Re: [PATCH bpf-next v2 6/6] docs, resolve_btfids: Document kfunc BTF annotation emission

On 8/6/26 12:47 PM, Eduard Zingerman wrote:
> On Wed, 2026-08-05 at 16:06 -0700, Ihor Solodrai wrote:
> 
> ...
> 
>> diff --git a/Documentation/bpf/kfuncs.rst b/Documentation/bpf/kfuncs.rst
>> index cbde86d082cc..c60fc574e8b0 100644
>> --- a/Documentation/bpf/kfuncs.rst
>> +++ b/Documentation/bpf/kfuncs.rst
>> @@ -472,6 +472,14 @@ type. An example is shown below::
>>          }
>>          late_initcall(init_subsystem);
>>  
>> +At kernel build time the ``resolve_btfids`` tool discovers all kfuncs from the
>> +registered ``BTF_SET8_KFUNCS`` sets and emits their BTF annotations into the
> 
> Note that this is a single occurrence of the word BTF_SET8_KFUNCS in
> this .rst file. Also The wording "registered" is confusing as the
> above code snippet shows the usage of register_btf_kfunc_id_set()
> function, which is completely unrelated.
> 
>> +kernel's BTF; these annotations were historically produced by pahole. For each
> 
> I'd skip a note about pahole, it does not convey usable information.

ack

> 
>> +discovered kfunc ``resolve_btfids`` emits a ``bpf_kfunc`` BTF decl tag, a
>> +``bpf_fastcall`` decl tag when the kfunc is flagged ``KF_FASTCALL``, and the
>> +``address_space(1)`` type attribute on the return value and/or arguments flagged
>> +``KF_ARENA_RET``, ``KF_ARENA_ARG1`` or ``KF_ARENA_ARG2`` (see section 2.8).
>> +
>>  2.7  Specifying no-cast aliases with ___init
>>  --------------------------------------------
>>  
>> diff --git a/Documentation/process/changes.rst b/Documentation/process/changes.rst
>> index 1ca8c5f73ad0..6d1dbe4abf0f 100644
>> --- a/Documentation/process/changes.rst
>> +++ b/Documentation/process/changes.rst
>> @@ -147,10 +147,9 @@ Since Linux 5.2, if CONFIG_DEBUG_INFO_BTF is selected, the build system
>>  generates BTF (BPF Type Format) from DWARF in vmlinux, a bit later from kernel
>>  modules as well.  This requires pahole v1.22 or later.
>>  
>> -Since Linux 7.0, kfuncs annotated with KF_IMPLICIT_ARGS require pahole v1.26
>> -or later.  Without it, such kfuncs will have incorrect BTF prototypes in
>> -vmlinux, causing BPF programs to fail to load with a "func_proto incompatible
>> -with vmlinux" error.  Many sched_ext kfuncs are affected.
>> +Kfunc BTF annotations (the bpf_kfunc and bpf_fastcall decl tags and the arena
>> +address_space(1) type attribute) are emitted in-tree by resolve_btfids from the
>> +BTF_KFUNCS sets, so they no longer depend on a specific pahole version.
> 
> Just drop the whole paragraph?

hmm... yeah you're right

> 
>>  
>>  It is found in the 'dwarves' or 'pahole' distro packages or from
>>  https://fedorapeople.org/~acme/dwarves/.
>> diff --git a/scripts/Makefile.btf b/scripts/Makefile.btf
>> index a1812985a61a..717e76ce96a7 100644
>> --- a/scripts/Makefile.btf
>> +++ b/scripts/Makefile.btf
>> @@ -14,6 +14,9 @@ pahole-flags-$(call test-ge, $(pahole-ver), 125)	+= --skip_encoding_btf_inconsis
>>  else
>>  
>>  # Switch to using --btf_features for v1.26 and later.
>> +#
>> +# kfunc BTF annotations (bpf_kfunc/bpf_fastcall decl tags and the arena
>> +# address_space(1) type attribute) are emitted by resolve_btfids, not pahole.
> 
> What's the point of this comment?

The point is to inform the reader "where did decl_tag_kfuncs go?".
Question is whether the git log will be enough, or is a comment
also appropriate?

> 
>>  pahole-flags-$(call test-ge, $(pahole-ver), 126)  = -j$(JOBS) --btf_features=encode_force,var,float,enum64,decl_tag,type_tag,optimized_func,consistent_func
>>  
>>  pahole-flags-$(call test-ge, $(pahole-ver), 131) += --btf_features=layout
> 
> ...