Remove some unused references to the pnacl-nativeld tool

Remove some unused references to the pnacl-nativeld tool

pnacl-nativeld was only used directly by service_runtime/nacl.scons, and
this use has been disabled for a while.

This will help us to remove pnacl-nativeld as a public entry point,
which will aid further cleanups.

BUG=https://code.google.com/p/nativeclient/issues/detail?id=3873
TEST=trybots
R=jvoung@chromium.org

Committed: https://src.chromium.org/viewvc/native_client?view=rev&revision=13340

[jvoung (off chromium), 2014-06-12 22:45:59]

LGTM

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
File src/trusted/service_runtime/nacl.scons (right):

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
src/trusted/service_runtime/nacl.scons:22: # TODO(dschuff): either re-enable
this test after we have fully transitioned
Should this not be a TODO() then?

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
src/trusted/service_runtime/nacl.scons:45: nacl_text_pad_asm =
'arch/mips/nacl_text_pad_test.S'
mips might end up being untested then (if the only untrusted toolchain is for
bitcode=1)

[Mark Seaborn, 2014-06-13 00:06:45]

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
File src/trusted/service_runtime/nacl.scons (right):

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
src/trusted/service_runtime/nacl.scons:22: # TODO(dschuff): either re-enable
this test after we have fully transitioned
On 2014/06/12 22:45:58, jvoung wrote:
> Should this not be a TODO() then?

I left it as a TODO because one of the options was "remove all pnacl-specific
stuff below and in the asm files", and I didn't do that fully.

I can do that if you insist (and go through the asm files to clean them up), but
it might be better to spend the time to rewrite the tests to be independent of a
linker.  But this is getting into yak-shave territory. :-)

(I initially just wanted to make pnacl-nativeld private so that I can refactor
it to make it easier to change.)

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
src/trusted/service_runtime/nacl.scons:45: nacl_text_pad_asm =
'arch/mips/nacl_text_pad_test.S'
On 2014/06/12 22:45:58, jvoung wrote:
> mips might end up being untested then (if the only untrusted toolchain is for
> bitcode=1)

Yes, arch/mips/nacl_text_pad_test.S is definitely unused at the moment, without
this change (but then the MIPS folks haven't enabled tests on the bots yet).

[jvoung (off chromium), 2014-06-13 01:01:52]

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
File src/trusted/service_runtime/nacl.scons (right):

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
src/trusted/service_runtime/nacl.scons:22: # TODO(dschuff): either re-enable
this test after we have fully transitioned
On 2014/06/13 00:06:44, Mark Seaborn wrote:
> On 2014/06/12 22:45:58, jvoung wrote:
> > Should this not be a TODO() then?
> 
> I left it as a TODO because one of the options was "remove all pnacl-specific
> stuff below and in the asm files", and I didn't do that fully.
> 
> I can do that if you insist (and go through the asm files to clean them up),
but
> it might be better to spend the time to rewrite the tests to be independent of
a
> linker.  But this is getting into yak-shave territory. :-)
> 
> (I initially just wanted to make pnacl-nativeld private so that I can refactor
> it to make it easier to change.)

Okay I see -- this is fine.

Usually, the linker will leave plenty of space for the IRT so these tests are
not a high priority thing anymore.  The only case that could be worrying is
generating the pnacl-llc.nexe/ld.nexe which don't use the IRT. Otherwise, the
service runtime's behavior is covered by gcc.

https://codereview.chromium.org/329793004/diff/40001/src/trusted/service_runt...
src/trusted/service_runtime/nacl.scons:45: nacl_text_pad_asm =
'arch/mips/nacl_text_pad_test.S'
On 2014/06/13 00:06:44, Mark Seaborn wrote:
> On 2014/06/12 22:45:58, jvoung wrote:
> > mips might end up being untested then (if the only untrusted toolchain is
for
> > bitcode=1)
> 
> Yes, arch/mips/nacl_text_pad_test.S is definitely unused at the moment,
without
> this change (but then the MIPS folks haven't enabled tests on the bots yet).

Okay, I guess there is already the early Return() from the if
env.Bit('bitcode'), so this was already the case.

[Mark Seaborn, 2014-06-13 16:43:05]

Committed patchset #3 manually as r13340 (presubmit successful).