Commit d0637c505f ("arm64: enable THP_SWAP for arm64") brings up
THP_SWAP on ARM64, but it doesn't enable THP_SWP on hardware with MTE as
the MTE code works with the assumption tags save/restore is always
handling a folio with only one page.
The limitation should be removed as more and more ARM64 SoCs have this
feature. Co-existence of MTE and THP_SWAP becomes more and more
important.
This patch makes MTE tags saving support large folios, then we don't need
to split large folios into base pages for swapping out on ARM64 SoCs with
MTE any more.
arch_prepare_to_swap() should take folio rather than page as parameter
because we support THP swap-out as a whole. It saves tags for all pages
in a large folio.
As now we are restoring tags based-on folio, in arch_swap_restore(), we
may increase some extra loops and early-exitings while refaulting a large
folio which is still in swapcache in do_swap_page(). In case a large
folio has nr pages, do_swap_page() will only set the PTE of the particular
page which is causing the page fault. Thus do_swap_page() runs nr times,
and each time, arch_swap_restore() will loop nr times for those subpages
in the folio. So right now the algorithmic complexity becomes O(nr^2).
Once we support mapping large folios in do_swap_page(), extra loops and
early-exitings will decrease while not being completely removed as a large
folio might get partially tagged in corner cases such as, 1. a large
folio in swapcache can be partially unmapped, thus, MTE tags for the
unmapped pages will be invalidated; 2. users might use mprotect() to set
MTEs on a part of a large folio.
arch_thp_swp_supported() is dropped since ARM64 MTE was the only one who
needed it.
Link: https://lkml.kernel.org/r/20240322114136.61386-2-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Chris Li <chrisl@kernel.org>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Kemeng Shi <shikemeng@huaweicloud.com>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Peter Collingbourne <pcc@google.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: "Mike Rapoport (IBM)" <rppt@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.ibm.com>
Cc: Rick Edgecombe <rick.p.edgecombe@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
PAT handling won't do the right thing in COW mappings: the first PTE (or,
in fact, all PTEs) can be replaced during write faults to point at anon
folios. Reliably recovering the correct PFN and cachemode using
follow_phys() from PTEs will not work in COW mappings.
Using follow_phys(), we might just get the address+protection of the anon
folio (which is very wrong), or fail on swap/nonswap entries, failing
follow_phys() and triggering a WARN_ON_ONCE() in untrack_pfn() and
track_pfn_copy(), not properly calling free_pfn_range().
In free_pfn_range(), we either wouldn't call memtype_free() or would call
it with the wrong range, possibly leaking memory.
To fix that, let's update follow_phys() to refuse returning anon folios,
and fallback to using the stored PFN inside vma->vm_pgoff for COW mappings
if we run into that.
We will now properly handle untrack_pfn() with COW mappings, where we
don't need the cachemode. We'll have to fail fork()->track_pfn_copy() if
the first page was replaced by an anon folio, though: we'd have to store
the cachemode in the VMA to make this work, likely growing the VMA size.
For now, lets keep it simple and let track_pfn_copy() just fail in that
case: it would have failed in the past with swap/nonswap entries already,
and it would have done the wrong thing with anon folios.
Simple reproducer to trigger the WARN_ON_ONCE() in untrack_pfn():
<--- C reproducer --->
#include <stdio.h>
#include <sys/mman.h>
#include <unistd.h>
#include <liburing.h>
int main(void)
{
struct io_uring_params p = {};
int ring_fd;
size_t size;
char *map;
ring_fd = io_uring_setup(1, &p);
if (ring_fd < 0) {
perror("io_uring_setup");
return 1;
}
size = p.sq_off.array + p.sq_entries * sizeof(unsigned);
/* Map the submission queue ring MAP_PRIVATE */
