Re: [PATCH] elf: Add elf checks for main executable

[Adhemerval Zanella <adhemerval.zanella@linaro.org>]

On 06/12/2021 17:37, Florian Weimer wrote:
> * Adhemerval Zanella:
> 
>> On 06/12/2021 16:22, H.J. Lu wrote:
>>> On Mon, Dec 6, 2021 at 11:09 AM Florian Weimer <fweimer@redhat.com> wrote:
>>>>
>>>> * Adhemerval Zanella:
>>>>
>>>>> +static void
>>>>> +modify_abiversion (ElfW(Ehdr) *ehdr)
>>>>> +{
>>>>> +  ehdr->e_ident[EI_ABIVERSION] = LIBC_ABI_MAX;
>>>>> +}
>>>>
>>>> So this is eventually controlled by the libc-abis file, right?
>>>>
>>>> I *thought* that the consensus was that binutils should bump version if
>>>> absolute symbols are used.  But I don't see that in the absolute symbol
>>>> tests.
>>>>
>>>> Is this really doing anything?
>>>>
>>>
>>> EI_ABIVERSION check works on executables created by the new linker
>>> which bumps EI_ABIVERSION.  This complements the existing
>>> EI_ABIVERSION check on DSOs.  This is orthogonal to the ABI version
>>> check for existing ld.so binaries which needs an ABIVERSION version.
>>>
>>
>> Currently only mips does actually set and checks different EI_ABIVERSION
>> through VALID_ELF_ABIVERSION. For instance, -Wl,--hash-style=gnu with
>> mips64 will set EI_ABIVERSION to 5.
> 
> What surprised me is that binutils does not seem to *generate* the bump
> for ABSOLUTE.  Not so much the lack of checking in glibc.

It does, but only for specific configurations.  bintutils does have a
testcase for it, pr21375*, but not all configurations does bump because
they do not require ABS relocations (for instance, for n64 -mmicromips
--defsym hidn=1 does set the ABI version to 4).