Re: Commit: PR 22875: Stop strip corrupting unknown relocs

[Alan Modra <amodra at gmail dot com>]

On Sat, Mar 03, 2018 at 05:31:55PM +0000, Maciej W. Rozycki wrote:
> Hi Nick,
> 
> >   The patch is quite large, because it touches almost all of the ELF
> >   files in the BFD library, but it is actually quite simple in
> >   execution.  I have tested it with a large number of different targets,
> >   and added a small testcase to the binutils testsuite as well.
> 
>  Nevertheless it has caused failures with the new test case and the 
> `mips64-openbsd' and `mips64el-openbsd' targets:
> 
> Executing on host: .../binutils/strip-new  -g tmpdir/bintest.o -o tmpdir/copy.o   (timeout = 300)
> .../binutils/strip-new: tmpdir/bintest.o: unsupported relocation type 0xf1
> .../binutils/strip-new: tmpdir/bintest.o: unsupported relocation type 0xf1
> 
> failed with: <.../binutils/strip-new: tmpdir/bintest.o: unsupported relocation type 0xf1>, expected: <.* bad value>
> .../binutils/strip-new: tmpdir/bintest.o: unsupported relocation type 0xf1
> FAIL: binutils-all/strip-13
> 
> which both default to the n64 rather than o32 or n32 MIPS ABI.
> 
>  Offhand the error message appears correct to me, and arguably more useful 
> than ambiguous and rather cryptic "bad value".

You're supposed to get the "unsupported relocation" error as well as
the "bad value" error.  The reason why that isn't the case for mips64
is that the error status isn't propagated up the call chain.
Something like the following ought to fix that.

diff --git a/bfd/elf64-mips.c b/bfd/elf64-mips.c
index ea2e921..9501613 100644
--- a/bfd/elf64-mips.c
+++ b/bfd/elf64-mips.c
@@ -3611,7 +3611,7 @@ mips_elf64_rtype_to_howto (bfd *abfd, unsigned int r_type, bfd_boolean rela_p)
 	  _bfd_error_handler (_("%pB: unsupported relocation type %#x"),
 			      abfd, r_type);
 	  bfd_set_error (bfd_error_bad_value);
-	  r_type = R_MIPS_NONE;
+	  return NULL;
 	}
       if (rela_p)
 	return &mips_elf64_howto_table_rela[r_type];
@@ -3800,6 +3800,8 @@ mips_elf64_slurp_one_reloc_table (bfd *abfd, asection *asect,
 	  relent->addend = rela.r_addend;
 
 	  relent->howto = mips_elf64_rtype_to_howto (abfd, type, rela_p);
+	  if (relent->howto == NULL)
+	    goto error_return;
 
 	  ++relent;
 	}


>  However I find it strange 
> that the MIPS n64 ABI backend is the only one to give this different error 
> message, despite that backends for the three MIPS ABIs are all structured 
> similarly.

An even more strange fact is that mips64el-openbsd creates really odd
relocs for the strip-13 testcase.  There seems to be a word swap..

Relocation section '.rela.text' at offset 0x90 contains 2 entries:
    Offset             Info             Type               Symbol's Value  Symbol's Name + Addend
0000000000000000  000000f100000000 R_MIPS_NONE           readelf: Error:  bad symbol index: 000000f1 in reloc
                    Type2: R_MIPS_NONE      
                    Type3: R_MIPS_NONE      
0000000000000000  0000000000000000 R_MIPS_NONE                               0
                    Type2: R_MIPS_NONE      
                    Type3: R_MIPS_NONE      

Oh, and the bad symbol index gets you a segfault on trying to run
strip -g, at this line
      if ((*ptr->sym_ptr_ptr)->the_bfd->xvec != abfd->xvec
in elf64-mip.c:mips_elf64_write_rela.  the_bfd is NULL.

-- 
Alan Modra
Australia Development Lab, IBM