map = mmap(0, size, PROT_READ | PROT_WRITE, MAP_PRIVATE,
ring_fd, IORING_OFF_SQ_RING);
if (map == MAP_FAILED) {
perror("mmap");
return 1;
}
/* We have at least one page. Let's COW it. */
*map = 0;
pause();
return 0;
}
<--- C reproducer --->
On a system with 16 GiB RAM and swap configured:
# ./iouring &
# memhog 16G
# killall iouring
[ 301.552930] ------------[ cut here ]------------
[ 301.553285] WARNING: CPU: 7 PID: 1402 at arch/x86/mm/pat/memtype.c:1060 untrack_pfn+0xf4/0x100
[ 301.553989] Modules linked in: binfmt_misc nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_g
[ 301.558232] CPU: 7 PID: 1402 Comm: iouring Not tainted 6.7.5-100.fc38.x86_64 #1
[ 301.558772] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebu4
[ 301.559569] RIP: 0010:untrack_pfn+0xf4/0x100
[ 301.559893] Code: 75 c4 eb cf 48 8b 43 10 8b a8 e8 00 00 00 3b 6b 28 74 b8 48 8b 7b 30 e8 ea 1a f7 000
[ 301.561189] RSP: 0018:ffffba2c0377fab8 EFLAGS: 00010282
[ 301.561590] RAX: 00000000ffffffea RBX: ffff9208c8ce9cc0 RCX: 000000010455e047
[ 301.562105] RDX: 07fffffff0eb1e0a RSI: 0000000000000000 RDI: ffff9208c391d200
[ 301.562628] RBP: 0000000000000000 R08: ffffba2c0377fab8 R09: 0000000000000000
[ 301.563145] R10: ffff9208d2292d50 R11: 0000000000000002 R12: 00007fea890e0000
[ 301.563669] R13: 0000000000000000 R14: ffffba2c0377fc08 R15: 0000000000000000
[ 301.564186] FS: 0000000000000000(0000) GS:ffff920c2fbc0000(0000) knlGS:0000000000000000
[ 301.564773] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 301.565197] CR2: 00007fea88ee8a20 CR3: 00000001033a8000 CR4: 0000000000750ef0
[ 301.565725] PKRU: 55555554
[ 301.565944] Call Trace:
[ 301.566148] <TASK>
[ 301.566325] ? untrack_pfn+0xf4/0x100
[ 301.566618] ? __warn+0x81/0x130
[ 301.566876] ? untrack_pfn+0xf4/0x100
[ 301.567163] ? report_bug+0x171/0x1a0
[ 301.567466] ? handle_bug+0x3c/0x80
[ 301.567743] ? exc_invalid_op+0x17/0x70
[ 301.568038] ? asm_exc_invalid_op+0x1a/0x20
[ 301.568363] ? untrack_pfn+0xf4/0x100
[ 301.568660] ? untrack_pfn+0x65/0x100
[ 301.568947] unmap_single_vma+0xa6/0xe0
[ 301.569247] unmap_vmas+0xb5/0x190
[ 301.569532] exit_mmap+0xec/0x340
[ 301.569801] __mmput+0x3e/0x130
[ 301.570051] do_exit+0x305/0xaf0
...
Link: https://lkml.kernel.org/r/20240403212131.929421-3-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reported-by: Wupeng Ma <mawupeng1@huawei.com>
Closes: https://lkml.kernel.org/r/20240227122814.3781907-1-mawupeng1@huawei.com
Fixes: b1a86e15dc ("x86, pat: remove the dependency on 'vm_pgoff' in track/untrack pfn vma routines")
Fixes: 5899329b19 ("x86: PAT: implement track/untrack of pfnmap regions for x86 - v3")
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Commit 0cf18e839f of large folio zap work broke uffd-wp. Now mm's uffd
unit test "wp-unpopulated" will trigger this WARN_ON_ONCE().
The WARN_ON_ONCE() asserts that an VMA cannot be registered with
userfaultfd-wp if it contains a !normal page, but it's actually possible.
One example is an anonymous vma, register with uffd-wp, read anything will
install a zero page. Then when zap on it, this should trigger.
What's more, removing that WARN_ON_ONCE may not be enough either, because
we should also not rely on "whether it's a normal page" to decide whether
pte marker is needed. For example, one can register wr-protect over some
DAX regions to track writes when UFFD_FEATURE_WP_ASYNC enabled, in which
case it can have page==NULL for a devmap but we may want to keep the
marker around.
Link: https://lkml.kernel.org/r/20240313213107.235067-1-peterx@redhat.com
Fixes: 0cf18e839f ("mm/memory: handle !page case in zap_present_pte() separately")
Signed-off-by: Peter Xu <peterx@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Muhammad Usama Anjum <usama.anjum@collabora.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
In a Copy-on-Write (CoW) scenario, the last subpage will reuse the entire
large folio, resulting in the waste of (nr_pages - 1) pages. This wasted
memory remains allocated until it is either unmapped or memory reclamation
occurs.
