tools/sched_ext: Receive updates from SCX repo

Receive tools/sched_ext updates form https://github.com/sched-ext/scx to
sync userspace bits:

- scx_bpf_dump_header() added which can be used to print out basic scheduler
  info on dump.

- BPF possible/online CPU iterators added.

- CO-RE enums added. The enums are autogenerated from vmlinux.h. Include the
  generated artifacts in tools/sched_ext to keep the Makefile simpler.

- Other misc changes.

Signed-off-by: Tejun Heo <tj@kernel.org>

tools/sched_ext/include/scx/common.bpf.h

@@ -9,7 +9,7 @@
 
 #ifdef LSP
 #define __bpf__
-#include "../vmlinux/vmlinux.h"
+#include "../vmlinux.h"
 #else
 #include "vmlinux.h"
 #endif
@@ -24,6 +24,10 @@
 #define PF_EXITING			0x00000004
 #define CLOCK_MONOTONIC			1
 
+extern int LINUX_KERNEL_VERSION __kconfig;
+extern const char CONFIG_CC_VERSION_TEXT[64] __kconfig __weak;
+extern const char CONFIG_LOCALVERSION[64] __kconfig __weak;
+
 /*
  * Earlier versions of clang/pahole lost upper 32bits in 64bit enums which can
  * lead to really confusing misbehaviors. Let's trigger a build failure.
@@ -98,7 +102,7 @@ void ___scx_bpf_bstr_format_checker(const char *fmt, ...) {}
 	_Pragma("GCC diagnostic push")						\
 	_Pragma("GCC diagnostic ignored \"-Wint-conversion\"")			\
 	___bpf_fill(___param, args);						\
-	_Pragma("GCC diagnostic pop")						\
+	_Pragma("GCC diagnostic pop")
 
 /*
  * scx_bpf_exit() wraps the scx_bpf_exit_bstr() kfunc with variadic arguments
@@ -136,6 +140,20 @@ void ___scx_bpf_bstr_format_checker(const char *fmt, ...) {}
 	___scx_bpf_bstr_format_checker(fmt, ##args);				\
 })
 
+/*
+ * scx_bpf_dump_header() is a wrapper around scx_bpf_dump that adds a header
+ * of system information for debugging.
+ */
+#define scx_bpf_dump_header()							\
+({										\
+	scx_bpf_dump("kernel: %d.%d.%d %s\ncc: %s\n",				\
+		     LINUX_KERNEL_VERSION >> 16,				\
+		     LINUX_KERNEL_VERSION >> 8 & 0xFF,				\
+		     LINUX_KERNEL_VERSION & 0xFF,				\
+		     CONFIG_LOCALVERSION,					\
+		     CONFIG_CC_VERSION_TEXT);					\
+})
+
 #define BPF_STRUCT_OPS(name, args...)						\
 SEC("struct_ops/"#name)								\
 BPF_PROG(name, ##args)
@@ -317,6 +335,66 @@ u32 bpf_cpumask_any_and_distribute(const struct cpumask *src1,
 				   const struct cpumask *src2) __ksym;
 u32 bpf_cpumask_weight(const struct cpumask *cpumask) __ksym;
 
