| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Wire up crc64_be_arch() and crc64_nvme_arch() for 64-bit RISC-V using
crc-clmul-template.h. This greatly improves the performance of these
CRCs on Zbc-capable CPUs in 64-bit kernels.
These optimized CRC64 functions are not yet supported in 32-bit kernels,
since crc-clmul-template.h assumes that the CRC fits in an unsigned
long. That implementation limitation could be addressed, but it would
add a fair bit of complexity, so it has been omitted for now.
Tested-by: Björn Töpel <bjorn@rivosinc.com>
Acked-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250216225530.306980-5-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Wire up crc_t10dif_arch() for RISC-V using crc-clmul-template.h. This
greatly improves CRC-T10DIF performance on Zbc-capable CPUs.
Tested-by: Björn Töpel <bjorn@rivosinc.com>
Acked-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250216225530.306980-4-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Delete the previous Zbc optimized CRC32 code, and re-implement it using
the new template. The new implementation is more optimized and shares
more code among CRC variants.
Tested-by: Björn Töpel <bjorn@rivosinc.com>
Acked-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250216225530.306980-3-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Add a "template" crc-clmul-template.h that can generate RISC-V Zbc
optimized CRC functions. Each generated CRC function is parameterized
by CRC length and bit order, and it accepts a pointer to the constants
struct required for the specific CRC polynomial desired. Update
gen-crc-consts.py to support generating the needed constants structs.
This makes it possible to easily wire up a Zbc optimized implementation
of almost any CRC.
The design generally follows what I did for x86, but it is simplified by
using RISC-V's scalar carryless multiplication Zbc, which has no
equivalent on x86. RISC-V's clmulr instruction is also helpful. A
potential switch to Zvbc (or support for Zvbc alongside Zbc) is left for
future work. For long messages Zvbc should be fastest, but it would
need to be shown to be worthwhile over just using Zbc which is
significantly more convenient to use, especially in the kernel context.
Compared to the existing Zbc-optimized CRC32 code and the earlier
proposed Zbc-optimized CRC-T10DIF code
(https://lore.kernel.org/r/20250211071101.181652-1-zhihang.shao.iscas@gmail.com),
this submission deduplicates the code among CRC variants and is
significantly more optimized. It uses "folding" to take better
advantage of instruction-level parallelism (to a more limited extent
than x86 for now, but it could be extended to more), it reworks the
Barrett reduction to eliminate unnecessary instructions, and it
documents all the math used and makes all the constants reproducible.
Tested-by: Björn Töpel <bjorn@rivosinc.com>
Acked-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Link: https://lore.kernel.org/r/20250216225530.306980-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
The assembly functions generated by crc-pclmul-template.S are called
only via static_call, so they do not need to begin with an endbr
instruction. But objtool still warns about a missing endbr by default.
Add ANNOTATE_NOENDBR to suppress these warnings:
vmlinux.o: warning: objtool: crc32_x86_init+0x1c0: relocation to !ENDBR: crc32_lsb_vpclmul_avx10_256+0x0
vmlinux.o: warning: objtool: crc64_x86_init+0x183: relocation to !ENDBR: crc64_msb_vpclmul_avx10_256+0x0
vmlinux.o: warning: objtool: crc_t10dif_x86_init+0x183: relocation to !ENDBR: crc16_msb_vpclmul_avx10_256+0x0
vmlinux.o: warning: objtool: __SCK__crc32_lsb_pclmul+0x0: data relocation to !ENDBR: crc32_lsb_pclmul_sse+0x0
vmlinux.o: warning: objtool: __SCK__crc64_lsb_pclmul+0x0: data relocation to !ENDBR: crc64_lsb_pclmul_sse+0x0
vmlinux.o: warning: objtool: __SCK__crc64_msb_pclmul+0x0: data relocation to !ENDBR: crc64_msb_pclmul_sse+0x0
vmlinux.o: warning: objtool: __SCK__crc16_msb_pclmul+0x0: data relocation to !ENDBR: crc16_msb_pclmul_sse+0x0
Fixes: 8d2d3e72e35b ("x86/crc: add "template" for [V]PCLMULQDQ based CRC functions")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Closes: https://lore.kernel.org/r/20250217170555.3d14df62@canb.auug.org.au/
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20250217193230.100443-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
crc32c_arch() is affected by
https://github.com/llvm/llvm-project/issues/20571 where clang
unnecessarily spills the inputs to "rm"-constrained operands to the
stack. Replace "rm" with ASM_INPUT_RM which partially works around this
by expanding to "r" when the compiler is clang. This results in better
code generation with clang, though still not optimal.
Link: https://lore.kernel.org/r/20250210210741.471725-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Add x86_64 [V]PCLMULQDQ optimized implementations of crc64_be() and
crc64_nvme() by wiring them up to crc-pclmul-template.S.
