Re: [PATCH v3] PE/COFF: raise normal PE section limit safely

On 10.07.2026 15:14, Oleg Tolmatcev wrote:
> Am Fr., 10. Juli 2026 um 15:03 Uhr schrieb Jan Beulich <[email protected]>:
>>
>> On 06.07.2026 00:07, Oleg Tolmatcev wrote:
>>> @@ -927,6 +933,8 @@ fill_comdat_hash (bfd *abfd)
>>>        flagword sec_flags = SEC_LINK_ONCE;
>>>
>>>        bfd_coff_swap_sym_in (abfd, esym, &isym);
>>> +      if (bfd_get_error () == bfd_error_bad_value)
>>> +     return false;
>>
>> Taking this as example (applies at all use sites): I'm not sure what to say.
>> What if bfd_coff_swap_sym_in() worked fine and something else left an error
>> latched in libbfd? This kind of indirect error checking feels pretty
>> fragile. If already all use sites of bfd_coff_swap_sym_in() need touching
>> anyway, can't the function be changed to properly report a success indicator?
>>
>> Having said "all" - you don't really amend all call sites, do you? The
>> description says nothing as to how the selection was made.
> 
> I wanted to keep the changes to error handling minimal, but now I think
> that I shouldn't have touched it at all. The original code did not do any
> error handling either. If the only thing wrong with v1 of the patch
> was the assert,
> should I just remove the assert from it and be done with the patch?

I think I could live with that.

> Otherwise I don't know what to do.

Well, I did outline one option. Sooner or later proper error handling will
want adding anyway, imo. But yes, it doesn't need to be you who does it.

Jan