Re: Strange symbols_lookup() behaviour in test-symbols on arm64 CI

[Jan Beulich <jbeulich@xxxxxxxx>]

On 11.12.2025 16:57, Mykola Kvach wrote:
> On Thu, Dec 11, 2025 at 5:31 PM Jan Beulich <jbeulich@xxxxxxxx> wrote:
>>
>> On 11.12.2025 13:06, Mykola Kvach wrote:
>>> On Thu, Dec 11, 2025 at 1:44 PM Mykola Kvach <xakep.amatop@xxxxxxxxx> wrote:
>>>> I tested the issue with the provided patch, and it is still reproducible.
>>>
>>> I also triggered a pipeline with the patch you provided on top of my
>>> changes [1],
>>> and it failed there as well. In fact, it exposed additional problems
>>> in other jobs.
>>>
>>> [1] 
>>> https://gitlab.com/xen-project/people/mykola_kvach/xen/-/pipelines/2209153728
>>
>> These are the failures we're trying to hunt down.
>>
>> Looking at one of the Arm build logs, I can't see any indication that you 
>> actually
>> extended the patch to Arm's linking rule. Hence my conclusion in the other 
>> reply
>> may be wrong - we simply still don't know what the situation is there.
> 
> You're right, I missed the necessary changes. Would this diff be sufficient?

Yes, this looks to be what is needed.

Jan

> --- a/xen/arch/arm/Makefile
> +++ b/xen/arch/arm/Makefile
> @@ -100,6 +100,7 @@ $(TARGET)-syms: $(objtree)/prelink.o $(obj)/xen.lds
>                 | $(objtree)/tools/symbols $(all_symbols) --sysv --sort \
>                 > $(dot-target).2.S
>         $(MAKE) $(build)=$(@D) $(dot-target).2.o
> +       $(call compare-symbol-tables,$(dot-target).1.o,$(dot-target).2.o)
>         $(LD) $(XEN_LDFLAGS) -T $(obj)/xen.lds $< $(build_id_linker) \
>             $(dot-target).2.o -o $@
>         $(NM) -pa --format=sysv $@ \
> 
> 
>>
>> Jan
> 
> Best regards,
> Mykola