crc64_be() is used by bcache and bcachefs, and crc64_nvme() is used by
blk-integrity. Both features can CRC large amounts of data, and the
developers of both features have expressed interest in having these CRCs
be optimized. So this optimization should be worthwhile. (See
https://lore.kernel.org/r/v36sousjd5ukqlkpdxslvpu7l37zbu7d7slgc2trjjqwty2bny@qgzew34feo2r
and
https://lore.kernel.org/r/20220222163144.1782447-11-kbusch@kernel.org)
Benchmark results on AMD Ryzen 9 9950X (Zen 5) using crc_kunit:
crc64_be:
Length Before After
------ ------ -----
1 633 MB/s 477 MB/s
16 717 MB/s 2517 MB/s
64 715 MB/s 7525 MB/s
127 714 MB/s 10002 MB/s
128 713 MB/s 13344 MB/s
200 715 MB/s 15752 MB/s
256 714 MB/s 22933 MB/s
511 715 MB/s 28025 MB/s
512 714 MB/s 49772 MB/s
1024 715 MB/s 65261 MB/s
3173 714 MB/s 78773 MB/s
4096 714 MB/s 83315 MB/s
16384 714 MB/s 89487 MB/s
crc64_nvme:
Length Before After
------ ------ -----
1 716 MB/s 474 MB/s
16 717 MB/s 3303 MB/s
64 713 MB/s 7940 MB/s
127 715 MB/s 9867 MB/s
128 714 MB/s 13698 MB/s
200 715 MB/s 15995 MB/s
256 714 MB/s 23479 MB/s
511 714 MB/s 28013 MB/s
512 715 MB/s 51533 MB/s
1024 715 MB/s 66788 MB/s
3173 715 MB/s 79182 MB/s
4096 715 MB/s 83966 MB/s
16384 715 MB/s 89739 MB/s
Acked-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250210174540.161705-7-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Instantiate crc-pclmul-template.S for crc_t10dif and delete the original
PCLMULQDQ optimized implementation. This has the following advantages:
- Less CRC-variant-specific code.
- VPCLMULQDQ support, greatly improving performance on sufficiently long
messages on newer CPUs.
- A faster reduction from 128 bits to the final CRC.
- Support for i386.
Benchmark results on AMD Ryzen 9 9950X (Zen 5) using crc_kunit:
Length Before After
------ ------ -----
1 440 MB/s 386 MB/s
16 1865 MB/s 2008 MB/s
64 4343 MB/s 6917 MB/s
127 5440 MB/s 8909 MB/s
128 5533 MB/s 12150 MB/s
200 5908 MB/s 14423 MB/s
256 15870 MB/s 21288 MB/s
511 14219 MB/s 25840 MB/s
512 18361 MB/s 37797 MB/s
1024 19941 MB/s 61374 MB/s
3173 20461 MB/s 74909 MB/s
4096 21310 MB/s 78919 MB/s
16384 21663 MB/s 85012 MB/s
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250210174540.161705-6-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Instantiate crc-pclmul-template.S for crc32_le, and delete the original
PCLMULQDQ optimized implementation. This has the following advantages:
- Less CRC-variant-specific code.
- VPCLMULQDQ support, greatly improving performance on sufficiently long
messages on newer CPUs.
- A faster reduction from 128 bits to the final CRC.
- Support for lengths not a multiple of 16 bytes, improving performance
for such lengths.
- Support for misaligned buffers, improving performance in such cases.
Benchmark results on AMD Ryzen 9 9950X (Zen 5) using crc_kunit:
Length Before After
------ ------ -----
1 427 MB/s 605 MB/s
16 710 MB/s 3631 MB/s
64 704 MB/s 7615 MB/s
127 3610 MB/s 9710 MB/s
128 8759 MB/s 12702 MB/s
200 7083 MB/s 15343 MB/s
256 17284 MB/s 22904 MB/s
511 10919 MB/s 27309 MB/s
512 19849 MB/s 48900 MB/s
1024 21216 MB/s 62630 MB/s
3173 22150 MB/s 72437 MB/s
4096 22496 MB/s 79593 MB/s
16384 22018 MB/s 85106 MB/s
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250210174540.161705-5-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
The Linux kernel implements many variants of CRC, such as crc16,
crc_t10dif, crc32_le, crc32c, crc32_be, crc64_nvme, and crc64_be. On
x86, except for crc32c which has special scalar instructions, the
fastest way to compute any of these CRCs on any message of length
roughly >= 16 bytes is to use the SIMD carryless multiplication
instructions PCLMULQDQ or VPCLMULQDQ. Depending on the available CPU
features this can mean PCLMULQDQ+SSE4.1, VPCLMULQDQ+AVX2,
VPCLMULQDQ+AVX10/256, or VPCLMULQDQ+AVX10/512 (or the AVX512 equivalents
to AVX10/*). This results in a total of 20+ CRC implementations being
potentially needed to properly optimize all CRCs that someone cares
about for x86. Besides crc32c, currently only crc32_le and crc_t10dif
are actually optimized for x86, and they only use PCLMULQDQ, which means
they can be 2-4x slower than what is possible with VPCLMULQDQ.
Fortunately, at a high level the code that is needed for any
[V]PCLMULQDQ based CRC implementation is mostly the same. Therefore,
this patch introduces an assembly macro that expands into the body of a
[V]PCLMULQDQ based CRC function for a given number of bits (8, 16, 32,
or 64), bit order (lsb or msb-first), vector length, and AVX level.
The function expects to be passed a constants table, specific to the
polynomial desired, that was generated by the script previously added.
When two CRC variants share the same number of bits and bit order, the
same functions can be reused, with only the constants table differing.
A new C header is also added to make it easy to integrate the new
assembly code using a static call.
The result is that it becomes straightforward to wire up an optimized
implementation of any CRC-8, CRC-16, CRC-32, or CRC-64 for x86. Later
patches will wire up specific CRC variants.
Although this new template allows easily generating many functions, care
was taken to still keep the binary size fairly low. Each generated
function is only 550 to 850 bytes depending on the CRC variant and
target CPU features. And only one function per CRC variant is actually
used at runtime (since all functions support all lengths >= 16 bytes).
Note that a similar approach should also work for other architectures
that have carryless multiplication instructions, such as arm64.
