88c83636d5a56bd9551577139786bdd3e74852c2 guix: Update documentation for time-machine (Carl Dong)
e6050884fdabfa6e51e6afce2041d91e60a5adec guix: Pin Guix using `guix time-machine` (Carl Dong)
Pull request description:
An alternative to #16519, pinning our version of Guix and eliminating a `guix pull` and changing the default Guix profile of builders.
I think this method might be superior, as it:
- Eliminates the possibility of future changes to the `guix environment` command line interface breaking our builds
- Eliminates the need to set up a separate channel repo
It is a more general pinning solution than #16519.
-----
The reason why I didn't originally propose this is because `guix time-machine` is a recent addition to Guix, only available since `f675f8dec73d02e319e607559ed2316c299ae8c7`
ACKs for top commit:
fanquake:
ACK 88c83636d5a56bd9551577139786bdd3e74852c2
Tree-SHA512: 85e03b0987ffa86da73e02801e1cd8b7622698d70c4ba4e60561611be1e9717d661c2811a59b3e137b1b8eef2d0ba37c313867d035ebc89c3bd06a23a078064a
ac831339cbfa65b1f7576c53b5d9a94841db9868 doc: Fix some misspellings (randymcmillan)
Pull request description:
Here is a more thorough lint-spelling update.
This PR takes care of easy to fix spelling errors to clean up the linting stages.
There are misspellings coded into the functional tests.
That is a whole separate job within itself.
ACKs for top commit:
practicalswift:
ACK ac831339cbfa65b1f7576c53b5d9a94841db9868 -- diff looks correct
Tree-SHA512: d8fad83fed083715655f148263ddeffc6752c8007d568fcf3dc2c418ccd5db70089ce3ccfd3994fcbd78043171402eb9cca5bdd5125287e22c42ea305aaa6e9d
751549b52a9a4cd27389d807ae67f02bbb39cd7f contrib: guix: Additional clarifications re: substitutes (Carl Dong)
cd3e947f50db7cfe05c05b368c25742193729a62 contrib: guix: Various improvements. (Carl Dong)
8dff3e48a9e03299468ed3b342642f01f70da9db contrib: guix: Clarify SOURCE_DATE_EPOCH. (Carl Dong)
3e80ec3ea9691c7c89173de922a113e643fe976b contrib: Add deterministic Guix builds. (Carl Dong)
Pull request description:
~~**This post is kept updated as this project progresses. Use this [latest update link](https://github.com/bitcoin/bitcoin/pull/15277#issuecomment-497303718) to see what's new.**~~
Please read the `README.md`.
-----
### Guix Introduction
This PR enables building bitcoin in Guix containers. [Guix](https://www.gnu.org/software/guix/manual/en/html_node/Features.html) is a transactional package manager much like Nix, but unlike Nix, it has more of a focus on [bootstrappability](https://www.gnu.org/software/guix/manual/en/html_node/Bootstrapping.html) and [reproducibility](https://www.gnu.org/software/guix/blog/tags/reproducible-builds/) which are attractive for security-sensitive projects like bitcoin.
### Guix Build Walkthrough
Please read the `README.md`.
[Old instructions no. 4](https://github.com/bitcoin/bitcoin/pull/15277#issuecomment-497303718)
[Old instructions no. 3](https://github.com/bitcoin/bitcoin/pull/15277#issuecomment-493827011)
[Old instructions no. 2](https://github.com/bitcoin/bitcoin/pull/15277#issuecomment-471658439)
<details>
<summary>Old instructions no. 1</summary>
In this PR, we define a Guix [manifest](https://www.gnu.org/software/guix/manual/en/html_node/Invoking-guix-package.html#profile_002dmanifest) in `contrib/guix/manifest.scm`, which declares what packages we want in our environment.
