From 7f267aeb228c470ce8f8fefdd324c1f03d028359 Mon Sep 17 00:00:00 2001 From: Peter Oskolkov Date: Fri, 22 Mar 2019 16:40:18 -0700 Subject: [PATCH 01/24] bpf: make bpf_skb_ecn_set_ce callable from BPF_PROG_TYPE_SCHED_ACT ANBZ: #5530 commit 315a202987dd2b2e0adebd13c83ceef44836e66f upstream. This helper is useful if a bpf tc filter sets skb->tstamp. Signed-off-by: Peter Oskolkov Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- net/core/filter.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/net/core/filter.c b/net/core/filter.c index e46497d08093..7c36fbde5c2c 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -6003,6 +6003,8 @@ tc_cls_act_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) return &bpf_skc_lookup_tcp_proto; case BPF_FUNC_tcp_check_syncookie: return &bpf_tcp_check_syncookie_proto; + case BPF_FUNC_skb_ecn_set_ce: + return &bpf_skb_ecn_set_ce_proto; #endif default: return bpf_base_func_proto(func_id); -- Gitee From 1b4be7bcee6ac9cad18408d7db7b3ceab92ef68e Mon Sep 17 00:00:00 2001 From: Matt Mullins Date: Wed, 12 Dec 2018 16:42:37 -0800 Subject: [PATCH 02/24] bpf: support raw tracepoints in modules ANBZ: #5530 commit a38d1107f937ca95dcf820161ef44ea683d6a0b1 upstream. Distributions build drivers as modules, including network and filesystem drivers which export numerous tracepoints. This enables bpf(BPF_RAW_TRACEPOINT_OPEN) to attach to those tracepoints. Signed-off-by: Matt Mullins Acked-by: Martin KaFai Lau Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- include/linux/module.h | 4 ++ include/linux/trace_events.h | 8 ++- kernel/bpf/syscall.c | 11 ++-- kernel/module.c | 5 ++ kernel/trace/bpf_trace.c | 99 +++++++++++++++++++++++++++++++++++- 5 files changed, 120 insertions(+), 7 deletions(-) diff --git a/include/linux/module.h b/include/linux/module.h index 9915397715fc..935bb918de43 100644 --- a/include/linux/module.h +++ b/include/linux/module.h @@ -433,6 +433,10 @@ struct module { unsigned int num_tracepoints; tracepoint_ptr_t *tracepoints_ptrs; #endif +#ifdef CONFIG_BPF_EVENTS + unsigned int num_bpf_raw_events; + struct bpf_raw_event_map *bpf_raw_events; +#endif #ifdef CONFIG_JUMP_LABEL struct jump_entry *jump_entries; unsigned int num_jump_entries; diff --git a/include/linux/trace_events.h b/include/linux/trace_events.h index 0643c083ed86..133cfea8413c 100644 --- a/include/linux/trace_events.h +++ b/include/linux/trace_events.h @@ -471,7 +471,8 @@ void perf_event_detach_bpf_prog(struct perf_event *event); int perf_event_query_prog_array(struct perf_event *event, void __user *info); int bpf_probe_register(struct bpf_raw_event_map *btp, struct bpf_prog *prog); int bpf_probe_unregister(struct bpf_raw_event_map *btp, struct bpf_prog *prog); -struct bpf_raw_event_map *bpf_find_raw_tracepoint(const char *name); +struct bpf_raw_event_map *bpf_get_raw_tracepoint(const char *name); +void bpf_put_raw_tracepoint(struct bpf_raw_event_map *btp); int bpf_get_perf_event_info(const struct perf_event *event, u32 *prog_id, u32 *fd_type, const char **buf, u64 *probe_offset, u64 *probe_addr); @@ -502,10 +503,13 @@ static inline int bpf_probe_unregister(struct bpf_raw_event_map *btp, struct bpf { return -EOPNOTSUPP; } -static inline struct bpf_raw_event_map *bpf_find_raw_tracepoint(const char *name) +static inline struct bpf_raw_event_map *bpf_get_raw_tracepoint(const char *name) { return NULL; } +static inline void bpf_put_raw_tracepoint(struct bpf_raw_event_map *btp) +{ +} static inline int bpf_get_perf_event_info(const struct perf_event *event, u32 *prog_id, u32 *fd_type, const char **buf, u64 *probe_offset, diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index cd7c9da22a56..644262530ea3 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -1769,6 +1769,7 @@ static int bpf_raw_tracepoint_release(struct inode *inode, struct file *filp) bpf_probe_unregister(raw_tp->btp, raw_tp->prog); bpf_prog_put(raw_tp->prog); } + bpf_put_raw_tracepoint(raw_tp->btp); kfree(raw_tp); return 0; } @@ -1794,13 +1795,15 @@ static int bpf_raw_tracepoint_open(const union bpf_attr *attr) return -EFAULT; tp_name[sizeof(tp_name) - 1] = 0; - btp = bpf_find_raw_tracepoint(tp_name); + btp = bpf_get_raw_tracepoint(tp_name); if (!btp) return -ENOENT; raw_tp = kzalloc(sizeof(*raw_tp), GFP_USER); - if (!raw_tp) - return -ENOMEM; + if (!raw_tp) { + err = -ENOMEM; + goto out_put_btp; + } raw_tp->btp = btp; prog = bpf_prog_get(attr->raw_tracepoint.prog_fd); @@ -1832,6 +1835,8 @@ static int bpf_raw_tracepoint_open(const union bpf_attr *attr) bpf_prog_put(prog); out_free_tp: kfree(raw_tp); +out_put_btp: + bpf_put_raw_tracepoint(btp); return err; } diff --git a/kernel/module.c b/kernel/module.c index 3eee7a55c038..ad51a4b42fc7 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -3131,6 +3131,11 @@ static int find_module_sections(struct module *mod, struct load_info *info) sizeof(*mod->tracepoints_ptrs), &mod->num_tracepoints); #endif +#ifdef CONFIG_BPF_EVENTS + mod->bpf_raw_events = section_objs(info, "__bpf_raw_tp_map", + sizeof(*mod->bpf_raw_events), + &mod->num_bpf_raw_events); +#endif #ifdef CONFIG_JUMP_LABEL mod->jump_entries = section_objs(info, "__jump_table", sizeof(*mod->jump_entries), diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index a29375b9d9a1..3df9fefb8bdd 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -19,6 +19,43 @@ #include "trace_probe.h" #include "trace.h" +#ifdef CONFIG_MODULES +struct bpf_trace_module { + struct module *module; + struct list_head list; +}; + +static LIST_HEAD(bpf_trace_modules); +static DEFINE_MUTEX(bpf_module_mutex); + +static struct bpf_raw_event_map *bpf_get_raw_tracepoint_module(const char *name) +{ + struct bpf_raw_event_map *btp, *ret = NULL; + struct bpf_trace_module *btm; + unsigned int i; + + mutex_lock(&bpf_module_mutex); + list_for_each_entry(btm, &bpf_trace_modules, list) { + for (i = 0; i < btm->module->num_bpf_raw_events; ++i) { + btp = &btm->module->bpf_raw_events[i]; + if (!strcmp(btp->tp->name, name)) { + if (try_module_get(btm->module)) + ret = btp; + goto out; + } + } + } +out: + mutex_unlock(&bpf_module_mutex); + return ret; +} +#else +static struct bpf_raw_event_map *bpf_get_raw_tracepoint_module(const char *name) +{ + return NULL; +} +#endif /* CONFIG_MODULES */ + u64 bpf_get_stackid(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5); u64 bpf_get_stack(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5); @@ -1195,7 +1232,7 @@ int perf_event_query_prog_array(struct perf_event *event, void __user *info) extern struct bpf_raw_event_map __start__bpf_raw_tp[]; extern struct bpf_raw_event_map __stop__bpf_raw_tp[]; -struct bpf_raw_event_map *bpf_find_raw_tracepoint(const char *name) +struct bpf_raw_event_map *bpf_get_raw_tracepoint(const char *name) { struct bpf_raw_event_map *btp = __start__bpf_raw_tp; @@ -1203,7 +1240,16 @@ struct bpf_raw_event_map *bpf_find_raw_tracepoint(const char *name) if (!strcmp(btp->tp->name, name)) return btp; } - return NULL; + + return bpf_get_raw_tracepoint_module(name); +} + +void bpf_put_raw_tracepoint(struct bpf_raw_event_map *btp) +{ + struct module *mod = __module_address((unsigned long)btp); + + if (mod) + module_put(mod); } static __always_inline @@ -1334,3 +1380,52 @@ int bpf_get_perf_event_info(const struct perf_event *event, u32 *prog_id, return err; } + +#ifdef CONFIG_MODULES +int bpf_event_notify(struct notifier_block *nb, unsigned long op, void *module) +{ + struct bpf_trace_module *btm, *tmp; + struct module *mod = module; + + if (mod->num_bpf_raw_events == 0 || + (op != MODULE_STATE_COMING && op != MODULE_STATE_GOING)) + return 0; + + mutex_lock(&bpf_module_mutex); + + switch (op) { + case MODULE_STATE_COMING: + btm = kzalloc(sizeof(*btm), GFP_KERNEL); + if (btm) { + btm->module = module; + list_add(&btm->list, &bpf_trace_modules); + } + break; + case MODULE_STATE_GOING: + list_for_each_entry_safe(btm, tmp, &bpf_trace_modules, list) { + if (btm->module == module) { + list_del(&btm->list); + kfree(btm); + break; + } + } + break; + } + + mutex_unlock(&bpf_module_mutex); + + return 0; +} + +static struct notifier_block bpf_module_nb = { + .notifier_call = bpf_event_notify, +}; + +int __init bpf_event_init(void) +{ + register_module_notifier(&bpf_module_nb); + return 0; +} + +fs_initcall(bpf_event_init); +#endif /* CONFIG_MODULES */ -- Gitee From 05773ce9922662ba9f03abff6eccd416460f5b9a Mon Sep 17 00:00:00 2001 From: Stanislav Fomichev Date: Mon, 13 May 2019 12:04:36 -0700 Subject: [PATCH 03/24] bpf: mark bpf_event_notify and bpf_event_init as static ANBZ: #5530 commit 390e99cfdda1334f45c718cc02cd26eb3135f233 upstream. Both of them are not declared in the headers and not used outside of bpf_trace.c file. Fixes: a38d1107f937c ("bpf: support raw tracepoints in modules") Signed-off-by: Stanislav Fomichev Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- kernel/trace/bpf_trace.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 3df9fefb8bdd..691be6f27ea6 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -1382,7 +1382,8 @@ int bpf_get_perf_event_info(const struct perf_event *event, u32 *prog_id, } #ifdef CONFIG_MODULES -int bpf_event_notify(struct notifier_block *nb, unsigned long op, void *module) +static int bpf_event_notify(struct notifier_block *nb, unsigned long op, + void *module) { struct bpf_trace_module *btm, *tmp; struct module *mod = module; @@ -1421,7 +1422,7 @@ static struct notifier_block bpf_module_nb = { .notifier_call = bpf_event_notify, }; -int __init bpf_event_init(void) +static int __init bpf_event_init(void) { register_module_notifier(&bpf_module_nb); return 0; -- Gitee From c6c8a71795e744c09ec4fc1fcf7de3952db7e4c6 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Thu, 3 Dec 2020 12:46:21 -0800 Subject: [PATCH 04/24] bpf: Fix bpf_put_raw_tracepoint()'s use of __module_address() ANBZ: #5530 commit 12cc126df82c96c89706aa207ad27c56f219047c upstream. __module_address() needs to be called with preemption disabled or with module_mutex taken. preempt_disable() is enough for read-only uses, which is what this fix does. Also, module_put() does internal check for NULL, so drop it as well. Fixes: a38d1107f937 ("bpf: support raw tracepoints in modules") Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Acked-by: Martin KaFai Lau Link: https://lore.kernel.org/bpf/20201203204634.1325171-2-andrii@kernel.org Signed-off-by: Yuanhe Shu --- kernel/trace/bpf_trace.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 691be6f27ea6..653bc8b2e9e4 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -1246,10 +1246,12 @@ struct bpf_raw_event_map *bpf_get_raw_tracepoint(const char *name) void bpf_put_raw_tracepoint(struct bpf_raw_event_map *btp) { - struct module *mod = __module_address((unsigned long)btp); + struct module *mod; - if (mod) - module_put(mod); + preempt_disable(); + mod = __module_address((unsigned long)btp); + module_put(mod); + preempt_enable(); } static __always_inline -- Gitee From 7654456d3c7ee529a274021138a1cd83be43e909 Mon Sep 17 00:00:00 2001 From: Yonghong Song Date: Thu, 23 May 2019 14:47:45 -0700 Subject: [PATCH 05/24] bpf: implement bpf_send_signal() helper ANBZ: #5530 commit 8b401f9ed2441ad9e219953927a842d24ed051fc upstream. This patch tries to solve the following specific use case. Currently, bpf program can already collect stack traces through kernel function get_perf_callchain() when certain events happens (e.g., cache miss counter or cpu clock counter overflows). But such stack traces are not enough for jitted programs, e.g., hhvm (jited php). To get real stack trace, jit engine internal data structures need to be traversed in order to get the real user functions. bpf program itself may not be the best place to traverse the jit engine as the traversing logic could be complex and it is not a stable interface either. Instead, hhvm implements a signal handler, e.g. for SIGALARM, and a set of program locations which it can dump stack traces. When it receives a signal, it will dump the stack in next such program location. Such a mechanism can be implemented in the following way: . a perf ring buffer is created between bpf program and tracing app. . once a particular event happens, bpf program writes to the ring buffer and the tracing app gets notified. . the tracing app sends a signal SIGALARM to the hhvm. But this method could have large delays and causing profiling results skewed. This patch implements bpf_send_signal() helper to send a signal to hhvm in real time, resulting in intended stack traces. Acked-by: Andrii Nakryiko Signed-off-by: Yonghong Song Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- include/uapi/linux/bpf.h | 17 +++++++++- kernel/trace/bpf_trace.c | 72 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 88 insertions(+), 1 deletion(-) diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 3d62b8beefef..6c6ff1cd2720 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -2664,6 +2664,20 @@ union bpf_attr { * 0 on success. * * **-ENOENT** if the bpf-local-storage cannot be found. + * + * int bpf_send_signal(u32 sig) + * Description + * Send signal *sig* to the current task. + * Return + * 0 on success or successfully queued. + * + * **-EBUSY** if work queue under nmi is full. + * + * **-EINVAL** if *sig* is invalid. + * + * **-EPERM** if no permission to send the *sig*. + * + * **-EAGAIN** if bpf program can try again. */ #define __BPF_FUNC_MAPPER(FN) \ FN(unspec), \ @@ -2774,7 +2788,8 @@ union bpf_attr { FN(strtol), \ FN(strtoul), \ FN(sk_storage_get), \ - FN(sk_storage_delete), + FN(sk_storage_delete), \ + FN(send_signal), /* integer value in 'imm' field of BPF_CALL instruction selects which helper * function eBPF program intends to call diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 653bc8b2e9e4..813dd089363f 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -608,6 +608,63 @@ static const struct bpf_func_proto bpf_probe_read_str_proto = { .arg3_type = ARG_ANYTHING, }; +struct send_signal_irq_work { + struct irq_work irq_work; + struct task_struct *task; + u32 sig; +}; + +static DEFINE_PER_CPU(struct send_signal_irq_work, send_signal_work); + +static void do_bpf_send_signal(struct irq_work *entry) +{ + struct send_signal_irq_work *work; + + work = container_of(entry, struct send_signal_irq_work, irq_work); + group_send_sig_info(work->sig, SEND_SIG_PRIV, work->task, PIDTYPE_TGID); +} + +BPF_CALL_1(bpf_send_signal, u32, sig) +{ + struct send_signal_irq_work *work = NULL; + + /* Similar to bpf_probe_write_user, task needs to be + * in a sound condition and kernel memory access be + * permitted in order to send signal to the current + * task. + */ + if (unlikely(current->flags & (PF_KTHREAD | PF_EXITING))) + return -EPERM; + if (unlikely(uaccess_kernel())) + return -EPERM; + if (unlikely(!nmi_uaccess_okay())) + return -EPERM; + + if (in_nmi()) { + work = this_cpu_ptr(&send_signal_work); + if (work->irq_work.flags & IRQ_WORK_BUSY) + return -EBUSY; + + /* Add the current task, which is the target of sending signal, + * to the irq_work. The current task may change when queued + * irq works get executed. + */ + work->task = current; + work->sig = sig; + irq_work_queue(&work->irq_work); + return 0; + } + + return group_send_sig_info(sig, SEND_SIG_PRIV, current, PIDTYPE_TGID); +} + +static const struct bpf_func_proto bpf_send_signal_proto = { + .func = bpf_send_signal, + .gpl_only = false, + .ret_type = RET_INTEGER, + .arg1_type = ARG_ANYTHING, +}; + static const struct bpf_func_proto * tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) { @@ -658,6 +715,8 @@ tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) case BPF_FUNC_get_current_cgroup_id: return &bpf_get_current_cgroup_id_proto; #endif + case BPF_FUNC_send_signal: + return &bpf_send_signal_proto; default: return NULL; } @@ -1430,5 +1489,18 @@ static int __init bpf_event_init(void) return 0; } +static int __init send_signal_irq_work_init(void) +{ + int cpu; + struct send_signal_irq_work *work; + + for_each_possible_cpu(cpu) { + work = per_cpu_ptr(&send_signal_work, cpu); + init_irq_work(&work->irq_work, do_bpf_send_signal); + } + return 0; +} + fs_initcall(bpf_event_init); +subsys_initcall(send_signal_irq_work_init); #endif /* CONFIG_MODULES */ -- Gitee From 95a514eab0774b56a37ad597760424e73b92c7b5 Mon Sep 17 00:00:00 2001 From: Yonghong Song Date: Thu, 23 May 2019 14:47:46 -0700 Subject: [PATCH 06/24] tools/bpf: sync bpf uapi header bpf.h to tools directory ANBZ: #5530 commit edaccf8985305967c22903a78283c8c837ea48dd upstream. The bpf uapi header include/uapi/linux/bpf.h is sync'ed to tools/include/uapi/linux/bpf.h. Acked-by: Andrii Nakryiko Signed-off-by: Yonghong Song Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- tools/include/uapi/linux/bpf.h | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index 763dcff87f92..e313a7dbc25c 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -2659,6 +2659,20 @@ union bpf_attr { * 0 on success. * * **-ENOENT** if the bpf-local-storage cannot be found. + * + * int bpf_send_signal(u32 sig) + * Description + * Send signal *sig* to the current task. + * Return + * 0 on success or successfully queued. + * + * **-EBUSY** if work queue under nmi is full. + * + * **-EINVAL** if *sig* is invalid. + * + * **-EPERM** if no permission to send the *sig*. + * + * **-EAGAIN** if bpf program can try again. */ #define __BPF_FUNC_MAPPER(FN) \ FN(unspec), \ @@ -2769,7 +2783,8 @@ union bpf_attr { FN(strtol), \ FN(strtoul), \ FN(sk_storage_get), \ - FN(sk_storage_delete), + FN(sk_storage_delete), \ + FN(send_signal), /* integer value in 'imm' field of BPF_CALL instruction selects which helper * function eBPF program intends to call -- Gitee From 9edfef897c09f1cafaa1166c234c5656ac4866ce Mon Sep 17 00:00:00 2001 From: Yonghong Song Date: Thu, 23 May 2019 14:47:47 -0700 Subject: [PATCH 07/24] tools/bpf: add selftest in test_progs for bpf_send_signal() helper ANBZ: #5530 commit 16f0efc3b46352018c297bbdb2c405e7d8a63095 upstream. The test covered both nmi and tracepoint perf events. $ ./test_progs ... test_send_signal_tracepoint:PASS:tracepoint 0 nsec ... test_send_signal_common:PASS:tracepoint 0 nsec ... test_send_signal_common:PASS:perf_event 0 nsec ... test_send_signal:OK Acked-by: Andrii Nakryiko Signed-off-by: Yonghong Song Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- tools/testing/selftests/bpf/bpf_helpers.h | 1 + .../bpf/progs/test_send_signal_kern.c | 51 +++++ tools/testing/selftests/bpf/test_progs.c | 197 ++++++++++++++++++ 3 files changed, 249 insertions(+) create mode 100644 tools/testing/selftests/bpf/progs/test_send_signal_kern.c diff --git a/tools/testing/selftests/bpf/bpf_helpers.h b/tools/testing/selftests/bpf/bpf_helpers.h index f5fefaaf343b..4f1caf1cf973 100644 --- a/tools/testing/selftests/bpf/bpf_helpers.h +++ b/tools/testing/selftests/bpf/bpf_helpers.h @@ -212,6 +212,7 @@ static void *(*bpf_sk_storage_get)(void *map, struct bpf_sock *sk, (void *) BPF_FUNC_sk_storage_get; static int (*bpf_sk_storage_delete)(void *map, struct bpf_sock *sk) = (void *)BPF_FUNC_sk_storage_delete; +static int (*bpf_send_signal)(unsigned sig) = (void *)BPF_FUNC_send_signal; /* llvm builtin functions that eBPF C program may use to * emit BPF_LD_ABS and BPF_LD_IND instructions diff --git a/tools/testing/selftests/bpf/progs/test_send_signal_kern.c b/tools/testing/selftests/bpf/progs/test_send_signal_kern.c new file mode 100644 index 000000000000..45a1a1a2c345 --- /dev/null +++ b/tools/testing/selftests/bpf/progs/test_send_signal_kern.c @@ -0,0 +1,51 @@ +// SPDX-License-Identifier: GPL-2.0 +// Copyright (c) 2019 Facebook +#include +#include +#include "bpf_helpers.h" + +struct bpf_map_def SEC("maps") info_map = { + .type = BPF_MAP_TYPE_ARRAY, + .key_size = sizeof(__u32), + .value_size = sizeof(__u64), + .max_entries = 1, +}; + +BPF_ANNOTATE_KV_PAIR(info_map, __u32, __u64); + +struct bpf_map_def SEC("maps") status_map = { + .type = BPF_MAP_TYPE_ARRAY, + .key_size = sizeof(__u32), + .value_size = sizeof(__u64), + .max_entries = 1, +}; + +BPF_ANNOTATE_KV_PAIR(status_map, __u32, __u64); + +SEC("send_signal_demo") +int bpf_send_signal_test(void *ctx) +{ + __u64 *info_val, *status_val; + __u32 key = 0, pid, sig; + int ret; + + status_val = bpf_map_lookup_elem(&status_map, &key); + if (!status_val || *status_val != 0) + return 0; + + info_val = bpf_map_lookup_elem(&info_map, &key); + if (!info_val || *info_val == 0) + return 0; + + sig = *info_val >> 32; + pid = *info_val & 0xffffFFFF; + + if ((bpf_get_current_pid_tgid() >> 32) == pid) { + ret = bpf_send_signal(sig); + if (ret == 0) + *status_val = 1; + } + + return 0; +} +char __license[] SEC("license") = "GPL"; diff --git a/tools/testing/selftests/bpf/test_progs.c b/tools/testing/selftests/bpf/test_progs.c index ab69a38ff8bd..02d9e9eab033 100644 --- a/tools/testing/selftests/bpf/test_progs.c +++ b/tools/testing/selftests/bpf/test_progs.c @@ -2737,6 +2737,202 @@ static void test_raw_tp_writable_test_run(void) close(bpf_fd); } +static volatile int sigusr1_received = 0; + +static void sigusr1_handler(int signum) +{ + sigusr1_received++; +} + +static int test_send_signal_common(struct perf_event_attr *attr, + int prog_type, + const char *test_name) +{ + int err = -1, pmu_fd, prog_fd, info_map_fd, status_map_fd; + const char *file = "./test_send_signal_kern.o"; + struct bpf_object *obj = NULL; + int pipe_c2p[2], pipe_p2c[2]; + __u32 key = 0, duration = 0; + char buf[256]; + pid_t pid; + __u64 val; + + if (CHECK(pipe(pipe_c2p), test_name, + "pipe pipe_c2p error: %s\n", strerror(errno))) + goto no_fork_done; + + if (CHECK(pipe(pipe_p2c), test_name, + "pipe pipe_p2c error: %s\n", strerror(errno))) { + close(pipe_c2p[0]); + close(pipe_c2p[1]); + goto no_fork_done; + } + + pid = fork(); + if (CHECK(pid < 0, test_name, "fork error: %s\n", strerror(errno))) { + close(pipe_c2p[0]); + close(pipe_c2p[1]); + close(pipe_p2c[0]); + close(pipe_p2c[1]); + goto no_fork_done; + } + + if (pid == 0) { + /* install signal handler and notify parent */ + signal(SIGUSR1, sigusr1_handler); + + close(pipe_c2p[0]); /* close read */ + close(pipe_p2c[1]); /* close write */ + + /* notify parent signal handler is installed */ + write(pipe_c2p[1], buf, 1); + + /* make sure parent enabled bpf program to send_signal */ + read(pipe_p2c[0], buf, 1); + + /* wait a little for signal handler */ + sleep(1); + + if (sigusr1_received) + write(pipe_c2p[1], "2", 1); + else + write(pipe_c2p[1], "0", 1); + + /* wait for parent notification and exit */ + read(pipe_p2c[0], buf, 1); + + close(pipe_c2p[1]); + close(pipe_p2c[0]); + exit(0); + } + + close(pipe_c2p[1]); /* close write */ + close(pipe_p2c[0]); /* close read */ + + err = bpf_prog_load(file, prog_type, &obj, &prog_fd); + if (CHECK(err < 0, test_name, "bpf_prog_load error: %s\n", + strerror(errno))) + goto prog_load_failure; + + pmu_fd = syscall(__NR_perf_event_open, attr, pid, -1, + -1 /* group id */, 0 /* flags */); + if (CHECK(pmu_fd < 0, test_name, "perf_event_open error: %s\n", + strerror(errno))) { + err = -1; + goto close_prog; + } + + err = ioctl(pmu_fd, PERF_EVENT_IOC_ENABLE, 0); + if (CHECK(err < 0, test_name, "ioctl perf_event_ioc_enable error: %s\n", + strerror(errno))) + goto disable_pmu; + + err = ioctl(pmu_fd, PERF_EVENT_IOC_SET_BPF, prog_fd); + if (CHECK(err < 0, test_name, "ioctl perf_event_ioc_set_bpf error: %s\n", + strerror(errno))) + goto disable_pmu; + + err = -1; + info_map_fd = bpf_object__find_map_fd_by_name(obj, "info_map"); + if (CHECK(info_map_fd < 0, test_name, "find map %s error\n", "info_map")) + goto disable_pmu; + + status_map_fd = bpf_object__find_map_fd_by_name(obj, "status_map"); + if (CHECK(status_map_fd < 0, test_name, "find map %s error\n", "status_map")) + goto disable_pmu; + + /* wait until child signal handler installed */ + read(pipe_c2p[0], buf, 1); + + /* trigger the bpf send_signal */ + key = 0; + val = (((__u64)(SIGUSR1)) << 32) | pid; + bpf_map_update_elem(info_map_fd, &key, &val, 0); + + /* notify child that bpf program can send_signal now */ + write(pipe_p2c[1], buf, 1); + + /* wait for result */ + err = read(pipe_c2p[0], buf, 1); + if (CHECK(err < 0, test_name, "reading pipe error: %s\n", strerror(errno))) + goto disable_pmu; + if (CHECK(err == 0, test_name, "reading pipe error: size 0\n")) { + err = -1; + goto disable_pmu; + } + + err = CHECK(buf[0] != '2', test_name, "incorrect result\n"); + + /* notify child safe to exit */ + write(pipe_p2c[1], buf, 1); + +disable_pmu: + close(pmu_fd); +close_prog: + bpf_object__close(obj); +prog_load_failure: + close(pipe_c2p[0]); + close(pipe_p2c[1]); + wait(NULL); +no_fork_done: + return err; +} + +static int test_send_signal_tracepoint(void) +{ + const char *id_path = "/sys/kernel/debug/tracing/events/syscalls/sys_enter_nanosleep/id"; + struct perf_event_attr attr = { + .type = PERF_TYPE_TRACEPOINT, + .sample_type = PERF_SAMPLE_RAW | PERF_SAMPLE_CALLCHAIN, + .sample_period = 1, + .wakeup_events = 1, + }; + __u32 duration = 0; + int bytes, efd; + char buf[256]; + + efd = open(id_path, O_RDONLY, 0); + if (CHECK(efd < 0, "tracepoint", + "open syscalls/sys_enter_nanosleep/id failure: %s\n", + strerror(errno))) + return -1; + + bytes = read(efd, buf, sizeof(buf)); + close(efd); + if (CHECK(bytes <= 0 || bytes >= sizeof(buf), "tracepoint", + "read syscalls/sys_enter_nanosleep/id failure: %s\n", + strerror(errno))) + return -1; + + attr.config = strtol(buf, NULL, 0); + + return test_send_signal_common(&attr, BPF_PROG_TYPE_TRACEPOINT, "tracepoint"); +} + +static int test_send_signal_nmi(void) +{ + struct perf_event_attr attr = { + .sample_freq = 50, + .freq = 1, + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + }; + + return test_send_signal_common(&attr, BPF_PROG_TYPE_PERF_EVENT, "perf_event"); +} + +static void test_send_signal(void) +{ + int ret = 0; + + ret |= test_send_signal_tracepoint(); + ret |= test_send_signal_nmi(); + if (!ret) + printf("test_send_signal:OK\n"); + else + printf("test_send_signal:FAIL\n"); +} + int main(int ac, char **av) { srand(time(NULL)); @@ -2777,6 +2973,7 @@ int main(int ac, char **av) test_global_data(); test_raw_tp_writable_reject_nbd_invalid(); test_raw_tp_writable_test_run(); + test_send_signal(); printf("Summary: %d PASSED, %d FAILED\n", pass_cnt, error_cnt); return error_cnt ? EXIT_FAILURE : EXIT_SUCCESS; -- Gitee From bd9fd845117900fdb60eb704d64b56003410ae1e Mon Sep 17 00:00:00 2001 From: Yonghong Song Date: Tue, 25 Jun 2019 17:35:03 -0700 Subject: [PATCH 08/24] bpf: fix compiler warning with CONFIG_MODULES=n MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit ANBZ: #5530 commit 9db1ff0a415c7de8eb67df5b2c56ac409ccefc37 upstream. With CONFIG_MODULES=n, the following compiler warning occurs: /data/users/yhs/work/net-next/kernel/trace/bpf_trace.c:605:13: warning: ‘do_bpf_send_signal’ defined but not used [-Wunused-function] static void do_bpf_send_signal(struct irq_work *entry) The __init function send_signal_irq_work_init(), which calls do_bpf_send_signal(), is defined under CONFIG_MODULES. Hence, when CONFIG_MODULES=n, nobody calls static function do_bpf_send_signal(), hence the warning. The init function send_signal_irq_work_init() should work without CONFIG_MODULES. Moving it out of CONFIG_MODULES code section fixed the compiler warning, and also make bpf_send_signal() helper work without CONFIG_MODULES. Fixes: 8b401f9ed244 ("bpf: implement bpf_send_signal() helper") Reported-By: Arnd Bergmann Signed-off-by: Yonghong Song Acked-by: Song Liu Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- kernel/trace/bpf_trace.c | 27 ++++++++++++++------------- 1 file changed, 14 insertions(+), 13 deletions(-) diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index 813dd089363f..feb088bd3c31 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -1442,6 +1442,20 @@ int bpf_get_perf_event_info(const struct perf_event *event, u32 *prog_id, return err; } +static int __init send_signal_irq_work_init(void) +{ + int cpu; + struct send_signal_irq_work *work; + + for_each_possible_cpu(cpu) { + work = per_cpu_ptr(&send_signal_work, cpu); + init_irq_work(&work->irq_work, do_bpf_send_signal); + } + return 0; +} + +subsys_initcall(send_signal_irq_work_init); + #ifdef CONFIG_MODULES static int bpf_event_notify(struct notifier_block *nb, unsigned long op, void *module) @@ -1489,18 +1503,5 @@ static int __init bpf_event_init(void) return 0; } -static int __init send_signal_irq_work_init(void) -{ - int cpu; - struct send_signal_irq_work *work; - - for_each_possible_cpu(cpu) { - work = per_cpu_ptr(&send_signal_work, cpu); - init_irq_work(&work->irq_work, do_bpf_send_signal); - } - return 0; -} - fs_initcall(bpf_event_init); -subsys_initcall(send_signal_irq_work_init); #endif /* CONFIG_MODULES */ -- Gitee From 80d8676da38358c63349529ba1c9119bb20ae7f2 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:58:56 -0700 Subject: [PATCH 09/24] libbpf: ensure libbpf.h is included along libbpf_internal.h ANBZ: #5530 commit 1d7a08b3bdaec1e25ba7979ff598272b7e34318f upstream. libbpf_internal.h expects a bunch of stuff defined in libbpf.h to be defined. This patch makes sure that libbpf.h is always included. Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/lib/bpf/libbpf_internal.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tools/lib/bpf/libbpf_internal.h b/tools/lib/bpf/libbpf_internal.h index dfab8012185c..61d90eb82ee6 100644 --- a/tools/lib/bpf/libbpf_internal.h +++ b/tools/lib/bpf/libbpf_internal.h @@ -9,6 +9,8 @@ #ifndef __LIBBPF_LIBBPF_INTERNAL_H #define __LIBBPF_LIBBPF_INTERNAL_H +#include "libbpf.h" + #define BTF_INFO_ENC(kind, kind_flag, vlen) \ ((!!(kind_flag) << 31) | ((kind) << 24) | ((vlen) & BTF_MAX_VLEN)) #define BTF_TYPE_ENC(name, info, size_or_type) (name), (info), (size_or_type) -- Gitee From c51658176f39968b632f0dde2eadc2fa12de144b Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:58:57 -0700 Subject: [PATCH 10/24] libbpf: add btf__parse_elf API to load .BTF and .BTF.ext ANBZ: #5530 commit e6c64855fd7ad565d87b2bd617f4a50d3bdad82f upstream. Loading BTF and BTF.ext from ELF file is a common need. Instead of requiring every user to re-implement it, let's provide this API from libbpf itself. It's mostly copy/paste from `bpftool btf dump` implementation, which will be switched to libbpf's version in next patch. btf__parse_elf allows to load BTF and optionally BTF.ext. This is also useful for tests that need to load/work with BTF, loaded from test ELF files. Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/lib/bpf/btf.c | 128 +++++++++++++++++++++++++++++++++++++++ tools/lib/bpf/btf.h | 2 + tools/lib/bpf/libbpf.map | 5 ++ 3 files changed, 135 insertions(+) diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c index 4cfffeb47fbd..21624a66e099 100644 --- a/tools/lib/bpf/btf.c +++ b/tools/lib/bpf/btf.c @@ -4,10 +4,12 @@ #include #include #include +#include #include #include #include #include +#include #include "btf.h" #include "bpf.h" #include "libbpf.h" @@ -416,6 +418,132 @@ struct btf *btf__new(__u8 *data, __u32 size) return btf; } +static bool btf_check_endianness(const GElf_Ehdr *ehdr) +{ +#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ + return ehdr->e_ident[EI_DATA] == ELFDATA2LSB; +#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ + return ehdr->e_ident[EI_DATA] == ELFDATA2MSB; +#else +# error "Unrecognized __BYTE_ORDER__" +#endif +} + +struct btf *btf__parse_elf(const char *path, struct btf_ext **btf_ext) +{ + Elf_Data *btf_data = NULL, *btf_ext_data = NULL; + int err = 0, fd = -1, idx = 0; + struct btf *btf = NULL; + Elf_Scn *scn = NULL; + Elf *elf = NULL; + GElf_Ehdr ehdr; + + if (elf_version(EV_CURRENT) == EV_NONE) { + pr_warning("failed to init libelf for %s\n", path); + return ERR_PTR(-LIBBPF_ERRNO__LIBELF); + } + + fd = open(path, O_RDONLY); + if (fd < 0) { + err = -errno; + pr_warning("failed to open %s: %s\n", path, strerror(errno)); + return ERR_PTR(err); + } + + err = -LIBBPF_ERRNO__FORMAT; + + elf = elf_begin(fd, ELF_C_READ, NULL); + if (!elf) { + pr_warning("failed to open %s as ELF file\n", path); + goto done; + } + if (!gelf_getehdr(elf, &ehdr)) { + pr_warning("failed to get EHDR from %s\n", path); + goto done; + } + if (!btf_check_endianness(&ehdr)) { + pr_warning("non-native ELF endianness is not supported\n"); + goto done; + } + if (!elf_rawdata(elf_getscn(elf, ehdr.e_shstrndx), NULL)) { + pr_warning("failed to get e_shstrndx from %s\n", path); + goto done; + } + + while ((scn = elf_nextscn(elf, scn)) != NULL) { + GElf_Shdr sh; + char *name; + + idx++; + if (gelf_getshdr(scn, &sh) != &sh) { + pr_warning("failed to get section(%d) header from %s\n", + idx, path); + goto done; + } + name = elf_strptr(elf, ehdr.e_shstrndx, sh.sh_name); + if (!name) { + pr_warning("failed to get section(%d) name from %s\n", + idx, path); + goto done; + } + if (strcmp(name, BTF_ELF_SEC) == 0) { + btf_data = elf_getdata(scn, 0); + if (!btf_data) { + pr_warning("failed to get section(%d, %s) data from %s\n", + idx, name, path); + goto done; + } + continue; + } else if (btf_ext && strcmp(name, BTF_EXT_ELF_SEC) == 0) { + btf_ext_data = elf_getdata(scn, 0); + if (!btf_ext_data) { + pr_warning("failed to get section(%d, %s) data from %s\n", + idx, name, path); + goto done; + } + continue; + } + } + + err = 0; + + if (!btf_data) { + err = -ENOENT; + goto done; + } + btf = btf__new(btf_data->d_buf, btf_data->d_size); + if (IS_ERR(btf)) + goto done; + + if (btf_ext && btf_ext_data) { + *btf_ext = btf_ext__new(btf_ext_data->d_buf, + btf_ext_data->d_size); + if (IS_ERR(*btf_ext)) + goto done; + } else if (btf_ext) { + *btf_ext = NULL; + } +done: + if (elf) + elf_end(elf); + close(fd); + + if (err) + return ERR_PTR(err); + /* + * btf is always parsed before btf_ext, so no need to clean up + * btf_ext, if btf loading failed + */ + if (IS_ERR(btf)) + return btf; + if (btf_ext && IS_ERR(*btf_ext)) { + btf__free(btf); + err = PTR_ERR(*btf_ext); + return ERR_PTR(err); + } + return btf; +} + static int compare_vsi_off(const void *_a, const void *_b) { const struct btf_var_secinfo *a = _a; diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h index c7b399e81fce..bded210df9e8 100644 --- a/tools/lib/bpf/btf.h +++ b/tools/lib/bpf/btf.h @@ -59,6 +59,8 @@ struct btf_ext_header { LIBBPF_API void btf__free(struct btf *btf); LIBBPF_API struct btf *btf__new(__u8 *data, __u32 size); +LIBBPF_API struct btf *btf__parse_elf(const char *path, + struct btf_ext **btf_ext); LIBBPF_API int btf__finalize_data(struct bpf_object *obj, struct btf *btf); LIBBPF_API int btf__load(struct btf *btf); LIBBPF_API __s32 btf__find_by_name(const struct btf *btf, diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map index 673001787cba..6ea5ce19b9e0 100644 --- a/tools/lib/bpf/libbpf.map +++ b/tools/lib/bpf/libbpf.map @@ -164,3 +164,8 @@ LIBBPF_0.0.3 { bpf_map_freeze; btf__finalize_data; } LIBBPF_0.0.2; + +LIBBPF_0.0.4 { + global: + btf__parse_elf; +} LIBBPF_0.0.3; -- Gitee From a4495f9bfb3d593debfb3f58a7f670685f02aa49 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:58:58 -0700 Subject: [PATCH 11/24] bpftool: use libbpf's btf__parse_elf API ANBZ: #5530 commit 58650cc47382af7d87282f53a5b508ac4bf734af upstream. Use btf__parse_elf() API, provided by libbpf, instead of implementing ELF parsing by itself. Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/bpf/bpftool/btf.c | 117 +++------------------------------------- 1 file changed, 8 insertions(+), 109 deletions(-) diff --git a/tools/bpf/bpftool/btf.c b/tools/bpf/bpftool/btf.c index 5c67bbcdf914..d861b088a947 100644 --- a/tools/bpf/bpftool/btf.c +++ b/tools/bpf/bpftool/btf.c @@ -8,8 +8,8 @@ #include #include #include -#include #include +#include #include #include "btf.h" @@ -340,112 +340,6 @@ static int dump_btf_raw(const struct btf *btf, return 0; } -static bool check_btf_endianness(GElf_Ehdr *ehdr) -{ - static unsigned int const endian = 1; - - switch (ehdr->e_ident[EI_DATA]) { - case ELFDATA2LSB: - return *(unsigned char const *)&endian == 1; - case ELFDATA2MSB: - return *(unsigned char const *)&endian == 0; - default: - return 0; - } -} - -static int btf_load_from_elf(const char *path, struct btf **btf) -{ - int err = -1, fd = -1, idx = 0; - Elf_Data *btf_data = NULL; - Elf_Scn *scn = NULL; - Elf *elf = NULL; - GElf_Ehdr ehdr; - - if (elf_version(EV_CURRENT) == EV_NONE) { - p_err("failed to init libelf for %s", path); - return -1; - } - - fd = open(path, O_RDONLY); - if (fd < 0) { - p_err("failed to open %s: %s", path, strerror(errno)); - return -1; - } - - elf = elf_begin(fd, ELF_C_READ, NULL); - if (!elf) { - p_err("failed to open %s as ELF file", path); - goto done; - } - if (!gelf_getehdr(elf, &ehdr)) { - p_err("failed to get EHDR from %s", path); - goto done; - } - if (!check_btf_endianness(&ehdr)) { - p_err("non-native ELF endianness is not supported"); - goto done; - } - if (!elf_rawdata(elf_getscn(elf, ehdr.e_shstrndx), NULL)) { - p_err("failed to get e_shstrndx from %s\n", path); - goto done; - } - - while ((scn = elf_nextscn(elf, scn)) != NULL) { - GElf_Shdr sh; - char *name; - - idx++; - if (gelf_getshdr(scn, &sh) != &sh) { - p_err("failed to get section(%d) header from %s", - idx, path); - goto done; - } - name = elf_strptr(elf, ehdr.e_shstrndx, sh.sh_name); - if (!name) { - p_err("failed to get section(%d) name from %s", - idx, path); - goto done; - } - if (strcmp(name, BTF_ELF_SEC) == 0) { - btf_data = elf_getdata(scn, 0); - if (!btf_data) { - p_err("failed to get section(%d, %s) data from %s", - idx, name, path); - goto done; - } - break; - } - } - - if (!btf_data) { - p_err("%s ELF section not found in %s", BTF_ELF_SEC, path); - goto done; - } - - *btf = btf__new(btf_data->d_buf, btf_data->d_size); - if (IS_ERR(*btf)) { - err = PTR_ERR(*btf); - *btf = NULL; - p_err("failed to load BTF data from %s: %s", - path, strerror(err)); - goto done; - } - - err = 0; -done: - if (err) { - if (*btf) { - btf__free(*btf); - *btf = NULL; - } - } - if (elf) - elf_end(elf); - close(fd); - return err; -} - static int do_dump(int argc, char **argv) { struct btf *btf = NULL; @@ -522,9 +416,14 @@ static int do_dump(int argc, char **argv) } NEXT_ARG(); } else if (is_prefix(src, "file")) { - err = btf_load_from_elf(*argv, &btf); - if (err) + btf = btf__parse_elf(*argv, NULL); + if (IS_ERR(btf)) { + err = PTR_ERR(btf); + btf = NULL; + p_err("failed to load BTF from %s: %s", + *argv, strerror(err)); goto done; + } NEXT_ARG(); } else { err = -1; -- Gitee From 968a4d1e8078dbdbcbff1a5cb0225d8c84a1995e Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:59:00 -0700 Subject: [PATCH 12/24] libbpf: add resizable non-thread safe internal hashmap ANBZ: #5530 commit e3b924224028c6fc31545e3812eecbe2ddbf35f6 upstream. There is a need for fast point lookups inside libbpf for multiple use cases (e.g., name resolution for BTF-to-C conversion, by-name lookups in BTF for upcoming BPF CO-RE relocation support, etc). This patch implements simple resizable non-thread safe hashmap using single linked list chains. Four different insert strategies are supported: - HASHMAP_ADD - only add key/value if key doesn't exist yet; - HASHMAP_SET - add key/value pair if key doesn't exist yet; otherwise, update value; - HASHMAP_UPDATE - update value, if key already exists; otherwise, do nothing and return -ENOENT; - HASHMAP_APPEND - always add key/value pair, even if key already exists. This turns hashmap into a multimap by allowing multiple values to be associated with the same key. Most useful read API for such hashmap is hashmap__for_each_key_entry() iteration. If hashmap__find() is still used, it will return last inserted key/value entry (first in a bucket chain). For HASHMAP_SET and HASHMAP_UPDATE, old key/value pair is returned, so that calling code can handle proper memory management, if necessary. Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/lib/bpf/Build | 2 +- tools/lib/bpf/hashmap.c | 228 ++++++++++++++++++++++++++++++++++++++++ tools/lib/bpf/hashmap.h | 173 ++++++++++++++++++++++++++++++ 3 files changed, 402 insertions(+), 1 deletion(-) create mode 100644 tools/lib/bpf/hashmap.c create mode 100644 tools/lib/bpf/hashmap.h diff --git a/tools/lib/bpf/Build b/tools/lib/bpf/Build index ee9d5362f35b..dcf0d36598e0 100644 --- a/tools/lib/bpf/Build +++ b/tools/lib/bpf/Build @@ -1 +1 @@ -libbpf-y := libbpf.o bpf.o nlattr.o btf.o libbpf_errno.o str_error.o netlink.o bpf_prog_linfo.o libbpf_probes.o xsk.o +libbpf-y := libbpf.o bpf.o nlattr.o btf.o libbpf_errno.o str_error.o netlink.o bpf_prog_linfo.o libbpf_probes.o xsk.o hashmap.o diff --git a/tools/lib/bpf/hashmap.c b/tools/lib/bpf/hashmap.c new file mode 100644 index 000000000000..e0d292f855d8 --- /dev/null +++ b/tools/lib/bpf/hashmap.c @@ -0,0 +1,228 @@ +// SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) + +/* + * Generic non-thread safe hash map implementation. + * + * Copyright (c) 2019 Facebook + */ +#include +#include +#include +#include +#include +#include "hashmap.h" + +/* start with 4 buckets */ +#define HASHMAP_MIN_CAP_BITS 2 + +static void hashmap_add_entry(struct hashmap_entry **pprev, + struct hashmap_entry *entry) +{ + entry->next = *pprev; + *pprev = entry; +} + +static void hashmap_del_entry(struct hashmap_entry **pprev, + struct hashmap_entry *entry) +{ + *pprev = entry->next; + entry->next = NULL; +} + +void hashmap__init(struct hashmap *map, hashmap_hash_fn hash_fn, + hashmap_equal_fn equal_fn, void *ctx) +{ + map->hash_fn = hash_fn; + map->equal_fn = equal_fn; + map->ctx = ctx; + + map->buckets = NULL; + map->cap = 0; + map->cap_bits = 0; + map->sz = 0; +} + +struct hashmap *hashmap__new(hashmap_hash_fn hash_fn, + hashmap_equal_fn equal_fn, + void *ctx) +{ + struct hashmap *map = malloc(sizeof(struct hashmap)); + + if (!map) + return ERR_PTR(-ENOMEM); + hashmap__init(map, hash_fn, equal_fn, ctx); + return map; +} + +void hashmap__clear(struct hashmap *map) +{ + free(map->buckets); + map->cap = map->cap_bits = map->sz = 0; +} + +void hashmap__free(struct hashmap *map) +{ + if (!map) + return; + + hashmap__clear(map); + free(map); +} + +size_t hashmap__size(const struct hashmap *map) +{ + return map->sz; +} + +size_t hashmap__capacity(const struct hashmap *map) +{ + return map->cap; +} + +static bool hashmap_needs_to_grow(struct hashmap *map) +{ + /* grow if empty or more than 75% filled */ + return (map->cap == 0) || ((map->sz + 1) * 4 / 3 > map->cap); +} + +static int hashmap_grow(struct hashmap *map) +{ + struct hashmap_entry **new_buckets; + struct hashmap_entry *cur, *tmp; + size_t new_cap_bits, new_cap; + size_t h; + int bkt; + + new_cap_bits = map->cap_bits + 1; + if (new_cap_bits < HASHMAP_MIN_CAP_BITS) + new_cap_bits = HASHMAP_MIN_CAP_BITS; + + new_cap = 1UL << new_cap_bits; + new_buckets = calloc(new_cap, sizeof(new_buckets[0])); + if (!new_buckets) + return -ENOMEM; + + hashmap__for_each_entry_safe(map, cur, tmp, bkt) { + h = hash_bits(map->hash_fn(cur->key, map->ctx), new_cap_bits); + hashmap_add_entry(&new_buckets[h], cur); + } + + map->cap = new_cap; + map->cap_bits = new_cap_bits; + free(map->buckets); + map->buckets = new_buckets; + + return 0; +} + +static bool hashmap_find_entry(const struct hashmap *map, + const void *key, size_t hash, + struct hashmap_entry ***pprev, + struct hashmap_entry **entry) +{ + struct hashmap_entry *cur, **prev_ptr; + + if (!map->buckets) + return false; + + for (prev_ptr = &map->buckets[hash], cur = *prev_ptr; + cur; + prev_ptr = &cur->next, cur = cur->next) { + if (map->equal_fn(cur->key, key, map->ctx)) { + if (pprev) + *pprev = prev_ptr; + *entry = cur; + return true; + } + } + + return false; +} + +int hashmap__insert(struct hashmap *map, const void *key, void *value, + enum hashmap_insert_strategy strategy, + const void **old_key, void **old_value) +{ + struct hashmap_entry *entry; + size_t h; + int err; + + if (old_key) + *old_key = NULL; + if (old_value) + *old_value = NULL; + + h = hash_bits(map->hash_fn(key, map->ctx), map->cap_bits); + if (strategy != HASHMAP_APPEND && + hashmap_find_entry(map, key, h, NULL, &entry)) { + if (old_key) + *old_key = entry->key; + if (old_value) + *old_value = entry->value; + + if (strategy == HASHMAP_SET || strategy == HASHMAP_UPDATE) { + entry->key = key; + entry->value = value; + return 0; + } else if (strategy == HASHMAP_ADD) { + return -EEXIST; + } + } + + if (strategy == HASHMAP_UPDATE) + return -ENOENT; + + if (hashmap_needs_to_grow(map)) { + err = hashmap_grow(map); + if (err) + return err; + h = hash_bits(map->hash_fn(key, map->ctx), map->cap_bits); + } + + entry = malloc(sizeof(struct hashmap_entry)); + if (!entry) + return -ENOMEM; + + entry->key = key; + entry->value = value; + hashmap_add_entry(&map->buckets[h], entry); + map->sz++; + + return 0; +} + +bool hashmap__find(const struct hashmap *map, const void *key, void **value) +{ + struct hashmap_entry *entry; + size_t h; + + h = hash_bits(map->hash_fn(key, map->ctx), map->cap_bits); + if (!hashmap_find_entry(map, key, h, NULL, &entry)) + return false; + + if (value) + *value = entry->value; + return true; +} + +bool hashmap__delete(struct hashmap *map, const void *key, + const void **old_key, void **old_value) +{ + struct hashmap_entry **pprev, *entry; + size_t h; + + h = hash_bits(map->hash_fn(key, map->ctx), map->cap_bits); + if (!hashmap_find_entry(map, key, h, &pprev, &entry)) + return false; + + if (old_key) + *old_key = entry->key; + if (old_value) + *old_value = entry->value; + + hashmap_del_entry(pprev, entry); + free(entry); + map->sz--; + + return true; +} diff --git a/tools/lib/bpf/hashmap.h b/tools/lib/bpf/hashmap.h new file mode 100644 index 000000000000..03748a742146 --- /dev/null +++ b/tools/lib/bpf/hashmap.h @@ -0,0 +1,173 @@ +/* SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) */ + +/* + * Generic non-thread safe hash map implementation. + * + * Copyright (c) 2019 Facebook + */ +#ifndef __LIBBPF_HASHMAP_H +#define __LIBBPF_HASHMAP_H + +#include +#include +#include "libbpf_internal.h" + +static inline size_t hash_bits(size_t h, int bits) +{ + /* shuffle bits and return requested number of upper bits */ + return (h * 11400714819323198485llu) >> (__WORDSIZE - bits); +} + +typedef size_t (*hashmap_hash_fn)(const void *key, void *ctx); +typedef bool (*hashmap_equal_fn)(const void *key1, const void *key2, void *ctx); + +struct hashmap_entry { + const void *key; + void *value; + struct hashmap_entry *next; +}; + +struct hashmap { + hashmap_hash_fn hash_fn; + hashmap_equal_fn equal_fn; + void *ctx; + + struct hashmap_entry **buckets; + size_t cap; + size_t cap_bits; + size_t sz; +}; + +#define HASHMAP_INIT(hash_fn, equal_fn, ctx) { \ + .hash_fn = (hash_fn), \ + .equal_fn = (equal_fn), \ + .ctx = (ctx), \ + .buckets = NULL, \ + .cap = 0, \ + .cap_bits = 0, \ + .sz = 0, \ +} + +void hashmap__init(struct hashmap *map, hashmap_hash_fn hash_fn, + hashmap_equal_fn equal_fn, void *ctx); +struct hashmap *hashmap__new(hashmap_hash_fn hash_fn, + hashmap_equal_fn equal_fn, + void *ctx); +void hashmap__clear(struct hashmap *map); +void hashmap__free(struct hashmap *map); + +size_t hashmap__size(const struct hashmap *map); +size_t hashmap__capacity(const struct hashmap *map); + +/* + * Hashmap insertion strategy: + * - HASHMAP_ADD - only add key/value if key doesn't exist yet; + * - HASHMAP_SET - add key/value pair if key doesn't exist yet; otherwise, + * update value; + * - HASHMAP_UPDATE - update value, if key already exists; otherwise, do + * nothing and return -ENOENT; + * - HASHMAP_APPEND - always add key/value pair, even if key already exists. + * This turns hashmap into a multimap by allowing multiple values to be + * associated with the same key. Most useful read API for such hashmap is + * hashmap__for_each_key_entry() iteration. If hashmap__find() is still + * used, it will return last inserted key/value entry (first in a bucket + * chain). + */ +enum hashmap_insert_strategy { + HASHMAP_ADD, + HASHMAP_SET, + HASHMAP_UPDATE, + HASHMAP_APPEND, +}; + +/* + * hashmap__insert() adds key/value entry w/ various semantics, depending on + * provided strategy value. If a given key/value pair replaced already + * existing key/value pair, both old key and old value will be returned + * through old_key and old_value to allow calling code do proper memory + * management. + */ +int hashmap__insert(struct hashmap *map, const void *key, void *value, + enum hashmap_insert_strategy strategy, + const void **old_key, void **old_value); + +static inline int hashmap__add(struct hashmap *map, + const void *key, void *value) +{ + return hashmap__insert(map, key, value, HASHMAP_ADD, NULL, NULL); +} + +static inline int hashmap__set(struct hashmap *map, + const void *key, void *value, + const void **old_key, void **old_value) +{ + return hashmap__insert(map, key, value, HASHMAP_SET, + old_key, old_value); +} + +static inline int hashmap__update(struct hashmap *map, + const void *key, void *value, + const void **old_key, void **old_value) +{ + return hashmap__insert(map, key, value, HASHMAP_UPDATE, + old_key, old_value); +} + +static inline int hashmap__append(struct hashmap *map, + const void *key, void *value) +{ + return hashmap__insert(map, key, value, HASHMAP_APPEND, NULL, NULL); +} + +bool hashmap__delete(struct hashmap *map, const void *key, + const void **old_key, void **old_value); + +bool hashmap__find(const struct hashmap *map, const void *key, void **value); + +/* + * hashmap__for_each_entry - iterate over all entries in hashmap + * @map: hashmap to iterate + * @cur: struct hashmap_entry * used as a loop cursor + * @bkt: integer used as a bucket loop cursor + */ +#define hashmap__for_each_entry(map, cur, bkt) \ + for (bkt = 0; bkt < map->cap; bkt++) \ + for (cur = map->buckets[bkt]; cur; cur = cur->next) + +/* + * hashmap__for_each_entry_safe - iterate over all entries in hashmap, safe + * against removals + * @map: hashmap to iterate + * @cur: struct hashmap_entry * used as a loop cursor + * @tmp: struct hashmap_entry * used as a temporary next cursor storage + * @bkt: integer used as a bucket loop cursor + */ +#define hashmap__for_each_entry_safe(map, cur, tmp, bkt) \ + for (bkt = 0; bkt < map->cap; bkt++) \ + for (cur = map->buckets[bkt]; \ + cur && ({tmp = cur->next; true; }); \ + cur = tmp) + +/* + * hashmap__for_each_key_entry - iterate over entries associated with given key + * @map: hashmap to iterate + * @cur: struct hashmap_entry * used as a loop cursor + * @key: key to iterate entries for + */ +#define hashmap__for_each_key_entry(map, cur, _key) \ + for (cur = ({ size_t bkt = hash_bits(map->hash_fn((_key), map->ctx),\ + map->cap_bits); \ + map->buckets ? map->buckets[bkt] : NULL; }); \ + cur; \ + cur = cur->next) \ + if (map->equal_fn(cur->key, (_key), map->ctx)) + +#define hashmap__for_each_key_entry_safe(map, cur, tmp, _key) \ + for (cur = ({ size_t bkt = hash_bits(map->hash_fn((_key), map->ctx),\ + map->cap_bits); \ + cur = map->buckets ? map->buckets[bkt] : NULL; }); \ + cur && ({ tmp = cur->next; true; }); \ + cur = tmp) \ + if (map->equal_fn(cur->key, (_key), map->ctx)) + +#endif /* __LIBBPF_HASHMAP_H */ -- Gitee From 8d68fa62b0c111e184ae1c29cfe97e1bf33f3076 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:59:02 -0700 Subject: [PATCH 13/24] libbpf: switch btf_dedup() to hashmap for dedup table ANBZ: #5530 commit 2fc3fc0bcdcc5429947d1e5ed8f6f0a77ba80997 upstream. Utilize libbpf's hashmap as a multimap fof dedup_table implementation. Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/lib/bpf/btf.c | 201 +++++++++++++++++++------------------------- 1 file changed, 85 insertions(+), 116 deletions(-) diff --git a/tools/lib/bpf/btf.c b/tools/lib/bpf/btf.c index 21624a66e099..2680799de25f 100644 --- a/tools/lib/bpf/btf.c +++ b/tools/lib/bpf/btf.c @@ -14,6 +14,7 @@ #include "bpf.h" #include "libbpf.h" #include "libbpf_internal.h" +#include "hashmap.h" #define max(a, b) ((a) > (b) ? (a) : (b)) #define min(a, b) ((a) < (b) ? (a) : (b)) @@ -1292,16 +1293,9 @@ int btf__dedup(struct btf *btf, struct btf_ext *btf_ext, return err; } -#define BTF_DEDUP_TABLE_DEFAULT_SIZE (1 << 14) -#define BTF_DEDUP_TABLE_MAX_SIZE_LOG 31 #define BTF_UNPROCESSED_ID ((__u32)-1) #define BTF_IN_PROGRESS_ID ((__u32)-2) -struct btf_dedup_node { - struct btf_dedup_node *next; - __u32 type_id; -}; - struct btf_dedup { /* .BTF section to be deduped in-place */ struct btf *btf; @@ -1317,7 +1311,7 @@ struct btf_dedup { * candidates, which is fine because we rely on subsequent * btf_xxx_equal() checks to authoritatively verify type equality. */ - struct btf_dedup_node **dedup_table; + struct hashmap *dedup_table; /* Canonical types map */ __u32 *map; /* Hypothetical mapping, used during type graph equivalence checks */ @@ -1342,30 +1336,18 @@ struct btf_str_ptrs { __u32 cap; }; -static inline __u32 hash_combine(__u32 h, __u32 value) +static long hash_combine(long h, long value) { -/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */ -#define GOLDEN_RATIO_PRIME 0x9e370001UL - return h * 37 + value * GOLDEN_RATIO_PRIME; -#undef GOLDEN_RATIO_PRIME + return h * 31 + value; } -#define for_each_dedup_cand(d, hash, node) \ - for (node = d->dedup_table[hash & (d->opts.dedup_table_size - 1)]; \ - node; \ - node = node->next) +#define for_each_dedup_cand(d, node, hash) \ + hashmap__for_each_key_entry(d->dedup_table, node, (void *)hash) -static int btf_dedup_table_add(struct btf_dedup *d, __u32 hash, __u32 type_id) +static int btf_dedup_table_add(struct btf_dedup *d, long hash, __u32 type_id) { - struct btf_dedup_node *node = malloc(sizeof(struct btf_dedup_node)); - int bucket = hash & (d->opts.dedup_table_size - 1); - - if (!node) - return -ENOMEM; - node->type_id = type_id; - node->next = d->dedup_table[bucket]; - d->dedup_table[bucket] = node; - return 0; + return hashmap__append(d->dedup_table, + (void *)hash, (void *)(long)type_id); } static int btf_dedup_hypot_map_add(struct btf_dedup *d, @@ -1394,36 +1376,10 @@ static void btf_dedup_clear_hypot_map(struct btf_dedup *d) d->hypot_cnt = 0; } -static void btf_dedup_table_free(struct btf_dedup *d) -{ - struct btf_dedup_node *head, *tmp; - int i; - - if (!d->dedup_table) - return; - - for (i = 0; i < d->opts.dedup_table_size; i++) { - while (d->dedup_table[i]) { - tmp = d->dedup_table[i]; - d->dedup_table[i] = tmp->next; - free(tmp); - } - - head = d->dedup_table[i]; - while (head) { - tmp = head; - head = head->next; - free(tmp); - } - } - - free(d->dedup_table); - d->dedup_table = NULL; -} - static void btf_dedup_free(struct btf_dedup *d) { - btf_dedup_table_free(d); + hashmap__free(d->dedup_table); + d->dedup_table = NULL; free(d->map); d->map = NULL; @@ -1437,40 +1393,43 @@ static void btf_dedup_free(struct btf_dedup *d) free(d); } -/* Find closest power of two >= to size, capped at 2^max_size_log */ -static __u32 roundup_pow2_max(__u32 size, int max_size_log) +static size_t btf_dedup_identity_hash_fn(const void *key, void *ctx) { - int i; + return (size_t)key; +} - for (i = 0; i < max_size_log && (1U << i) < size; i++) - ; - return 1U << i; +static size_t btf_dedup_collision_hash_fn(const void *key, void *ctx) +{ + return 0; } +static bool btf_dedup_equal_fn(const void *k1, const void *k2, void *ctx) +{ + return k1 == k2; +} static struct btf_dedup *btf_dedup_new(struct btf *btf, struct btf_ext *btf_ext, const struct btf_dedup_opts *opts) { struct btf_dedup *d = calloc(1, sizeof(struct btf_dedup)); + hashmap_hash_fn hash_fn = btf_dedup_identity_hash_fn; int i, err = 0; - __u32 sz; if (!d) return ERR_PTR(-ENOMEM); d->opts.dont_resolve_fwds = opts && opts->dont_resolve_fwds; - sz = opts && opts->dedup_table_size ? opts->dedup_table_size - : BTF_DEDUP_TABLE_DEFAULT_SIZE; - sz = roundup_pow2_max(sz, BTF_DEDUP_TABLE_MAX_SIZE_LOG); - d->opts.dedup_table_size = sz; + /* dedup_table_size is now used only to force collisions in tests */ + if (opts && opts->dedup_table_size == 1) + hash_fn = btf_dedup_collision_hash_fn; d->btf = btf; d->btf_ext = btf_ext; - d->dedup_table = calloc(d->opts.dedup_table_size, - sizeof(struct btf_dedup_node *)); - if (!d->dedup_table) { - err = -ENOMEM; + d->dedup_table = hashmap__new(hash_fn, btf_dedup_equal_fn, NULL); + if (IS_ERR(d->dedup_table)) { + err = PTR_ERR(d->dedup_table); + d->dedup_table = NULL; goto done; } @@ -1781,9 +1740,9 @@ static int btf_dedup_strings(struct btf_dedup *d) return err; } -static __u32 btf_hash_common(struct btf_type *t) +static long btf_hash_common(struct btf_type *t) { - __u32 h; + long h; h = hash_combine(0, t->name_off); h = hash_combine(h, t->info); @@ -1799,10 +1758,10 @@ static bool btf_equal_common(struct btf_type *t1, struct btf_type *t2) } /* Calculate type signature hash of INT. */ -static __u32 btf_hash_int(struct btf_type *t) +static long btf_hash_int(struct btf_type *t) { __u32 info = *(__u32 *)(t + 1); - __u32 h; + long h; h = btf_hash_common(t); h = hash_combine(h, info); @@ -1822,9 +1781,9 @@ static bool btf_equal_int(struct btf_type *t1, struct btf_type *t2) } /* Calculate type signature hash of ENUM. */ -static __u32 btf_hash_enum(struct btf_type *t) +static long btf_hash_enum(struct btf_type *t) { - __u32 h; + long h; /* don't hash vlen and enum members to support enum fwd resolving */ h = hash_combine(0, t->name_off); @@ -1876,11 +1835,11 @@ static bool btf_compat_enum(struct btf_type *t1, struct btf_type *t2) * as referenced type IDs equivalence is established separately during type * graph equivalence check algorithm. */ -static __u32 btf_hash_struct(struct btf_type *t) +static long btf_hash_struct(struct btf_type *t) { struct btf_member *member = (struct btf_member *)(t + 1); __u32 vlen = BTF_INFO_VLEN(t->info); - __u32 h = btf_hash_common(t); + long h = btf_hash_common(t); int i; for (i = 0; i < vlen; i++) { @@ -1923,10 +1882,10 @@ static bool btf_shallow_equal_struct(struct btf_type *t1, struct btf_type *t2) * under assumption that they were already resolved to canonical type IDs and * are not going to change. */ -static __u32 btf_hash_array(struct btf_type *t) +static long btf_hash_array(struct btf_type *t) { struct btf_array *info = (struct btf_array *)(t + 1); - __u32 h = btf_hash_common(t); + long h = btf_hash_common(t); h = hash_combine(h, info->type); h = hash_combine(h, info->index_type); @@ -1977,11 +1936,11 @@ static bool btf_compat_array(struct btf_type *t1, struct btf_type *t2) * under assumption that they were already resolved to canonical type IDs and * are not going to change. */ -static inline __u32 btf_hash_fnproto(struct btf_type *t) +static long btf_hash_fnproto(struct btf_type *t) { struct btf_param *member = (struct btf_param *)(t + 1); __u16 vlen = BTF_INFO_VLEN(t->info); - __u32 h = btf_hash_common(t); + long h = btf_hash_common(t); int i; for (i = 0; i < vlen; i++) { @@ -1999,7 +1958,7 @@ static inline __u32 btf_hash_fnproto(struct btf_type *t) * This function is called during reference types deduplication to compare * FUNC_PROTO to potential canonical representative. */ -static inline bool btf_equal_fnproto(struct btf_type *t1, struct btf_type *t2) +static bool btf_equal_fnproto(struct btf_type *t1, struct btf_type *t2) { struct btf_param *m1, *m2; __u16 vlen; @@ -2025,7 +1984,7 @@ static inline bool btf_equal_fnproto(struct btf_type *t1, struct btf_type *t2) * IDs. This check is performed during type graph equivalence check and * referenced types equivalence is checked separately. */ -static inline bool btf_compat_fnproto(struct btf_type *t1, struct btf_type *t2) +static bool btf_compat_fnproto(struct btf_type *t1, struct btf_type *t2) { struct btf_param *m1, *m2; __u16 vlen; @@ -2056,11 +2015,12 @@ static inline bool btf_compat_fnproto(struct btf_type *t1, struct btf_type *t2) static int btf_dedup_prim_type(struct btf_dedup *d, __u32 type_id) { struct btf_type *t = d->btf->types[type_id]; + struct hashmap_entry *hash_entry; struct btf_type *cand; - struct btf_dedup_node *cand_node; /* if we don't find equivalent type, then we are canonical */ __u32 new_id = type_id; - __u32 h; + __u32 cand_id; + long h; switch (BTF_INFO_KIND(t->info)) { case BTF_KIND_CONST: @@ -2077,10 +2037,11 @@ static int btf_dedup_prim_type(struct btf_dedup *d, __u32 type_id) case BTF_KIND_INT: h = btf_hash_int(t); - for_each_dedup_cand(d, h, cand_node) { - cand = d->btf->types[cand_node->type_id]; + for_each_dedup_cand(d, hash_entry, h) { + cand_id = (__u32)(long)hash_entry->value; + cand = d->btf->types[cand_id]; if (btf_equal_int(t, cand)) { - new_id = cand_node->type_id; + new_id = cand_id; break; } } @@ -2088,10 +2049,11 @@ static int btf_dedup_prim_type(struct btf_dedup *d, __u32 type_id) case BTF_KIND_ENUM: h = btf_hash_enum(t); - for_each_dedup_cand(d, h, cand_node) { - cand = d->btf->types[cand_node->type_id]; + for_each_dedup_cand(d, hash_entry, h) { + cand_id = (__u32)(long)hash_entry->value; + cand = d->btf->types[cand_id]; if (btf_equal_enum(t, cand)) { - new_id = cand_node->type_id; + new_id = cand_id; break; } if (d->opts.dont_resolve_fwds) @@ -2099,21 +2061,22 @@ static int btf_dedup_prim_type(struct btf_dedup *d, __u32 type_id) if (btf_compat_enum(t, cand)) { if (btf_is_enum_fwd(t)) { /* resolve fwd to full enum */ - new_id = cand_node->type_id; + new_id = cand_id; break; } /* resolve canonical enum fwd to full enum */ - d->map[cand_node->type_id] = type_id; + d->map[cand_id] = type_id; } } break; case BTF_KIND_FWD: h = btf_hash_common(t); - for_each_dedup_cand(d, h, cand_node) { - cand = d->btf->types[cand_node->type_id]; + for_each_dedup_cand(d, hash_entry, h) { + cand_id = (__u32)(long)hash_entry->value; + cand = d->btf->types[cand_id]; if (btf_equal_common(t, cand)) { - new_id = cand_node->type_id; + new_id = cand_id; break; } } @@ -2514,12 +2477,12 @@ static void btf_dedup_merge_hypot_map(struct btf_dedup *d) */ static int btf_dedup_struct_type(struct btf_dedup *d, __u32 type_id) { - struct btf_dedup_node *cand_node; struct btf_type *cand_type, *t; + struct hashmap_entry *hash_entry; /* if we don't find equivalent type, then we are canonical */ __u32 new_id = type_id; __u16 kind; - __u32 h; + long h; /* already deduped or is in process of deduping (loop detected) */ if (d->map[type_id] <= BTF_MAX_NR_TYPES) @@ -2532,7 +2495,8 @@ static int btf_dedup_struct_type(struct btf_dedup *d, __u32 type_id) return 0; h = btf_hash_struct(t); - for_each_dedup_cand(d, h, cand_node) { + for_each_dedup_cand(d, hash_entry, h) { + __u32 cand_id = (__u32)(long)hash_entry->value; int eq; /* @@ -2545,17 +2509,17 @@ static int btf_dedup_struct_type(struct btf_dedup *d, __u32 type_id) * creating a loop (FWD -> STRUCT and STRUCT -> FWD), because * FWD and compatible STRUCT/UNION are considered equivalent. */ - cand_type = d->btf->types[cand_node->type_id]; + cand_type = d->btf->types[cand_id]; if (!btf_shallow_equal_struct(t, cand_type)) continue; btf_dedup_clear_hypot_map(d); - eq = btf_dedup_is_equiv(d, type_id, cand_node->type_id); + eq = btf_dedup_is_equiv(d, type_id, cand_id); if (eq < 0) return eq; if (!eq) continue; - new_id = cand_node->type_id; + new_id = cand_id; btf_dedup_merge_hypot_map(d); break; } @@ -2605,12 +2569,12 @@ static int btf_dedup_struct_types(struct btf_dedup *d) */ static int btf_dedup_ref_type(struct btf_dedup *d, __u32 type_id) { - struct btf_dedup_node *cand_node; + struct hashmap_entry *hash_entry; + __u32 new_id = type_id, cand_id; struct btf_type *t, *cand; /* if we don't find equivalent type, then we are representative type */ - __u32 new_id = type_id; int ref_type_id; - __u32 h; + long h; if (d->map[type_id] == BTF_IN_PROGRESS_ID) return -ELOOP; @@ -2633,10 +2597,11 @@ static int btf_dedup_ref_type(struct btf_dedup *d, __u32 type_id) t->type = ref_type_id; h = btf_hash_common(t); - for_each_dedup_cand(d, h, cand_node) { - cand = d->btf->types[cand_node->type_id]; + for_each_dedup_cand(d, hash_entry, h) { + cand_id = (__u32)(long)hash_entry->value; + cand = d->btf->types[cand_id]; if (btf_equal_common(t, cand)) { - new_id = cand_node->type_id; + new_id = cand_id; break; } } @@ -2656,10 +2621,11 @@ static int btf_dedup_ref_type(struct btf_dedup *d, __u32 type_id) info->index_type = ref_type_id; h = btf_hash_array(t); - for_each_dedup_cand(d, h, cand_node) { - cand = d->btf->types[cand_node->type_id]; + for_each_dedup_cand(d, hash_entry, h) { + cand_id = (__u32)(long)hash_entry->value; + cand = d->btf->types[cand_id]; if (btf_equal_array(t, cand)) { - new_id = cand_node->type_id; + new_id = cand_id; break; } } @@ -2687,10 +2653,11 @@ static int btf_dedup_ref_type(struct btf_dedup *d, __u32 type_id) } h = btf_hash_fnproto(t); - for_each_dedup_cand(d, h, cand_node) { - cand = d->btf->types[cand_node->type_id]; + for_each_dedup_cand(d, hash_entry, h) { + cand_id = (__u32)(long)hash_entry->value; + cand = d->btf->types[cand_id]; if (btf_equal_fnproto(t, cand)) { - new_id = cand_node->type_id; + new_id = cand_id; break; } } @@ -2717,7 +2684,9 @@ static int btf_dedup_ref_types(struct btf_dedup *d) if (err < 0) return err; } - btf_dedup_table_free(d); + /* we won't need d->dedup_table anymore */ + hashmap__free(d->dedup_table); + d->dedup_table = NULL; return 0; } -- Gitee From 8c2ac7e7556919fc02e4ae355cbbc2663ecaacc8 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:59:03 -0700 Subject: [PATCH 14/24] libbpf: add btf_dump API for BTF-to-C conversion ANBZ: #5530 commit 351131b51c7a27daf0fbdce80b619b8d130374c6 upstream. BTF contains enough type information to allow generating valid compilable C header w/ correct layout of structs/unions and all the typedef/enum definitions. This patch adds a new "object" - btf_dump to facilitate dumping BTF as valid C. btf_dump__dump_type() is the main API which takes care of dumping out (through user-provided printf-like callback function) C definitions for given type ID and it's required dependencies. This allows for not just dumping out entirety of BTF types, but also selective filtering based on user-provided criterias w/ minimal set of dependent types. Signed-off-by: Andrii Nakryiko Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/lib/bpf/Build | 4 +- tools/lib/bpf/btf.h | 17 + tools/lib/bpf/btf_dump.c | 1336 ++++++++++++++++++++++++++++++++++++++ tools/lib/bpf/libbpf.map | 3 + 4 files changed, 1359 insertions(+), 1 deletion(-) create mode 100644 tools/lib/bpf/btf_dump.c diff --git a/tools/lib/bpf/Build b/tools/lib/bpf/Build index dcf0d36598e0..e3962cfbc9a6 100644 --- a/tools/lib/bpf/Build +++ b/tools/lib/bpf/Build @@ -1 +1,3 @@ -libbpf-y := libbpf.o bpf.o nlattr.o btf.o libbpf_errno.o str_error.o netlink.o bpf_prog_linfo.o libbpf_probes.o xsk.o hashmap.o +libbpf-y := libbpf.o bpf.o nlattr.o btf.o libbpf_errno.o str_error.o \ + netlink.o bpf_prog_linfo.o libbpf_probes.o xsk.o hashmap.o \ + btf_dump.o diff --git a/tools/lib/bpf/btf.h b/tools/lib/bpf/btf.h index bded210df9e8..ba4ffa831aa4 100644 --- a/tools/lib/bpf/btf.h +++ b/tools/lib/bpf/btf.h @@ -4,6 +4,7 @@ #ifndef __LIBBPF_BTF_H #define __LIBBPF_BTF_H +#include #include #ifdef __cplusplus @@ -102,6 +103,22 @@ struct btf_dedup_opts { LIBBPF_API int btf__dedup(struct btf *btf, struct btf_ext *btf_ext, const struct btf_dedup_opts *opts); +struct btf_dump; + +struct btf_dump_opts { + void *ctx; +}; + +typedef void (*btf_dump_printf_fn_t)(void *ctx, const char *fmt, va_list args); + +LIBBPF_API struct btf_dump *btf_dump__new(const struct btf *btf, + const struct btf_ext *btf_ext, + const struct btf_dump_opts *opts, + btf_dump_printf_fn_t printf_fn); +LIBBPF_API void btf_dump__free(struct btf_dump *d); + +LIBBPF_API int btf_dump__dump_type(struct btf_dump *d, __u32 id); + #ifdef __cplusplus } /* extern "C" */ #endif diff --git a/tools/lib/bpf/btf_dump.c b/tools/lib/bpf/btf_dump.c new file mode 100644 index 000000000000..4b22db77e2cc --- /dev/null +++ b/tools/lib/bpf/btf_dump.c @@ -0,0 +1,1336 @@ +// SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) + +/* + * BTF-to-C type converter. + * + * Copyright (c) 2019 Facebook + */ + +#include +#include +#include +#include +#include +#include +#include +#include "btf.h" +#include "hashmap.h" +#include "libbpf.h" +#include "libbpf_internal.h" + +#define min(x, y) ((x) < (y) ? (x) : (y)) +#define max(x, y) ((x) < (y) ? (y) : (x)) + +static const char PREFIXES[] = "\t\t\t\t\t\t\t\t\t\t\t\t\t"; +static const size_t PREFIX_CNT = sizeof(PREFIXES) - 1; + +static const char *pfx(int lvl) +{ + return lvl >= PREFIX_CNT ? PREFIXES : &PREFIXES[PREFIX_CNT - lvl]; +} + +enum btf_dump_type_order_state { + NOT_ORDERED, + ORDERING, + ORDERED, +}; + +enum btf_dump_type_emit_state { + NOT_EMITTED, + EMITTING, + EMITTED, +}; + +/* per-type auxiliary state */ +struct btf_dump_type_aux_state { + /* topological sorting state */ + enum btf_dump_type_order_state order_state: 2; + /* emitting state used to determine the need for forward declaration */ + enum btf_dump_type_emit_state emit_state: 2; + /* whether forward declaration was already emitted */ + __u8 fwd_emitted: 1; + /* whether unique non-duplicate name was already assigned */ + __u8 name_resolved: 1; +}; + +struct btf_dump { + const struct btf *btf; + const struct btf_ext *btf_ext; + btf_dump_printf_fn_t printf_fn; + struct btf_dump_opts opts; + + /* per-type auxiliary state */ + struct btf_dump_type_aux_state *type_states; + /* per-type optional cached unique name, must be freed, if present */ + const char **cached_names; + + /* topo-sorted list of dependent type definitions */ + __u32 *emit_queue; + int emit_queue_cap; + int emit_queue_cnt; + + /* + * stack of type declarations (e.g., chain of modifiers, arrays, + * funcs, etc) + */ + __u32 *decl_stack; + int decl_stack_cap; + int decl_stack_cnt; + + /* maps struct/union/enum name to a number of name occurrences */ + struct hashmap *type_names; + /* + * maps typedef identifiers and enum value names to a number of such + * name occurrences + */ + struct hashmap *ident_names; +}; + +static size_t str_hash_fn(const void *key, void *ctx) +{ + const char *s = key; + size_t h = 0; + + while (*s) { + h = h * 31 + *s; + s++; + } + return h; +} + +static bool str_equal_fn(const void *a, const void *b, void *ctx) +{ + return strcmp(a, b) == 0; +} + +static __u16 btf_kind_of(const struct btf_type *t) +{ + return