ba8950ee01 build: optionally skip external warnings (Vasil Dimov)
Pull request description:
Add an option to `./configure` to suppress compilation warnings from
external headers. The option is off by default (no change in behavior,
show warnings from external headers).
This option is useful if e.g. Boost or Qt is installed outside of
`/usr/include` (warnings from headers in `/usr/include` are already
suppressed by default) and those warnings stand in the way of compiling
Bitcoin Core with `-Werror[=...]` or they just clutter the build output
too much and make our own warnings hard to spot.
`-isystem /usr/include` bricks GCC's `#include_next`, so we use
`-idirafter` instead. This way we don't have to treat `/usr/include`
specially.
ACKs for top commit:
practicalswift:
ACK ba8950ee0134a7958e3e9b041cd54d222feb09a1: diff looks correct!
hebasto:
ACK ba8950ee01, tested on Linux Mint 20 (x86_64).
luke-jr:
utACK ba8950ee01
Tree-SHA512: 9b54fae8590be6c79f2688a5aca09e0a9067f481dabecdd49bb278c08a62ac2b0cc704c894fbd53240e77ac84da0c7a237845df0a696cfbdb0359e1c8e2e10c9
Add an option to `./configure` to suppress compilation warnings from
external headers. The option is off by default (no change in behavior,
show warnings from external headers).
This option is useful if e.g. Boost or Qt is installed outside of
`/usr/include` (warnings from headers in `/usr/include` are already
suppressed by default) and those warnings stand in the way of compiling
Bitcoin Core with `-Werror[=...]` or they just clutter the build output
too much and make our own warnings hard to spot.
c4b85ba704 Bugfix: Define and use HAVE_FDATASYNC correctly outside LevelDB (Luke Dashjr)
Pull request description:
Fixes a bug introduced in #19614
The LevelDB-specific fdatasync check was only using `AC_SUBST`, which works for Makefiles, but doesn't define anything for C++. Furthermore, the #define is typically 0 or 1, never undefined.
This fixes both issues by defining it and checking its value instead of whether it is merely defined.
Pulled out of #14501 by fanquake's request
ACKs for top commit:
fanquake:
ACK c4b85ba704 - thanks for catching and fixing my mistake.
laanwj:
Code review ACK c4b85ba704
Tree-SHA512: 91d5d426ba000b4f3ee7e2315635e24bbb23ceff16269ddf4f65a63d25fc9e9cf94a3b236eed2f8031cc36ddcf78aeb5916efcb244f415943a8a12f907ede8f9
b536813cef build: add -fstack-clash-protection to hardening flags (fanquake)
076183b36b build: add -fcf-protection=full to hardening options (fanquake)
Pull request description:
Beginning with Ubuntu `19.10`, it's packaged GCC now has some additional hardening options enabled by default (in addition to existing defaults like `-fstack-protector-strong` and reducing the minimum ssp buffer size). The new additions are`-fcf-protection=full` and `-fstack-clash-protection`.
> -fcf-protection=[full|branch|return|none]
> Enable code instrumentation of control-flow transfers to increase program security by checking that target addresses of control-flow transfer instructions (such as indirect function call, function return, indirect jump) are valid. This prevents diverting the flow of control to an unexpected target. This is intended to protect against such threats as Return-oriented Programming (ROP), and similarly call/jmp-oriented programming (COP/JOP).
> -fstack-clash-protection
> Generate code to prevent stack clash style attacks. When this option is enabled, the compiler will only allocate one page of stack space at a time and each page is accessed immediately after allocation. Thus, it prevents allocations from jumping over any stack guard page provided by the operating system.
If your interested you can grab `gcc-9_9.3.0-10ubuntu2.debian.tar.xz` from https://packages.ubuntu.com/focal/g++-9. The relevant changes are part of the `gcc-distro-specs` patches, along with the relevant additions to the gcc manages:
> NOTE: In Ubuntu 19.10 and later versions, -fcf-protection is enabled by default for C, C++, ObjC, ObjC++, if none of -fno-cf-protection nor -fcf-protection=* are found.
> NOTE: In Ubuntu 19.10 and later versions, -fstack-clash-protection is enabled by default for C, C++, ObjC, ObjC++, unless -fno-stack-clash-protection is found.
