Uprev NSS to 3.18 RTM

README.chromium

 Name: Network Security Services (NSS)
 Short Name: nss
 URL: http://www.mozilla.org/projects/security/pki/nss/
 Version: 3.16.5
 License: MPL 2
 License File: nss/COPYING
 Security Critical: yes
 
 Description:
-NSS 3.16.5 with NSPR 4.10.7
+NSS 3.18 with NSPR 4.10.7
 
 This copy of NSS has been customized for Chromium.  NSPR is also put here
 rather than in a separate directory to emphasize the fact that Chromium is
 using NSPR strictly as an NSS dependency.
 
 We took a subset of NSS, omitting the SSL and SMIME libraries and the
 built-in root CA certificates module.  This NSS subset satisfies the
 dependencies of the NSS SSL library in src/net/third_party/nss.  Do NOT use
 this copy of NSS on platforms that have NSS as system libraries, such as
 Linux.
 
 The source code was checked out from the mozilla.org CVS or hg repository using
 the nspr-checkout.sh and nss-checkout.sh scripts in the scripts directory.
-The current source code was checked out with the hg tag NSS_3_16_5_RTM
+The current source code was checked out with the hg tag NSS_3_18_RTM
 and the hg tag NSPR_4_10_8_RTM.
 
 Local Modifications:
 
 We made the following local changes to NSPR.
 - patches/nspr-static.patch: to build NSPR as static libraries.  See NSPR
   bug 533014 (https://bugzilla.mozilla.org/show_bug.cgi?id=533014).
 - patches/prcpucfg.h: added to the nspr/pr/include directory.
 - patches/nspr-attach-as-system-thread.patch: attach a "foreign" thread
   (a thread not created by NSPR) to NSPR as a "system" thread rather than
   a "user" thread, which needs to terminate before PR_Cleanup can return.
   (The "system" vs. "user" thread distinction comes from Java, and
   ultimately from Solaris threads.)  This is a workaround for
   http://crbug.com/40663.
 - patches/nspr-remove-io.patch: Remove IO operations in NSPR to allow NSS
   to work in the sandbox.  Do not initialize IO when initializing NSPR.
   Windows version of NSPR also tried to use getaddrinfo to resolve hostname
   in a SSL connection.  By removing _PR_HAVE_GETADDRINFO this will force it
   to use PR_GetHostByName.  Removing _PR_INET6_PROBE will prevent it from
   creating an IPv6 socket to probe if IPv6 is there.
   DO NOT upstream this patch.
 
 We made the following local changes to NSS.
-- patches/nss-static.patch: to build NSS as static libraries and omit
-  libpkix (the new certification path validation library) and
-  softoken/legacydb (support for the old Berkeley DB databases).  See NSS
-  bug 534471 (https://bugzilla.mozilla.org/show_bug.cgi?id=534471).
 - nss/exports_win.def: The list of exports to use when building nss as a
   dynamic library (crnss.dll).
 - nss/lib/ckfw/builtins/certdata.c: a generated file. Do an upstream NSS
   build and copy the generated certdata.c.
 - nss/lib/freebl/nss_build_config_mac.h: a header that defines the target
   arch specific configuration macros for lib/freebl on iOS and Mac OS X.
   This works around the lack of support for the xcode_settings
   GCC_PREPROCESSOR_DEFINITIONS[arch=foo] by the ninja GYP generator
   (http://crbug.com/122592).
 - nss/lib/freebl/mpi/mpi_arm_mac.c: a wrapper file for mpi_arm.c for iOS
   and Mac OS X. This works around the inability to specify target arch
   specific source files in Xcode.
 - patches/nss-remove-fortezza.patch: remove Fortezza certificate support
   from PK11_ImportPublicKey.  See NSS bug 668397
   (https://bugzilla.mozilla.org/show_bug.cgi?id=668397).
 - patches/nss-urandom-abort.patch: call abort() if NSS cannot read from
   /dev/urandom.  See Chromium issue 244661 (http://crbug.com/244661).
 - patches/nss-chacha20-poly1305.patch: Support ChaCha20+Poly1305 cipher
   suites.  See NSS bug 917571
   (https://bugzilla.mozilla.org/show_bug.cgi?id=917571).
+- patches/nss-static.patch: to build NSS as static libraries and omit

[wtc, 2015/03/20 00:09:46]

I guess you moved this patch here so that all the patches are listed together.
If so, I suggest we list this patch first, before nss-remove-fortezza.patch,
because this is the biggest/most permanent (in the sense that it is most
difficult to upstream) one among the patches.

[Ryan Sleevi, 2015/03/20 00:15:36]

No, because I forgot to apply it first when merging all the bits. It also made
it easier to track the changes, by merging the small bits first, resolving the
conflicts, and then the larger change.

As you note, because this one is the most likely to create conflicts, any
further conflicts will propagate to the rest of the patches; applying this last
reduces the diff hunk fuzzing of the earlier patches.

+  libpkix (the new certification path validation library) and
+  softoken/legacydb (support for the old Berkeley DB databases).  See NSS
+  bug 534471 (https://bugzilla.mozilla.org/show_bug.cgi?id=534471).