[PATCH bpf-next v3 01/10] bpf: btf: Add btf_type_by_name_kind

[KP Singh]

On 23-Jan 12:06, Andrii Nakryiko wrote:
> On Thu, Jan 23, 2020 at 7:25 AM KP Singh <kpsingh at chromium.org> wrote:
> >
> > From: KP Singh <kpsingh at google.com>
> >
> > - The LSM code does the combination of btf_find_by_name_kind and
> >   btf_type_by_id a couple of times to figure out the BTF type for
> >   security_hook_heads and security_list_options.
> > - Add an extern for btf_vmlinux in btf.h
> >
> > Signed-off-by: KP Singh <kpsingh at google.com>
> > Reviewed-by: Brendan Jackman <jackmanb at google.com>
> > Reviewed-by: Florent Revest <revest at google.com>
> > Reviewed-by: Thomas Garnier <thgarnie at google.com>
> > ---
> >  include/linux/btf.h |  3 +++
> >  kernel/bpf/btf.c    | 12 ++++++++++++
> >  2 files changed, 15 insertions(+)
> >
> > diff --git a/include/linux/btf.h b/include/linux/btf.h
> > index 5c1ea99b480f..d4e859f90a39 100644
> > --- a/include/linux/btf.h
> > +++ b/include/linux/btf.h
> > @@ -15,6 +15,7 @@ struct btf_type;
> >  union bpf_attr;
> >
> >  extern const struct file_operations btf_fops;
> > +extern struct btf *btf_vmlinux;
> >
> >  void btf_put(struct btf *btf);
> >  int btf_new_fd(const union bpf_attr *attr);
> > @@ -66,6 +67,8 @@ const struct btf_type *
> >  btf_resolve_size(const struct btf *btf, const struct btf_type *type,
> >                  u32 *type_size, const struct btf_type **elem_type,
> >                  u32 *total_nelems);
> > +const struct btf_type *btf_type_by_name_kind(
> > +       struct btf *btf, const char *name, u8 kind);
> >
> >  #define for_each_member(i, struct_type, member)                        \
> >         for (i = 0, member = btf_type_member(struct_type);      \
> > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> > index 32963b6d5a9c..ea53c16802cb 100644
> > --- a/kernel/bpf/btf.c
> > +++ b/kernel/bpf/btf.c
> > @@ -441,6 +441,18 @@ const struct btf_type *btf_type_resolve_func_ptr(const struct btf *btf,
> >         return NULL;
> >  }
> >
> > +const struct btf_type *btf_type_by_name_kind(
> > +       struct btf *btf, const char *name, u8 kind)
> > +{
> > +       s32 type_id;
> > +
> > +       type_id = btf_find_by_name_kind(btf, name, kind);
> > +       if (type_id < 0)
> > +               return ERR_PTR(type_id);
> > +
> > +       return btf_type_by_id(btf, type_id);
> > +}
> > +
> 
> is it worth having this as a separate global function? If
> btf_find_by_name_kind returns valid ID, then you don't really need to
> check btf_type_by_id result, it is always going to be valid. So the
> pattern becomes:

Yeah you're right. We went from using this a few times in separate
functions to 2 times in the same function (based on the changes in
v2 -> v3)

I will drop this from the next revision.

 
> type_id = btf_find_by_name_kind(btf, name, kind);
> if (type_id < 0)
>   goto handle_error;
> t = btf_type_by_id(btf, type_id);
> /* now just use t */
> 
> which is not much more verbose than:
> 
> t = btf_type_by_name_kind(btf, name, kind);
> if (IS_ERR(t))
>   goto handle_error
> /* now use t */
> 

Agreed.

- KP

> 
> >  /* Types that act only as a source, not sink or intermediate
> >   * type when resolving.
> >   */
> > --
> > 2.20.1
> >