Re: [PATCH dwarves v6 1/6] bpf, x86: Generate trampolines from bpf_tramp_links

[Kui-Feng Lee <kuifeng@fb.com>]

On Wed, 2022-04-20 at 10:37 -0700, Andrii Nakryiko wrote:
> On Fri, Apr 15, 2022 at 9:30 PM Kui-Feng Lee <kuifeng@fb.com> wrote:
> > 
> > Replace struct bpf_tramp_progs with struct bpf_tramp_links to
> > collect
> > struct bpf_tramp_link(s) for a trampoline.  struct bpf_tramp_link
> > extends bpf_link to act as a linked list node.
> > 
> > arch_prepare_bpf_trampoline() accepts a struct bpf_tramp_links to
> > collects all bpf_tramp_link(s) that a trampoline should call.
> > 
> > Change BPF trampoline and bpf_struct_ops to pass bpf_tramp_links
> > instead of bpf_tramp_progs.
> > 
> > Signed-off-by: Kui-Feng Lee <kuifeng@fb.com>
> > ---
> >  arch/x86/net/bpf_jit_comp.c    | 36 +++++++++--------
> >  include/linux/bpf.h            | 36 +++++++++++------
> >  include/linux/bpf_types.h      |  1 +
> >  include/uapi/linux/bpf.h       |  1 +
> >  kernel/bpf/bpf_struct_ops.c    | 69 ++++++++++++++++++++++--------
> > --
> >  kernel/bpf/syscall.c           | 23 ++++-------
> >  kernel/bpf/trampoline.c        | 73 +++++++++++++++++++-----------
> > ----
> >  net/bpf/bpf_dummy_struct_ops.c | 37 ++++++++++++++---
> >  tools/bpf/bpftool/link.c       |  1 +
> >  tools/include/uapi/linux/bpf.h |  1 +
> >  10 files changed, 175 insertions(+), 103 deletions(-)
> > 
> 
> [...]
> 
> > @@ -385,6 +399,7 @@ static int
> > bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
> >         for_each_member(i, t, member) {
> >                 const struct btf_type *mtype, *ptype;
> >                 struct bpf_prog *prog;
> > +               struct bpf_tramp_link *link;
> >                 u32 moff;
> > 
> >                 moff = __btf_member_bit_offset(t, member) / 8;
> > @@ -438,16 +453,26 @@ static int
> > bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
> >                         err = PTR_ERR(prog);
> >                         goto reset_unlock;
> >                 }
> > -               st_map->progs[i] = prog;
> > 
> >                 if (prog->type != BPF_PROG_TYPE_STRUCT_OPS ||
> >                     prog->aux->attach_btf_id != st_ops->type_id ||
> >                     prog->expected_attach_type != i) {
> > +                       bpf_prog_put(prog);
> >                         err = -EINVAL;
> >                         goto reset_unlock;
> >                 }
> > 
> > -               err = bpf_struct_ops_prepare_trampoline(tprogs,
> > prog,
> > +               link = kzalloc(sizeof(*link), GFP_USER);
> 
> seems like you are leaking this link and all the links allocated in
> previous successful iterations of this loop?
> 
> > +               if (!link) {
> > +                       bpf_prog_put(prog);
> > +                       err = -ENOMEM;
> > +                       goto reset_unlock;
> > +               }
> > +               bpf_link_init(&link->link,
> > BPF_LINK_TYPE_STRUCT_OPS,
> > +                             &bpf_struct_ops_link_lops, prog);
> > +               st_map->links[i] = &link->link;
> > +
> > +               err = bpf_struct_ops_prepare_trampoline(tlinks,
> > link,
> >                                                         &st_ops-
> > >func_models[i],
> >                                                         image,
> > image_end);
> >                 if (err < 0)
> > @@ -490,7 +515,7 @@ static int
> > bpf_struct_ops_map_update_elem(struct bpf_map *map, void *key,
> >         memset(uvalue, 0, map->value_size);
> >         memset(kvalue, 0, map->value_size);
> >  unlock:
> > -       kfree(tprogs);
> > +       kfree(tlinks);
> 
> so you'll need to free those links inside tlinks (or wherever else
> they are stored)
> 
> >         mutex_unlock(&st_map->lock);
> >         return err;
> >  }
> > @@ -545,9 +570,9 @@ static void bpf_struct_ops_map_free(struct
> > bpf_map *map)
> >  {
> >         struct bpf_struct_ops_map *st_map = (struct
> > bpf_struct_ops_map *)map;
> > 
> > -       if (st_map->progs)
> > +       if (st_map->links)
> >                 bpf_struct_ops_map_put_progs(st_map);
> > -       bpf_map_area_free(st_map->progs);
> > +       bpf_map_area_free(st_map->links);
> >         bpf_jit_free_exec(st_map->image);
> >         bpf_map_area_free(st_map->uvalue);
> >         bpf_map_area_free(st_map);
> 
> [...]
> 
> > @@ -105,10 +120,20 @@ int bpf_struct_ops_test_run(struct bpf_prog
> > *prog, const union bpf_attr *kattr,
> >         }
> >         set_vm_flush_reset_perms(image);
> > 
> > +       link = kzalloc(sizeof(*link), GFP_USER);
> > +       if (!link) {
> > +               err = -ENOMEM;
> > +               goto out;
> > +       }
> > +       /* prog doesn't take the ownership of the reference from
> > caller */
> > +       bpf_prog_inc(prog);
> > +       bpf_link_init(&link->link, BPF_LINK_TYPE_STRUCT_OPS,
> > &bpf_struct_ops_link_lops, prog);
> > +
> >         op_idx = prog->expected_attach_type;
> > -       err = bpf_struct_ops_prepare_trampoline(tprogs, prog,
> > +       err = bpf_struct_ops_prepare_trampoline(tlinks, link,
> >                                                 &st_ops-
> > >func_models[op_idx],
> >                                                 image, image +
> > PAGE_SIZE);
> > +
> 
> nit: no need for extra empty line here
> 
> >         if (err < 0)
> >                 goto out;
> > 
> > @@ -124,7 +149,9 @@ int bpf_struct_ops_test_run(struct bpf_prog
> > *prog, const union bpf_attr *kattr,
> >  out:
> >         kfree(args);
> >         bpf_jit_free_exec(image);
> > -       kfree(tprogs);
> > +       if (link)
> > +               bpf_link_put(&link->link);
> 
> you never to bpf_link_prime() and bpf_link_settle() for these "pseudo
> links" for struct_ops, so there is no need for bpf_link_put(), it can
> be just bpf_link_free(), right?

