Patch series "mm: Support huge pfnmaps", v2.
Overview
========
This series implements huge pfnmaps support for mm in general. Huge
pfnmap allows e.g. VM_PFNMAP vmas to map in either PMD or PUD levels,
similar to what we do with dax / thp / hugetlb so far to benefit from TLB
hits. Now we extend that idea to PFN mappings, e.g. PCI MMIO bars where
it can grow as large as 8GB or even bigger.
Currently, only x86_64 (1G+2M) and arm64 (2M) are supported. The last
patch (from Alex Williamson) will be the first user of huge pfnmap, so as
to enable vfio-pci driver to fault in huge pfn mappings.
Implementation
==============
In reality, it's relatively simple to add such support comparing to many
other types of mappings, because of PFNMAP's specialties when there's no
vmemmap backing it, so that most of the kernel routines on huge mappings
should simply already fail for them, like GUPs or old-school follow_page()
(which is recently rewritten to be folio_walk* APIs by David).
One trick here is that we're still unmature on PUDs in generic paths here
and there, as DAX is so far the only user. This patchset will add the 2nd
user of it. Hugetlb can be a 3rd user if the hugetlb unification work can
go on smoothly, but to be discussed later.
The other trick is how to allow gup-fast working for such huge mappings
even if there's no direct sign of knowing whether it's a normal page or
MMIO mapping. This series chose to keep the pte_special solution, so that
it reuses similar idea on setting a special bit to pfnmap PMDs/PUDs so
that gup-fast will be able to identify them and fail properly.
Along the way, we'll also notice that the major pgtable pfn walker, aka,
follow_pte(), will need to retire soon due to the fact that it only works
with ptes. A new set of simple API is introduced (follow_pfnmap* API) to
be able to do whatever follow_pte() can already do, plus that it can also
process huge pfnmaps now. Half of this series is about that and
converting all existing pfnmap walkers to use the new API properly.
Hopefully the new API also looks better to avoid exposing e.g. pgtable
lock details into the callers, so that it can be used in an even more
straightforward way.
Here, three more options will be introduced and involved in huge pfnmap:
- ARCH_SUPPORTS_HUGE_PFNMAP
Arch developers will need to select this option when huge pfnmap is
supported in arch's Kconfig. After this patchset applied, both x86_64
and arm64 will start to enable it by default.
- ARCH_SUPPORTS_PMD_PFNMAP / ARCH_SUPPORTS_PUD_PFNMAP
These options are for driver developers to identify whether current
arch / config supports huge pfnmaps, making decision on whether it can
use the huge pfnmap APIs to inject them. One can refer to the last
vfio-pci patch from Alex on the use of them properly in a device
driver.
So after the whole set applied, and if one would enable some dynamic debug
lines in vfio-pci core files, we should observe things like:
vfio-pci 0000:00:06.0: vfio_pci_mmap_huge_fault(,order = 9) BAR 0 page offset 0x0: 0x100
vfio-pci 0000:00:06.0: vfio_pci_mmap_huge_fault(,order = 9) BAR 0 page offset 0x200: 0x100
vfio-pci 0000:00:06.0: vfio_pci_mmap_huge_fault(,order = 9) BAR 0 page offset 0x400: 0x100
In this specific case, it says that vfio-pci faults in PMDs properly for a
few BAR0 offsets.
Patch Layout
============
Patch 1: Introduce the new options mentioned above for huge PFNMAPs
Patch 2: A tiny cleanup
Patch 3-8: Preparation patches for huge pfnmap (include introduce
special bit for pmd/pud)
Patch 9-16: Introduce follow_pfnmap*() API, use it everywhere, and
then drop follow_pte() API
Patch 17: Add huge pfnmap support for x86_64
Patch 18: Add huge pfnmap support for arm64
Patch 19: Add vfio-pci support for all kinds of huge pfnmaps (Alex)
TODO
====
More architectures / More page sizes
------------------------------------
Currently only x86_64 (2M+1G) and arm64 (2M) are supported. There seems
to have plan to support arm64 1G later on top of this series [2].
Any arch will need to first support THP / THP_1G, then provide a special
bit in pmds/puds to support huge pfnmaps.
remap_pfn_range() support
-------------------------
Currently, remap_pfn_range() still only maps PTEs. With the new option,
remap_pfn_range() can logically start to inject either PMDs or PUDs when
the alignment requirements match on the VAs.