So, if you're C++ using GCC on Ubuntu 19.10 or later, these options will be active unless you explicitly opt out. This can be observed with a small test:
```c++
int main() { return 0; }
```
```bash
g++ --version
g++ (Ubuntu 9.3.0-10ubuntu2) 9.3.0
g++ test.cpp
objdump -dC a.out
..
0000000000001129 <main>:
1129: f3 0f 1e fa endbr64
112d: 55 push %rbp
112e: 48 89 e5 mov %rsp,%rbp
1131: b8 00 00 00 00 mov $0x0,%eax
1136: 5d pop %rbp
1137: c3 retq
1138: 0f 1f 84 00 00 00 00 nopl 0x0(%rax,%rax,1)
113f: 00
# recompile opting out of control flow protection
g++ test.cpp -fcf-protection=none
objdump -dC a.out
...
0000000000001129 <main>:
1129: 55 push %rbp
112a: 48 89 e5 mov %rsp,%rbp
112d: b8 00 00 00 00 mov $0x0,%eax
1132: 5d pop %rbp
1133: c3 retq
1134: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1)
113b: 00 00 00
113e: 66 90 xchg %ax,%ax
```
Note the insertion of an `endbr64` instruction when compiling and _not_ opting out. This instruction is part of the Intel Control-flow Enforcement Technology [spec](https://software.intel.com/sites/default/files/managed/4d/2a/control-flow-enforcement-technology-preview.pdf), which the GCC control flow implementation is based on.
If we're still doing gitian builds for the `0.21.0` and `0.22.0` releases, we'd likely update the gitian image to Ubuntu Focal, which would mean that the GCC used for gitian builds would also be using these options by default. So we should decide whether we want to explicitly turn these options on as part of our hardening options (although not just for this reason), or, we should be opting-out.
GCC has supported both options since 8.0.0. Clang has supported `-fcf-protection` from 7.0.0 and will support `-fstack-clash-protection` in it's upcoming [11.0.0 release](https://clang.llvm.org/docs/ReleaseNotes.html#id6).
ACKs for top commit:
jamesob:
ACK b536813cef ([`jamesob/ackr/18921.1.fanquake.build_add_stack_clash_an`](https://github.com/jamesob/bitcoin/tree/ackr/18921.1.fanquake.build_add_stack_clash_an))
laanwj:
Code review ACK b536813cef
Tree-SHA512: abc9adf23cdf1be384f5fb9aa5bfffdda86b9ecd671064298d4cda0440828b509f070f9b19c88c7ce50ead9ff32afff9f14c5e78d75f01241568fbfa077be0b7
1ccb9f30c0 Move Win32 defines to configure.ac to ensure they are globally defined (Luke Dashjr)
Pull request description:
#9245 no longer needs this, since the main `_WIN32_WINNT` got bumped by something else.
So rather than just lose it, might as well get it merged in independently.
I'm not aware of any practical effects, but it seems safer to use the same API versions everywhere.
ACKs for top commit:
fanquake:
ACK 1ccb9f30c0 - checked that the binaries produced are the same.
Tree-SHA512: 273e9186579197be01b443b6968e26b9a8031d356fabc5b73aa967fcdb837df195b7ce0fc4e4529c85d9b86da6f2d7ff1bf56a3ff0cbbcd8cee8a9c2bf70a244
2f8a4c9a06 build: Enable some commonly enabled compiler diagnostics (practicalswift)
Pull request description:
Enable some commonly enabled compiler diagnostics as discussed in #17344.
| Compiler diagnostic | no# of emitted unique GCC warnings in `master` | no# of emitted unique Clang warnings in `master` |
| ------------- | ------------- | ------------- |
| `-Wduplicated-branches`: Warn if `if`/`else` branches have duplicated code | 0 | Not supported |
| `-Wduplicated-cond`: Warn if `if`/`else` chain has duplicated conditions | 0 | Not supported |
| `-Wlogical-op`: Warn about logical operations being used where bitwise were probably wanted | 0 | Not supported |
| `-Woverloaded-virtual`: Warn if you overload (not `override`) a virtual function | 0 | 0 |
| ~~`-Wunused-member-function`: Warn on unused member function~~ | Not supported | 2 |
| ~~`-Wunused-template`: Warn on unused template~~ | Not supported | 1 |
There is a large overlap between this list and [Jason Turner's list of recommended compiler diagnostics in the Collaborative Collection of C++ Best Practices (`cppbestpractices`) project](https://github.com/lefticus/cppbestpractices/blob/master/02-Use_the_Tools_Available.md#gcc--clang). There is also an overlap with the recommendations given in the [C++ Core Guidelines](https://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines) (with editors Bjarne Stroustrup and Herb Sutter).
