed in soft_offline_page.  While we are it, we can also remove the
'pfn' parameter as it is no longer used.

Link: https://lkml.kernel.org/r/20201204102558.31607-1-osalvador@suse.de
Link: https://lkml.kernel.org/r/20201204102558.31607-2-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Acked-by: Vlastimil Babka <Vbabka@suse.cz>
Cc: Qian Cai <qcai@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: drop unneeded pcplist draining

memory_failure and soft_offline_path paths now drain pcplists by calling
get_hwpoison_page.

memory_failure flags the page as HWPoison before, so that page cannot
longer go into a pcplist, and soft_offline_page only flags a page as
HWPoison if 1) we took the page off a buddy freelist 2) the page was
in-use and we migrated it 3) was a clean pagecache.

Because of that, a page cannot longer be poisoned and be in a pcplist.

Link: https://lkml.kernel.org/r/20201013144447.6706-5-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: take free pages off the buddy freelists

The crux of the matter is that historically we left poisoned pages in the
buddy system because we have some checks in place when allocating a page
that are gatekeeper for poisoned pages.  Unfortunately, we do have other
users (e.g: compaction [1]) that scan buddy freelists and try to get a
page from there without checking whether the page is HWPoison.

As I stated already, I think it is fundamentally wrong to keep HWPoison
pages within the buddy systems, checks in place or not.

Let us fix this the same way we did for soft_offline [2], taking the page
off the buddy freelist so it is completely unreachable.

Note that this is fairly simple to trigger, as we only need to poison free
buddy pages (madvise MADV_HWPOISON) and then run some sort of memory
stress system.

Just for a matter of reference, I put a dump_page() in compaction_alloc()
to trigger for HWPoison patches:

    page:0000000012b2982b refcount:1 mapcount:0 mapping:0000000000000000 index:0x1 pfn:0x1d5db
    flags: 0xfffffc0800000(hwpoison)
    raw: 000fffffc0800000 ffffea00007573c8 ffffc90000857de0 0000000000000000
    raw: 0000000000000001 0000000000000000 00000001ffffffff 0000000000000000
    page dumped because: compaction_alloc

    CPU: 4 PID: 123 Comm: kcompactd0 Tainted: G            E     5.9.0-rc2-mm1-1-default+ #5
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.10.2-0-g5f4c7b1-prebuilt.qemu-project.org 04/01/2014
    Call Trace:
     dump_stack+0x6d/0x8b
     compaction_alloc+0xb2/0xc0
     migrate_pages+0x2a6/0x12a0
     compact_zone+0x5eb/0x11c0
     proactive_compact_node+0x89/0xf0
     kcompactd+0x2d0/0x3a0
     kthread+0x118/0x130
     ret_from_fork+0x22/0x30

After that, if e.g: a process faults in the page,  it will get killed
unexpectedly.
Fix it by containing the page immediatelly.

Besides that, two more changes can be noticed:

* MF_DELAYED no longer suits as we are fixing the issue by containing
  the page immediately, so it does no longer rely on the allocation-time
  checks to stop HWPoison to be handed over.
  gain unless it is unpoisoned, so we fixed the situation.
  Because of that, let us use MF_RECOVERED from now on.

* The second block that handles PageBuddy pages is no longer needed:
  We call shake_page and then check whether the page is Buddy
  because shake_page calls drain_all_pages, which sends pcp-pages back to
  the buddy freelists, so we could have a chance to handle free pages.
  Currently, get_hwpoison_page already calls drain_all_pages, and we call
  get_hwpoison_page right before coming here, so we should be on the safe
  side.

[1] https://lore.kernel.org/linux-mm/20190826104144.GA7849@linux/T/#u
[2] https://patchwork.kernel.org/cover/11792607/

[osalvador@suse.de: take the poisoned subpage off the buddy frelists]
Link: https://lkml.kernel.org/r/20201013144447.6706-4-osalvador@suse.de
Link: https://lkml.kernel.org/r/20201013144447.6706-3-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm,hwpoison: drain pcplists before bailing out for non-buddy zero-refcount page

Patch series "HWpoison: further fixes and cleanups", v5.

This patchset includes some more fixes and a cleanup.

Patch#2 and patch#3 are both fixes for taking a HWpoison page off a buddy
freelist, since having them there has proved to be bad (see [1] and
pathch#2's commit log).  Patch#3 does the same for hugetlb pages.

[1] https://lkml.org/lkml/2020/9/22/565

This patch (of 4):

A page with 0-refcount and !PageBuddy could perfectly be a pcppage.
Currently, we bail out with an error if we encounter such a page, meaning
that we do not handle pcppages neither from hard-offline nor from
soft-offline path.

Fix this by draining pcplists whenever we find this kind of page and retry
the check again.  It might be that pcplists have been spilled into the
buddy allocator and so we can handle it.

Link: https://lkml.kernel.org/r/20201013144447.6706-1-osalvador@suse.de
Link: https://lkml.kernel.org/r/20201013144447.6706-2-osalvador@suse.de
Signed-off-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page_alloc: speed up the iteration of max_order

When we free a page whose order is very close to MAX_ORDER and greater
than pageblock_order, it wastes some CPU cycles to increase max_order to
MAX_ORDER one by one and check the pageblock migratetype of that page
repeatedly especially when MAX_ORDER is much larger than pageblock_order.

We also should not be checking migratetype of buddy when "order ==
MAX_ORDER - 1" as the buddy pfn may be invalid, so adjust the condition.
With the new check, we don't need the max_order check anymore, so we
replace it.

Also adjust max_order initialization so that it's lower by one than
previously, which makes the code hopefully more clear.

Link: https://lkml.kernel.org/r/20201204155109.55451-1-songmuchun@bytedance.com
Fixes: d9dddbf55667 ("mm/page_alloc: prevent merging between isolated and other pageblocks")
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: page_alloc: refactor setup_per_zone_lowmem_reserve()

setup_per_zone_lowmem_reserve() iterates through each zone setting
zone->lowmem_reserve[j] = 0 (where j is the zone's index) then iterates
backwards through all preceding zones, setting
lower_zone->lowmem_reserve[j] = sum(managed pages of higher zones) /
lowmem_reserve_ratio[idx] for each (where idx is the lower zone's index).

If the lower zone has no managed pages or its ratio is 0 then all of its
lowmem_reserve[] entries are effectively zeroed.

As these arrays are only assigned here and all lowmem_reserve[] entries
for index < this zone's index are implicitly assumed to be 0 (as these are
specifically output in show_free_areas() and zoneinfo_show_print() for
example) there is no need to additionally zero index == this zone's index
too.  This patch avoids zeroing unnecessarily.

Rather than iterating through zones and setting lowmem_reserve[j] for each
lower zone this patch reverse the process and populates each zone's
lowmem_reserve[] values in ascending order.

This clarifies what is going on especially in the case of zero managed
pages or ratio which is now explicitly shown to clear these values.