BTF_INFO_KIND(t->info); +} + +static __u16 btf_vlen_of(const struct btf_type *t) +{ + return BTF_INFO_VLEN(t->info); +} + +static bool btf_kflag_of(const struct btf_type *t) +{ + return BTF_INFO_KFLAG(t->info); +} + +static const char *btf_name_of(const struct btf_dump *d, __u32 name_off) +{ + return btf__name_by_offset(d->btf, name_off); +} + +static void btf_dump_printf(const struct btf_dump *d, const char *fmt, ...) +{ + va_list args; + + va_start(args, fmt); + d->printf_fn(d->opts.ctx, fmt, args); + va_end(args); +} + +struct btf_dump *btf_dump__new(const struct btf *btf, + const struct btf_ext *btf_ext, + const struct btf_dump_opts *opts, + btf_dump_printf_fn_t printf_fn) +{ + struct btf_dump *d; + int err; + + d = calloc(1, sizeof(struct btf_dump)); + if (!d) + return ERR_PTR(-ENOMEM); + + d->btf = btf; + d->btf_ext = btf_ext; + d->printf_fn = printf_fn; + d->opts.ctx = opts ? opts->ctx : NULL; + + d->type_names = hashmap__new(str_hash_fn, str_equal_fn, NULL); + if (IS_ERR(d->type_names)) { + err = PTR_ERR(d->type_names); + d->type_names = NULL; + btf_dump__free(d); + return ERR_PTR(err); + } + d->ident_names = hashmap__new(str_hash_fn, str_equal_fn, NULL); + if (IS_ERR(d->ident_names)) { + err = PTR_ERR(d->ident_names); + d->ident_names = NULL; + btf_dump__free(d); + return ERR_PTR(err); + } + + return d; +} + +void btf_dump__free(struct btf_dump *d) +{ + int i, cnt; + + if (!d) + return; + + free(d->type_states); + if (d->cached_names) { + /* any set cached name is owned by us and should be freed */ + for (i = 0, cnt = btf__get_nr_types(d->btf); i <= cnt; i++) { + if (d->cached_names[i]) + free((void *)d->cached_names[i]); + } + } + free(d->cached_names); + free(d->emit_queue); + free(d->decl_stack); + hashmap__free(d->type_names); + hashmap__free(d->ident_names); + + free(d); +} + +static int btf_dump_order_type(struct btf_dump *d, __u32 id, bool through_ptr); +static void btf_dump_emit_type(struct btf_dump *d, __u32 id, __u32 cont_id); + +/* + * Dump BTF type in a compilable C syntax, including all the necessary + * dependent types, necessary for compilation. If some of the dependent types + * were already emitted as part of previous btf_dump__dump_type() invocation + * for another type, they won't be emitted again. This API allows callers to + * filter out BTF types according to user-defined criterias and emitted only + * minimal subset of types, necessary to compile everything. Full struct/union + * definitions will still be emitted, even if the only usage is through + * pointer and could be satisfied with just a forward declaration. + * + * Dumping is done in two high-level passes: + * 1. Topologically sort type definitions to satisfy C rules of compilation. + * 2. Emit type definitions in C syntax. + * + * Returns 0 on success; <0, otherwise. + */ +int btf_dump__dump_type(struct btf_dump *d, __u32 id) +{ + int err, i; + + if (id > btf__get_nr_types(d->btf)) + return -EINVAL; + + /* type states are lazily allocated, as they might not be needed */ + if (!d->type_states) { + d->type_states = calloc(1 + btf__get_nr_types(d->btf), + sizeof(d->type_states[0])); + if (!d->type_states) + return -ENOMEM; + d->cached_names = calloc(1 + btf__get_nr_types(d->btf), + sizeof(d->cached_names[0])); + if (!d->cached_names) + return -ENOMEM; + + /* VOID is special */ + d->type_states[0].order_state = ORDERED; + d->type_states[0].emit_state = EMITTED; + } + + d->emit_queue_cnt = 0; + err = btf_dump_order_type(d, id, false); + if (err < 0) + return err; + + for (i = 0; i < d->emit_queue_cnt; i++) + btf_dump_emit_type(d, d->emit_queue[i], 0 /*top-level*/); + + return 0; +} + +static int btf_dump_add_emit_queue_id(struct btf_dump *d, __u32 id) +{ + __u32 *new_queue; + size_t new_cap; + + if (d->emit_queue_cnt >= d->emit_queue_cap) { + new_cap = max(16, d->emit_queue_cap * 3 / 2); + new_queue = realloc(d->emit_queue, + new_cap * sizeof(new_queue[0])); + if (!new_queue) + return -ENOMEM; + d->emit_queue = new_queue; + d->emit_queue_cap = new_cap; + } + + d->emit_queue[d->emit_queue_cnt++] = id; + return 0; +} + +/* + * Determine order of emitting dependent types and specified type to satisfy + * C compilation rules. This is done through topological sorting with an + * additional complication which comes from C rules. The main idea for C is + * that if some type is "embedded" into a struct/union, it's size needs to be + * known at the time of definition of containing type. E.g., for: + * + * struct A {}; + * struct B { struct A x; } + * + * struct A *HAS* to be defined before struct B, because it's "embedded", + * i.e., it is part of struct B layout. But in the following case: + * + * struct A; + * struct B { struct A *x; } + * struct A {}; + * + * it's enough to just have a forward declaration of struct A at the time of + * struct B definition, as struct B has a pointer to struct A, so the size of + * field x is known without knowing struct A size: it's sizeof(void *). + * + * Unfortunately, there are some trickier cases we need to handle, e.g.: + * + * struct A {}; // if this was forward-declaration: compilation error + * struct B { + * struct { // anonymous struct + * struct A y; + * } *x; + * }; + * + * In this case, struct B's field x is a pointer, so it's size is known + * regardless of the size of (anonymous) struct it points to. But because this + * struct is anonymous and thus defined inline inside struct B, *and* it + * embeds struct A, compiler requires full definition of struct A to be known + * before struct B can be defined. This creates a transitive dependency + * between struct A and struct B. If struct A was forward-declared before + * struct B definition and fully defined after struct B definition, that would + * trigger compilation error. + * + * All this means that while we are doing topological sorting on BTF type + * graph, we need to determine relationships between different types (graph + * nodes): + * - weak link (relationship) between X and Y, if Y *CAN* be + * forward-declared at the point of X definition; + * - strong link, if Y *HAS* to be fully-defined before X can be defined. + * + * The rule is as follows. Given a chain of BTF types from X to Y, if there is + * BTF_KIND_PTR type in the chain and at least one non-anonymous type + * Z (excluding X, including Y), then link is weak. Otherwise, it's strong. + * Weak/strong relationship is determined recursively during DFS traversal and + * is returned as a result from btf_dump_order_type(). + * + * btf_dump_order_type() is trying to avoid unnecessary forward declarations, + * but it is not guaranteeing that no extraneous forward declarations will be + * emitted. + * + * To avoid extra work, algorithm marks some of BTF types as ORDERED, when + * it's done with them, but not for all (e.g., VOLATILE, CONST, RESTRICT, + * ARRAY, FUNC_PROTO), as weak/strong semantics for those depends on the + * entire graph path, so depending where from one came to that BTF type, it + * might cause weak or strong ordering. For types like STRUCT/UNION/INT/ENUM, + * once they are processed, there is no need to do it again, so they are + * marked as ORDERED. We can mark PTR as ORDERED as well, as it semi-forces + * weak link, unless subsequent referenced STRUCT/UNION/ENUM is anonymous. But + * in any case, once those are processed, no need to do it again, as the + * result won't change. + * + * Returns: + * - 1, if type is part of strong link (so there is strong topological + * ordering requirements); + * - 0, if type is part of weak link (so can be satisfied through forward + * declaration); + * - <0, on error (e.g., unsatisfiable type loop detected). + */ +static int btf_dump_order_type(struct btf_dump *d, __u32 id, bool through_ptr) +{ + /* + * Order state is used to detect strong link cycles, but only for BTF + * kinds that are or could be an independent definition (i.e., + * stand-alone fwd decl, enum, typedef, struct, union). Ptrs, arrays, + * func_protos, modifiers are just means to get to these definitions. + * Int/void don't need definitions, they are assumed to be always + * properly defined. We also ignore datasec, var, and funcs for now. + * So for all non-defining kinds, we never even set ordering state, + * for defining kinds we set ORDERING and subsequently ORDERED if it + * forms a strong link. + */ + struct btf_dump_type_aux_state *tstate = &d->type_states[id]; + const struct btf_type *t; + __u16 kind, vlen; + int err, i; + + /* return true, letting typedefs know that it's ok to be emitted */ + if (tstate->order_state == ORDERED) + return 1; + + t = btf__type_by_id(d->btf, id); + kind = btf_kind_of(t); + + if (tstate->order_state == ORDERING) { + /* type loop, but resolvable through fwd declaration */ + if ((kind == BTF_KIND_STRUCT || kind == BTF_KIND_UNION) && + through_ptr && t->name_off != 0) + return 0; + pr_warning("unsatisfiable type cycle, id:[%u]\n", id); + return -ELOOP; + } + + switch (kind) { + case BTF_KIND_INT: + tstate->order_state = ORDERED; + return 0; + + case BTF_KIND_PTR: + err = btf_dump_order_type(d, t->type, true); + tstate->order_state = ORDERED; + return err; + + case BTF_KIND_ARRAY: { + const struct btf_array *a = (void *)(t + 1); + + return btf_dump_order_type(d, a->type, through_ptr); + } + case BTF_KIND_STRUCT: + case BTF_KIND_UNION: { + const struct btf_member *m = (void *)(t + 1); + /* + * struct/union is part of strong link, only if it's embedded + * (so no ptr in a path) or it's anonymous (so has to be + * defined inline, even if declared through ptr) + */ + if (through_ptr && t->name_off != 0) + return 0; + + tstate->order_state = ORDERING; + + vlen = btf_vlen_of(t); + for (i = 0; i < vlen; i++, m++) { + err = btf_dump_order_type(d, m->type, false); + if (err < 0) + return err; + } + + if (t->name_off != 0) { + err = btf_dump_add_emit_queue_id(d, id); + if (err < 0) + return err; + } + + tstate->order_state = ORDERED; + return 1; + } + case BTF_KIND_ENUM: + case BTF_KIND_FWD: + if (t->name_off != 0) { + err = btf_dump_add_emit_queue_id(d, id); + if (err) + return err; + } + tstate->order_state = ORDERED; + return 1; + + case BTF_KIND_TYPEDEF: { + int is_strong; + + is_strong = btf_dump_order_type(d, t->type, through_ptr); + if (is_strong < 0) + return is_strong; + + /* typedef is similar to struct/union w.r.t. fwd-decls */ + if (through_ptr && !is_strong) + return 0; + + /* typedef is always a named definition */ + err = btf_dump_add_emit_queue_id(d, id); + if (err) + return err; + + d->type_states[id].order_state = ORDERED; + return 1; + } + case BTF_KIND_VOLATILE: + case BTF_KIND_CONST: + case BTF_KIND_RESTRICT: + return btf_dump_order_type(d, t->type, through_ptr); + + case BTF_KIND_FUNC_PROTO: { + const struct btf_param *p = (void *)(t + 1); + bool is_strong; + + err = btf_dump_order_type(d, t->type, through_ptr); + if (err < 0) + return err; + is_strong = err > 0; + + vlen = btf_vlen_of(t); + for (i = 0; i < vlen; i++, p++) { + err = btf_dump_order_type(d, p->type, through_ptr); + if (err < 0) + return err; + if (err > 0) + is_strong = true; + } + return is_strong; + } + case BTF_KIND_FUNC: + case BTF_KIND_VAR: + case BTF_KIND_DATASEC: + d->type_states[id].order_state = ORDERED; + return 0; + + default: + return -EINVAL; + } +} + +static void btf_dump_emit_struct_fwd(struct btf_dump *d, __u32 id, + const struct btf_type *t); +static void btf_dump_emit_struct_def(struct btf_dump *d, __u32 id, + const struct btf_type *t, int lvl); + +static void btf_dump_emit_enum_fwd(struct btf_dump *d, __u32 id, + const struct btf_type *t); +static void btf_dump_emit_enum_def(struct btf_dump *d, __u32 id, + const struct btf_type *t, int lvl); + +static void btf_dump_emit_fwd_def(struct btf_dump *d, __u32 id, + const struct btf_type *t); + +static void btf_dump_emit_typedef_def(struct btf_dump *d, __u32 id, + const struct btf_type *t, int lvl); + +/* a local view into a shared stack */ +struct id_stack { + const __u32 *ids; + int cnt; +}; + +static void btf_dump_emit_type_decl(struct btf_dump *d, __u32 id, + const char *fname, int lvl); +static void btf_dump_emit_type_chain(struct btf_dump *d, + struct id_stack *decl_stack, + const char *fname, int lvl); + +static const char *btf_dump_type_name(struct btf_dump *d, __u32 id); +static const char *btf_dump_ident_name(struct btf_dump *d, __u32 id); +static size_t btf_dump_name_dups(struct btf_dump *d, struct hashmap *name_map, + const char *orig_name); + +static bool btf_dump_is_blacklisted(struct btf_dump *d, __u32 id) +{ + const struct btf_type *t = btf__type_by_id(d->btf, id); + + /* __builtin_va_list is a compiler built-in, which causes compilation + * errors, when compiling w/ different compiler, then used to compile + * original code (e.g., GCC to compile kernel, Clang to use generated + * C header from BTF). As it is built-in, it should be already defined + * properly internally in compiler. + */ + if (t->name_off == 0) + return false; + return strcmp(btf_name_of(d, t->name_off), "__builtin_va_list") == 0; +} + +/* + * Emit C-syntax definitions of types from chains of BTF types. + * + * High-level handling of determining necessary forward declarations are handled + * by btf_dump_emit_type() itself, but all nitty-gritty details of emitting type + * declarations/definitions in C syntax are handled by a combo of + * btf_dump_emit_type_decl()/btf_dump_emit_type_chain() w/ delegation to + * corresponding btf_dump_emit_*_{def,fwd}() functions. + * + * We also keep track of "containing struct/union type ID" to determine when + * we reference it from inside and thus can avoid emitting unnecessary forward + * declaration. + * + * This algorithm is designed in such a way, that even if some error occurs + * (either technical, e.g., out of memory, or logical, i.e., malformed BTF + * that doesn't comply to C rules completely), algorithm will try to proceed + * and produce as much meaningful output as possible. + */ +static void btf_dump_emit_type(struct btf_dump *d, __u32 id, __u32 cont_id) +{ + struct btf_dump_type_aux_state *tstate = &d->type_states[id]; + bool top_level_def = cont_id == 0; + const struct btf_type *t; + __u16 kind; + + if (tstate->emit_state == EMITTED) + return; + + t = btf__type_by_id(d->btf, id); + kind = btf_kind_of(t); + + if (top_level_def && t->name_off == 0) { + pr_warning("unexpected nameless definition, id:[%u]\n", id); + return; + } + + if (tstate->emit_state == EMITTING) { + if (tstate->fwd_emitted) + return; + + switch (kind) { + case BTF_KIND_STRUCT: + case BTF_KIND_UNION: + /* + * if we are referencing a struct/union that we are + * part of - then no need for fwd declaration + */ + if (id == cont_id) + return; + if (t->name_off == 0) { + pr_warning("anonymous struct/union loop, id:[%u]\n", + id); + return; + } + btf_dump_emit_struct_fwd(d, id, t); + btf_dump_printf(d, ";\n\n"); + tstate->fwd_emitted = 1; + break; + case BTF_KIND_TYPEDEF: + /* + * for typedef fwd_emitted means typedef definition + * was emitted, but it can be used only for "weak" + * references through pointer only, not for embedding + */ + if (!btf_dump_is_blacklisted(d, id)) { + btf_dump_emit_typedef_def(d, id, t, 0); + btf_dump_printf(d, ";\n\n"); + }; + tstate->fwd_emitted = 1; + break; + default: + break; + } + + return; + } + + switch (kind) { + case BTF_KIND_INT: + tstate->emit_state = EMITTED; + break; + case BTF_KIND_ENUM: + if (top_level_def) { + btf_dump_emit_enum_def(d, id, t, 0); + btf_dump_printf(d, ";\n\n"); + } + tstate->emit_state = EMITTED; + break; + case BTF_KIND_PTR: + case BTF_KIND_VOLATILE: + case BTF_KIND_CONST: + case BTF_KIND_RESTRICT: + btf_dump_emit_type(d, t->type, cont_id); + break; + case BTF_KIND_ARRAY: { + const struct btf_array *a = (void *)(t + 1); + + btf_dump_emit_type(d, a->type, cont_id); + break; + } + case BTF_KIND_FWD: + btf_dump_emit_fwd_def(d, id, t); + btf_dump_printf(d, ";\n\n"); + tstate->emit_state = EMITTED; + break; + case BTF_KIND_TYPEDEF: + tstate->emit_state = EMITTING; + btf_dump_emit_type(d, t->type, id); + /* + * typedef can server as both definition and forward + * declaration; at this stage someone depends on + * typedef as a forward declaration (refers to it + * through pointer), so unless we already did it, + * emit typedef as a forward declaration + */ + if (!tstate->fwd_emitted && !btf_dump_is_blacklisted(d, id)) { + btf_dump_emit_typedef_def(d, id, t, 0); + btf_dump_printf(d, ";\n\n"); + } + tstate->emit_state = EMITTED; + break; + case BTF_KIND_STRUCT: + case BTF_KIND_UNION: + tstate->emit_state = EMITTING; + /* if it's a top-level struct/union definition or struct/union + * is anonymous, then in C we'll be emitting all fields and + * their types (as opposed to just `struct X`), so we need to + * make sure that all types, referenced from struct/union + * members have necessary forward-declarations, where + * applicable + */ + if (top_level_def || t->name_off == 0) { + const struct btf_member *m = (void *)(t + 1); + __u16 vlen = btf_vlen_of(t); + int i, new_cont_id; + + new_cont_id = t->name_off == 0 ? cont_id : id; + for (i = 0; i < vlen; i++, m++) + btf_dump_emit_type(d, m->type, new_cont_id); + } else if (!tstate->fwd_emitted && id != cont_id) { + btf_dump_emit_struct_fwd(d, id, t); + btf_dump_printf(d, ";\n\n"); + tstate->fwd_emitted = 1; + } + + if (top_level_def) { + btf_dump_emit_struct_def(d, id, t, 0); + btf_dump_printf(d, ";\n\n"); + tstate->emit_state = EMITTED; + } else { + tstate->emit_state = NOT_EMITTED; + } + break; + case BTF_KIND_FUNC_PROTO: { + const struct