Just realize that bpf_link_free() is a static function of
bpf/syscall.c.  And, this code is for testing only.  So, I don't touch
this line.

> 
> > +       kfree(tlinks);
> >         return err;
> >  }
> > 
> > diff --git a/tools/bpf/bpftool/link.c b/tools/bpf/bpftool/link.c
> > index 8fb0116f9136..6353a789322b 100644
> > --- a/tools/bpf/bpftool/link.c
> > +++ b/tools/bpf/bpftool/link.c
> > @@ -23,6 +23,7 @@ static const char * const link_type_name[] = {
> >         [BPF_LINK_TYPE_XDP]                     = "xdp",
> >         [BPF_LINK_TYPE_PERF_EVENT]              = "perf_event",
> >         [BPF_LINK_TYPE_KPROBE_MULTI]            = "kprobe_multi",
> > +       [BPF_LINK_TYPE_STRUCT_OPS]               = "struct_ops",
> >  };
> > 
> >  static struct hashmap *link_table;
> > diff --git a/tools/include/uapi/linux/bpf.h
> > b/tools/include/uapi/linux/bpf.h
> > index d14b10b85e51..a4f557338af7 100644
> > --- a/tools/include/uapi/linux/bpf.h
> > +++ b/tools/include/uapi/linux/bpf.h
> > @@ -1013,6 +1013,7 @@ enum bpf_link_type {
> >         BPF_LINK_TYPE_XDP = 6,
> >         BPF_LINK_TYPE_PERF_EVENT = 7,
> >         BPF_LINK_TYPE_KPROBE_MULTI = 8,
> > +       BPF_LINK_TYPE_STRUCT_OPS = 9,
> > 
> >         MAX_BPF_LINK_TYPE,
> >  };
> > --
> > 2.30.2
> >