[PATCH v2] coff: Ignore corrupt relocation table in linker input

[H.J. Lu]

On Mon, Sep 22, 2025 at 6:46 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Mon, Sep 22, 2025 at 6:14 AM Alan Modra <amodra@gmail.com> wrote:
> >
> > On Sun, Sep 21, 2025 at 07:52:01PM +0200, Jan Beulich wrote:
> > > On 21.09.2025 13:45, H.J. Lu wrote:
> > > > On Sun, Sep 21, 2025 at 6:12 PM Alan Modra <amodra@gmail.com> wrote:
> > > >> On Sun, Sep 21, 2025 at 03:47:11PM +0800, H.J. Lu wrote:
> > > >>> Any comments or objections?
> > > >>
> > > >> All three of your patches for these fuzzed object file linker bugs
> > > >> reduce the capability of the binutils to display corrupted object
> > > >
> > > > Such information displayed by binutils is unreliable and corrupted
> > > > objects may lead to corrupted executable.   I think benefits of
> > > > such corrupted information are very limited if any.
> > >
> > > I agree with Alan. Corruption or anomalies may be diagnosed (by
> > > warnings, in particular when linking), but we still should aim at
> > > displaying to users whatever we can sensibly display.
> >
> > Yes, it would be quite OK to refuse to link objects that are
> > corrupted, but we shouldn't refuse to objdump them.
> >
>
> I will see what I can do.
>

Here is the patch to ignore slurp corrupt relocation table
in linker input to avoid linker crash later.

OK for master?

Thanks.

-- 
H.J.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: v2-0001-coff-Ignore-corrupt-relocation-table-in-linker-in.patch
Type: text/x-patch
Size: 1093 bytes
Desc: not available
URL: <https://sourceware.org/pipermail/binutils/attachments/20250922/6cbd3e22/attachment.bin>