On Tue, Oct 03, 2017 at 05:20:25PM -0700, Jakub Kicinski wrote:
> The only global state protected by the verifier lock is
> the verifier log. Move that log to struct bpf_verifier_env
> and allow verification of multiple programs in parallel.
>
> Signed-off-by: Jakub Kicinski <[email protected]>
> ---
> Like this? :)
>
> Compile-tested, I'm rebasing things and will test shortly.
>
> include/linux/bpf_verifier.h | 5 +
> kernel/bpf/verifier.c | 520
> ++++++++++++++++++++++---------------------
> 2 files changed, 276 insertions(+), 249 deletions(-)
>
> diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h
> index b8d200f60a40..598802dd1897 100644
> --- a/include/linux/bpf_verifier.h
> +++ b/include/linux/bpf_verifier.h
> @@ -139,6 +139,11 @@ struct bpf_verifier_env {
> bool allow_ptr_leaks;
> bool seen_direct_write;
> struct bpf_insn_aux_data *insn_aux_data; /* array of per-insn state */
> +
> + u32 log_level;
> + u32 log_size;
> + u32 log_len;
> + char *log_buf;
> };
>
> int bpf_analyzer(struct bpf_prog *prog, const struct bpf_ext_analyzer_ops
> *ops,
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 4cf9b72c59a0..450f60e6229d 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -153,27 +153,21 @@ struct bpf_call_arg_meta {
> int access_size;
> };
>
> -/* verbose verifier prints what it's seeing
> - * bpf_check() is called under lock, so no race to access these global vars
> - */
> -static u32 log_level, log_size, log_len;
> -static char *log_buf;
> -
> -static DEFINE_MUTEX(bpf_verifier_lock);
yep. looks great.
Please test it and submit officially :)
The commit aafe6ae9cee3 ("bpf: dynamically allocate digest scratch buffer")
fixed the other case where we were relying on the above mutex.
The only other spot to be adjusted is to add spin_lock/mutex or DO_ONCE() to
bpf_get_skb_set_tunnel_proto() to protect md_dst init.
imo that would be it.
Daniel, anything else comes to mind?