Link: https://lkml.kernel.org/r/20201129162758.115907-1-lstoakes@gmail.com
Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Roman Gushchin <guro@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

init/main: fix broken buffer_init when DEFERRED_STRUCT_PAGE_INIT set

In the booting phase if CONFIG_DEFERRED_STRUCT_PAGE_INIT is set,
we have following callchain:

start_kernel
...
  mm_init
    mem_init
     memblock_free_all
       reset_all_zones_managed_pages
       free_low_memory_core_early
...
  buffer_init
    nr_free_buffer_pages
      zone->managed_pages
...
  rest_init
    kernel_init
      kernel_init_freeable
        page_alloc_init_late
          kthread_run(deferred_init_memmap, NODE_DATA(nid), "pgdatinit%d", nid);
          wait_for_completion(&pgdat_init_all_done_comp);
          ...
          files_maxfiles_init

It's clear that buffer_init depends on zone->managed_pages, but it's reset
in reset_all_zones_managed_pages after that pages are readded into
zone->managed_pages, but when buffer_init runs this process is half done
and most of them will finally be added till deferred_init_memmap done.  In
large memory couting of nr_free_buffer_pages drifts too much, also
drifting from kernels to kernels on same hardware.

Fix is simple, it delays buffer_init run till deferred_init_memmap all
done.

But as corrected by this patch, max_buffer_heads becomes very large, the
value is roughly as many as 4 times of totalram_pages, formula:
max_buffer_heads = nrpages * (10%) * (PAGE_SIZE / sizeof(struct
buffer_head));

Say in a 64GB memory box we have 16777216 pages, then max_buffer_heads
turns out to be roughly 67,108,864.  In common cases, should a buffer_head
be mapped to one page/block(4KB)?  So max_buffer_heads never exceeds
totalram_pages.  IMO it's likely to make buffer_heads_over_limit bool
value alwasy false, then make codes 'if (buffer_heads_over_limit)' test in
vmscan unnecessary.

So this patch will change the original behavior related to
buffer_heads_over_limit in vmscan since we used a half done value of
zone->managed_pages before, or should we use a smaller factor(<10%) in
previous formula.

akpm: I think this is OK - the max_buffer_heads code is only needed on
highmem machines, to prevent ZONE_NORMAL from being consumed by large
amounts of buffer_heads attached to highmem pagecache.  This problem will
not occur on 64-bit machines, so this feature's non-functionality on such
machines is a feature, not a bug.

Link: https://lkml.kernel.org/r/20201123110500.103523-1-linf@wangsu.com
Signed-off-by: Lin Feng <linf@wangsu.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page_alloc: clear all pages in post_alloc_hook() with init_on_alloc=1

commit 6471384af2a6 ("mm: security: introduce init_on_alloc=1 and
init_on_free=1 boot options") resulted with init_on_alloc=1 in all pages
leaving the buddy via alloc_pages() and friends to be
initialized/cleared/zeroed on allocation.

However, the same logic is currently not applied to alloc_contig_pages():
allocated pages leaving the buddy aren't cleared with init_on_alloc=1 and
init_on_free=0.  Let's also properly clear pages on that allocation path.

To achieve that, let's move clearing into post_alloc_hook().  This will
not only affect alloc_contig_pages() allocations but also any pages used
as migration target in compaction code via compaction_alloc().

While this sounds sub-optimal, it's the very same handling as when
allocating migration targets via alloc_migration_target() - pages will get
properly cleared with init_on_free=1.  In case we ever want to optimize
migration in that regard, we should tackle all such migration users - if
we believe migration code can be fully trusted.

With this change, we will see double clearing of pages in some cases.  One
example are gigantic pages (either allocated via CMA, or allocated
dynamically via alloc_contig_pages()) - which is the right thing to do
(and to be optimized outside of the buddy in the callers) as discussed in:
https://lkml.kernel.org/r/20201019182853.7467-1-gpiccoli@canonical.com

This change implies that with init_on_alloc=1

 - All CMA allocations will be cleared

 - Gigantic pages allocated via alloc_contig_pages() will be cleared

 - virtio-mem memory to be unplugged will be cleared. While this is
   suboptimal, it's similar to memory balloon drivers handling, where
   all pages to be inflated will get cleared as well.

 - Pages isolated for compaction will be cleared

Link: https://lkml.kernel.org/r/20201120180452.19071-1-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Kees Cook <keescook@chromium.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page_alloc: mark some symbols with static keyword

Fix the following sparse warnings:

  mm/page_alloc.c:3040:6: warning: symbol '__drain_all_pages' was not declared. Should it be static?
  mm/page_alloc.c:6349:6: warning: symbol '__zone_set_pageset_high_and_batch' was not declared. Should it be static?

Link: https://lkml.kernel.org/r/1605517365-65858-1-git-send-email-zou_wei@huawei.com
Signed-off-by: Zou Wei <zou_wei@huawei.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page_alloc: add __free_pages() documentation

Provide some guidance towards when this might not be the right interface
to use.

Link: https://lkml.kernel.org/r/20201027025523.3235-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: William Kucharski <william.kucharski@oracle.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/page-flags: fix comment

We haven't had 'dontuse' flags since 2002.  Replace this obsolete warning
with a hopefully more useful one.

Link: https://lkml.kernel.org/r/20201027025823.3704-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: William Kucharski <william.kucharski@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

include/linux/page-flags.h: remove unused __[Set|Clear]PagePrivate

They are not used anymore.

Link: https://lkml.kernel.org/r/20201009135914.64826-1-linmiaohe@huawei.com
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: disable pcplists during memory offline

Memory offlining relies on page isolation to guarantee a forward progress
because pages cannot be reused while they are isolated.  But the page
isolation itself doesn't prevent from races while freed pages are stored
on pcp lists and thus can be reused.  This can be worked around by
repeated draining of pcplists, as done by commit 968318261221
("mm/memory_hotplug: drain per-cpu pages again during memory offline").

David and Michal would prefer that this race was closed in a way that
callers of page isolation who need stronger guarantees don't need to
repeatedly drain.  David suggested disabling pcplists usage completely
during page isolation, instead of repeatedly draining them.

To achieve this without adding special cases in alloc/free fastpath, we
can use the same approach as boot pagesets - when pcp->high is 0, any
pcplist addition will be immediately flushed.

The race can thus be closed by setting pcp->high to 0 and draining
pcplists once, before calling start_isolate_page_range().  The draining
will serialize after processes that already disabled interrupts and read
the old value of pcp->high in free_unref_page_commit(), and processes that
have not yet disabled interrupts, will observe pcp->high == 0 when they
are rescheduled, and skip pcplists.  This guarantees no stray pages on
pcplists in zones where isolation happens.

This patch thus adds zone_pcp_disable() and zone_pcp_enable() functions
that page isolation users can call before start_isolate_page_range() and
after unisolating (or offlining) the isolated pages.

Also, drain_all_pages() is optimized to only execute on cpus where
pcplists are not empty.  The check can however race with a free to pcplist
that has not yet increased the pcp->count from 0 to 1.  Thus make the
drain optionally skip the racy check and drain on all cpus, and use this
option in zone_pcp_disable().

As we have to avoid external updates to high and batch while pcplists are
disabled, we take pcp_batch_high_lock in zone_pcp_disable() and release it
in zone_pcp_enable().  This also synchronizes multiple users of
zone_pcp_disable()/enable().

Currently the only user of this functionality is offline_pages().

