From 74a2bd0bfb0675cb8f276f2fe3ede0c7f9e78fa3 Mon Sep 17 00:00:00 2001 From: Yury Norov Date: Wed, 12 Mar 2025 22:19:49 -0400 Subject: nodemask: drop nodes_shift nodes_shift_{left,right} are not used. Drop them. Signed-off-by: Yury Norov [NVIDIA] --- include/linux/nodemask.h | 19 ------------------- 1 file changed, 19 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nodemask.h b/include/linux/nodemask.h index f0ac0633366b..0aa6b63aa747 100644 --- a/include/linux/nodemask.h +++ b/include/linux/nodemask.h @@ -39,9 +39,6 @@ * int nodes_full(mask) Is mask full (all bits sets)? * int nodes_weight(mask) Hamming weight - number of set bits * - * void nodes_shift_right(dst, src, n) Shift right - * void nodes_shift_left(dst, src, n) Shift left - * * unsigned int first_node(mask) Number lowest set bit, or MAX_NUMNODES * unsigend int next_node(node, mask) Next node past 'node', or MAX_NUMNODES * unsigned int next_node_in(node, mask) Next node past 'node', or wrap to first, @@ -247,22 +244,6 @@ static __always_inline int __nodes_weight(const nodemask_t *srcp, unsigned int n return bitmap_weight(srcp->bits, nbits); } -#define nodes_shift_right(dst, src, n) \ - __nodes_shift_right(&(dst), &(src), (n), MAX_NUMNODES) -static __always_inline void __nodes_shift_right(nodemask_t *dstp, - const nodemask_t *srcp, int n, int nbits) -{ - bitmap_shift_right(dstp->bits, srcp->bits, n, nbits); -} - -#define nodes_shift_left(dst, src, n) \ - __nodes_shift_left(&(dst), &(src), (n), MAX_NUMNODES) -static __always_inline void __nodes_shift_left(nodemask_t *dstp, - const nodemask_t *srcp, int n, int nbits) -{ - bitmap_shift_left(dstp->bits, srcp->bits, n, nbits); -} - /* FIXME: better would be to fix all architectures to never return > MAX_NUMNODES, then the silly min_ts could be dropped. */ -- cgit v1.2.3 From 4923c2c5b66fe8dea1df5e16d61f15c1dbea5ba1 Mon Sep 17 00:00:00 2001 From: Yury Norov Date: Wed, 12 Mar 2025 22:19:50 -0400 Subject: cpumask: add non-atomic __assign_cpu() Similarly to atomic, add a non-atomic version. Signed-off-by: Yury Norov [NVIDIA] --- include/linux/cpumask.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h index f9a868384083..ba1e232e0200 100644 --- a/include/linux/cpumask.h +++ b/include/linux/cpumask.h @@ -1074,6 +1074,9 @@ void init_cpu_possible(const struct cpumask *src); #define assign_cpu(cpu, mask, val) \ assign_bit(cpumask_check(cpu), cpumask_bits(mask), (val)) +#define __assign_cpu(cpu, mask, val) \ + __assign_bit(cpumask_check(cpu), cpumask_bits(mask), (val)) + #define set_cpu_possible(cpu, possible) assign_cpu((cpu), &__cpu_possible_mask, (possible)) #define set_cpu_enabled(cpu, enabled) assign_cpu((cpu), &__cpu_enabled_mask, (enabled)) #define set_cpu_present(cpu, present) assign_cpu((cpu), &__cpu_present_mask, (present)) -- cgit v1.2.3 From 791a9b25ce2e6ecbe404ee32eed8a96a17e52896 Mon Sep 17 00:00:00 2001 From: Yury Norov Date: Wed, 12 Mar 2025 22:19:52 -0400 Subject: cpumask: drop cpumask_assign_cpu() Commit decde1fa209323c7 ("cpumask: Add assign cpu") was merged bypassing cpumasks reviewers. It adds atomic and non-atomic cpumask_assign_cpu() helpers. In the same merge window, commit 5c563ee90a22d3 ("cpumask: introduce assign_cpu() macro") added the same functionality. So now we have it duplicated. __cpumask_assign_cpu() has never been used since introducing, and because this series reworks the only user of cpumask_assign_cpu(), both functions become a dead code. Signed-off-by: Yury Norov [NVIDIA] --- include/linux/cpumask.h | 16 ---------------- 1 file changed, 16 deletions(-) (limited to 