The following small program can serve as evidence of this behavior
main()
{
#define SIZE 1024 * 1024 * 1024UL
void *p = malloc(SIZE);
memset(p, 0x11, SIZE);
if (fork() == 0)
_exit(0);
memset(p, 0x12, SIZE);
printf("done\n");
while(1);
}
For example, using a 1024KiB mTHP by:
echo always > /sys/kernel/mm/transparent_hugepage/hugepages-1024kB/enabled
(1) w/o the patch, it takes 2GiB,
Before running the test program,
/ # free -m
total used free shared buff/cache available
Mem: 5754 84 5692 0 17 5669
Swap: 0 0 0
/ # /a.out &
/ # done
After running the test program,
/ # free -m
total used free shared buff/cache available
Mem: 5754 2149 3627 0 19 3605
Swap: 0 0 0
(2) w/ the patch, it takes 1GiB only,
Before running the test program,
/ # free -m
total used free shared buff/cache available
Mem: 5754 89 5687 0 17 5664
Swap: 0 0 0
/ # /a.out &
/ # done
After running the test program,
/ # free -m
total used free shared buff/cache available
Mem: 5754 1122 4655 0 17 4632
Swap: 0 0 0
This patch migrates the last subpage to a small folio and immediately
returns the large folio to the system. It benefits both memory availability
and anti-fragmentation.
Link: https://lkml.kernel.org/r/20240308092721.144735-1-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Lance Yang <ioworker0@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The rounddown_pow_of_two(0) is undefined, so val = 0 is not allowed in the
fault_around_bytes_set(), and leads to shift-out-of-bounds,
UBSAN: shift-out-of-bounds in include/linux/log2.h:67:13
shift exponent 4294967295 is too large for 64-bit type 'long unsigned int'
CPU: 7 PID: 107 Comm: sh Not tainted 6.8.0-rc6-next-20240301 #294
Hardware name: QEMU QEMU Virtual Machine, BIOS 0.0.0 02/06/2015
Call trace:
dump_backtrace+0x94/0xec
show_stack+0x18/0x24
dump_stack_lvl+0x78/0x90
dump_stack+0x18/0x24
ubsan_epilogue+0x10/0x44
__ubsan_handle_shift_out_of_bounds+0x98/0x134
fault_around_bytes_set+0xa4/0xb0
simple_attr_write_xsigned.isra.0+0xe4/0x1ac
simple_attr_write+0x18/0x24
debugfs_attr_write+0x4c/0x98
vfs_write+0xd0/0x4b0
ksys_write+0x6c/0xfc
__arm64_sys_write+0x1c/0x28
invoke_syscall+0x44/0x104
el0_svc_common.constprop.0+0x40/0xe0
do_el0_svc+0x1c/0x28
el0_svc+0x34/0xdc
el0t_64_sync_handler+0xc0/0xc4
el0t_64_sync+0x190/0x194
---[ end trace ]---
Fix it by setting the minimum val to PAGE_SIZE.
Link: https://lkml.kernel.org/r/20240302064312.2358924-1-wangkefeng.wang@huawei.com
Fixes: 53d36a56d8 ("mm: prefer fault_around_pages to fault_around_bytes")
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reported-by: Yue Sun <samsun1006219@gmail.com>
Closes: https://lore.kernel.org/all/CAEkJfYPim6DQqW1GqCiHLdh2-eweqk1fGyXqs3JM+8e1qGge8w@mail.gmail.com/
Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
do_numa_page() is reading from the same page table entry, twice, while
holding the page table lock: once while checking that the pte hasn't
changed, and again in order to modify the pte.
Instead, just read the pte once, and save it in the same old_pte variable
that already exists. This has no effect on behavior, other than to
provide a tiny potential improvement to performance, by avoiding the
redundant memory read (which the compiler cannot elide, due to
READ_ONCE()).
Also improve the associated comments nearby.
Link: https://lkml.kernel.org/r/20240228034151.459370-1-jhubbard@nvidia.com
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "Handle hugetlb faults under the VMA lock", v2.
It is generally safe to handle hugetlb faults under the VMA lock. The
only time this is unsafe is when no anon_vma has been allocated to this
vma yet, so we can use vmf_anon_prepare() instead of anon_vma_prepare() to
bailout if necessary. This should only happen for the first hugetlb page
in the vma.
Additionally, this patchset begins to use struct vm_fault within
hugetlb_fault(). This works towards cleaning up hugetlb code, and should
significantly reduce the number of arguments passed to functions.
The last patch in this series may cause ltp hugemmap10 to "fail". This is
because vmf_anon_prepare() may bailout with no anon_vma under the VMA lock
after allocating a folio for the hugepage. In free_huge_folio(), this
folio is completely freed on bailout iff there is a surplus of hugetlb
pages. This will remove a folio off the freelist and decrement the number
of hugepages while ltp expects these counters to remain unchanged on
failure. The rest of the ltp testcases pass.
This patch (of 2):
In order to handle hugetlb faults under the VMA lock, hugetlb can use
vmf_anon_prepare() to ensure we can safely prepare an anon_vma. Change it
to be a non-static function so it can be used within hugetlb as well.
