Message ID | 20200924011951.408313-2-songliubraving@fb.com |
---|---|
State | Changes Requested |
Delegated to: | BPF Maintainers |
Headers | show |
Series | enable BPF_PROG_TEST_RUN for raw_tp | expand |
On Wed, Sep 23, 2020 at 6:46 PM Song Liu <songliubraving@fb.com> wrote: > > Add .test_run for raw_tracepoint. Also, introduce a new feature that runs > the target program on a specific CPU. This is achieved by a new flag in > bpf_attr.test, BPF_F_TEST_RUN_ON_CPU. When this flag is set, the program > is triggered on cpu with id bpf_attr.test.cpu. This feature is needed for > BPF programs that handle perf_event and other percpu resources, as the > program can access these resource locally. > > Acked-by: John Fastabend <john.fastabend@gmail.com> > Signed-off-by: Song Liu <songliubraving@fb.com> > --- > include/linux/bpf.h | 3 ++ > include/uapi/linux/bpf.h | 7 +++ > kernel/bpf/syscall.c | 2 +- > kernel/trace/bpf_trace.c | 1 + > net/bpf/test_run.c | 89 ++++++++++++++++++++++++++++++++++ > tools/include/uapi/linux/bpf.h | 7 +++ > 6 files changed, 108 insertions(+), 1 deletion(-) > [...] > +int bpf_prog_test_run_raw_tp(struct bpf_prog *prog, > + const union bpf_attr *kattr, > + union bpf_attr __user *uattr) > +{ > + void __user *ctx_in = u64_to_user_ptr(kattr->test.ctx_in); > + __u32 ctx_size_in = kattr->test.ctx_size_in; > + struct bpf_raw_tp_test_run_info info; > + int cpu, err = 0; > + > + /* doesn't support data_in/out, ctx_out, duration, or repeat */ > + if (kattr->test.data_in || kattr->test.data_out || > + kattr->test.ctx_out || kattr->test.duration || > + kattr->test.repeat) duration and repeat sound generally useful (benchmarking raw_tp programs), so it's a pity you haven't implemented them. But it can be added later, so not a deal breaker. > + return -EINVAL; > + > + if (ctx_size_in < prog->aux->max_ctx_offset) > + return -EINVAL; > + > + if (ctx_size_in) { > + info.ctx = kzalloc(ctx_size_in, GFP_USER); > + if (!info.ctx) > + return -ENOMEM; > + if (copy_from_user(info.ctx, ctx_in, ctx_size_in)) { > + err = -EFAULT; > + goto out; > + } > + } else { > + info.ctx = NULL; > + } > + > + info.prog = prog; > + cpu = kattr->test.cpu; > + > + if ((kattr->test.flags & BPF_F_TEST_RUN_ON_CPU) == 0 || > + cpu == smp_processor_id()) { should we enforce that cpu == 0 if BPF_F_TEST_RUN_ON_CPU is not set? > + __bpf_prog_test_run_raw_tp(&info); > + } else { > + /* smp_call_function_single() also checks cpu_online() > + * after csd_lock(). However, since cpu_plus is from user cpu_plus leftover in a comment > + * space, let's do an extra quick check to filter out > + * invalid value before smp_call_function_single(). > + */ > + if (!cpu_online(cpu)) { briefly looking at cpu_online() code, it seems like it's not checking that cpu is < NR_CPUS. Should we add a selftest that validates that passing unreasonable cpu index doesn't generate warning or invalid memory access? > + err = -ENXIO; > + goto out; > + } > + > + err = smp_call_function_single(cpu, __bpf_prog_test_run_raw_tp, > + &info, 1); > + if (err) > + goto out; > + } > + [...]