'include/linux') diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h index ba1e232e0200..beff4d26e605 100644 --- a/include/linux/cpumask.h +++ b/include/linux/cpumask.h @@ -558,22 +558,6 @@ static __always_inline void __cpumask_clear_cpu(int cpu, struct cpumask *dstp) __clear_bit(cpumask_check(cpu), cpumask_bits(dstp)); } -/** - * cpumask_assign_cpu - assign a cpu in a cpumask - * @cpu: cpu number (< nr_cpu_ids) - * @dstp: the cpumask pointer - * @bool: the value to assign - */ -static __always_inline void cpumask_assign_cpu(int cpu, struct cpumask *dstp, bool value) -{ - assign_bit(cpumask_check(cpu), cpumask_bits(dstp), value); -} - -static __always_inline void __cpumask_assign_cpu(int cpu, struct cpumask *dstp, bool value) -{ - __assign_bit(cpumask_check(cpu), cpumask_bits(dstp), value); -} - /** * cpumask_test_cpu - test for a cpu in a cpumask * @cpu: cpu number (< nr_cpu_ids) -- cgit v1.2.3 From 31299a5e0211241171b2222c5633aad4763bf700 Mon Sep 17 00:00:00 2001 From: Vincent Mailhol Date: Wed, 26 Mar 2025 00:59:56 +0900 Subject: bits: add comments and newlines to #if, #else and #endif directives This is a preparation for the upcoming GENMASK_U*() and BIT_U*() changes. After introducing those new macros, there will be a lot of scrolling between the #if, #else and #endif. Add a comment to the #else and #endif preprocessor macros to help keep track of which context we are in. Also, add new lines to better visually separate the non-asm and asm sections. Signed-off-by: Vincent Mailhol Reviewed-by: Andy Shevchenko Signed-off-by: Yury Norov --- include/linux/bits.h | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/bits.h b/include/linux/bits.h index 14fd0ca9a6cd..e1e517769140 100644 --- a/include/linux/bits.h +++ b/include/linux/bits.h @@ -19,16 +19,21 @@ * GENMASK_ULL(39, 21) gives us the 64bit vector 0x000000ffffe00000. */ #if !defined(__ASSEMBLY__) + #include #include + #define GENMASK_INPUT_CHECK(h, l) BUILD_BUG_ON_ZERO(const_true((l) > (h))) -#else + +#else /* defined(__ASSEMBLY__) */ + /* * BUILD_BUG_ON_ZERO is not available in h files included from asm files, * disable the input check if that is the case. */ #define GENMASK_INPUT_CHECK(h, l) 0 -#endif + +#endif /* !defined(__ASSEMBLY__) */ #define GENMASK(h, l) \ (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l)) -- cgit v1.2.3 From 19408200c094858d952a90bf4977733dc89a4df5 Mon Sep 17 00:00:00 2001 From: Vincent Mailhol Date: Wed, 26 Mar 2025 00:59:57 +0900 Subject: bits: introduce fixed-type GENMASK_U*() Add GENMASK_TYPE() which generalizes __GENMASK() to support different types, and implement fixed-types versions of GENMASK() based on it. The fixed-type version allows more strict checks to the min/max values accepted, which is useful for defining registers like implemented by i915 and xe drivers with their REG_GENMASK*() macros. The strict checks rely on shift-count-overflow compiler check to fail the build if a number outside of the range allowed is passed. Example: #define FOO_MASK GENMASK_U32(33, 4) will generate a warning like: include/linux/bits.h:51:27: error: right shift count >= width of type [-Werror=shift-count-overflow] 51 | type_max(t) >> (BITS_PER_TYPE(t) - 1 - (h))))) | ^~ The result is casted to the corresponding fixed width type. For example, GENMASK_U8() returns an u8. Note that because of the C promotion rules, GENMASK_U8() and GENMASK_U16() will immediately be promoted to int if used in an expression. Regardless, the main goal is not to get the correct type, but rather to enforce more checks at compile time. While GENMASK_TYPE() is crafted to cover all variants, including the already