We can then invoke
```
guix environment --manifest=contrib/guix/manifest.scm --container --pure --no-grafts --no-substitutes
```
To have Guix:
1. Build an environment containing the packages we defined in our `contrib/guix/manifest.scm` manifest from the Guix bootstrap binaries (see [bootstrappability](https://www.gnu.org/software/guix/manual/en/html_node/Bootstrapping.html) for more details).
2. Start a container with that environment that has no network access, and no access to the host's filesystem except to the `pwd` that it was started in.
3. Drop you into a shell in that container.
> Note: if you don't want to wait hours for Guix to build the entire world from scratch, you can eliminate the `--no-substitutes` option to have Guix download from available binary sources. Note that this convenience doesn't necessarily compromise your security, as you can check that a package was built correctly after the fact using `guix build --check <packagename>`
Therefore, we can perform a build of bitcoin much like in Gitian by invoking the following:
```
make -C depends -j"$(nproc)" download && \
cat contrib/guix/build.sh | guix environment --manifest=contrib/guix/manifest.scm --container --pure --no-grafts --no-substitutes
```
We don't include `make -C depends -j"$(nproc)" download` inside `contrib/guix/build.sh` because `contrib/guix/build.sh` is run inside the container, which has no network access (which is a good thing).
</details>
### Rationale
I believe that this represents a substantial improvement for the "supply chain security" of bitcoin because:
1. We no longer have to rely on Ubuntu for our build environment for our releases ([oh the horror](72bd4ab867/contrib/gitian-descriptors/gitian-linux.yml (L10))), because Guix builds everything about the container, we can perform this on almost any Linux distro/system.
2. It is now much easier to determine what trusted binaries are in our supply chain, and even make a nice visualization! (see [bootstrappability](https://www.gnu.org/software/guix/manual/en/html_node/Bootstrapping.html)).
3. There is active effort among Guix folks to minimize the number of trusted binaries even further. OriansJ's [stage0](https://github.com/oriansj/stage0), and janneke's [Mes](https://www.gnu.org/software/mes/) all aim to achieve [reduced binary boostrap](http://joyofsource.com/reduced-binary-seed-bootstrap.html) for Guix. In fact, I believe if OriansJ gets his way, we will end up some day with only a single trusted binary: hex0 (a ~500 byte self-hosting hex assembler).
### Steps to Completion
- [x] Successfully build bitcoin inside the Guix environment
- [x] Make `check-symbols` pass
- [x] Do the above but without nasty hacks
- [x] Solve some of the more innocuous hacks
- [ ] Make it cross-compile (HELP WANTED HERE)
- [x] Linux
- [x] x86_64-linux-gnu
- [x] i686-linux-gnu
- [x] aarch64-linux-gnu
- [x] arm-linux-gnueabihf
- [x] riscv64-linux-gnu
- [ ] OS X
- [ ] x86_64-apple-darwin14
- [ ] Windows
- [ ] x86_64-w64-mingw32
- [ ] Maybe make importer for depends syntax
- [ ] Document build process for future releases
- [ ] Extra: Pin the revision of Guix that we build with with Guix [inferiors](https://www.gnu.org/software/guix/manual/en/html_node/Inferiors.html)
### Help Wanted
[Old content no. 3](https://github.com/bitcoin/bitcoin/pull/15277#issuecomment-483318210)
[Old content no. 2](https://github.com/bitcoin/bitcoin/pull/15277#issuecomment-471658439)
<details>
<summary>Old content no. 1</summary>
As of now, the command described above to perform a build of bitcoin a lot like Gitian works, but fails at the `check-symbols` stage. This is because a few dynamic libraries are linked in that shouldn't be.