Closes#17344.
ACKs for top commit:
jonatack:
ACK 2f8a4c9a06 no warnings for me with these locally on debian 5.7.10-1 (2020-07-26) x86_64 with gcc 10 and clang 12
fanquake:
ACK 2f8a4c9a06 - no-longer seeing any obvious issues with doing this.
hebasto:
ACK 2f8a4c9a06, no new warnings in Travis jobs.
Tree-SHA512: f669ea22b31263a555f999eff6a9d65750662e95831b188c3192a2cf0127fb7b5136deb762a6b0b7bbdfb0dc6a40caf48251a62b164fffb81dd562bdd15ec3c8
70452a070b build: set minimum required Boost to 1.58 (fanquake)
Pull request description:
Any systems which only have an older installable Boost can use depends.
1.58.0 retains compatibility with the packages [installable on Ubuntu 16.04](https://packages.ubuntu.com/xenial/libboost-dev).
The projects usage of Boost wont be going away any time soon, if ever (i.e #15382), and our usage of the test framework.
Fixes: #19506
ACKs for top commit:
practicalswift:
ACK 70452a070b -- patch looks correct
laanwj:
ACK 70452a070b
hebasto:
ACK 70452a070b, tested on Linux Mint 20 (x86_64).
Tree-SHA512: d290415e3c70a394b3d7659c0480a35b4082bdce8d48b1c64a0025f7ad6e21567b4dc85813869513ad246d27f950706930410587c11c1aa3693ae6245084765c
This flag was added to binutils/ld in the 2.30 release,
see commit c11c786f0b45617bb8807ab6a57220d5ff50e414:
> The new "-z separate-code" option will generate separate code LOAD
segment which must be in wholly disjoint pages from any other data.
It was made the default for Linux/x86 targets in the 2.31 release, see commit
f6aec96dce1ddbd8961a3aa8a2925db2021719bb:
> This patch adds --enable-separate-code to ld configure to turn on
-z separate-code by default and enables it by default for Linux/x86.
This avoids mixing code pages with data to improve cache performance
as well as security.
> To reduce x86-64 executable and shared object sizes, the maximum page
size is reduced from 2MB to 4KB when -z separate-code is turned on by
default. Note: -z max-page-size= can be used to set the maximum page
size.
> We compared SPEC CPU 2017 performance before and after this change on
Skylake server. There are no any significant performance changes.
Everything is mostly below +/-1%.
Support was also added to LLVMs lld: https://reviews.llvm.org/D64903, however
there is remains off by default.
There were concerns about an increase in binary size, however in our case, the
increase (1 page worth of bytes) would seem negligible, given we are shipping a
multi-megabyte binary, which then downloads 100's of GBs of data.
Also note that most recent versions of distros are shipping a new enough version
of binutils that this is available and/or on by default (assuming the distro has
not turned it off, I haven't checked everywhere):
CentOS 8: 2.30
Debian Buster 2.31.1
Fedora 29: 2.31.1
FreeBSD: 2.33
GNU Guix: 2.33 / 2.34
Ubuntu 18.04: 2.30
Related threads / discussion:
https://bugzilla.redhat.com/show_bug.cgi?id=1623218
While testing #19530 I noticed that we couldn't call dsymutil after LTO:
```bash
../libtool: line 10643: x86_64-apple-darwin16-dsymutil: command not found
```
This updates configure to call `AC_PATH_TOOL` so that we end up with the
full path to dsymutil, similar to `otool` and `install_name_tool`, ie:
`/bitcoin/depends/x86_64-apple-darwin16/share/../native/bin/x86_64-apple-darwin16-otool`.
