Remove _Unwind_* stubs from pnacl_abi.bc

Remove _Unwind_* stubs from pnacl_abi.bc

For stable builds, they are provided by unwind_stubs.bc
Also, when doing the final bitcode link, gold accepts flags that
tell it to allow these symbols to remain undefined, which makes
the unstable-with-exceptions builds work.

BUG=https://code.google.com/p/nativeclient/issues/detail?id=3339
R=jvoung@chromium.org, mseaborn@chromium.org

Committed: https://src.chromium.org/viewvc/native_client?view=rev&revision=11315

[jvoung (off chromium), 2013-05-07 18:11:54]

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py
File pnacl/driver/pnacl-ld.py (right):

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py#newc...
pnacl/driver/pnacl-ld.py:105: '--undef-sym-check
--allow-unresolved=_Unwind_DeleteException ' +
keep one exception per line so that they are easier to see?

Should the list of exceptions only be there when ALLOW_CXX_EXCEPTIONS is true?

[Mark Seaborn, 2013-05-07 18:16:06]

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py
File pnacl/driver/pnacl-ld.py (right):

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py#newc...
pnacl/driver/pnacl-ld.py:105: '--undef-sym-check
--allow-unresolved=_Unwind_DeleteException ' +
On 2013/05/07 18:11:55, jvoung (cr) wrote:
> keep one exception per line so that they are easier to see?

Also, you can omit the "+": Python does C-style string literal concatenation.

> Should the list of exceptions only be there when ALLOW_CXX_EXCEPTIONS is true?

Right.  It would be better to get a bitcode-link-time error than an ABI verifier
error if you're using these symbols wrongly.

[eliben, 2013-05-07 18:55:04]

PTAL

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py
File pnacl/driver/pnacl-ld.py (right):

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py#newc...
pnacl/driver/pnacl-ld.py:105: '--undef-sym-check
--allow-unresolved=_Unwind_DeleteException ' +
On 2013/05/07 18:11:55, jvoung (cr) wrote:
> keep one exception per line so that they are easier to see?
> 
> Should the list of exceptions only be there when ALLOW_CXX_EXCEPTIONS is true?

Done.

https://codereview.chromium.org/14780006/diff/1/pnacl/driver/pnacl-ld.py#newc...
pnacl/driver/pnacl-ld.py:105: '--undef-sym-check
--allow-unresolved=_Unwind_DeleteException ' +
On 2013/05/07 18:16:06, Mark Seaborn wrote:
> On 2013/05/07 18:11:55, jvoung (cr) wrote:
> > keep one exception per line so that they are easier to see?
> 
> Also, you can omit the "+": Python does C-style string literal concatenation.
>

Done, although it was consistent with the surrounding script's style.
 
> > Should the list of exceptions only be there when ALLOW_CXX_EXCEPTIONS is
true?
> 
> Right.  It would be better to get a bitcode-link-time error than an ABI
verifier
> error if you're using these symbols wrongly.

Done

[Mark Seaborn, 2013-05-07 19:59:05]

https://codereview.chromium.org/14780006/diff/9001/pnacl/driver/pnacl-ld.py
File pnacl/driver/pnacl-ld.py (right):

https://codereview.chromium.org/14780006/diff/9001/pnacl/driver/pnacl-ld.py#n...
pnacl/driver/pnacl-ld.py:105: '{ALLOW_CXX_EXCEPTIONS ? '
Does there need to be a "$" here for this to work?

https://codereview.chromium.org/14780006/diff/9001/pnacl/driver/pnacl-ld.py#n...
pnacl/driver/pnacl-ld.py:106: '--allow-unresolved=_Unwind_DeleteException '
Maybe sort the list?

[eliben, 2013-05-07 21:03:14]

https://codereview.chromium.org/14780006/diff/9001/pnacl/driver/pnacl-ld.py
File pnacl/driver/pnacl-ld.py (right):

https://codereview.chromium.org/14780006/diff/9001/pnacl/driver/pnacl-ld.py#n...
pnacl/driver/pnacl-ld.py:105: '{ALLOW_CXX_EXCEPTIONS ? '
On 2013/05/07 19:59:05, Mark Seaborn wrote:
> Does there need to be a "$" here for this to work?

Good catch, thanks. I'm wondering why this hasn't come up in tests...

https://codereview.chromium.org/14780006/diff/9001/pnacl/driver/pnacl-ld.py#n...
pnacl/driver/pnacl-ld.py:106: '--allow-unresolved=_Unwind_DeleteException '
On 2013/05/07 19:59:05, Mark Seaborn wrote:
> Maybe sort the list?

Will do

[eliben, 2013-05-07 23:09:13]

PTAL

[jvoung (off chromium), 2013-05-07 23:12:09]

lgtm

[Mark Seaborn, 2013-05-08 02:33:18]

LGTM

[eliben, 2013-05-08 15:23:01]

Committed patchset #3 manually as r11315 (presubmit successful).