[vbabka@suse.cz: add comment, per David]
Link: https://lkml.kernel.org/r/527480ef-ed72-e1c1-52a0-1c5b0113df45@suse.cz
Link: https://lkml.kernel.org/r/20201111092812.11329-8-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Suggested-by: David Hildenbrand <david@redhat.com>
Suggested-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: move draining pcplists to page isolation users

Currently, pcplists are drained during set_migratetype_isolate() which
means once per pageblock processed start_isolate_page_range().  This is
somewhat wasteful.  Moreover, the callers might need different guarantees,
and the draining is currently prone to races and does not guarantee that
no page from isolated pageblock will end up on the pcplist after the
drain.

Better guarantees are added by later patches and require explicit actions
by page isolation users that need them.  Thus it makes sense to move the
current imperfect draining to the callers also as a preparation step.

Link: https://lkml.kernel.org/r/20201111092812.11329-7-vbabka@suse.cz
Suggested-by: David Hildenbrand <david@redhat.com>
Suggested-by: Pavel Tatashin <pasha.tatashin@soleen.com>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: cache pageset high and batch in struct zone

All per-cpu pagesets for a zone use the same high and batch values, that
are duplicated there just for performance (locality) reasons.  This patch
adds the same variables also to struct zone as a shared copy.

This will be useful later for making possible to disable pcplists
temporarily by setting high value to 0, while remembering the values for
restoring them later.  But we can also immediately benefit from not
updating pagesets of all possible cpus in case the newly recalculated
values (after sysctl change or memory online/offline) are actually
unchanged from the previous ones.

Link: https://lkml.kernel.org/r/20201111092812.11329-6-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: simplify pageset_update()

pageset_update() attempts to update pcplist's high and batch values in a
way that readers don't observe batch > high.  It uses smp_wmb() to order
the updates in a way to achieve this.  However, without proper pairing
read barriers in readers this guarantee doesn't hold, and there are no
such barriers in e.g.  free_unref_page_commit().

Commit 88e8ac11d2ea ("mm, page_alloc: fix core hung in
free_pcppages_bulk()") already showed this is problematic, and solved this
by ultimately only trusing pcp->count of the current cpu with interrupts
disabled.

The update dance with unpaired write barriers thus makes no sense.
Replace them with plain WRITE_ONCE to prevent store tearing, and document
that the values can change asynchronously and should not be trusted for
correctness.

All current readers appear to be OK after 88e8ac11d2ea.  Convert them to
READ_ONCE to prevent unnecessary read tearing, but mainly to alert anybody
making future changes to the code that special care is needed.

Link: https://lkml.kernel.org/r/20201111092812.11329-5-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: remove setup_pageset()

We initialize boot-time pagesets with setup_pageset(), which sets high and
batch values that effectively disable pcplists.

We can remove this wrapper if we just set these values for all pagesets in
pageset_init().  Non-boot pagesets then subsequently update them to the
proper values.

No functional change.

Link: https://lkml.kernel.org/r/20201111092812.11329-4-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: calculate pageset high and batch once per zone

We currently call pageset_set_high_and_batch() for each possible cpu,
which repeats the same calculations of high and batch values.

Instead call the function just once per zone, and make it apply the
calculated values to all per-cpu pagesets of the zone.

This also allows removing the zone_pageset_init() and __zone_pcp_update()
wrappers.

No functional change.

Link: https://lkml.kernel.org/r/20201111092812.11329-3-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm, page_alloc: clean up pageset high and batch update

Patch series "disable pcplists during memory offline", v3.

As per the discussions [1] [2] this is an attempt to implement David's
suggestion that page isolation should disable pcplists to avoid races with
page freeing in progress.  This is done without extra checks in fast
paths, as explained in Patch 9.  The repeated draining done by [2] is then
no longer needed.  Previous version (RFC) is at [3].

The RFC tried to hide pcplists disabling/enabling into page isolation, but
it wasn't completely possible, as memory offline does not unisolation.
Michal suggested an explicit API in [4] so that's the current
implementation and it seems indeed nicer.

Once we accept that page isolation users need to do explicit actions
around it depending on the needed guarantees, we can also IMHO accept that
the current pcplist draining can be also done by the callers, which is
more effective.  After all, there are only two users of page isolation.
So patch 6 does effectively the same thing as Pavel proposed in [5], and
patch 7 implement stronger guarantees only for memory offline.  If CMA
decides to opt-in to the stronger guarantee, it can be added later.

Patches 1-5 are preparatory cleanups for pcplist disabling.

Patchset was briefly tested in QEMU so that memory online/offline works,
but I haven't done a stress test that would prove the race fixed by [2] is
eliminated.

Note that patch 7 could be avoided if we instead adjusted page freeing in
shown in [6], but I believe the current implementation of disabling
pcplists is not too much complex, so I would prefer this instead of adding
new checks and longer irq-disabled section into page freeing hotpaths.

[1] https://lore.kernel.org/linux-mm/20200901124615.137200-1-pasha.tatashin@soleen.com/
[2] https://lore.kernel.org/linux-mm/20200903140032.380431-1-pasha.tatashin@soleen.com/
[3] https://lore.kernel.org/linux-mm/20200907163628.26495-1-vbabka@suse.cz/
[4] https://lore.kernel.org/linux-mm/20200909113647.GG7348@dhcp22.suse.cz/
[5] https://lore.kernel.org/linux-mm/20200904151448.100489-3-pasha.tatashin@soleen.com/
[6] https://lore.kernel.org/linux-mm/3d3b53db-aeaa-ff24-260b-36427fac9b1c@suse.cz/
[7] https://lore.kernel.org/linux-mm/20200922143712.12048-1-vbabka@suse.cz/
[8] https://lore.kernel.org/linux-mm/20201008114201.18824-1-vbabka@suse.cz/

This patch (of 7):

The updates to pcplists' high and batch values are handled by multiple
functions that make the calculations hard to follow.  Consolidate
everything to pageset_set_high_and_batch() and remove pageset_set_batch()
and pageset_set_high() wrappers.

The only special case using one of the removed wrappers was:
build_all_zonelists_init()

  setup_pageset()
    pageset_set_batch()

which was hardcoding batch as 0, so we can just open-code a call to
pageset_update() with constant parameters instead.

No functional change.

Link: https://lkml.kernel.org/r/20201111092812.11329-1-vbabka@suse.cz
Link: https://lkml.kernel.org/r/20201111092812.11329-2-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Pankaj Gupta <pankaj.gupta@cloud.ionos.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arch, mm: make kernel_page_present() always available

For architectures that enable ARCH_HAS_SET_MEMORY having the ability to
verify that a page is mapped in the kernel direct map can be useful
regardless of hibernation.

Add RISC-V implementation of kernel_page_present(), update its forward
declarations and stubs to be a part of set_memory API and remove ugly
ifdefery in inlcude/linux/mm.h around current declarations of
kernel_page_present().

Link: https://lkml.kernel.org/r/20201109192128.960-5-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: "Edgecombe, Rick P" <rick.p.edgecombe@intel.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arch, mm: restore dependency of __kernel_map_pages() on DEBUG_PAGEALLOC

The design of DEBUG_PAGEALLOC presumes that __kernel_map_pages() must
never fail.  With this assumption is wouldn't be safe to allow general
usage of this function.

