ld64 is threaded, and uses a worker for each CPU to parse input files. But
there's a bug in the parser causing dependencies to be calculated differently
based on which files have already been parsed.
As a result, builders with more CPUs are more likely to see non-determinism.
This looks to have been fixed in a newer version of ld64, so just disable
threading for now. There's no noticible slowdown.
This contains a few hacks very specific to Qt's buildsystem. These can be
reverted once we split the build between native and target builds.
Qt's build contains a circular dependency when not using a system zlib.
By far the easiest fix is to switch to a system zlib, rather than Qt's own.
However, that confuses Qt's cross build which assumes that when using a system
zlib, it should also find a system (native) zlib for native tools. The build
breaks if that zlib is not present.
To solve this:
1. Always use a system zlib rather than the one provided by qt
2. Set force_bootstrap, which instructs the build tools to be built as though
we're cross-compiling (build != target)
3. For build tools, use qt's internal zlib so that a native zlib is not
required.
Step 3 means that if any zlib headers are found by the native build, it will
confuse Qt's internal zlib build. So we also need to make sure that the target
headers/libs aren't found. To do so, specify that our
cflags/cxxflags/cppflags/ldflags only apply for non-host builds.
qt5.7 changed the location of some of its symbols, creating a circular
dependency in Qt5Core. Rather than trying to fix that up, build our own zlib
rather than having it built for us.
Their buildsystem insists on using the installed ltranslate, but gets confused
about how to find it. Since we manually control the build order, just drop the
dependency.
Add a patch that seems to be necessary for compatibilty of libevent
2.0.22 with recent mingw-w64 gcc versions (at least GCC 5.3.1 from Ubuntu
16.04).
Without this patch the Content-Length in the HTTP header ends up as
`Content-Length: zu`, causing communication between the RPC
client and server to break down. See discussion in #8653.
Source: https://sourceforge.net/p/levent/bugs/363/
Thanks to @sstone for the suggestion.
- create a script to handle split debug. This will also eventually need to check
targets, and use dsymutil for osx.
- update config.guess/config.sub for bdb for aarch64.
- temporarily disable symbol checks for arm/aarch64
- quit renaming to linux32/linux64 and use the host directly
This also adds a hack to work around an Ubuntu bug in the gcc-multilib package:
https://bugs.launchpad.net/ubuntu/+source/gcc-defaults-armhf-cross/+bug/1347820
The problem is that gcc-multilib conflicts with the aarch toolchain.
gcc-multilib installs a symlink that points
/usr/include/asm -> /usr/include/x86_64-linux-gnu/asm.
Without this link, gcc -m32 can't find asm/errno.h (and others), since
/usr/include/x86_64-linux-gnu isn't in its default include path. But
/usr/include/i386-linux-gnu is (though it doesn't exist on disk).
So work around the problem by linking
/usr/include/i386-linux-gnu/asm -> /usr/include/x86_64-linux-gnu/asm.
The symlink fix is actually quite reasonable, but echoing the password into
sudo is nasty, and should probably be addressed in gitian itself. It makes more
sense to enable passwordless sudo for the build user by default.
2016/01/24:
Change miniwget to return HTTP status code
Increments API_VERSION to 16
2016/01/22:
Improve UPNPIGD_IsConnected() to check if WAN address is not private.
Parse HTTP response status line in miniwget.c
Remove sed-based qt PIDLIST_ABSOLUTE workaround, replace by a patch that
works for both old (such as used by Travis and Ubuntu Precise) and new
mingw (Ubuntu Trusty).
This passes `-Wa,--noexecstack` to the assembler when building
platform-specific assembly files, to signal that a non-executable stack
can be used. This is the same approach as used by Debian
(see https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=430583)
This version of miniupnpc fixes a buffer overflow in the XML (ugh)
parser during initial network discovery.
http://talosintel.com/reports/TALOS-2015-0035/
The commit fixing the vulnerability is:
79cca974a4
Reported by timothy on IRC.
This should fix the spurious comparison tool failures.
See discussion here: https://github.com/bitcoin/bitcoin/pull/6305
The race fix was cherry-picked on top of the version we're currently using, so
it should be functionally identical otherwise.
This should be functionally identical to what's in place now. It was built from
be0eef7744
That commit is the same as this pruned commit in TheBlueMatt's repo:
https://github.com/TheBlueMatt/bitcoinj/commit/0f7b5d8
Now we'll be able to trust the line numbers in the stack traces.
Boost assumes variadic templates are always available in GCC 4.4+, but
they aren't since we don't build with -std=c++11.
This applies the patch that fixed the issue in boost 1.57:
eec8085549
See also: https://svn.boost.org/trac/boost/ticket/10500
In some cases (Travis), sources and build caches may be moved around in-between
builds, and we can't necessarily trust that everything is still intact.
