Arnd Bergmann [Wed, 15 Feb 2023 13:00:58 +0000 (14:00 +0100)]
objtool: add UACCESS exceptions for __tsan_volatile_read/write
A lot of the tsan helpers are already excempt from the UACCESS warnings,
but some more functions were added that need the same thing:
kernel/kcsan/core.o: warning: objtool: __tsan_volatile_read16+0x0: call to __tsan_unaligned_read16() with UACCESS enabled
kernel/kcsan/core.o: warning: objtool: __tsan_volatile_write16+0x0: call to __tsan_unaligned_write16() with UACCESS enabled
vmlinux.o: warning: objtool: __tsan_unaligned_volatile_read16+0x4: call to __tsan_unaligned_read16() with UACCESS enabled
vmlinux.o: warning: objtool: __tsan_unaligned_volatile_write16+0x4: call to __tsan_unaligned_write16() with UACCESS enabled
As Marco points out, these functions don't even call each other
explicitly but instead gcc (but not clang) notices the functions
being identical and turns one symbol into a direct branch to the
other.
Link: https://lkml.kernel.org/r/20230215130058.3836177-4-arnd@kernel.org Fixes: 75d75b7a4d54 ("kcsan: Support distinguishing volatile accesses") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Marco Elver <elver@google.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: Josh Poimboeuf <jpoimboe@kernel.org> Cc: Kuan-Ying Lee <Kuan-Ying.Lee@mediatek.com> Cc: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Arnd Bergmann [Wed, 15 Feb 2023 13:00:57 +0000 (14:00 +0100)]
kmsan: disable ftrace in kmsan core code
objtool warns about some suspicous code inside of kmsan:
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_load_n+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_store_n+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_load_1+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_store_1+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_load_2+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_store_2+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_load_4+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_store_4+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_load_8+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_metadata_ptr_for_store_8+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_instrument_asm_store+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_chain_origin+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_poison_alloca+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_warning+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: __msan_get_context_state+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: kmsan_copy_to_user+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: kmsan_unpoison_memory+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: kmsan_unpoison_entry_regs+0x4: call to __fentry__() with UACCESS enabled
vmlinux.o: warning: objtool: kmsan_report+0x4: call to __fentry__() with UACCESS enabled
The Makefile contained a line to turn off ftrace for the entire directory,
but this does not work. Replace it with individual lines, matching the
approach in kasan.
Arnd Bergmann [Wed, 15 Feb 2023 13:00:56 +0000 (14:00 +0100)]
kasan: mark addr_has_metadata __always_inline
Patch series "objtool warning fixes", v2.
These are three of the easier fixes for objtool warnings around
kasan/kmsan/kcsan. I dropped one patch since Peter had come up with a
better fix, and adjusted the changelog text based on feedback.
This patch (of 3):
When the compiler decides not to inline this function, objtool complains
about incorrect UACCESS state:
mm/kasan/generic.o: warning: objtool: __asan_load2+0x11: call to addr_has_metadata() with UACCESS enabled
Link: https://lore.kernel.org/all/20230208164011.2287122-1-arnd@kernel.org/ Link: https://lkml.kernel.org/r/20230215130058.3836177-2-arnd@kernel.org Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Marco Elver <elver@google.com> Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com> Cc: Alexander Potapenko <glider@google.com> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Josh Poimboeuf <jpoimboe@kernel.org> Cc: Kuan-Ying Lee <Kuan-Ying.Lee@mediatek.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Roman Gushchin [Mon, 13 Feb 2023 19:29:22 +0000 (11:29 -0800)]
mm: memcontrol: rename memcg_kmem_enabled()
Currently there are two kmem-related helper functions with a confusing
semantics: memcg_kmem_enabled() and mem_cgroup_kmem_disabled().
The problem is that an obvious expectation
memcg_kmem_enabled() == !mem_cgroup_kmem_disabled(),
can be false.
mem_cgroup_kmem_disabled() is similar to mem_cgroup_disabled(): it returns
true only if CONFIG_MEMCG_KMEM is not set or the kmem accounting is
disabled using a boot time kernel option "cgroup.memory=nokmem". It never
changes the value dynamically.
memcg_kmem_enabled() is different: it always returns false until the first
non-root memory cgroup will get online (assuming the kernel memory
accounting is enabled). It's goal is to improve the performance on
systems without the cgroupfs mounted/memory controller enabled or on the
systems with only the root memory cgroup.
To make things more obvious and avoid potential bugs, let's rename
memcg_kmem_enabled() to memcg_kmem_online().
Link: https://lkml.kernel.org/r/20230213192922.1146370-1-roman.gushchin@linux.dev Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev> Acked-by: Muchun Song <songmuchun@bytedance.com> Acked-by: Michal Hocko <mhocko@suse.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Shakeel Butt <shakeelb@google.com> Cc: Dennis Zhou <dennis@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
On m68k/nommu RAM does not necessarily start at 0x0 and when it does not
pfn_valid() uses a wrong offset into the memory map which causes silent
boot failures.
Define ARCH_PFN_OFFSET to make pfn_valid() use the correct offset.
Link: https://lkml.kernel.org/r/20230214140729.1649961-1-rppt@kernel.org Link: https://lkml.kernel.org/r/20230214140729.1649961-2-rppt@kernel.org Fixes: d82f07f06cf8 ("m68k: use asm-generic/memory_model.h for both MMU and !MMU") Reported-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Mike Rapoport (IBM) <rppt@kernel.org> Acked-by: Greg Ungerer <gerg@linux-m68k.org> Reviewed-by: David Hildenbrand <david@redhat.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Rich Felker <dalias@libc.org> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Yafang Shao [Tue, 14 Feb 2023 15:35:49 +0000 (15:35 +0000)]
mm: percpu: fix incorrect size in pcpu_obj_full_size()
The extra space which is used to store the obj_cgroup membership is only
valid when kmemcg is enabled. The kmemcg can be disabled via the kernel
parameter "cgroup.memory=nokmem" at boot time. This helper is also used
in non-memcg code, for example the tracepoint, so we should fix it.