+int bpf_iter_bits_new(struct bpf_iter_bits *it, const u64 *unsafe_ptr__ign, u32 nr_words) __ksym;
+int *bpf_iter_bits_next(struct bpf_iter_bits *it) __ksym;
+void bpf_iter_bits_destroy(struct bpf_iter_bits *it) __ksym;
+
+#define def_iter_struct(name)							\
+struct bpf_iter_##name {							\
+    struct bpf_iter_bits it;							\
+    const struct cpumask *bitmap;						\
+};
+
+#define def_iter_new(name)							\
+static inline int bpf_iter_##name##_new(					\
+	struct bpf_iter_##name *it, const u64 *unsafe_ptr__ign, u32 nr_words)	\
+{										\
+	it->bitmap = scx_bpf_get_##name##_cpumask();				\
+	return bpf_iter_bits_new(&it->it, (const u64 *)it->bitmap,		\
+				 sizeof(struct cpumask) / 8);			\
+}
+
+#define def_iter_next(name)							\
+static inline int *bpf_iter_##name##_next(struct bpf_iter_##name *it) {		\
+	return bpf_iter_bits_next(&it->it);					\
+}
+
+#define def_iter_destroy(name)							\
+static inline void bpf_iter_##name##_destroy(struct bpf_iter_##name *it) {	\
+	scx_bpf_put_cpumask(it->bitmap);					\
+	bpf_iter_bits_destroy(&it->it);						\
+}
+#define def_for_each_cpu(cpu, name) for_each_##name##_cpu(cpu)
+
+/// Provides iterator for possible and online cpus.
+///
+/// # Example
+///
+/// ```
+/// static inline void example_use() {
+///     int *cpu;
+///
+///     for_each_possible_cpu(cpu){
+///         bpf_printk("CPU %d is possible", *cpu);
+///     }
+///
+///     for_each_online_cpu(cpu){
+///         bpf_printk("CPU %d is online", *cpu);
+///     }
+/// }
+/// ```
+def_iter_struct(possible);
+def_iter_new(possible);
+def_iter_next(possible);
+def_iter_destroy(possible);
+#define for_each_possible_cpu(cpu) bpf_for_each(possible, cpu, NULL, 0)
+
+def_iter_struct(online);
+def_iter_new(online);
+def_iter_next(online);
+def_iter_destroy(online);
+#define for_each_online_cpu(cpu) bpf_for_each(online, cpu, NULL, 0)
+
 /*
  * Access a cpumask in read-only mode (typically to check bits).
  */
@@ -423,5 +501,6 @@ static inline u32 log2_u64(u64 v)
 }
 
 #include "compat.bpf.h"
+#include "enums.bpf.h"
 
 #endif	/* __SCX_COMMON_BPF_H */

tools/sched_ext/include/scx/common.h

@@ -71,5 +71,11 @@ typedef int64_t s64;
 
 #include "user_exit_info.h"
 #include "compat.h"
+#include "enums.h"
+
+/* not available when building kernel tools/sched_ext */
+#if __has_include(<lib/sdt_task.h>)
+#include <lib/sdt_task.h>
+#endif
 
 #endif	/* __SCHED_EXT_COMMON_H */

tools/sched_ext/include/scx/compat.h

@@ -149,6 +149,7 @@ static inline long scx_hotplug_seq(void)
 	__skel = __scx_name##__open();						\
 	SCX_BUG_ON(!__skel, "Could not open " #__scx_name);			\
 	__skel->struct_ops.__ops_name->hotplug_seq = scx_hotplug_seq();		\
+	SCX_ENUM_INIT(__skel);							\
 	__skel; 								\
 })
 

