PATCH: PR ld/16428: autoconf tests don't terminate on ix86-linux-gnu with -static -fPIE -pie on glibc-2.18 based systems

Richard Sandiford rdsandiford@googlemail.com
Wed Jan 15 18:36:00 GMT 2014


Alan Modra <amodra@gmail.com> writes:
> On Wed, Jan 15, 2014 at 07:49:57PM +1030, Alan Modra wrote:
>> I really don't want to hack all the backends.  So, lets see if this
>> generic hack passes testing..
>
> The test results are in, with failures on frv-linux, crisv32-linux,
> i370-linux and all mips-linux targets.  The first two are caused by
> complaints about relocation types in shared libs, so I'll just xfail
> the test.  i370-linux has so many failures that I reckon the port
> should be declared dead.  The mips failures
>
> mips-linux  +FAIL: MIPS magic __ehdr_start symbol test 1 (n32)
> mips-linux  +FAIL: MIPS magic __ehdr_start symbol test 1 (n64)
>
> are due to this code in mips_elf_record_got_page_entry
>
>       entry = bfd_zalloc (sec->owner, sizeof (*entry));
>
> When a symbol is absolute, sec == bfd_abs_section_ptr and
> sec->owner == NULL.  Methinks I've found a bug in the mips backend,
> rather than there being anything wrong with my __ehdr_start patch.
> Not being a mips expert, I'm asking for an OK from a mips maintainer
> before committing this part of the patch.

Oops, serves me right for taking a shortcut. :-/

The MIPS part looks good to me, thanks.

Richard



More information about the Binutils mailing list