Generate Breakpad symbols for Remoting Host plugin on Mac.

Generate Breakpad symbols for Remoting Host plugin on Mac.

See also http://crrev.com/95103

BUG=87632
TEST=Official build logs show Breakpad symbols being uploaded to crash server.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=95498

[Lambros, 2011-08-03 02:49:31]

http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_pro...
File chrome/tools/build/mac/dump_product_syms (left):

http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_pro...
chrome/tools/build/mac/dump_product_syms:64: SRC_STEM=$(echo "${SRC_NAME}" | sed
-Ee 's/^(.*)\.(app|framework)$/\1/')
This may need correction, depending on where the Mac build places the dSYMs.

[Mark Mentovai, 2011-08-03 04:04:18]

Additional comment:

remoting_host_plugin’s target will need to be declared as mac_real_dsym
conditionally based on the value of mac_breakpad. It doesn’t appear that this
has been done. If you go ask codesearch.chromium.org for examples of
mac_real_dsym, you’ll find several examples.

Note to self: target is remoting/remoting.gyp:remoting_host_plugin.

http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_pro...
File chrome/tools/build/mac/dump_product_syms (left):

http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_pro...
chrome/tools/build/mac/dump_product_syms:64: SRC_STEM=$(echo "${SRC_NAME}" | sed
-Ee 's/^(.*)\.(app|framework)$/\1/')
Lambros wrote:
> This may need correction, depending on where the Mac build places the dSYMs.

This comment makes it sound like this patch is untested. Bad!

Are you asking to have this tested or are you saying you just haven’t had a
chance yet?

http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_pro...
File chrome/tools/build/mac/dump_product_syms (right):

http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_pro...
chrome/tools/build/mac/dump_product_syms:59: "remoting_host_plugin.plugin" ; do
Is this present in all builds? I did an official-flavored release build from the
trunk earlier today and don’t see this in my output directory.

[Lambros, 2011-08-03 17:10:01]

On Tue, Aug 2, 2011 at 9:04 PM, <mark@chromium.org> wrote:

> Lambros wrote:
>
>> This may need correction, depending on where the Mac build places the
>>
> dSYMs.
>
> This comment makes it sound like this patch is untested. Bad!
>
> Are you asking to have this tested or are you saying you just haven’t
> had a chance yet?

Sorry for uploading an untested patch, I wanted to get my CL uploaded so
people could have a look at it while I figure out how to test it.  My own
build wasn't producing dSYMs, presumably the "mac_real_dsyms" will fix this?


>
> http://codereview.chromium.**org/7562002/diff/1/chrome/**
>
tools/build/mac/dump_product_**syms<http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_product_syms>
> File chrome/tools/build/mac/dump_**product_syms (right):
>
> http://codereview.chromium.**org/7562002/diff/1/chrome/**
>
tools/build/mac/dump_product_**syms#newcode59<http://codereview.chromium.org/7562002/diff/1/chrome/tools/build/mac/dump_product_syms#newcode59>
> chrome/tools/build/mac/dump_**product_syms:59:
>
> "remoting_host_plugin.plugin" ; do
> Is this present in all builds? I did an official-flavored release build
> from the trunk earlier today and don’t see this in my output directory.

When I'm testing, I normally build the "remoting/remoting.gyp:webapp_it2me"
target, which builds the plugin as well.  The development snapshot builds
are somehow building this target on Mac, Win, Linux.  On Linux, the default
Make target seems to build it, if I'm interpreting the bot logs correctly.


>
>
>
http://codereview.chromium.**org/7562002/<http://codereview.chromium.org/7562...
>

[Mark Mentovai, 2011-08-03 17:13:19]

You need to do a release build. For the proper end-to-end testing, it
needs to be an official release build. This is one in which the
following GYP variables are set:

    'branding': 'Chrome',
    'buildtype': 'Official',

and obviously it must be built in release mode. Note that buildtype:
Official requires that you have a src-internal checkout.

This gets you .dSYMs, makes them real .dSYMs for things that need
symbols archived and sent to the crash servers, and builds the
Breakpad symbol files and .dSYM archives in the script you’re
modifying.

With respect to the remoting_host_plugin target, the reason I bring it
up is that the script that produced Breakpad symbol files and archives
.dSYMs is hung off of the chrome target, so the chrome target needs to
depend somehow on remoting_host_plugin in order to ensure that the
plugin and its .dSYM is actually built by the time the script runs—or
that it’s even built at all.

[Lambros, 2011-08-04 02:20:05]

PTAL, I've addressed the comments raised.

http://codereview.chromium.org/7562002/diff/5001/chrome/tools/build/mac/dump_...
File chrome/tools/build/mac/dump_product_syms (right):

http://codereview.chromium.org/7562002/diff/5001/chrome/tools/build/mac/dump_...
chrome/tools/build/mac/dump_product_syms:65: SRC_STEM=$(echo "${SRC_NAME}" | sed
-Ee 's/\.(app|framework|plugin)$//')
Simplified sed so it still fits on same line :)

[Lambros, 2011-08-04 16:59:57]

I've confirmed the dSYMs are real with the remoting.gyp changes (and without the
change, the fake_dsym key is present in Info.plist).  The real dSYMs are much
bigger as well.

[Mark Mentovai, 2011-08-04 17:03:26]

LGTM!

http://codereview.chromium.org/7562002/diff/5001/chrome/chrome_exe.gypi
File chrome/chrome_exe.gypi (right):

http://codereview.chromium.org/7562002/diff/5001/chrome/chrome_exe.gypi#newco...
chrome/chrome_exe.gypi:238: # In order to process symbols for the Remoting Host
plugin,
Blank line before.