Moreover, some architectures that implement __kernel_map_pages() have this
function guarded by #ifdef DEBUG_PAGEALLOC and some refuse to map/unmap
pages when page allocation debugging is disabled at runtime.

As all the users of __kernel_map_pages() were converted to use
debug_pagealloc_map_pages() it is safe to make it available only when
DEBUG_PAGEALLOC is set.

Link: https://lkml.kernel.org/r/20201109192128.960-4-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: "Edgecombe, Rick P" <rick.p.edgecombe@intel.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

PM: hibernate: make direct map manipulations more explicit

When DEBUG_PAGEALLOC or ARCH_HAS_SET_DIRECT_MAP is enabled a page may be
not present in the direct map and has to be explicitly mapped before it
could be copied.

Introduce hibernate_map_page() and hibernation_unmap_page() that will
explicitly use set_direct_map_{default,invalid}_noflush() for
ARCH_HAS_SET_DIRECT_MAP case and debug_pagealloc_{map,unmap}_pages() for
DEBUG_PAGEALLOC case.

The remapping of the pages in safe_copy_page() presumes that it only
changes protection bits in an existing PTE and so it is safe to ignore
return value of set_direct_map_{default,invalid}_noflush().

Still, add a pr_warn() so that future changes in set_memory APIs will not
silently break hibernation.

Link: https://lkml.kernel.org/r/20201109192128.960-3-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: "Edgecombe, Rick P" <rick.p.edgecombe@intel.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: introduce debug_pagealloc_{map,unmap}_pages() helpers

Patch series "arch, mm: improve robustness of direct map manipulation", v7.

During recent discussion about KVM protected memory, David raised a
concern about usage of __kernel_map_pages() outside of DEBUG_PAGEALLOC
scope [1].

Indeed, for architectures that define CONFIG_ARCH_HAS_SET_DIRECT_MAP it is
possible that __kernel_map_pages() would fail, but since this function is
void, the failure will go unnoticed.

Moreover, there's lack of consistency of __kernel_map_pages() semantics
across architectures as some guard this function with #ifdef
DEBUG_PAGEALLOC, some refuse to update the direct map if page allocation
debugging is disabled at run time and some allow modifying the direct map
regardless of DEBUG_PAGEALLOC settings.

This set straightens this out by restoring dependency of
__kernel_map_pages() on DEBUG_PAGEALLOC and updating the call sites
accordingly.

Since currently the only user of __kernel_map_pages() outside
DEBUG_PAGEALLOC is hibernation, it is updated to make direct map accesses
there more explicit.

[1] https://lore.kernel.org/lkml/2759b4bf-e1e3-d006-7d86-78a40348269d@redhat.com

This patch (of 4):

When CONFIG_DEBUG_PAGEALLOC is enabled, it unmaps pages from the kernel
direct mapping after free_pages().  The pages than need to be mapped back
before they could be used.  Theese mapping operations use
__kernel_map_pages() guarded with with debug_pagealloc_enabled().

The only place that calls __kernel_map_pages() without checking whether
DEBUG_PAGEALLOC is enabled is the hibernation code that presumes
availability of this function when ARCH_HAS_SET_DIRECT_MAP is set.  Still,
on arm64, __kernel_map_pages() will bail out when DEBUG_PAGEALLOC is not
enabled but set_direct_map_invalid_noflush() may render some pages not
present in the direct map and hibernation code won't be able to save such
pages.

To make page allocation debugging and hibernation interaction more robust,
the dependency on DEBUG_PAGEALLOC or ARCH_HAS_SET_DIRECT_MAP has to be
made more explicit.

Start with combining the guard condition and the call to
__kernel_map_pages() into debug_pagealloc_map_pages() and
debug_pagealloc_unmap_pages() functions to emphasize that
__kernel_map_pages() should not be called without DEBUG_PAGEALLOC and use
these new functions to map/unmap pages when page allocation debugging is
enabled.

Link: https://lkml.kernel.org/r/20201109192128.960-1-rppt@kernel.org
Link: https://lkml.kernel.org/r/20201109192128.960-2-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "Edgecombe, Rick P" <rick.p.edgecombe@intel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Len Brown <len.brown@intel.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: "Rafael J. Wysocki" <rjw@rjwysocki.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m68k: deprecate DISCONTIGMEM

DISCONTIGMEM was intended to provide more efficient support for systems
with holes in their physical address space that FLATMEM did.

Yet, it's overhead in terms of the memory consumption seems to
overweight the savings on the unused memory map.

For a ARAnyM system with 16 MBytes of FastRAM configured, the memory
usage reported after page allocator initialization is

  Memory: 23828K/30720K available (3206K kernel code, 535K rwdata, 936K rodata, 768K init, 193K bss, 6892K reserved, 0K cma-reserved)

and with DISCONTIGMEM disabled and with relatively large hole in the memory
map it is:

  Memory: 23864K/30720K available (3197K kernel code, 516K rwdata, 936K rodata, 764K init, 179K bss, 6856K reserved, 0K cma-reserved)

Moreover, since m68k already has custom pfn_valid() it is possible to
define HAVE_ARCH_PFN_VALID to enable freeing of unused memory map.  The
minimal size of a hole that can be freed should not be less than
MAX_ORDER_NR_PAGES so to achieve more substantial memory savings let
m68k also define custom FORCE_MAX_ZONEORDER.

With FORCE_MAX_ZONEORDER set to 9 memory usage becomes:

  Memory: 23880K/30720K available (3197K kernel code, 516K rwdata, 936K rodata, 764K init, 179K bss, 6840K reserved, 0K cma-reserved)

Link: https://lkml.kernel.org/r/20201101170454.9567-14-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m68k/mm: enable use of generic memory_model.h for !DISCONTIGMEM

The pg_data_map and pg_data_table arrays as well as page_to_pfn() and
pfn_to_page() are required only for DISCONTIGMEM. Other memory models can
use the generic definitions in asm-generic/memory_model.h.

Link: https://lkml.kernel.org/r/20201101170454.9567-13-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

m68k/mm: make node data and node setup depend on CONFIG_DISCONTIGMEM

The pg_data_t node structures and their initialization currently depends on
!CONFIG_SINGLE_MEMORY_CHUNK. Since they are required only for DISCONTIGMEM
make this dependency explicit and replace usage of
CONFIG_SINGLE_MEMORY_CHUNK with CONFIG_DISCONTIGMEM where appropriate.

The CONFIG_SINGLE_MEMORY_CHUNK was implicitly disabled on the ColdFire MMU
variant, although it always presumed a single memory bank. As there is no
actual need for DISCONTIGMEM in this case, make sure that ColdFire MMU
systems set CONFIG_SINGLE_MEMORY_CHUNK to 'y'.

Link: https://lkml.kernel.org/r/20201101170454.9567-12-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arc: use FLATMEM with freeing of unused memory map instead of DISCONTIGMEM

Currently ARC uses DISCONTIGMEM to cope with sparse physical memory address
space on systems with 2 memory banks. While DISCONTIGMEM avoids wasting
memory on unpopulated memory map, it adds both memory and CPU overhead
relatively to FLATMEM. Moreover, DISCONTINGMEM is generally considered
deprecated.