When the support is there, it should be able to silently benefit all
drivers that is using remap_pfn_range() in its mmap() handler on better
TLB hit rate and overall faster MMIO accesses similar to processor on
hugepages.
More driver support
-------------------
VFIO is so far the only consumer for the huge pfnmaps after this series
applied. Besides above remap_pfn_range() generic optimization, device
driver can also try to optimize its mmap() on a better VA alignment for
either PMD/PUD sizes. This may, iiuc, normally require userspace changes,
as the driver doesn't normally decide the VA to map a bar. But I don't
think I know all the drivers to know the full picture.
Credits all go to Alex on help testing the GPU/NIC use cases above.
[0] https://lore.kernel.org/r/73ad9540-3fb8-4154-9a4f-30a0a2b03d41@lucifer.local
[1] https://lore.kernel.org/r/20240807194812.819412-1-peterx@redhat.com
[2] https://lore.kernel.org/r/498e0731-81a4-4f75-95b4-a8ad0bcc7665@huawei.com
This patch (of 19):
This patch introduces the option to introduce special pte bit into
pmd/puds. Archs can start to define pmd_special / pud_special when
supported by selecting the new option. Per-arch support will be added
later.
Before that, create fallbacks for these helpers so that they are always
available.
Link: https://lkml.kernel.org/r/20240826204353.2228736-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20240826204353.2228736-2-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Gavin Shan <gshan@redhat.com>
Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason Gunthorpe <jgg@nvidia.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Niklas Schnelle <schnelle@linux.ibm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Add pgalloc_tag_copy() to transfer the codetag from the old folio to the
new one during migration. This makes original allocation sites persist
cross migration rather than lump into the get_new_folio callbacks passed
into migrate_pages(), e.g., compaction_alloc():
# echo 1 >/proc/sys/vm/compact_memory
# grep compaction_alloc /proc/allocinfo
Before this patch:
132968448 32463 mm/compaction.c:1880 func:compaction_alloc
After this patch:
0 0 mm/compaction.c:1880 func:compaction_alloc
Link: https://lkml.kernel.org/r/20240906042108.1150526-3-yuzhao@google.com
Fixes: dcfe378c81 ("lib: introduce support for page allocation tagging")
Signed-off-by: Yu Zhao <yuzhao@google.com>
Acked-by: Suren Baghdasaryan <surenb@google.com>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The current assumption is that a large folio can only be split into
order-0 folios. That is not the case for hugeTLB demotion, nor for THP
split: see commit c010d47f10 ("mm: thp: split huge page to any lower
order pages").
When a large folio is split into ones of a lower non-zero order, only the
new head pages should be tagged. Tagging tail pages can cause imbalanced
"calls" counters, since only head pages are untagged by pgalloc_tag_sub()
and the "calls" counts on tail pages are leaked, e.g.,
# echo 2048kB >/sys/kernel/mm/hugepages/hugepages-1048576kB/demote_size
# echo 700 >/sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages
# time echo 700 >/sys/kernel/mm/hugepages/hugepages-1048576kB/demote
# echo 0 >/sys/kernel/mm/hugepages/hugepages-2048kB/nr_hugepages
# grep alloc_gigantic_folio /proc/allocinfo
Before this patch:
0 549427200 mm/hugetlb.c:1549 func:alloc_gigantic_folio
real 0m2.057s
user 0m0.000s
sys 0m2.051s
After this patch:
0 0 mm/hugetlb.c:1549 func:alloc_gigantic_folio
real 0m1.711s
user 0m0.000s
sys 0m1.704s
Not tagging tail pages also improves the splitting time, e.g., by about
15% when demoting 1GB hugeTLB folios to 2MB ones, as shown above.
