[PATCH v3 8/8] riscv: check relaxation delete range, propagate failures

Harshit Kumar harshit7kr@gmail.com
Fri Nov 20 10:50:13 GMT 2026


On Fri, Nov 20, 2026 at 06:01:56PM +0800, Kito Cheng wrote:
> Thanks for the patch, the fix generally seems good to me, but I am not
> comfortable about the testcase...could you try to convert that to .s
> file instead of hard coded hex?

Thanks, Kito.  Yes: gas builds both objects from your .s (ALIGN at
offset 8 and at 22), so the next version will use .s/.d tests instead
of the hex.

Harshit

On Fri, 20 Nov 2026 at 15:32, Kito Cheng <kito.cheng@gmail.com> wrote:
>
> Hi Harshit:
>
> Thanks for the patch, the fix generally seems good to me, but I am not
> comfortable about the testcase...could you try to convert that to .s
> file instead of hard coded hex?
>
> I guess it should be something like that:
>
>         .text
>         .globl _start
> _start:
>         .rept 6
>         nop
>         .endr
>         .reloc _start + OFF, R_RISCV_ALIGN, 6
>
>
> Thanks.


More information about the Binutils mailing list