existing GENMASK(), GENMASK_ULL() and GENMASK_U128(), for the moment, only use it for the newly introduced GENMASK_U*(). The consolidation will be done in a separate change. Co-developed-by: Yury Norov Signed-off-by: Lucas De Marchi Acked-by: Jani Nikula Signed-off-by: Vincent Mailhol Reviewed-by: Andy Shevchenko Signed-off-by: Yury Norov --- include/linux/bitops.h | 1 - include/linux/bits.h | 30 ++++++++++++++++++++++++++++++ 2 files changed, 30 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/bitops.h b/include/linux/bitops.h index c1cb53cf2f0f..9be2d50da09a 100644 --- a/include/linux/bitops.h +++ b/include/linux/bitops.h @@ -8,7 +8,6 @@ #include -#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE) #define BITS_TO_LONGS(nr) __KERNEL_DIV_ROUND_UP(nr, BITS_PER_TYPE(long)) #define BITS_TO_U64(nr) __KERNEL_DIV_ROUND_UP(nr, BITS_PER_TYPE(u64)) #define BITS_TO_U32(nr) __KERNEL_DIV_ROUND_UP(nr, BITS_PER_TYPE(u32)) diff --git a/include/linux/bits.h b/include/linux/bits.h index e1e517769140..9718c5ae5fc3 100644 --- a/include/linux/bits.h +++ b/include/linux/bits.h @@ -12,6 +12,7 @@ #define BIT_ULL_MASK(nr) (ULL(1) << ((nr) % BITS_PER_LONG_LONG)) #define BIT_ULL_WORD(nr) ((nr) / BITS_PER_LONG_LONG) #define BITS_PER_BYTE 8 +#define BITS_PER_TYPE(type) (sizeof(type) * BITS_PER_BYTE) /* * Create a contiguous bitmask starting at bit position @l and ending at @@ -20,11 +21,40 @@ */ #if !defined(__ASSEMBLY__) +/* + * Missing asm support + * + * GENMASK_U*() depend on BITS_PER_TYPE() which relies on sizeof(), + * something not available in asm. Nevertheless, fixed width integers is a C + * concept. Assembly code can rely on the long and long long versions instead. + */ + #include #include +#include #define GENMASK_INPUT_CHECK(h, l) BUILD_BUG_ON_ZERO(const_true((l) > (h))) +/* + * Generate a mask for the specified type @t. Additional checks are made to + * guarantee the value returned fits in that type, relying on + * -Wshift-count-overflow compiler check to detect incompatible arguments. + * For example, all these create build errors or warnings: + * + * - GENMASK(15, 20): wrong argument order + * - GENMASK(72, 15): doesn't fit unsigned long + * - GENMASK_U32(33, 15): doesn't fit in a u32 + */ +#define GENMASK_TYPE(t, h, l) \ + ((t)(GENMASK_INPUT_CHECK(h, l) + \ + (type_max(t) << (l) & \ + type_max(t) >> (BITS_PER_TYPE(t) - 1 - (h))))) + +#define GENMASK_U8(h, l) GENMASK_TYPE(u8, h, l) +#define GENMASK_U16(h, l) GENMASK_TYPE(u16, h, l) +#define GENMASK_U32(h, l) GENMASK_TYPE(u32, h, l) +#define GENMASK_U64(h, l) GENMASK_TYPE(u64, h, l) + #else /* defined(__ASSEMBLY__) */ /* -- cgit v1.2.3 From 5b572e8a9f3dcd6e3ba80ec5b5ec85472c88cb4c Mon Sep 17 00:00:00 2001 From: Lucas De Marchi Date: Wed, 26 Mar 2025 00:59:58 +0900 Subject: bits: introduce fixed-type BIT_U*() Implement fixed-type BIT_U*() to help drivers add stricter checks, like it was done for GENMASK_U*(). Signed-off-by: Lucas De Marchi Acked-by: Jani Nikula Co-developed-by: Vincent Mailhol Signed-off-by: Vincent Mailhol Reviewed-by: Andy Shevchenko Signed-off-by: Yury Norov --- include/linux/bits.h | 20 +++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/bits.h b/include/linux/bits.h index 9718c5ae5fc3..7ad056219115 100644 --- a/include/linux/bits.h +++ b/include/linux/bits.h @@ -24,7 +24,7 @@ /* * Missing asm support * - * GENMASK_U*() depend on BITS_PER_TYPE() which relies on sizeof(), + * GENMASK_U*() and BIT_U*() depend on BITS_PER_TYPE() which relies on sizeof(), * something not available in asm. Nevertheless, fixed width integers is a C * concept. Assembly code can rely on the long and