Link: https://lkml.kernel.org/r/20240221234732.187629-6-vishal.moola@gmail.com
Link: https://lkml.kernel.org/r/20240221234732.187629-2-vishal.moola@gmail.com
Signed-off-by: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Similar to how we optimized fork(), let's implement PTE batching when
consecutive (present) PTEs map consecutive pages of the same large folio.
Most infrastructure we need for batching (mmu gather, rmap) is already
there. We only have to add get_and_clear_full_ptes() and
clear_full_ptes(). Similarly, extend zap_install_uffd_wp_if_needed() to
process a PTE range.
We won't bother sanity-checking the mapcount of all subpages, but only
check the mapcount of the first subpage we process. If there is a real
problem hiding somewhere, we can trigger it simply by using small folios,
or when we zap single pages of a large folio. Ideally, we had that check
in rmap code (including for delayed rmap), but then we cannot print the
PTE. Let's keep it simple for now. If we ever have a cheap
folio_mapcount(), we might just want to check for underflows there.
To keep small folios as fast as possible force inlining of a specialized
variant using __always_inline with nr=1.
Link: https://lkml.kernel.org/r/20240214204435.167852-11-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Naveen N. Rao" <naveen.n.rao@linux.ibm.com>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm/memory: optimize unmap/zap with PTE-mapped THP", v3.
This series is based on [1]. Similar to what we did with fork(), let's
implement PTE batching during unmap/zap when processing PTE-mapped THPs.
We collect consecutive PTEs that map consecutive pages of the same large
folio, making sure that the other PTE bits are compatible, and (a) adjust
the refcount only once per batch, (b) call rmap handling functions only
once per batch, (c) perform batch PTE setting/updates and (d) perform TLB
entry removal once per batch.
Ryan was previously working on this in the context of cont-pte for arm64,
int latest iteration [2] with a focus on arm6 with cont-pte only. This
series implements the optimization for all architectures, independent of
such PTE bits, teaches MMU gather/TLB code to be fully aware of such
large-folio-pages batches as well, and amkes use of our new rmap batching
function when removing the rmap.
To achieve that, we have to enlighten MMU gather / page freeing code
(i.e., everything that consumes encoded_page) to process unmapping of
consecutive pages that all belong to the same large folio. I'm being very
careful to not degrade order-0 performance, and it looks like I managed to
achieve that.
While this series should -- similar to [1] -- be beneficial for adding
cont-pte support on arm64[2], it's one of the requirements for maintaining
a total mapcount[3] for large folios with minimal added overhead and
further changes[4] that build up on top of the total mapcount.
Independent of all that, this series results in a speedup during munmap()
and similar unmapping (process teardown, MADV_DONTNEED on larger ranges)
with PTE-mapped THP, which is the default with THPs that are smaller than
a PMD (for example, 16KiB to 1024KiB mTHPs for anonymous memory[5]).
On an Intel Xeon Silver 4210R CPU, munmap'ing a 1GiB VMA backed by
PTE-mapped folios of the same size (stddev < 1%) results in the following
runtimes for munmap() in seconds (shorter is better):
Folio Size | mm-unstable | New | Change
---------------------------------------------
4KiB | 0.058110 | 0.057715 | - 1%
16KiB | 0.044198 | 0.035469 | -20%
32KiB | 0.034216 | 0.023522 | -31%
64KiB | 0.029207 | 0.018434 | -37%
128KiB | 0.026579 | 0.014026 | -47%
256KiB | 0.025130 | 0.011756 | -53%
512KiB | 0.024292 | 0.010703 | -56%
1024KiB | 0.023812 | 0.010294 | -57%
2048KiB | 0.023785 | 0.009910 | -58%
[1] https://lkml.kernel.org/r/20240129124649.189745-1-david@redhat.com
[2] https://lkml.kernel.org/r/20231218105100.172635-1-ryan.roberts@arm.com
[3] https://lkml.kernel.org/r/20230809083256.699513-1-david@redhat.com
[4] https://lkml.kernel.org/r/20231124132626.235350-1-david@redhat.com
[5] https://lkml.kernel.org/r/20231207161211.2374093-1-ryan.roberts@arm.com
This patch (of 10):
Let's prepare for further changes by factoring out processing of present
PTEs.
Link: https://lkml.kernel.org/r/20240214204435.167852-1-david@redhat.com
Link: https://lkml.kernel.org/r/20240214204435.167852-2-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Naveen N. Rao" <naveen.n.rao@linux.ibm.com>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The GFP flags from vma_thp_gfp_mask() according to user configuration only
used for large folio allocation but not for memory cgroup charge, and
GFP_KERNEL is used for both order-0 and large order folio when memory
cgroup charge at present. However, mem_cgroup_charge() uses the GFP flags
in a fairly sophisticated way. In addition to checking
gfpflags_allow_blocking(), it pays attention to __GFP_NORETRY and
__GFP_RETRY_MAYFAIL to ensure that processes within this memcg do not
exceed their quotas.
