[PATCH] ld/aarch64: prune a PE/COFF test

[evgeny@kmaps.co]

Fri Aut 8 2025
Jan Beulich <jbeulich@suse.com> wrote:

> Expecting ___tls_end__ there is entirely random, when there are multiple
> symbols all at the same address. Help objdump to pick the intended
> symbol, by making it global and giving it a type.

> Disassembly of section .text:
>
> -0000000140001000 <___tls_end__>:
> +0000000140001000 <_start>:
>    140001000:	d2800281 	mov	x1, #0x14                  	// #20

> .section .text
>
> +    .global _start
> +    .def _start
> +    .scl 2 /* C_EXT */
> +    .type 32 /* DT_FUNC */
> +    .endef
>  _start:
>      mov x1, 20
>      b foo

Changing only __tls_end__ to _start should be enough to validate correctness.
The target was actively being upstreamed and had issues such as resolving
symbols that look fixed now.

Fri Aug 8 2025
Maciej W. Rozycki <macro@orcam.me.uk> wrote:

> Does the symbol reported here have any meaning for this test?  What's the
> purpose of the test?  The usual approach is just to match anything with
> disassembly:

Most likely, it was one of the first tests used to validate minimal compilation
for the new target aarch64-w64-mingw32.

Regards,
Evgeny