long long versions instead. */ @@ -55,6 +55,24 @@ #define GENMASK_U32(h, l) GENMASK_TYPE(u32, h, l) #define GENMASK_U64(h, l) GENMASK_TYPE(u64, h, l) +/* + * Fixed-type variants of BIT(), with additional checks like GENMASK_TYPE(). The + * following examples generate compiler warnings due to -Wshift-count-overflow: + * + * - BIT_U8(8) + * - BIT_U32(-1) + * - BIT_U32(40) + */ +#define BIT_INPUT_CHECK(type, nr) \ + BUILD_BUG_ON_ZERO(const_true((nr) >= BITS_PER_TYPE(type))) + +#define BIT_TYPE(type, nr) ((type)(BIT_INPUT_CHECK(type, nr) + BIT_ULL(nr))) + +#define BIT_U8(nr) BIT_TYPE(u8, nr) +#define BIT_U16(nr) BIT_TYPE(u16, nr) +#define BIT_U32(nr) BIT_TYPE(u32, nr) +#define BIT_U64(nr) BIT_TYPE(u64, nr) + #else /* defined(__ASSEMBLY__) */ /* -- cgit v1.2.3 From 243c90e917f5cfc99821e5104d1c8a81c11cda4c Mon Sep 17 00:00:00 2001 From: Vincent Mailhol Date: Sat, 29 Mar 2025 01:48:50 +0900 Subject: build_bug.h: more user friendly error messages in BUILD_BUG_ON_ZERO() __BUILD_BUG_ON_ZERO_MSG(), as introduced in [1], makes it possible to do a static assertions in expressions. The direct benefit is to provide a meaningful error message instead of the cryptic negative bitfield size error message currently returned by BUILD_BUG_ON_ZERO(): ./include/linux/build_bug.h:16:51: error: negative width in bit-field '' 16 | #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); }))) | ^ Get rid of BUILD_BUG_ON_ZERO()'s bitfield size hack. Instead rely on __BUILD_BUG_ON_ZERO_MSG() which in turn relies on C11's _Static_assert(). Use some macro magic, similarly to static_assert(), to either use an optional error message provided by the user or, when omitted, to produce a default error message by stringifying the tested expression. With this, for example: BUILD_BUG_ON_ZERO(1 > 0) would now throw: ./include/linux/compiler.h:197:62: error: static assertion failed: "1 > 0 is true" 197 | define __BUILD_BUG_ON_ZERO_MSG(e, msg) ((int)sizeof(struct {_Static_assert(!(e), msg);})) | ^~~~~~~~~~~~~~ Finally, __BUILD_BUG_ON_ZERO_MSG() is already guarded by an: #ifdef __CHECKER__ So no need any more for that guard clause for BUILD_BUG_ON_ZERO(). Remove it. [1] commit d7a516c6eeae ("compiler.h: Fix undefined BUILD_BUG_ON_ZERO()") Link: https://git.kernel.org/torvalds/c/d7a516c6eeae Signed-off-by: Vincent Mailhol Link: https://git.kernel.org/next/linux-next/c/b88937277df Reviewed-by: Kees Cook Signed-off-by: Yury Norov --- include/linux/build_bug.h | 10 +++++----- include/linux/compiler.h | 4 ++-- 2 files changed, 7 insertions(+), 7 deletions(-) (limited to 'include/linux') diff --git a/include/linux/build_bug.h b/include/linux/build_bug.h index 3aa3640f8c18..2cfbb4c65c78 100644 --- a/include/linux/build_bug.h +++ b/include/linux/build_bug.h @@ -4,17 +4,17 @@ #include -#ifdef __CHECKER__ -#define BUILD_BUG_ON_ZERO(e) (0) -#else /* __CHECKER__ */ /* * Force a compilation error if condition is true, but also produce a * result (of value 0 and type int), so the expression can be used * e.g. in a structure initializer (or where-ever else comma expressions * aren't permitted). + * + * Take an error message as an optional second argument. If omitted, + * default to the stringification of the tested expression. */ -#define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); }))) -#endif /* __CHECKER__ */ +#define BUILD_BUG_ON_ZERO(e, ...) \ + __BUILD_BUG_ON_ZERO_MSG(e, ##__VA_ARGS__, #e " is true") /* Force a compilation error if a constant expression is not a power of 2 */ #define __BUILD_BUG_ON_NOT_POWER_OF_2(n) \ diff --git a/include/linux/compiler.h b/include/linux/compiler.h index 27725f1ab5ab..6f04a1d8c720 100644 --- a/include/linux/compiler.h +++ b/include/linux/compiler.h @@ -192,9 +192,9 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val, }) #ifdef __CHECKER__ -#define __BUILD_BUG_ON_ZERO_MSG(e, msg) (0) +#define __BUILD_BUG_ON_ZERO_MSG(e, msg, ...) (0) #else /* __CHECKER__ */ -#define __BUILD_BUG_ON_ZERO_MSG(e, msg) ((int)sizeof(struct {_Static_assert(!