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250210174540.161705-4-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Add a Python script that generates constants for computing the given CRC
variant(s) using x86's pclmulqdq or vpclmulqdq instructions.
This is specifically tuned for x86's crc-pclmul-template.S. However,
other architectures with a 64x64 => 128-bit carryless multiplication
instruction should be able to use the generated constants too. (Some
tweaks may be warranted based on the exact instructions available on
each arch, so the script may grow an arch argument in the future.)
The script also supports generating the tables needed for table-based
CRC computation. Thus, it can also be used to reproduce the tables like
t10_dif_crc_table[] and crc16_table[] that are currently hardcoded in
the source with no generation script explicitly documented.
Python is used rather than C since it enables implementing the CRC math
in the simplest way possible, using arbitrary precision integers. The
outputs of this script are intended to be checked into the repo, so
Python will continue to not be required to build the kernel, and the
script has been optimized for simplicity rather than performance.
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250210174540.161705-3-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Lift zmm_exclusion_list in aesni-intel_glue.c into the x86 CPU setup
code, and add a new x86 CPU feature flag X86_FEATURE_PREFER_YMM that is
set when the CPU is on this list.
This allows other code in arch/x86/, such as the CRC library code, to
apply the same exclusion list when deciding whether to execute 256-bit
or 512-bit optimized functions.
Note that full AVX512 support including ZMM registers is still exposed
to userspace and is still supported for in-kernel use. This flag just
indicates whether in-kernel code should prefer to use YMM registers.
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250210174540.161705-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
With the "crct10dif" algorithm having been removed from the crypto API,
crc_t10dif_is_optimized() is no longer used.
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208175647.12333-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Remove the "crct10dif" shash algorithm from the crypto API. It has no
known user now that the lib is no longer built on top of it. It has no
remaining references in kernel code. The only other potential users
would be the usual components that allow specifying arbitrary hash
algorithms by name, namely AF_ALG and dm-integrity. However there are
no indications that "crct10dif" is being used with these components.
Debian Code Search and web searches don't find anything relevant, and
explicitly grepping the source code of the usual suspects (cryptsetup,
libell, iwd) finds no matches either. "crc32" and "crc32c" are used in
a few more places, but that doesn't seem to be the case for "crct10dif".
crc_t10dif_update() is also tested by crc_kunit now, so the test
coverage provided via the crypto self-tests is no longer needed.
Also note that the "crct10dif" shash algorithm was inconsistent with the
rest of the shash API in that it wrote the digest in CPU endianness,
making the resulting byte array differ on little endian vs. big endian
platforms. This means it was effectively just built for use by the lib
functions, and it was not actually correct to treat it as "just another
hash function" that could be dropped in via the shash API.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250206173857.39794-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Following the standardization on crc32c() as the lib entry point for the
Castagnoli CRC32 instead of the previous mix of crc32c(), crc32c_le(),
and __crc32c_le(), make the same change to the underlying base and arch
functions that implement it.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208024911.14936-7-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Since the Castagnoli CRC32 is now always just crc32c(), rename
__crc32c_le_combine() and __crc32c_le_shift() accordingly.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208024911.14936-6-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
For historical reasons, the Castagnoli CRC32 is available under 3 names:
crc32c(), crc32c_le(), and __crc32c_le(). Most callers use crc32c().
The more verbose versions are not really warranted; there is no "_be"
version that the "_le" version needs to be differentiated from, and the
leading underscores are pointless.
Therefore, let's standardize on just crc32c(). Remove the other two
names, and update callers accordingly.
Specifically, the new crc32c() comes from what was previously
__crc32c_le(), so compared to the old crc32c() it now takes a size_t
length rather than unsigned int, and it's now in linux/crc32.h instead
of just linux/crc32c.h (which includes linux/crc32.h).
Later patches will also rename __crc32c_le_combine(), crc32c_le_base(),
and crc32c_le_arch().
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208024911.14936-5-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Drop the use of __pure and __attribute_const__ from the CRC32 library
functions that had them. Both of these are unusual optimizations that
don't help properly written code. They seem more likely to cause
problems than have any real benefit.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208024911.14936-4-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Update crc32_le(), crc32_be(), and __crc32c_le() to take the data as a
'const void *' instead of 'const u8 *'.
This makes them slightly easier to use, as it can eliminate the need for
casts in the calling code. It's the only pointer argument, so there is
no possibility for confusion with another pointer argument.
Also, some of the CRC library functions, for example crc32c() and
crc64_be(), already used 'const void *'. Let's standardize on that, as
it seems like a better choice.
The underlying base and arch functions continue to use 'const u8 *', as
that is often more convenient for the implementation.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208024911.14936-3-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Remove enum crc_op_size and enum crc_type, since they are never actually
used. Tokens with the names of the enum values do appear in the file,
but they are only used for token concatenation with the preprocessor.
This prevents a conflict with the addition of crc32c() to linux/crc32.h.
Reported-by: Nathan Chancellor <nathan@kernel.org>
Closes: https://lore.kernel.org/r/20250207224233.GA1261167@ax162
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250208024911.14936-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Remove all remaining references to CONFIG_CRC32_BIT,
CONFIG_CRC32_SARWATE, CONFIG_CRC32_SLICEBY4, and CONFIG_CRC32_SLICEBY8.
These options no longer exist, now that we've standardized on a single
generic CRC32 implementation.
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20250205000424.75149-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Add support for architecture-optimized implementations of the CRC64
library functions, following the approach taken for the CRC32 and
CRC-T10DIF library functions.
