[PATCH] LD/PE: Remove remains of MIPS target support

[Jan Beulich]

On 07.08.2025 17:00, Jan Beulich wrote:
> On 07.08.2025 16:44, Maciej W. Rozycki wrote:
>> Support for ECOFF MIPS targets, including `mips*-*-pe*' among others, 
>> has been removed from GAS and LD with commit e8044f355dc9 ("Remove gas 
>> and ld support for MIPS ECOFF"), 
>> <https://inbox.sourceware.org/binutils/8761x65bzx.fsf@talisman.default/>.
>> However bits in PEI-format DLL support code have been left behind.
>>
>> Remove the relevant pieces then, originally from commit 344a211f9995 
>> ("Add support for WinCE based toolchains."), including MIPS architecture 
>> parts and HI16/LO16 relocation support in particular.
>> ---
>> Hi,
>>
>>  This has passed my usual verification, including `aarch64-pe', `arm-pe', 
>> `arm-wince-pe', `i686-pe', `mcore-pe', `sh-pe', and `x86_64-mingw32' 
>> targets among others.  It can't be ruled out however that there are gaps 
>> in coverage, so please let me know if you think there's anything I may 
>> have overlooked with this change.
>>
>>  Otherwise OK to apply?
> 
> One question first: Is ...
> 
>> @@ -1699,19 +1688,6 @@ generate_reloc (bfd *abfd, struct bfd_li
>>  		      reloc_data[total_relocs].type = IMAGE_REL_BASED_HIGHLOW;
>>  		      total_relocs++;
>>  		      break;
>> -		    case BITS_AND_SHIFT (16, 0):
>> -		      reloc_data[total_relocs].type = IMAGE_REL_BASED_LOW;
>> -		      total_relocs++;
>> -		      break;
>> -		    case BITS_AND_SHIFT (16, 16):
>> -		      reloc_data[total_relocs].type = IMAGE_REL_BASED_HIGHADJ;
>> -		      /* FIXME: we can't know the symbol's right value
>> -			 yet, but we probably can safely assume that
>> -			 CE will relocate us in 64k blocks, so leaving
>> -			 it zero is safe.  */
>> -		      reloc_data[total_relocs].extra = 0;
>> -		      total_relocs++;
>> -		      break;
>>  		    case BITS_AND_SHIFT (26, 2):
>>  		      reloc_data[total_relocs].type =
>>                          IMAGE_REL_BASED_ARM_MOV32;
>> @@ -1765,9 +1741,6 @@ generate_reloc (bfd *abfd, struct bfd_li
>>  	}
>>  
>>        reloc_sz += 2;
>> -
>> -      if (reloc_data[i].type == IMAGE_REL_BASED_HIGHADJ)
>> -	reloc_sz += 2;
>>      }
>>  
>>    reloc_sz = (reloc_sz + 3) & ~3;	/* 4-byte align.  */
> 
> ... all of this positively MIPS-only, i.e. not (unexpectedly) re-used by any
> other target?

As I'd like to see this go in, for me to then re-base my patch over it, can we
get this aspect sorted?

Jan