mm: memory-failure: refactor add_to_kill()

Patch series "mm: ksm: support hwpoison for ksm page", v2.

Currently, ksm does not support hwpoison.  As ksm is being used more
widely for deduplication at the system level, container level, and process
level, supporting hwpoison for ksm has become increasingly important.
However, ksm pages were not processed by hwpoison in 2009 [1].

The main method of implementation:

1. Refactor add_to_kill() and add new add_to_kill_*() to better
   accommodate the handling of different types of pages.

2.  Add collect_procs_ksm() to collect processes when the error hit an
   ksm page.

3. Add task_in_to_kill_list() to avoid duplicate addition of tsk to
   the to_kill list.

4. Try_to_unmap ksm page (already supported).

5. Handle related processes such as sending SIGBUS.

Tested with poisoning to ksm page from
1) different process
2) one process

and with/without memory_failure_early_kill set, the processes are killed
as expected with the patchset.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/
commit/?h=01e00f880ca700376e1845cf7a2524ebe68e47d6

This patch (of 2):

The page_address_in_vma() is used to find the user virtual address of page
in add_to_kill(), but it doesn't support ksm due to the ksm page->index
unusable, add an ksm_addr as parameter to add_to_kill(), let's the caller
to pass it, also rename the function to __add_to_kill(), and adding
add_to_kill_anon_file() for handling anonymous pages and file pages,
adding add_to_kill_fsdax() for handling fsdax pages.

Link: https://lkml.kernel.org/r/20230414021741.2597273-1-xialonglong1@huawei.com
Link: https://lkml.kernel.org/r/20230414021741.2597273-2-xialonglong1@huawei.com
Signed-off-by: Longlong Xia <xialonglong1@huawei.com>
Tested-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Reviewed-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/memfd: fix test_sysctl

sysctl memfd_noexec is pid-namespaced, non-reservable, and inherent to the
child process.

Move the inherence test from init ns to child ns, so init ns can keep the
default value.

Link: https://lkml.kernel.org/r/20230414022801.2545257-1-jeffxu@google.com
Signed-off-by: Jeff Xu <jeffxu@google.com>
Reported-by: kernel test robot <yujie.liu@intel.com>
Link: https://lore.kernel.org/oe-lkp/202303312259.441e35db-yujie.liu@intel.com
Tested-by: Yujie Liu <yujie.liu@intel.com>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Jorge Lucangeli Obes <jorgelo@chromium.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: run hugetlb testcases of va switch

The va_high_addr_switch selftest is used to test mmap across 128TB
boundary.  It divides the selftest cases into two main categories on the
basis of size.  One set is used to create mappings that are multiples of
PAGE_SIZE while the other creates mappings that are multiples of
HUGETLB_SIZE.

In order to run the hugetlb testcases the binary must be appended with
"--run-hugetlb" but the file that used to run the test only invokes the
binary, thereby completely skipping the hugetlb testcases.  Hence, the
required statement has been added.

Link: https://lkml.kernel.org/r/20230323105243.2807166-6-chaitanyas.prakash@arm.com
Signed-off-by: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: configure nr_hugepages for arm64

Arm64 has a default hugepage size of 512MB when CONFIG_ARM64_64K_PAGES=y
is enabled.  While testing on arm64 platforms having up to 4PB of virtual
address space, a minimum of 6 hugepages were required for all test cases
to pass.  Support for this requirement has been added.

Link: https://lkml.kernel.org/r/20230323105243.2807166-5-chaitanyas.prakash@arm.com
Signed-off-by: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: add platform independent in code comments

The in code comments for the selftest were made on the basis of 128TB
switch, an architecture feature specific to PowerPc and x86 platforms.
Keeping in mind the support added for arm64 platforms which implements a
256TB switch, a more generic explanation has been provided.

Link: https://lkml.kernel.org/r/20230323105243.2807166-4-chaitanyas.prakash@arm.com
Signed-off-by: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: rename va_128TBswitch to va_high_addr_switch

As the initial selftest only took into consideration PowperPC and x86
architectures, on adding support for arm64, a platform independent naming
convention is chosen.

Link: https://lkml.kernel.org/r/20230323105243.2807166-3-chaitanyas.prakash@arm.com
Signed-off-by: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: add support for arm64 platform on va switch

Patch series "selftests/mm: Implement support for arm64 on va".

The va_128TBswitch selftest is designed and implemented for PowerPC and
x86 architectures which support a 128TB switch, up to 256TB of virtual
address space and hugepage sizes of 16MB and 2MB respectively.  Arm64
platforms on the other hand support a 256Tb switch, up to 4PB of virtual
address space and a default hugepage size of 512MB when 64k pagesize is
enabled.

These architectural differences require introducing support for arm64
platforms, after which a more generic naming convention is suggested.  The
in code comments are amended to provide a more platform independent
explanation of the working of the code and nr_hugepages are configured as
required.  Finally, the file running the testcase is modified in order to
prevent skipping of hugetlb testcases of va_high_addr_switch.

This patch (of 5):

Arm64 platforms have the ability to support 64kb pagesize, 512MB default
hugepage size and up to 4PB of virtual address space.  The address switch
occurs at 256TB as opposed to 128TB.  Hence, the necessary support has
been added.

Link: https://lkml.kernel.org/r/20230323105243.2807166-1-chaitanyas.prakash@arm.com
Link: https://lkml.kernel.org/r/20230323105243.2807166-2-chaitanyas.prakash@arm.com
Signed-off-by: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

memfd: pass argument of memfd_fcntl as int

The interface for fcntl expects the argument passed for the command
F_ADD_SEALS to be of type int.  The current code wrongly treats it as a
long.  In order to avoid access to undefined bits, we should explicitly
cast the argument to int.

This commit changes the signature of all the related and helper functions
so that they treat the argument as int instead of long.

Link: https://lkml.kernel.org/r/20230414152459.816046-5-Luca.Vizzarro@arm.com
Signed-off-by: Luca Vizzarro <Luca.Vizzarro@arm.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Jeff Layton <jlayton@kernel.org>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Kevin Brodsky <Kevin.Brodsky@arm.com>
Cc: Vincenzo Frascino <Vincenzo.Frascino@arm.com>
Cc: Szabolcs Nagy <Szabolcs.Nagy@arm.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: David Laight <David.Laight@ACULAB.com>
Cc: Mark Rutland <Mark.Rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: Multi-gen LRU: remove wait_event_killable()

Android 14 and later default to MGLRU [1] and field telemetry showed
occasional long tail latency (>100ms) in the reclaim path.

Tracing revealed priority inversion in the reclaim path.  In
try_to_inc_max_seq(), when high priority tasks were blocked on
wait_event_killable(), the preemption of the low priority task to call
wake_up_all() caused those high priority tasks to wait longer than
necessary.  In general, this problem is not different from others of its
kind, e.g., one caused by mutex_lock().  However, it is specific to MGLRU
because it introduced the new wait queue lruvec->mm_state.wait.

The purpose of this new wait queue is to avoid the thundering herd
problem.  If many direct reclaimers rush into try_to_inc_max_seq(), only
one can succeed, i.e., the one to wake up the rest, and the rest who
failed might cause premature OOM kills if they do not wait.  So far there
is no evidence supporting this scenario, based on how often the wait has
been hit.  And this begs the question how useful the wait queue is in
practice.

Based on Minchan's recommendation, which is in line with his commit
6d4675e60135 ("mm: don't be stuck to rmap lock on reclaim path") and the
rest of the MGLRU code which also uses trylock when possible, remove the
wait queue.

[1] https://android-review.googlesource.com/q/I7ed7fbfd6ef9ce10053347528125dd98c39e50bf

Link: https://lkml.kernel.org/r/20230413214326.2147568-1-kaleshsingh@google.com
Fixes: bd74fdaea146 ("mm: multi-gen LRU: support page table walks")
Signed-off-by: Kalesh Singh <kaleshsingh@google.com>
Suggested-by: Minchan Kim <minchan@kernel.org>
Reported-by: Wei Wang <wvw@google.com>
Acked-by: Yu Zhao <yuzhao@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Jan Alexander Steffens (heftig) <heftig@archlinux.org>
Cc: Oleksandr Natalenko <oleksandr@natalenko.name>
Cc: Suleiman Souhlal <suleiman@google.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: workingset: update description of the source file

The calculation of workingset size is the core logic of handling refault,
it had been updated several times[1][2] after workingset.c was created[3].
But the description hadn't been updated accordingly, this mismatch may
confuse the readers.  So we update the description to make it consistent
to the code.

[1] commit 34e58cac6d8f ("mm: workingset: let cache workingset challenge anon")
[2] commit aae466b0052e ("mm/swap: implement workingset detection for anonymous LRU")
[3] commit a528910e12ec ("mm: thrash detection-based file cache sizing")

Link: https://lkml.kernel.org/r/202304131634494948454@zte.com.cn
Signed-off-by: Yang Yang <yang.yang29@zte.com.cn>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

printk: export console trace point for kcsan/kasan/kfence/kmsan

The console tracepoint is used by kcsan/kasan/kfence/kmsan test modules.
Since this tracepoint is not exported, these modules iterate over all
available tracepoints to find the console trace point.  Export the trace
point so that it can be directly used.

Link: https://lkml.kernel.org/r/20230413100859.1492323-1-quic_pkondeti@quicinc.com
Signed-off-by: Pavankumar Kondeti <quic_pkondeti@quicinc.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: John Ogness <john.ogness@linutronix.de>
Cc: Marco Elver <elver@google.com>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: vmscan: refactor updating current->reclaim_state

During reclaim, we keep track of pages reclaimed from other means than
LRU-based reclaim through scan_control->reclaim_state->reclaimed_slab,
which we stash a pointer to in current task_struct.

However, we keep track of more than just reclaimed slab pages through
this.  We also use it for clean file pages dropped through pruned inodes,
and xfs buffer pages freed.  Rename reclaimed_slab to reclaimed, and add a
helper function that wraps updating it through current, so that future
changes to this logic are contained within include/linux/swap.h.

Link: https://lkml.kernel.org/r/20230413104034.1086717-4-yosryahmed@google.com
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Lameter <cl@linux.com>
Cc: Darrick J. Wong <djwong@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: NeilBrown <neilb@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: vmscan: move set_task_reclaim_state() near flush_reclaim_state()

Move set_task_reclaim_state() near flush_reclaim_state() so that all
helpers manipulating reclaim_state are in close proximity.

