Re: [PATCH bpf-next v2 01/35] bpf: memcg-based memory accounting for bpf progs

[Roman Gushchin <guro@xxxxxx>]

On Mon, Jul 27, 2020 at 03:11:42PM -0700, Song Liu wrote:
> On Mon, Jul 27, 2020 at 12:20 PM Roman Gushchin <guro@xxxxxx> wrote:
> >
> > Include memory used by bpf programs into the memcg-based accounting.
> > This includes the memory used by programs itself, auxiliary data
> > and statistics.
> >
> > Signed-off-by: Roman Gushchin <guro@xxxxxx>
> > ---
> >  kernel/bpf/core.c | 8 ++++----
> >  1 file changed, 4 insertions(+), 4 deletions(-)
> >
> > diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
> > index bde93344164d..daab8dcafbd4 100644
> > --- a/kernel/bpf/core.c
> > +++ b/kernel/bpf/core.c
> > @@ -77,7 +77,7 @@ void *bpf_internal_load_pointer_neg_helper(const struct sk_buff *skb, int k, uns
> >
> >  struct bpf_prog *bpf_prog_alloc_no_stats(unsigned int size, gfp_t gfp_extra_flags)
> >  {
> > -       gfp_t gfp_flags = GFP_KERNEL | __GFP_ZERO | gfp_extra_flags;
> > +       gfp_t gfp_flags = GFP_KERNEL_ACCOUNT | __GFP_ZERO | gfp_extra_flags;
> >         struct bpf_prog_aux *aux;
> >         struct bpf_prog *fp;
> >
> > @@ -86,7 +86,7 @@ struct bpf_prog *bpf_prog_alloc_no_stats(unsigned int size, gfp_t gfp_extra_flag
> >         if (fp == NULL)
> >                 return NULL;
> >
> > -       aux = kzalloc(sizeof(*aux), GFP_KERNEL | gfp_extra_flags);
> > +       aux = kzalloc(sizeof(*aux), GFP_KERNEL_ACCOUNT | gfp_extra_flags);
> >         if (aux == NULL) {
> >                 vfree(fp);
> >                 return NULL;
> > @@ -104,7 +104,7 @@ struct bpf_prog *bpf_prog_alloc_no_stats(unsigned int size, gfp_t gfp_extra_flag
> >
> >  struct bpf_prog *bpf_prog_alloc(unsigned int size, gfp_t gfp_extra_flags)
> >  {
> > -       gfp_t gfp_flags = GFP_KERNEL | __GFP_ZERO | gfp_extra_flags;
> > +       gfp_t gfp_flags = GFP_KERNEL_ACCOUNT | __GFP_ZERO | gfp_extra_flags;
> >         struct bpf_prog *prog;
> >         int cpu;
> >
> > @@ -217,7 +217,7 @@ void bpf_prog_free_linfo(struct bpf_prog *prog)
> >  struct bpf_prog *bpf_prog_realloc(struct bpf_prog *fp_old, unsigned int size,
> >                                   gfp_t gfp_extra_flags)
> >  {
> > -       gfp_t gfp_flags = GFP_KERNEL | __GFP_ZERO | gfp_extra_flags;
> > +       gfp_t gfp_flags = GFP_KERNEL_ACCOUNT | __GFP_ZERO | gfp_extra_flags;
> >         struct bpf_prog *fp;
> >         u32 pages, delta;
> >         int ret;
> > --

Hi Song!

Thank you for looking into the patchset!

> 
> Do we need similar changes in
> 
> bpf_prog_array_copy()
> bpf_prog_alloc_jited_linfo()
> bpf_prog_clone_create()
> 
> and maybe a few more?

I've tried to follow the rlimit-based accounting, so those objects which were
skipped are mostly skipped now and vice versa. The main reason for that is
simple: I don't know many parts of bpf code well enough to decide whether
we need accounting or not.

In general with memcg-based accounting we can easily cover places which were
not covered previously: e.g. the memory used by the verifier. But I guess it's
better to do it case-by-case.

But if you're aware of any big objects which should be accounted for sure,
please, let me know.

Thanks!