The obvious replacement for DISCONTIGMEM would be SPARSEMEM, but it is also
less efficient than FLATMEM in pfn_to_page() and page_to_pfn() conversions.
Besides it requires tuning of SECTION_SIZE which is not trivial for
possible ARC memory configuration.

Since the memory map for both banks is always allocated from the "lowmem"
bank, it is possible to use FLATMEM for two-bank configuration and simply
free the unused hole in the memory map. All is required for that is to
provide ARC-specific pfn_valid() that will take into account actual
physical memory configuration and define HAVE_ARCH_PFN_VALID.

The resulting kernel image configured with defconfig + HIGHMEM=y is
smaller:

  $ size a/vmlinux b/vmlinux
     text    data     bss     dec     hex filename
  4673503 1245456  279756 6198715  5e95bb a/vmlinux
  4658706 1246864  279756 6185326  5e616e b/vmlinux

  $ ./scripts/bloat-o-meter a/vmlinux b/vmlinux
  add/remove: 28/30 grow/shrink: 42/399 up/down: 10986/-29025 (-18039)
  ...
  Total: Before=4709315, After = 4691276, chg -0.38%

Booting nSIM with haps_ns.dts results in the following memory usage
reports:

  a:
  Memory: 1559104K/1572864K available (3531K kernel code, 595K rwdata, 752K rodata, 136K init, 275K bss, 13760K reserved, 0K cma-reserved, 1048576K highmem)

  b:
  Memory: 1559112K/1572864K available (3519K kernel code, 594K rwdata, 752K rodata, 136K init, 280K bss, 13752K reserved, 0K cma-reserved, 1048576K highmem)

Link: https://lkml.kernel.org/r/20201101170454.9567-11-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arm, arm64: move free_unused_memmap() to generic mm

ARM and ARM64 free unused parts of the memory map just before the
initialization of the page allocator. To allow holes in the memory map both
architectures overload pfn_valid() and define HAVE_ARCH_PFN_VALID.

Allowing holes in the memory map for FLATMEM may be useful for small
machines, such as ARC and m68k and will enable those architectures to cease
using DISCONTIGMEM and still support more than one memory bank.

Move the functions that free unused memory map to generic mm and enable
them in case HAVE_ARCH_PFN_VALID=y.

Link: https://lkml.kernel.org/r/20201101170454.9567-10-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arm: remove CONFIG_ARCH_HAS_HOLES_MEMORYMODEL

ARM is the only architecture that defines CONFIG_ARCH_HAS_HOLES_MEMORYMODEL
which in turn enables memmap_valid_within() function that is intended to
verify existence  of struct page associated with a pfn when there are holes
in the memory map.

However, the ARCH_HAS_HOLES_MEMORYMODEL also enables HAVE_ARCH_PFN_VALID
and arch-specific pfn_valid() implementation that also deals with the holes
in the memory map.

The only two users of memmap_valid_within() call this function after
a call to pfn_valid() so the memmap_valid_within() check becomes redundant.

Remove CONFIG_ARCH_HAS_HOLES_MEMORYMODEL and memmap_valid_within() and rely
entirely on ARM's implementation of pfn_valid() that is now enabled
unconditionally.

Link: https://lkml.kernel.org/r/20201101170454.9567-9-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ia64: make SPARSEMEM default and disable DISCONTIGMEM

SPARSEMEM memory model suitable for systems with large holes in their
phyiscal memory layout. With SPARSEMEM_VMEMMAP enabled it provides
pfn_to_page() and page_to_pfn() as fast as FLATMEM.

Make it the default memory model for IA-64 and disable DISCONTIGMEM which
is considered obsolete for quite some time.

Link: https://lkml.kernel.org/r/20201101170454.9567-8-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ia64: forbid using VIRTUAL_MEM_MAP with FLATMEM

Virtual memory map was intended to avoid wasting memory on the memory map
on systems with large holes in the physical memory layout. Long ago it been
superseded first by DISCONTIGMEM and then by SPARSEMEM. Moreover,
SPARSEMEM_VMEMMAP provide the same functionality in much more portable way.

As the first step to removing the VIRTUAL_MEM_MAP forbid it's usage with
FLATMEM and panic on systems with large holes in the physical memory
layout that try to run FLATMEM kernels.

Link: https://lkml.kernel.org/r/20201101170454.9567-7-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ia64: split virtual map initialization out of paging_init()

For both FLATMEM and DISCONTIGMEM/SPARSEMEM the virtual map initialization
is spread over paging_init() for no good reason.

Split out the bits related to virtual map initialization to a helper
functions, one for FLATMEM and another for !FLATMEM configurations.

Link: https://lkml.kernel.org/r/20201101170454.9567-6-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ia64: discontig: paging_init(): remove local max_pfn calculation

The maximal PFN in the system is calculated during find_memory() time and
it is stored at max_low_pfn then.

Use this value in paging_init() and remove the redundant detection of
max_pfn in that function.

Link: https://lkml.kernel.org/r/20201101170454.9567-5-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ia64: remove 'ifdef CONFIG_ZONE_DMA32' statements

After the removal of SN2 platform (commit cf07cb1ff4ea ("ia64: remove
support for the SGI SN2 platform") IA-64 always has ZONE_DMA32 and there is
no point to guard code with this configuration option.

Remove ifdefery associated with CONFIG_ZONE_DMA32

Link: https://lkml.kernel.org/r/20201101170454.9567-4-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

ia64: remove custom __early_pfn_to_nid()

The ia64 implementation of __early_pfn_to_nid() essentially relies on the
same data as the generic implementation.

The correspondence between memory ranges and nodes is set in memblock
during early memory initialization in register_active_ranges() function.

The initialization of sparsemem that requires early_pfn_to_nid() happens
later and it can use the memblock information like the other architectures.

Link: https://lkml.kernel.org/r/20201101170454.9567-3-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

alpha: switch from DISCONTIGMEM to SPARSEMEM

Patch series "arch, mm: deprecate DISCONTIGMEM", v2.

It's been a while since DISCONTIGMEM is generally considered deprecated,
but it is still used by four architectures.  This set replaces
DISCONTIGMEM with a different way to handle holes in the memory map and
marks DISCONTIGMEM configuration as BROKEN in Kconfigs of these
architectures with the intention to completely remove it in several
releases.

While for 64-bit alpha and ia64 the switch to SPARSEMEM is quite obvious
and was a matter of moving some bits around, for smaller 32-bit arc and
m68k SPARSEMEM is not necessarily the best thing to do.

On 32-bit machines SPARSEMEM would require large sections to make section
index fit in the page flags, but larger sections mean that more memory is
wasted for unused memory map.

Besides, pfn_to_page() and page_to_pfn() become less efficient, at least
on arc.

So I've decided to generalize arm's approach for freeing of unused parts
of the memory map with FLATMEM and enable it for both arc and m68k.  The
details are in the description of patches 10 (arc) and 13 (m68k).

This patch (of 13):

Enable SPARSEMEM support on alpha and deprecate DISCONTIGMEM.

The required changes are mostly around moving duplicated definitions of
page access and address conversion macros to a common place and making sure
they are available for all memory models.

The DISCONTINGMEM support is marked as BROKEN an will be removed in a
couple of releases.