(e), msg);})) +#define __BUILD_BUG_ON_ZERO_MSG(e, msg, ...) ((int)sizeof(struct {_Static_assert(!(e), msg);})) #endif /* __CHECKER__ */ /* &a[0] degrades to a pointer: a different type from an array */ -- cgit v1.2.3 From 99c712d788c46452289beada638476b68b5a773b Mon Sep 17 00:00:00 2001 From: Andy Shevchenko Date: Thu, 17 Apr 2025 19:17:15 +0300 Subject: bitmap-str: Get rid of 'extern' for function prototypes The bitmap-str.h uses mixed style for function prototypes. Drop the 'extern' as it easier to read and makes style aligned with a new code in the kernel. Signed-off-by: Andy Shevchenko Signed-off-by: Yury Norov --- include/linux/bitmap-str.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'include/linux') diff --git a/include/linux/bitmap-str.h b/include/linux/bitmap-str.h index 17caeca94cab..d758b4809a3a 100644 --- a/include/linux/bitmap-str.h +++ b/include/linux/bitmap-str.h @@ -4,10 +4,10 @@ int bitmap_parse_user(const char __user *ubuf, unsigned int ulen, unsigned long *dst, int nbits); int bitmap_print_to_pagebuf(bool list, char *buf, const unsigned long *maskp, int nmaskbits); -extern int bitmap_print_bitmask_to_buf(char *buf, const unsigned long *maskp, - int nmaskbits, loff_t off, size_t count); -extern int bitmap_print_list_to_buf(char *buf, const unsigned long *maskp, - int nmaskbits, loff_t off, size_t count); +int bitmap_print_bitmask_to_buf(char *buf, const unsigned long *maskp, int nmaskbits, + loff_t off, size_t count); +int bitmap_print_list_to_buf(char *buf, const unsigned long *maskp, int nmaskbits, + loff_t off, size_t count); int bitmap_parse(const char *buf, unsigned int buflen, unsigned long *dst, int nbits); int bitmap_parselist(const char *buf, unsigned long *maskp, int nmaskbits); int bitmap_parselist_user(const char __user *ubuf, unsigned int ulen, -- cgit v1.2.3 From 89a44a808814d4717a8bf945ddebd5c39bfffcf4 Mon Sep 17 00:00:00 2001 From: Andy Shevchenko Date: Thu, 17 Apr 2025 19:17:16 +0300 Subject: bitmap-str: Add missing header(s) bitmap-str.h is not self-contained, it uses bool type that is provided in types.h and it uses __user annotation that is guaranteed to be included with types.h. Add missing header(s) to follow IWYU (Include What You Use) principle. Signed-off-by: Andy Shevchenko Signed-off-by: Yury Norov --- include/linux/bitmap-str.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include/linux') diff --git a/include/linux/bitmap-str.h b/include/linux/bitmap-str.h index d758b4809a3a..53d3e1b32d3d 100644 --- a/include/linux/bitmap-str.h +++ b/include/linux/bitmap-str.h @@ -2,6 +2,8 @@ #ifndef __LINUX_BITMAP_STR_H #define __LINUX_BITMAP_STR_H +#include + int bitmap_parse_user(const char __user *ubuf, unsigned int ulen, unsigned long *dst, int nbits); int bitmap_print_to_pagebuf(bool list, char *buf, const unsigned long *maskp, int nmaskbits); int bitmap_print_bitmask_to_buf(char *buf, const unsigned long *maskp, int nmaskbits, -- cgit v1.2.3 From a256ae22570ee4c3427fdc703a58a89afee6a332 Mon Sep 17 00:00:00 2001 From: Luo Jie Date: Thu, 17 Apr 2025 18:47:08 +0800 Subject: bitfield: Add FIELD_MODIFY() helper Add a helper for replacing the contents of bitfield in memory with the specified value. Even though a helper