Here's what `ldd src/bitcoind` looks like when built in a Guix container:
```
linux-vdso.so.1 (0x00007ffcc2d90000)
libdl.so.2 => /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/libdl.so.2 (0x00007fb7eda09000)
librt.so.1 => /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/librt.so.1 (0x00007fb7ed9ff000)
libstdc++.so.6 => /gnu/store/4sqps8dczv3g7rwbdibfz6rf5jlk7w90-gcc-5.5.0-lib/lib/libstdc++.so.6 (0x00007fb7ed87c000)
libpthread.so.0 => /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/libpthread.so.0 (0x00007fb7ed85b000)
libm.so.6 => /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/libm.so.6 (0x00007fb7ed6da000)
libgcc_s.so.1 => /gnu/store/4sqps8dczv3g7rwbdibfz6rf5jlk7w90-gcc-5.5.0-lib/lib/libgcc_s.so.1 (0x00007fb7ed6bf000)
libc.so.6 => /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/libc.so.6 (0x00007fb7ed506000)
/gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007fb7ee3a0000)
```
And here's what it looks in one of our releases:
```
linux-vdso.so.1 (0x00007ffff52cd000)
libpthread.so.0 => /usr/lib/libpthread.so.0 (0x00007f87726b4000)
librt.so.1 => /usr/lib/librt.so.1 (0x00007f87726aa000)
libm.so.6 => /usr/lib/libm.so.6 (0x00007f8772525000)
libgcc_s.so.1 => /usr/lib/libgcc_s.so.1 (0x00007f877250b000)
libc.so.6 => /usr/lib/libc.so.6 (0x00007f8772347000)
/lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007f8773392000)
```
~~I suspect it is because my script does not apply the gitian-input patches [described in the release process](https://github.com/bitcoin/bitcoin/blob/master/doc/release-process.md#fetch-and-create-inputs-first-time-or-when-dependency-versions-change) but there is no description as to how these patches are applied.~~ It might also be something else entirely.
Edit: It is something else. It appears that the gitian inputs are only used by [`gitian-win-signer.yml`](d6e700e40f/contrib/gitian-descriptors/gitian-win-signer.yml (L14))
</details>
### How to Help
1. Install Guix on your distro either [from source](https://www.gnu.org/software/guix/manual/en/html_node/Requirements.html) or perform a [binary installation](https://www.gnu.org/software/guix/manual/en/html_node/Binary-Installation.html#Binary-Installation)
2. Try out my branch and the command described above!
ACKs for top commit:
MarcoFalke:
Thanks for the replies. ACK 751549b52a9a4cd27389d807ae67f02bbb39cd7f
laanwj:
ACK 751549b52a9a4cd27389d807ae67f02bbb39cd7f
Tree-SHA512: 50e6ab58c6bda9a67125b6271daf7eff0ca57d0efa8941ed3cd951e5bf78b31552fc5e537b1e1bcf2d3cc918c63adf19d685aa117a0f851024dc67e697890a8d
fab558612278909df93bdf88f5727b04f13aef0f doc: Use precise permission flags where possible (MarcoFalke)
Pull request description:
Instead of mentioning the all-encompassing `-whitelist*` settings, change the docs to mention the exact permission flag that will influence the behaviour.
This is needed because in the future, the too-broad `-whitelist*` settings (they either include *all* permission flags or apply to *all* peers) might be deprecated to require the permission flags to be enumerated.
Alternatively, in the future there could be an RPC to set the net permission flags on an existing connection, in which case the `-whitelist*` terminology is of no help.
ACKs for top commit:
jnewbery:
reACK fab558612278909df93bdf88f5727b04f13aef0f
fjahr:
Code review ACK fab558612278909df93bdf88f5727b04f13aef0f
jonatack:
ACK fab558612278909df93bdf88f5727b04f13aef0f
Tree-SHA512: c7dea3e577d90103bb2b0ffab7b7c8640b388932a3a880f69e2b70747fc9213dc1f437085671fd54c902ec2a578458b8a2fae6dbe076642fb88efbf9fa9e679c
* Squashed 'src/dashbls/' content from commit 66ee820fbc
git-subtree-dir: src/dashbls
git-subtree-split: 66ee820fbc9e3b97370db8c164904af48327a124
* build: stop tracking build-system generated relic_conf.h.in
* build: add support for building bls-signatures from local subtree
* build: add exclusions to linting scripts and filters
* build: drop bls-signatures (bls-dash) from depends
e1c582cbaa4c094d204da34c3b1fdd0d4c557519 contrib: makeseeds: Read suspicious hosts from a file instead of hardcoding (Sanjay K)
Pull request description:
referring to: https://github.com/bitcoin/bitcoin/issues/17020
good first issue: reading SUSPICIOUS_HOSTS from a file.