tools/sched_ext/include/scx/enums.autogen.bpf.h

@@ -0,0 +1,105 @@
+/*
+ * WARNING: This file is autogenerated from scripts/gen_enums.py. If you would
+ * like to access an enum that is currently missing, add it to the script
+ * and run it from the root directory to update this file.
+ */
+
+const volatile u64 __SCX_OPS_NAME_LEN __weak;
+#define SCX_OPS_NAME_LEN __SCX_OPS_NAME_LEN
+
+const volatile u64 __SCX_SLICE_DFL __weak;
+#define SCX_SLICE_DFL __SCX_SLICE_DFL
+
+const volatile u64 __SCX_SLICE_INF __weak;
+#define SCX_SLICE_INF __SCX_SLICE_INF
+
+const volatile u64 __SCX_DSQ_FLAG_BUILTIN __weak;
+#define SCX_DSQ_FLAG_BUILTIN __SCX_DSQ_FLAG_BUILTIN
+
+const volatile u64 __SCX_DSQ_FLAG_LOCAL_ON __weak;
+#define SCX_DSQ_FLAG_LOCAL_ON __SCX_DSQ_FLAG_LOCAL_ON
+
+const volatile u64 __SCX_DSQ_INVALID __weak;
+#define SCX_DSQ_INVALID __SCX_DSQ_INVALID
+
+const volatile u64 __SCX_DSQ_GLOBAL __weak;
+#define SCX_DSQ_GLOBAL __SCX_DSQ_GLOBAL
+
+const volatile u64 __SCX_DSQ_LOCAL __weak;
+#define SCX_DSQ_LOCAL __SCX_DSQ_LOCAL
+
+const volatile u64 __SCX_DSQ_LOCAL_ON __weak;
+#define SCX_DSQ_LOCAL_ON __SCX_DSQ_LOCAL_ON
+
+const volatile u64 __SCX_DSQ_LOCAL_CPU_MASK __weak;
+#define SCX_DSQ_LOCAL_CPU_MASK __SCX_DSQ_LOCAL_CPU_MASK
+
+const volatile u64 __SCX_TASK_QUEUED __weak;
+#define SCX_TASK_QUEUED __SCX_TASK_QUEUED
+
+const volatile u64 __SCX_TASK_RESET_RUNNABLE_AT __weak;
+#define SCX_TASK_RESET_RUNNABLE_AT __SCX_TASK_RESET_RUNNABLE_AT
+
+const volatile u64 __SCX_TASK_DEQD_FOR_SLEEP __weak;
+#define SCX_TASK_DEQD_FOR_SLEEP __SCX_TASK_DEQD_FOR_SLEEP
+
+const volatile u64 __SCX_TASK_STATE_SHIFT __weak;
+#define SCX_TASK_STATE_SHIFT __SCX_TASK_STATE_SHIFT
+
+const volatile u64 __SCX_TASK_STATE_BITS __weak;
+#define SCX_TASK_STATE_BITS __SCX_TASK_STATE_BITS
+
+const volatile u64 __SCX_TASK_STATE_MASK __weak;
+#define SCX_TASK_STATE_MASK __SCX_TASK_STATE_MASK
+
+const volatile u64 __SCX_TASK_CURSOR __weak;
+#define SCX_TASK_CURSOR __SCX_TASK_CURSOR
+
+const volatile u64 __SCX_TASK_NONE __weak;
+#define SCX_TASK_NONE __SCX_TASK_NONE
+
+const volatile u64 __SCX_TASK_INIT __weak;
+#define SCX_TASK_INIT __SCX_TASK_INIT
+
+const volatile u64 __SCX_TASK_READY __weak;
+#define SCX_TASK_READY __SCX_TASK_READY
+
+const volatile u64 __SCX_TASK_ENABLED __weak;
+#define SCX_TASK_ENABLED __SCX_TASK_ENABLED
+
+const volatile u64 __SCX_TASK_NR_STATES __weak;
+#define SCX_TASK_NR_STATES __SCX_TASK_NR_STATES
+
+const volatile u64 __SCX_TASK_DSQ_ON_PRIQ __weak;
+#define SCX_TASK_DSQ_ON_PRIQ __SCX_TASK_DSQ_ON_PRIQ
+
+const volatile u64 __SCX_KICK_IDLE __weak;
+#define SCX_KICK_IDLE __SCX_KICK_IDLE
+
+const volatile u64 __SCX_KICK_PREEMPT __weak;
+#define SCX_KICK_PREEMPT __SCX_KICK_PREEMPT
+
+const volatile u64 __SCX_KICK_WAIT __weak;
+#define SCX_KICK_WAIT __SCX_KICK_WAIT
+
+const volatile u64 __SCX_ENQ_WAKEUP __weak;
+#define SCX_ENQ_WAKEUP __SCX_ENQ_WAKEUP
+
+const volatile u64 __SCX_ENQ_HEAD __weak;
+#define SCX_ENQ_HEAD __SCX_ENQ_HEAD
+
+const volatile u64 __SCX_ENQ_PREEMPT __weak;
+#define SCX_ENQ_PREEMPT __SCX_ENQ_PREEMPT
+
+const volatile u64 __SCX_ENQ_REENQ __weak;
+#define SCX_ENQ_REENQ __SCX_ENQ_REENQ
+
+const volatile u64 __SCX_ENQ_LAST __weak;
+#define SCX_ENQ_LAST __SCX_ENQ_LAST
+
+const volatile u64 __SCX_ENQ_CLEAR_OPSS __weak;
+#define SCX_ENQ_CLEAR_OPSS __SCX_ENQ_CLEAR_OPSS
+
+const volatile u64 __SCX_ENQ_DSQ_PRIQ __weak;
+#define SCX_ENQ_DSQ_PRIQ __SCX_ENQ_DSQ_PRIQ
+