It was found by code review when I was implementing bpf memory usage[1].
No real issue happens in production environment.
[1]. https://lwn.net/Articles/921991/
Link: https://lkml.kernel.org/r/20230214153549.12291-1-laoar.shao@gmail.com Signed-off-by: Yafang Shao <laoar.shao@gmail.com> Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev> Acked-by: Dennis Zhou <dennis@kernel.org> Cc: Tejun Heo <tj@kernel.org> Cc: Christoph Lameter <cl@linux.com> Cc: Vasily Averin <vvs@openvz.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Arnd Bergmann [Tue, 14 Feb 2023 10:30:24 +0000 (11:30 +0100)]
maple_tree: reduce stack usage with gcc-9 and earlier
gcc-10 changed the way inlining works to be less aggressive, but older
versions run into an oversized stack frame warning whenever
CONFIG_KASAN_STACK is enabled, as that forces variables from inlined
callees to be non-overlapping:
lib/maple_tree.c: In function 'mas_wr_bnode':
lib/maple_tree.c:4320:1: error: the frame size of 1424 bytes is larger than 1024 bytes [-Werror=frame-larger-than=]
Change the annotations on mas_store_b_node() and mas_commit_b_node()
to explicitly forbid inlining in this configuration, which is
the same behavior that newer versions already have.
Link: https://lkml.kernel.org/r/20230214103030.1051950-1-arnd@kernel.org Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: David Hildenbrand <david@redhat.com> Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com> Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com> Cc: Alexander Potapenko <glider@google.com> Cc: Andrey Konovalov <andreyknvl@gmail.com> Cc: Dmitry Vyukov <dvyukov@google.com> Cc: Vincenzo Frascino <vincenzo.frascino@arm.com> Cc: Vernon Yang <vernon2gm@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Qi Zheng [Sun, 12 Feb 2023 11:10:27 +0000 (19:10 +0800)]
mm: page_alloc: call panic() when memoryless node allocation fails
In free_area_init(), we will continue to run after allocation of
memoryless node pgdat fails. However, in the subsequent process (such as
when initializing zonelist), the case that NODE_DATA(nid) is NULL is not
handled, which will cause panic. Instead of this, it's better to call
panic() directly when the memory allocation fails during system boot.
Yu Zhao [Mon, 13 Feb 2023 07:53:22 +0000 (00:53 -0700)]
mm: multi-gen LRU: avoid futile retries
Recall that the per-node memcg LRU has two generations and they alternate
when the last memcg (of a given node) is moved from one to the other.
Each generation is also sharded into multiple bins to improve scalability.
A reclaimer starts with a random bin (in the old generation) and, if it
fails, it will retry, i.e., to try the rest of the bins.
If a reclaimer fails with the last memcg, it should move this memcg to the
young generation first, which causes the generations to alternate, and
then retry. Otherwise, the retries will be futile because all other bins
are empty.
Huang Ying [Mon, 13 Feb 2023 12:34:44 +0000 (20:34 +0800)]
migrate_pages: move THP/hugetlb migration support check to simplify code
This is a code cleanup patch, no functionality change is expected. After
the change, the line number reduces especially in the long
migrate_pages_batch().
Link: https://lkml.kernel.org/r/20230213123444.155149-10-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Suggested-by: Alistair Popple <apopple@nvidia.com> Reviewed-by: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Xin Hao <xhao@linux.alibaba.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:43 +0000 (20:34 +0800)]
migrate_pages: batch flushing TLB
The TLB flushing will cost quite some CPU cycles during the folio
migration in some situations. For example, when migrate a folio of a
process with multiple active threads that run on multiple CPUs. After
batching the _unmap and _move in migrate_pages(), the TLB flushing can be
batched easily with the existing TLB flush batching mechanism. This patch
implements that.
We use the following test case to test the patch.
On a 2-socket Intel server,
- Run pmbench memory accessing benchmark
- Run `migratepages` to migrate pages of pmbench between node 0 and
node 1 back and forth.
With the patch, the TLB flushing IPI reduces 99.1% during the test and the
number of pages migrated successfully per second increases 291.7%.
Haoxin helped to test the patchset on an ARM64 server with 128 cores, 2
NUMA nodes. Test results show that the page migration performance
increases up to 78%.
NOTE: TLB flushing is batched only for normal folios, not for THP folios.
Because the overhead of TLB flushing for THP folios is much lower than
that for normal folios (about 1/512 on x86 platform).
Link: https://lkml.kernel.org/r/20230213123444.155149-9-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Tested-by: Xin Hao <xhao@linux.alibaba.com> Reviewed-by: Zi Yan <ziy@nvidia.com> Reviewed-by: Xin Hao <xhao@linux.alibaba.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:42 +0000 (20:34 +0800)]
migrate_pages: share more code between _unmap and _move
This is a code cleanup patch to reduce the duplicated code between the
_unmap and _move stages of migrate_pages(). No functionality change is
expected.
Link: https://lkml.kernel.org/r/20230213123444.155149-8-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Xin Hao <xhao@linux.alibaba.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:41 +0000 (20:34 +0800)]
migrate_pages: move migrate_folio_unmap()
Just move the position of the functions. There's no any functionality
change. This is to make it easier to review the next patch via putting
code near its position in the next patch.