I haven't changed the base hosts that were included in the original source, just made it readable from a file.
ACKs for top commit:
practicalswift:
ACK e1c582cbaa4c094d204da34c3b1fdd0d4c557519 -- diff looks correct
Tree-SHA512: 18684abc1c02cf52d63f6f6ecd98df01a9574a7c470524c37e152296504e2e3ffbabd6f3208214b62031512aeb809a6d37446af82c9f480ff14ce4c42c98e7c2
2d23082cbe4641175d752a5969f67cdadf1afcea bump test timeouts so that functional tests run in valgrind (Micky Yun Chan)
Pull request description:
ci/tests: Bump timeouts so all functional tests run on travis in valgrind #17763
Top commit has no ACKs.
Tree-SHA512: 5a8c6e2ea02b715facfcb58c761577be15ae58c45a61654beb98c2c2653361196c2eec521bcae4a9a1bab8e409d6807de771ef4c46d3d05996ae47a22d499d54
c491368d8cfddf3a5b6d574f10ed67492fcecbed scripts: add MACHO dylib checking to symbol-check.py (fanquake)
76bf97213f4b153dd3ccf1314088a73c4804601d scripts: fix check-symbols & check-security argument passing (fanquake)
Pull request description:
Based on #17857.
This adds dynamic library checks for MACHO executables to symbol-check.py. The script has been modified to function more like `security-check.py`. The error output is now also slightly different. i.e:
```bash
# Linux x86
bitcoin-cli: symbol operator new[](unsigned long) from unsupported version GLIBCXX_3.4
bitcoin-cli: export of symbol vtable for std::basic_ios<char, std::char_traits<char> > not allowed
bitcoin-cli: NEEDED library libstdc++.so.6 is not allowed
bitcoin-cli: failed IMPORTED_SYMBOLS EXPORTED_SYMBOLS LIBRARY_DEPENDENCIES
# RISCV (skips exported symbols checks)
bitcoin-tx: symbol operator new[](unsigned long) from unsupported version GLIBCXX_3.4
bitcoin-tx: NEEDED library libstdc++.so.6 is not allowed
bitcoin-tx: failed IMPORTED_SYMBOLS LIBRARY_DEPENDENCIES
# macOS
Checking macOS dynamic libraries...
libboost_filesystem.dylib is not in ALLOWED_LIBRARIES!