Also take the opportunity to tweak the function prototypes:
- Use 'const void *' for the lib entry points (since this is easier for
users) but 'const u8 *' for the underlying arch and generic functions
(since this is easier for the implementations of these functions).
- Don't bother with __pure. It's an unusual optimization that doesn't
help properly written code. It's a weird quirk we can do without.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250130035130.180676-6-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Wire up crc64_nvme() to the new CRC unit test and benchmark.
This replaces and improves on the test coverage that was lost by
removing this CRC variant from the crypto API.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250130035130.180676-5-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
This CRC64 variant comes from the NVME NVM Command Set Specification
(https://nvmexpress.org/wp-content/uploads/NVM-Express-NVM-Command-Set-Specification-1.0e-2024.07.29-Ratified.pdf).
The "Rocksoft Model CRC Algorithm", published in 1993 and available at
https://www.zlib.net/crc_v3.txt, is a generalized CRC algorithm that can
calculate any variant of CRC, given a list of parameters such as
polynomial, bit order, etc. It is not a CRC variant.
The NVME NVM Command Set Specification has a table that gives the
"Rocksoft Model Parameters" for the CRC variant it uses. When support
for this CRC variant was added to Linux, this table seems to have been
misinterpreted as naming the CRC variant the "Rocksoft" CRC. In fact,
the table names the CRC variant as the "NVM Express 64b CRC".
Most implementations of this CRC variant outside Linux have been calling
it CRC64-NVME. Therefore, update Linux to match.
While at it, remove the superfluous "update" from the function name, so
crc64_rocksoft_update() is now just crc64_nvme(), matching most of the
other CRC library functions.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250130035130.180676-4-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Remove crc64-rocksoft from the crypto API. It has no known user now
that the lib is no longer built on top of it. It was also added much
more recently than the longstanding crc32 and crc32c. Unlike crc32 and
crc32c, crc64-rocksoft is also not mentioned in the dm-integrity
documentation and there are no references to it in anywhere in the
cryptsetup git repo, so it is unlikely to have any user there either.
Also, this CRC variant is named incorrectly; it has nothing to do with
Rocksoft and should be called crc64-nvme. That is yet another reason to
remove it from the crypto API; we would not want anyone to start
depending on the current incorrect algorithm name of crc64-rocksoft.
Note that this change temporarily makes this CRC variant not be covered
by any tests, as previously it was relying on the crypto self-tests.
This will be fixed by adding this CRC variant to crc_kunit.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250130035130.180676-3-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
Following what was done for the CRC32 and CRC-T10DIF library functions,
get rid of the pointless use of the crypto API and make
crc64_rocksoft_update() call into the library directly. This is faster
and simpler.
Remove crc64_rocksoft() (the version of the function that did not take a
'crc' argument) since it is unused.
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: "Martin K. Petersen" <martin.petersen@oracle.com>
Acked-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20250130035130.180676-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
|
|
|
|
Summary of Changes since 2024.11.30:
Fix regression in 2023.11.07 that affinitized forked child
in one-shot mode.
Harden one-shot mode against hotplug online/offline
Enable RAPL SysWatt column by default.
Add initial PTL, CWF platform support.
Harden initial PMT code in response to early use.
Enable first built-in PMT counter: CWF c1e residency
Refuse to run on unsupported platforms without --force,
to encourage updating to a version that supports the system,
and to avoid no-so-useful measurement results.
Signed-off-by: Len Brown <len.brown@intel.com>
|
|
MM developers have an interest in the xarray code.
Cc: David Gow <davidgow@google.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Tamir Duberstein <tamird@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Revert c7bb5cf9fc4e ("xarray: port tests to kunit"). It broke the build
when compiing the xarray userspace test harness code.
Reported-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Closes: https://lkml.kernel.org/r/07cf896e-adf8-414f-a629-a808fc26014a@oracle.com
Cc: David Gow <davidgow@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Tamir Duberstein <tamird@gmail.com>
Cc: "Liam R. Howlett" <Liam.Howlett@oracle.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Ensure test-only changes are sent to the relevant maintainer.
Link: https://lkml.kernel.org/r/20250129-xarray-test-maintainer-v1-1-482e31f30f47@gmail.com
Signed-off-by: Tamir Duberstein <tamird@gmail.com>
Cc: Mattew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Update .mailmap to reflect my new (and final) primary email address,
carlos.bilbao@kernel.org. Also update contact information in files
Documentation/translations/sp_SP/index.rst and MAINTAINERS.
Link: https://lkml.kernel.org/r/20250130012248.1196208-1-carlos.bilbao@kernel.org
Signed-off-by: Carlos Bilbao <carlos.bilbao@kernel.org>
Cc: Carlos Bilbao <bilbao@vt.edu>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Mattew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
gather_bootmem_prealloc() assumes the start nid as 0 and size as
num_node_state(N_MEMORY). That means in case if memory attached numa
nodes are interleaved, then gather_bootmem_prealloc_parallel() will fail
to scan few of these nodes.