Link: https://lkml.kernel.org/r/20230413104034.1086717-3-yosryahmed@google.com
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Lameter <cl@linux.com>
Cc: Darrick J. Wong <djwong@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: NeilBrown <neilb@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: vmscan: ignore non-LRU-based reclaim in memcg reclaim

Patch series "Ignore non-LRU-based reclaim in memcg reclaim", v6.

Upon running some proactive reclaim tests using memory.reclaim, we noticed
some tests flaking where writing to memory.reclaim would be successful
even though we did not reclaim the requested amount fully Looking further
into it, I discovered that *sometimes* we overestimate the number of
reclaimed pages in memcg reclaim.

Reclaimed pages through other means than LRU-based reclaim are tracked
through reclaim_state in struct scan_control, which is stashed in current
task_struct.  These pages are added to the number of reclaimed pages
through LRUs.  For memcg reclaim, these pages generally cannot be linked
to the memcg under reclaim and can cause an overestimated count of
reclaimed pages.  This short series tries to address that.

Patch 1 ignores pages reclaimed outside of LRU reclaim in memcg reclaim.
The pages are uncharged anyway, so even if we end up under-reporting
reclaimed pages we will still succeed in making progress during charging.

Patches 2-3 are just refactoring.  Patch 2 moves set_reclaim_state()
helper next to flush_reclaim_state().  Patch 3 adds a helper that wraps
updating current->reclaim_state, and renames reclaim_state->reclaimed_slab
to reclaim_state->reclaimed.

This patch (of 3):

We keep track of different types of reclaimed pages through
reclaim_state->reclaimed_slab, and we add them to the reported number of
reclaimed pages.  For non-memcg reclaim, this makes sense.  For memcg
reclaim, we have no clue if those pages are charged to the memcg under
reclaim.

Slab pages are shared by different memcgs, so a freed slab page may have
only been partially charged to the memcg under reclaim.  The same goes for
clean file pages from pruned inodes (on highmem systems) or xfs buffer
pages, there is no simple way to currently link them to the memcg under
reclaim.

Stop reporting those freed pages as reclaimed pages during memcg reclaim.
This should make the return value of writing to memory.reclaim, and may
help reduce unnecessary reclaim retries during memcg charging.  Writing to
memory.reclaim on the root memcg is considered as cgroup_reclaim(), but
for this case we want to include any freed pages, so use the
global_reclaim() check instead of !cgroup_reclaim().

Generally, this should make the return value of
try_to_free_mem_cgroup_pages() more accurate.  In some limited cases (e.g.
freed a slab page that was mostly charged to the memcg under reclaim),
the return value of try_to_free_mem_cgroup_pages() can be underestimated,
but this should be fine.  The freed pages will be uncharged anyway, and we
can charge the memcg the next time around as we usually do memcg reclaim
in a retry loop.

Link: https://lkml.kernel.org/r/20230413104034.1086717-1-yosryahmed@google.com
Link: https://lkml.kernel.org/r/20230413104034.1086717-2-yosryahmed@google.com
Fixes: f2fe7b09a52b ("mm: memcg/slab: charge individual slab objects
instead of pages")
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Lameter <cl@linux.com>
Cc: Darrick J. Wong <djwong@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: NeilBrown <neilb@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: apply __must_check to vmap_pages_range_noflush()

To prevent errors when vmap_pages_range_noflush() or
__vmap_pages_range_noflush() silently fail (see the link below for an
example), annotate them with __must_check so that the callers do not
unconditionally assume the mapping succeeded.

Link: https://lkml.kernel.org/r/20230413131223.4135168-4-glider@google.com
Signed-off-by: Alexander Potapenko <glider@google.com>
Reported-by: Dipanjan Das <mail.dipanjan.das@gmail.com>
Link: https://lore.kernel.org/linux-mm/CANX2M5ZRrRA64k0hOif02TjmY9kbbO2aCBPyq79es34RXZ=cAw@mail.gmail.com/
Reviewed-by: Marco Elver <elver@google.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Uladzislau Rezki (Sony) <urezki@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: kmsan: apply __must_check to non-void functions

Non-void KMSAN hooks may return error codes that indicate that KMSAN
failed to reflect the changed memory state in the metadata (e.g.  it could
not create the necessary memory mappings).  In such cases the callers
should handle the errors to prevent the tool from using the inconsistent
metadata in the future.

We mark non-void hooks with __must_check so that error handling is not
skipped.

Link: https://lkml.kernel.org/r/20230413131223.4135168-3-glider@google.com
Signed-off-by: Alexander Potapenko <glider@google.com>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dipanjan Das <mail.dipanjan.das@gmail.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Uladzislau Rezki (Sony) <urezki@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: hwpoison: support recovery from HugePage copy-on-write faults

copy-on-write of hugetlb user pages with uncorrectable errors will result
in a kernel crash.  This is because the copy is performed in kernel mode
and in general we can not handle accessing memory with such errors while
in kernel mode.  Commit a873dfe1032a ("mm, hwpoison: try to recover from
copy-on write faults") introduced the routine copy_user_highpage_mc() to
gracefully handle copying of user pages with uncorrectable errors.
However, the separate hugetlb copy-on-write code paths were not modified
as part of commit a873dfe1032a.

Modify hugetlb copy-on-write code paths to use copy_mc_user_highpage() so
that they can also gracefully handle uncorrectable errors in user pages.
This involves changing the hugetlb specific routine
copy_user_large_folio() from type void to int so that it can return an
error.  Modify the hugetlb userfaultfd code in the same way so that it can
return -EHWPOISON if it encounters an uncorrectable error.

Link: https://lkml.kernel.org/r/20230413131349.2524210-1-liushixin2@huawei.com
Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Acked-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

memcg: page_cgroup_ino() get memcg from the page's folio

In a kernel with added WARN_ON_ONCE(PageTail) in page_memcg_check(), we
observed a warning from page_cgroup_ino() when reading /proc/kpagecgroup.
This warning was added to catch fragile reads of a page memcg.  Make
page_cgroup_ino() get memcg from the page's folio using
folio_memcg_check(): that gives it the correct memcg for each page of a
folio, so is the right fix.

Note that page_folio() is racy, the page's folio can change from under us,
but the entire function is racy and documented as such.

I dithered between the right fix and the safer "fix": it's unlikely but
conceivable that some userspace has learnt that /proc/kpagecgroup gives no
memcg on tail pages, and compensates for that in some (racy) way: so
continuing to give no memcg on tails, without warning, might be safer.

But hwpoison_filter_task(), the only other user of page_cgroup_ino(),
persuaded me.  It looks as if it currently leaves out tail pages of the
selected memcg, by mistake: whereas hwpoison_inject() uses compound_head()
and expects the tails to be included.  So hwpoison testing coverage has
probably been restricted by the wrong output from page_cgroup_ino() (if
that memcg filter is used at all): in the short term, it might be safer
not to enable wider coverage there, but long term we would regret that.

This is based on a patch originally written by Hugh Dickins and retains
most of the original commit log [1]

The patch was changed to use folio_memcg_check(page_folio(page)) instead
of page_memcg_check(compound_head(page)) based on discussions with Matthew
Wilcox; where he stated that callers of page_memcg_check() should stop
using it due to the ambiguity around tail pages -- instead they should use
folio_memcg_check() and handle tail pages themselves.

Link: https://lkml.kernel.org/r/20230412003451.4018887-1-yosryahmed@google.com
Link: https://lore.kernel.org/linux-mm/20230313083452.1319968-1-yosryahmed@google.com/
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/hugetlb_vmemmap: rename ARCH_WANT_HUGETLB_PAGE_OPTIMIZE_VMEMMAP

Now we use ARCH_WANT_HUGETLB_PAGE_OPTIMIZE_VMEMMAP config option to
indicate devdax and hugetlb vmemmap optimization support.  Hence rename
that to a generic ARCH_WANT_OPTIMIZE_VMEMMAP

Link: https://lkml.kernel.org/r/20230412050025.84346-2-aneesh.kumar@linux.ibm.com
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Reviewed-by: Muchun Song <songmuchun@bytedance.com>
Cc: Joao Martins <joao.m.martins@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Tarun Sahu <tsahu@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/vmemmap/devdax: fix kernel crash when probing devdax devices

commit 4917f55b4ef9 ("mm/sparse-vmemmap: improve memory savings for
compound devmaps") added support for using optimized vmmemap for devdax
devices.  But how vmemmap mappings are created are architecture specific.
For example, powerpc with hash translation doesn't have vmemmap mappings
in init_mm page table instead they are bolted table entries in the
hardware page table

vmemmap_populate_compound_pages() used by vmemmap optimization code is not
aware of these architecture-specific mapping.  Hence allow architecture to
opt for this feature.  I selected architectures supporting
HUGETLB_PAGE_OPTIMIZE_VMEMMAP option as also supporting this feature.

This patch fixes the below crash on ppc64.