So we'd better to move mem_cgroup_charge() into alloc_anon_folio(),
1) it will make us to allocate as much as possible large order folio,
because we could try the next order if mem_cgroup_charge() fails,
although the memcg's memory usage is close to its limits.
2) using same GFP flags for allocation and charge is to be consistent
with PMD THP firstly, in addition, according to GFP flag returned from
vma_thp_gfp_mask(), GFP_TRANSHUGE_LIGHT could make us skip direct
reclaim, _GFP_NORETRY will make us skip mem_cgroup_oom() and won't
trigger memory cgroup oom from large order(order <= COSTLY_ORDER) folio
charging.
Link: https://lkml.kernel.org/r/20240122011612.501029-1-wangkefeng.wang@huawei.com
Link: https://lkml.kernel.org/r/20240117103954.2756050-1-wangkefeng.wang@huawei.com
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
When skipping swapcache for SWP_SYNCHRONOUS_IO, if two or more threads
swapin the same entry at the same time, they get different pages (A, B).
Before one thread (T0) finishes the swapin and installs page (A) to the
PTE, another thread (T1) could finish swapin of page (B), swap_free the
entry, then swap out the possibly modified page reusing the same entry.
It breaks the pte_same check in (T0) because PTE value is unchanged,
causing ABA problem. Thread (T0) will install a stalled page (A) into the
PTE and cause data corruption.
One possible callstack is like this:
CPU0 CPU1
---- ----
do_swap_page() do_swap_page() with same entry
<direct swapin path> <direct swapin path>
<alloc page A> <alloc page B>
swap_read_folio() <- read to page A swap_read_folio() <- read to page B
<slow on later locks or interrupt> <finished swapin first>
... set_pte_at()
swap_free() <- entry is free
<write to page B, now page A stalled>
<swap out page B to same swap entry>
pte_same() <- Check pass, PTE seems
unchanged, but page A
is stalled!
swap_free() <- page B content lost!
set_pte_at() <- staled page A installed!
And besides, for ZRAM, swap_free() allows the swap device to discard the
entry content, so even if page (B) is not modified, if swap_read_folio()
on CPU0 happens later than swap_free() on CPU1, it may also cause data
loss.
To fix this, reuse swapcache_prepare which will pin the swap entry using
the cache flag, and allow only one thread to swap it in, also prevent any
parallel code from putting the entry in the cache. Release the pin after
PT unlocked.
Racers just loop and wait since it's a rare and very short event. A
schedule_timeout_uninterruptible(1) call is added to avoid repeated page
faults wasting too much CPU, causing livelock or adding too much noise to
perf statistics. A similar livelock issue was described in commit
029c4628b2 ("mm: swap: get rid of livelock in swapin readahead")
Reproducer:
This race issue can be triggered easily using a well constructed
reproducer and patched brd (with a delay in read path) [1]:
With latest 6.8 mainline, race caused data loss can be observed easily:
$ gcc -g -lpthread test-thread-swap-race.c && ./a.out
Polulating 32MB of memory region...
Keep swapping out...
Starting round 0...
Spawning 65536 workers...
32746 workers spawned, wait for done...
Round 0: Error on 0x5aa00, expected 32746, got 32743, 3 data loss!
Round 0: Error on 0x395200, expected 32746, got 32743, 3 data loss!
Round 0: Error on 0x3fd000, expected 32746, got 32737, 9 data loss!
Round 0 Failed, 15 data loss!
This reproducer spawns multiple threads sharing the same memory region
using a small swap device. Every two threads updates mapped pages one by
one in opposite direction trying to create a race, with one dedicated
thread keep swapping out the data out using madvise.
The reproducer created a reproduce rate of about once every 5 minutes, so
the race should be totally possible in production.
After this patch, I ran the reproducer for over a few hundred rounds and
no data loss observed.