Link: https://lkml.kernel.org/r/20230213123444.155149-7-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Xin Hao <xhao@linux.alibaba.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:40 +0000 (20:34 +0800)]
migrate_pages: batch _unmap and _move
In this patch the _unmap and _move stage of the folio migration is
batched. That for, previously, it is,
for each folio
_unmap()
_move()
Now, it is,
for each folio
_unmap()
for each folio
_move()
Based on this, we can batch the TLB flushing and use some hardware
accelerator to copy folios between batched _unmap and batched _move
stages.
Link: https://lkml.kernel.org/r/20230213123444.155149-6-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Tested-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Xin Hao <xhao@linux.alibaba.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:39 +0000 (20:34 +0800)]
migrate_pages: split unmap_and_move() to _unmap() and _move()
This is a preparation patch to batch the folio unmapping and moving.
In this patch, unmap_and_move() is split to migrate_folio_unmap() and
migrate_folio_move(). So, we can batch _unmap() and _move() in different
loops later. To pass some information between unmap and move, the
original unused dst->mapping and dst->private are used.
Link: https://lkml.kernel.org/r/20230213123444.155149-5-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by: Xin Hao <xhao@linux.alibaba.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:38 +0000 (20:34 +0800)]
migrate_pages: restrict number of pages to migrate in batch
This is a preparation patch to batch the folio unmapping and moving for
non-hugetlb folios.
If we had batched the folio unmapping, all folios to be migrated would be
unmapped before copying the contents and flags of the folios. If the
folios that were passed to migrate_pages() were too many in unit of pages,
the execution of the processes would be stopped for too long time, thus
too long latency. For example, migrate_pages() syscall will call
migrate_pages() with all folios of a process. To avoid this possible
issue, in this patch, we restrict the number of pages to be migrated to be
no more than HPAGE_PMD_NR. That is, the influence is at the same level of
THP migration.
Link: https://lkml.kernel.org/r/20230213123444.155149-4-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Xin Hao <xhao@linux.alibaba.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:37 +0000 (20:34 +0800)]
migrate_pages: separate hugetlb folios migration
This is a preparation patch to batch the folio unmapping and moving for
the non-hugetlb folios. Based on that we can batch the TLB shootdown
during the folio migration and make it possible to use some hardware
accelerator for the folio copying.
In this patch the hugetlb folios and non-hugetlb folios migration is
separated in migrate_pages() to make it easy to change the non-hugetlb
folios migration implementation.
Link: https://lkml.kernel.org/r/20230213123444.155149-3-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by: Xin Hao <xhao@linux.alibaba.com> Cc: Zi Yan <ziy@nvidia.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Huang Ying [Mon, 13 Feb 2023 12:34:36 +0000 (20:34 +0800)]
migrate_pages: organize stats with struct migrate_pages_stats
Patch series "migrate_pages(): batch TLB flushing", v5.
Now, migrate_pages() migrates folios one by one, like the fake code as
follows,
for each folio
unmap
flush TLB
copy
restore map
If multiple folios are passed to migrate_pages(), there are opportunities
to batch the TLB flushing and copying. That is, we can change the code to
something as follows,
for each folio
unmap
for each folio
flush TLB
for each folio
copy
for each folio
restore map
The total number of TLB flushing IPI can be reduced considerably. And we
may use some hardware accelerator such as DSA to accelerate the folio
copying.
So in this patch, we refactor the migrate_pages() implementation and
implement the TLB flushing batching. Base on this, hardware accelerated
folio copying can be implemented.
If too many folios are passed to migrate_pages(), in the naive batched
implementation, we may unmap too many folios at the same time. The
possibility for a task to wait for the migrated folios to be mapped again
increases. So the latency may be hurt. To deal with this issue, the max
number of folios be unmapped in batch is restricted to no more than
HPAGE_PMD_NR in the unit of page. That is, the influence is at the same
level of THP migration.
We use the following test to measure the performance impact of the
patchset,
On a 2-socket Intel server,
- Run pmbench memory accessing benchmark
- Run `migratepages` to migrate pages of pmbench between node 0 and
node 1 back and forth.
With the patch, the TLB flushing IPI reduces 99.1% during the test and
the number of pages migrated successfully per second increases 291.7%.
Xin Hao helped to test the patchset on an ARM64 server with 128 cores,
2 NUMA nodes. Test results show that the page migration performance
increases up to 78%.
This patch (of 9):
Define struct migrate_pages_stats to organize the various statistics in
migrate_pages(). This makes it easier to collect and consume the
statistics in multiple functions. This will be needed in the following
patches in the series.
Link: https://lkml.kernel.org/r/20230213123444.155149-1-ying.huang@intel.com Link: https://lkml.kernel.org/r/20230213123444.155149-2-ying.huang@intel.com Signed-off-by: "Huang, Ying" <ying.huang@intel.com> Reviewed-by: Alistair Popple <apopple@nvidia.com> Reviewed-by: Zi Yan <ziy@nvidia.com> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com> Reviewed-by: Xin Hao <xhao@linux.alibaba.com> Cc: Yang Shi <shy828301@gmail.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Matthew Wilcox <willy@infradead.org> Cc: Bharata B Rao <bharata@amd.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Andrey Konovalov [Fri, 10 Feb 2023 21:16:04 +0000 (22:16 +0100)]
lib/stackdepot: annotate racy pool_index accesses
Accesses to pool_index are protected by pool_lock everywhere except
in a sanity check in stack_depot_fetch. The read access there can race
with the write access in depot_alloc_stack.
Use WRITE/READ_ONCE() to annotate the racy accesses.