BUG: Unable to handle kernel data access on write at 0xc00c000100400038
Faulting instruction address: 0xc000000001269d90
Oops: Kernel access of bad area, sig: 11 [#1]
LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
Modules linked in:
CPU: 7 PID: 1 Comm: swapper/0 Not tainted 6.3.0-rc5-150500.34-default+ #2 5c90a668b6bbd142599890245c2fb5de19d7d28a
Hardware name: IBM,9009-42G POWER9 (raw) 0x4e0202 0xf000005 of:IBM,FW950.40 (VL950_099) hv:phyp pSeries
NIP:  c000000001269d90 LR: c0000000004c57d4 CTR: 0000000000000000
REGS: c000000003632c30 TRAP: 0300   Not tainted  (6.3.0-rc5-150500.34-default+)
MSR:  8000000000009033 <SF,EE,ME,IR,DR,RI,LE>  CR: 24842228  XER: 00000000
CFAR: c0000000004c57d0 DAR: c00c000100400038 DSISR: 42000000 IRQMASK: 0
....
NIP [c000000001269d90] __init_single_page.isra.74+0x14/0x4c
LR [c0000000004c57d4] __init_zone_device_page+0x44/0xd0
Call Trace:
[c000000003632ed0] [c000000003632f60] 0xc000000003632f60 (unreliable)
[c000000003632f10] [c0000000004c5ca0] memmap_init_zone_device+0x170/0x250
[c000000003632fe0] [c0000000005575f8] memremap_pages+0x2c8/0x7f0
[c0000000036330c0] [c000000000557b5c] devm_memremap_pages+0x3c/0xa0
[c000000003633100] [c000000000d458a8] dev_dax_probe+0x108/0x3e0
[c0000000036331a0] [c000000000d41430] dax_bus_probe+0xb0/0x140
[c0000000036331d0] [c000000000cef27c] really_probe+0x19c/0x520
[c000000003633260] [c000000000cef6b4] __driver_probe_device+0xb4/0x230
[c0000000036332e0] [c000000000cef888] driver_probe_device+0x58/0x120
[c000000003633320] [c000000000cefa6c] __device_attach_driver+0x11c/0x1e0
[c0000000036333a0] [c000000000cebc58] bus_for_each_drv+0xa8/0x130
[c000000003633400] [c000000000ceefcc] __device_attach+0x15c/0x250
[c0000000036334a0] [c000000000ced458] bus_probe_device+0x108/0x110
[c0000000036334f0] [c000000000ce92dc] device_add+0x7fc/0xa10
[c0000000036335b0] [c000000000d447c8] devm_create_dev_dax+0x1d8/0x530
[c000000003633640] [c000000000d46b60] __dax_pmem_probe+0x200/0x270
[c0000000036337b0] [c000000000d46bf0] dax_pmem_probe+0x20/0x70
[c0000000036337d0] [c000000000d2279c] nvdimm_bus_probe+0xac/0x2b0
[c000000003633860] [c000000000cef27c] really_probe+0x19c/0x520
[c0000000036338f0] [c000000000cef6b4] __driver_probe_device+0xb4/0x230
[c000000003633970] [c000000000cef888] driver_probe_device+0x58/0x120
[c0000000036339b0] [c000000000cefd08] __driver_attach+0x1d8/0x240
[c000000003633a30] [c000000000cebb04] bus_for_each_dev+0xb4/0x130
[c000000003633a90] [c000000000cee564] driver_attach+0x34/0x50
[c000000003633ab0] [c000000000ced878] bus_add_driver+0x218/0x300
[c000000003633b40] [c000000000cf1144] driver_register+0xa4/0x1b0
[c000000003633bb0] [c000000000d21a0c] __nd_driver_register+0x5c/0x100
[c000000003633c10] [c00000000206a2e8] dax_pmem_init+0x34/0x48
[c000000003633c30] [c0000000000132d0] do_one_initcall+0x60/0x320
[c000000003633d00] [c0000000020051b0] kernel_init_freeable+0x360/0x400
[c000000003633de0] [c000000000013764] kernel_init+0x34/0x1d0
[c000000003633e50] [c00000000000de14] ret_from_kernel_thread+0x5c/0x64

Link: https://lkml.kernel.org/r/20230411142214.64464-1-aneesh.kumar@linux.ibm.com
Fixes: 4917f55b4ef9 ("mm/sparse-vmemmap: improve memory savings for compound devmaps")
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Reported-by: Tarun Sahu <tsahu@linux.ibm.com>
Reviewed-by: Joao Martins <joao.m.martins@oracle.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: add uffdio register ioctls test

This new test tests against the returned ioctls from UFFDIO_REGISTER,
where put into uffdio_register.ioctls.

This also tests the expected failure cases of UFFDIO_REGISTER, aka:

  - Register with empty mode should fail with -EINVAL
  - Register minor without page cache (anon) should fail with -EINVAL

Link: https://lkml.kernel.org/r/20230412164548.329376-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: add shmem-private test to uffd-stress

The userfaultfd stress test never tested private shmem, which I think was
overlooked long due.  Add it so it matches with uffd unit test and it'll
cover all memory supported with the three memory types.

Meanwhile, rename the memory types a bit.  Considering shared mem is the
major use case for both shmem / hugetlbfs, changing from:

  anon, hugetlb, hugetlb_shared, shmem

To (with shmem-private added):

  anon, hugetlb, hugetlb-private, shmem, shmem-private

Add the shmem-private to run_vmtests.sh too.

Link: https://lkml.kernel.org/r/20230412164546.329355-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: drop sys/dev test in uffd-stress test

With the new uffd unit test covering the /dev/userfaultfd path and syscall
path of uffd initializations, we can safely drop the devnode test in the
old stress test.

One thing is to avoid duplication of running the stress test twice which is
an overkill to only test the /dev/ interface in run_vmtests.sh.

The other benefit is now all uffd tests (that uses userfaultfd_open) can
run automatically as long as any type of interface is enabled (either
syscall or dev), so it's more likely to succeed rather than fail due to
unprivilege.

With this patch lands, we can drop all the "mem_type:XXX" handlings too.

Link: https://lkml.kernel.org/r/20230412164525.329176-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: allow uffd test to skip properly with no privilege

Allow skip a unit test properly due to no privilege (e.g.  sigbus and
events tests).

[colin.i.king@gmail.com: fix spelling mistake "priviledge" -> "privilege"]
Link: https://lkml.kernel.org/r/20230414081506.1678998-1-colin.i.king@gmail.com
Link: https://lkml.kernel.org/r/20230412164520.329163-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: workaround no way to detect uffd-minor + wp

Userfaultfd minor+wp mode was very recently added.  The test will fail on
the old kernels at ioctl(UFFDIO_CONTINUE) which is misterious.
Unfortunately there's no feature bit to detect for this support.

Add a hack to leverage WP_UNPOPULATED to detect whether that feature
existed, since WP_UNPOPULATED was merged right after minor+wp.

Link: https://lkml.kernel.org/r/20230412164517.329152-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: move zeropage test into uffd unit tests

Simplifies it a bit along the way, e.g., drop the never used offset field
(which was always the 1st page so offset=0).

Introduce uffd_register_with_ioctls() out of uffd_register() to detect
uffdio_register.ioctls got returned.  Check that automatically when testing
UFFDIO_ZEROPAGE on different types of memory (and kernel).

Link: https://lkml.kernel.org/r/20230412164404.328815-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: move uffd sig/events tests into uffd unit tests

Move the two tests into the unit test, and convert it into 20 standalone
tests:

  - events test on all 5 mem types, with wp on/off
  - signal test on all 5 mem types, with wp on/off

  Testing sigbus on anon... done
  Testing sigbus on shmem... done
  Testing sigbus on shmem-private... done
  Testing sigbus on hugetlb... done
  Testing sigbus on hugetlb-private... done
  Testing sigbus-wp on anon... done
  Testing sigbus-wp on shmem... done
  Testing sigbus-wp on shmem-private... done
  Testing sigbus-wp on hugetlb... done
  Testing sigbus-wp on hugetlb-private... done
  Testing events on anon... done
  Testing events on shmem... done
  Testing events on shmem-private... done
  Testing events on hugetlb... done
  Testing events on hugetlb-private... done
  Testing events-wp on anon... done
  Testing events-wp on shmem... done
  Testing events-wp on shmem-private... done
  Testing events-wp on hugetlb... done
  Testing events-wp on hugetlb-private... done

It'll also remove a lot of global references along the way,
e.g. test_uffdio_wp will be replaced with the wp value passed over.

Link: https://lkml.kernel.org/r/20230412164400.328798-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: move uffd minor test to unit test

This moves the minor test to the new unit test.

Rewrite the content check with char* opeartions to avoid fiddling with
my_bcmp().

Drop global vars test_uffdio_minor and test_collapse, just assume test them
always in common code for now.

OTOH make this single test into five tests:

  - minor test on [shmem, hugetlb] with wp=false
  - minor test on [shmem, hugetlb] with wp=true
  - minor test + collapse on shmem only

One thing to mention that we used to test COLLAPSE+WP but that doesn't
sound right at all.  It's possible it's silently broken but unnoticed
because COLLAPSE is not part of the default test suite.

Make the MADV_COLLAPSE test fail-able (by skip it when failing), because
it's not guaranteed to success anyway.

Drop a bunch of useless code after the move, because the unit test always
use aligned num of pages and has nothing to do with n_cpus.

Link: https://lkml.kernel.org/r/20230412164357.328779-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: move uffd pagemap test to unit test

Move it over and make it split into two tests, one for pagemap and one for
the new WP_UNPOPULATED (to be a separate one).

The thp pagemap test wasn't really working (with MADV_HUGEPAGE).  Let's
just drop it (since it never really worked anyway..) and leave that for
later.

Link: https://lkml.kernel.org/r/20230412164352.328733-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: add framework for uffd-unit-test

Add a framework to be prepared to move unit tests from uffd-stress.c into
uffd-unit-tests.c.  The goal is to allow detection of uffd features for
each test, and also loop over specified types of memory that a test
support.

Link: https://lkml.kernel.org/r/20230412164348.328710-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: allow allocate_area() to fail properly

Mostly to detect hugetlb allocation errors and skip hugetlb tests when
pages are not allocated.

Link: https://lkml.kernel.org/r/20230412164345.328659-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: let uffd_handle_page_fault() take wp parameter

Make the handler optionally apply WP bit when resolving page faults for
either missing or minor page faults.  This moves towards removing global
test_uffdio_wp outside of the common code.

Link: https://lkml.kernel.org/r/20230412164341.328618-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: rename uffd_stats to uffd_args

Prepare for adding more fields into the struct.

Link: https://lkml.kernel.org/r/20230412164337.328607-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Suggested-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: drop global hpage_size in uffd tests

hpage_size was wrongly used.  Sometimes it means hugetlb default size,
sometimes it was used as thp size.

Remove the global variable and use the right one at each place.

Link: https://lkml.kernel.org/r/20230412164333.328596-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: drop global mem_fd in uffd tests

Drop it by creating the memfd dynamically in the tests.

Link: https://lkml.kernel.org/r/20230412164331.328584-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: UFFDIO_API test

Add one simple test for UFFDIO_API.  With that, I also added a bunch of
small but handy helpers along the way.

Link: https://lkml.kernel.org/r/20230412164257.328375-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: uffd_open_{dev|sys}()

Provide two helpers to open an uffd handle.  Drop the error checks around
SKIPs because it's inside an errexit() anyway, which IMHO doesn't really
help much if the test will not continue.

Link: https://lkml.kernel.org/r/20230412164254.328335-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: uffd_[un]register()

Add two helpers to register/unregister to an uffd.  Use them to drop
duplicate codes.