Performance overhead is minimal, microbenchmark swapin 10G from 32G
zram:
Before: 10934698 us
After: 11157121 us
Cached: 13155355 us (Dropping SWP_SYNCHRONOUS_IO flag)
[kasong@tencent.com: v4]
Link: https://lkml.kernel.org/r/20240219082040.7495-1-ryncsn@gmail.com
Link: https://lkml.kernel.org/r/20240206182559.32264-1-ryncsn@gmail.com
Fixes: 0bcac06f27 ("mm, swap: skip swapcache for swapin of synchronous device")
Reported-by: "Huang, Ying" <ying.huang@intel.com>
Closes: https://lore.kernel.org/lkml/87bk92gqpx.fsf_-_@yhuang6-desk2.ccr.corp.intel.com/
Link: https://github.com/ryncsn/emm-test-project/tree/master/swap-stress-race [1]
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Acked-by: Yu Zhao <yuzhao@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Chris Li <chrisl@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Yosry Ahmed <yosryahmed@google.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Barry Song <21cnbao@gmail.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Pull MM updates from Andrew Morton:
"Many singleton patches against the MM code. The patch series which are
included in this merge do the following:
- Peng Zhang has done some mapletree maintainance work in the series
'maple_tree: add mt_free_one() and mt_attr() helpers'
'Some cleanups of maple tree'
- In the series 'mm: use memmap_on_memory semantics for dax/kmem'
Vishal Verma has altered the interworking between memory-hotplug
and dax/kmem so that newly added 'device memory' can more easily
have its memmap placed within that newly added memory.
- Matthew Wilcox continues folio-related work (including a few fixes)
in the patch series
'Add folio_zero_tail() and folio_fill_tail()'
'Make folio_start_writeback return void'
'Fix fault handler's handling of poisoned tail pages'
'Convert aops->error_remove_page to ->error_remove_folio'
'Finish two folio conversions'
'More swap folio conversions'
- Kefeng Wang has also contributed folio-related work in the series
'mm: cleanup and use more folio in page fault'
- Jim Cromie has improved the kmemleak reporting output in the series
'tweak kmemleak report format'.
- In the series 'stackdepot: allow evicting stack traces' Andrey
Konovalov to permits clients (in this case KASAN) to cause eviction
of no longer needed stack traces.
- Charan Teja Kalla has fixed some accounting issues in the page
allocator's atomic reserve calculations in the series 'mm:
page_alloc: fixes for high atomic reserve caluculations'.
- Dmitry Rokosov has added to the samples/ dorectory some sample code
for a userspace memcg event listener application. See the series
'samples: introduce cgroup events listeners'.
- Some mapletree maintanance work from Liam Howlett in the series
'maple_tree: iterator state changes'.
- Nhat Pham has improved zswap's approach to writeback in the series
'workload-specific and memory pressure-driven zswap writeback'.
- DAMON/DAMOS feature and maintenance work from SeongJae Park in the
series
'mm/damon: let users feed and tame/auto-tune DAMOS'
'selftests/damon: add Python-written DAMON functionality tests'
'mm/damon: misc updates for 6.8'
- Yosry Ahmed has improved memcg's stats flushing in the series 'mm:
memcg: subtree stats flushing and thresholds'.
- In the series 'Multi-size THP for anonymous memory' Ryan Roberts
has added a runtime opt-in feature to transparent hugepages which
improves performance by allocating larger chunks of memory during
anonymous page faults.
- Matthew Wilcox has also contributed some cleanup and maintenance
work against eh buffer_head code int he series 'More buffer_head
cleanups'.
- Suren Baghdasaryan has done work on Andrea Arcangeli's series
'userfaultfd move option'. UFFDIO_MOVE permits userspace heap
compaction algorithms to move userspace's pages around rather than
UFFDIO_COPY'a alloc/copy/free.
- Stefan Roesch has developed a 'KSM Advisor', in the series 'mm/ksm:
Add ksm advisor'. This is a governor which tunes KSM's scanning
aggressiveness in response to userspace's current needs.
- Chengming Zhou has optimized zswap's temporary working memory use
in the series 'mm/zswap: dstmem reuse optimizations and cleanups'.
- Matthew Wilcox has performed some maintenance work on the writeback
code, both code and within filesystems. The series is 'Clean up the
writeback paths'.
- Andrey Konovalov has optimized KASAN's handling of alloc and free
stack traces for secondary-level allocators, in the series 'kasan:
save mempool stack traces'.
- Andrey also performed some KASAN maintenance work in the series
'kasan: assorted clean-ups'.
- David Hildenbrand has gone to town on the rmap code. Cleanups, more
pte batching, folio conversions and more. See the series 'mm/rmap:
interface overhaul'.
- Kinsey Ho has contributed some maintenance work on the MGLRU code
in the series 'mm/mglru: Kconfig cleanup'.