As the sanity check is only used to print a warning in case of a
violation of the stack depot interface usage, it does not make a lot
of sense to use proper synchronization.
The current implementation of the extra_bits interface is confusing:
passing extra_bits to __stack_depot_save makes it seem that the extra
bits are somehow stored in stack depot. In reality, they are only
embedded into a stack depot handle and are not used within stack depot.
Drop the extra_bits argument from __stack_depot_save and instead provide
a new stack_depot_set_extra_bits function (similar to the exsiting
stack_depot_get_extra_bits) that saves extra bits into a stack depot
handle.
Update the callers of __stack_depot_save to use the new interace.
This change also fixes a minor issue in the old code: __stack_depot_save
does not return NULL if saving stack trace fails and extra_bits is used.
Andrey Konovalov [Fri, 10 Feb 2023 21:16:02 +0000 (22:16 +0100)]
lib/stackdepot: rename next_pool_inited to next_pool_required
Stack depot uses next_pool_inited to mark that either the next pool is
initialized or the limit on the number of pools is reached. However,
the flag name only reflects the former part of its purpose, which is
confusing.
Rename next_pool_inited to next_pool_required and invert its value.
Also annotate usages of next_pool_required with comments.
Andrey Konovalov [Fri, 10 Feb 2023 21:16:00 +0000 (22:16 +0100)]
lib/stacktrace: drop impossible WARN_ON for depot_init_pool
depot_init_pool has two call sites:
1. In depot_alloc_stack with a potentially NULL prealloc.
2. In __stack_depot_save with a non-NULL prealloc.
At the same time depot_init_pool can only return false when prealloc is
NULL.
As the second call site makes sure that prealloc is not NULL, the WARN_ON
there can never trigger. Thus, drop the WARN_ON and also move the prealloc
check from depot_init_pool to its first call site.
Also change the return type of depot_init_pool to void as it now always
returns true.
Andrey Konovalov [Fri, 10 Feb 2023 21:15:57 +0000 (22:15 +0100)]
lib/stackdepot: rename slab to pool
Use "pool" instead of "slab" for naming memory regions stack depot
uses to store stack traces. Using "slab" is confusing, as stack depot
pools have nothing to do with the slab allocator.
Also give better names to pool-related global variables: change
"depot_" prefix to "pool_" to point out that these variables are
related to stack depot pools.
Also rename the slabindex (poolindex) field in handle_parts to pool_index
to align its name with the pool_index global variable.
Rename stack_depot_want_early_init to stack_depot_request_early_init.
The old name is confusing, as it hints at returning some kind of intention
of stack depot. The new name reflects that this function requests an
action from stack depot instead.
Andrey Konovalov [Fri, 10 Feb 2023 21:15:49 +0000 (22:15 +0100)]
lib/stackdepot: put functions in logical order
Patch series "lib/stackdepot: fixes and clean-ups", v2.
A set of fixes, comments, and clean-ups I came up with while reading
the stack depot code.
This patch (of 18):
Put stack depot functions' declarations and definitions in a more logical
order:
1. Functions that save stack traces into stack depot.
2. Functions that fetch and print stack traces.
3. stack_depot_get_extra_bits that operates on stack depot handles
and does not interact with the stack depot storage.
Jakub Wilk [Fri, 10 Feb 2023 20:33:16 +0000 (21:33 +0100)]
mm: fix typo in __vm_enough_memory warning
Link: https://lkml.kernel.org/r/20230210203316.5613-1-jwilk@jwilk.net Signed-off-by: Jakub Wilk <jwilk@jwilk.net> Acked-by: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
DAMON debugfs interface has announced to be deprecated after >v5.15 LTS
kernel is released. And, v6.1.y has announced to be an LTS[1].
Though the announcement was there for a while, some people might not
noticed that so far. Also, some users could depend on it and have
problems at movng to the alternative (DAMON sysfs interface).
For such cases, warn DAMON debugfs interface deprecation with contacts
to ask helps when any DAMON debugfs interface file is opened.
DAMON debugfs interface has announced to be deprecated after >v5.15 LTS
kernel is released. And, v6.1.y has announced to be an LTS[1].
Though the announcement was there for a while, some people might not
noticed that so far. Also, some users could depend on it and have
problems at movng to the alternative (DAMON sysfs interface).
For such cases, note DAMON debugfs interface as deprecated, and contacts
to ask helps on the Kconfig.
Patch series "mm/damon: deprecate DAMON debugfs interface".
DAMON debugfs interface has announced to be deprecated after >v5.15 LTS
kernel is released. And v6.1.y has been announced to be an LTS[1].
Though the announcement was there for a while, some people might not have
noticed that so far. Also, some users could depend on it and have
problems at movng to the alternative (DAMON sysfs interface).
For such cases, keep the code and documents with warning messages and
contacts to ask helps for the deprecation.
DAMON debugfs interface has announced to be deprecated after >v5.15 LTS
kernel is released. And, v6.1.y has announced to be an LTS[1].
Though the announcement was there for a while, some people might not
noticed that so far. Also, some users could depend on it and have
problems at movng to the alternative (DAMON sysfs interface).
For such cases, note DAMON debugfs interface as deprecated, and contacts
to ask helps on the document.
folio_movable_ops() does the same as page_movable_ops() except uses folios
instead of pages. This function will help make folio conversions in
migrate.c more readable.
Patch series "Convert a couple migrate functions to use folios", v2.
This patchset introduces folio_movable_ops() and converts 3 functions in
mm/migrate.c to use folios. It also introduces folio_get_nontail_page()
for folio conversions which may want to distinguish between head and tail
pages.
