Roll src/third_party/boringssl/src/ to 209b2562235f7dab66b8260624e7b3c5b00d14a6.

Roll src/third_party/boringssl/src/ to 209b2562235f7dab66b8260624e7b3c5b00d14a6.

$ git log 7f15ff53d..209b25622 --date=short --format='%ad %ae %s'
2015-07-20 davidben Fix o2i_ECPublicKey documentation.
2015-07-20 davidben Surround immintrin.h includes with warning pragmas.
2015-07-12 davidben Simplify the AMD-specific codepath.
2015-07-11 davidben Replace cpuid assembly with C code.
2015-06-29 davidben Reserve ex_data index zero for app_data.
2015-07-11 davidben Remove incorrect TODO(fork).
2015-07-11 davidben Extra documentation.
2015-07-16 agl Fix typo in |HMAC_CTX_cleanup|.
2015-06-29 davidben ERR_LIB_USER should be the last error.
2015-06-29 davidben Remove the func parameter to OPENSSL_PUT_ERROR.
2015-06-28 davidben Get rid of err function codes.

BUG=none

Committed: https://crrev.com/2cc798746473a6a002006e0adc24afb1b6c05f59
Cr-Commit-Position: refs/heads/master@{#339724}

[davidben, 2015-07-21 18:24:11]

davidben@chromium.org changed reviewers:
+ agl@chromium.org

[davidben, 2015-07-21 18:27:56]

https://codereview.chromium.org/1244093002/diff/20001/net/android/keystore_op...
File net/android/keystore_openssl.cc (right):

https://codereview.chromium.org/1244093002/diff/20001/net/android/keystore_op...
net/android/keystore_openssl.cc:179: OPENSSL_PUT_ERROR(RSA,
RSA_R_UNKNOWN_ALGORITHM_TYPE);
This will be going away once the Chrome/Android code uses the async signing
stuff (Android and everything else go through slightly different codepaths right
now. I'm probably going to have my new minion finish this cleanup as his first
task.), but it occurs to me you'll probably have to do something similar in
Conscrypt. That should have occurred to me. Oops.

[agl, 2015-07-21 19:25:58]

lgtm

https://codereview.chromium.org/1244093002/diff/20001/net/android/keystore_op...
File net/android/keystore_openssl.cc (right):

https://codereview.chromium.org/1244093002/diff/20001/net/android/keystore_op...
net/android/keystore_openssl.cc:179: OPENSSL_PUT_ERROR(RSA,
RSA_R_UNKNOWN_ALGORITHM_TYPE);
On 2015/07/21 18:27:56, David Benjamin wrote:
> This will be going away once the Chrome/Android code uses the async signing
> stuff (Android and everything else go through slightly different codepaths
right
> now. I'm probably going to have my new minion finish this cleanup as his first
> task.), but it occurs to me you'll probably have to do something similar in
> Conscrypt. That should have occurred to me. Oops.

You mean that the code to bridge BoringSSL to Java-backed certs (in Conscrypt)
and keystore-backed certs (in keystore) will have to change? The ENGINE-based
code will still work, won't it? I don't have any problem changing Conscrypt, but
things like wpa_supplicant expect an ENGINE at the moment and changing that code
is more of a pain.

[davidben, 2015-07-21 19:43:25]

https://codereview.chromium.org/1244093002/diff/20001/net/android/keystore_op...
File net/android/keystore_openssl.cc (right):

https://codereview.chromium.org/1244093002/diff/20001/net/android/keystore_op...
net/android/keystore_openssl.cc:179: OPENSSL_PUT_ERROR(RSA,
RSA_R_UNKNOWN_ALGORITHM_TYPE);
On 2015/07/21 19:25:58, agl wrote:
> On 2015/07/21 18:27:56, David Benjamin wrote:
> > This will be going away once the Chrome/Android code uses the async signing
> > stuff (Android and everything else go through slightly different codepaths
> right
> > now. I'm probably going to have my new minion finish this cleanup as his
first
> > task.), but it occurs to me you'll probably have to do something similar in
> > Conscrypt. That should have occurred to me. Oops.
> 
> You mean that the code to bridge BoringSSL to Java-backed certs (in Conscrypt)
> and keystore-backed certs (in keystore) will have to change? The ENGINE-based
> code will still work, won't it? I don't have any problem changing Conscrypt,
but
> things like wpa_supplicant expect an ENGINE at the moment and changing that
code
> is more of a pain.

