From 0d6526f82c3cdefcca47f73f5fc08dc6f335eac6 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:22 -0700 Subject: sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling bug alb_break_llc() decides whether to break LLC preference during active load balance. It does so by testing that every runnable fair task on the source rq prefers its LLC: env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable But the two counters cover different sets. nr_pref_llc_running is updated in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued, so it follows queued tasks. h_nr_runnable is updated in set_delayed()/ clear_delayed() and drops delay-dequeued tasks. So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks, alb_break_llc() returns false, and active balance is free to pull a task off its preferred LLC. Active balance only moves runnable tasks, and this is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB skips the per-task test in can_migrate_task(). The runnable set is the one we want. Fix it on the counter side. A task should be counted in nr_pref_llc_running exactly while it is both queued on its preferred LLC (pref_llc_queued) and runnable (!sched_delayed). Define that membership once in task_pref_llc_runnable(), and adjust the counter only through pref_llc_running_inc()/pref_llc_running_dec() from the four sites that change either input: account_llc_enqueue(), account_llc_dequeue(), set_delayed() and clear_delayed(). Gating every update on the same predicate keeps the delay, wake and dequeue paths from double-counting or underflowing; see the comments at those sites for the ordering. nr_llc_running and sd->llc_counts are not touched and stay on queued semantics. Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC") Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/ Reported-by: Zhan Xusheng Suggested-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Kayra Cizmeci Cc: # v7.2.x Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com --- scripts/elf-parse.h | 305 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 305 insertions(+) create mode 100644 scripts/elf-parse.h (limited to 'scripts/elf-parse.h') diff --git a/scripts/elf-parse.h b/scripts/elf-parse.h new file mode 100644 index 000000000..f4411e030 --- /dev/null +++ b/scripts/elf-parse.h @@ -0,0 +1,305 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +#ifndef _SCRIPTS_ELF_PARSE_H +#define _SCRIPTS_ELF_PARSE_H + +#include + +#include +#include + +typedef union { + Elf32_Ehdr e32; + Elf64_Ehdr e64; +} Elf_Ehdr; + +typedef union { + Elf32_Shdr e32; + Elf64_Shdr e64; +} Elf_Shdr; + +typedef union { + Elf32_Sym e32; + Elf64_Sym e64; +} Elf_Sym; + +typedef union { + Elf32_Rela e32; + Elf64_Rela e64; +} Elf_Rela; + +struct elf_funcs { + int (*compare_extable)(const void *a, const void *b); + uint64_t (*ehdr_shoff)(Elf_Ehdr *ehdr); + uint16_t (*ehdr_shstrndx)(Elf_Ehdr *ehdr); + uint16_t (*ehdr_shentsize)(Elf_Ehdr *ehdr); + uint16_t (*ehdr_shnum)(Elf_Ehdr *ehdr); + uint64_t (*shdr_addr)(Elf_Shdr *shdr); + uint64_t (*shdr_offset)(Elf_Shdr *shdr); + uint64_t (*shdr_size)(Elf_Shdr *shdr); + uint64_t (*shdr_entsize)(Elf_Shdr *shdr); + uint32_t (*shdr_link)(Elf_Shdr *shdr); + uint32_t (*shdr_name)(Elf_Shdr *shdr); + uint32_t (*shdr_type)(Elf_Shdr *shdr); + uint8_t (*sym_type)(Elf_Sym *sym); + uint32_t (*sym_name)(Elf_Sym *sym); + uint64_t (*sym_value)(Elf_Sym *sym); + uint16_t (*sym_shndx)(Elf_Sym *sym); + uint64_t (*rela_offset)(Elf_Rela *rela); + uint64_t (*rela_info)(Elf_Rela *rela); + uint64_t (*rela_addend)(Elf_Rela *rela); + void (*rela_write_addend)(Elf_Rela *rela, uint64_t val); + uint32_t (*r)(const uint32_t *); + uint16_t (*r2)(const uint16_t *); + uint64_t (*r8)(const uint64_t *); + void (*w)(uint32_t, uint32_t *); + void (*w8)(uint64_t, uint64_t *); +}; + +extern struct elf_funcs elf_parser; + +static inline uint64_t ehdr64_shoff(Elf_Ehdr *ehdr) +{ + return elf_parser.r8(&ehdr->e64.e_shoff); +} + +static inline uint64_t ehdr32_shoff(Elf_Ehdr *ehdr) +{ + return elf_parser.r(&ehdr->e32.e_shoff); +} + +static inline uint64_t ehdr_shoff(Elf_Ehdr *ehdr) +{ + return elf_parser.ehdr_shoff(ehdr); +} + +#define EHDR_HALF(fn_name) \ +static inline uint16_t ehdr64_##fn_name(Elf_Ehdr *ehdr) \ +{ \ + return elf_parser.r2(&ehdr->e64.e_##fn_name); \ +} \ + \ +static inline uint16_t ehdr32_##fn_name(Elf_Ehdr *ehdr) \ +{ \ + return elf_parser.r2(&ehdr->e32.e_##fn_name); \ +} \ + \ +static inline uint16_t ehdr_##fn_name(Elf_Ehdr *ehdr) \ +{ \ + return elf_parser.ehdr_##fn_name(ehdr); \ +} + +EHDR_HALF(shentsize) +EHDR_HALF(shstrndx) +EHDR_HALF(shnum) + +#define SHDR_WORD(fn_name) \ +static inline