Track undefined sym in the symtab. Remove hack for missing relocs against undef.

Track undefined sym in the symtab. Remove hack for missing relocs against undef.

Preliminary linking tests, seems to show that the linker
and objcopy are happy to use 'em on spec2k, and the
result runs! (Had to be careful to clobber the old .s
and .o files to make it's testing the right copy).

Haven't tried crosstests yet.

BUG=none
R=stichnot@chromium.org

Committed: https://gerrit.chromium.org/gerrit/gitweb?p=native_client/pnacl-subzero.git;a=commit;h=261cae38fd492fbe90b8431d11d2a36014101d71

[jvoung (off chromium), 2015-01-31 05:17:46]

jvoung@chromium.org changed reviewers:
+ stichnot@chromium.org

[Jim Stichnoth, 2015-01-31 20:58:42]

Cool!

See https://codereview.chromium.org/892803002/ for making elf testing more
automated.

I also see spec2k working with --elf.

The crosstests ran into a couple problems:
  - Align=0 should probably be bumped to Align=1
  - It doesn't like some reference to myFremff, but I didn't pursue further.

[Jim Stichnoth, 2015-02-01 03:23:56]

otherwise lgtm

https://codereview.chromium.org/889613004/diff/20001/src/IceELFObjectWriter.cpp
File src/IceELFObjectWriter.cpp (right):

https://codereview.chromium.org/889613004/diff/20001/src/IceELFObjectWriter.c...
src/IceELFObjectWriter.cpp:539: IceString Name = Sym->getName();
The fact that this is IceString instead of IceString& makes me think it's too
bad all the getName() methods don't return references.  Most of the getName()
calls are only used for dumping or textual emission, but this use is probably
unavoidable.

Nothing to do about it yet, I guess.

https://codereview.chromium.org/889613004/diff/20001/src/IceELFObjectWriter.c...
src/IceELFObjectWriter.cpp:540: assert(Sym->getOffset() == 0 &&
Sym->getSuppressMangling());
I realize there are "violations" elsewhere, but I think it's better to use two
asserts here since you get more precise information on a failure.

[jvoung (off chromium), 2015-02-01 17:07:56]

https://codereview.chromium.org/889613004/diff/20001/src/IceELFObjectWriter.cpp
File src/IceELFObjectWriter.cpp (right):

https://codereview.chromium.org/889613004/diff/20001/src/IceELFObjectWriter.c...
src/IceELFObjectWriter.cpp:539: IceString Name = Sym->getName();
On 2015/02/01 03:23:55, stichnot wrote:
> The fact that this is IceString instead of IceString& makes me think it's too
> bad all the getName() methods don't return references.  Most of the getName()
> calls are only used for dumping or textual emission, but this use is probably
> unavoidable.
> 
> Nothing to do about it yet, I guess.

Hmm I can check if they can be converted to return IceString& in a separate CL.

https://codereview.chromium.org/889613004/diff/20001/src/IceELFObjectWriter.c...
src/IceELFObjectWriter.cpp:540: assert(Sym->getOffset() == 0 &&
Sym->getSuppressMangling());
On 2015/02/01 03:23:55, stichnot wrote:
> I realize there are "violations" elsewhere, but I think it's better to use two
> asserts here since you get more precise information on a failure.

Oops done.

[jvoung (off chromium), 2015-02-01 17:18:29]

On 2015/01/31 20:58:42, stichnot wrote:
> Cool!
> 
> See https://codereview.chromium.org/892803002/ for making elf testing more
> automated.
> 
> I also see spec2k working with --elf.
> 
> The crosstests ran into a couple problems:
>   - Align=0 should probably be bumped to Align=1

This is for globals? I guess there isn't much type info left, either a data
array or a ptrtoint. For ptrtoint (or complex inits w/ a ptrtoint), it could be
4 otherwise 1?

>   - It doesn't like some reference to myFremff, but I didn't pursue further.

My guess is that  IceConverter and PNaClTranslator need to handle getConstantSym
vs getConstantExternSym if the function declaration is only a proto / not
defined. Will try that out.

[jvoung (off chromium), 2015-02-01 18:31:11]

Committed patchset #3 (id:40001) manually as
261cae38fd492fbe90b8431d11d2a36014101d71 (presubmit successful).