- Matthew Wilcox has contributed lruvec page accounting code cleanups
in the series 'Remove some lruvec page accounting functions'"
* tag 'mm-stable-2024-01-08-15-31' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (361 commits)
mm, treewide: rename MAX_ORDER to MAX_PAGE_ORDER
mm, treewide: introduce NR_PAGE_ORDERS
selftests/mm: add separate UFFDIO_MOVE test for PMD splitting
selftests/mm: skip test if application doesn't has root privileges
selftests/mm: conform test to TAP format output
selftests: mm: hugepage-mmap: conform to TAP format output
selftests/mm: gup_test: conform test to TAP format output
mm/selftests: hugepage-mremap: conform test to TAP format output
mm/vmstat: move pgdemote_* out of CONFIG_NUMA_BALANCING
mm: zsmalloc: return -ENOSPC rather than -EINVAL in zs_malloc while size is too large
mm/memcontrol: remove __mod_lruvec_page_state()
mm/khugepaged: use a folio more in collapse_file()
slub: use a folio in __kmalloc_large_node
slub: use folio APIs in free_large_kmalloc()
slub: use alloc_pages_node() in alloc_slab_page()
mm: remove inc/dec lruvec page state functions
mm: ratelimit stat flush from workingset shrinker
kasan: stop leaking stack trace handles
mm/mglru: remove CONFIG_TRANSPARENT_HUGEPAGE
mm/mglru: add dummy pmd_dirty()
...
Let's get rid of the compound parameter and instead define explicitly
which mappings we're adding. That is more future proof, easier to read
and harder to mess up.
Use an enum to express the granularity internally. Make the compiler
always special-case on the granularity by using __always_inline. Replace
the "compound" check by a switch-case that will be removed by the compiler
completely.
Add plenty of sanity checks with CONFIG_DEBUG_VM. Replace the
folio_test_pmd_mappable() check by a config check in the caller and sanity
checks. Convert the single user of folio_add_file_rmap_range().
While at it, consistently use "int" instead of "unisgned int" in rmap code
when dealing with mapcounts and the number of pages.
This function design can later easily be extended to PUDs and to batch
PMDs. Note that for now we don't support anything bigger than PMD-sized
folios (as we cleanly separated hugetlb handling). Sanity checks will
catch if that ever changes.
Next up is removing page_remove_rmap() along with its "compound" parameter
and smilarly converting all other rmap functions.
Link: https://lkml.kernel.org/r/20231220224504.646757-8-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The bug happens when highest bit of holebegin is 1, suppose holebegin is
0x8000000111111000, after shift, hba would be 0xfff8000000111111, then
vma_interval_tree_foreach would look it up fail or leads to the wrong
result.
error call seq e.g.:
- mmap(..., offset=0x8000000111111000)
|- syscall(mmap, ... unsigned long, off):
|- ksys_mmap_pgoff( ... , off >> PAGE_SHIFT);
here pgoff is correctly shifted to 0x8000000111111,
but pass 0x8000000111111000 as holebegin to unmap
would then cause terrible result, as shown below:
- unmap_mapping_range(..., loff_t const holebegin)
|- pgoff_t hba = holebegin >> PAGE_SHIFT;
/* hba = 0xfff8000000111111 unexpectedly */
The issue happens in Heterogeneous computing, where the device(e.g.
gpu) and host share the same virtual address space.
A simple workflow pattern which hit the issue is:
/* host */
1. userspace first mmap a file backed VA range with specified offset.
e.g. (offset=0x800..., mmap return: va_a)
2. write some data to the corresponding sys page
e.g. (va_a = 0xAABB)
/* device */
3. gpu workload touches VA, triggers gpu fault and notify the host.
/* host */
4. reviced gpu fault notification, then it will:
4.1 unmap host pages and also takes care of cpu tlb
(use unmap_mapping_range with offset=0x800...)
4.2 migrate sys page to device
4.3 setup device page table and resolve device fault.
/* device */
5. gpu workload continued, it accessed va_a and got 0xAABB.
6. gpu workload continued, it wrote 0xBBCC to va_a.
/* host */
7. userspace access va_a, as expected, it will:
7.1 trigger cpu vm fault.
7.2 driver handling fault to migrate gpu local page to host.
8. userspace then could correctly get 0xBBCC from va_a
9. done
But in step 4.1, if we hit the bug this patch mentioned, then userspace
would never trigger cpu fault, and still get the old value: 0xAABB.
Making holebegin unsigned first fixes the bug.
Link: https://lkml.kernel.org/r/20231220052839.26970-1-jiajun.xie.sh@gmail.com
Signed-off-by: Jiajun Xie <jiajun.xie.sh@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
kmap() has been deprecated in favor of kmap_local_page().
Therefore, replace kmap() with kmap_local_page() in mm/memory.c.
There are two main problems with kmap(): (1) It comes with an overhead as
the mapping space is restricted and protected by a global lock for
synchronization and (2) it also requires global TLB invalidation when the
kmap's pool wraps and it might block when the mapping space is fully
utilized until a slot becomes available.