bitcoind: failed DYNAMIC_LIBRARIES
```
Compared to `v0.19.0.1` the macOS allowed dylibs has been slimmed down somewhat:
```diff
src/qt/bitcoin-qt:
/usr/lib/libSystem.B.dylib
-/System/Library/Frameworks/DiskArbitration.framework/Versions/A/DiskArbitration
/System/Library/Frameworks/IOKit.framework/Versions/A/IOKit
/System/Library/Frameworks/Foundation.framework/Versions/C/Foundation
/System/Library/Frameworks/CoreServices.framework/Versions/A/CoreServices
/System/Library/Frameworks/AppKit.framework/Versions/C/AppKit
/System/Library/Frameworks/ApplicationServices.framework/Versions/A/ApplicationServices
/System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation
-/System/Library/Frameworks/Security.framework/Versions/A/Security
-/System/Library/Frameworks/SystemConfiguration.framework/Versions/A/SystemConfiguration
/System/Library/Frameworks/CoreGraphics.framework/Versions/A/CoreGraphics
-/System/Library/Frameworks/OpenGL.framework/Versions/A/OpenGL
-/System/Library/Frameworks/AGL.framework/Versions/A/AGL
/System/Library/Frameworks/Carbon.framework/Versions/A/Carbon
/usr/lib/libc++.1.dylib
-/System/Library/Frameworks/CFNetwork.framework/Versions/A/CFNetwork
/System/Library/Frameworks/CoreText.framework/Versions/A/CoreText
/System/Library/Frameworks/ImageIO.framework/Versions/A/ImageIO
/usr/lib/libobjc.A.dylib
```
ACKs for top commit:
laanwj:
ACK c491368d8cfddf3a5b6d574f10ed67492fcecbed
Tree-SHA512: f8624e4964e80b3e0d34e8d3cc33f3107938f3ef7a01c07828f09b902b5ea31a53c50f9be03576e1896ed832cf2c399e03a7943a4f537a1e1c705f3804aed979
4455949d6f0218b40d33d7fe6de6555f8f62192f Make test DoS_mapOrphans deterministic (David Reikher)
Pull request description:
This pull request proposes a solution to make the test `DoS_mapOrphans` in denialofservice_tests.cpp have deterministic coverage.
The `RandomOrphan` function in denialofservice_tests.cpp and the implicitly called function `ecdsa_signature_parse_der_lax` in pubkey.cpp were causing the non-deterministic test coverage.
In the former, if a random orphan was selected the index of which is bigger than the max. orphan index in `mapOrphanTransactions`, the last orphan was returned from `RandomOrphan`. If the random number generated was never large enough, this condition would not be fulfilled and the corresponding branch wouldn't run. The proposed solution is to force one of the 50 dependant orphans to depend on the last orphan in `mapOrphanTransactions` using the newly introduced function `OrphanByIndex` (and passing it a large uint256), forcing this branch to run at least once.
In the latter, if values for ECDSA `R` or `S` (or both) had no leading zeros, some code would not be executed. The solution was to find a constant signature that would be comprised of `R` and `S` values with leading zeros and calling `CPubKey::Verify` at the end of the test with this signature forcing this code to always run at least once at the end even if it hadn't throughout the test.
To test that the coverage is (at least highly likely) deterministic, I ran
`contrib/devtools/test_deterministic_coverage.sh denialofservice_tests/DoS_mapOrphans 1000`
and the result was deterministic coverage across 1000 runs.
Also - removed denialofservice_tests test entry from the list of non-deterministic tests in the coverage script.
ACKs for top commit:
MarcoFalke:
ACK 4455949d6f0218b40d33d7fe6de6555f8f62192f
Tree-SHA512: 987eb1f94b80d5bec4d4944e91ef43b9b8603055750362d4b4665b7f011be27045808aa9f4c6ccf8ae009b61405f9a1b8671d65a843c3328e5b8acce1f1c00a6
* contrib: set the working directory to /src/dash to allow for cloning gitian dependencies
* contrib: place the home directory inside /home instead of root
* contrib: add notes about sharing ccache across the network
* contrib: chown based on the (u/g)id env vars instead of the associated username
* contrib: reduce layer count by reducing run invocations
* contrib: develop container cleanup and maintenance
- add apt-cacher-ng, gpg, lsb-release, screen as a package dependencies
- reorder packages in alphabetical order
- correct documentation
- create and add user to the docker group to satisfy Gitian's needs
- reduce the number of RUN calls to reduce layer count
683d197970a533690ca1bd4d06d021900e87cb8b Use latest signapple commit (Andrew Chow)
Pull request description:
Update gitian and guix to use the same latest signapple commit.
Also changed guix to use the actual repo. The changes from the fork were incorporated upstream.