btf_param *p = (void *)(t + 1); + __u16 vlen = btf_vlen_of(t); + int i; + + btf_dump_emit_type(d, t->type, cont_id); + for (i = 0; i < vlen; i++, p++) + btf_dump_emit_type(d, p->type, cont_id); + + break; + } + default: + break; + } +} + +static int btf_align_of(const struct btf *btf, __u32 id) +{ + const struct btf_type *t = btf__type_by_id(btf, id); + __u16 kind = btf_kind_of(t); + + switch (kind) { + case BTF_KIND_INT: + case BTF_KIND_ENUM: + return min(sizeof(void *), t->size); + case BTF_KIND_PTR: + return sizeof(void *); + case BTF_KIND_TYPEDEF: + case BTF_KIND_VOLATILE: + case BTF_KIND_CONST: + case BTF_KIND_RESTRICT: + return btf_align_of(btf, t->type); + case BTF_KIND_ARRAY: { + const struct btf_array *a = (void *)(t + 1); + + return btf_align_of(btf, a->type); + } + case BTF_KIND_STRUCT: + case BTF_KIND_UNION: { + const struct btf_member *m = (void *)(t + 1); + __u16 vlen = btf_vlen_of(t); + int i, align = 1; + + for (i = 0; i < vlen; i++, m++) + align = max(align, btf_align_of(btf, m->type)); + + return align; + } + default: + pr_warning("unsupported BTF_KIND:%u\n", btf_kind_of(t)); + return 1; + } +} + +static bool btf_is_struct_packed(const struct btf *btf, __u32 id, + const struct btf_type *t) +{ + const struct btf_member *m; + int align, i, bit_sz; + __u16 vlen; + bool kflag; + + align = btf_align_of(btf, id); + /* size of a non-packed struct has to be a multiple of its alignment*/ + if (t->size % align) + return true; + + m = (void *)(t + 1); + kflag = btf_kflag_of(t); + vlen = btf_vlen_of(t); + /* all non-bitfield fields have to be naturally aligned */ + for (i = 0; i < vlen; i++, m++) { + align = btf_align_of(btf, m->type); + bit_sz = kflag ? BTF_MEMBER_BITFIELD_SIZE(m->offset) : 0; + if (bit_sz == 0 && m->offset % (8 * align) != 0) + return true; + } + + /* + * if original struct was marked as packed, but its layout is + * naturally aligned, we'll detect that it's not packed + */ + return false; +} + +static int chip_away_bits(int total, int at_most) +{ + return total % at_most ? : at_most; +} + +static void btf_dump_emit_bit_padding(const struct btf_dump *d, + int cur_off, int m_off, int m_bit_sz, + int align, int lvl) +{ + int off_diff = m_off - cur_off; + int ptr_bits = sizeof(void *) * 8; + + if (off_diff <= 0) + /* no gap */ + return; + if (m_bit_sz == 0 && off_diff < align * 8) + /* natural padding will take care of a gap */ + return; + + while (off_diff > 0) { + const char *pad_type; + int pad_bits; + + if (ptr_bits > 32 && off_diff > 32) { + pad_type = "long"; + pad_bits = chip_away_bits(off_diff, ptr_bits); + } else if (off_diff > 16) { + pad_type = "int"; + pad_bits = chip_away_bits(off_diff, 32); + } else if (off_diff > 8) { + pad_type = "short"; + pad_bits = chip_away_bits(off_diff, 16); + } else { + pad_type = "char"; + pad_bits = chip_away_bits(off_diff, 8); + } + btf_dump_printf(d, "\n%s%s: %d;", pfx(lvl), pad_type, pad_bits); + off_diff -= pad_bits; + } +} + +static void btf_dump_emit_struct_fwd(struct btf_dump *d, __u32 id, + const struct btf_type *t) +{ + btf_dump_printf(d, "%s %s", + btf_kind_of(t) == BTF_KIND_STRUCT ? "struct" : "union", + btf_dump_type_name(d, id)); +} + +static void btf_dump_emit_struct_def(struct btf_dump *d, + __u32 id, + const struct btf_type *t, + int lvl) +{ + const struct btf_member *m = (void *)(t + 1); + bool kflag = btf_kflag_of(t), is_struct; + int align, i, packed, off = 0; + __u16 vlen = btf_vlen_of(t); + + is_struct = btf_kind_of(t) == BTF_KIND_STRUCT; + packed = is_struct ? btf_is_struct_packed(d->btf, id, t) : 0; + align = packed ? 1 : btf_align_of(d->btf, id); + + btf_dump_printf(d, "%s%s%s {", + is_struct ? "struct" : "union", + t->name_off ? " " : "", + btf_dump_type_name(d, id)); + + for (i = 0; i < vlen; i++, m++) { + const char *fname; + int m_off, m_sz; + + fname = btf_name_of(d, m->name_off); + m_sz = kflag ? BTF_MEMBER_BITFIELD_SIZE(m->offset) : 0; + m_off = kflag ? BTF_MEMBER_BIT_OFFSET(m->offset) : m->offset; + align = packed ? 1 : btf_align_of(d->btf, m->type); + + btf_dump_emit_bit_padding(d, off, m_off, m_sz, align, lvl + 1); + btf_dump_printf(d, "\n%s", pfx(lvl + 1)); + btf_dump_emit_type_decl(d, m->type, fname, lvl + 1); + + if (m_sz) { + btf_dump_printf(d, ": %d", m_sz); + off = m_off + m_sz; + } else { + m_sz = max(0, btf__resolve_size(d->btf, m->type)); + off = m_off + m_sz * 8; + } + btf_dump_printf(d, ";"); + } + + if (vlen) + btf_dump_printf(d, "\n"); + btf_dump_printf(d, "%s}", pfx(lvl)); + if (packed) + btf_dump_printf(d, " __attribute__((packed))"); +} + +static void btf_dump_emit_enum_fwd(struct btf_dump *d, __u32 id, + const struct btf_type *t) +{ + btf_dump_printf(d, "enum %s", btf_dump_type_name(d, id)); +} + +static void btf_dump_emit_enum_def(struct btf_dump *d, __u32 id, + const struct btf_type *t, + int lvl) +{ + const struct btf_enum *v = (void *)(t+1); + __u16 vlen = btf_vlen_of(t); + const char *name; + size_t dup_cnt; + int i; + + btf_dump_printf(d, "enum%s%s", + t->name_off ? " " : "", + btf_dump_type_name(d, id)); + + if (vlen) { + btf_dump_printf(d, " {"); + for (i = 0; i < vlen; i++, v++) { + name = btf_name_of(d, v->name_off); + /* enumerators share namespace with typedef idents */ + dup_cnt = btf_dump_name_dups(d, d->ident_names, name); + if (dup_cnt > 1) { + btf_dump_printf(d, "\n%s%s___%zu = %d,", + pfx(lvl + 1), name, dup_cnt, + (__s32)v->val); + } else { + btf_dump_printf(d, "\n%s%s = %d,", + pfx(lvl + 1), name, + (__s32)v->val); + } + } + btf_dump_printf(d, "\n%s}", pfx(lvl)); + } +} + +static void btf_dump_emit_fwd_def(struct btf_dump *d, __u32 id, + const struct btf_type *t) +{ + const char *name = btf_dump_type_name(d, id); + + if (btf_kflag_of(t)) + btf_dump_printf(d, "union %s", name); + else + btf_dump_printf(d, "struct %s", name); +} + +static void btf_dump_emit_typedef_def(struct btf_dump *d, __u32 id, + const struct btf_type *t, int lvl) +{ + const char *name = btf_dump_ident_name(d, id); + + btf_dump_printf(d, "typedef "); + btf_dump_emit_type_decl(d, t->type, name, lvl); +} + +static int btf_dump_push_decl_stack_id(struct btf_dump *d, __u32 id) +{ + __u32 *new_stack; + size_t new_cap; + + if (d->decl_stack_cnt >= d->decl_stack_cap) { + new_cap = max(16, d->decl_stack_cap * 3 / 2); + new_stack = realloc(d->decl_stack, + new_cap * sizeof(new_stack[0])); + if (!new_stack) + return -ENOMEM; + d->decl_stack = new_stack; + d->decl_stack_cap = new_cap; + } + + d->decl_stack[d->decl_stack_cnt++] = id; + + return 0; +} + +/* + * Emit type declaration (e.g., field type declaration in a struct or argument + * declaration in function prototype) in correct C syntax. + * + * For most types it's trivial, but there are few quirky type declaration + * cases worth mentioning: + * - function prototypes (especially nesting of function prototypes); + * - arrays; + * - const/volatile/restrict for pointers vs other types. + * + * For a good discussion of *PARSING* C syntax (as a human), see + * Peter van der Linden's "Expert C Programming: Deep C Secrets", + * Ch.3 "Unscrambling Declarations in C". + * + * It won't help with BTF to C conversion much, though, as it's an opposite + * problem. So we came up with this algorithm in reverse to van der Linden's + * parsing algorithm. It goes from structured BTF representation of type + * declaration to a valid compilable C syntax. + * + * For instance, consider this C typedef: + * typedef const int * const * arr[10] arr_t; + * It will be represented in BTF with this chain of BTF types: + * [typedef] -> [array] -> [ptr] -> [const] -> [ptr] -> [const] -> [int] + * + * Notice how [const] modifier always goes before type it modifies in BTF type + * graph, but in C syntax, const/volatile/restrict modifiers are written to + * the right of pointers, but to the left of other types. There are also other + * quirks, like function pointers, arrays of them, functions returning other + * functions, etc. + * + * We handle that by pushing all the types to a stack, until we hit "terminal" + * type (int/enum/struct/union/fwd). Then depending on the kind of a type on + * top of a stack, modifiers are handled differently. Array/function pointers + * have also wildly different syntax and how nesting of them are done. See + * code for authoritative definition. + * + * To avoid allocating new stack for each independent chain of BTF types, we + * share one bigger stack, with each chain working only on its own local view + * of a stack frame. Some care is required to "pop" stack frames after + * processing type declaration chain. + */ +static void btf_dump_emit_type_decl(struct btf_dump *d, __u32 id, + const char *fname, int lvl) +{ + struct id_stack decl_stack; + const struct btf_type *t; + int err, stack_start; + __u16 kind; + + stack_start = d->decl_stack_cnt; + for (;;) { + err = btf_dump_push_decl_stack_id(d, id); + if (err < 0) { + /* + * if we don't have enough memory for entire type decl + * chain, restore stack, emit warning, and try to + * proceed nevertheless + */ + pr_warning("not enough memory for decl stack:%d", err); + d->decl_stack_cnt = stack_start; + return; + } + + /* VOID */ + if (id == 0) + break; + + t = btf__type_by_id(d->btf, id); + kind = btf_kind_of(t); + switch (kind) { + case BTF_KIND_PTR: + case BTF_KIND_VOLATILE: + case BTF_KIND_CONST: + case BTF_KIND_RESTRICT: + case BTF_KIND_FUNC_PROTO: + id = t->type; + break; + case BTF_KIND_ARRAY: { + const struct btf_array *a = (void *)(t + 1); + + id = a->type; + break; + } + case BTF_KIND_INT: + case BTF_KIND_ENUM: + case BTF_KIND_FWD: + case BTF_KIND_STRUCT: + case BTF_KIND_UNION: + case BTF_KIND_TYPEDEF: + goto done; + default: + pr_warning("unexpected type in decl chain, kind:%u, id:[%u]\n", + kind, id); + goto done; + } + } +done: + /* + * We might be inside a chain of declarations (e.g., array of function + * pointers returning anonymous (so inlined) structs, having another + * array field). Each of those needs its own "stack frame" to handle + * emitting of declarations. Those stack frames are non-overlapping + * portions of shared btf_dump->decl_stack. To make it a bit nicer to + * handle this set of nested stacks, we create a view corresponding to + * our own "stack frame" and work with it as an independent stack. + * We'll need to clean up after emit_type_chain() returns, though. + */ + decl_stack.ids = d->decl_stack + stack_start; + decl_stack.cnt = d->decl_stack_cnt - stack_start; + btf_dump_emit_type_chain(d, &decl_stack, fname, lvl); + /* + * emit_type_chain() guarantees that it will pop its entire decl_stack + * frame before returning. But it works with a read-only view into + * decl_stack, so it doesn't actually pop anything from the + * perspective of shared btf_dump->decl_stack, per se. We need to + * reset decl_stack state to how it was before us to avoid it growing + * all the time. + */ + d->decl_stack_cnt = stack_start; +} + +static void btf_dump_emit_mods(struct btf_dump *d, struct id_stack *decl_stack) +{ + const struct btf_type *t; + __u32 id; + + while (decl_stack->cnt) { + id = decl_stack->ids[decl_stack->cnt - 1]; + t = btf__type_by_id(d->btf, id); + + switch (btf_kind_of(t)) { + case BTF_KIND_VOLATILE: + btf_dump_printf(d, "volatile "); + break; + case BTF_KIND_CONST: + btf_dump_printf(d, "const "); + break; + case BTF_KIND_RESTRICT: + btf_dump_printf(d, "restrict "); + break; + default: + return; + } + decl_stack->cnt--; + } +} + +static bool btf_is_mod_kind(const struct btf *btf, __u32 id) +{ + const struct btf_type *t = btf__type_by_id(btf, id); + + switch (btf_kind_of(t)) { + case BTF_KIND_VOLATILE: + case BTF_KIND_CONST: + case BTF_KIND_RESTRICT: + return true; + default: + return false; + } +} + +static void btf_dump_emit_name(const struct btf_dump *d, + const char *name, bool last_was_ptr) +{ + bool separate = name[0] && !last_was_ptr; + + btf_dump_printf(d, "%s%s", separate ? " " : "", name); +} + +static void btf_dump_emit_type_chain(struct btf_dump *d, + struct id_stack *decls, + const char *fname, int lvl) +{ + /* + * last_was_ptr is used to determine if we need to separate pointer + * asterisk (*) from previous part of type signature with space, so + * that we get `int ***`, instead of `int * * *`. We default to true + * for cases where we have single pointer in a chain. E.g., in ptr -> + * func_proto case. func_proto will start a new emit_type_chain call + * with just ptr, which should be emitted as (*) or (*), so we + * don't want to prepend space for that last pointer. + */ + bool last_was_ptr = true; + const struct btf_type *t; + const char *name; + __u16 kind; + __u32 id; + + while (decls->cnt) { + id = decls->ids[--decls->cnt]; + if (id == 0) { + /* VOID is a special snowflake */ + btf_dump_emit_mods(d, decls); + btf_dump_printf(d, "void"); + last_was_ptr = false; + continue; + } + + t = btf__type_by_id(d->btf, id); + kind = btf_kind_of(t); + + switch (kind) { + case BTF_KIND_INT: + btf_dump_emit_mods(d, decls); + name = btf_name_of(d, t->name_off); + btf_dump_printf(d, "%s", name); + break; + case BTF_KIND_STRUCT: + case BTF_KIND_UNION: + btf_dump_emit_mods(d, decls); + /* inline anonymous struct/union */ + if (t->name_off == 0) + btf_dump_emit_struct_def(d, id, t, lvl); + else + btf_dump_emit_struct_fwd(d, id, t); + break; + case BTF_KIND_ENUM: + btf_dump_emit_mods(d, decls); + /* inline anonymous enum */ + if (t->name_off == 0) + btf_dump_emit_enum_def(d, id, t, lvl); + else + btf_dump_emit_enum_fwd(d, id, t); + break; + case BTF_KIND_FWD: + btf_dump_emit_mods(d, decls); + btf_dump_emit_fwd_def(d, id, t); + break; + case BTF_KIND_TYPEDEF: + btf_dump_emit_mods(d, decls); + btf_dump_printf(d, "%s", btf_dump_ident_name(d, id)); + break; + case BTF_KIND_PTR: + btf_dump_printf(d, "%s", last_was_ptr ? "*" : " *"); + break; + case BTF_KIND_VOLATILE: + btf_dump_printf(d, " volatile"); + break; + case BTF_KIND_CONST: + btf_dump_printf(d, " const"); + break; + case BTF_KIND_RESTRICT: + btf_dump_printf(d, " restrict"); + break; + case BTF_KIND_ARRAY: { + const struct btf_array *a = (void *)(t + 1); + const struct btf_type *next_t; + __u32 next_id; + bool multidim; + /* + * GCC has a bug + * (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=8354) + * which causes it to emit extra const/volatile + * modifiers for an array, if array's element type has + * const/volatile modifiers. Clang doesn't do that. + * In general, it doesn't seem very meaningful to have + * a const/volatile modifier for array, so we are + * going to silently skip them here. + */ + while (decls->cnt) { + next_id = decls->ids[decls->cnt - 1]; + if (btf_is_mod_kind(d->btf, next_id)) + decls->cnt--; + else + break; + } + + if (decls->cnt == 0) { + btf_dump_emit_name(d, fname, last_was_ptr); + btf_dump_printf(d, "[%u]", a->nelems); + return; + } + + next_t = btf__type_by_id(d->btf, next_id); + multidim = btf_kind_of(next_t) == BTF_KIND_ARRAY; + /* we need space if we have named non-pointer */ + if (fname[0] && !last_was_ptr) + btf_dump_printf(d, " "); + /* no parentheses for multi-dimensional array */ + if (!multidim) + btf_dump_printf(d, "("); + btf_dump_emit_type_chain(d, decls, fname, lvl); + if (!multidim) + btf_dump_printf(d, ")"); + btf_dump_printf(d, "[%u]", a->nelems); + return; + } + case BTF_KIND_FUNC_PROTO: { + const struct btf_param *p = (void *)(t + 1); + __u16 vlen = btf_vlen_of(t); + int i; + + btf_dump_emit_mods(d, decls); + if (decls->cnt) { + btf_dump_printf(d, " ("); + btf_dump_emit_type_chain(d, decls, fname, lvl); + btf_dump_printf(d, ")"); + } else { + btf_dump_emit_name(d, fname, last_was_ptr); + } + btf_dump_printf(d, "("); + /* + * Clang for BPF target generates func_proto with no + * args as a func_proto with a single void arg (e.g., + * `int (*f)(void)` vs just `int (*f)()`). We are + * going to pretend there are no args for such case. + */ + if (vlen == 1 && p->type == 0) { + btf_dump_printf(d, ")"); + return; + } + + for (i = 0; i < vlen; i++, p++) { + if (i > 0) + btf_dump_printf(d, ", "); + + /* last arg of type void is vararg */ + if (i == vlen - 1 && p->type == 0) { + btf_dump_printf(d, "..."); + break; + } + + name = btf_name_of(d, p->name_off); + btf_dump_emit_type_decl(d, p->type, name, lvl); + } + + btf_dump_printf(d, ")"); + return; + } + default: + pr_warning("unexpected type in decl chain, kind:%u, id:[%u]\n", + kind, id); + return; + } + + last_was_ptr = kind == BTF_KIND_PTR; + } + + btf_dump_emit_name(d, fname, last_was_ptr); +} + +/* return number of duplicates (occurrences) of a given name */ +static size_t btf_dump_name_dups(struct btf_dump *d, struct hashmap *name_map, + const char *orig_name) +{ + size_t dup_cnt = 0; + + hashmap__find(name_map, orig_name, (void **)&dup_cnt); + dup_cnt++; + hashmap__set(name_map, orig_name, (void *)dup_cnt, NULL, NULL); + + return dup_cnt; +} + +static const char *btf_dump_resolve_name(struct btf_dump *d, __u32 id, + struct hashmap *name_map) +{ + struct btf_dump_type_aux_state *s = &d->type_states[id]; + const struct btf_type *t = btf__type_by_id(d->btf, id); + const char *orig_name = btf_name_of(d, t->name_off); + const char **cached_name = &d->cached_names[id]; + size_t dup_cnt; + + if (t->name_off == 0) + return ""; + + if (s->name_resolved) + return *cached_name ? *cached_name : orig_name; + + dup_cnt = btf_dump_name_dups(d, name_map, orig_name); + if (dup_cnt > 1) { + const size_t max_len = 256; + char new_name[max_len]; + + snprintf(new_name, max_len, "%s___%zu", orig_name, dup_cnt); + *cached_name = strdup(new_name); + } + + s->name_resolved = 1; + return *cached_name ? *cached_name : orig_name; +} + +static const char *btf_dump_type_name(struct btf_dump *d, __u32 id) +{ + return btf_dump_resolve_name(d, id, d->type_names); +} + +static const char *btf_dump_ident_name(struct btf_dump *d, __u32 id) +{ + return btf_dump_resolve_name(d, id, d->ident_names); +} diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map index 6ea5ce19b9e0..8bf51d0a6072 100644 --- a/tools/lib/bpf/libbpf.map +++ b/tools/lib/bpf/libbpf.map @@ -167,5 +167,8 @@ LIBBPF_0.0.3 { LIBBPF_0.0.4 { global: + btf_dump__dump_type; + btf_dump__free; + btf_dump__new; btf__parse_elf; } LIBBPF_0.0.3; -- Gitee From e893bae2c371434d2f3bb168a270b4dd9f9eaa2b Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:59:05 -0700 Subject: [PATCH 15/24] bpftool: add C output format option to btf dump subcommand ANBZ: #5530 commit 2119f2189df1836d28909fe5a90b3d93638ecf65 upstream. Utilize new libbpf's btf_dump API to emit BTF as a C definitions. Acked-by: Jakub Kicinski Signed-off-by: Andrii Nakryiko Reviewed-by: Quentin Monnet Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/bpf/bpftool/btf.c | 75 +++++++++++++++++++++++++++++++++++++++-- 1 file changed, 73 insertions(+), 2 deletions(-) diff --git a/tools/bpf/bpftool/btf.c b/tools/bpf/bpftool/btf.c index d861b088a947..a86e0b8864a3 100644 --- a/tools/bpf/bpftool/btf.c +++ b/tools/bpf/bpftool/btf.c @@ -340,11 +340,49 @@ static int dump_btf_raw(const struct btf *btf, return 0; } +static void __printf(2, 0) btf_dump_printf(void *ctx, + const char *fmt, va_list args) +{ + vfprintf(stdout, fmt, args); +} + +static int dump_btf_c(const struct btf *btf, + __u32 *root_type_ids, int root_type_cnt) +{ + struct btf_dump *d; + int err = 0, i; + + d = btf_dump__new(btf, NULL, NULL, btf_dump_printf); + if (IS_ERR(d)) + return PTR_ERR(d); + + if (root_type_cnt) { + for (i = 0; i < root_type_cnt; i++) { + err = btf_dump__dump_type(d, root_type_ids[i]); + if (err) + goto done; + } + } else { + int cnt = btf__get_nr_types(btf); + + for (i = 1; i <= cnt; i++) { + err = btf_dump__dump_type(d, i); + if (err) + goto done; + } + } + +done: + btf_dump__free(d); + return err; +} + static int do_dump(int argc, char **argv) { struct btf *btf = NULL; __u32 root_type_ids[2]; int root_type_cnt = 0; + bool dump_c = false; __u32 btf_id = -1; const char *src; int fd = -1; @@ -431,6 +469,29 @@ static int do_dump(int argc, char **argv) goto done; } + while (argc) { + if (is_prefix(*argv, "format")) { + NEXT_ARG(); + if (argc < 1) { + p_err("expecting value for 'format' option\n"); + goto done; + } + if (strcmp(*argv, "c") == 0) { + dump_c = true; + } else if (strcmp(*argv, "raw") == 0) { + dump_c = false; + } else { + p_err("unrecognized format specifier: '%s', possible values: raw, c", + *argv); + goto done; + } + NEXT_ARG(); + } else { + p_err("unrecognized option: '%s'", *argv); + goto done; + } + } + if (!btf) { err = btf__get_from_id(btf_id, &btf); if (err) { @@ -444,7 +505,16 @@ static int do_dump(int argc, char **argv) } } - dump_btf_raw(btf, root_type_ids, root_type_cnt); + if (dump_c) { + if (json_output) { + p_err("JSON output for C-syntax dump is not supported"); + err = -ENOTSUP; + goto done; + } + err = dump_btf_c(btf, root_type_ids, root_type_cnt); + } else { + err = dump_btf_raw(btf, root_type_ids, root_type_cnt); + } done: close(fd); @@ -460,10 +530,11 @@ static int do_help(int argc, char **argv) } fprintf(stderr, - "Usage: %s btf dump BTF_SRC\n" + "Usage: %s btf dump BTF_SRC [format FORMAT]\n" " %s btf help\n" "\n" " BTF_SRC := { id BTF_ID | prog PROG | map MAP [{key | value | kv | all}] | file FILE }\n" + " FORMAT := { raw | c }\n" " " HELP_SPEC_MAP "\n" " " HELP_SPEC_PROGRAM "\n" " " HELP_SPEC_OPTIONS "\n" -- Gitee From 7a0d21fcbe0abae776e80ca00e098916557d7e5a Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:59:06 -0700 Subject: [PATCH 16/24] bpftool/docs: add description of btf dump C option ANBZ: #5530 commit 220ba451124e275ff548bec33b0dc01e40895b06 upstream. Document optional **c** option for btf dump subcommand. Cc: Quentin Monnet Signed-off-by: Andrii Nakryiko Reviewed-by: Quentin Monnet Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- .../bpf/bpftool/Documentation/bpftool-btf.rst | 35 +++++++++++-------- 1 file changed, 20 insertions(+), 15 deletions(-) diff --git a/tools/bpf/bpftool/Documentation/bpftool-btf.rst b/tools/bpf/bpftool/Documentation/bpftool-btf.rst index 2dbc1413fabd..3daed9eba766 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-btf.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-btf.rst @@ -19,10 +19,11 @@ SYNOPSIS BTF COMMANDS ============= -| **bpftool** **btf dump** *BTF_SRC* +| **bpftool** **btf dump** *BTF_SRC* [**format** *FORMAT*] | **bpftool** **btf help** | | *BTF_SRC* := { **id** *BTF_ID* | **prog** *PROG* | **map** *MAP* [{**key** | **value** | **kv** | **all**}] | **file** *FILE* } +| *FORMAT* := { **raw** | **c** } | *MAP* := { **id** *MAP_ID* | **pinned** *FILE* } | *PROG* := { **id** *PROG_ID* | **pinned** *FILE* | **tag** *PROG_TAG* } @@ -31,23 +32,27 @@ DESCRIPTION **bpftool btf dump** *BTF_SRC* Dump BTF entries from a given *BTF_SRC*. - When **id** is specified, BTF object with that ID will be - loaded and all its BTF types emitted. + When **id** is specified, BTF object with that ID will be + loaded and all its BTF types emitted. - When **map** is provided, it's expected that map has - associated BTF object with BTF types describing key and - value. It's possible to select whether to dump only BTF - type(s) associated with key (**key**), value (**value**), - both key and value (**kv**), or all BTF types present in - associated BTF object (**all**). If not specified, **kv** - is assumed. + When **map** is provided, it's expected that map has + associated BTF object with BTF types describing key and + value. It's possible to select whether to dump only BTF + type(s) associated with key (**key**), value (**value**), + both key and value (**kv**), or all BTF types present in + associated BTF object (**all**). If not specified, **kv** + is assumed. - When **prog** is provided, it's expected that program has - associated BTF object with BTF types. + When **prog** is provided, it's expected that program has + associated BTF object with BTF types. - When specifying *FILE*, an ELF file is expected, containing - .BTF section with well-defined BTF binary format data, - typically produced by clang or pahole. + When specifying *FILE*, an ELF file is expected, containing + .BTF section with well-defined BTF binary format data, + typically produced by clang or pahole. + + **format** option can be used to override default (raw) + output format. Raw (**raw**) or C-syntax (**c**) output + formats are supported. **bpftool btf help** Print short help message. -- Gitee From f61a5bc4df5650146957ebb1aa98297379628136 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Fri, 24 May 2019 11:59:07 -0700 Subject: [PATCH 17/24] bpftool: update bash-completion w/ new c option for btf dump ANBZ: #5530 commit 90eea4086d5ed31936889a44d536bf77afa4ca8a upstream. Add bash completion for new C btf dump option. Cc: Quentin Monnet Signed-off-by: Andrii Nakryiko Reviewed-by: Quentin Monnet Signed-off-by: Alexei Starovoitov Signed-off-by: Yuanhe Shu --- tools/bpf/bpftool/bash-completion/bpftool | 21 +++++++++++++++++---- 1 file changed, 17 insertions(+), 4 deletions(-) diff --git a/tools/bpf/bpftool/bash-completion/bpftool b/tools/bpf/bpftool/bash-completion/bpftool index 50e402a5a9c8..75c01eafd3a1 100644 --- a/tools/bpf/bpftool/bash-completion/bpftool +++ b/tools/bpf/bpftool/bash-completion/bpftool @@ -638,11 +638,24 @@ _bpftool() esac return 0 ;; + format) + COMPREPLY=( $( compgen -W "c raw" -- "$cur" ) ) + ;; *) - if [[ $cword == 6 ]] && [[ ${words[3]} == "map" ]]; then - COMPREPLY+=( $( compgen -W 'key value kv all' -- \ - "$cur" ) ) - fi + # emit extra options + case ${words[3]} in + id|file) + _bpftool_once_attr 'format' + ;; + map|prog) + if [[ ${words[3]} == "map" ]] && [[ $cword == 6 ]]; then + COMPREPLY+=( $( compgen -W "key value kv all" -- "$cur" ) ) + fi + _bpftool_once_attr 'format' + ;; + *) + ;; + esac return 0 ;; esac -- Gitee From b7416fb0182e2861dd85c340c0da960e3b982e13 Mon Sep 17 00:00:00 2001 From: Andrii Nakryiko Date: Thu, 18 Jul 2019 10:30:21 -0700 Subject: [PATCH 18/24] libbpf: fix missing __WORDSIZE definition ANBZ: #5530 commit 8aa259b10a6a759c50137bbbf225df0c17ca5d27 upstream. hashmap.h depends on __WORDSIZE being defined. It is defined by glibc/musl in different headers. It's an explicit goal for musl to be "non-detectable" at compilation time, so instead include glibc header if glibc is explicitly detected and fall back to musl header otherwise. Reported-by: Arnaldo Carvalho de Melo Signed-off-by: Andrii Nakryiko Tested-by: Arnaldo Carvalho de Melo Cc: Alexei Starovoitov Cc: Andrii Nakryiko Cc: Daniel Borkmann Fixes: e3b924224028 ("libbpf: add resizable non-thread safe internal hashmap") Link: https://lkml.kernel.org/r/20190718173021.2418606-1-andriin@fb.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Yuanhe Shu --- tools/lib/bpf/hashmap.h | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/tools/lib/bpf/hashmap.h b/tools/lib/bpf/hashmap.h index 03748a742146..bae8879cdf58 100644 --- a/tools/lib/bpf/hashmap.h +++ b/tools/lib/bpf/hashmap.h @@ -10,6 +10,11 @@ #include #include +#ifdef __GLIBC__ +#include +#else +#include +#endif #include "libbpf_internal.h" static inline size_t hash_bits(size_t h, int bits) -- Gitee From 4ee469e15d959037d7bb1cbeee9c6d0c030829e6 Mon Sep 17 00:00:00 2001 From: Quentin Monnet Date: Fri, 24 May 2019 11:36:46 +0100 Subject: [PATCH 19/24] tools: bpftool: add -d option to get debug output from libbpf ANBZ: #5530 commit 775bc8ada89b376b4bbbce31aba47f4117fe1d9c upstream. libbpf has three levels of priority for output messages: warn, info, debug. By default, debug output is not printed to the console. Add a new "--debug" (short name: "-d") option to bpftool to print libbpf logs for all three levels. Internally, we simply use the function provided by libbpf to replace the default printing function by one that prints logs regardless of their level. v2: - Remove the possibility to select the log-levels to use (v1 offered a combination of "warn", "info" and "debug"). - Rename option and offer a short name: -d|--debug. - Add option description to all bpftool manual pages (instead of bpftool-prog.rst only), as all commands use libbpf. Signed-off-by: Quentin Monnet Reviewed-by: Jakub Kicinski Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- tools/bpf/bpftool/Documentation/bpftool-btf.rst | 4 ++++ tools/bpf/bpftool/Documentation/bpftool-cgroup.rst | 4 ++++ .../bpf/bpftool/Documentation/bpftool-feature.rst | 4 ++++ tools/bpf/bpftool/Documentation/bpftool-map.rst | 4 ++++ tools/bpf/bpftool/Documentation/bpftool-net.rst | 4 ++++ tools/bpf/bpftool/Documentation/bpftool-perf.rst | 4 ++++ tools/bpf/bpftool/Documentation/bpftool-prog.rst | 4 ++++ tools/bpf/bpftool/Documentation/bpftool.rst | 3 +++ tools/bpf/bpftool/bash-completion/bpftool | 2 +- tools/bpf/bpftool/main.c | 14 +++++++++++++- 10 files changed, 45 insertions(+), 2 deletions(-) diff --git a/tools/bpf/bpftool/Documentation/bpftool-btf.rst b/tools/bpf/bpftool/Documentation/bpftool-btf.rst index 3daed9eba766..6694a0fc8f99 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-btf.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-btf.rst @@ -72,6 +72,10 @@ OPTIONS -p, --pretty Generate human-readable JSON output. Implies **-j**. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + EXAMPLES ======== **# bpftool btf dump id 1226** diff --git a/tools/bpf/bpftool/Documentation/bpftool-cgroup.rst b/tools/bpf/bpftool/Documentation/bpftool-cgroup.rst index ac26876389c2..36807735e2a5 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-cgroup.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-cgroup.rst @@ -113,6 +113,10 @@ OPTIONS -f, --bpffs Show file names of pinned programs. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + EXAMPLES ======== | diff --git a/tools/bpf/bpftool/Documentation/bpftool-feature.rst b/tools/bpf/bpftool/Documentation/bpftool-feature.rst index 14180e887082..4d08f35034a2 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-feature.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-feature.rst @@ -73,6 +73,10 @@ OPTIONS -p, --pretty Generate human-readable JSON output. Implies **-j**. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + SEE ALSO ======== **bpf**\ (2), diff --git a/tools/bpf/bpftool/Documentation/bpftool-map.rst b/tools/bpf/bpftool/Documentation/bpftool-map.rst index 13ef27b39f20..490b4501cb6e 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-map.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-map.rst @@ -152,6 +152,10 @@ OPTIONS Do not automatically attempt to mount any virtual file system (such as tracefs or BPF virtual file system) when necessary. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + EXAMPLES ======== **# bpftool map show** diff --git a/tools/bpf/bpftool/Documentation/bpftool-net.rst b/tools/bpf/bpftool/Documentation/bpftool-net.rst index 934580850f42..d8e5237a2085 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-net.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-net.rst @@ -65,6 +65,10 @@ OPTIONS -p, --pretty Generate human-readable JSON output. Implies **-j**. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + EXAMPLES ======== diff --git a/tools/bpf/bpftool/Documentation/bpftool-perf.rst b/tools/bpf/bpftool/Documentation/bpftool-perf.rst index 0c7576523a21..e252bd0bc434 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-perf.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-perf.rst @@ -53,6 +53,10 @@ OPTIONS -p, --pretty Generate human-readable JSON output. Implies **-j**. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + EXAMPLES ======== diff --git a/tools/bpf/bpftool/Documentation/bpftool-prog.rst b/tools/bpf/bpftool/Documentation/bpftool-prog.rst index e8118544d118..9a92614569e6 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-prog.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-prog.rst @@ -174,6 +174,10 @@ OPTIONS Do not automatically attempt to mount any virtual file system (such as tracefs or BPF virtual file system) when necessary. + -d, --debug + Print all logs available from libbpf, including debug-level + information. + EXAMPLES ======== **# bpftool prog show** diff --git a/tools/bpf/bpftool/Documentation/bpftool.rst b/tools/bpf/bpftool/Documentation/bpftool.rst index 3e562d7fd56f..43dba0717953 100644 --- a/tools/bpf/bpftool/Documentation/bpftool.rst +++ b/tools/bpf/bpftool/Documentation/bpftool.rst @@ -66,6 +66,9 @@ OPTIONS Do not automatically attempt to mount any virtual file system (such as tracefs or BPF virtual file system) when necessary. + -d, --debug + Print all logs available from libbpf, including debug-level + information. SEE ALSO ======== diff --git a/tools/bpf/bpftool/bash-completion/bpftool b/tools/bpf/bpftool/bash-completion/bpftool index 75c01eafd3a1..dbbd9519ead1 100644 --- a/tools/bpf/bpftool/bash-completion/bpftool +++ b/tools/bpf/bpftool/bash-completion/bpftool @@ -181,7 +181,7 @@ _bpftool() # Deal with options if [[ ${words[cword]} == -* ]]; then - local c='--version --json --pretty --bpffs --mapcompat' + local c='--version --json --pretty --bpffs --mapcompat --debug' COMPREPLY=( $( compgen -W "$c" -- "$cur" ) ) return 0 fi diff --git a/tools/bpf/bpftool/main.c b/tools/bpf/bpftool/main.c index c85809a5c2b5..c4e2b3f20be0 100644 --- a/tools/bpf/bpftool/main.c +++ b/tools/bpf/bpftool/main.c @@ -10,6 +10,7 @@ #include #include +#include #include "main.h" @@ -77,6 +78,13 @@ static int do_version(int argc, char **argv) return 0; } +static int __printf(2, 0) +print_all_levels(__maybe_unused enum libbpf_print_level level, + const char *format, va_list args) +{ + return vfprintf(stderr, format, args); +} + int cmd_select(const struct cmd *cmds, int argc, char **argv, int (*help)(int argc, char **argv)) { @@ -317,6 +325,7 @@ int main(int argc, char **argv) { "bpffs", no_argument, NULL, 'f' }, { "mapcompat", no_argument, NULL, 'm' }, { "nomount", no_argument, NULL, 'n' }, + { "debug", no_argument, NULL, 'd' }, { 0 } }; int opt, ret; @@ -334,7 +343,7 @@ int main(int argc, char **argv) hash_init(map_table.table); opterr = 0; - while ((opt = getopt_long(argc, argv, "Vhpjfmn", + while ((opt = getopt_long(argc, argv, "Vhpjfmnd", options, NULL)) >= 0) { switch (opt) { case 'V': @@ -364,6 +373,9 @@ int main(int argc, char **argv) case 'n': block_mount = true; break; + case 'd': + libbpf_set_print(print_all_levels); + break; default: p_err("unrecognized option '%s'", argv[optind - 1]); if (json_output) -- Gitee From f291bfbf8b6525e6d53364e9dd70ddf4a511812b Mon Sep 17 00:00:00 2001 From: Quentin Monnet Date: Fri, 24 May 2019 11:36:47 +0100 Subject: [PATCH 20/24] libbpf: add bpf_object__load_xattr() API function to pass log_level ANBZ: #5530 commit 60276f9849988d3d3a54943c9ec27222c5819dae upstream. libbpf was recently made aware of the log_level attribute for programs, used to specify the level of information expected to be dumped by the verifier. Function bpf_prog_load_xattr() got support for this log_level parameter. But some applications using libbpf rely on another function to load programs, bpf_object__load(), which does accept any parameter for log level. Create an API function based on bpf_object__load(), but accepting an "attr" object as a parameter. Then add a log_level