Home Original page

[PATCH,V3 6/9] bfd: gas: ld: libsframe: adopt new encoding for FDE func start addr field

Jan Beulich jbeulich@suse.com
Tue Jun 17 05:28:06 GMT 2025 
On 16.06.2025 23:02, Indu Bhagat wrote:
> On 6/16/25 12:23 AM, Jan Beulich wrote:
>> On 16.06.2025 08:43, Indu Bhagat wrote:
>>> On 6/13/25 5:39 AM, Jan Beulich wrote:
>>>> On 13.06.2025 09:32, Indu Bhagat wrote:
>>>>> --- a/bfd/elf-sframe.c
>>>>> +++ b/bfd/elf-sframe.c
>>>>> @@ -328,6 +328,8 @@ _bfd_elf_merge_section_sframe (bfd *abfd,
>>>>>      int8_t sfd_ctx_fixed_ra_offset;
>>>>>      uint8_t dctx_version;
>>>>>      uint8_t ectx_version;
>>>>> +  uint8_t dctx_flags;
>>>>> +  uint8_t ectx_flags;
>>>>>      int encerr = 0;
>>>>>         struct elf_link_hash_table *htab;
>>>>> @@ -351,6 +353,8 @@ _bfd_elf_merge_section_sframe (bfd *abfd,
>>>>>      if (sfd_ctx == NULL || sfe_info == NULL)
>>>>>        return false;
>>>>>    +  dctx_flags = sframe_decoder_get_flags (sfd_ctx);
>>>>> +
>>>>>      if (htab->sfe_info.sfe_ctx == NULL)
>>>>>        {
>>>>>          sfd_ctx_abi_arch = sframe_decoder_get_abi_arch (sfd_ctx);
>>>>> @@ -361,8 +365,12 @@ _bfd_elf_merge_section_sframe (bfd *abfd,
>>>>>          if (!sfd_ctx_abi_arch)
>>>>>        return false;
>>>>>    +      /* Reset SFRAME_F_FDE_SORTED for the encoder context.  This will be set
>>>>> +     later when FDEs are finally sorted before emission in the output
>>>>> +     section.  */
>>>>> +      uint8_t tflags = dctx_flags & ~SFRAME_F_FDE_SORTED;
>>>>>          htab->sfe_info.sfe_ctx = sframe_encode (SFRAME_VERSION_2,
>>>>> -                          0, /* SFrame flags.  */
>>>>> +                          tflags, /* SFrame flags.  */
>>>>>                              sfd_ctx_abi_arch,
>>>>>                              sfd_ctx_fixed_fp_offset,
>>>>>                              sfd_ctx_fixed_ra_offset,
>>>>
>>>> I'm likely confused here, too: Why would the one flag need special casing?
>>
>> You didn't address this question.
>>
> 
> (Aplogies I mixed it up in the other paragraph, but here is why we special case SFRAME_F_FDE_SORTED...)
> 
> Currently the three flags defined are such that _only_ SFRAME_F_FDE_SORTED needs reset at the time of linking, because this property in broken while the sections are being linked in memory (as the SFrame FDEs are being placed in the output section) until the eventual sort at write time.

And why does the intermediate state of the flag matter?

>>>> What about flags which are set, but which aren't defined right now?
>>>
>>> Flags which arent defined are not set.  The rest of the bits (bits at
>>> position [4:8] of the flags field should be zero.  Perhaps I should add
>>> a line in the spec document that undefined bits in the flags field are
>>> expected to be zero.
>>
>> And then, if another flag gets defined later, how are older binutils
>> supposed to be dealing with such an object? Imo they simply ought to
>> refuse handling it.
>>
> 
> We have a check in sframe_header_sanity_check_p () in libsframe/sframe.c where we check (hp->sfh_preamble.sfp_flags | all_flags) != all_flags), where all_flags was defined as (SFRAME_F_FDE_SORTED | SFRAME_F_FRAME_POINTER | SFRAME_F_FDE_FUNC_START_ADDR_PCREL).
> 
> The sanity check is done for every sframe_decode () (used by ld and the dumping code).  So older binutils are effectively rejecting objects with new flags set.

Ah, okay, that's what I failed to spot then.

Jan
More information about the Binutils mailing list