Run real dsymutil to get a real .dSYM for Breakpad dump_syms

Run real dsymutil to get a real .dSYM for Breakpad dump_syms.  This will
allow dump_syms to access DWARF data for line numbers and other stuff, and not
just public symbol table data.

BUG=12776
TEST=set branding to Chrome, dsymutil should run and take forever

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=17316

[Mark Mentovai, 2009-05-29 14:51:28]

NOT FOR CHECKIN.  At least not yet.  dump_syms is outputting like ten lines and
then dying.  That's not really fantastic for symbol data.

[TVL, 2009-05-29 15:02:44]

http://codereview.chromium.org/113999/diff/1/2
File build/common.gypi (right):

http://codereview.chromium.org/113999/diff/1/2#newcode391
Line 391: 'DEPLOYMENT_POSTPROCESSING': 'YES',
DEAD_CODE_STRIPPING = YES?
STRIP_STYLE = ?
Or do you trust the xcode defaults for these?

http://codereview.chromium.org/113999/diff/1/2#newcode402
Line 402: # Define strip_from_xcode in a variable ending in _path so
alignment/rewrap

[TVL, 2009-05-29 15:04:29]

http://codereview.chromium.org/113999/diff/1/2
File build/common.gypi (right):

http://codereview.chromium.org/113999/diff/1/2#newcode349
Line 349: 'mac_real_dsym': 0,  # Fake .dSYMs are fine in most cases.
actually, should this be mac_real_dysm% so it can be overridden?

[Mark Mentovai, 2009-05-29 15:11:47]

New version uploaded.

http://codereview.chromium.org/113999/diff/1/2
File build/common.gypi (right):

http://codereview.chromium.org/113999/diff/1/2#newcode349
Line 349: 'mac_real_dsym': 0,  # Fake .dSYMs are fine in most cases.
On 2009/05/29 15:04:33, TVL wrote:
> actually, should this be mac_real_dysm% so it can be overridden?

It seems like there's a bug in % - when I initially had it at %, all targets
processed after app had mac_real_dsym set to 1, which was not intended.  I'll
look into that bug and then we can move this to a %.  I added a comment here. 
In the mean time, mac_real_dsym isn't really something that we need to override
on a global basis, so I'm OK with this.

http://codereview.chromium.org/113999/diff/1/2#newcode391
Line 391: 'DEPLOYMENT_POSTPROCESSING': 'YES',
TVL wrote:
> DEAD_CODE_STRIPPING = YES?

DEAD_CODE_STRIPPING is entirely unrelated to stripping.  Also, we set it
elsewhere.

> STRIP_STYLE = ?
> Or do you trust the xcode defaults for these?

I trust the default for STRIP_STYLE ("all" for executables).

http://codereview.chromium.org/113999/diff/1/2#newcode402
Line 402: # Define strip_from_xcode in a variable ending in _path so
TVL wrote:
> alignment/rewrap

Fixed.

[TVL, 2009-05-29 15:20:04]

lgtm

[Mark Mentovai, 2009-05-29 19:10:28]

The breakpad dump_syms problem has been fixed and is rolled into this change
now.  Have another look?

[TVL, 2009-05-29 19:17:49]

lgtm