I messed up the mailing lists and will be resending this patch series in a moment. Apologies for the noise. Best, Joe Burton On Tue, Oct 26, 2021 at 3:27 PM Joe Burton <jevburton.kernel@xxxxxxxxx> wrote: > > From: Joe Burton <jevburton@xxxxxxxxxx> > > Add a flag to `enum libbpf_strict_mode' to disable the global > `bpf_objects_list', preventing race conditions when concurrent threads > call bpf_object__open() or bpf_object__close(). > > bpf_object__next() will return NULL if this option is set. > > Callers may achieve the same workflow by tracking bpf_objects in > application code. > > [0] Closes: https://github.com/libbpf/libbpf/issues/293 > > Signed-off-by: Joe Burton <jevburton@xxxxxxxxxx> > --- > tools/lib/bpf/libbpf.c | 8 +++++++- > tools/lib/bpf/libbpf.h | 3 ++- > tools/lib/bpf/libbpf_legacy.h | 6 ++++++ > 3 files changed, 15 insertions(+), 2 deletions(-) > > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c > index 2fbed2d4a645..59d39ce9f375 100644 > --- a/tools/lib/bpf/libbpf.c > +++ b/tools/lib/bpf/libbpf.c > @@ -1148,6 +1148,7 @@ static struct bpf_object *bpf_object__new(const char *path, > size_t obj_buf_sz, > const char *obj_name) > { > + bool strict = (libbpf_mode & LIBBPF_STRICT_NO_OBJECT_LIST); > struct bpf_object *obj; > char *end; > > @@ -1188,7 +1189,8 @@ static struct bpf_object *bpf_object__new(const char *path, > obj->loaded = false; > > INIT_LIST_HEAD(&obj->list); > - list_add(&obj->list, &bpf_objects_list); > + if (!strict) > + list_add(&obj->list, &bpf_objects_list); > return obj; > } > > @@ -7935,6 +7937,10 @@ struct bpf_object * > bpf_object__next(struct bpf_object *prev) > { > struct bpf_object *next; > + bool strict = (libbpf_mode & LIBBPF_STRICT_NO_OBJECT_LIST); > + > + if (strict) > + return NULL; > > if (!prev) > next = list_first_entry(&bpf_objects_list, > diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h > index e1900819bfab..defabdbe7760 100644 > --- a/tools/lib/bpf/libbpf.h > +++ b/tools/lib/bpf/libbpf.h > @@ -168,7 +168,8 @@ LIBBPF_API struct bpf_program * > bpf_object__find_program_by_name(const struct bpf_object *obj, > const char *name); > > -LIBBPF_API struct bpf_object *bpf_object__next(struct bpf_object *prev); > +LIBBPF_API LIBBPF_DEPRECATED_SINCE(0, 7, "track bpf_objects in application code instead") > +struct bpf_object *bpf_object__next(struct bpf_object *prev); > #define bpf_object__for_each_safe(pos, tmp) \ > for ((pos) = bpf_object__next(NULL), \ > (tmp) = bpf_object__next(pos); \ > diff --git a/tools/lib/bpf/libbpf_legacy.h b/tools/lib/bpf/libbpf_legacy.h > index 29ccafab11a8..5ba5c9beccfa 100644 > --- a/tools/lib/bpf/libbpf_legacy.h > +++ b/tools/lib/bpf/libbpf_legacy.h > @@ -57,6 +57,12 @@ enum libbpf_strict_mode { > * function name instead of section name. > */ > LIBBPF_STRICT_SEC_NAME = 0x04, > + /* > + * Disable the global 'bpf_objects_list'. Maintaining this list adds > + * a race condition to bpf_object__open() and bpf_object__close(). > + * Clients can maintain it on their own if it is valuable for them. > + */ > + LIBBPF_STRICT_NO_OBJECT_LIST = 0x08, > > __LIBBPF_STRICT_LAST, > }; > -- > 2.33.0.1079.g6e70778dc9-goog >