This patch also drops assert_expected_ioctls_present() and
get_expected_ioctls().  Reasons:

  - It'll need a lot of effort to pass test_type==HUGETLB into it from
    the upper, so it's the simplest way to get rid of another global var

  - The ioctls returned in UFFDIO_REGISTER is hardly useful at all,
    because any app can already detect kernel support on any ioctl via its
    corresponding UFFD_FEATURE_*.  The check here is for sanity mostly but
    it's probably destined no user app will even use it.

  - It's not friendly to one future goal of uffd to run on old
    kernels, the problem is get_expected_ioctls() compiles against
    UFFD_API_RANGE_IOCTLS, which is a value that can change depending on
    where the test is compiled, rather than reflecting what the kernel
    underneath has.  It means it'll report false negatives on old kernels
    so it's against our will.

So let's make our lives easier.

[peterx@redhat.com; tools/testing/selftests/mm/hugepage-mremap.c: add headers]
Link: https://lkml.kernel.org/r/ZDxrvZh/cw357D8P@x1n
Link: https://lkml.kernel.org/r/20230412164247.328293-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: split uffd tests into uffd-stress and uffd-unit-tests

In many ways it's weird and unwanted to keep all the tests in the same
userfaultfd.c at least when still in the current way.

For example, it doesn't make much sense to run the stress test for each
method we can create an userfaultfd handle (either via syscall or /dev/
node).  It's a waste of time running this twice for the whole stress as
the stress paths are the same, only the open path is different.

It's also just weird to need to manually specify different types of memory
to run all unit tests for the userfaultfd interface.  We should be able to
just run a single program and that should go through all functional uffd
tests without running the stress test at all.  The stress test was more
for torturing and finding race conditions.  We don't want to wait for
stress to finish just to regress test a functional test.

When we start to pile up more things on top of the same file and same
functions, things start to go a bit chaos and the code is just harder to
maintain too with tons of global variables.

This patch creates a new test uffd-unit-tests to keep userfaultfd unit
tests in the future, currently empty.

Meanwhile rename the old userfaultfd.c test to uffd-stress.c.

Link: https://lkml.kernel.org/r/20230412164244.328270-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: create uffd-common.[ch]

Move common utility functions into uffd-common.[ch] files from the
original userfaultfd.c.  This prepares for a split of userfaultfd.c into
two tests: one to only cover the old but powerful stress test, the other
one covers all the functional tests.

This movement is kind of a brute-force effort for now, with light
touch-ups but nothing should really change.  There's chances to optimize
more, but let's leave that for later.

Link: https://lkml.kernel.org/r/20230412164241.328259-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: drop test_uffdio_zeropage_eexist

The idea was trying to flip this var in the alarm handler from time to
time to test -EEXIST of UFFDIO_ZEROPAGE, but firstly it's only used in the
zeropage test so probably only used once, meanwhile we passed
"retry==false" so it'll never got tested anyway.

Drop both sides so we always test UFFDIO_ZEROPAGE retries if has_zeropage
is set (!hugetlb).

One more thing to do is doing UFFDIO_REGISTER for the alias buffer too,
because otherwise the test won't even pass!  We were just lucky that this
test never really got ran at all.

Link: https://lkml.kernel.org/r/20230412164238.328238-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: test UFFDIO_ZEROPAGE only when !hugetlb

Make the check as simple as "test_type == TEST_HUGETLB" because that's the
only mem that doesn't support ZEROPAGE.

Link: https://lkml.kernel.org/r/20230412164234.328168-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: reuse pagemap_get_entry() in vm_util.h

Meanwhile drop pagemap_read_vaddr().

Link: https://lkml.kernel.org/r/20230412164231.328157-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: use PM_* macros in vm_utils.h

We've got the macros in uffd-stress.c, move it over and use it in
vm_util.h.

Link: https://lkml.kernel.org/r/20230412164227.328145-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: merge default_huge_page_size() into one

There're already 3 same definitions of the three functions.  Move it into
vm_util.[ch].

Link: https://lkml.kernel.org/r/20230412164223.328134-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: link vm_util.c always

We do have plenty of files that want to link against vm_util.c.  Just make
it simple by linking it always.

Link: https://lkml.kernel.org/r/20230412164220.328123-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: use TEST_GEN_PROGS where proper

TEST_GEN_PROGS and TEST_GEN_FILES are used randomly in the mm/Makefile to
specify programs that need to build.  Logically all these binaries should
all fall into TEST_GEN_PROGS.

Replace those TEST_GEN_FILES with TEST_GEN_PROGS, so that we can reference
all the tests easily later.

[peterx@redhat.com: tools/testing/selftests/mm/Makefile: don't wipe out TEST_GEN_PROGS]
Link: https://lkml.kernel.org/r/ZDxrvZh/cw357D8P@x1n
Link: https://lkml.kernel.org/r/20230412164218.328104-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: merge util.h into vm_util.h

There're two util headers under mm/ kselftest.  Merge one with another.
It turns out util.h is the easy one to move.

When merging, drop PAGE_SIZE / PAGE_SHIFT because they're unnecessary
wrappers to page_size() / page_shift(), meanwhile rename them to psize()
and pshift() so as to not conflict with some existing definitions in some
test files that includes vm_util.h.

Link: https://lkml.kernel.org/r/20230412164120.327731-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: dump a summary in run_vmtests.sh

Dump a summary after running whatever test specified.  Useful for human
runners to identify any kind of failures (besides exit code).

Link: https://lkml.kernel.org/r/20230412164117.327720-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: update .gitignore with two missing tests

Patch series "selftests/mm: Split / Refactor userfault test", v2.

This patchset splits userfaultfd.c into two tests:

  - uffd-stress: the "vanilla", old and powerful stress test
  - uffd-unit-tests: all the unit tests will be moved here

This is on my todo list for a long time but I never did it for real.  The
uffd test is growing into a small and cute monster.  I start to notice it's
going harder to maintain such a test and make it useful.

A few issues I found when looking at userfaultfd test:

  - We have a bunch of unit tests in userfaultfd.c, but they always need to
    be run only after a stress type.  No way to not do it.

  - We can only run an unit test for one memory type only, if we want to
    do a quick smoke test to check regressions, there's no good way.  The
    best to come currently is "bash ./run_vmtests.sh -t userfaultfd" thanks
    to the most recent changes to run_vmtests.sh on tagging.  Still, that
    needs to run the stress tests always and hard to see what's wrong.

  - It's hard to add a new unit test to userfaultfd.c, we don't really know
    what's happening, not until we mostly read the whole file.

  - We did a bunch of useless tests, e.g. we run twice the whole suite of
    stress test just to verify both syscall and /dev/userfaultfd.  They're
    all using userfaultfd_new() to create the handle, everything should
    really be the same underneath.  One simple unit test should cover that!

  - We have tens of global variables in one file but shared with all the
    tests.  Some of them are not suitable to be a global var from
    maintainance pov.  It enforces every unit test to consider how these
    vars affects the stress test and vice versa, but that's logically not
    necessary.

  - Userfaultfd test is not friendly to old kernels.  Mostly it only works
    on the latest kernel tree.  It's preferrable to be run on all kernels
    and properly report what's missing.

I'll stop here, I feel like I can still list some..

This patchset should resolve all issues above, and actually we can do even
more on top.  I stopped doing that until I found I already got 29 patches
and 2000+ LOC changes.  That's already a patchset terrible enough so we
should move in small steps.

After the whole set applied, "./run_vmtests.sh -t userfaultfd" looks like
this:

===8<===
vm.nr_hugepages = 1024
-------------------------
running ./uffd-unit-tests
-------------------------
Testing UFFDIO_API (with syscall)... done
Testing UFFDIO_API (with /dev/userfaultfd)... done
Testing register-ioctls on anon... done
Testing register-ioctls on shmem... done
Testing register-ioctls on shmem-private... done
Testing register-ioctls on hugetlb... done
Testing register-ioctls on hugetlb-private... done
Testing zeropage on anon... done
Testing zeropage on shmem... done
Testing zeropage on shmem-private... done
Testing zeropage on hugetlb... done
Testing zeropage on hugetlb-private... done
Testing pagemap on anon... done
Testing wp-unpopulated on anon... done
Testing minor on shmem... done
Testing minor on hugetlb... done
Testing minor-wp on shmem... done
Testing minor-wp on hugetlb... done
Testing minor-collapse on shmem... done
Testing sigbus on anon... done
Testing sigbus on shmem... done
Testing sigbus on shmem-private... done
Testing sigbus on hugetlb... done
Testing sigbus on hugetlb-private... done
Testing sigbus-wp on anon... done
Testing sigbus-wp on shmem... done
Testing sigbus-wp on shmem-private... done
Testing sigbus-wp on hugetlb... done
Testing sigbus-wp on hugetlb-private... done
Testing events on anon... done
Testing events on shmem... done
Testing events on shmem-private... done
Testing events on hugetlb... done
Testing events on hugetlb-private... done
Testing events-wp on anon... done
Testing events-wp on shmem... done
Testing events-wp on shmem-private... done
Testing events-wp on hugetlb... done
Testing events-wp on hugetlb-private... done
Userfaults unit tests: pass=39, skip=0, fail=0 (total=39)
[PASS]
--------------------------------
running ./uffd-stress anon 20 16
--------------------------------
nr_pages: 5120, nr_pages_per_cpu: 640
bounces: 15, mode: rnd racing ver poll, userfaults: 345 missing (26+48+61+102+30+12+59+7) 1596 wp (120+139+317+346+215+67+306+86)
[...]
[PASS]
------------------------------------
running ./uffd-stress hugetlb 128 32
------------------------------------
nr_pages: 64, nr_pages_per_cpu: 8
bounces: 31, mode: rnd racing ver poll, userfaults: 29 missing (6+6+6+5+4+2+0+0) 104 wp (20+19+22+18+7+12+5+1)
[...]
[PASS]
--------------------------------------------
running ./uffd-stress hugetlb-private 128 32
--------------------------------------------
nr_pages: 64, nr_pages_per_cpu: 8
bounces: 31, mode: rnd racing ver poll, userfaults: 33 missing (12+9+7+0+5+0+0+0) 111 wp (24+25+14+14+11+17+5+1)
[...]
[PASS]
---------------------------------
running ./uffd-stress shmem 20 16
---------------------------------
nr_pages: 5120, nr_pages_per_cpu: 640
bounces: 15, mode: rnd racing ver poll, userfaults: 247 missing (15+17+34+60+81+37+3+0) 2038 wp (180+114+276+400+381+318+165+204)
[...]
[PASS]
-----------------------------------------
running ./uffd-stress shmem-private 20 16
-----------------------------------------
nr_pages: 5120, nr_pages_per_cpu: 640
bounces: 15, mode: rnd racing ver poll, userfaults: 235 missing (52+29+55+56+13+9+16+5) 2849 wp (218+406+461+531+328+284+430+191)
[...]
[PASS]
SUMMARY: PASS=6 SKIP=0 FAIL=0
===8<===