> On Sep 24, 2020, at 12:56 PM, Andrii Nakryiko <andrii.nakryiko@gmail.com> wrote: > > On Wed, Sep 23, 2020 at 6:46 PM Song Liu <songliubraving@fb.com> wrote: >> >> Add .test_run for raw_tracepoint. Also, introduce a new feature that runs >> the target program on a specific CPU. This is achieved by a new flag in >> bpf_attr.test, BPF_F_TEST_RUN_ON_CPU. When this flag is set, the program >> is triggered on cpu with id bpf_attr.test.cpu. This feature is needed for >> BPF programs that handle perf_event and other percpu resources, as the >> program can access these resource locally. >> >> Acked-by: John Fastabend <john.fastabend@gmail.com> >> Signed-off-by: Song Liu <songliubraving@fb.com> >> --- >> include/linux/bpf.h | 3 ++ >> include/uapi/linux/bpf.h | 7 +++ >> kernel/bpf/syscall.c | 2 +- >> kernel/trace/bpf_trace.c | 1 + >> net/bpf/test_run.c | 89 ++++++++++++++++++++++++++++++++++ >> tools/include/uapi/linux/bpf.h | 7 +++ >> 6 files changed, 108 insertions(+), 1 deletion(-) >> > > [...] > >> +int bpf_prog_test_run_raw_tp(struct bpf_prog *prog, >> + const union bpf_attr *kattr, >> + union bpf_attr __user *uattr) >> +{ >> + void __user *ctx_in = u64_to_user_ptr(kattr->test.ctx_in); >> + __u32 ctx_size_in = kattr->test.ctx_size_in; >> + struct bpf_raw_tp_test_run_info info; >> + int cpu, err = 0; >> + >> + /* doesn't support data_in/out, ctx_out, duration, or repeat */ >> + if (kattr->test.data_in || kattr->test.data_out || >> + kattr->test.ctx_out || kattr->test.duration || >> + kattr->test.repeat) > > duration and repeat sound generally useful (benchmarking raw_tp > programs), so it's a pity you haven't implemented them. But it can be > added later, so not a deal breaker. > >> + return -EINVAL; >> + >> + if (ctx_size_in < prog->aux->max_ctx_offset) >> + return -EINVAL; >> + >> + if (ctx_size_in) { >> + info.ctx = kzalloc(ctx_size_in, GFP_USER); >> + if (!info.ctx) >> + return -ENOMEM; >> + if (copy_from_user(info.ctx, ctx_in, ctx_size_in)) { >> + err = -EFAULT; >> + goto out; >> + } >> + } else { >> + info.ctx = NULL; >> + } >> + >> + info.prog = prog; >> + cpu = kattr->test.cpu; >> + >> + if ((kattr->test.flags & BPF_F_TEST_RUN_ON_CPU) == 0 || >> + cpu == smp_processor_id()) { > > should we enforce that cpu == 0 if BPF_F_TEST_RUN_ON_CPU is not set? Added a test. > > >> + __bpf_prog_test_run_raw_tp(&info); >> + } else { >> + /* smp_call_function_single() also checks cpu_online() >> + * after csd_lock(). However, since cpu_plus is from user > > cpu_plus leftover in a comment Fixed. > >> + * space, let's do an extra quick check to filter out >> + * invalid value before smp_call_function_single(). >> + */ >> + if (!cpu_online(cpu)) { > > briefly looking at cpu_online() code, it seems like it's not checking > that cpu is < NR_CPUS. Should we add a selftest that validates that > passing unreasonable cpu index doesn't generate warning or invalid > memory access? Good catch! We need extra "cpu >= nr_cpu_ids". Thanks, Song
diff --git a/include/linux/bpf.h b/include/linux/bpf.h index d7c5a6ed87e30..23758c282eb4b 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -1376,6 +1376,9 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog, int bpf_prog_test_run_flow_dissector(struct bpf_prog *prog, const union bpf_attr *kattr, union bpf_attr __user *uattr); +int bpf_prog_test_run_raw_tp(struct bpf_prog *prog, + const union bpf_attr *kattr, + union bpf_attr __user *uattr); bool btf_ctx_access(int off, int size, enum bpf_access_type type, const struct bpf_prog *prog, struct bpf_insn_access_aux *info); diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index a22812561064a..05e480f66f475 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -424,6 +424,11 @@ enum { */ #define BPF_F_QUERY_EFFECTIVE (1U << 0) +/* Flags for BPF_PROG_TEST_RUN */ + +/* If set, run the test on the cpu specified by bpf_attr.test.cpu */ +#define BPF_F_TEST_RUN_ON_CPU (1U << 0) + /* type for BPF_ENABLE_STATS */ enum bpf_stats_type { /* enabled run_time_ns and run_cnt */ @@ -566,6 +571,8 @@ union bpf_attr { */ __aligned_u64 ctx_in; __aligned_u64 ctx_out; + __u32 flags; + __u32 cpu; } test; struct { /* anonymous struct used by BPF_*_GET_*_ID */ diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index ec68d3a23a2b7..684bc29515664 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -2975,7 +2975,7 @@ static int bpf_prog_query(const union bpf_attr *attr, } } -#define