Patch series "mm: folio_zero_user: clear page ranges", v11.
This series adds clearing of contiguous page ranges for hugepages.
The series improves on the current discontiguous clearing approach in two
ways:
- clear pages in a contiguous fashion.
- use batched clearing via clear_pages() wherever exposed.
The first is useful because it allows us to make much better use of
hardware prefetchers.
The second, enables advertising the real extent to the processor. Where
specific instructions support it (ex. string instructions on x86; "mops"
on arm64 etc), a processor can optimize based on this because, instead of
seeing a sequence of 8-byte stores, or a sequence of 4KB pages, it sees a
larger unit being operated on.
For instance, AMD Zen uarchs (for extents larger than LLC-size) switch to
a mode where they start eliding cacheline allocation. This is helpful not
just because it results in higher bandwidth, but also because now the
cache is not evicting useful cachelines and replacing them with zeroes.
Demand faulting a 64GB region shows performance improvement:
$ perf bench mem mmap -p $pg-sz -f demand -s 64GB -l 5
baseline +series
(GBps +- %stdev) (GBps +- %stdev)
pg-sz=2MB 11.76 +- 1.10% 25.34 +- 1.18% [*] +115.47% preempt=*
pg-sz=1GB 24.85 +- 2.41% 39.22 +- 2.32% + 57.82% preempt=none|voluntary
pg-sz=1GB (similar) 52.73 +- 0.20% [#] +112.19% preempt=full|lazy
[*] This improvement is because switching to sequential clearing
allows the hardware prefetchers to do a much better job.
[#] For pg-sz=1GB a large part of the improvement is because of the
cacheline elision mentioned above. preempt=full|lazy improves upon
that because, not needing explicit invocations of cond_resched() to
ensure reasonable preemption latency, it can clear the full extent
as a single unit. In comparison the maximum extent used for
preempt=none|voluntary is PROCESS_PAGES_NON_PREEMPT_BATCH (32MB).
When provided the full extent the processor forgoes allocating
cachelines on this path almost entirely.
(The hope is that eventually, in the fullness of time, the lazy
preemption model will be able to do the same job that none or
voluntary models are used for, allowing us to do away with
cond_resched().)
Raghavendra also tested previous version of the series on AMD Genoa and
sees similar improvement [1] with preempt=lazy.
$ perf bench mem map -p $page-size -f populate -s 64GB -l 10
base patched change
pg-sz=2MB 12.731939 GB/sec 26.304263 GB/sec 106.6%
pg-sz=1GB 26.232423 GB/sec 61.174836 GB/sec 133.2%
This patch (of 8):
Let's drop all variants that effectively map to clear_page() and provide
it in a generic variant instead.
We'll use the macro clear_user_page to indicate whether an architecture
provides it's own variant.
Also, clear_user_page() is only called from the generic variant of
clear_user_highpage(), so define it only if the architecture does not
provide a clear_user_highpage(). And, for simplicity define it in
linux/highmem.h.
Note that for parisc, clear_page() and clear_user_page() map to
clear_page_asm(), so we can just get rid of the custom clear_user_page()
implementation. There is a clear_user_page_asm() function on parisc, that
seems to be unused. Not sure what's up with that.
Link: https://lkml.kernel.org/r/20260107072009.1615991-1-ankur.a.arora@oracle.com
Link: https://lkml.kernel.org/r/20260107072009.1615991-2-ankur.a.arora@oracle.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Co-developed-by: Ankur Arora <ankur.a.arora@oracle.com>
Signed-off-by: Ankur Arora <ankur.a.arora@oracle.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Ankur Arora <ankur.a.arora@oracle.com>
Cc: "Borislav Petkov (AMD)" <bp@alien8.de>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: David Hildenbrand <david@kernel.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Konrad Rzessutek Wilk <konrad.wilk@oracle.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Li Zhe <lizhe.67@bytedance.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Mateusz Guzik <mjguzik@gmail.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Raghavendra K T <raghavendra.kt@amd.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
While the GCC and Clang compilers already define __ASSEMBLER__
automatically when compiling assembly code, __ASSEMBLY__ is a
macro that only gets defined by the Makefiles in the kernel.
This can be very confusing when switching between userspace
and kernelspace coding, or when dealing with uapi headers that
rather should use __ASSEMBLER__ instead. So let's standardize on
the __ASSEMBLER__ macro that is provided by the compilers now.
This is a completely mechanical patch (done with a simple "sed -i"
statement).
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Stafford Horne <shorne@gmail.com>
Cc: linux-openrisc@vger.kernel.org
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Stafford Horne <shorne@gmail.com>
Pull asm-generic updates from Arnd Bergmann:
"These are a number of unrelated cleanups, generally simplifying the
architecture specific header files:
- A series from Al Viro simplifies asm/vga.h, after it turns out that
most of it can be generalized.
- A series from Julian Vetter adds a common version of
memcpy_{to,from}io() and memset_io() and changes most architectures
to use that instead of their own implementation
- A series from Niklas Schnelle concludes his work to make PC style
inb()/outb() optional
- Nicolas Pitre contributes improvements for the generic do_div()
helper
- Christoph Hellwig adds a generic version of page_to_phys() and
phys_to_page(), replacing the slightly different architecture
specific definitions.