tools/sched_ext/include/scx/enums.autogen.h

@@ -0,0 +1,41 @@
+/*
+ * WARNING: This file is autogenerated from scripts/gen_enums.py. If you would
+ * like to access an enum that is currently missing, add it to the script
+ * and run it from the root directory to update this file.
+ */
+
+#define SCX_ENUM_INIT(skel) do { \
+	SCX_ENUM_SET(skel, scx_public_consts, SCX_OPS_NAME_LEN); \
+	SCX_ENUM_SET(skel, scx_public_consts, SCX_SLICE_DFL); \
+	SCX_ENUM_SET(skel, scx_public_consts, SCX_SLICE_INF); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_FLAG_BUILTIN); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_FLAG_LOCAL_ON); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_INVALID); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_GLOBAL); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_LOCAL); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_LOCAL_ON); \
+	SCX_ENUM_SET(skel, scx_dsq_id_flags, SCX_DSQ_LOCAL_CPU_MASK); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_QUEUED); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_RESET_RUNNABLE_AT); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_DEQD_FOR_SLEEP); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_STATE_SHIFT); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_STATE_BITS); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_STATE_MASK); \
+	SCX_ENUM_SET(skel, scx_ent_flags, SCX_TASK_CURSOR); \
+	SCX_ENUM_SET(skel, scx_task_state, SCX_TASK_NONE); \
+	SCX_ENUM_SET(skel, scx_task_state, SCX_TASK_INIT); \
+	SCX_ENUM_SET(skel, scx_task_state, SCX_TASK_READY); \
+	SCX_ENUM_SET(skel, scx_task_state, SCX_TASK_ENABLED); \
+	SCX_ENUM_SET(skel, scx_task_state, SCX_TASK_NR_STATES); \
+	SCX_ENUM_SET(skel, scx_ent_dsq_flags, SCX_TASK_DSQ_ON_PRIQ); \
+	SCX_ENUM_SET(skel, scx_kick_flags, SCX_KICK_IDLE); \
+	SCX_ENUM_SET(skel, scx_kick_flags, SCX_KICK_PREEMPT); \
+	SCX_ENUM_SET(skel, scx_kick_flags, SCX_KICK_WAIT); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_WAKEUP); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_HEAD); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_PREEMPT); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_REENQ); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_LAST); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_CLEAR_OPSS); \
+	SCX_ENUM_SET(skel, scx_enq_flags, SCX_ENQ_DSQ_PRIQ); \
+} while (0)

tools/sched_ext/include/scx/enums.bpf.h

@@ -0,0 +1,12 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Convenience macros for getting/setting struct scx_enums instances.
+ *
+ * Copyright (c) 2024 Meta Platforms, Inc. and affiliates.
+ */
+#ifndef __SCX_ENUMS_BPF_H
+#define __SCX_ENUMS_BPF_H
+
+#include "enums.autogen.bpf.h"
+
+#endif /* __SCX_ENUMS_BPF_H */

tools/sched_ext/include/scx/enums.h

@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Define struct scx_enums that stores the load-time values of enums
+ * used by the BPF program.
+ *
+ * Copyright (c) 2024 Meta Platforms, Inc. and affiliates.
+ */
+
+#ifndef __SCX_ENUMS_H
+#define __SCX_ENUMS_H
+
+static inline void __ENUM_set(u64 *val, char *type, char *name)
+{
+	bool res;
+
+	res = __COMPAT_read_enum(type, name, val);
+	SCX_BUG_ON(!res, "enum not found(%s)", name);
+}
+
+#define SCX_ENUM_SET(skel, type, name) do {			\
+	__ENUM_set(&skel->rodata->__##name, #type, #name);	\
+	} while (0)
+
+
+#include "enums.autogen.h"
+
+#endif /* __SCX_ENUMS_H */