The output may be different if we miss some features (e.g., hugetlb not
allocated, old kernel, less privilege of uffd handle), but they should show
up with good reasons.  E.g., I tried to run the unit test on my Fedora
kernel and it gives me:

===8<===
UFFDIO_API (with syscall)... failed [reason: UFFDIO_API should fail with wrong api but didn't]
UFFDIO_API (with /dev/userfaultfd)... skipped [reason: cannot open userfaultfd handle]
zeropage on anon... done
zeropage on shmem... done
zeropage on shmem-private... done
zeropage-hugetlb on hugetlb... done
zeropage-hugetlb on hugetlb-private... done
pagemap on anon... pagemap on anon... pagemap on anon... done
wp-unpopulated on anon... skipped [reason: feature missing]
minor on shmem... done
minor on hugetlb... done
minor-wp on shmem... skipped [reason: feature missing]
minor-wp on hugetlb... skipped [reason: feature missing]
minor-collapse on shmem... done
sigbus on anon... skipped [reason: possible lack of priviledge]
sigbus on shmem... skipped [reason: possible lack of priviledge]
sigbus on shmem-private... skipped [reason: possible lack of priviledge]
sigbus on hugetlb... skipped [reason: possible lack of priviledge]
sigbus on hugetlb-private... skipped [reason: possible lack of priviledge]
sigbus-wp on anon... skipped [reason: possible lack of priviledge]
sigbus-wp on shmem... skipped [reason: possible lack of priviledge]
sigbus-wp on shmem-private... skipped [reason: possible lack of priviledge]
sigbus-wp on hugetlb... skipped [reason: possible lack of priviledge]
sigbus-wp on hugetlb-private... skipped [reason: possible lack of priviledge]
events on anon... skipped [reason: possible lack of priviledge]
events on shmem... skipped [reason: possible lack of priviledge]
events on shmem-private... skipped [reason: possible lack of priviledge]
events on hugetlb... skipped [reason: possible lack of priviledge]
events on hugetlb-private... skipped [reason: possible lack of priviledge]
events-wp on anon... skipped [reason: possible lack of priviledge]
events-wp on shmem... skipped [reason: possible lack of priviledge]
events-wp on shmem-private... skipped [reason: possible lack of priviledge]
events-wp on hugetlb... skipped [reason: possible lack of priviledge]
events-wp on hugetlb-private... skipped [reason: possible lack of priviledge]
Userfaults unit tests: pass=9, skip=24, fail=1 (total=34)
===8<===

Patch layout:

- Revert "userfaultfd: don't fail on unrecognized features"

  Something I found when I got the UFFDIO_API test below.  Axel, I still
  propose to revert it as a whole, but feel free to continue the discussion
  from the original patch thread.

- selftests/mm: Update .gitignore with two missing tests
- selftests/mm: Dump a summary in run_vmtests.sh
- selftests/mm: Merge util.h into vm_util.h
- selftests/mm: Use TEST_GEN_PROGS where proper
- selftests/mm: Link vm_util.c always
- selftests/mm: Merge default_huge_page_size() into one
- selftests/mm: Use PM_* macros in vm_utils.h
- selftests/mm: Reuse pagemap_get_entry() in vm_util.h
- selftests/mm: Test UFFDIO_ZEROPAGE only when !hugetlb
- selftests/mm: Drop test_uffdio_zeropage_eexist

  Until here, all cleanups here and there.  I wanted to keep going, but I
  found that maybe it'll take a few more days to split the test.  Hence I
  did a split starting from the next one, so we have a working thing first.

- selftests/mm: Create uffd-common.[ch]
- selftests/mm: Split uffd tests into uffd-stress and uffd-unit-tests

  This did the major brute force split of common codes into
  uffd-common.[ch].  That'll be the so far common base for stress and unit
  tests.  Then a new unit test is created.

- selftests/mm: uffd_[un]register()
- selftests/mm: uffd_open_{dev|sys}()
- selftests/mm: UFFDIO_API test

  This patch hides here to start writting the 1st unit test with
  UFFDIO_API, also detection of userfaultfd privileges.

- selftests/mm: Drop global mem_fd in uffd tests
- selftests/mm: Drop global hpage_size in uffd tests
- selftests/mm: Rename uffd_stats to uffd_args
- selftests/mm: Let uffd_handle_page_fault() takes wp parameter
- selftests/mm: Allow allocate_area() to fail properly

  Some further cleanup that I noticed otherwise hard to move the tests.

- selftests/mm: Add framework for uffd-unit-test

  The major patch provides the framework for most of the rest unit tests.

- selftests/mm: Move uffd pagemap test to unit test
- selftests/mm: Move uffd minor test to unit test
- selftests/mm: Move uffd sig/events tests into uffd unit tests
- selftests/mm: Move zeropage test into uffd unit tests

  Move unit tests and suite them into the new file.

- selftests/mm: Workaround no way to detect uffd-minor + wp
- selftests/mm: Allow uffd test to skip properly with no privilege
- selftests/mm: Drop sys/dev test in uffd-stress test
- selftests/mm: Add shmem-private test to uffd-stress

  A bunch of changes to do better on error reportings, and add
  shmem-private to the stress test which was long missing.

- selftests/mm: Add uffdio register ioctls test

  One more patch to test uffdio_register.ioctls.

This patch (of 30):

Update .gitignore with two missing tests.

Link: https://lkml.kernel.org/r/20230412163922.327282-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20230412164114.327709-1-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Dmitry Safonov <0x7f454c46@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Zach O'Keefe <zokeefe@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/vmscan: simplify shrink_node()

The difference between sc->nr_reclaimed and nr_reclaimed is computed three
times.  Introduce a new variable to record the value, so it only needs to
be computed once.

Link: https://lkml.kernel.org/r/20230411061757.12041-1-haifeng.xu@shopee.com
Signed-off-by: Haifeng Xu <haifeng.xu@shopee.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mpage: use folios in bio end_io handler

Use folios in the bio end_io handler.  This conversion does the
appropriate handling on the folios in the respective end_io callback and
removes the call to page_endio(), which is soon to be removed.

Link: https://lkml.kernel.org/r/20230411122920.30134-4-p.raghav@samsung.com
Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Martin Brandenburg <martin@omnibond.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mike Marshall <hubcap@omnibond.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mpage: split submit_bio and bio end_io handler for reads and writes

Split the submit_bio() and bio end_io handler for reads and writes similar
to other aops.

This is a prep patch before we convert end_io handlers to use folios.

Link: https://lkml.kernel.org/r/20230411122920.30134-3-p.raghav@samsung.com
Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Suggested-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Martin Brandenburg <martin@omnibond.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mike Marshall <hubcap@omnibond.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

orangefs: use folios in orangefs_readahead

Patch series "remove page_endio()", v3.

It was decided to remove the page_endio() as per the previous RFC
discussion[1] of this series and move that functionality into the caller
itself.  One of the side benefit of doing that is the callers have been
modified to directly work on folios as page_endio() already worked on
folios.

As Christoph is doing ZRAM cleanups[4] which will get rid of page_endio()
function usage, I removed the final patch that removes page_endio()[5].  I
will send it separately after rc-1 once the zram cleanups are merged.

mpage changes were tested with a simple boot testing and running a fio
workload on ext2 filesystem.  orangefs was tested by Mike Marshall (No
code changes since he tested).

This patch (of 3):

Convert orangefs_readahead() from using struct page to struct folio.  This
conversion removes the call to page_endio() which is soon to be removed,
and simplifies the final page handling.

The page error flags is not required to be set in the error case as
orangefs doesn't depend on them.

Link: https://lkml.kernel.org/r/20230411122920.30134-1-p.raghav@samsung.com
Link: https://lkml.kernel.org/r/20230411122920.30134-2-p.raghav@samsung.com
Link: https://lore.kernel.org/linux-mm/ZBHcl8Pz2ULb4RGD@infradead.org/
Link: https://lore.kernel.org/linux-mm/20230322135013.197076-1-p.raghav@samsung.com/
Link: https://lore.kernel.org/linux-mm/8adb0770-6124-e11f-2551-6582db27ed32@samsung.com/
Link: https://lore.kernel.org/linux-block/20230404150536.2142108-1-hch@lst.de/T/#t
Link: https://lore.kernel.org/lkml/20230403132221.94921-6-p.raghav@samsung.com/
Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Tested-by: Mike Marshall <hubcap@omnibond.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Martin Brandenburg <martin@omnibond.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/huge_memory: conditionally call maybe_mkwrite() and drop pte_wrprotect() in __split_huge_pmd_locked()

No need to call maybe_mkwrite() to then wrprotect if the source PMD was not
writable.

It's worth nothing that this now allows for PTEs to be writable even if
the source PMD was not writable: if vma->vm_page_prot includes write
permissions.

As documented in commit 931298e103c2 ("mm/userfaultfd: rely on
vma->vm_page_prot in uffd_wp_range()"), any mechanism that intends to
have pages wrprotected (COW, writenotify, mprotect, uffd-wp, softdirty,
...) has to properly adjust vma->vm_page_prot upfront, to not include
write permissions. If vma->vm_page_prot includes write permissions, the
PTE/PMD can be writable as default.

This now mimics the handling in mm/migrate.c:remove_migration_pte() and in
mm/huge_memory.c:remove_migration_pmd(), which has been in place for a
long time (except that 96a9c287e25d ("mm/migrate: fix wrongly apply write
bit after mkdirty on sparc64") temporarily changed it).

Link: https://lkml.kernel.org/r/20230411142512.438404-7-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/huge_memory: revert "Partly revert "mm/thp: carry over dirty bit when thp splits on pmd""

This reverts commit 624a2c94f5b7 ("Partly revert "mm/thp: carry over dirty
bit when thp splits on pmd"") and the fixup in commit e833bc503405
("mm/thp: re-apply mkdirty for small pages after split").

Now that sparc64 mkdirty handling is fixed and no longer sets a PTE/PMD
writable that shouldn't be writable, let's revert the temporary fix and
remove the stale comment.