ACKs for top commit:
fanquake:
ACK 683d197970a533690ca1bd4d06d021900e87cb8b - sanity checked that the updated package is built:
Tree-SHA512: a4981f8bbe33e6c5654632bc9b9f6f2f1e675741a19ac7296205e370f1e64a747101ecb632e0cc82a0134e4c2e9ce47b3f7b4d8c8f75f0f06dd069c078303759
2c403279e2f0f7c8c27c56d4e7b0573c59571f0a gitian: Remove codesign_allocate and pagestuff from MacOS build (Andrew Chow)
f55eed251488d70d5e2e3a2965a4f8ec0c476853 gitian: use signapple to create the MacOS code signature (Andrew Chow)
95b06d21852b28712db6c710e420a58bdc1a0944 gitian: use signapple to apply the MacOS code signature (Andrew Chow)
42bb1ea363286b088257cabccb686ef1887c1d3b gitian: install signapple in gitian-osx-signer.yml (Andrew Chow)
Pull request description:
The MacOS code signing issues that were encountered during the 0.21.0 release cycle have shown that it is necessary for us to use a code signing tool for which the source code is available and modifiable by us. Given that there appears to not be such a tool available, I have written such a tool, [signapple](https://github.com/achow101/signapple), that we can use. This tool is able to create a valid MacOS code signature, detach it in a way that we were doing previously, and attach it to the unsigned binary. This tool can also verify that the signature is correct.
This PR implements the usage of that tool in the gitian build for the code signed MacOS binary. The code signer will use this tool to create the detached signature. Gitian builders will use this tool to apply the detached signature. The `gitian-osx-signer.yml` descriptor has been modified to install this tool so that the `detached-sig-apply.sh` script can use it. Additionally, the `codesign_allocate` and `pagestuff` tools are no longer necessary so they are no longer added to the tarball used in code signing. Lastly, both the `detached-sig-create.sh` and `detached-sig-apply.sh` scripts are made to be significantly less complex and to not do unexpected things such as unpacking an already unpacked tarball.
The detached code signature that signapple creates is almost identical to that which we were previously creating. The only difference is that the cpu architecture name is included in the extension (e.g. we have `bitcoin-qt.x86_64sign` instead of `bitcoin-qt.sign`). This was done in order to support signing universal binaries which we may want to do in the future. However signapple can still apply existing code signatures as it will accept the `.sign` extension. If it is desired, it can be modified to produce signatures with just the `.sign` extension. However I do not think it is necessary to maintain compatibility with the old process.
ACKs for top commit:
laanwj:
Code review ACK 2c403279e2f0f7c8c27c56d4e7b0573c59571f0a
Tree-SHA512: 2a0e01e9133f8859b9de26e7e8fe1d2610d2cbdee2845e6008b12c083c7e3622cbb2d9b83c50a269e2c3074ab95914a8225d3cd4108017f58b77a62bf10951e0
eacedfb0230978748cbcfb13817fed7e7c756ba7 scripts: add additional type annotations to security-check.py (fanquake)
83d063e9541cc9ea41ea86919eb9435c73efb14e scripts: add run_command to security-check.py (fanquake)
13f606b4f940e5820ff21ea62fc27a5a91774b05 scripts: remove NONFATAL from security-check.py (fanquake)
061acf62a15ad3dbb9f055b7c2569b9832ed623a scripts: no-longer check for 32 bit windows in security-check.py (fanquake)
Pull request description:
* Remove 32-bit Windows checks.
* Remove NONFATAL checking. Added in #8249, however unused since #13764.
* Add `run_command` to de-duplicate all of the subprocess calls. Mentioned in #18713.
* Add additional type annotations.
* Print stderr when there is an issue running a command.
ACKs for top commit:
laanwj:
ACK eacedfb0230978748cbcfb13817fed7e7c756ba7
Tree-SHA512: 69a7ccfdf346ee202b3e8f940634c5daed1d2b5a5d15ac9800252866ba3284ec66e391a66a0b341f5a4e5e8482fe1b614d4671e8e766112ff059405081184a85
3e38023af724a76972d39cbccfb0bba4c54a0323 scripts: add PE .reloc section check to security-check.py (fanquake)
Pull request description:
The `ld` in binutils has historically had a few issues with PE binaries, there's a good summary in this [thread](https://sourceware.org/bugzilla/show_bug.cgi?id=19011).