Link: https://lkml.kernel.org/r/20201101170454.9567-1-rppt@kernel.org
Link: https://lkml.kernel.org/r/20201101170454.9567-2-rppt@kernel.org
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Meelis Roos <mroos@linux.ee>
Cc: Michael Schmitz <schmitzmic@gmail.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lkdtm: disable KASAN for rodata.o

Building lkdtm with KASAN and Clang 11 or later results in the following
error when attempting to load the module:

  kernel tried to execute NX-protected page - exploit attempt? (uid: 0)
  BUG: unable to handle page fault for address: ffffffffc019cd70
  #PF: supervisor instruction fetch in kernel mode
  #PF: error_code(0x0011) - permissions violation
  ...
  RIP: 0010:asan.module_ctor+0x0/0xffffffffffffa290 [lkdtm]
  ...
  Call Trace:
   do_init_module+0x17c/0x570
   load_module+0xadee/0xd0b0
   __x64_sys_finit_module+0x16c/0x1a0
   do_syscall_64+0x34/0x50
   entry_SYSCALL_64_after_hwframe+0x44/0xa9

The reason is that rodata.o generates a dummy function that lives in
.rodata to validate that .rodata can't be executed; however, Clang 11 adds
KASAN globals support by generating module constructors to initialize
globals redzones.  When Clang 11 adds a module constructor to rodata.o, it
is also added to .rodata: any attempt to call it on initialization results
in the above error.

Therefore, disable KASAN instrumentation for rodata.o.

Link: https://lkml.kernel.org/r/20201214191413.3164796-1-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kasan: update documentation for generic kasan

Generic KASAN also supports to record the last two workqueue stacks and
print them in KASAN report.  So that need to update documentation.

Link: https://lkml.kernel.org/r/20201203023037.30792-1-walter-zh.wu@mediatek.com
Signed-off-by: Walter Wu <walter-zh.wu@mediatek.com>
Suggested-by: Marco Elver <elver@google.com>
Acked-by: Marco Elver <elver@google.com>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

lib/test_kasan.c: add workqueue test case

Adds a test to verify workqueue stack recording and print it in
KASAN report.

The KASAN report was as follows(cleaned up slightly):

 BUG: KASAN: use-after-free in kasan_workqueue_uaf

 Freed by task 54:
  kasan_save_stack+0x24/0x50
  kasan_set_track+0x24/0x38
  kasan_set_free_info+0x20/0x40
  __kasan_slab_free+0x10c/0x170
  kasan_slab_free+0x10/0x18
  kfree+0x98/0x270
  kasan_workqueue_work+0xc/0x18

 Last potentially related work creation:
  kasan_save_stack+0x24/0x50
  kasan_record_wq_stack+0xa8/0xb8
  insert_work+0x48/0x288
  __queue_work+0x3e8/0xc40
  queue_work_on+0xf4/0x118
  kasan_workqueue_uaf+0xfc/0x190

Link: https://lkml.kernel.org/r/20201203022748.30681-1-walter-zh.wu@mediatek.com
Signed-off-by: Walter Wu <walter-zh.wu@mediatek.com>
Acked-by: Marco Elver <elver@google.com>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kasan: print workqueue stack

The aux_stack[2] is reused to record the call_rcu() call stack and
enqueuing work call stacks.  So that we need to change the auxiliary stack
title for common title, print them in KASAN report.

Link: https://lkml.kernel.org/r/20201203022715.30635-1-walter-zh.wu@mediatek.com
Signed-off-by: Walter Wu <walter-zh.wu@mediatek.com>
Suggested-by: Marco Elver <elver@google.com>
Acked-by: Marco Elver <elver@google.com>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

workqueue: kasan: record workqueue stack

Patch series "kasan: add workqueue stack for generic KASAN", v5.

Syzbot reports many UAF issues for workqueue, see [1].

In some of these access/allocation happened in process_one_work(), we
see the free stack is useless in KASAN report, it doesn't help
programmers to solve UAF for workqueue issue.

This patchset improves KASAN reports by making them to have workqueue
queueing stack.  It is useful for programmers to solve use-after-free or
double-free memory issue.

Generic KASAN also records the last two workqueue stacks and prints them
in KASAN report.  It is only suitable for generic KASAN.

[1] https://groups.google.com/g/syzkaller-bugs/search?q=%22use-after-free%22+process_one_work
[2] https://bugzilla.kernel.org/show_bug.cgi?id=198437

This patch (of 4):

When analyzing use-after-free or double-free issue, recording the
enqueuing work stacks is helpful to preserve usage history which
potentially gives a hint about the affected code.

For workqueue it has turned out to be useful to record the enqueuing work
call stacks.  Because user can see KASAN report to determine whether it is
root cause.  They don't need to enable debugobjects, but they have a
chance to find out the root cause.

Link: https://lkml.kernel.org/r/20201203022148.29754-1-walter-zh.wu@mediatek.com
Link: https://lkml.kernel.org/r/20201203022442.30006-1-walter-zh.wu@mediatek.com
Signed-off-by: Walter Wu <walter-zh.wu@mediatek.com>
Suggested-by: Marco Elver <elver@google.com>
Acked-by: Marco Elver <elver@google.com>
Acked-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Marco Elver <elver@google.com>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc.c: fix kasan shadow poisoning size

The size of vm area can be affected by the presence or not of the guard
page.  In particular when VM_NO_GUARD is present, the actual accessible
size has to be considered like the real size minus the guard page.

Currently kasan does not keep into account this information during the
poison operation and in particular tries to poison the guard page as well.

This approach, even if incorrect, does not cause an issue because the tags
for the guard page are written in the shadow memory.  With the future
introduction of the Tag-Based KASAN, being the guard page inaccessible by
nature, the write tag operation on this page triggers a fault.

Fix kasan shadow poisoning size invoking get_vm_area_size() instead of
accessing directly the field in the data structure to detect the correct
value.

Link: https://lkml.kernel.org/r/20201027160213.32904-1-vincenzo.frascino@arm.com
Fixes: d98c9e83b5e7c ("kasan: fix crashes on access to memory mapped by vm_map_ram()")
Signed-off-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Marco Elver <elver@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

docs/vm: remove unused 3 items explanation for /proc/vmstat

Commit 5647bc293ab1 ("mm: compaction: Move migration fail/success
stats to migrate.c"), removed 3 items in /proc/vmstat. but the docs
still has their explanation. let's remove them.

"compact_blocks_moved",
"compact_pages_moved",
"compact_pagemigrate_failed",

Link: https://lkml.kernel.org/r/1605520282-51993-1-git-send-email-alex.shi@linux.alibaba.com
Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com>
Reviewed-by: Zi Yan <ziy@nvidia.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc: Fix unlock order in s_stop()

When multiple locks are acquired, they should be released in reverse
order. For s_start() and s_stop() in mm/vmalloc.c, that is not the
case.

  s_start: mutex_lock(&vmap_purge_lock); spin_lock(&vmap_area_lock);
  s_stop : mutex_unlock(&vmap_purge_lock); spin_unlock(&vmap_area_lock);

This unlock sequence, though allowed, is not optimal. If a waiter is
present, mutex_unlock() will need to go through the slowpath of waking
up the waiter with preemption disabled. Fix that by releasing the
spinlock first before the mutex.