Since memory attached numa nodes can be interleaved in any fashion, hence
ensure that the current code checks for all numa node ids
(.size = nr_node_ids). Let's still keep max_threads as N_MEMORY, so that
it can distributes all nr_node_ids among the these many no. threads.
e.g. qemu cmdline
========================
numa_cmd="-numa node,nodeid=1,memdev=mem1,cpus=2-3 -numa node,nodeid=0,cpus=0-1 -numa dist,src=0,dst=1,val=20"
mem_cmd="-object memory-backend-ram,id=mem1,size=16G"
w/o this patch for cmdline (default_hugepagesz=1GB hugepagesz=1GB hugepages=2):
==========================
~ # cat /proc/meminfo |grep -i huge
AnonHugePages: 0 kB
ShmemHugePages: 0 kB
FileHugePages: 0 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 1048576 kB
Hugetlb: 0 kB
with this patch for cmdline (default_hugepagesz=1GB hugepagesz=1GB hugepages=2):
===========================
~ # cat /proc/meminfo |grep -i huge
AnonHugePages: 0 kB
ShmemHugePages: 0 kB
FileHugePages: 0 kB
HugePages_Total: 2
HugePages_Free: 2
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 1048576 kB
Hugetlb: 2097152 kB
Link: https://lkml.kernel.org/r/f8d8dad3a5471d284f54185f65d575a6aaab692b.1736592534.git.ritesh.list@gmail.com
Fixes: b78b27d02930 ("hugetlb: parallelize 1G hugetlb initialization")
Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Reported-by: Pavithra Prakash <pavrampu@linux.ibm.com>
Suggested-by: Muchun Song <muchun.song@linux.dev>
Tested-by: Sourabh Jain <sourabhjain@linux.ibm.com>
Reviewed-by: Luiz Capitulino <luizcap@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Donet Tom <donettom@linux.ibm.com>
Cc: Gang Li <gang.li@linux.dev>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
We can run into an infinite loop in __get_longterm_locked() when
collect_longterm_unpinnable_folios() finds only folios that are isolated
from the LRU or were never added to the LRU. This can happen when all
folios to be pinned are never added to the LRU, for example when
vm_ops->fault allocated pages using cma_alloc() and never added them to
the LRU.
Fix it by simply taking a look at the list in the single caller, to see if
anything was added.
[zhaoyang.huang@unisoc.com: move definition of local]
Link: https://lkml.kernel.org/r/20250122012604.3654667-1-zhaoyang.huang@unisoc.com
Link: https://lkml.kernel.org/r/20250121020159.3636477-1-zhaoyang.huang@unisoc.com
Fixes: 67e139b02d99 ("mm/gup.c: refactor check_and_migrate_movable_pages()")
Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
Reviewed-by: John Hubbard <jhubbard@nvidia.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Aijun Sun <aijun.sun@unisoc.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
There is a code error that will cause the swap entry allocator to reclaim
and check the whole cluster with an unexpected tail offset instead of the
part that needs to be reclaimed. This may cause corruption of the swap
map, so fix it.
Link: https://lkml.kernel.org/r/20250130115131.37777-1-ryncsn@gmail.com
Fixes: 3b644773eefd ("mm, swap: reduce contention on device lock")
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Chris Li <chrisl@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Update my email address.
Link: https://lkml.kernel.org/r/20250122-wip-obbardc-update-email-v2-1-12bde6b79ad0@linaro.org
Signed-off-by: Christopher Obbard <christopher.obbard@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
On NUMA systems, __GFP_THISNODE indicates that an allocation _must_ be on
a particular node, and failure to allocate on the desired node will result
in a failed allocation.
Skip __GFP_THISNODE allocations if we are running on a NUMA system, since
KFENCE can't guarantee which node its pool pages are allocated on.
Link: https://lkml.kernel.org/r/20250124120145.410066-1-elver@google.com
Fixes: 236e9f153852 ("kfence: skip all GFP_ZONEMASK allocations")
Signed-off-by: Marco Elver <elver@google.com>
Reported-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Christoph Lameter <cl@linux.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Chistoph Lameter <cl@linux.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Since nilfs_bmap_lookup_contig() in nilfs_fiemap() calculates its result
by being prepared to go through potentially maxblocks == INT_MAX blocks,
the value in n may experience an overflow caused by left shift of blkbits.
While it is extremely unlikely to occur, play it safe and cast right hand
expression to wider type to mitigate the issue.
Found by Linux Verification Center (linuxtesting.org) with static analysis
tool SVACE.
Link: https://lkml.kernel.org/r/20250124222133.5323-1-konishi.ryusuke@gmail.com
Fixes: 622daaff0a89 ("nilfs2: fiemap support")
Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
There are 4 NUMA nodes on my machine, and each NUMA node has 32GB of
memory. I have configured 16GB of CMA memory on each NUMA node, and
starting a 32GB virtual machine with device passthrough is extremely slow,
taking almost an hour.
Long term GUP cannot allocate memory from CMA area, so a maximum of 16 GB
of no-CMA memory on a NUMA node can be used as virtual machine memory.
There is 16GB of free CMA memory on a NUMA node, which is sufficient to
pass the order-0 watermark check, causing the __compaction_suitable()
function to consistently return true.
For costly allocations, if the __compaction_suitable() function always
returns true, it causes the __alloc_pages_slowpath() function to fail to
exit at the appropriate point. This prevents timely fallback to
allocating memory on other nodes, ultimately resulting in excessively long
virtual machine startup times.
Call trace:
__alloc_pages_slowpath
if (compact_result == COMPACT_SKIPPED ||
compact_result == COMPACT_DEFERRED)
goto nopage; // should exit __alloc_pages_slowpath() from here
We could use the real unmovable allocation context to have
__zone_watermark_unusable_free() subtract CMA pages, and thus we won't
pass the order-0 check anymore once the non-CMA part is exhausted. There
is some risk that in some different scenario the compaction could in fact
migrate pages from the exhausted non-CMA part of the zone to the CMA part
and succeed, and we'll skip it instead. But only __GFP_NORETRY
allocations should be affected in the immediate "goto nopage" when
compaction is skipped, others will attempt with DEF_COMPACT_PRIORITY
anyway and won't fail without trying to compact-migrate the non-CMA
pageblocks into CMA pageblocks first, so it should be fine.