Just the Conscrypt stuff. You'll need to make the analogous change to
OPENSSL_PUT_ERROR lines when pulling the new BoringSSL in. wpa_supplicant
doesn't have any of those.

[davidben, 2015-07-21 19:43:30]

The CQ bit was checked by davidben@chromium.org

[commit-bot: I haz the power, 2015-07-21 19:44:03]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1244093002/20001

[commit-bot: I haz the power, 2015-07-21 19:50:38]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-07-21 19:51:37]

Patchset 2 (id:??) landed as
https://crrev.com/2cc798746473a6a002006e0adc24afb1b6c05f59
Cr-Commit-Position: refs/heads/master@{#339724}

[achuithb, 2015-07-21 21:15:00]

achuith@chromium.org changed reviewers:
+ achuith@chromium.org

[achuithb, 2015-07-21 21:15:00]

This is a compile error on chromeos at TOT

https://codereview.chromium.org/1244093002/diff/20001/net/ssl/openssl_ssl_uti...
File net/ssl/openssl_ssl_util.cc (right):

https://codereview.chromium.org/1244093002/diff/20001/net/ssl/openssl_ssl_uti...
net/ssl/openssl_ssl_util.cc:182: ERR_put_error(OpenSSLNetErrorLib(), err,
location.function_name(),
../../net/ssl/openssl_ssl_util.cc: In function 'void
net::OpenSSLPutNetError(const tracked_objects::Location&, int)':
../../net/ssl/openssl_ssl_util.cc:182:67: error: invalid conversion from 'const
char*' to 'int' [-fpermissive]
   ERR_put_error(OpenSSLNetErrorLib(), err, location.function_name(),
                                                                   ^
In file included from ../../net/ssl/openssl_ssl_util.cc:9:0:
../../third_party/boringssl/src/include/openssl/err.h:313:21: note: initializing
argument 3 of 'void ERR_put_error(int, int, int, const char*, unsigned int)'
 OPENSSL_EXPORT void ERR_put_error(int library, int func, int reason,

[achuithb, 2015-07-21 21:17:12]

Also this:
In file included from ../../third_party/boringssl/err_data.c:19:0:
../../third_party/boringssl/src/include/openssl/type_check.h:83:16: error: size
of array 'OPENSSL_COMPILE_ASSERT_library_values_changed_31' is negative
   typedef char OPENSSL_COMPILE_ASSERT_##msg[((cond) ? 1 : -1)]
                ^
../../third_party/boringssl/err_data.c:52:1: note: in expansion of macro
'OPENSSL_COMPILE_ASSERT'
 OPENSSL_COMPILE_ASSERT(ERR_LIB_HKDF == 31, library_values_changed_31);
 ^
../../third_party/boringssl/src/include/openssl/type_check.h:83:16: error: size
of array 'OPENSSL_COMPILE_ASSERT_library_values_changed_32' is negative
   typedef char OPENSSL_COMPILE_ASSERT_##msg[((cond) ? 1 : -1)]
                ^
../../third_party/boringssl/err_data.c:53:1: note: in expansion of macro
'OPENSSL_COMPILE_ASSERT'
 OPENSSL_COMPILE_ASSERT(ERR_LIB_USER == 32, library_values_changed_32);
 ^
[807/8721] CXX obj/net/ssl/net.openssl_ssl_util.o

[agl, 2015-07-21 21:19:56]

achuithb: are you sure that you did a gclient sync?

[achuithb, 2015-07-21 21:22:00]

On 2015/07/21 21:19:56, agl wrote:
> achuithb: are you sure that you did a gclient sync?

Sorry for the noise, don't think I did :(