BPF_PROG_TEST_RUN_LAST_FIELD test.ctx_out +#define BPF_PROG_TEST_RUN_LAST_FIELD test.cpu static int bpf_prog_test_run(const union bpf_attr *attr, union bpf_attr __user *uattr) diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c index b2a5380eb1871..4553aebf53862 100644 --- a/kernel/trace/bpf_trace.c +++ b/kernel/trace/bpf_trace.c @@ -1675,6 +1675,7 @@ const struct bpf_verifier_ops raw_tracepoint_verifier_ops = { }; const struct bpf_prog_ops raw_tracepoint_prog_ops = { + .test_run = bpf_prog_test_run_raw_tp, }; const struct bpf_verifier_ops tracing_verifier_ops = { diff --git a/net/bpf/test_run.c b/net/bpf/test_run.c index 99eb8c6c0fbcc..1411bc665be6b 100644 --- a/net/bpf/test_run.c +++ b/net/bpf/test_run.c @@ -11,6 +11,7 @@ #include <net/sock.h> #include <net/tcp.h> #include <linux/error-injection.h> +#include <linux/smp.h> #define CREATE_TRACE_POINTS #include <trace/events/bpf_test_run.h> @@ -204,6 +205,9 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog, int b = 2, err = -EFAULT; u32 retval = 0; + if (kattr->test.flags || kattr->test.cpu) + return -EINVAL; + switch (prog->expected_attach_type) { case BPF_TRACE_FENTRY: case BPF_TRACE_FEXIT: @@ -236,6 +240,85 @@ int bpf_prog_test_run_tracing(struct bpf_prog *prog, return err; } +struct bpf_raw_tp_test_run_info { + struct bpf_prog *prog; + void *ctx; + u32 retval; +}; + +static void +__bpf_prog_test_run_raw_tp(void *data) +{ + struct bpf_raw_tp_test_run_info *info = data; + + rcu_read_lock(); + migrate_disable(); + info->retval = BPF_PROG_RUN(info->prog, info->ctx); + migrate_enable(); + rcu_read_unlock(); +} + +int bpf_prog_test_run_raw_tp(struct bpf_prog *prog, + const union bpf_attr *kattr, + union bpf_attr __user *uattr) +{ + void __user *ctx_in = u64_to_user_ptr(kattr->test.ctx_in); + __u32 ctx_size_in = kattr->test.ctx_size_in; + struct bpf_raw_tp_test_run_info info; + int cpu, err = 0; + + /* doesn't support data_in/out, ctx_out, duration, or repeat */ + if (kattr->test.data_in || kattr->test.data_out || + kattr->test.ctx_out || kattr->test.duration || + kattr->test.repeat) + return -EINVAL; + + if (ctx_size_in < prog->aux->max_ctx_offset) + return -EINVAL; + + if (ctx_size_in) { + info.ctx = kzalloc(ctx_size_in, GFP_USER); + if (!info.ctx) + return -ENOMEM; + if (copy_from_user(info.ctx, ctx_in, ctx_size_in)) { + err = -EFAULT; + goto out; + } + } else { + info.ctx = NULL; + } + + info.prog = prog; + cpu = kattr->test.cpu; + + if ((kattr->test.flags & BPF_F_TEST_RUN_ON_CPU) == 0 || + cpu == smp_processor_id()) { + __bpf_prog_test_run_raw_tp(&info); + } else { + /* smp_call_function_single() also checks cpu_online() + * after csd_lock(). However, since cpu_plus is from user + * space, let's do an extra quick check to filter out + * invalid value before smp_call_function_single(). + */ + if (!cpu_online(cpu)) { + err = -ENXIO; + goto out; + } + + err = smp_call_function_single(cpu, __bpf_prog_test_run_raw_tp, + &info, 1); + if (err) + goto out; + } + + if (copy_to_user(&uattr->test.retval, &info.retval, sizeof(u32))) + err = -EFAULT; + +out: + kfree(info.ctx); + return err; +} + static void *bpf_ctx_init(const union bpf_attr *kattr, u32 max_size) { void __user *data_in = u64_to_user_ptr(kattr->test.ctx_in); @@ -410,6 +493,9 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr, void *data; int ret; + if (kattr->test.flags || kattr->test.cpu) + return -EINVAL; + data = bpf_test_init(kattr, size, NET_SKB_PAD + NET_IP_ALIGN, SKB_DATA_ALIGN(sizeof(struct skb_shared_info))); if (IS_ERR(data)) @@ -607,6 +693,9 @@ int bpf_prog_test_run_flow_dissector(struct bpf_prog *prog, if (prog->type != BPF_PROG_TYPE_FLOW_DISSECTOR) return -EINVAL; + if (kattr->test.flags || kattr->test.cpu) + return -EINVAL; + if (size < ETH_HLEN) return -EINVAL; diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index a22812561064a..05e480f66f475 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -424,6 +424,11 @@ enum { */ #define BPF_F_QUERY_EFFECTIVE (1U << 0) +/* Flags for BPF_PROG_TEST_RUN */ + +/* If set, run the test on the cpu specified by bpf_attr.test.cpu */ +#define BPF_F_TEST_RUN_ON_CPU (1U << 0) + /* type for BPF_ENABLE_STATS */ enum bpf_stats_type { /* enabled run_time_ns and run_cnt */ @@ -566,6 +571,8 @@ union bpf_attr { */ __aligned_u64 ctx_in; __aligned_u64 ctx_out; + __u32 flags; + __u32 cpu; } test; struct { /* anonymous struct used by BPF_*_GET_*_ID */