- Uwe Kleine-Koenig has a minor cleanup for ioctl definitions"
* tag 'asm-generic-3.13' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic: (24 commits)
empty include/asm-generic/vga.h
sparc: get rid of asm/vga.h
asm/vga.h: don't bother with scr_mem{cpy,move}v() unless we need to
vt_buffer.h: get rid of dead code in default scr_...() instances
tty: serial: export serial_8250_warn_need_ioport
lib/iomem_copy: fix kerneldoc format style
hexagon: simplify asm/io.h for !HAS_IOPORT
loongarch: Use new fallback IO memcpy/memset
csky: Use new fallback IO memcpy/memset
arm64: Use new fallback IO memcpy/memset
New implementation for IO memcpy and IO memset
watchdog: Add HAS_IOPORT dependency for SBC8360 and SBC7240
__arch_xprod64(): make __always_inline when optimizing for performance
ARM: div64: improve __arch_xprod_64()
asm-generic/div64: optimize/simplify __div64_const32()
lib/math/test_div64: add some edge cases relevant to __div64_const32()
asm-generic: add an optional pfn_valid check to page_to_phys
asm-generic: provide generic page_to_phys and phys_to_page implementations
asm-generic/io.h: Remove I/O port accessors for HAS_IOPORT=n
tty: serial: handle HAS_IOPORT dependencies
...
page_to_phys is duplicated by all architectures, and from some strange
reason placed in <asm/io.h> where it doesn't fit at all.
phys_to_page is only provided by a few architectures despite having a lot
of open coded users.
Provide generic versions in <asm-generic/memory_model.h> to make these
helpers more easily usable.
Note with this patch powerpc loses the CONFIG_DEBUG_VIRTUAL pfn_valid
check. It will be added back in a generic version later.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Most architectures only support a single hardcoded page size. In order
to ensure that each one of these sets the corresponding Kconfig symbols,
change over the PAGE_SHIFT definition to the common one and allow
only the hardware page size to be selected.
Acked-by: Guo Ren <guoren@kernel.org>
Acked-by: Heiko Carstens <hca@linux.ibm.com>
Acked-by: Stafford Horne <shorne@gmail.com>
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Remove the broken pfn_to_virt() on architectures csky/hexagon/openrisc.
The pfn_to_virt() on those architectures takes PFN instead of PA as the
input to macro __va(), with PAGE_SHIFT applying to the converted VA, which
is not right, especially when there's a non-zero offset in __va(). [1]
The broken pfn_to_virt() was noticed when checking how page_to_virt() is
unfolded on x86. It turns out that the one in linux/mm.h instead of in
asm-generic/page.h is compiled for x86. However, page_to_virt() in
asm-generic/page.h is found out expanding to pfn_to_virt() with a bug
described above. The pfn_to_virt() is found out not right as well on
architectures csky/hexagon/openrisc.
Since there's no single caller on csky/hexagon/openrisc to pfn_to_virt()
and there are functions doing similar things as pfn_to_virt() --
- pfn_to_kaddr() in asm/page.h for csky,
- page_to_virt() in asm/page.h for hexagon, and
- page_to_virt() in linux/mm.h for openrisc,
just delete the pfn_to_virt() on those architectures.
The pfn_to_virt() in asm-generic/page.h is not touched in this patch as
it's referenced by page_to_virt() in that header while the whole header is
going to be removed as a whole due to no one including it.
Link:https://lore.kernel.org/all/20240131055159.2506-1-yan.y.zhao@intel.com [1]
Cc: Linus Walleij <linus.walleij@linaro.org>
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Yan Zhao <yan.y.zhao@intel.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Making virt_to_pfn() a static inline taking a strongly typed
(const void *) makes the contract of a passing a pointer of that
type to the function explicit and exposes any misuse of the
macro virt_to_pfn() acting polymorphic and accepting many types
such as (void *), (unitptr_t) or (unsigned long) as arguments
without warnings.
For symmetry, do the same with pfn_to_virt().
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Stafford Horne <shorne@gmail.com>
Based on 1 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license as published by
the free software foundation either version 2 of the license or at
your option any later version
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-or-later
has been chosen to replace the boilerplate/reference in 3029 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Allison Randal <allison@lohutok.net>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190527070032.746973796@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To align with generic code and other architectures that expect the macro
page_to_virt to produce an expression whose type is 'void*', drop the
arch specific definition, which is never referenced anyway.
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
They are not symmetric with each other, neither are used in real world
(can not be found by grep command in source code root directory), so
remove them.
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
OpenRISC does not have global memory_start and memory_end
symbols.
The prototypes are in vain.
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Jonas Bonn <jonas@southpole.se>
virt_addr_valid() shouldn't be comparing the address to memory_end which is
a phys_addr_t. Change this to do like other arches and check that the
address falls within a valid page frame.
Signed-off-by: Jonas Bonn <jonas@southpole.se>