Link: https://lkml.kernel.org/r/20240906042108.1150526-2-yuzhao@google.com
Fixes: be25d1d4e8 ("mm: create new codetag references during page splitting")
Signed-off-by: Yu Zhao <yuzhao@google.com>
Acked-by: Suren Baghdasaryan <surenb@google.com>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
random.h is not needed since commit 6c542ab757 ("mm/demotion: build
demotion targets based on explicit memory tiers"), all functions moved
into memory-tiers.
nsproxy.h is not needed since commit 228ebcbe63 ("Uninline
find_task_by_xxx set of functions"), no nsproxy, we only call
find_task_by_vpid() now.
hugetlb_cgroup.h is not needed since commit ab5ac90aec ("mm, hugetlb: do
not rely on overcommit limit during migration"), move_hugetlb_state() is
called and it belongs to hugetlb.h, which is already included.
balloon_compaction.h, we have more general movable_operations for non-lru
movable page migration, so it could be dropped.
memremap.h, userfaultfd_k.h and oom.h are introduced for zone device page
migration, but all functions are moved into migrate_device.c, so no needed
anymore too.
Link: https://lkml.kernel.org/r/20240905152432.626877-1-wangkefeng.wang@huawei.com
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Use find_get_task_by_vpid() to replace the task_struct find logic in
find_mm_struct(), note that this patch move the ptrace_may_access() call
out from rcu_read_lock() scope, this is ok because it actually does not
need it, find_get_task_by_vpid() already get the pid and task safely,
ptrace_may_access() can use the task safely, like what
sched_core_share_pid() similarly do.
Link: https://lkml.kernel.org/r/20240905153118.1205173-1-sunnanyong@huawei.com
Signed-off-by: Nanyong Sun <sunnanyong@huawei.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The aggregation interval of test purpose damon_attrs for
damon_test_nr_accesses_to_accesses_bp() becomes zero on 32 bit
architecture, since size of int and long types are same. As a result,
damon_nr_accesses_to_accesses_bp() call with the test data triggers
divide-by-zero exception. damon_nr_accesses_to_accesses_bp() shouldn't
be called with such data, and the non-test code avoids that by checking
the case on damon_update_monitoring_results(). Skip the test code in
the case, and add an explicit caution of the case on the comment for the
test target function.
Link: https://lkml.kernel.org/r/20240905162423.74053-1-sj@kernel.org
Fixes: 5e06ad5900 ("mm/damon/core-test: test max_nr_accesses overflow caused divide-by-zero")
Signed-off-by: SeongJae Park <sj@kernel.org>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Closes: https://lore.kernel.org/c771b962-a58f-435b-89e4-1211a9323181@roeck-us.net
Cc: Brendan Higgins <brendanhiggins@google.com>
Cc: David Gow <davidgow@google.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
PGFREE is currently updated in two code paths:
- __free_pages_ok(): for pages freed to the buddy allocator.
- free_unref_page_commit(): for pages freed to the pcplists.
Before commit df1acc8569 ("mm/page_alloc: avoid conflating IRQs disabled
with zone->lock"), free_unref_page_commit() used to fallback to freeing
isolated pages directly to the buddy allocator through free_one_page().
This was done _after_ updating PGFREE, so the counter was correctly
updated.
However, that commit moved the fallback logic to its callers (now called
free_unref_page() and free_unref_folios()), so PGFREE was no longer
updated in this fallback case.
Now that the code has developed, there are more cases in free_unref_page()
and free_unref_folios() where we fallback to calling free_one_page() (e.g.
!pcp_allowed_order(), pcp_spin_trylock() fails). These cases also miss
updating PGFREE.
To make sure PGFREE is updated in all cases where pages are freed to the
buddy allocator, move the update down the stack to free_one_page().
This was noticed through code inspection, although it should be noticeable
at runtime (at least with some workloads).
Link: https://lkml.kernel.org/r/20240904205419.821776-1-yosryahmed@google.com
Fixes: df1acc8569 ("mm/page_alloc: avoid conflating IRQs disabled with zone->lock")
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Cc: Brendan Jackman <jackmanb@google.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
damon_test_three_regions_in_vmas() initializes a maple tree with
MM_MT_FLAGS. The flags contains MT_FLAGS_LOCK_EXTERN, which means mt_lock
of the maple tree will not be used. And therefore the maple tree
initialization code skips initialization of the mt_lock. However,
__link_vmas(), which adds vmas for test to the maple tree, uses the
mt_lock. In other words, the uninitialized spinlock is used. The problem
becomes clear when spinlock debugging is turned on, since it reports
spinlock bad magic bug.