tools/sched_ext/include/scx/user_exit_info.h

@@ -10,6 +10,11 @@
 #ifndef __USER_EXIT_INFO_H
 #define __USER_EXIT_INFO_H
 
+#ifdef LSP
+#define __bpf__
+#include "../vmlinux.h"
+#endif
+
 enum uei_sizes {
 	UEI_REASON_LEN		= 128,
 	UEI_MSG_LEN		= 1024,
@@ -25,9 +30,7 @@ struct user_exit_info {
 
 #ifdef __bpf__
 
-#ifdef LSP
+#ifndef LSP
-#include "../vmlinux/vmlinux.h"
-#else
 #include "vmlinux.h"
 #endif
 #include <bpf/bpf_core_read.h>

tools/sched_ext/scx_central.bpf.c

@@ -57,7 +57,7 @@ enum {
 
 const volatile s32 central_cpu;
 const volatile u32 nr_cpu_ids = 1;	/* !0 for veristat, set during init */
-const volatile u64 slice_ns = SCX_SLICE_DFL;
+const volatile u64 slice_ns;
 
 bool timer_pinned = true;
 u64 nr_total, nr_locals, nr_queued, nr_lost_pids;

tools/sched_ext/scx_central.c

@@ -58,6 +58,7 @@ restart:
 
 	skel->rodata->central_cpu = 0;
 	skel->rodata->nr_cpu_ids = libbpf_num_possible_cpus();
+	skel->rodata->slice_ns = __COMPAT_ENUM_OR_ZERO("scx_public_consts", "SCX_SLICE_DFL");
 
 	while ((opt = getopt(argc, argv, "s:c:pvh")) != -1) {
 		switch (opt) {

tools/sched_ext/scx_flatcg.bpf.c

@@ -57,7 +57,7 @@ enum {
 char _license[] SEC("license") = "GPL";
 
 const volatile u32 nr_cpus = 32;	/* !0 for veristat, set during init */
-const volatile u64 cgrp_slice_ns = SCX_SLICE_DFL;
+const volatile u64 cgrp_slice_ns;
 const volatile bool fifo_sched;
 
 u64 cvtime_now;

tools/sched_ext/scx_flatcg.c

@@ -137,6 +137,7 @@ restart:
 	skel = SCX_OPS_OPEN(flatcg_ops, scx_flatcg);
 
 	skel->rodata->nr_cpus = libbpf_num_possible_cpus();
+	skel->rodata->cgrp_slice_ns = __COMPAT_ENUM_OR_ZERO("scx_public_consts", "SCX_SLICE_DFL");
 
 	while ((opt = getopt(argc, argv, "s:i:dfvh")) != -1) {
 		double v;

tools/sched_ext/scx_qmap.bpf.c

@@ -33,7 +33,7 @@ enum consts {
 
 char _license[] SEC("license") = "GPL";
 
-const volatile u64 slice_ns = SCX_SLICE_DFL;
+const volatile u64 slice_ns;
 const volatile u32 stall_user_nth;
 const volatile u32 stall_kernel_nth;
 const volatile u32 dsp_inf_loop_after;

tools/sched_ext/scx_qmap.c

@@ -64,6 +64,8 @@ int main(int argc, char **argv)
 
 	skel = SCX_OPS_OPEN(qmap_ops, scx_qmap);
 
+	skel->rodata->slice_ns = __COMPAT_ENUM_OR_ZERO("scx_public_consts", "SCX_SLICE_DFL");
+
 	while ((opt = getopt(argc, argv, "s:e:t:T:l:b:PHd:D:Spvh")) != -1) {
 		switch (opt) {
 		case 's':