field to that object, so that applications calling the new bpf_object__load_xattr() can pick the desired log level. v3: - Rewrite commit log. v2: - We are in a new cycle, bump libbpf extraversion number. Signed-off-by: Quentin Monnet Reviewed-by: Jakub Kicinski Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- tools/lib/bpf/Makefile | 2 +- tools/lib/bpf/libbpf.c | 20 +++++++++++++++++--- tools/lib/bpf/libbpf.h | 6 ++++++ tools/lib/bpf/libbpf.map | 1 + 4 files changed, 25 insertions(+), 4 deletions(-) diff --git a/tools/lib/bpf/Makefile b/tools/lib/bpf/Makefile index 23879feef1e8..fa609bfdd116 100644 --- a/tools/lib/bpf/Makefile +++ b/tools/lib/bpf/Makefile @@ -3,7 +3,7 @@ BPF_VERSION = 0 BPF_PATCHLEVEL = 0 -BPF_EXTRAVERSION = 3 +BPF_EXTRAVERSION = 4 MAKEFLAGS += --no-print-directory diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index e277d20c1671..79202fa9a4cf 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -2259,7 +2259,7 @@ static bool bpf_program__is_function_storage(struct bpf_program *prog, } static int -bpf_object__load_progs(struct bpf_object *obj) +bpf_object__load_progs(struct bpf_object *obj, int log_level) { size_t i; int err; @@ -2267,6 +2267,7 @@ bpf_object__load_progs(struct bpf_object *obj) for (i = 0; i < obj->nr_programs; i++) { if (bpf_program__is_function_storage(&obj->programs[i], obj)) continue; + obj->programs[i].log_level = log_level; err = bpf_program__load(&obj->programs[i], obj->license, obj->kern_version); @@ -2418,10 +2419,14 @@ int bpf_object__unload(struct bpf_object *obj) return 0; } -int bpf_object__load(struct bpf_object *obj) +int bpf_object__load_xattr(struct bpf_object_load_attr *attr) { + struct bpf_object *obj; int err; + if (!attr) + return -EINVAL; + obj = attr->obj; if (!obj) return -EINVAL; @@ -2434,7 +2439,7 @@ int bpf_object__load(struct bpf_object *obj) CHECK_ERR(bpf_object__create_maps(obj), err, out); CHECK_ERR(bpf_object__relocate(obj), err, out); - CHECK_ERR(bpf_object__load_progs(obj), err, out); + CHECK_ERR(bpf_object__load_progs(obj, attr->log_level), err, out); return 0; out: @@ -2443,6 +2448,15 @@ int bpf_object__load(struct bpf_object *obj) return err; } +int bpf_object__load(struct bpf_object *obj) +{ + struct bpf_object_load_attr attr = { + .obj = obj, + }; + + return bpf_object__load_xattr(&attr); +} + static int check_path(const char *path) { char *cp, errmsg[STRERR_BUFSIZE]; diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h index c5ff00515ce7..e1c748db44f6 100644 --- a/tools/lib/bpf/libbpf.h +++ b/tools/lib/bpf/libbpf.h @@ -89,8 +89,14 @@ LIBBPF_API int bpf_object__unpin_programs(struct bpf_object *obj, LIBBPF_API int bpf_object__pin(struct bpf_object *object, const char *path); LIBBPF_API void bpf_object__close(struct bpf_object *object); +struct bpf_object_load_attr { + struct bpf_object *obj; + int log_level; +}; + /* Load/unload object into/from kernel */ LIBBPF_API int bpf_object__load(struct bpf_object *obj); +LIBBPF_API int bpf_object__load_xattr(struct bpf_object_load_attr *attr); LIBBPF_API int bpf_object__unload(struct bpf_object *obj); LIBBPF_API const char *bpf_object__name(struct bpf_object *obj); LIBBPF_API unsigned int bpf_object__kversion(struct bpf_object *obj); diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map index 8bf51d0a6072..46dcda89df21 100644 --- a/tools/lib/bpf/libbpf.map +++ b/tools/lib/bpf/libbpf.map @@ -171,4 +171,5 @@ LIBBPF_0.0.4 { btf_dump__free; btf_dump__new; btf__parse_elf; + bpf_object__load_xattr; } LIBBPF_0.0.3; -- Gitee From fe8bb889b8347b5da5bdb10853ae33d75c820ca2 Mon Sep 17 00:00:00 2001 From: Quentin Monnet Date: Fri, 24 May 2019 11:36:48 +0100 Subject: [PATCH 21/24] tools: bpftool: make -d option print debug output from verifier ANBZ: #5530 commit 55d778076b0354b088a9a16d9ff584c887e17f42 upstream. The "-d" option is used to require all logs available for bpftool. So far it meant telling libbpf to print even debug-level information. But there is another source of info that can be made more verbose: when we attemt to load programs with bpftool, we can pass a log_level parameter to the verifier in order to control the amount of information that is printed to the console. Reuse the "-d" option to print all information the verifier can tell. At this time, this means logs related to BPF_LOG_LEVEL1, BPF_LOG_LEVEL2 and BPF_LOG_STATS. As mentioned in the discussion on the first version of this set, these macros are internal to the kernel (include/linux/bpf_verifier.h) and are not meant to be part of the stable user API, therefore we simply use the related constants to print whatever we can at this time, without trying to tell users what is log_level1 or what is statistics. Verifier logs are only used when loading programs for now (In the future: for loading BTF objects with bpftool? Although libbpf does not currently offer to print verifier info at debug level if no error occurred when loading BTF objects), so bpftool.rst and bpftool-prog.rst are the only man pages to get the update. v3: - Add details on log level and BTF loading at the end of commit log. v2: - Remove the possibility to select the log levels to use (v1 offered a combination of "log_level1", "log_level2" and "stats"). - The macros from kernel header bpf_verifier.h are not used (and therefore not moved to UAPI header). - In v1 this was a distinct option, but is now merged in the only "-d" switch to activate libbpf and verifier debug-level logs all at the same time. Signed-off-by: Quentin Monnet Reviewed-by: Jakub Kicinski Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- .../bpftool/Documentation/bpftool-prog.rst | 5 ++-- tools/bpf/bpftool/Documentation/bpftool.rst | 5 ++-- tools/bpf/bpftool/main.c | 2 ++ tools/bpf/bpftool/main.h | 1 + tools/bpf/bpftool/prog.c | 27 ++++++++++++------- 5 files changed, 26 insertions(+), 14 deletions(-) diff --git a/tools/bpf/bpftool/Documentation/bpftool-prog.rst b/tools/bpf/bpftool/Documentation/bpftool-prog.rst index 9a92614569e6..228a5c863cc7 100644 --- a/tools/bpf/bpftool/Documentation/bpftool-prog.rst +++ b/tools/bpf/bpftool/Documentation/bpftool-prog.rst @@ -175,8 +175,9 @@ OPTIONS (such as tracefs or BPF virtual file system) when necessary. -d, --debug - Print all logs available from libbpf, including debug-level - information. + Print all logs available, even debug-level information. This + includes logs from libbpf as well as from the verifier, when + attempting to load programs. EXAMPLES ======== diff --git a/tools/bpf/bpftool/Documentation/bpftool.rst b/tools/bpf/bpftool/Documentation/bpftool.rst index 43dba0717953..6a9c52ef84a9 100644 --- a/tools/bpf/bpftool/Documentation/bpftool.rst +++ b/tools/bpf/bpftool/Documentation/bpftool.rst @@ -67,8 +67,9 @@ OPTIONS (such as tracefs or BPF virtual file system) when necessary. -d, --debug - Print all logs available from libbpf, including debug-level - information. + Print all logs available, even debug-level information. This + includes logs from libbpf as well as from the verifier, when + attempting to load programs. SEE ALSO ======== diff --git a/tools/bpf/bpftool/main.c b/tools/bpf/bpftool/main.c index c4e2b3f20be0..d26c498b376c 100644 --- a/tools/bpf/bpftool/main.c +++ b/tools/bpf/bpftool/main.c @@ -26,6 +26,7 @@ bool pretty_output; bool json_output; bool show_pinned; bool block_mount; +bool verifier_logs; int bpf_flags; struct pinned_obj_table prog_table; struct pinned_obj_table map_table; @@ -375,6 +376,7 @@ int main(int argc, char **argv) break; case 'd': libbpf_set_print(print_all_levels); + verifier_logs = true; break; default: p_err("unrecognized option '%s'", argv[optind - 1]); diff --git a/tools/bpf/bpftool/main.h b/tools/bpf/bpftool/main.h index ce75b76d17b1..8bb591f16737 100644 --- a/tools/bpf/bpftool/main.h +++ b/tools/bpf/bpftool/main.h @@ -92,6 +92,7 @@ extern json_writer_t *json_wtr; extern bool json_output; extern bool show_pinned; extern bool block_mount; +extern bool verifier_logs; extern int bpf_flags; extern struct pinned_obj_table prog_table; extern struct pinned_obj_table map_table; diff --git a/tools/bpf/bpftool/prog.c b/tools/bpf/bpftool/prog.c index 0ce441fae446..37cb4a6c2790 100644 --- a/tools/bpf/bpftool/prog.c +++ b/tools/bpf/bpftool/prog.c @@ -752,10 +752,11 @@ static int do_detach(int argc, char **argv) static int load_with_options(int argc, char **argv, bool first_prog_only) { - enum bpf_attach_type expected_attach_type; - struct bpf_object_open_attr attr = { - .prog_type = BPF_PROG_TYPE_UNSPEC, + struct bpf_object_load_attr load_attr = { 0 }; + struct bpf_object_open_attr open_attr = { + .prog_type = BPF_PROG_TYPE_UNSPEC, }; + enum bpf_attach_type expected_attach_type; struct map_replace *map_replace = NULL; struct bpf_program *prog = NULL, *pos; unsigned int old_map_fds = 0; @@ -769,7 +770,7 @@ static int load_with_options(int argc, char **argv, bool first_prog_only) if (!REQ_ARGS(2)) return -1; - attr.file = GET_ARG(); + open_attr.file = GET_ARG(); pinfile = GET_ARG(); while (argc) { @@ -778,7 +779,7 @@ static int load_with_options(int argc, char **argv, bool first_prog_only) NEXT_ARG(); - if (attr.prog_type != BPF_PROG_TYPE_UNSPEC) { + if (open_attr.prog_type != BPF_PROG_TYPE_UNSPEC) { p_err("program type already specified"); goto err_free_reuse_maps; } @@ -795,7 +796,8 @@ static int load_with_options(int argc, char **argv, bool first_prog_only) strcat(type, *argv); strcat(type, "/"); - err = libbpf_prog_type_by_name(type, &attr.prog_type, + err = libbpf_prog_type_by_name(type, + &open_attr.prog_type, &expected_attach_type); free(type); if (err < 0) @@ -883,16 +885,16 @@ static int load_with_options(int argc, char **argv, bool first_prog_only) set_max_rlimit(); - obj = __bpf_object__open_xattr(&attr, bpf_flags); + obj = __bpf_object__open_xattr(&open_attr, bpf_flags); if (IS_ERR_OR_NULL(obj)) { p_err("failed to open object file"); goto err_free_reuse_maps; } bpf_object__for_each_program(pos, obj) { - enum bpf_prog_type prog_type = attr.prog_type; + enum bpf_prog_type prog_type = open_attr.prog_type; - if (attr.prog_type == BPF_PROG_TYPE_UNSPEC) { + if (open_attr.prog_type == BPF_PROG_TYPE_UNSPEC) { const char *sec_name = bpf_program__title(pos, false); err = libbpf_prog_type_by_name(sec_name, &prog_type, @@ -962,7 +964,12 @@ static int load_with_options(int argc, char **argv, bool first_prog_only) goto err_close_obj; } - err = bpf_object__load(obj); + load_attr.obj = obj; + if (verifier_logs) + /* log_level1 + log_level2 + stats, but not stable UAPI */ + load_attr.log_level = 1 + 2 + 4; + + err = bpf_object__load_xattr(&load_attr); if (err) { p_err("failed to load object file"); goto err_close_obj; -- Gitee From c4cc6d4688d4090cda22b8fe1bbe04bac79c2b6f Mon Sep 17 00:00:00 2001 From: Quentin Monnet Date: Wed, 29 May 2019 15:26:41 +0100 Subject: [PATCH 22/24] libbpf: prevent overwriting of log_level in bpf_object__load_progs() ANBZ: #5530 commit 501b125a29f7cd8983960c5f3c80b337b1359cc3 upstream. There are two functions in libbpf that support passing a log_level parameter for the verifier for loading programs: bpf_object__load_xattr() and bpf_prog_load_xattr(). Both accept an attribute object containing the log_level, and apply it to the programs to load. It turns out that to effectively load the programs, the latter function eventually relies on the former. This was not taken into account when adding support for log_level in bpf_object__load_xattr(), and the log_level passed to bpf_prog_load_xattr() later gets overwritten with a zero value, thus disabling verifier logs for the program in all cases: bpf_prog_load_xattr() // prog->log_level = attr1->log_level; -> bpf_object__load() // attr2->log_level = 0; -> bpf_object__load_xattr() // -> bpf_object__load_progs() // prog->log_level = attr2->log_level; Fix this by OR-ing the log_level in bpf_object__load_progs(), instead of overwriting it. v2: Fix commit log description (confusion on function names in v1). Fixes: 60276f984998 ("libbpf: add bpf_object__load_xattr() API function to pass log_level") Reported-by: Alexei Starovoitov Signed-off-by: Quentin Monnet Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- tools/lib/bpf/libbpf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index 79202fa9a4cf..34e5ee0845f4 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -2267,7 +2267,7 @@ bpf_object__load_progs(struct bpf_object *obj, int log_level) for (i = 0; i < obj->nr_programs; i++) { if (bpf_program__is_function_storage(&obj->programs[i], obj)) continue; - obj->programs[i].log_level = log_level; + obj->programs[i].log_level |= log_level; err = bpf_program__load(&obj->programs[i], obj->license, obj->kern_version); -- Gitee From 89aafaf33961a98341630cb28ca2378c1f1421b1 Mon Sep 17 00:00:00 2001 From: Andrey Ignatov Date: Wed, 3 Apr 2019 23:22:39 -0700 Subject: [PATCH 23/24] bpf: Reject indirect var_off stack access in unpriv mode ANBZ: #5530 commit 088ec26d9c2da9d879ab73e3f4117f9df6c566ee upstream. Proper support of indirect stack access with variable offset in unprivileged mode (!root) requires corresponding support in Spectre masking for stack ALU in retrieve_ptr_limit(). There are no use-case for variable offset in unprivileged mode though so make verifier reject such accesses for simplicity. Pointer arithmetics is one (and only?) way to cause variable offset and it's already rejected in unpriv mode so that verifier won't even get to helper function whose argument contains variable offset, e.g.: 0: (7a) *(u64 *)(r10 -16) = 0 1: (7a) *(u64 *)(r10 -8) = 0 2: (61) r2 = *(u32 *)(r1 +0) 3: (57) r2 &= 4 4: (17) r2 -= 16 5: (0f) r2 += r10 variable stack access var_off=(0xfffffffffffffff0; 0x4) off=-16 size=1R2 stack pointer arithmetic goes out of range, prohibited for !root Still it looks like a good idea to reject variable offset indirect stack access for unprivileged mode in check_stack_boundary() explicitly. [backport note] Ignore explaination in retrieve_ptr_limit as e399fd9c3ac6a0(fixes CVE-2021-29155, CVE-2021-33200) and f8fee261df05d6(fixes CVE-2020-27170, CVE-2020-27171) has been introduced. Fixes: 2011fccfb61b ("bpf: Support variable offset stack access from helpers") Reported-by: Daniel Borkmann Signed-off-by: Andrey Ignatov Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- kernel/bpf/verifier.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index bff1e01282e7..fd79c7a2bd42 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -2279,6 +2279,19 @@ static int check_stack_boundary(struct bpf_verifier_env *env, int regno, if (err) return err; } else { + /* Variable offset is prohibited for unprivileged mode for + * simplicity since it requires corresponding support in + * Spectre masking for stack ALU. + * See also retrieve_ptr_limit(). + */ + if (!env->allow_ptr_leaks) { + char tn_buf[48]; + + tnum_strn(tn_buf, sizeof(tn_buf), reg->var_off); + verbose(env, "R%d indirect variable offset stack access prohibited for !root, var_off=%s\n", + regno, tn_buf); + return -EACCES; + } /* Only initialized buffer on stack is allowed to be accessed * with variable offset. With uninitialized buffer it's hard to * guarantee that whole memory is marked as initialized on -- Gitee From b39e58da74375f98cacccf087f70f721523be4d1 Mon Sep 17 00:00:00 2001 From: Andrey Ignatov Date: Wed, 3 Apr 2019 23:22:40 -0700 Subject: [PATCH 24/24] selftests/bpf: Test indirect var_off stack access in unpriv mode ANBZ: #5530 commit 2c6927dbdc3fbd41207e671212f53a98bbebf6ba upstream. Test that verifier rejects indirect stack access with variable offset in unprivileged mode and accepts same code in privileged mode. Since pointer arithmetics is prohibited in unprivileged mode verifier should reject the program even before it gets to helper call that uses variable offset, at the time when that variable offset is trying to be constructed. Example of output: # ./test_verifier ... #859/u indirect variable-offset stack access, priv vs unpriv OK #859/p indirect variable-offset stack access, priv vs unpriv OK Signed-off-by: Andrey Ignatov Signed-off-by: Daniel Borkmann Signed-off-by: Yuanhe Shu --- tools/testing/selftests/bpf/test_verifier.c | 27 +++++++++++++++++++++ 1 file changed, 27 insertions(+) diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c index 3e0502af3307..22bff19472f5 100644 --- a/tools/testing/selftests/bpf/test_verifier.c +++ b/tools/testing/selftests/bpf/test_verifier.c @@ -9548,6 +9548,33 @@ static struct bpf_test tests[] = { .result = REJECT, .prog_type = BPF_PROG_TYPE_LWT_IN, }, + { + "indirect variable-offset stack access, priv vs unpriv", + .insns = { + /* Fill the top 16 bytes of the stack. */ + BPF_ST_MEM(BPF_DW, BPF_REG_10, -16, 0), + BPF_ST_MEM(BPF_DW, BPF_REG_10, -8, 0), + /* Get an unknown value. */ + BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1, 0), + /* Make it small and 4-byte aligned. */ + BPF_ALU64_IMM(BPF_AND, BPF_REG_2, 4), + BPF_ALU64_IMM(BPF_SUB, BPF_REG_2, 16), + /* Add it to fp. We now have either fp-12 or fp-16, we don't know + * which, but either way it points to initialized stack. + */ + BPF_ALU64_REG(BPF_ADD, BPF_REG_2, BPF_REG_10), + /* Dereference it indirectly. */ + BPF_LD_MAP_FD(BPF_REG_1, 0), + BPF_EMIT_CALL(BPF_FUNC_map_lookup_elem), + BPF_MOV64_IMM(BPF_REG_0, 0), + BPF_EXIT_INSN(), + }, + .fixup_map_hash_8b = { 6 }, + .errstr_unpriv = "R2 stack pointer arithmetic goes out of range, prohibited for !root", + .result_unpriv = REJECT, + .result = ACCEPT, + .prog_type = BPF_PROG_TYPE_CGROUP_SKB, + }, { "indirect variable-offset stack access, ok", .insns = { -- Gitee