Fix the issue by excluding MT_FLAGS_LOCK_EXTERN from the maple tree
initialization flags. Note that we don't use empty flags to make it
further similar to the usage of mm maple tree, and to be prepared for
possible future changes, as suggested by Liam.
Link: https://lkml.kernel.org/r/20240904172931.1284-1-sj@kernel.org
Fixes: d0cf3dd47f ("damon: convert __damon_va_three_regions to use the VMA iterator")
Signed-off-by: SeongJae Park <sj@kernel.org>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Closes: https://lore.kernel.org/1453b2b2-6119-4082-ad9e-f3c5239bf87e@roeck-us.net
Suggested-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Take the end of a file write into consideration when deciding whether or
not to use huge pages for tmpfs files when the tmpfs filesystem is mounted
with huge=within_size
This allows large writes that append to the end of a file to automatically
use large pages.
Doing 4MB sequential writes without fallocate to a 16GB tmpfs file with
fio. The numbers without THP or with huge=always stay the same, but the
performance with huge=within_size now matches that of huge=always.
huge before after
4kB pages 1560 MB/s 1560 MB/s
within_size 1560 MB/s 4720 MB/s
always: 4720 MB/s 4720 MB/s
[akpm@linux-foundation.org: coding-style cleanups]
Link: https://lkml.kernel.org/r/20240903111928.7171e60c@imladris.surriel.com
Signed-off-by: Rik van Riel <riel@surriel.com>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Tested-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Darrick J. Wong <djwong@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
recompress device attribute supports alg=NAME parameter so that we can
specify only one particular algorithm we want to perform recompression
with. However, with algo params we now can have several exactly same
secondary algorithms but each with its own params tuning (e.g. priority 1
configured to use more aggressive level, and priority 2 configured to use
a pre-trained dictionary). Support priority=NUM parameter so that we can
correctly determine which secondary algorithm we want to use.
Link: https://lkml.kernel.org/r/20240902105656.1383858-25-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nick Terrell <terrelln@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Support pre-trained dictionary param. lz4 doesn't mandate specific format
of the dictionary and even zstd --train can be used to train a dictionary
for lz4, according to [1].
TEST
====
*** lz4
/sys/block/zram0/mm_stat
1750654976 664188565 676864000 0 676864000 1 0 34288 34288
*** lz4 dict=/etc/lz4-dict-amd64
/sys/block/zram0/mm_stat
1750638592 619891141 632053760 0 632053760 1 0 34278 34278
*** lz4 level=5 dict=/etc/lz4-dict-amd64
/sys/block/zram0/mm_stat
1750638592 727174243 740810752 0 740810752 1 0 34437 34437
[1] https://github.com/lz4/lz4/issues/557
Link: https://lkml.kernel.org/r/20240902105656.1383858-21-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nick Terrell <terrelln@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Immutable params never change once comp has been allocated and setup, so
we don't need to store multiple copies of them in each per-CPU backend
context. Move those to per-comp zcomp_params and pass it to backends
callbacks for requests execution. Basically, this means parameters
sharing between different contexts.
Also introduce two new backends callbacks: setup_params() and
release_params(). First, we need to validate params in a driver-specific
way; second, driver may want to allocate its specific representation of
the params which is needed to execute requests.
Link: https://lkml.kernel.org/r/20240902105656.1383858-20-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nick Terrell <terrelln@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This attribute is used to setup compression algorithms' parameters, so we
can tweak algorithms' characteristics. At this point only 'level' is
supported (to be extended in the future).
Each call sets up parameters for one particular algorithm, which should be
specified either by the algorithm's priority or algo name. This is
expected to be called after corresponding algorithm is selected via
comp_algorithm or recomp_algorithm.
echo "priority=0 level=1" > /sys/block/zram0/algorithm_params
or
echo "algo=zstd level=1" > /sys/block/zram0/algorithm_params
Link: https://lkml.kernel.org/r/20240902105656.1383858-16-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nick Terrell <terrelln@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Peter Xu
Yu Zhao
Uladzislau Rezki (Sony)
ZhangPeng
Shakeel Butt
Liam R. Howlett
Kinsey Ho
Andrew Morton
Kefeng Wang
Nanyong Sun
SeongJae Park
Sven Schnelle
Yosry Ahmed
Mark Brown
Sergey Senozhatsky
Takaya Saeki
Rik van Riel