This introduces pre-build checks that verify against stashed checksums.
Note that this will cause all sources to be re-downloaded, since cached sources
weren't trustworthy before this.
See here for background: https://bugreports.qt.io/browse/QTBUG-34748
libxcb temporarily had an abi breakage which caused crashes when qt was
compiled against a non-compatible version. Building qt with -qt-xcb should have
shielded us from this issue, except that incompatible headers were used when
building qt's wrapper.
Make sure those headers aren't picked up by qt's build.
Details:
qt's build adds a wrapper around the xcb libs when -qt-xcb is used. This is
done to avoid having to link to a handful of different libs, which may not be
api/abi stable. This build depends on include-order, so that its files are
found before the real libxcb headers.
Our build (for other reasons related to qt's complicated build-system) injects
our prefix into CXXFLAGS. Because libxcb is found in this path, that reverses
the include-order, negating the purpose of the wrapper.
To fix, libxcb's includes are simply moved to a subdir. pkg-config ensures that
they're still found properly when needed.
To make things even more interesting, this behavior in qt's .pro files is broken:
INCLUDEPATH += $$QMAKE_CFLAGS_XCB
The INCLUDEPATH variable is processed by qmake which automatically prefixes each
entry with "-I". The QMAKE_CFLAGS_XCB variable comes from pkg-config and
already contains -I, making the path look like "-I-I/path/to/xcb/headers".
To work around that, CFLAGS/CXXFLAGS are used here rather than INCLUDEPATH.
tl;dr: Update to the newer stable toolchain and SDK for OSX without giving up
any backwards compatibility. We can move to clang 3.5 as a next step which
allows use to use libc++ and the 10.10 sdk, but we'll need to find a build that
works in gitian/travis first.
Switch to a new, better maintained fork of cctools:
https://github.com/tpoechtrager/cctools-port
I've forked this and will be working on it some as well:
https://github.com/theuni/cctools-port
This brings in:
cctools v862
ld64: v241.9
It also fixes 64bit builds, so there's no longer any need to use a 32bit clang.
Since clang is no longer tied to an old/crusty 32bit build, clang has been
upgraded to 3.3. Unfortunately, there's a bug in 3.4 that breaks builds. 3.5
works fine, but there are no binary builds compatible with precise, which is
currently used for gitian and travis. We could always build our own if
necessary.
After updating to stable clang/linker/cctools, it's possible to use a more
recent SDK. The current SDK (10.7) through the most recent 10.10 have all been
built/tested successfully, both with and without 10.6 compatibility. However,
10.10 requires clang 3.5.
SDKs >= 10.9 use libc++ rather than libstdc++. This is verified working as well.
Fixes default hidden symbol visibility for our linux->osx cross build. Without
this change, the check for working -fvisibility=hidden fails, and all symbols
are visible by default.
Ugly as this is, it's just a simple find/replace to fix a bug in Qt's configure.
They assume in an "XPLATFORM_MAC" block that the builder is capable of running
osx programs. This should be "BUILD_ON_MAC" instead.
Descriptors now make use of the dependencies builder, so results are cached.
A very new version (>= e9741525c) of Gitian should be used in order to take
advantage of caching.
We're not ready to switch to a static qt5 for Linux yet due to missing plugin
support. This adds a recipe for building a shared qt4 that we build and link
against, but don't distribute.
make USE_LINUX_STATIC_QT5=1 can be used to build static qt5 as before.
tl;dr: This solves boost visibility problems for default/release build configs
on non-Linux platforms.
When Bitcoin builds against boost's header-only classes, it ends up with
objects containing symbols that the upstream boost libs also have. Since
Bitcoin builds by default with hidden symbol visibility, it can end up trying
to link against a copy of the same symbols with default visibility.
This is not a problem on Linux because 3rd party static libs are un-exported
by default (--exclude-libs,ALL), but that is not available for MinGW and OSX.
Those platforms (and maybe others?) end up confused about which version to use.
The OSX linker spews hundreds of: "ld: warning: direct access in <foo> to
global weak symbol guard variable for <bar> means the weak symbol cannot be
overridden at runtime. This was likely caused by different translation units
being compiled with different visibility settings."
MinGW's linker complains similarly.
Since the default symbol visibility for Bitcoin is hidden and releases are
built that way as well, build Boost with hidden visibility. Linux builds Boost
this way also, but only for the sake of continuity.
This means that the linker confusion logic is reversed, so the problem will
will now be encountered if Bitcoin is built with --disable-reduce-exports, but
that's better than the current situation.
Bumps the OpenSSL version to the latest release, and kills SSL2. (SSL3 was already killed here, so I'm not sure why SSL2 was left around?)
No other changes.