This patch (of 4):
folio_get_nontail_page() returns the folio associated with a head page.
This is necessary for folio conversions where the behavior of that
function differs between head pages and tail pages.
mm/mempolicy: convert migrate_page_add() to migrate_folio_add()
Replace migrate_page_add() with migrate_folio_add(). migrate_folio_add()
does the same a migrate_page_add() but takes in a folio instead of a page.
This removes a couple of calls to compound_head().
Link: https://lkml.kernel.org/r/20230130201833.27042-7-vishal.moola@gmail.com Signed-off-by: Vishal Moola (Oracle) <vishal.moola@gmail.com> Reviewed-by: Yin Fengwei <fengwei.yin@intel.com> Cc: David Hildenbrand <david@redhat.com> Cc: Jane Chu <jane.chu@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
mm/mempolicy: convert queue_pages_required() to queue_folio_required()
Replace queue_pages_required() with queue_folio_required().
queue_folio_required() does the same as queue_pages_required(), except
takes in a folio instead of a page.
Link: https://lkml.kernel.org/r/20230130201833.27042-6-vishal.moola@gmail.com Signed-off-by: Vishal Moola (Oracle) <vishal.moola@gmail.com> Cc: David Hildenbrand <david@redhat.com> Cc: Jane Chu <jane.chu@oracle.com> Cc: "Yin, Fengwei" <fengwei.yin@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "Convert various mempolicy.c functions to use folios", v4.
This patch series converts migrate_page_add() and queue_pages_required()
to migrate_folio_add() and queue_page_required(). It also converts the
callers of the functions to use folios as well, and introduces a helper
function to estimate the number of sharers of a folio.
This patch (of 6):
folio_estimated_sharers() takes in a folio and returns the precise number
of times the first subpage of the folio is mapped.
This function aims to provide an estimate for the number of sharers of a
folio. This is necessary for folio conversions where we care about the
number of processes that share a folio, but don't necessarily want to
check every single page within that folio.
This is in contrast to folio_mapcount() which calculates the total number
of the times a folio and all its subpages are mapped.
Sidhartha Kumar [Wed, 25 Jan 2023 17:05:35 +0000 (09:05 -0800)]
mm/hugetlb: convert hugetlb_add_to_page_cache to take in a folio
Every caller of hugetlb_add_to_page_cache() is now passing in
&folio->page, change the function to take in a folio directly and clean up
the call sites.
Link: https://lkml.kernel.org/r/20230125170537.96973-7-sidhartha.kumar@oracle.com Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Wed, 25 Jan 2023 17:05:34 +0000 (09:05 -0800)]
mm/hugetlb: convert restore_reserve_on_error to take in a folio
Every caller of restore_reserve_on_error() is now passing in &folio->page,
change the function to take in a folio directly and clean up the call
sites.
Link: https://lkml.kernel.org/r/20230125170537.96973-6-sidhartha.kumar@oracle.com Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Wed, 25 Jan 2023 17:05:33 +0000 (09:05 -0800)]
mm/hugetlb: convert hugetlb fault paths to use alloc_hugetlb_folio()
Change alloc_huge_page() to alloc_hugetlb_folio() by changing all callers
to handle the now folio return type of the function. In this conversion,
alloc_huge_page_vma() is also changed to alloc_hugetlb_folio_vma() and
hugepage_add_new_anon_rmap() is changed to take in a folio directly. Many
additions of '&folio->page' are cleaned up in subsequent patches.
hugetlbfs_fallocate() is also refactored to use the RCU +
page_cache_next_miss() API.
Link: https://lkml.kernel.org/r/20230125170537.96973-5-sidhartha.kumar@oracle.com Suggested-by: Mike Kravetz <mike.kravetz@oracle.com> Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Wed, 25 Jan 2023 17:05:32 +0000 (09:05 -0800)]
mm/hugetlb: convert putback_active_hugepage to take in a folio
Convert putback_active_hugepage() to folio_putback_active_hugetlb(), this
removes one user of the Huge Page macros which take in a page. The
callers in migrate.c are also cleaned up by being able to directly use the
src and dst folio variables.
Link: https://lkml.kernel.org/r/20230125170537.96973-4-sidhartha.kumar@oracle.com Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Wed, 25 Jan 2023 17:05:30 +0000 (09:05 -0800)]
mm/hugetlb: convert hugetlb_install_page to folios
Patch series "convert hugetlb fault functions to folios", v2.
This series converts the hugetlb page faulting functions to operate on
folios. These include hugetlb_no_page(), hugetlb_wp(),
copy_hugetlb_page_range(), and hugetlb_mcopy_atomic_pte().
This patch (of 8):
Change hugetlb_install_page() to hugetlb_install_folio(). This reduces
one user of the Huge Page flag macros which take in a page.
Sidhartha Kumar [Fri, 13 Jan 2023 22:30:55 +0000 (16:30 -0600)]
mm/hugetlb: convert alloc_migrate_huge_page to folios
Change alloc_huge_page_nodemask() to alloc_hugetlb_folio_nodemask() and
alloc_migrate_huge_page() to alloc_migrate_hugetlb_folio(). Both
functions now return a folio rather than a page.
Link: https://lkml.kernel.org/r/20230113223057.173292-7-sidhartha.kumar@oracle.com Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Fri, 13 Jan 2023 22:30:54 +0000 (16:30 -0600)]
mm/hugetlb: increase use of folios in alloc_huge_page()
Change hugetlb_cgroup_commit_charge{,_rsvd}(), dequeue_huge_page_vma() and
alloc_buddy_huge_page_with_mpol() to use folios so alloc_huge_page() is
cleaned by operating on folios until its return.
