[PATCH] readelf: Display and return true for empty note section/segment

H.J. Lu hjl.tools@gmail.com
Wed Jan 14 09:32:11 GMT 2026


On Wed, Jan 14, 2026 at 4:40 PM Jan Beulich <jbeulich@suse.com> wrote:
>
> On 14.01.2026 02:02, H.J. Lu wrote:
> > Change readelf to display and return true for empty note section and
> > segment.
>
> I don't mind this part, but I also don't view it as strictly necessary. The
> patch suggested in [1] would imo do.
>
> >  Add a linker test to verify that empty note section is discarded.
>
> This part, otoh, I disagree with: Rather than adding a test to verify this
> behavior, shouldn't it be corrected? Apart from you, everyone else now
> looks to converge to empty note sections being legitimate. If the linker

My patch doesn't change the linker nor the linker needs change for empty
sections.

> discards them blindly (there may want / need to be exceptions, like for
> .note.GNU-stack), that may need changing then.
>
> > binutils/
> >
> > PR ld/33780
> > * readelf.c (process_notes_at): Display and return true for empty
> > note section and segment.
> >
> > gas/
> >
> > PR ld/33780
> > * testsuite/gas/elf/elf.exp: Run empty-note.
> > * testsuite/gas/elf/empty-note.d: New file.
> > * testsuite/gas/elf/empty-note.s: Likewise.
> >
> > ld/
> >
> > PR ld/33780
> > * testsuite/ld-elf/empty-note.d: New file.
>
> As to this file's use of #failif: I don't think it can be used like this,
> i.e. with multiple expectation lines different from "#...". Aiui you want
> the test to fail if either line shows up, yet the effect would be that it
> would fail only if both show up.
>
> Jan
>
> [1] https://sourceware.org/pipermail/binutils/2026-January/147115.html



-- 
H.J.


More information about the Binutils mailing list