Re: [PATCH v2 1/1] aarch64: mingw: Auto-import implementation

On Fri, 14 Aug 2026, Jan Beulich wrote:
> >      {
> >      case 8:
> > @@ -1186,7 +1198,6 @@ read_addend (arelent *rel, asection *s)
> >  	    addend = bfd_get_16 (s->owner, buf);
> >  	}
> >        break;
> > -    case 26:
> 
> How come this can go away here?

For 26 bit relocation, an addend is not needed.
For 12 bit relocations, the same. It is visible in the example that these
instructions do not require relocation in this case.

For 21 bit relocation, it is calculated in pe_create_import_fixup.
It can be moved to read_addend if it is required.

> > +      if (rel->howto->bitsize == 12)
> > +	return;
> > +      else if (rel->howto->bitsize == 26)
> > +	{
> > +/* On AArch64, a single opcode is not sufficient for relocation
> > +   in dynamic linking. The linker generates a jump stub instead.  */
> 
> Missing indentation again.

The indentations will be fixed.

Regards,
Evgeny