Debian GCC ignores -Wformat-security, without -Wformat, which
means when we test for it, it currently fails:
```bash
checking whether C++ compiler accepts -Wformat-security... no
...
configure:15907: checking whether C++ compiler accepts -Wformat-security
configure:15926: g++ -std=c++11 -c -g -O2 -Werror -Wformat-security conftest.cpp >&5
cc1plus: error: '-Wformat-security' ignored without '-Wformat' [-Werror=format-security]
cc1plus: all warnings being treated as errors
```
Fix this by just combining the -Wformat and -Wformat-security checks
together.
92bc268e4a build: Detect missed pkg-config early (Hennadii Stepanov)
1739eb23d8 build: Drop unused use_pkgconfig variable (Hennadii Stepanov)
a661449a2e build: Drop use_pkgconfig check for libmultiprocess check (Hennadii Stepanov)
90b95e7929 build: Drop dead non-pkg-config code for libevent check (Hennadii Stepanov)
44a14afbb8 build: Drop dead non-pkg-config code for qrencode check (Hennadii Stepanov)
10cbae0c39 build: Drop dead non-pkg-config code for ZMQ check (Hennadii Stepanov)
06cfc9cadf build: Fix indentation in UNIVALUE check (Hennadii Stepanov)
6fd2118e77 build: Drop dead non-pkg-config code for UNIVALUE check (Hennadii Stepanov)
e9edbe4dbd build: Always use pkg-config (Hennadii Stepanov)
9e2e753b06 build: Always define ZMQ_STATIC for MinGW (Hennadii Stepanov)
Pull request description:
This PR:
- is based on #18297 (already merged)
- drops all of the non-pkg-config paths from the `configure` script
Ref: #17768
ACKs for top commit:
fanquake:
ACK 92bc268e4a. I re-gitian-built. There are a couple follow-ups that I'll PR shortly. Thanks for addressing my feedback above. I took too long to get back to this.
laanwj:
ACK 92bc268e4a
Tree-SHA512: 83c2d9cf03518867a1ebf7e26a8fc5b6dd8962ef983fe0d84e0c7eb74717f4c36a834da02faf0e503ffd87167005351671cf040c0d4ddae57ee152a6ff84012b
c4ffcf07af build: remove BIP70 configure option (fanquake)
Pull request description:
This was left in after #17165, so that anyone who had been compiling
with (already disabled by default) BIP70 would realise that support
had been completely removed in 0.20.0. However we should be able to
remove it for 0.21.0.
ACKs for top commit:
jnewbery:
utACK c4ffcf07af
MarcoFalke:
ACK c4ffcf07af with or without the "catch-all reject"
Tree-SHA512: a5dd4231ed97c9dd1984fb90d69a8725df2fdda0b963269b0575601c74528e5d820a4a863c428f8ede86eaae2a1606671fe1fcebdeb96b1023f7a5f899270284
9952242c03 build: improve builtin_clz* detection (fanquake)
Pull request description:
Fixes#19402.
The way we currently test for `__builtin_clz*` support with `AC_CHECK_DECLS` does not work with Clang:
```bash
configure:21492: clang++-10 -std=c++11 -c -g -O2 -DHAVE_BUILD_INFO -D__STDC_FORMAT_MACROS conftest.cpp >&5
conftest.cpp💯10: error: builtin functions must be directly called
(void) __builtin_clz;
^
1 error generated.
```
This also removes the `__builtin_clz()` check, as we don't actually use it anywhere, and it's trvial to re-add detection if we do start using it at some point. If this is controversial then I'll add a test for it as well.
ACKs for top commit:
sipa:
ACK 9952242c03
laanwj:
ACK 9952242c03
Tree-SHA512: 695abb1a694a01a25aaa483b4fffa7d598842f2ba4fe8630fbed9ce5450b915c33bf34bb16ad16a16b702dd7c91ebf49fe509a2498b9e28254fe0ec5177bbac0
The way we currently test with AC_CHECK_DECLS do not work with Clang:
```bash
configure:21492: clang++-10 -std=c++11 -c -g -O2 -DHAVE_BUILD_INFO -D__STDC_FORMAT_MACROS conftest.cpp >&5
conftest.cpp💯10: error: builtin functions must be directly called
(void) __builtin_clz;
^
1 error generated.
```
This also removes the __builtin_clz() check, as we don't actually use
it anywhere, and it's trvial to re-add detection if we do start using
it at some point.