Link: https://lkml.kernel.org/r/20201213180843.16938-1-longman@redhat.com
Fixes: e36176be1c39 ("mm/vmalloc: rework vmap_area_lock")
Signed-off-by: Waiman Long <longman@redhat.com>
Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc.c: remove unnecessary return statement

Remove unnecessary return statement for void function.

Link: https://lkml.kernel.org/r/ca23f89259c80c3562700ae6e227b2815a195853.1606891153.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc: add 'align' parameter explanation for pvm_determine_end_from_reverse

Kernel-doc markup has a issue on pvm_determine_end_from_reverse:

  mm/vmalloc.c:3145: warning: Function parameter or member 'align' not described in 'pvm_determine_end_from_reverse'

Add a explanation for it to remove the warning.

Link: https://lkml.kernel.org/r/1605605088-30668-3-git-send-email-alex.shi@linux.alibaba.com
Signed-off-by: Alex Shi <alex.shi@linux.alibaba.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc: rework the drain logic

A current "lazy drain" model suffers from at least two issues.

First one is related to the unsorted list of vmap areas, thus in order to
identify the [min:max] range of areas to be drained, it requires a full
list scan.  What is a time consuming if the list is too long.

Second one and as a next step is about merging all fragments with a free
space.  What is also a time consuming because it has to iterate over
entire list which holds outstanding lazy areas.

See below the "preemptirqsoff" tracer that illustrates a high latency.  It
is ~24676us.  Our workloads like audio and video are effected by such long
latency:

<snip>
  tracer: preemptirqsoff

  preemptirqsoff latency trace v1.1.5 on 4.9.186-perf+
  --------------------------------------------------------------------
  latency: 24676 us, #4/4, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 P:8)
     -----------------
     | task: crtc_commit:112-261 (uid:0 nice:0 policy:1 rt_prio:16)
     -----------------
   => started at: __purge_vmap_area_lazy
   => ended at:   __purge_vmap_area_lazy

                   _------=> CPU#
                  / _-----=> irqs-off
                 | / _----=> need-resched
                 || / _---=> hardirq/softirq
                 ||| / _--=> preempt-depth
                 |||| /     delay
   cmd     pid   ||||| time  |   caller
      \   /      |||||  \    |   /
crtc_com-261     1...1    1us*: _raw_spin_lock <-__purge_vmap_area_lazy
[...]
crtc_com-261     1...1 24675us : _raw_spin_unlock <-__purge_vmap_area_lazy
crtc_com-261     1...1 24677us : trace_preempt_on <-__purge_vmap_area_lazy
crtc_com-261     1...1 24683us : <stack trace>
 => free_vmap_area_noflush
 => remove_vm_area
 => __vunmap
 => vfree
 => drm_property_free_blob
 => drm_mode_object_unreference
 => drm_property_unreference_blob
 => __drm_atomic_helper_crtc_destroy_state
 => sde_crtc_destroy_state
 => drm_atomic_state_default_clear
 => drm_atomic_state_clear
 => drm_atomic_state_free
 => complete_commit
 => _msm_drm_commit_work_cb
 => kthread_worker_fn
 => kthread
 => ret_from_fork
<snip>

To address those two issues we can redesign a purging of the outstanding
lazy areas.  Instead of queuing vmap areas to the list, we replace it by
the separate rb-tree.  In hat case an area is located in the tree/list in
ascending order.  It will give us below advantages:

a) Outstanding vmap areas are merged creating bigger coalesced blocks,
   thus it becomes less fragmented.

b) It is possible to calculate a flush range [min:max] without scanning
   all elements.  It is O(1) access time or complexity;

c) The final merge of areas with the rb-tree that represents a free
   space is faster because of (a).  As a result the lock contention is
   also reduced.

Link: https://lkml.kernel.org/r/20201116220033.1837-2-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Oleksiy Avramchenko <oleksiy.avramchenko@sonymobile.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: huang ying <huang.ying.caritas@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc: use free_vm_area() if an allocation fails

There is a dedicated and separate function that finds and removes a
continuous kernel virtual area.  As a final step it also releases the
"area", a descriptor of corresponding vm_struct.

Use free_vmap_area() in the __vmalloc_node_range() instead of open coded
steps which are exactly the same, to perform a cleanup.

Link: https://lkml.kernel.org/r/20201116220033.1837-1-urezki@gmail.com
Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Oleksiy Avramchenko <oleksiy.avramchenko@sonymobile.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmalloc.c:__vmalloc_area_node(): avoid 32-bit overflow

With a machine with 3 TB (more than 2 TB memory).  If you use vmalloc to
allocate > 2 TB memory, the array_size below will be overflowed.

The array_size is an unsigned int and can only be used to allocate less
than 2 TB memory.  If you pass 2*1028*1028*1024*1024 = 2 * 2^40 in the
argument of vmalloc.  The array_size will become 2*2^31 = 2^32.  The 2^32
cannot be store with a 32 bit integer.

The fix is to change the type of array_size to unsigned long.

[akpm@linux-foundation.org: rework for current mainline]

Link: https://bugzilla.kernel.org/show_bug.cgi?id=210023
Reported-by: <hsinhuiwu@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

locking/selftests: add testcases for fs_reclaim

Since I butchered this I figured better to make sure we have testcases for
this now.  Since we only have a locking context for __GFP_FS that's the
only thing we're testing right now.

Link: https://lkml.kernel.org/r/20201125162532.1299794-4-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Thomas Hellström (Intel) <thomas_os@shipmail.org>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Will Deacon <will@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Waiman Long <longman@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: extract might_alloc() debug check

Extracted from slab.h, which seems to have the most complete version
including the correct might_sleep() check.  Roll it out to slob.c.

Motivated by a discussion with Paul about possibly changing call_rcu
behaviour to allocate memory, but only roughly every 500th call.

There are a lot fewer places in the kernel that care about whether
allocating memory is allowed or not (due to deadlocks with reclaim code)
than places that care whether sleeping is allowed.  But debugging these
also tends to be a lot harder, so nice descriptive checks could come in
handy.  I might have some use eventually for annotations in drivers/gpu.

Note that unlike fs_reclaim_acquire/release gfpflags_allow_blocking does
not consult the PF_MEMALLOC flags.  But there is no flag equivalent for
GFP_NOWAIT, hence this check can't go wrong due to
memalloc_no*_save/restore contexts.  Willy is working on a patch series
which might change this:

https://lore.kernel.org/linux-mm/20200625113122.7540-7-willy@infradead.org/

I think best would be if that updates gfpflags_allow_blocking(), since
there's a ton of callers all over the place for that already.

Link: https://lkml.kernel.org/r/20201125162532.1299794-3-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Michel Lespinasse <walken@google.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Waiman Long <longman@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Qian Cai <cai@lca.pw>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: Christian König <christian.koenig@amd.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Thomas Hellström (Intel) <thomas_os@shipmail.org>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: track mmu notifiers in fs_reclaim_acquire/release

fs_reclaim_acquire/release nicely catch recursion issues when allocating
GFP_KERNEL memory against shrinkers (which gpu drivers tend to use to keep
the excessive caches in check).  For mmu notifier recursions we do have
lockdep annotations since 23b68395c7c7 ("mm/mmu_notifiers: add a lockdep
map for invalidate_range_start/end").