uint32_t shdr64_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.r(&shdr->e64.sh_##fn_name); \ +} \ + \ +static inline uint32_t shdr32_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.r(&shdr->e32.sh_##fn_name); \ +} \ + \ +static inline uint32_t shdr_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.shdr_##fn_name(shdr); \ +} + +#define SHDR_ADDR(fn_name) \ +static inline uint64_t shdr64_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.r8(&shdr->e64.sh_##fn_name); \ +} \ + \ +static inline uint64_t shdr32_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.r(&shdr->e32.sh_##fn_name); \ +} \ + \ +static inline uint64_t shdr_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.shdr_##fn_name(shdr); \ +} + +#define SHDR_WORD(fn_name) \ +static inline uint32_t shdr64_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.r(&shdr->e64.sh_##fn_name); \ +} \ + \ +static inline uint32_t shdr32_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.r(&shdr->e32.sh_##fn_name); \ +} \ +static inline uint32_t shdr_##fn_name(Elf_Shdr *shdr) \ +{ \ + return elf_parser.shdr_##fn_name(shdr); \ +} + +SHDR_ADDR(addr) +SHDR_ADDR(offset) +SHDR_ADDR(size) +SHDR_ADDR(entsize) + +SHDR_WORD(link) +SHDR_WORD(name) +SHDR_WORD(type) + +#define SYM_ADDR(fn_name) \ +static inline uint64_t sym64_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.r8(&sym->e64.st_##fn_name); \ +} \ + \ +static inline uint64_t sym32_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.r(&sym->e32.st_##fn_name); \ +} \ + \ +static inline uint64_t sym_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.sym_##fn_name(sym); \ +} + +#define SYM_WORD(fn_name) \ +static inline uint32_t sym64_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.r(&sym->e64.st_##fn_name); \ +} \ + \ +static inline uint32_t sym32_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.r(&sym->e32.st_##fn_name); \ +} \ + \ +static inline uint32_t sym_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.sym_##fn_name(sym); \ +} + +#define SYM_HALF(fn_name) \ +static inline uint16_t sym64_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.r2(&sym->e64.st_##fn_name); \ +} \ + \ +static inline uint16_t sym32_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.r2(&sym->e32.st_##fn_name); \ +} \ + \ +static inline uint16_t sym_##fn_name(Elf_Sym *sym) \ +{ \ + return elf_parser.sym_##fn_name(sym); \ +} + +static inline uint8_t sym64_type(Elf_Sym *sym) +{ + return ELF64_ST_TYPE(sym->e64.st_info); +} + +static inline uint8_t sym32_type(Elf_Sym *sym) +{ + return ELF32_ST_TYPE(sym->e32.st_info); +} + +static inline uint8_t sym_type(Elf_Sym *sym) +{ + return elf_parser.sym_type(sym); +} + +SYM_ADDR(value) +SYM_WORD(name) +SYM_HALF(shndx) + +#define __maybe_unused __attribute__((__unused__)) + +#define RELA_ADDR(fn_name) \ +static inline uint64_t rela64_##fn_name(Elf_Rela *rela) \ +{ \ + return elf_parser.r8((uint64_t *)&rela->e64.r_##fn_name); \ +} \ + \ +static inline uint64_t rela32_##fn_name(Elf_Rela *rela) \ +{ \ + return elf_parser.r((uint32_t *)&rela->e32.r_##fn_name); \ +} \ + \ +static inline uint64_t __maybe_unused rela_##fn_name(Elf_Rela *rela) \ +{ \ + return elf_parser.rela_##fn_name(rela); \ +} + +RELA_ADDR(offset) +RELA_ADDR(info) +RELA_ADDR(addend) + +static inline void rela64_write_addend(Elf_Rela *rela, uint64_t val) +{ + elf_parser.w8(val, (uint64_t *)&rela->e64.r_addend); +} + +static inline void rela32_write_addend(Elf_Rela *rela, uint64_t val) +{ + elf_parser.w(val, (uint32_t *)&rela->e32.r_addend); +} + +static inline uint32_t rbe(const uint32_t *x) +{ + return get_unaligned_be32(x); +} + +static inline uint16_t r2be(const uint16_t *x) +{ + return get_unaligned_be16(x); +} + +static inline uint64_t r8be(const uint64_t *x) +{ + return get_unaligned_be64(x); +} + +static inline uint32_t rle(const uint32_t *x) +{ + return get_unaligned_le32(x); +} + +static inline uint16_t r2le(const uint16_t *x) +{ + return get_unaligned_le16(x); +} + +static inline uint64_t r8le(const uint64_t *x) +{ + return get_unaligned_le64(x); +} + +static inline void wbe(uint32_t val, uint32_t *x) +{ + put_unaligned_be32(val, x); +} + +static inline void wle(uint32_t val, uint32_t *x) +{ + put_unaligned_le32(val, x); +} + +static inline void w8be(uint64_t val, uint64_t *x) +{ + put_unaligned_be64(val, x); +} + +static inline void w8le(uint64_t val, uint64_t *x) +{ + put_unaligned_le64(val, x); +} + +void *elf_map(char const *fname, size_t *size, uint32_t types); +void elf_unmap(void *addr, size_t size); +int elf_map_machine(void *addr); +int elf_map_long_size(void *addr); + +#endif /* _SCRIPTS_ELF_PARSE_H */ -- cgit v1.3.1