Update LTO flags for Mac.

Update LTO flags for Mac.

Specifically:

 - Use -all_load to work around http://openradar.appspot.com/20356002
 - Disable the verify_order postbuild step when using LTO, as this
   step currently fails.

Also update the documentation for the use_lto and use_lto_o2 build flags.

Requires https://codereview.chromium.org/1003273007

Requires https://codereview.chromium.org/1047853005 to build Chrome. We can
build simple targets even without this change.

BUG=471146
R=thakis@chromium.org

Committed: https://crrev.com/059ec677a02907ceac79dd68741b6ed820cd7ff0
Cr-Commit-Position: refs/heads/master@{#323134}

[Nico, 2015-03-31 21:45:04]

lgtm

I wonder if this'll Just Work for chrome/ios too.

[pcc1, 2015-03-31 21:52:15]

The CQ bit was checked by pcc@chromium.org

[commit-bot: I haz the power, 2015-03-31 21:53:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1045113004/1

[commit-bot: I haz the power, 2015-03-31 22:49:14]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-03-31 22:50:03]

Patchset 1 (id:??) landed as
https://crrev.com/059ec677a02907ceac79dd68741b6ed820cd7ff0
Cr-Commit-Position: refs/heads/master@{#323134}

[pcc1, 2015-03-31 22:52:33]

On 2015/03/31 21:45:04, Nico wrote:
> lgtm
> 
> I wonder if this'll Just Work for chrome/ios too.

Doesn't look like it. For Release-iphonesimulator I get:

ld: reference to symbol (which has not been assigned an address) _poll in
'__ZN4mojo8embedder12_GLOBAL__N_112WaitReadableENS0_14PlatformHandleE' from
obj/third_party/mojo/src/mojo/edk/embedder/mojo_system_unittests.platform_channel_pair_posix_unittest.o
for architecture x86_64

and for Release-iphoneos I get:

'corei7-avx' is not a recognized processor for this target (ignoring processor)
'corei7-avx' is not a recognized processor for this target (ignoring processor)
'corei7-avx' is not a recognized processor for this target (ignoring processor)
'corei7-avx' is not a recognized processor for this target (ignoring processor)
'corei7-avx' is not a recognized processor for this target (ignoring processor)
'corei7-avx' is not a recognized processor for this target (ignoring processor)
Undefined symbols for architecture arm64:
  "___fixsfsi", referenced from:
      __uhash_allocate in libicuuc.a(icuuc.uhash.o)
  "___ledf2", referenced from:
      -[NSString(CrStringDrawing) cr_stringByElidingToFitSize:] in
libgtest.a(gtest.coverage_util_ios.o)
  "___fixunsdfsi", referenced from:
      base::PlatformThread::SetThreadPriority(base::PlatformThreadHandle,
base::ThreadPriority) in libbase.a(base.platform_thread_mac.o)
[ and several more undefined symbol errors ]

The corei7-avx errors seem weird as that flag is only supposed to be used for
x86 targets. If I remove it I later get the _poll error like with the simulator.
Might be another case like https://codereview.chromium.org/1036123003 but I
couldn't immediately find anything that is overriding the poll function.

[pcc1, 2015-03-31 23:13:02]

On 2015/03/31 22:52:33, pcc1 wrote:
> On 2015/03/31 21:45:04, Nico wrote:
> > lgtm
> > 
> > I wonder if this'll Just Work for chrome/ios too.
> 
> Doesn't look like it. For Release-iphonesimulator I get:
> 
> ld: reference to symbol (which has not been assigned an address) _poll in
> '__ZN4mojo8embedder12_GLOBAL__N_112WaitReadableENS0_14PlatformHandleE' from
>
obj/third_party/mojo/src/mojo/edk/embedder/mojo_system_unittests.platform_channel_pair_posix_unittest.o
> for architecture x86_64
> 
> and for Release-iphoneos I get:
> 
> 'corei7-avx' is not a recognized processor for this target (ignoring
processor)
> 'corei7-avx' is not a recognized processor for this target (ignoring
processor)
> 'corei7-avx' is not a recognized processor for this target (ignoring
processor)
> 'corei7-avx' is not a recognized processor for this target (ignoring
processor)
> 'corei7-avx' is not a recognized processor for this target (ignoring
processor)
> 'corei7-avx' is not a recognized processor for this target (ignoring
processor)
> Undefined symbols for architecture arm64:
>   "___fixsfsi", referenced from:
>       __uhash_allocate in libicuuc.a(icuuc.uhash.o)
>   "___ledf2", referenced from:
>       -[NSString(CrStringDrawing) cr_stringByElidingToFitSize:] in
> libgtest.a(gtest.coverage_util_ios.o)
>   "___fixunsdfsi", referenced from:
>       base::PlatformThread::SetThreadPriority(base::PlatformThreadHandle,
> base::ThreadPriority) in libbase.a(base.platform_thread_mac.o)
> [ and several more undefined symbol errors ]
> 
> The corei7-avx errors seem weird as that flag is only supposed to be used for
> x86 targets. If I remove it I later get the _poll error like with the
simulator.
> Might be another case like https://codereview.chromium.org/1036123003 but I
> couldn't immediately find anything that is overriding the poll function.