Link: https://lkml.kernel.org/r/20230113223057.173292-6-sidhartha.kumar@oracle.com Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Fri, 13 Jan 2023 22:30:52 +0000 (16:30 -0600)]
mm/hugetlb: convert dequeue_hugetlb_page functions to folios
dequeue_huge_page_node_exact() is changed to dequeue_hugetlb_folio_node_
exact() and dequeue_huge_page_nodemask() is changed to dequeue_hugetlb_
folio_nodemask(). Update their callers to pass in a folio.
Link: https://lkml.kernel.org/r/20230113223057.173292-4-sidhartha.kumar@oracle.com Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Matthew Wilcox <willy@infradead.org> Cc: Mike Kravetz <mike.kravetz@oracle.com> Cc: Muchun Song <songmuchun@bytedance.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Sidhartha Kumar [Fri, 13 Jan 2023 22:30:50 +0000 (16:30 -0600)]
mm/hugetlb: convert isolate_hugetlb to folios
Patch series "continue hugetlb folio conversion", v3.
This series continues the conversion of core hugetlb functions to use
folios. This series converts many helper funtions in the hugetlb fault
path. This is in preparation for another series to convert the hugetlb
fault code paths to operate on folios.
This patch (of 8):
Convert isolate_hugetlb() to take in a folio and convert its callers to
pass a folio. Use page_folio() to convert the callers to use a folio is
safe as isolate_hugetlb() operates on a head page.
mm/khugepaged: fix invalid page access in release_pte_pages()
release_pte_pages() converts from a pfn to a folio by using pfn_folio().
If the pte is not mapped, pfn_folio() will result in undefined behavior
which ends up causing a kernel panic[1].
Only call pfn_folio() once we have validated that the pte is both valid
and mapped to fix the issue.
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:34 +0000 (16:34 -0400)]
mm/gup: move private gup FOLL_ flags to internal.h
Move the flags that should not/are not used outside gup.c and related into
mm/internal.h to discourage driver abuse.
To make this more maintainable going forward compact the two FOLL ranges
with new bit numbers from 0 to 11 and 16 to 21, using shifts so it is
explicit.
Switch to an enum so the whole thing is easier to read.
Link: https://lkml.kernel.org/r/13-v2-987e91b59705+36b-gup_tidy_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Reviewed-by: John Hubbard <jhubbard@nvidia.com> Acked-by: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Claudio Imbrenda <imbrenda@linux.ibm.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:29 +0000 (16:34 -0400)]
mm/gup: add FOLL_UNLOCKABLE
Setting FOLL_UNLOCKABLE allows GUP to lock/unlock the mmap lock on its
own. It is a more explicit replacement for locked != NULL. This clears
the way for passing in locked = 1, without intending that the lock can be
unlocked.
Set the flag in all cases where it is used, eg locked is present in the
external interface or locked is used internally with locked = 0.
Link: https://lkml.kernel.org/r/8-v2-987e91b59705+36b-gup_tidy_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Acked-by: Mike Rapoport (IBM) <rppt@kernel.org> Reviewed-by: John Hubbard <jhubbard@nvidia.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Claudio Imbrenda <imbrenda@linux.ibm.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:27 +0000 (16:34 -0400)]
mm/gup: add an assertion that the mmap lock is locked
Since commit 5b78ed24e8ec ("mm/pagemap: add mmap_assert_locked()
annotations to find_vma*()") we already have this assertion, it is just
buried in find_vma():
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:26 +0000 (16:34 -0400)]
mm/gup: simplify the external interface functions and consolidate invariants
The GUP family of functions have a complex, but fairly well defined, set
of invariants for their arguments. Currently these are sprinkled about,
sometimes in duplicate through many functions.
Internally we don't follow all the invariants that the external interface
has to follow, so place these checks directly at the exported interface.
This ensures the internal functions never reach a violated invariant.
Remove the duplicated invariant checks.
The end result is to make these functions fully internal:
__get_user_pages_locked()
internal_get_user_pages_fast()
__gup_longterm_locked()
And all the other functions call directly into one of these.
Link: https://lkml.kernel.org/r/5-v2-987e91b59705+36b-gup_tidy_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Suggested-by: John Hubbard <jhubbard@nvidia.com> Reviewed-by: John Hubbard <jhubbard@nvidia.com> Acked-by: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Alistair Popple <apopple@nvidia.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Claudio Imbrenda <imbrenda@linux.ibm.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:24 +0000 (16:34 -0400)]
mm/gup: don't call __gup_longterm_locked() if FOLL_LONGTERM cannot be set
get_user_pages_remote(), get_user_pages_unlocked() and get_user_pages()
are never called with FOLL_LONGTERM, so directly call
__get_user_pages_locked()
The next patch will add an assertion for this.
Link: https://lkml.kernel.org/r/3-v2-987e91b59705+36b-gup_tidy_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Suggested-by: John Hubbard <jhubbard@nvidia.com> Reviewed-by: John Hubbard <jhubbard@nvidia.com> Acked-by: Mike Rapoport (IBM) <rppt@kernel.org> Cc: Alistair Popple <apopple@nvidia.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Claudio Imbrenda <imbrenda@linux.ibm.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:23 +0000 (16:34 -0400)]
mm/gup: remove obsolete FOLL_LONGTERM comment
These days FOLL_LONGTERM is not allowed at all on any get_user_pages*()
functions, it must be only be used with pin_user_pages*(), plus it now has
universal support for all the pin_user_pages*() functions.
