Re: [PATCH v2 dwarves 2/5] btf_encoder: Add true_signature feature support for "."-suffixed functions

On 26/01/2026 11:18, Matt Bobrowski wrote:
> On Mon, Jan 26, 2026 at 10:49:00AM +0000, Alan Maguire wrote:
>> On 26/01/2026 09:52, Matt Bobrowski wrote:
>>> On Fri, Jan 23, 2026 at 05:26:47PM +0000, Alan Maguire wrote:
>>>> Currently we collate function information by name and add functions
>>>> provided there are no inconsistencies across various representations.
>>>>
>>>> For true_signature support - where we wish to add the real signature
>>>> of a function even if it differs from source level - we need to do
>>>> a few things:
>>>>
>>>> 1. For "."-suffixed functions, we need to match from DWARF->ELF;
>>>>    we can do this via the address associated with the function.
>>>>    In doing this, we can then be confident that the debug info
>>>>    for foo.isra.0 is the right info for the function at that
>>>>    address.
>>>>
>>>> 2. When adding saved functions we need to look for such cases
>>>>    and provided they do not violate other constraints around BTF
>>>>    representation - unexpected reg usage for function, uncertain
>>>>    parameter location or ambiguous address - we add them with
>>>>    their "."-suffixed name.  The latter can be used as a signal
>>>>    that the function is transformed from the original.
>>>>
>>>> Doing this adds 500 functions to BTF.  These are traceable with
>>>> their "."-suffix names and because we have excluded ambiguous
>>>> address cases we know exactly which function address they refer
>>>> to.
>>>>
>>>> Signed-off-by: Alan Maguire <[email protected]>
>>>
>>> Some minor nits, but apart from that it looks OK to me.
>>>
>>> Acked-by: Matt Bobrowski <[email protected]>
>>>
>>>> ---
>>>>  btf_encoder.c | 76 ++++++++++++++++++++++++++++++++++++++++++++++-----
>>>>  dwarves.h     |  1 +
>>>>  pahole.c      |  1 +
>>>>  3 files changed, 71 insertions(+), 7 deletions(-)
>>>>
>>>> diff --git a/btf_encoder.c b/btf_encoder.c
>>>> index 9a567e4..c1002c3 100644
>>>> --- a/btf_encoder.c
>>>> +++ b/btf_encoder.c
>>>> @@ -77,9 +77,16 @@ struct btf_encoder_func_annot {
>>>>  	int16_t component_idx;
>>>>  };
>>>>  
>>>> +struct elf_function_sym {
>>>> +	const char *name;
>>>> +	uint64_t addr;
>>>> +};
>>>> +
>>>>  /* state used to do later encoding of saved functions */
>>>>  struct btf_encoder_func_state {
>>>>  	struct elf_function *elf;
>>>> +	struct elf_function_sym *sym;
>>>> +	uint64_t addr;
>>>   		 ^
>>> 		 This appears to have leaked into wrong commit?
>>> 		 This member should've been introduced within patch
>>> 		 5/5.
>>>
>>
>> It's used here too though; specifically in btf_encoder__save_func()
>> below. We need to use addresses to map between a DWARF representation
>> and the associated ELF function to ensure we're using debug info
>> for the correct "."-suffixed function. We need to do this because
>> the late DWARF generated after optimizations are applied uses
>> the original name ("foo" rather than "foo.isra.0").
> 
> Sorry, I thought I only saw usage of member "addr" within
> btf_encoder__save_func() against fn->lenblock.ip.addr which maps to
> member "addr" within struct ip_tag and func->syms[i].addr which maps
> to member "addr" within struct elf_function_sym. Member "addr" within
> struct btf_encoder_func_state remains unused until patch 5/5 where we
> assign it the value returned from function__addr().
> 
> With that said, within btf_encoder__save_func() why don't you do the
> following instead:
> 
> state->addr = function__addr(fn);
> 
> ...
> 
> if (encoder->true_signature && state->addr) {
> 	if (state->addr != func->syms[i].addr) {
> 		...
> 	}
> }
> 
> There's no need to open code fn->lexblock.ip.addr everywhere as it
> deminishes readability.
>

yep, good suggestion, will do this for v3 which should be ready shortly..