Looks like this was third_party/nss/nspr/pr/src/md/unix/unix.c . Commenting out
that definition gets me a little further, now I get:

duplicate symbol
__ZN3web25BuildUserAgentFromProductERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEE
in:
    libios_web.a(ios_web.user_agent.o)
    libuser_agent.a(user_agent.user_agent.o)
duplicate symbol
__ZN7OSCrypt13EncryptStringERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEPS6_
in:
    libos_crypt.a(os_crypt.os_crypt_mac.o)
    libos_crypt.a(os_crypt.os_crypt_posix.o)
duplicate symbol
__ZN7OSCrypt13DecryptStringERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEPS6_
in:
    libos_crypt.a(os_crypt.os_crypt_mac.o)
    libos_crypt.a(os_crypt.os_crypt_posix.o)
duplicate symbol
__ZN7OSCrypt15DecryptString16ERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEEPNS1_ItN4base20string16_char_traitsENS4_ItEEEE
in:
    libos_crypt.a(os_crypt.os_crypt_mac.o)
    libos_crypt.a(os_crypt.os_crypt_posix.o)
duplicate symbol
__ZN7OSCrypt15EncryptString16ERKNSt3__112basic_stringItN4base20string16_char_traitsENS0_9allocatorItEEEEPNS1_IcNS0_11char_traitsIcEENS4_IcEEEE
in:
    libos_crypt.a(os_crypt.os_crypt_mac.o)
    libos_crypt.a(os_crypt.os_crypt_posix.o)
duplicate symbol
__ZN3web25BuildUserAgentFromProductERKNSt3__112basic_stringIcNS0_11char_traitsIcEENS0_9allocatorIcEEEE
in:
    libios_web.a(ios_web.user_agent.o)
    libuser_agent.a(user_agent.user_agent.o)

Giving up now, but these should be easy to fix.

[Nico, 2015-03-31 23:14:32]

maybe file a tracking bug with the current state, so that if someone wants
to do this they have a continuation they can call

On Tue, Mar 31, 2015 at 4:13 PM, <pcc@chromium.org> wrote:

> On 2015/03/31 22:52:33, pcc1 wrote:
>
>> On 2015/03/31 21:45:04, Nico wrote:
>> > lgtm
>> >
>> > I wonder if this'll Just Work for chrome/ios too.
>>
>
>  Doesn't look like it. For Release-iphonesimulator I get:
>>
>
>  ld: reference to symbol (which has not been assigned an address) _poll in
>> '__ZN4mojo8embedder12_GLOBAL__N_112WaitReadableENS0_14PlatformHandleE'
>> from
>>
>
> obj/third_party/mojo/src/mojo/edk/embedder/mojo_system_
> unittests.platform_channel_pair_posix_unittest.o
>
>> for architecture x86_64
>>
>
>  and for Release-iphoneos I get:
>>
>
>  'corei7-avx' is not a recognized processor for this target (ignoring
>>
> processor)
>
>> 'corei7-avx' is not a recognized processor for this target (ignoring
>>
> processor)
>
>> 'corei7-avx' is not a recognized processor for this target (ignoring
>>
> processor)
>
>> 'corei7-avx' is not a recognized processor for this target (ignoring
>>
> processor)
>
>> 'corei7-avx' is not a recognized processor for this target (ignoring
>>
> processor)
>
>> 'corei7-avx' is not a recognized processor for this target (ignoring
>>
> processor)
>
>> Undefined symbols for architecture arm64:
>>    "___fixsfsi", referenced from:
>>        __uhash_allocate in libicuuc.a(icuuc.uhash.o)
>>    "___ledf2", referenced from:
>>        -[NSString(CrStringDrawing) cr_stringByElidingToFitSize:] in
>> libgtest.a(gtest.coverage_util_ios.o)
>>    "___fixunsdfsi", referenced from:
>>        base::PlatformThread::SetThreadPriority(base::
>> PlatformThreadHandle,
>> base::ThreadPriority) in libbase.a(base.platform_thread_mac.o)
>> [ and several more undefined symbol errors ]
>>
>
>  The corei7-avx errors seem weird as that flag is only supposed to be used
>> for
>> x86 targets. If I remove it I later get the _poll error like with the
>>
> simulator.
>
>> Might be another case like https://codereview.chromium.org/1036123003
>> but I
>> couldn't immediately find anything that is overriding the poll function.
>>
>
> Looks like this was third_party/nss/nspr/pr/src/md/unix/unix.c .
> Commenting out
> that definition gets me a little further, now I get:
>
> duplicate symbol
> __ZN3web25BuildUserAgentFromProductERKNSt3__112basic_stringIcNS0_11char_
> traitsIcEENS0_9allocatorIcEEEE
> in:
>     libios_web.a(ios_web.user_agent.o)
>     libuser_agent.a(user_agent.user_agent.o)
> duplicate symbol
> __ZN7OSCrypt13EncryptStringERKNSt3__112basic_stringIcNS0_
> 11char_traitsIcEENS0_9allocatorIcEEEEPS6_
> in:
>     libos_crypt.a(os_crypt.os_crypt_mac.o)
>     libos_crypt.a(os_crypt.os_crypt_posix.o)
> duplicate symbol
> __ZN7OSCrypt13DecryptStringERKNSt3__112basic_stringIcNS0_
> 11char_traitsIcEENS0_9allocatorIcEEEEPS6_
> in:
>     libos_crypt.a(os_crypt.os_crypt_mac.o)
>     libos_crypt.a(os_crypt.os_crypt_posix.o)
> duplicate symbol
> __ZN7OSCrypt15DecryptString16ERKNSt3__112basic_stringIcNS0_
> 11char_traitsIcEENS0_9allocatorIcEEEEPNS1_ItN4base20string16_char_
> traitsENS4_ItEEEE
> in:
>     libos_crypt.a(os_crypt.os_crypt_mac.o)
>     libos_crypt.a(os_crypt.os_crypt_posix.o)
> duplicate symbol
> __ZN7OSCrypt15EncryptString16ERKNSt3__112basic_
> stringItN4base20string16_char_traitsENS0_9allocatorItEEEEPNS1_IcNS0_
> 11char_traitsIcEENS4_IcEEEE
> in:
>     libos_crypt.a(os_crypt.os_crypt_mac.o)
>     libos_crypt.a(os_crypt.os_crypt_posix.o)
> duplicate symbol
> __ZN3web25BuildUserAgentFromProductERKNSt3__112basic_stringIcNS0_11char_
> traitsIcEENS0_9allocatorIcEEEE
> in:
>     libios_web.a(ios_web.user_agent.o)
>     libuser_agent.a(user_agent.user_agent.o)
>
> Giving up now, but these should be easy to fix.
>
> https://codereview.chromium.org/1045113004/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[pcc1, 2015-03-31 23:24:59]