The mkdirty mm selftest still passes with this change on sparc64.

Note that loongarch handling was fixed in commit bf2f34a506e6 ("LoongArch:
Set _PAGE_DIRTY only if _PAGE_WRITE is set in {pmd,pte}_mkdirty()")

Link: https://lkml.kernel.org/r/20230411142512.438404-6-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/migrate: revert "mm/migrate: fix wrongly apply write bit after mkdirty on sparc64"

This reverts commit 96a9c287e25d ("mm/migrate: fix wrongly apply write bit
after mkdirty on sparc64").

Now that sparc64 mkdirty handling is fixed and no longer sets a PTE/PMD
writable that shouldn't be writable, let's revert the temporary fix.

The mkdirty mm selftest still passes with this change on sparc64.

Note that loongarch handling was fixed in commit bf2f34a506e6 ("LoongArch:
Set _PAGE_DIRTY only if _PAGE_WRITE is set in {pmd,pte}_mkdirty()").

Link: https://lkml.kernel.org/r/20230411142512.438404-5-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

sparc/mm: don't unconditionally set HW writable bit when setting PTE dirty on 64bit

On sparc64, there is no HW modified bit, therefore, SW tracks via a SW bit
if the PTE is dirty via pte_mkdirty().  However, pte_mkdirty() currently
also unconditionally sets the HW writable bit, which is wrong.

pte_mkdirty() is not supposed to make a PTE actually writable, unless the
SW writable bit -- pte_write() -- indicates that the PTE is not
write-protected.  Fortunately, sparc64 also defines a SW writable bit.

For example, this already turned into a problem in the context of THP
splitting as documented in commit 624a2c94f5b7 ("Partly revert "mm/thp:
carry over dirty bit when thp splits on pmd""), and for page migration, as
documented in commit 96a9c287e25d ("mm/migrate: fix wrongly apply write
bit after mkdirty on sparc64").

Also, we might want to use the dirty PTE bit in the context of KSM with
shared zeropage [1], whereby setting the page writable would be
problematic.

But more general, any code that might end up setting a PTE/PMD dirty
inside a VM without write permissions is possibly broken,

Before this commit (sun4u in QEMU):
root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty
# [INFO] detected THP size: 8192 KiB
TAP version 13
1..6
# [INFO] PTRACE write access
not ok 1 SIGSEGV generated, page not modified
# [INFO] PTRACE write access to THP
not ok 2 SIGSEGV generated, page not modified
# [INFO] Page migration
ok 3 SIGSEGV generated, page not modified
# [INFO] Page migration of THP
ok 4 SIGSEGV generated, page not modified
# [INFO] PTE-mapping a THP
ok 5 SIGSEGV generated, page not modified
# [INFO] UFFDIO_COPY
not ok 6 SIGSEGV generated, page not modified
Bail out! 3 out of 6 tests failed
# Totals: pass:3 fail:3 xfail:0 xpass:0 skip:0 error:0

Test #3,#4,#5 pass ever since we added some MM workarounds, the
underlying issue remains.

Let's fix the remaining issues and prepare for reverting the workarounds
by setting the HW writable bit only if both, the SW dirty bit and the SW
writable bit are set.

We have to move pte_dirty() and pte_write() up. The code patching
mechanism and handling constants > 22bit is a bit special on sparc64.

The ASM logic in pte_mkdirty() and pte_mkwrite() match the logic in
pte_mkold() to create the mask depending on the machine type. The ASM
logic in __pte_mkhwwrite() matches the logic in pte_present(), just
using an "or" instead of an "and" instruction.

With this commit (sun4u in QEMU):
root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty
# [INFO] detected THP size: 8192 KiB
TAP version 13
1..6
# [INFO] PTRACE write access
ok 1 SIGSEGV generated, page not modified
# [INFO] PTRACE write access to THP
ok 2 SIGSEGV generated, page not modified
# [INFO] Page migration
ok 3 SIGSEGV generated, page not modified
# [INFO] Page migration of THP
ok 4 SIGSEGV generated, page not modified
# [INFO] PTE-mapping a THP
ok 5 SIGSEGV generated, page not modified
# [INFO] UFFDIO_COPY
ok 6 SIGSEGV generated, page not modified
# Totals: pass:6 fail:0 xfail:0 xpass:0 skip:0 error:0

This handling seems to have been in place forever.

[1] https://lkml.kernel.org/r/533a7c3d-3a48-b16b-b421-6e8386e0b142@redhat.com

Link: https://lkml.kernel.org/r/20230411142512.438404-4-david@redhat.com
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: mkdirty: test behavior of (pte|pmd)_mkdirty on VMAs without write permissions

Let's add some tests that trigger (pte|pmd)_mkdirty on VMAs without write
permissions.  If an architecture implementation is wrong, we might
accidentally set the PTE/PMD writable and allow for write access in a VMA
without write permissions.

The tests include reproducers for the two issues recently discovered
and worked-around in core-MM for now:

(1) commit 624a2c94f5b7 ("Partly revert "mm/thp: carry over dirty
    bit when thp splits on pmd"")
(2) commit 96a9c287e25d ("mm/migrate: fix wrongly apply write bit
    after mkdirty on sparc64")

In addition, some other tests that reveal further issues.

All tests pass under x86_64:
./mkdirty
# [INFO] detected THP size: 2048 KiB
TAP version 13
1..6
# [INFO] PTRACE write access
ok 1 SIGSEGV generated, page not modified
# [INFO] PTRACE write access to THP
ok 2 SIGSEGV generated, page not modified
# [INFO] Page migration
ok 3 SIGSEGV generated, page not modified
# [INFO] Page migration of THP
ok 4 SIGSEGV generated, page not modified
# [INFO] PTE-mapping a THP
ok 5 SIGSEGV generated, page not modified
# [INFO] UFFDIO_COPY
ok 6 SIGSEGV generated, page not modified
# Totals: pass:6 fail:0 xfail:0 xpass:0 skip:0 error:0

But some fail on sparc64:
./mkdirty
# [INFO] detected THP size: 8192 KiB
TAP version 13
1..6
# [INFO] PTRACE write access
not ok 1 SIGSEGV generated, page not modified
# [INFO] PTRACE write access to THP
not ok 2 SIGSEGV generated, page not modified
# [INFO] Page migration
ok 3 SIGSEGV generated, page not modified
# [INFO] Page migration of THP
ok 4 SIGSEGV generated, page not modified
# [INFO] PTE-mapping a THP
ok 5 SIGSEGV generated, page not modified
# [INFO] UFFDIO_COPY
not ok 6 SIGSEGV generated, page not modified
Bail out! 3 out of 6 tests failed
# Totals: pass:3 fail:3 xfail:0 xpass:0 skip:0 error:0

Reverting both above commits makes all tests fail on sparc64:
./mkdirty
# [INFO] detected THP size: 8192 KiB
TAP version 13
1..6
# [INFO] PTRACE write access
not ok 1 SIGSEGV generated, page not modified
# [INFO] PTRACE write access to THP
not ok 2 SIGSEGV generated, page not modified
# [INFO] Page migration
not ok 3 SIGSEGV generated, page not modified
# [INFO] Page migration of THP
not ok 4 SIGSEGV generated, page not modified
# [INFO] PTE-mapping a THP
not ok 5 SIGSEGV generated, page not modified
# [INFO] UFFDIO_COPY
not ok 6 SIGSEGV generated, page not modified
Bail out! 6 out of 6 tests failed
# Totals: pass:0 fail:6 xfail:0 xpass:0 skip:0 error:0

The tests are useful to detect other problematic archs, to verify new
arch fixes, and to stop such issues from reappearing in the future.

For now, we don't add any hugetlb tests.

Link: https://lkml.kernel.org/r/20230411142512.438404-3-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: reuse read_pmd_pagesize() in COW selftest

Patch series "mm: (pte|pmd)_mkdirty() should not unconditionally allow for
write access".

This is the follow-up on [1], adding selftests (testing for known issues
we added workarounds for and other issues that haven't been fixed yet),
fixing sparc64, reverting the workarounds, and perform one cleanup.

The patch from [1] was modified slightly (updated/extended patch
description, dropped one unnecessary NOP instruction from the ASM in
__pte_mkhwwrite()).

Retested on x86_64 and sparc64 (sun4u in QEMU).

I scanned most architectures to make sure their (pte|pmd)_mkdirty()
handling is correct.  To be sure, we can run the selftests and find out if
other architectures are still affectes (loongarch was fixed recently as
well).

Based on master for now. I don't expect surprises regarding mm-tress, but
I can rebase if there are any problems.

This patch (of 6):

The COW selftest can deal with THP not being configured.  So move error
handling of read_pmd_pagesize() into the callers such that we can reuse it
in the COW selftest.

Link: https://lkml.kernel.org/r/20230411142512.438404-1-david@redhat.com
Link: https://lkml.kernel.org/r/20221212130213.136267-1-david@redhat.com
Link: https://lkml.kernel.org/r/20230411142512.438404-2-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: return errors from read_from_bdev_sync

Propagate read errors to the caller instead of dropping them on the floor,
and stop returning the somewhat dangerous 1 on success from
read_from_bdev*.

Link: https://lkml.kernel.org/r/20230411171459.567614-18-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: fix synchronous reads

Currently nothing waits for the synchronous reads before accessing the
data.  Switch them to an on-stack bio and submit_bio_wait to make sure the
I/O has actually completed when the work item has been flushed.  This also
removes the call to page_endio that would unlock a page that has never
been locked.

Drop the partial_io/sync flag, as chaining only makes sense for the
asynchronous reads of the entire page.

Link: https://lkml.kernel.org/r/20230411171459.567614-17-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: don't return errors from read_from_bdev_async

bio_alloc will never return a NULL bio when it is allowed to sleep, and
adding a single page to bio with a single vector also can't fail, so
switch to the asserting __bio_add_page variant and drop the error returns.

Link: https://lkml.kernel.org/r/20230411171459.567614-16-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: pass a page to read_from_bdev

read_from_bdev always reads a whole page, so pass a page to it instead of
the bvec and remove the now pointless zram_bvec_read_from_bdev wrapper.

Link: https://lkml.kernel.org/r/20230411171459.567614-15-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: refactor zram_bdev_write

Split the read/modify/write case into a separate helper.