On OS X, when searching Homebrew keg-only packages for BDB 4.8, if we find it,
use BDB_CPPFLAGS and BDB_LIBS instead of CFLAGS and LIBS for the result. This
is (1) more correct, and (2) necessary in order to give this location
priority over other directories in the include search path, which may include
system include directories with other versions of BDB.
This option causes the compiler to insert probes whenever stack space
is allocated statically or dynamically to reliably detect stack overflows
and thus mitigate the attack vector that relies on jumping over a stack
guard page as provided by the operating system.
This option is now enabled by default in Ubuntu GCC as of 19.10.
Available in GCC 8 and Clang 11.
8a26848c46 build: Fix m4 escaping (Hennadii Stepanov)
9123ec15db build: Remove extra tokens warning (Hennadii Stepanov)
fded4f48c3 build: Remove duplicated QT_STATICPLUGIN define (Hennadii Stepanov)
05a93d5d96 build: Fix indentation in bitcoin_qt.m4 (Hennadii Stepanov)
ddbb419310 build: Use pkg-config in BITCOIN_QT_CONFIGURE for all hosts (Hennadii Stepanov)
492971de35 build: Fix mingw pkgconfig file and dependency naming (Hennadii Stepanov)
Pull request description:
This PR makes `bitcoin_qt.m4` to use `pkg-config` for all hosts and removes non-pkg-config paths from it. This is a step towards the idea which was clear [stated](https://github.com/bitcoin/bitcoin/pull/8314#issue-76644643) by Cory Fields:
> I believe the consensus is to treat Windows like the others and require pkg-config across the board. We can drop all of the non-pkg-config paths, and simply AC_REQUIRE(PKG_PROG_PKG_CONFIG)
There are two unsolved problems with this PR. If depends is built with `DEBUG=1` the `configure` script fails to pickup Qt:
- for macOS host (similar to, but not the same as #16391)
- for Windows host (regression)
The fix is ~on its way~ submitted in #18298 (as a followup).
Also this PR picks some small improvements from #17820.
ACKs for top commit:
theuni:
Code review ACK 8a26848c46
dongcarl:
Code Review ACK 8a26848c46
laanwj:
Code review ACK 8a26848c46
Tree-SHA512: 3b25990934b939121983df7707997b31d61063b1207d909f539d69494c7cb85212f353092956d09ecffebb9fef28b869914dd1216a596d102fcb9744bb5487f7
This was left in after #17165, so that anyone who had been compiling
with (already disabled by default) BIP70 would realise that support
had been completely removed in 0.20.0. However we should be able to
remove it for 0.21.0.
Fuzzing code uses C++17 specific code (e.g. std::optional), so it is not
possible to compile with --enable-fuzz and without --enable-c++17.
Thus, turn on --enable-c++17 whenever --enable-fuzz is used.
eea8114657 build: Enable unreachable-code-loop-increment (Jonathan Schoeller)
d15db4b1fc refactor: Fix unreachable code in init arg checks (Jonathan Schoeller)
Pull request description:
Closes: #19017
In #19015 it's been suggested that we add some new compiler warnings to our build. Some of these, such as `-Wunreachable-code-loop-increment`, generate warnings. We'll likely want to fix these up if we're going to turn these warnings on.
```shell
init.cpp:969:5: warning: loop will run at most once (loop increment never executed) [-Wunreachable-code-loop-increment]
for (const auto& arg : gArgs.GetUnsuitableSectionOnlyArgs()) {
^~~
1 warning generated.
```
aa8d76806c/src/init.cpp (L968-L972)
To fix this, collect all errors, and output them in a single error message after the loop completes. This resolves the unreachable code warning, and avoids popup hell that could result from outputting a seperate message for each error or warning one by one.
ACKs for top commit:
laanwj:
Code review ACK eea8114657
hebasto:
re-ACK eea8114657, only suggested changes applied since the [previous](https://github.com/bitcoin/bitcoin/pull/19131#pullrequestreview-421772387) review.
Tree-SHA512: 2aa3ceb7fab581b6ba2580900668388d8eba1c3001c8ff9c11c1f4a9a10fbc37f30e590249862676858446e3f4950140a252953ba1643ba3bfd772f8eae20583
Andrew Chow