But these only fire if a path actually results in some pte invalidation -
for most small allocations that's very rarely the case.  The other trouble
is that pte invalidation can happen any time when __GFP_RECLAIM is set.
Which means only really GFP_ATOMIC is a safe choice, GFP_NOIO isn't good
enough to avoid potential mmu notifier recursion.

I was pondering whether we should just do the general annotation, but
there's always the risk for false positives.  Plus I'm assuming that the
core fs and io code is a lot better reviewed and tested than random mmu
notifier code in drivers.  Hence why I decide to only annotate for that
specific case.

Furthermore even if we'd create a lockdep map for direct reclaim, we'd
still need to explicit pull in the mmu notifier map - there's a lot more
places that do pte invalidation than just direct reclaim, these two
contexts arent the same.

Note that the mmu notifiers needing their own independent lockdep map is
also the reason we can't hold them from fs_reclaim_acquire to
fs_reclaim_release - it would nest with the acquistion in the pte
invalidation code, causing a lockdep splat.  And we can't remove the
annotations from pte invalidation and all the other places since they're
called from many other places than page reclaim.  Hence we can only do the
equivalent of might_lock, but on the raw lockdep map.

With this we can also remove the lockdep priming added in 66204f1d2d1b
("mm/mmu_notifiers: prime lockdep") since the new annotations are strictly
more powerful.

Link: https://lkml.kernel.org/r/20201125162532.1299794-2-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Qian Cai <cai@lca.pw>
Cc: Thomas Hellström (Intel) <thomas_os@shipmail.org>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Waiman Long <longman@redhat.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: forbid splitting special mappings

Don't allow splitting of vm_special_mapping's.  It affects vdso/vvar
areas.  Uprobes have only one page in xol_area so they aren't affected.

Those restrictions were enforced by checks in .mremap() callbacks.
Restrict resizing with generic .split() callback.

Link: https://lkml.kernel.org/r/20201013013416.390574-7-dima@arista.com
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mremap: check if it's possible to split original vma

If original VMA can't be split at the desired address, do_munmap() will
fail and leave both new-copied VMA and old VMA.  De-facto it's
MREMAP_DONTUNMAP behaviour, which is unexpected.

Currently, it may fail such way for hugetlbfs and dax device mappings.

Minimize such unpleasant situations to OOM by checking .may_split() before
attempting to create a VMA copy.

Link: https://lkml.kernel.org/r/20201013013416.390574-6-dima@arista.com
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

vm_ops: rename .split() callback to .may_split()

Rename the callback to reflect that it's not called *on* or *after* split,
but rather some time before the splitting to check if it's possible.

Link: https://lkml.kernel.org/r/20201013013416.390574-5-dima@arista.com
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mremap: don't allow MREMAP_DONTUNMAP on special_mappings and aio

As kernel expect to see only one of such mappings, any further operations
on the VMA-copy may be unexpected by the kernel.  Maybe it's being on the
safe side, but there doesn't seem to be any expected use-case for this, so
restrict it now.

Link: https://lkml.kernel.org/r/20201013013416.390574-4-dima@arista.com
Fixes: commit e346b3813067 ("mm/mremap: add MREMAP_DONTUNMAP to mremap()")
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/mremap: for MREMAP_DONTUNMAP check security_vm_enough_memory_mm()

Currently memory is accounted post-mremap() with MREMAP_DONTUNMAP, which
may break overcommit policy.  So, check if there's enough memory before
doing actual VMA copy.

Don't unset VM_ACCOUNT on MREMAP_DONTUNMAP.  By semantics, such mremap()
is actually a memory allocation.  That also simplifies the error-path a
little.

Also, as it's memory allocation on success don't reset hiwater_vm value.

Link: https://lkml.kernel.org/r/20201013013416.390574-3-dima@arista.com
Fixes: commit e346b3813067 ("mm/mremap: add MREMAP_DONTUNMAP to mremap()")
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/mremap: account memory on do_munmap() failure

Patch series "mremap: move_vma() fixes".

This patch (of 6):

move_vma() copies VMA without adding it to account, then unmaps old part
of VMA.  On failure it unmaps the new VMA.  With hacks accounting in
munmap is disabled as it's a copy of existing VMA.

Account the memory on munmap() failure which was previously copied into
a new VMA.

Link: https://lkml.kernel.org/r/20201013013416.390574-1-dima@arista.com
Link: https://lkml.kernel.org/r/20201013013416.390574-2-dima@arista.com
Fixes: commit e2ea83742133 ("[PATCH] mremap: move_vma fixes and cleanup")
Signed-off-by: Dmitry Safonov <dima@arista.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Brian Geffon <bgeffon@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Will Deacon <will@kernel.org>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: move free_unref_page to mm/internal.h

Code outside mm/ should not be calling free_unref_page().  Also move
free_unref_page_list().

Link: https://lkml.kernel.org/r/20201125034655.27687-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

sparc: fix handling of page table constructor failure

The page has just been allocated, so its refcount is 1.  free_unref_page()
is for use on pages which have a zero refcount.  Use __free_page() like
the other implementations of pte_alloc_one().

Link: https://lkml.kernel.org/r/20201125034655.27687-1-willy@infradead.org
Fixes: 1ae9ae5f7df7 ("sparc: handle pgtable_page_ctor() fail")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm: mmap_lock: add tracepoints around lock acquisition

The goal of these tracepoints is to be able to debug lock contention
issues.  This lock is acquired on most (all?) mmap / munmap / page fault
operations, so a multi-threaded process which does a lot of these can
experience significant contention.

We trace just before we start acquisition, when the acquisition returns
(whether it succeeded or not), and when the lock is released (or
downgraded).  The events are broken out by lock type (read / write).

The events are also broken out by memcg path.  For container-based
workloads, users often think of several processes in a memcg as a single
logical "task", so collecting statistics at this level is useful.

The end goal is to get latency information.  This isn't directly included
in the trace events.  Instead, users are expected to compute the time
between "start locking" and "acquire returned", using e.g.  synthetic
events or BPF.  The benefit we get from this is simpler code.

Because we use tracepoint_enabled() to decide whether or not to trace,
this patch has effectively no overhead unless tracepoints are enabled at
runtime.  If tracepoints are enabled, there is a performance impact, but
how much depends on exactly what e.g.  the BPF program does.

[axelrasmussen@google.com: fix use-after-free race and css ref leak in tracepoints]
Link: https://lkml.kernel.org/r/20201130233504.3725241-1-axelrasmussen@google.com
[axelrasmussen@google.com: v3]
Link: https://lkml.kernel.org/r/20201207213358.573750-1-axelrasmussen@google.com
[rostedt@goodmis.org: in-depth examples of tracepoint_enabled() usage, and per-cpu-per-context buffer design]

Link: https://lkml.kernel.org/r/20201105211739.568279-2-axelrasmussen@google.com
Signed-off-by: Axel Rasmussen <axelrasmussen@google.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Jann Horn <jannh@google.com>
Cc: Chinwen Chang <chinwen.chang@mediatek.com>
Cc: Davidlohr Bueso <dbueso@suse.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Yafang Shao <laoar.shao@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>