crbug.com/472311

On 2015/03/31 23:14:32, Nico wrote:
> maybe file a tracking bug with the current state, so that if someone wants
> to do this they have a continuation they can call
> 
> On Tue, Mar 31, 2015 at 4:13 PM, <mailto:pcc@chromium.org> wrote:
> 
> > On 2015/03/31 22:52:33, pcc1 wrote:
> >
> >> On 2015/03/31 21:45:04, Nico wrote:
> >> > lgtm
> >> >
> >> > I wonder if this'll Just Work for chrome/ios too.
> >>
> >
> >  Doesn't look like it. For Release-iphonesimulator I get:
> >>
> >
> >  ld: reference to symbol (which has not been assigned an address) _poll in
> >> '__ZN4mojo8embedder12_GLOBAL__N_112WaitReadableENS0_14PlatformHandleE'
> >> from
> >>
> >
> > obj/third_party/mojo/src/mojo/edk/embedder/mojo_system_
> > unittests.platform_channel_pair_posix_unittest.o
> >
> >> for architecture x86_64
> >>
> >
> >  and for Release-iphoneos I get:
> >>
> >
> >  'corei7-avx' is not a recognized processor for this target (ignoring
> >>
> > processor)
> >
> >> 'corei7-avx' is not a recognized processor for this target (ignoring
> >>
> > processor)
> >
> >> 'corei7-avx' is not a recognized processor for this target (ignoring
> >>
> > processor)
> >
> >> 'corei7-avx' is not a recognized processor for this target (ignoring
> >>
> > processor)
> >
> >> 'corei7-avx' is not a recognized processor for this target (ignoring
> >>
> > processor)
> >
> >> 'corei7-avx' is not a recognized processor for this target (ignoring
> >>
> > processor)
> >
> >> Undefined symbols for architecture arm64:
> >>    "___fixsfsi", referenced from:
> >>        __uhash_allocate in libicuuc.a(icuuc.uhash.o)
> >>    "___ledf2", referenced from:
> >>        -[NSString(CrStringDrawing) cr_stringByElidingToFitSize:] in
> >> libgtest.a(gtest.coverage_util_ios.o)
> >>    "___fixunsdfsi", referenced from:
> >>        base::PlatformThread::SetThreadPriority(base::
> >> PlatformThreadHandle,
> >> base::ThreadPriority) in libbase.a(base.platform_thread_mac.o)
> >> [ and several more undefined symbol errors ]
> >>
> >
> >  The corei7-avx errors seem weird as that flag is only supposed to be used
> >> for
> >> x86 targets. If I remove it I later get the _poll error like with the
> >>
> > simulator.
> >
> >> Might be another case like https://codereview.chromium.org/1036123003
> >> but I
> >> couldn't immediately find anything that is overriding the poll function.
> >>
> >
> > Looks like this was third_party/nss/nspr/pr/src/md/unix/unix.c .
> > Commenting out
> > that definition gets me a little further, now I get:
> >
> > duplicate symbol
> > __ZN3web25BuildUserAgentFromProductERKNSt3__112basic_stringIcNS0_11char_
> > traitsIcEENS0_9allocatorIcEEEE
> > in:
> >     libios_web.a(ios_web.user_agent.o)
> >     libuser_agent.a(user_agent.user_agent.o)
> > duplicate symbol
> > __ZN7OSCrypt13EncryptStringERKNSt3__112basic_stringIcNS0_
> > 11char_traitsIcEENS0_9allocatorIcEEEEPS6_
> > in:
> >     libos_crypt.a(os_crypt.os_crypt_mac.o)
> >     libos_crypt.a(os_crypt.os_crypt_posix.o)
> > duplicate symbol
> > __ZN7OSCrypt13DecryptStringERKNSt3__112basic_stringIcNS0_
> > 11char_traitsIcEENS0_9allocatorIcEEEEPS6_
> > in:
> >     libos_crypt.a(os_crypt.os_crypt_mac.o)
> >     libos_crypt.a(os_crypt.os_crypt_posix.o)
> > duplicate symbol
> > __ZN7OSCrypt15DecryptString16ERKNSt3__112basic_stringIcNS0_
> > 11char_traitsIcEENS0_9allocatorIcEEEEPNS1_ItN4base20string16_char_
> > traitsENS4_ItEEEE
> > in:
> >     libos_crypt.a(os_crypt.os_crypt_mac.o)
> >     libos_crypt.a(os_crypt.os_crypt_posix.o)
> > duplicate symbol
> > __ZN7OSCrypt15EncryptString16ERKNSt3__112basic_
> > stringItN4base20string16_char_traitsENS0_9allocatorItEEEEPNS1_IcNS0_
> > 11char_traitsIcEENS4_IcEEEE
> > in:
> >     libos_crypt.a(os_crypt.os_crypt_mac.o)
> >     libos_crypt.a(os_crypt.os_crypt_posix.o)
> > duplicate symbol
> > __ZN3web25BuildUserAgentFromProductERKNSt3__112basic_stringIcNS0_11char_
> > traitsIcEENS0_9allocatorIcEEEE
> > in:
> >     libios_web.a(ios_web.user_agent.o)
> >     libuser_agent.a(user_agent.user_agent.o)
> >
> > Giving up now, but these should be easy to fix.
> >
> > https://codereview.chromium.org/1045113004/
> >
> 
> To unsubscribe from this group and stop receiving emails from it, send an
email
> to mailto:chromium-reviews+unsubscribe@chromium.org.