From 5febf432df1cfa5b25d99b54c32103fafbdd0eb9 Mon Sep 17 00:00:00 2001 From: Miguel Ojeda Date: Sun, 16 Aug 2026 15:32:33 +0200 Subject: kbuild: rust: keep Rust objects out of Clang LTO with inline helpers Under `CONFIG_LTO_CLANG` + `CONFIG_RUST_INLINE_HELPERS`, one may hit `objtool` errors such as: vmlinux.o: error: objtool: _R..._3Gsp4boot+0xd6a: can't find jump dest instruction at .text._R..._3Gsp4boot+0x1dfd The reason is that in such builds, the Clang invocation that compiles the combined Rust plus helpers bitcode emits LLVM bitcode (again) -- the final code generation happens in the linker's LTO step, which the `-mllvm` trap options passed to Clang do not reach. This, in turn, means that unreachable traps are missing, and the impossible paths do not merely fallthrough to the next symbol, but past the end of their own section, since LTO builds place each function in its own section. Thus filter `CC_FLAGS_LTO` out of the Clang invocation, so that it always emits machine code directly, with the traps in place. Assisted-by: LLM Cc: Gary Guo Cc: Boqun Feng Cc: Alice Ryhl Cc: Matthew Maurer Cc: Josh Poimboeuf Cc: Peter Zijlstra Cc: stable@vger.kernel.org Fixes: 3a2486cc1da5 ("kbuild: rust: provide an option to inline C helpers into Rust") Acked-by: Gary Guo Link: https://patch.msgid.link/20260816133233.197500-2-ojeda@kernel.org Signed-off-by: Miguel Ojeda --- include/net/rps.h | 197 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 197 insertions(+) create mode 100644 include/net/rps.h (limited to 'include/net/rps.h') diff --git a/include/net/rps.h b/include/net/rps.h new file mode 100644 index 000000000..e33c6a2fa --- /dev/null +++ b/include/net/rps.h @@ -0,0 +1,197 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +#ifndef _NET_RPS_H +#define _NET_RPS_H + +#include +#include +#include +#include + +#ifdef CONFIG_RPS +#include + +extern struct static_key_false rps_needed; +extern struct static_key_false rfs_needed; + +/* + * This structure holds an RPS map which can be of variable length. The + * map is an array of CPUs. + */ +struct rps_map { + unsigned int len; + struct rcu_head rcu; + u16 cpus[]; +}; +#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16))) + +/* + * The rps_dev_flow structure contains the mapping of a flow to a CPU, the + * tail pointer for that CPU's input queue at the time of last enqueue, a + * hardware filter index, and the hash of the flow if aRFS is enabled. + */ +struct rps_dev_flow { + u16 cpu; + u16 filter; + unsigned int last_qtail; +#ifdef CONFIG_RFS_ACCEL + u32 hash; +#endif +}; +#define RPS_NO_FILTER 0xffff + +/* + * The rps_sock_flow_table contains mappings of flows to the last CPU + * on which they were processed by the application (set in recvmsg). + * Each entry is a 32bit value. Upper part is the high-order bits + * of flow hash, lower part is CPU number. + * rps_cpu_mask is used to partition the space, depending on number of + * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1 + * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f, + * meaning we use 32-6=26 bits for the hash. + */ +struct rps_sock_flow_table { + u32 ent; +}; + +#define RPS_NO_CPU 0xffff + +static inline void rps_record_sock_flow(rps_tag_ptr tag_ptr, u32 hash) +{ + unsigned int index = hash & rps_tag_to_mask(tag_ptr); + u32 val = hash & ~net_hotdata.rps_cpu_mask; + struct rps_sock_flow_table *table; + + /* We only give a hint, preemption can change CPU under us */ + val |= raw_smp_processor_id(); + + table = rps_tag_to_table(tag_ptr); + /* The following WRITE_ONCE() is paired with the READ_ONCE() + * here, and another one in get_rps_cpu(). + */ + if (READ_ONCE(table[index].ent) != val) + WRITE_ONCE(table[index].ent, val); +} + +static inline void _sock_rps_record_flow_hash(__u32 hash) +{ + rps_tag_ptr tag_ptr; + + if (!hash) + return; + rcu_read_lock(); + tag_ptr = READ_ONCE(net_hotdata.rps_sock_flow_table); + if (tag_ptr) + rps_record_sock_flow(tag_ptr, hash); + rcu_read_unlock(); +} + +static inline void _sock_rps_record_flow(const struct sock *sk) +{ + /* Reading sk->sk_rxhash might incur an expensive cache line + * miss. + * + * TCP_ESTABLISHED does cover almost all states where RFS + * might be useful, and is cheaper [1] than testing : + * IPv4: inet_sk(sk)->inet_daddr + * IPv6: ipv6_addr_any(&sk->sk_v6_daddr) + * OR an additional socket flag + * [1] : sk_state and sk_prot are in the same cache line. + */ + if (sk->sk_state == TCP_ESTABLISHED) { + /* This READ_ONCE() is paired with the WRITE_ONCE() + * from sock_rps_save_rxhash() and sock_rps_reset_rxhash(). + */ + _sock_rps_record_flow_hash(READ_ONCE(sk->sk_rxhash)); + } +} + +static inline void _sock_rps_delete_flow(const struct sock *sk) +{ + struct rps_sock_flow_table *table; + rps_tag_ptr tag_ptr; + u32 hash, index; + + hash = READ_ONCE(sk->sk_rxhash); + if (!hash) + return; + + rcu_read_lock(); + tag_ptr = READ_ONCE(net_hotdata.rps_sock_flow_table); + if (tag_ptr) { + index = hash & rps_tag_to_mask(tag_ptr); + table = rps_tag_to_table(tag_ptr); + if (READ_ONCE(table[index].ent) != RPS_NO_CPU) + WRITE_ONCE(table[index].ent, RPS_NO_CPU); + } + rcu_read_unlock(); +} +#endif /* CONFIG_RPS */ + +static inline bool rfs_is_needed(void) +{ +#ifdef CONFIG_RPS + return static_branch_unlikely(&rfs_needed); +#else + return false; +#endif +} + +static inline void sock_rps_record_flow_hash(__u32 hash) +{ +#ifdef CONFIG_RPS + if (!rfs_is_needed()) + return; + + _sock_rps_record_flow_hash(hash); +#endif +} + +static inline void sock_rps_record_flow(const struct sock *sk) +{ +#ifdef CONFIG_RPS + if (!rfs_is_needed()) + return; + + _sock_rps_record_flow(sk); +#endif +} + +static inline void sock_rps_delete_flow(const struct sock *sk) +{ +#ifdef CONFIG_RPS + if (!rfs_is_needed()) + return; + + _sock_rps_delete_flow(sk); +#endif +} + +static inline u32 rps_input_queue_tail_incr(struct softnet_data *sd) +{ +#ifdef CONFIG_RPS + return ++sd->input_queue_tail; +#else + return 0; +#endif +} + +static inline void rps_input_queue_tail_save(u32 *dest, u32 tail) +{ +#ifdef CONFIG_RPS + WRITE_ONCE(*dest, tail); +#endif +} + +static inline void rps_input_queue_head_add(struct softnet_data *sd, int val) +{ +#ifdef CONFIG_RPS + WRITE_ONCE(sd->input_queue_head, sd->input_queue_head + val); +#endif +} + +static inline void rps_input_queue_head_incr(struct softnet_data *sd) +{ + rps_input_queue_head_add(sd, 1); +} + +#endif /* _NET_RPS_H */ -- cgit v1.3.1