Link: https://lkml.kernel.org/r/20230411171459.567614-14-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Minchan Kim <minchan@kernel.org>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: don't pass a bvec to __zram_bvec_write

__zram_bvec_write only extracts the page from __zram_bvec_write and always
expects a full page of input.  Pass the page directly instead of the bvec
and rename the function to zram_write_page.

Link: https://lkml.kernel.org/r/20230411171459.567614-13-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: refactor zram_bdev_read

Split the partial read into a separate helper.

Link: https://lkml.kernel.org/r/20230411171459.567614-12-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: directly call zram_read_page in writeback_store

writeback_store always reads a full page, so just call zram_read_page
directly and bypass the boune buffer handling.

Link: https://lkml.kernel.org/r/20230411171459.567614-11-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: rename __zram_bvec_read to zram_read_page

__zram_bvec_read doesn't get passed a bvec, but always read a whole page.
Rename it to make the usage more clear.

Link: https://lkml.kernel.org/r/20230411171459.567614-10-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: don't use highmem for the bounce buffer in zram_bvec_{read,write}

There is no point in allocation a highmem page when we instantly need to
copy from it.

Link: https://lkml.kernel.org/r/20230411171459.567614-9-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: refactor highlevel read and write handling

Instead of having an outer loop in __zram_make_request and then branch out
for reads vs writes for each loop iteration in zram_bvec_rw, split the
main handler into separat zram_bio_read and zram_bio_write handlers that
also include the functionality formerly in zram_bvec_rw.

Link: https://lkml.kernel.org/r/20230411171459.567614-8-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Minchan Kim <minchan@kernel.org>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: return early on error in zram_bvec_rw

When the low-level access fails, don't clear the idle flag or clear the
caches, and just return.

Link: https://lkml.kernel.org/r/20230411171459.567614-7-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Minchan Kim <minchan@kernel.org>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: move discard handling to zram_submit_bio

Switch on the bio operation in zram_submit_bio and only call into
__zram_make_request for read and write operations.

Link: https://lkml.kernel.org/r/20230411171459.567614-6-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: simplify bvec iteration in __zram_make_request

bio_for_each_segment synthetize bvecs that never cross page boundaries, so
don't duplicate that work in an inner loop.

Link: https://lkml.kernel.org/r/20230411171459.567614-5-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: make zram_bio_discard more self-contained

Derive the index and offset variables inside the function, and complete
the bio directly in preparation for cleaning up the I/O path.

Link: https://lkml.kernel.org/r/20230411171459.567614-4-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: remove valid_io_request

All bios hande to drivers from the block layer are checked against the
device size and for logical block alignment already (and have been since
long before zram was merged), so don't duplicate those checks.

Link: https://lkml.kernel.org/r/20230411171459.567614-3-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

zram: always compile read_from_bdev_sync

Patch series "zram I/O path cleanups and fixups", v3.

This series cleans up the zram I/O path, and fixes the handling of
synchronous I/O to the underlying device in the writeback_store function
or for > 4K PAGE_SIZE systems.

The fixes are at the end, as I could not fully reason about them being
safe before untangling the callchain.

This patch (of 17):

read_from_bdev_sync is currently only compiled for non-4k PAGE_SIZE, which
means it won't be built with the most common configurations.

Replace the ifdef with a check for the PAGE_SIZE in an if instead.  The
check uses an extra symbol and IS_ENABLED to allow the compiler to
eliminate the dead code, leading to the same generated code size:

   text    data     bss     dec     hex filename
  16709    1428      12   18149    46e5 drivers/block/zram/zram_drv.o.old
  16709    1428      12   18149    46e5 drivers/block/zram/zram_drv.o.new

Link: https://lkml.kernel.org/r/20230411171459.567614-1-hch@lst.de
Link: https://lkml.kernel.org/r/20230411171459.567614-2-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

maple_tree: add a test case to check maple_alloc

Add a test case to check whether the number of maple_alloc structures is
actually equal to mas->alloc->total.

Link: https://lkml.kernel.org/r/20230411041005.26205-2-zhangpeng.00@bytedance.com
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Cc: Liam R. Howlett <Liam.Howlett@Oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: backing-dev: set variables dev_attr_min,max_bytes storage-class-specifier to static

smatch reports
mm/backing-dev.c:266:1: warning: symbol
  'dev_attr_min_bytes' was not declared. Should it be static?
mm/backing-dev.c:294:1: warning: symbol
  'dev_attr_max_bytes' was not declared. Should it be static?

These variables are only used in one file so should be static.

Link: https://lkml.kernel.org/r/20230408141609.2703647-1-trix@redhat.com
Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

maple_tree: use correct variable type in sizeof

The type of variable pointed to by pivs is unsigned long, but the type
used in sizeof is a pointer type.  Change it to unsigned long.

This change has no runtime effect, as sizeof(ul) == sizeof(ul *).

Link: https://lkml.kernel.org/r/20230411023513.15227-1-zhangpeng.00@bytedance.com
Fixes: 54a611b60590 ("Maple Tree: add new data structure")
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reported-by: David Binderman <dcb314@hotmail.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

userfaultfd: convert mfill_atomic() to use a folio

Convert mfill_atomic_pte_copy(), shmem_mfill_atomic_pte() and
mfill_atomic_pte() to take in a folio pointer.

Convert mfill_atomic() to use a folio.  Convert page_kaddr to kaddr in
mfill_atomic().

Link: https://lkml.kernel.org/r/20230410133932.32288-7-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert copy_user_huge_page() to copy_user_large_folio()

Replace copy_user_huge_page() with copy_user_large_folio().
copy_user_large_folio() does the same as copy_user_huge_page(), but takes
in folios instead of pages.  Remove pages_per_huge_page from
copy_user_large_folio(), because we can get that from folio_nr_pages(dst).

Convert copy_user_gigantic_page() to take in folios.

Link: https://lkml.kernel.org/r/20230410133932.32288-6-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

userfaultfd: convert mfill_atomic_hugetlb() to use a folio

Convert hugetlb_mfill_atomic_pte() to take in a folio pointer instead of
a page pointer.

Convert mfill_atomic_hugetlb() to use a folio.

Link: https://lkml.kernel.org/r/20230410133932.32288-5-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

userfaultfd: convert copy_huge_page_from_user() to copy_folio_from_user()

Replace copy_huge_page_from_user() with copy_folio_from_user().
copy_folio_from_user() does the same as copy_huge_page_from_user(), but
takes in a folio instead of a page.

Convert page_kaddr to kaddr in copy_folio_from_user() to do indenting
cleanup.

Link: https://lkml.kernel.org/r/20230410133932.32288-4-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

userfaultfd: use kmap_local_page() in copy_huge_page_from_user()

kmap() and kmap_atomic() are being deprecated in favor of
kmap_local_page() which is appropriate for any thread local context.[1]

Let's replace the kmap() and kmap_atomic() with kmap_local_page() in
copy_huge_page_from_user().  When allow_pagefault is false, disable page
faults to prevent potential deadlock.[2]

[1] https://lore.kernel.org/all/20220813220034.806698-1-ira.weiny@intel.com/
[2] https://lkml.kernel.org/r/20221025220136.2366143-1-ira.weiny@intel.com

Link: https://lkml.kernel.org/r/20230410133932.32288-3-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

userfaultfd: convert mfill_atomic_pte_copy() to use a folio

Patch series "userfaultfd: convert userfaultfd functions to use folios",
v6.

This patch series converts several userfaultfd functions to use folios.

This patch (of 6):

Call vma_alloc_folio() directly instead of alloc_page_vma() and convert
page_kaddr to kaddr in mfill_atomic_pte_copy().  Removes several calls to
compound_head().

Link: https://lkml.kernel.org/r/20230410133932.32288-1-zhangpeng362@huawei.com
Link: https://lkml.kernel.org/r/20230410133932.32288-2-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

smaps: fix defined but not used smaps_shmem_walk_ops

When !CONFIG_SHMEM smaps_shmem_walk_ops is defined but not used,
triggering a compiler warning.  To avoid the warning remove the #ifdef
around the usage.  This has no effect because shmem_mapping() is a stub
returning false when !CONFIG_SHMEM so the code will be compiled out,
however we now need to also provide a stub for shmem_swap_usage().

Link: https://lkml.kernel.org/r/20230405103819.151246-1-steven.price@arm.com
Fixes: 7b86ac3371b7 ("pagewalk: separate function pointers from iterator data")
Signed-off-by: Steven Price <steven.price@arm.com>
Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/oe-kbuild-all/202304031749.UiyJpxzF-lkp@intel.com/
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm, page_alloc: use check_pages_enabled static key to check tail pages

Commit 700d2e9a36b9 ("mm, page_alloc: reduce page alloc/free sanity
checks") has introduced a new static key check_pages_enabled to control
when struct pages are sanity checked during allocation and freeing.  Mel
Gorman suggested that free_tail_pages_check() could use this static key as
well, instead of relying on CONFIG_DEBUG_VM.  That makes sense, so do
that.  Also rename the function to free_tail_page_prepare() because it
works on a single tail page and has a struct page preparation component as
well as the optional checking component.

Also remove some unnecessary unlikely() within static_branch_unlikely()
statements that Mel pointed out for commit 700d2e9a36b9.

Link: https://lkml.kernel.org/r/20230405142840.11068-1-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Suggested-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Alexander Halbuer <halbuer@sra.uni-hannover.de>
Cc: Kees Cook <keescook@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/userfaultfd: don't consider uffd-wp bit of writable migration entries

If we end up with a writable migration entry that has the uffd-wp bit set,
we already messed up: the source PTE/PMD was writable, which means we
could have modified the page without notifying uffd first.  Setting the
uffd-wp bit always implies converting migration entries to !writable
migration entries.

Commit 8f34f1eac382 ("mm/userfaultfd: fix uffd-wp special cases for
fork()") documents that "3.  Forget to carry over uffd-wp bit for a write
migration huge pmd entry", but it doesn't really say why that should be
relevant.

So let's remove that code to avoid hiding an eventual underlying issue (in
the future, we might want to warn when creating writable migration entries
that have the uffd-wp bit set -- or even better when turning a PTE
writable that still has the uffd-wp bit set).

This now matches the handling for hugetlb migration entries in
hugetlb_change_protection().

In copy_huge_pmd()/copy_nonpresent_pte()/copy_hugetlb_page_range(), we
still transfer the uffd-bit also for writable migration entries, but
simply because we have unified handling for "writable" and
"readable-exclusive" migration entries, and we care about transferring the
uffd-wp bit for the latter.