After this fix, it only takes a few tens of seconds to start a 32GB
virtual machine with device passthrough functionality.
Link: https://lore.kernel.org/lkml/1736335854-548-1-git-send-email-yangge1116@126.com/
Link: https://lkml.kernel.org/r/1737788037-8439-1-git-send-email-yangge1116@126.com
Signed-off-by: yangge <yangge1116@126.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Barry Song <21cnbao@gmail.com>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
If a memory allocation fails during dup_mmap(), the maple tree can be left
in an unsafe state for other iterators besides the exit path. All the
locks are dropped before the exit_mmap() call (in mm/mmap.c), but the
incomplete mm_struct can be reached through (at least) the rmap finding
the vmas which have a pointer back to the mm_struct.
Up to this point, there have been no issues with being able to find an
mm_struct that was only partially initialised. Syzbot was able to make
the incomplete mm_struct fail with recent forking changes, so it has been
proven unsafe to use the mm_struct that hasn't been initialised, as
referenced in the link below.
Although 8ac662f5da19f ("fork: avoid inappropriate uprobe access to
invalid mm") fixed the uprobe access, it does not completely remove the
race.
This patch sets the MMF_OOM_SKIP to avoid the iteration of the vmas on the
oom side (even though this is extremely unlikely to be selected as an oom
victim in the race window), and sets MMF_UNSTABLE to avoid other potential
users from using a partially initialised mm_struct.
When registering vmas for uprobe, skip the vmas in an mm that is marked
unstable. Modifying a vma in an unstable mm may cause issues if the mm
isn't fully initialised.
Link: https://lore.kernel.org/all/6756d273.050a0220.2477f.003d.GAE@google.com/
Link: https://lkml.kernel.org/r/20250127170221.1761366-1-Liam.Howlett@oracle.com
Fixes: d24062914837 ("fork: use __mt_dup() to duplicate maple tree in dup_mmap()")
Signed-off-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Jann Horn <jannh@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Peng Zhang <zhangpeng.00@bytedance.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Handle more gracefully cases where no SRAT information is available, like
in VMs with no Numa support, and allow fake-numa configuration to complete
successfully in these cases
Link: https://lkml.kernel.org/r/20250127171623.1523171-1-bfaccini@nvidia.com
Fixes: 63db8170bf34 (“mm/fake-numa: allow later numa node hotplug”)
Signed-off-by: Bruno Faccini <bfaccini@nvidia.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hyeonggon Yoo <hyeonggon.yoo@sk.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Len Brown <lenb@kernel.org>
Cc: "Mike Rapoport (IBM)" <rppt@kernel.org>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Memblock allocations are registered by kmemleak separately, based on their
physical address. During the scanning stage, it checks whether an object
is within the min_low_pfn and max_low_pfn boundaries and ignores it
otherwise.
With the recent addition of __percpu pointer leak detection (commit
6c99d4eb7c5e ("kmemleak: enable tracking for percpu pointers")), kmemleak
started reporting leaks in setup_zone_pageset() and
setup_per_cpu_pageset(). These were caused by the node_data[0] object
(initialised in alloc_node_data()) ending on the PFN_PHYS(max_low_pfn)
boundary. The non-strict upper boundary check introduced by commit
84c326299191 ("mm: kmemleak: check physical address when scan") causes the
pg_data_t object to be ignored (not scanned) and the __percpu pointers it
contains to be reported as leaks.
Make the max_low_pfn upper boundary check strict when deciding whether to
ignore a physical address object and not scan it.
Link: https://lkml.kernel.org/r/20250127184233.2974311-1-catalin.marinas@arm.com
Fixes: 84c326299191 ("mm: kmemleak: check physical address when scan")
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Reported-by: Jakub Kicinski <kuba@kernel.org>
Tested-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Cc: Patrick Wang <patrick.wang.shcn@gmail.com>
Cc: <stable@vger.kernel.org> [6.0.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Map my previous work email to my current one.
Link: https://lkml.kernel.org/r/20250120205659.139027-1-hamzamahfooz@linux.microsoft.com
Signed-off-by: Hamza Mahfooz <hamzamahfooz@linux.microsoft.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Hans verkuil <hverkuil@xs4all.nl>
Cc: Matthieu Baerts <matttbe@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Moving to a linux.dev email address.
Link: https://lkml.kernel.org/r/20250123231344.817358-1-yosry.ahmed@linux.dev
Signed-off-by: Yosry Ahmed <yosry.ahmed@linux.dev>
Cc: Chengming Zhou <chengming.zhou@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
At least recent gdb releases (seen with 14.2) return SP_EL0 as signed long
which lets the right-shift always return 0.
Link: https://lkml.kernel.org/r/dcd2fabc-9131-4b48-8419-6444e2d67454@siemens.com
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Barry Song <baohua@kernel.org>
Cc: Kieran Bingham <kbingham@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
In shrink_folio_list(), demote_folio_list() can be called 2 times.
Currently stat->nr_demoted will only store the last nr_demoted( the later
nr_demoted is always zero, the former nr_demoted will get lost), as a
result number of demoted pages is not accurate.
Accumulate the nr_demoted count across multiple calls to
demote_folio_list(), ensuring accurate reporting of demotion statistics.
