On 30-Mar 19:54, KP Singh wrote:

So, it looks like bpf_tracing_func_proto is only defined when
CONFIG_BPF_EVENTS is set:

        obj-$(CONFIG_BPF_EVENTS) += bpf_trace.o

We have a few options:

* Add a __weak symbol for bpf_tracing_func_proto which we have done in
  the past for similar issues. This however, does not make much sense,
  as CONFIG_BPF_LSM cannot really do much without its helpers.
* Make CONFIG_BPF_LSM depend on CONFIG_BPF_EVENTS, this should solve
  it, but not for this particular Kconfig that was generated. Randy,
  I am assuming if we add the dependency, this particular Kconfig
  won't be generated.

I am assuming this patch now needs to be sent for "bpf" and not
"bpf-next" as the merge window has opened?

- KP

Thanks for adding me Daniel, taking a look.

- KP

On Mon, Mar 30, 2020 at 7:25 PM Daniel Borkmann [off-list ref] wrote:

quoted

[Cc KP, ptal]

On 3/30/20 7:15 PM, Randy Dunlap wrote:

quoted

On 3/30/20 2:43 AM, Stephen Rothwell wrote:

quoted

Hi all,

The merge window has opened, so please do not add any material for the
next release into your linux-next included trees/branches until after
the merge window closes.

Changes since 20200327:

(note: linux-next is based on linux 5.6-rc7)


on i386:

ld: kernel/bpf/bpf_lsm.o:(.rodata+0x0): undefined reference to `bpf_tracing_func_proto'


Full randconfig file is attached.