Link: https://lkml.kernel.org/r/2-v2-987e91b59705+36b-gup_tidy_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Reviewed-by: John Hubbard <jhubbard@nvidia.com> Cc: Alistair Popple <apopple@nvidia.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Claudio Imbrenda <imbrenda@linux.ibm.com> Cc: David Hildenbrand <david@redhat.com> Cc: David Howells <dhowells@redhat.com> Cc: Mike Rapoport (IBM) <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Jason Gunthorpe [Tue, 24 Jan 2023 20:34:22 +0000 (16:34 -0400)]
mm/gup: have internal functions get the mmap_read_lock()
Patch series "Simplify the external interface for GUP", v2.
It is quite a maze of EXPORTED symbols leading up to the three actual
worker functions of GUP. Simplify this by reorganizing some of the code so
the EXPORTED symbols directly call the correct internal function with
validated and consistent arguments.
Consolidate all the assertions into one place at the top of the call
chains.
Remove some dead code.
Move more things into the mm/internal.h header
This patch (of 13):
__get_user_pages_locked() and __gup_longterm_locked() both require the
mmap lock to be held. They have a slightly unusual locked parameter that
is used to allow these functions to unlock and relock the mmap lock and
convey that fact to the caller.
Several places wrap these functions with a simple mmap_read_lock() just so
they can follow the optimized locked protocol.
Consolidate this internally to the functions. Allow internal callers to
set locked = 0 to cause the functions to acquire and release the lock on
their own.
Reorganize __gup_longterm_locked() to use the autolocking in
__get_user_pages_locked().
Replace all the places obtaining the mmap_read_lock() just to call
__get_user_pages_locked() with the new mechanism. Replace all the
internal callers of get_user_pages_unlocked() with direct calls to
__gup_longterm_locked() using the new mechanism.
A following patch will add assertions ensuring the external interface
continues to always pass in locked = 1.
Baoquan He [Mon, 6 Feb 2023 08:40:18 +0000 (16:40 +0800)]
mm/vmalloc: skip the uninitilized vmalloc areas
For areas allocated via vmalloc_xxx() APIs, it searches for unmapped area
to reserve and allocates new pages to map into, please see function
__vmalloc_node_range(). During the process, flag VM_UNINITIALIZED is set
in vm->flags to indicate that the pages allocation and mapping haven't
been done, until clear_vm_uninitialized_flag() is called to clear
VM_UNINITIALIZED.
For this kind of area, if VM_UNINITIALIZED is still set, let's ignore it
in vread() because pages newly allocated and being mapped in that area
only contains zero data. reading them out by aligned_vread() is wasting
time.
Link: https://lkml.kernel.org/r/20230206084020.174506-6-bhe@redhat.com Signed-off-by: Baoquan He <bhe@redhat.com> Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com> Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com> Cc: Dan Carpenter <error27@gmail.com> Cc: Stephen Brennan <stephen.s.brennan@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Baoquan He [Mon, 6 Feb 2023 08:40:17 +0000 (16:40 +0800)]
mm/vmalloc: explicitly identify vm_map_ram area when shown in /proc/vmcoreinfo
Now, by marking VMAP_RAM in vmap_area->flags for vm_map_ram area, we can
clearly differentiate it with other vmalloc areas. So identify
vm_map_area area by checking VMAP_RAM of vmap_area->flags when shown in
/proc/vmcoreinfo.
Meanwhile, the code comment above vm_map_ram area checking in s_show() is
not needed any more, remove it here.
Link: https://lkml.kernel.org/r/20230206084020.174506-5-bhe@redhat.com Signed-off-by: Baoquan He <bhe@redhat.com> Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com> Cc: Dan Carpenter <error27@gmail.com> Cc: Stephen Brennan <stephen.s.brennan@oracle.com> Cc: Uladzislau Rezki (Sony) <urezki@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Baoquan He [Mon, 6 Feb 2023 08:40:16 +0000 (16:40 +0800)]
mm/vmalloc.c: allow vread() to read out vm_map_ram areas
Currently, vread can read out vmalloc areas which is associated with a
vm_struct. While this doesn't work for areas created by vm_map_ram()
interface because it doesn't have an associated vm_struct. Then in
vread(), these areas are all skipped.
Here, add a new function vmap_ram_vread() to read out vm_map_ram areas.
The area created with vmap_ram_vread() interface directly can be handled
like the other normal vmap areas with aligned_vread(). While areas which
will be further subdivided and managed with vmap_block need carefully read
out page-aligned small regions and zero fill holes.
Link: https://lkml.kernel.org/r/20230206084020.174506-4-bhe@redhat.com Reported-by: Stephen Brennan <stephen.s.brennan@oracle.com> Signed-off-by: Baoquan He <bhe@redhat.com> Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com> Tested-by: Stephen Brennan <stephen.s.brennan@oracle.com> Cc: Dan Carpenter <error27@gmail.com> Cc: Uladzislau Rezki (Sony) <urezki@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Baoquan He [Mon, 6 Feb 2023 08:40:15 +0000 (16:40 +0800)]
mm/vmalloc.c: add flags to mark vm_map_ram area
Through vmalloc API, a virtual kernel area is reserved for physical
address mapping. And vmap_area is used to track them, while vm_struct is
allocated to associate with the vmap_area to store more information and
passed out.
However, area reserved via vm_map_ram() is an exception. It doesn't have
vm_struct to associate with vmap_area. And we can't recognize the
vmap_area with '->vm == NULL' as a vm_map_ram() area because the normal
freeing path will set va->vm = NULL before unmapping, please see function
remove_vm_area().
Meanwhile, there are two kinds of handling for vm_map_ram area. One is
the whole vmap_area being reserved and mapped at one time through
vm_map_area() interface; the other is the whole vmap_area with
VMAP_BLOCK_SIZE size being reserved, while mapped into split regions with
smaller size via vb_alloc().