[lizhijian@fujitsu.com: introduce local nr_demoted to fix nr_reclaimed double counting]
Link: https://lkml.kernel.org/r/20250111015253.425693-1-lizhijian@fujitsu.com
Link: https://lkml.kernel.org/r/20250110122133.423481-1-lizhijian@fujitsu.com
Fixes: f77f0c751478 ("mm,memcg: provide per-cgroup counters for NUMA balancing operations")
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
Acked-by: Kaiyang Zhao <kaiyang2@cs.cmu.edu>
Tested-by: Donet Tom <donettom@linux.ibm.com>
Reviewed-by: Donet Tom <donettom@linux.ibm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Commit 23aab037106d ("ocfs2: fix UBSAN warning in ocfs2_verify_volume()")
introduced a regression bug. The blksz_bits value is already converted to
CPU endian in the previous code; therefore, the code shouldn't use
le32_to_cpu() anymore.
Link: https://lkml.kernel.org/r/20250121112204.12834-1-heming.zhao@suse.com
Fixes: 23aab037106d ("ocfs2: fix UBSAN warning in ocfs2_verify_volume()")
Signed-off-by: Heming Zhao <heming.zhao@suse.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Jun Piao <piaojun@huawei.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
Commit c1b3bb73d55e ("mm/zsmalloc: use zpdesc in
trylock_zspage()/lock_zspage()") introduces is_first_zpdesc() function.
However, the function is only used when CONFIG_DEBUG_VM=y.
When building with LLVM=1 and W=1 option, the following warning is
generated:
$ make -j12 W=1 LLVM=1 mm/zsmalloc.o
mm/zsmalloc.c:455:20: error: function 'is_first_zpdesc' is not needed and will not be emitted [-Werror,-Wunneeded-internal-declaration]
455 | static inline bool is_first_zpdesc(struct zpdesc *zpdesc)
| ^~~~~~~~~~~~~~~
1 error generated.
Fix the warning by adding __maybe_unused attribute to the function.
No functional change intended.
Link: https://lkml.kernel.org/r/20250127231631.4363-1-42.hyeyoo@gmail.com
Fixes: c1b3bb73d55e ("mm/zsmalloc: use zpdesc in trylock_zspage()/lock_zspage()")
Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202501240958.4ILzuBrH-lkp@intel.com/
Cc: Alex Shi <alexs@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
This fixes the following hard lockup in isolate_lru_folios() during memory
reclaim. If the LRU mostly contains ineligible folios this may trigger
watchdog.
watchdog: Watchdog detected hard LOCKUP on cpu 173
RIP: 0010:native_queued_spin_lock_slowpath+0x255/0x2a0
Call Trace:
_raw_spin_lock_irqsave+0x31/0x40
folio_lruvec_lock_irqsave+0x5f/0x90
folio_batch_move_lru+0x91/0x150
lru_add_drain_per_cpu+0x1c/0x40
process_one_work+0x17d/0x350
worker_thread+0x27b/0x3a0
kthread+0xe8/0x120
ret_from_fork+0x34/0x50
ret_from_fork_asm+0x1b/0x30
lruvec->lru_lock owner：
PID: 2865 TASK: ffff888139214d40 CPU: 40 COMMAND: "kswapd0"
#0 [fffffe0000945e60] crash_nmi_callback at ffffffffa567a555
#1 [fffffe0000945e68] nmi_handle at ffffffffa563b171
#2 [fffffe0000945eb0] default_do_nmi at ffffffffa6575920
#3 [fffffe0000945ed0] exc_nmi at ffffffffa6575af4
#4 [fffffe0000945ef0] end_repeat_nmi at ffffffffa6601dde
[exception RIP: isolate_lru_folios+403]
RIP: ffffffffa597df53 RSP: ffffc90006fb7c28 RFLAGS: 00000002
RAX: 0000000000000001 RBX: ffffc90006fb7c60 RCX: ffffea04a2196f88
RDX: ffffc90006fb7c60 RSI: ffffc90006fb7c60 RDI: ffffea04a2197048
RBP: ffff88812cbd3010 R8: ffffea04a2197008 R9: 0000000000000001
R10: 0000000000000000 R11: 0000000000000001 R12: ffffea04a2197008
R13: ffffea04a2197048 R14: ffffc90006fb7de8 R15: 0000000003e3e937
ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0018
<NMI exception stack>
#5 [ffffc90006fb7c28] isolate_lru_folios at ffffffffa597df53
#6 [ffffc90006fb7cf8] shrink_active_list at ffffffffa597f788
#7 [ffffc90006fb7da8] balance_pgdat at ffffffffa5986db0
#8 [ffffc90006fb7ec0] kswapd at ffffffffa5987354
#9 [ffffc90006fb7ef8] kthread at ffffffffa5748238
crash>
Scenario:
User processe are requesting a large amount of memory and keep page active.
Then a module continuously requests memory from ZONE_DMA32 area.
Memory reclaim will be triggered due to ZONE_DMA32 watermark alarm reached.
However pages in the LRU(active_anon) list are mostly from
the ZONE_NORMAL area.
Reproduce:
Terminal 1: Construct to continuously increase pages active(anon).
mkdir /tmp/memory
mount -t tmpfs -o size=1024000M tmpfs /tmp/memory
dd if=/dev/zero of=/tmp/memory/block bs=4M
tail /tmp/memory/block
Terminal 2:
vmstat -a 1
active will increase.
procs ---memory--- ---swap-- ---io---- -system-- ---cpu--- ...
r b swpd free inact active si so bi bo
1 0 0 1445623076 45898836 83646008 0 0 0
1 0 0 1445623076 43450228 86094616 0 0 0
1 0 0 1445623076 41003480 88541364 0 0 0
1 0 0 1445623076 38557088 90987756 0 0 0
1 0 0 1445623076 36109688 93435156 0 0 0
1 0 0 1445619552 33663256 95881632 0 0 0
1 0 0 1445619804 31217140 98327792 0 0 0
1 0 0 1445619804 28769988 100774944 0 0 0
1 0 0 1445619804 26322348 103222584 0 0 0
1 0 0 1445619804 23875592 105669340 0 0 0
cat /proc/meminfo | head
Active(anon) increase.