With kmap_local_page() the mappings are per thread, CPU local, can take
page-faults, and can be called from any context (including interrupts).
It is faster than kmap() in kernels with HIGHMEM enabled. The tasks can
be preempted and, when they are scheduled to run again, the kernel virtual
addresses are restored and still valid.
Obviously, thread locality implies that the kernel virtual addresses
returned by kmap_local_page() are only valid in the context of the callers
(i.e., they cannot be handed to other threads).
The use of kmap_local_page() in mm/memory.c does not break the
above-mentioned assumption, so it is allowed and preferred.
Link: https://lkml.kernel.org/r/20231215084417.2002370-1-fabio.maria.de.francesco@linux.intel.com
Link: https://lkml.kernel.org/r/20231214081039.1919328-1-fabio.maria.de.francesco@linux.intel.com
Signed-off-by: Fabio M. De Francesco <fabio.maria.de.francesco@linux.intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Introduce the logic to allow THP to be configured (through the new sysfs
interface we just added) to allocate large folios to back anonymous
memory, which are larger than the base page size but smaller than
PMD-size. We call this new THP extension "multi-size THP" (mTHP).
mTHP continues to be PTE-mapped, but in many cases can still provide
similar benefits to traditional PMD-sized THP: Page faults are
significantly reduced (by a factor of e.g. 4, 8, 16, etc. depending on
the configured order), but latency spikes are much less prominent because
the size of each page isn't as huge as the PMD-sized variant and there is
less memory to clear in each page fault. The number of per-page
operations (e.g. ref counting, rmap management, lru list management) are
also significantly reduced since those ops now become per-folio.
Some architectures also employ TLB compression mechanisms to squeeze more
entries in when a set of PTEs are virtually and physically contiguous and
approporiately aligned. In this case, TLB misses will occur less often.
The new behaviour is disabled by default, but can be enabled at runtime by
writing to /sys/kernel/mm/transparent_hugepage/hugepage-XXkb/enabled (see
documentation in previous commit). The long term aim is to change the
default to include suitable lower orders, but there are some risks around
internal fragmentation that need to be better understood first.
[ryan.roberts@arm.com: resolve some multi-size THP review nits]
Link: https://lkml.kernel.org/r/20231214160251.3574571-1-ryan.roberts@arm.com
Link: https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm.com
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Tested-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Tested-by: John Hubbard <jhubbard@nvidia.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Itaru Kitayama <itaru.kitayama@gmail.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Yin Fengwei <fengwei.yin@intel.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
In preparation for adding support for anonymous multi-size THP, introduce
new sysfs structure that will be used to control the new behaviours. A
new directory is added under transparent_hugepage for each supported THP
size, and contains an `enabled` file, which can be set to "inherit" (to
inherit the global setting), "always", "madvise" or "never". For now, the
kernel still only supports PMD-sized anonymous THP, so only 1 directory is
populated.
The first half of the change converts transhuge_vma_suitable() and
hugepage_vma_check() so that they take a bitfield of orders for which the
user wants to determine support, and the functions filter out all the
orders that can't be supported, given the current sysfs configuration and
the VMA dimensions. The resulting functions are renamed to
thp_vma_suitable_orders() and thp_vma_allowable_orders() respectively.
Convenience functions that take a single, unencoded order and return a
boolean are also defined as thp_vma_suitable_order() and
thp_vma_allowable_order().
The second half of the change implements the new sysfs interface. It has
been done so that each supported THP size has a `struct thpsize`, which
describes the relevant metadata and is itself a kobject. This is pretty
minimal for now, but should make it easy to add new per-thpsize files to
the interface if needed in future (e.g. per-size defrag). Rather than
keep the `enabled` state directly in the struct thpsize, I've elected to
directly encode it into huge_anon_orders_[always|madvise|inherit]
bitfields since this reduces the amount of work required in
thp_vma_allowable_orders() which is called for every page fault.
See Documentation/admin-guide/mm/transhuge.rst, as modified by this
commit, for details of how the new sysfs interface works.
[ryan.roberts@arm.com: fix build warning when CONFIG_SYSFS is disabled]
Link: https://lkml.kernel.org/r/20231211125320.3997543-1-ryan.roberts@arm.com
Link: https://lkml.kernel.org/r/20231207161211.2374093-4-ryan.roberts@arm.com
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Barry Song <v-songbaohua@oppo.com>
Tested-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Tested-by: John Hubbard <jhubbard@nvidia.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Itaru Kitayama <itaru.kitayama@gmail.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Yin Fengwei <fengwei.yin@intel.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