To mark the area reserved through vm_map_ram(), add flags field into
struct vmap_area. Bit 0 indicates this is vm_map_ram area created through
vm_map_ram() interface, while bit 1 marks out the type of vm_map_ram area
which makes use of vmap_block to manage split regions via vb_alloc/free().
This is a preparation for later use.
Link: https://lkml.kernel.org/r/20230206084020.174506-3-bhe@redhat.com Signed-off-by: Baoquan He <bhe@redhat.com> Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com> Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com> Cc: Dan Carpenter <error27@gmail.com> Cc: Stephen Brennan <stephen.s.brennan@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Baoquan He [Mon, 6 Feb 2023 08:40:14 +0000 (16:40 +0800)]
mm/vmalloc.c: add used_map into vmap_block to track space of vmap_block
Patch series "mm/vmalloc.c: allow vread() to read out vm_map_ram areas", v5.
Problem:
***
Stephen reported vread() will skip vm_map_ram areas when reading out
/proc/kcore with drgn utility. Please see below link to get more details.
/proc/kcore reads 0's for vmap_block
https://lore.kernel.org/all/87ilk6gos2.fsf@oracle.com/T/#u
Root cause:
***
The normal vmalloc API uses struct vmap_area to manage the virtual kernel
area allocated, and associate a vm_struct to store more information and
pass out. However, area reserved through vm_map_ram() interface doesn't
allocate vm_struct to associate with. So the current code in vread() will
skip the vm_map_ram area through 'if (!va->vm)' conditional checking.
Solution:
***
To mark the area reserved through vm_map_ram() interface, add field
'flags' into struct vmap_area. Bit 0 indicates this is vm_map_ram area
created through vm_map_ram() interface, bit 1 marks out the type of
vm_map_ram area which makes use of vmap_block to manage split regions via
vb_alloc/free().
And also add bitmap field 'used_map' into struct vmap_block to mark those
further subdivided regions being used to differentiate with dirty and free
regions in vmap_block.
With the help of above vmap_area->flags and vmap_block->used_map, we can
recognize and handle vm_map_ram areas successfully. All these are done in
patch 1~3.
Meanwhile, do some improvement on areas related to vm_map_ram areas in
patch 4, 5. And also change area flag from VM_ALLOC to VM_IOREMAP in
patch 6, 7 because this will show them as 'ioremap' in /proc/vmallocinfo,
and exclude them from /proc/kcore.
This patch (of 7):
In one vmap_block area, there could be three types of regions: region
being used which is allocated through vb_alloc(), dirty region which is
freed via vb_free() and free region. Among them, only used region has
available data. While there's no way to track those used regions
currently.
Here, add bitmap field used_map into vmap_block, and set/clear it during
allocation or freeing regions of vmap_block area.
Yajun Deng [Fri, 3 Feb 2023 10:01:32 +0000 (18:01 +0800)]
mm/page_alloc: reduce fallbacks to (MIGRATE_PCPTYPES - 1)
The commit 1dd214b8f21c ("mm: page_alloc: avoid merging non-fallbackable
pageblocks with others") has removed MIGRATE_CMA and MIGRATE_ISOLATE from
fallbacks list. so there is no need to add an element at the end of every
type.
Reduce fallbacks to (MIGRATE_PCPTYPES - 1).
Link: https://lkml.kernel.org/r/20230203100132.1627787-1-yajun.deng@linux.dev Signed-off-by: Yajun Deng <yajun.deng@linux.dev> Acked-by: Vlastimil Babka <vbabka@suse.cz> Cc: Zi Yan <ziy@nvidia.com> Cc: Mel Gorman <mgorman@techsingularity.net> Cc: David Hildenbrand <david@redhat.com> Cc: Mike Rapoport <rppt@linux.ibm.com> Cc: Oscar Salvador <osalvador@suse.de> Cc: Mike Rapoport <rppt@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Hyunmin Lee [Wed, 1 Feb 2023 11:51:42 +0000 (20:51 +0900)]
mm/vmalloc: replace BUG_ON with a simple if statement
As per the coding standards, in the event of an abnormal condition that
should not occur under normal circumstances, the kernel should attempt
recovery and proceed with execution, rather than halting the machine.
Specifically, in the alloc_vmap_area() function, use a simple if()
instead of using BUG_ON() halting the machine.
Link: https://lkml.kernel.org/r/20230201115142.GA7772@min-iamroot Co-developed-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com> Signed-off-by: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com> Co-developed-by: Jeungwoo Yoo <casionwoo@gmail.com> Signed-off-by: Jeungwoo Yoo <casionwoo@gmail.com> Co-developed-by: Sangyun Kim <sangyun.kim@snu.ac.kr> Signed-off-by: Sangyun Kim <sangyun.kim@snu.ac.kr> Signed-off-by: Hyunmin Lee <hn.min.lee@gmail.com> Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu> Reviewed-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org> Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Longlong Xia [Tue, 31 Jan 2023 07:10:35 +0000 (07:10 +0000)]
mm/swapfile: remove pr_debug in get_swap_pages()
It's known that get_swap_pages() may fail to find available space under
some extreme case, but pr_debug() provides useless information. Let's
remove it.
Link: https://lkml.kernel.org/r/20230131071035.1085968-1-xialonglong1@huawei.com Signed-off-by: Longlong Xia <xialonglong1@huawei.com> Reviewed-by: "Huang, Ying" <ying.huang@intel.com> Cc: Chen Wandun <chenwandun@huawei.com> Cc: Kefeng Wang <wangkefeng.wang@huawei.com> Cc: Nanyong Sun <sunnanyong@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>