MemTotal: 1579941036 kB
MemFree: 1445618500 kB
MemAvailable: 1453013224 kB
Buffers: 6516 kB
Cached: 128653956 kB
SwapCached: 0 kB
Active: 118110812 kB
Inactive: 11436620 kB
Active(anon): 115345744 kB
Inactive(anon): 945292 kB
When the Active(anon) is 115345744 kB, insmod module triggers
the ZONE_DMA32 watermark.
perf record -e vmscan:mm_vmscan_lru_isolate -aR
perf script
isolate_mode=0 classzone=1 order=1 nr_requested=32 nr_scanned=2
nr_skipped=2 nr_taken=0 lru=active_anon
isolate_mode=0 classzone=1 order=1 nr_requested=32 nr_scanned=0
nr_skipped=0 nr_taken=0 lru=active_anon
isolate_mode=0 classzone=1 order=0 nr_requested=32 nr_scanned=28835844
nr_skipped=28835844 nr_taken=0 lru=active_anon
isolate_mode=0 classzone=1 order=1 nr_requested=32 nr_scanned=28835844
nr_skipped=28835844 nr_taken=0 lru=active_anon
isolate_mode=0 classzone=1 order=0 nr_requested=32 nr_scanned=29
nr_skipped=29 nr_taken=0 lru=active_anon
isolate_mode=0 classzone=1 order=0 nr_requested=32 nr_scanned=0
nr_skipped=0 nr_taken=0 lru=active_anon
See nr_scanned=28835844.
28835844 * 4k = 115343376KB approximately equal to 115345744 kB.
If increase Active(anon) to 1000G then insmod module triggers
the ZONE_DMA32 watermark. hard lockup will occur.
In my device nr_scanned = 0000000003e3e937 when hard lockup.
Convert to memory size 0x0000000003e3e937 * 4KB = 261072092 KB.
[ffffc90006fb7c28] isolate_lru_folios at ffffffffa597df53
ffffc90006fb7c30: 0000000000000020 0000000000000000
ffffc90006fb7c40: ffffc90006fb7d40 ffff88812cbd3000
ffffc90006fb7c50: ffffc90006fb7d30 0000000106fb7de8
ffffc90006fb7c60: ffffea04a2197008 ffffea0006ed4a48
ffffc90006fb7c70: 0000000000000000 0000000000000000
ffffc90006fb7c80: 0000000000000000 0000000000000000
ffffc90006fb7c90: 0000000000000000 0000000000000000
ffffc90006fb7ca0: 0000000000000000 0000000003e3e937
ffffc90006fb7cb0: 0000000000000000 0000000000000000
ffffc90006fb7cc0: 8d7c0b56b7874b00 ffff88812cbd3000
About the Fixes:
Why did it take eight years to be discovered?
The problem requires the following conditions to occur:
1. The device memory should be large enough.
2. Pages in the LRU(active_anon) list are mostly from the ZONE_NORMAL area.
3. The memory in ZONE_DMA32 needs to reach the watermark.
If the memory is not large enough, or if the usage design of ZONE_DMA32
area memory is reasonable, this problem is difficult to detect.
notes:
The problem is most likely to occur in ZONE_DMA32 and ZONE_NORMAL,
but other suitable scenarios may also trigger the problem.
Link: https://lkml.kernel.org/r/20241119060842.274072-1-liuye@kylinos.cn
Fixes: b2e18757f2c9 ("mm, vmscan: begin reclaiming pages on a per-node basis")
Signed-off-by: liuye <liuye@kylinos.cn>
Cc: Hugh Dickins <hughd@google.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Yang Shi <yang@os.amperecomputing.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
|
|
If CONFIG_I2C=n:
WARNING: unmet direct dependencies detected for SND_SOC_AK4642
Depends on [n]: SOUND [=y] && SND [=y] && SND_SOC [=y] && I2C [=n]
Selected by [y]:
- SH_7724_SOLUTION_ENGINE [=y] && CPU_SUBTYPE_SH7724 [=y] && SND_SIMPLE_CARD [=y]
WARNING: unmet direct dependencies detected for SND_SOC_DA7210
Depends on [n]: SOUND [=y] && SND [=y] && SND_SOC [=y] && SND_SOC_I2C_AND_SPI [=n]
Selected by [y]:
- SH_ECOVEC [=y] && CPU_SUBTYPE_SH7724 [=y] && SND_SIMPLE_CARD [=y]
Fix this by replacing select by imply, instead of adding a dependency on
I2C.
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202501240836.OvXqmANX-lkp@intel.com/
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Signed-off-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
|
Eric Biggers
Linus Torvalds
Len Brown
Andrew Morton
Tamir Duberstein
Carlos Bilbao
Ritesh Harjani (IBM)
Zhaoyang Huang
Kairui Song
Christopher Obbard
Marco Elver
Nikita Zhandarovich
yangge
Liam R. Howlett
Bruno Faccini
Catalin Marinas
Hamza Mahfooz
Yosry Ahmed
Jan Kiszka
Li Zhijian
Heming Zhao
Hyeonggon Yoo
liuye
Geert Uytterhoeven