[Bug ld/29592] [2.39 Regression][bisected] Exec stack warning causes go programs with cgo to fail to link on mips

[matoro_bugzilla_glibc at matoro dot tk]

https://sourceware.org/bugzilla/show_bug.cgi?id=29592

--- Comment #2 from matoro <matoro_bugzilla_glibc at matoro dot tk> ---
(In reply to Nick Clifton from comment #1)
> (In reply to matoro from comment #0)
> Hi Matoro,
> 
> > On mips, since this commit, all linking prints the following warning:
> > 
> > /usr/lib/gcc/mips64-unknown-linux-gnu/12.2.0/../../../../mips64-unknown-
> > linux-gnu/bin/ld: warning:
> > /usr/lib/gcc/mips64-unknown-linux-gnu/12.2.0/../../../../lib64/Scrt1.o:
> > requires executable stack (because the .note.GNU-stack section is 
> > executable)
> 
> This was supposed to have been fixed by commit
> ed1c7ad8e31a2cfca1d8c8c898ee7cde6ae340ca  
> Do you have that commit in your sources ?
> 
> 
>  
> > /var/tmp/portage/dev-lang/go-1.19.1/temp/go-link-3757955373/go.o: in
> > function `internal/cpu.Initialize':
> > /usr/lib/go/src/internal/cpu/cpu.go:123:(.text+0x14): relocation R_MIPS_26
> > against `a local symbol' cannot be used when making a shared object;
> > recompile with -fPIC
> 
> Are you sure that it was that commit which triggered this problem ?
> 
> The commit only added code to generate the new warning messages, it did not
> - or at least should not - affect the linking process in any other way.
> 
> Also - is the warning correct ?  Is the -fPIC option missing from the
> compilation command line ?  Or is the linker mistaken when it complains
> about attempting to create a shared object ?  (IE is it supposed to be
> creating a static executable rather than a dynamic executable ?)
> 
> Cheers
>   Nick

Hi Nick, thank you for taking a look at this.  I checked and found that the
distro (Gentoo) is actually explicitly passing --enable-warn-execstack, so any
change to the defaults doesn't help: 
https://github.com/gentoo/gentoo/blob/master/sys-devel/binutils/binutils-9999.ebuild#L251

I tested  at that commit and when --enable-warn-execstack is passed, the issue
is still present.  What I don't understand is why toggling this warning on or
off changes the behavior of the linker - shouldn't it just be a purely
informational warning?

-- 
You are receiving this mail because:
You are on the CC list for the bug.