Link: https://lkml.kernel.org/r/20230405160236.587705-3-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Cc: Muhammad Usama Anjum <usama.anjum@collabora.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: mlock: use folios_put() in mlock_folio_batch()

Since we have updated mlock to use folios, it's better to call
folios_put() instead of calling release_pages() directly.

Link: https://lkml.kernel.org/r/20230405161854.6931-2-zhengqi.arch@bytedance.com
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

prctl: add PR_GET_AUXV to copy auxv to userspace

If a library wants to get information from auxv (for instance,
AT_HWCAP/AT_HWCAP2), it has a few options, none of them perfectly reliable
or ideal:

- Be main or the pre-main startup code, and grub through the stack above
  main. Doesn't work for a library.
- Call libc getauxval. Not ideal for libraries that are trying to be
  libc-independent and/or don't otherwise require anything from other
  libraries.
- Open and read /proc/self/auxv. Doesn't work for libraries that may run
  in arbitrarily constrained environments that may not have /proc
  mounted (e.g. libraries that might be used by an init program or a
  container setup tool).
- Assume you're on the main thread and still on the original stack, and
  try to walk the stack upwards, hoping to find auxv. Extremely bad
  idea.
- Ask the caller to pass auxv in for you. Not ideal for a user-friendly
  library, and then your caller may have the same problem.

Add a prctl that copies current->mm->saved_auxv to a userspace buffer.

Link: https://lkml.kernel.org/r/d81864a7f7f43bca6afa2a09fc2e850e4050ab42.1680611394.git.josh@joshtriplett.org
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

maple_tree: simplify mas_wr_node_walk()

Simplify code of mas_wr_node_walk() without changing functionality, and
improve readability.  Remove some special judgments.  Instead of
dynamically recording the min and max in the loop, get the final min and
max directly at the end.

Link: https://lkml.kernel.org/r/20230314124203.91572-3-zhangpeng.00@bytedance.com
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

m68k/mm: use correct bit number in _PAGE_SWP_EXCLUSIVE comment

As noticed by Geert, commit b5c88f21531c ("microblaze/mm: support
__HAVE_ARCH_PTE_SWP_EXCLUSIVE") modified m68k code by accident.  While
replacing 0x080 by CF_PAGE_NOCACHE is correct, although it should have
been part of commit ed4154067a08 ("m68k/mm: support
__HAVE_ARCH_PTE_SWP_EXCLUSIVE"), replacing "bit 7" by "bit 24" in the
comment was wrong.

Let's revert to the previous, correct, comment.

Link: https://lkml.kernel.org/r/20230404085636.121409-1-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/madvise: use vma_lookup() instead of find_vma()

Using vma_lookup() verifies the address is contained in the found vma.
This results in easier to read the code.

Link: https://lkml.kernel.org/r/20230404094515.1883552-1-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

memcg v1: provide read access to memory.pressure_level

cgroups v1 has a unique way of setting up memory pressure notifications:
the user opens "memory.pressure_level" of the cgroup they want to monitor
for pressure, then open "cgroup.event_control" and write the fd (among
other things) to that file.  memory.pressure_level has no other use,
specifically it does not support any read or write operations.
Consequently, no handlers are provided, and cgroup_file_mode() sets the
permissions to 000.  However, to actually use the mechanism, the
subscribing user must have read access to the file and open the fd for
reading, see memcg_write_event_control().

This is all fine as long as the subscribing process runs as root and is
otherwise unconfined by further restrictions.  However, if you add strict
access controls such as selinux, the permission bits will be enforced, and
opening memory.pressure_level for reading will fail, preventing the
process from subscribing, even as root.

To work around this issue, introduce a dummy read handler.  When
memory.pressure_level is created, cgroup_file_mode() will notice the
existence of a handler, and therefore add read permissions to the file.

Link: https://lkml.kernel.org/r/20230404105900.2005-1-flosch@nutanix.com
Signed-off-by: Florian Schmidt <flosch@nutanix.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/khugepaged: maintain page cache uptodate flag

Make sure that collapse_file doesn't interfere with checking the uptodate
flag in the page cache by only inserting hpage into the page cache after
it has been updated and marked uptodate.  This is achieved by simply not
replacing present pages with hpage when iterating over the target range.

The present pages are already locked, so replacing them with the locked
hpage before the collapse is finalized is unnecessary.  However, it is
necessary to stop freezing the present pages after validating them, since
leaving long-term frozen pages in the page cache can lead to deadlocks.
Simply checking the reference count is sufficient to ensure that there are
no long-term references hanging around that would the collapse would
break.  Similar to hpage, there is no reason that the present pages
actually need to be frozen in addition to being locked.

This fixes a race where folio_seek_hole_data would mistake hpage for an
fallocated but unwritten page.  This race is visible to userspace via data
temporarily disappearing from SEEK_DATA/SEEK_HOLE.  This also fixes a
similar race where pages could temporarily disappear from mincore.

Link: https://lkml.kernel.org/r/20230404120117.2562166-5-stevensd@google.com
Fixes: f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages")
Signed-off-by: David Stevens <stevensd@chromium.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jiaqi Yan <jiaqiyan@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Peter Xu <peterx@redhat.com>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/khugepaged: skip shmem with userfaultfd

Make sure that collapse_file respects any userfaultfds registered with
MODE_MISSING.  If userspace has any such userfaultfds registered, then for
any page which it knows to be missing, it may expect a
UFFD_EVENT_PAGEFAULT.  This means collapse_file needs to be careful when
collapsing a shmem range would result in replacing an empty page with a
THP, to avoid breaking userfaultfd.

Synchronization when checking for userfaultfds in collapse_file is tricky
because the mmap locks can't be used to prevent races with the
registration of new userfaultfds.  Instead, we provide synchronization by
ensuring that userspace cannot observe the fact that pages are missing
before we check for userfaultfds.  Although this allows registration of a
userfaultfd to race with collapse_file, it ensures that userspace cannot
observe any pages transition from missing to present after such a race
occurs.  This makes such a race indistinguishable to the collapse
occurring immediately before the userfaultfd registration.

The first step to provide this synchronization is to stop filling gaps
during the loop iterating over the target range, since the page cache lock
can be dropped during that loop.  The second step is to fill the gaps with
XA_RETRY_ENTRY after the page cache lock is acquired the final time, to
avoid races with accesses to the page cache that only take the RCU read
lock.

The fact that we don't fill holes during the initial iteration means that
collapse_file now has to handle faults occurring during the collapse.
This is done by re-validating the number of missing pages after acquiring
the page cache lock for the final time.

This fix is targeted at khugepaged, but the change also applies to
MADV_COLLAPSE.  MADV_COLLAPSE on a range with a userfaultfd will now
return EBUSY if there are any missing pages (instead of succeeding on
shmem and returning EINVAL on anonymous memory).  There is also now a
window during MADV_COLLAPSE where a fault on a missing page will cause the
syscall to fail with EAGAIN.

The fact that intermediate page cache state can no longer be observed
before the rollback of a failed collapse is also technically a
userspace-visible change (via at least SEEK_DATA and SEEK_END), but it is
exceedingly unlikely that anything relies on being able to observe that
transient state.

Link: https://lkml.kernel.org/r/20230404120117.2562166-4-stevensd@google.com
Signed-off-by: David Stevens <stevensd@chromium.org>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jiaqi Yan <jiaqiyan@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/khugepaged: refactor collapse_file control flow

Add a rollback label to deal with failure, instead of continuously
checking for RESULT_SUCCESS, to make it easier to add more failure cases.
The refactoring also allows the collapse_file tracepoint to include hpage
on success (instead of NULL).

Link: https://lkml.kernel.org/r/20230404120117.2562166-3-stevensd@google.com
Signed-off-by: David Stevens <stevensd@chromium.org>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jiaqi Yan <jiaqiyan@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/khugepaged: drain lru after swapping in shmem

Patch series "mm/khugepaged: fixes for khugepaged+shmem", v6.

This series reworks collapse_file so that the intermediate state of the
collapse does not leak out of collapse_file. Although this makes
collapse_file a bit more complicated, it means that the rest of the
kernel doesn't have to deal with the unusual state. This directly fixes
races with both lseek and mincore.

This series also fixes the fact that khugepaged completely breaks
userfaultfd+shmem. The rework of collapse_file provides a convenient
place to check for registered userfaultfds without making the shmem
userfaultfd implementation care about khugepaged.

Finally, this series adds a lru_add_drain after swapping in shmem pages,
which makes the subsequent folio_isolate_lru significantly more likely to
succeed.

This patch (of 4):

Call lru_add_drain after swapping in shmem pages so that isolate_lru_page
is more likely to succeed.

Link: https://lkml.kernel.org/r/20230404120117.2562166-1-stevensd@google.com
Link: https://lkml.kernel.org/r/20230404120117.2562166-2-stevensd@google.com
Signed-off-by: David Stevens <stevensd@chromium.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jiaqi Yan <jiaqiyan@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Peter Xu <peterx@redhat.com>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/khugepaged: recover from poisoned file-backed memory

Make collapse_file roll back when copying pages failed. More concretely:
- extract copying operations into a separate loop
- postpone the updates for nr_none until both scanning and copying
  succeeded
- postpone joining small xarray entries until both scanning and copying
  succeeded
- postpone the update operations to NR_XXX_THPS until both scanning and
  copying succeeded
- for non-SHMEM file, roll back filemap_nr_thps_inc if scan succeeded but
  copying failed

Tested manually:
0. Enable khugepaged on system under test. Mount tmpfs at /mnt/ramdisk.
1. Start a two-thread application. Each thread allocates a chunk of
   non-huge memory buffer from /mnt/ramdisk.
2. Pick 4 random buffer address (2 in each thread) and inject
   uncorrectable memory errors at physical addresses.
3. Signal both threads to make their memory buffer collapsible, i.e.
   calling madvise(MADV_HUGEPAGE).
4. Wait and then check kernel log: khugepaged is able to recover from
   poisoned pages by skipping them.
5. Signal both threads to inspect their buffer contents and make sure no
   data corruption.

Link: https://lkml.kernel.org/r/20230329151121.949896-4-jiaqiyan@google.com
Signed-off-by: Jiaqi Yan <jiaqiyan@google.com>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: David Stevens <stevensd@chromium.org>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Tong Tiangen <tongtiangen@huawei.com>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>