xxx_replace_bits() is available, it is not well documented, and only reports errors at the run time, which will not be helpful to catch possible overflow errors due to incorrect parameter types used. FIELD_MODIFY(REG_FIELD_C, ®, c) is the wrapper to the code below. reg &= ~REG_FIELD_C; reg |= FIELD_PREP(REG_FIELD_C, c); Yury: trim commit message, align backslashes. Signed-off-by: Luo Jie Signed-off-by: Yury Norov --- include/linux/bitfield.h | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/bitfield.h b/include/linux/bitfield.h index 63928f173223..6d9a53db54b6 100644 --- a/include/linux/bitfield.h +++ b/include/linux/bitfield.h @@ -8,6 +8,7 @@ #define _LINUX_BITFIELD_H #include +#include #include /* @@ -38,8 +39,7 @@ * FIELD_PREP(REG_FIELD_D, 0x40); * * Modify: - * reg &= ~REG_FIELD_C; - * reg |= FIELD_PREP(REG_FIELD_C, c); + * FIELD_MODIFY(REG_FIELD_C, ®, c); */ #define __bf_shf(x) (__builtin_ffsll(x) - 1) @@ -156,6 +156,23 @@ (typeof(_mask))(((_reg) & (_mask)) >> __bf_shf(_mask)); \ }) +/** + * FIELD_MODIFY() - modify a bitfield element + * @_mask: shifted mask defining the field's length and position + * @_reg_p: pointer to the memory that should be updated + * @_val: value to store in the bitfield + * + * FIELD_MODIFY() modifies the set of bits in @_reg_p specified by @_mask, + * by replacing them with the bitfield value passed in as @_val. + */ +#define FIELD_MODIFY(_mask, _reg_p, _val) \ + ({ \ + typecheck_pointer(_reg_p); \ + __BF_FIELD_CHECK(_mask, *(_reg_p), _val, "FIELD_MODIFY: "); \ + *(_reg_p) &= ~(_mask); \ + *(_reg_p) |= (((typeof(_mask))(_val) << __bf_shf(_mask)) & (_mask)); \ + }) + extern void __compiletime_error("value doesn't fit into mask") __field_overflow(void); extern void __compiletime_error("bad bitfield mask") -- cgit v1.2.3 From 895ee6a22e3195b7c1fee140c842bdeedb89ed33 Mon Sep 17 00:00:00 2001 From: "Yury Norov [NVIDIA]" Date: Fri, 9 May 2025 12:20:08 -0400 Subject: topology: make for_each_node_with_cpus() O(N) for_each_node_with_cpus() calls nr_cpus_node() at every iteration, which makes it O(N^2). Kernel tracks such nodes with N_CPU record in node_states array. Switching to it makes for_each_node_with_cpus() O(N). Andrea: Now we can include also offline nodes with CPUs assigned (assuming it's possible). If checking the online state is required, the user must use node_online() within the loop. CC: Andrea Righi CC:Tejun Heo Signed-off-by: Yury Norov [NVIDIA] --- include/linux/nodemask.h | 1 + include/linux/topology.h | 5 +---- 2 files changed, 2 insertions(+), 4 deletions(-) (limited to 'include/linux') diff --git a/include/linux/nodemask.h b/include/linux/nodemask.h index 0aa6b63aa747..f08ae71585fa 100644 --- a/include/linux/nodemask.h +++ b/include/linux/nodemask.h @@ -522,6 +522,7 @@ static __always_inline int node_random(const nodemask_t *maskp) #define for_each_node(node) for_each_node_state(node, N_POSSIBLE) #define for_each_online_node(node) for_each_node_state(node, N_ONLINE) +#define for_each_node_with_cpus(node) for_each_node_state(node, N_CPU) /* * For nodemask scratch area. diff --git a/include/linux/topology.h b/include/linux/topology.h index 24e715f0f6d2..ffee6b4a071a 100644 --- a/include/linux/topology.h +++ b/include/linux/topology.h @@ -29,6 +29,7 @@ #include #include +#include #include #include #include @@ -39,10 +40,6 @@ #define nr_cpus_node(node) cpumask_weight(cpumask_of_node(node)) #endif -#define for_each_node_with_cpus(node) \ - for_each_online_node(node) \ - if (nr_cpus_node(node)) - int arch_update_cpu_topology(void); /* Conform to ACPI 2.0 SLIT distance definitions */ -- cgit v1.2.3