[Bug gold/22221] add --no-dynamic-linker option

[markus at trippelsdorf dot de]

https://sourceware.org/bugzilla/show_bug.cgi?id=22221

--- Comment #4 from Markus Trippelsdorf <markus at trippelsdorf dot de> ---
(In reply to Cary Coutant from comment #3)
> I'm not sure I understand exactly what this option is for. If "-static -pie"
> will produce a static-pie executable, wouldn't such an executable already
> lack a dynamic linker? That's (part of) what -static does.

The problem is that when both "-static -pie" are used -pie currently forces
PT_INTERP to be set.

Generally "-static -pie" could be thought of as a hardened -static.

 % cat aslr.c
#include <stdio.h>

static void foo() {}
static int bar = 5;

int main() {
  int baz = 5;
  printf("function: %p, library function: %p, data: %p, stack: %p\n", foo,
         &printf, &bar, &baz);
}

(gcc trunk supports -static-pie, glibc with PIE libc.a is needed)               
 ~ % gcc -fuse-ld=bfd -static-pie -fPIE aslr.c

 ~ % file a.out
a.out: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically
linked, for GNU/Linux 4.10.0, with debug_info, not stripped                     
 ~ % ldd ./a.out
        statically linked

 ~ % ./a.out
function: 0x7f1c3fd130e5, library function: 0x7f1c3fd21b00, data:
0x7f1c3ffaf070, stack: 0x7ffeb1d24c0c                                           
 ~ % ./a.out
function: 0x7f137df070e5, library function: 0x7f137df15b00, data:
0x7f137e1a3070, stack: 0x7ffeeeb59eac

> What does the option have to do with the PT_PHDR header and segment? From
> the name & description, it sounds like it's intended to suppress the
> PT_INTERP header.

ld.bfd outputs no PT_PHDR header with -static-pie, so I just tried to copy its
behavior.

> If the option is also supposed to suppress the PT_PHDR header, shouldn't we
> issue a warning or an error if this option is used and we see a PHDRS clause
> in the linker script?

Probably yes. But first we have to get the basic functionality working.

(Also the labels for the output from comment 2 are wrong.
They should be the other way round. ld.bfd output is ld.gold output and vice
versa. Sorry.)

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