One issue in particular was `ld` stripping the `.reloc` section out of PE binaries, even though it's required for functioning ASLR. This was [reported by a Tor developer in 2014](https://sourceware.org/bugzilla/show_bug.cgi?id=17321) and they have been patching their [own binutils](https://gitweb.torproject.org/builders/tor-browser-build.git/tree/projects/binutils) ever since. However their patch only made it into binutils at the [start of this year](https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=dc9bd8c92af67947db44b3cb428c050259b15cd0). It adds an `--enable-reloc-section` flag, which is turned on by default if you are using `--dynamic-base`. In the mean time this issue has also been worked around by other projects, such as FFmpeg, see [this commit](91b668acd6).
I have checked our recent supported Windows release binaries, and they do contain a `.reloc` section. From what I understand, we are using all the right compile/linker flags, including `-pie` & `-fPIE`, and have never run into the crashing/entrypoint issues that other projects might have seen.
One other thing worth noting here, it how Debian/Ubuntu patch the binutils that they distribute, because that's what we end up using in our gitian builds.
In the binutils-mingw-w64 in Bionic (18.04), which we currently use in gitian, PE hardening options/security flags are enabled by default. See the [changelog](https://changelogs.ubuntu.com/changelogs/pool/universe/b/binutils-mingw-w64/binutils-mingw-w64_8ubuntu1/changelog) and the [relevant commit](452b3013b8).
However in Focal (20.04), this has now been reversed. PE hardening options are no-longer the default. See the [changelog](https://changelogs.ubuntu.com/changelogs/pool/universe/b/binutils-mingw-w64/binutils-mingw-w64_8.8/changelog) and [relevant commit](7bd8b2fbc2), which cites same .reloc issue mentioned here.
Given that we explicitly specify/opt-in to everything that we want to use, the defaults aren't necessarily an issue for us. However I think it highlights the importance of continuing to be explicit about what we want, and not falling-back or relying on upstream.
This was also prompted by the possibility of us doing link time garbage collection, see #18579 & #18605. It seemed some sanity checks would be worthwhile in-case the linker goes haywire while garbage collecting.
I think Guix is going to bring great benefits when dealing with these kinds of issues. Carl you might have something to say in that regard.
ACKs for top commit:
dongcarl:
ACK 3e38023af724a76972d39cbccfb0bba4c54a0323
Tree-SHA512: af14d63bdb334bde548dd7de3e0946556b7e2598d817b56eb4e75b3f56c705c26aa85dd9783134c4b6a7aeb7cb4de567eed996e94d533d31511f57ed332287da
8334ee31f868f0f9baf0920d14d20174ed889dbe scripts: add MACHO LAZY_BINDINGS test to test-security-check.py (fanquake)
7b99c7454cdb74cd9cd7a5eedc2fb9d0a19df456 scripts: add MACHO Canary check to security-check.py (fanquake)
Pull request description:
7b99c7454cdb74cd9cd7a5eedc2fb9d0a19df456 uses `otool -Iv` to check for `___stack_chk_fail` in the macOS binaries. Similar to the [ELF check](https://github.com/bitcoin/bitcoin/blob/master/contrib/devtools/security-check.py#L105). Note that looking for a triple underscore prefixed function (as opposed to two for ELF) is correct for the macOS binaries. i.e:
```bash
otool -Iv bitcoind | grep chk
0x00000001006715b8 509 ___memcpy_chk
0x00000001006715be 510 ___snprintf_chk
0x00000001006715c4 511 ___sprintf_chk
0x00000001006715ca 512 ___stack_chk_fail
0x00000001006715d6 517 ___vsnprintf_chk
0x0000000100787898 513 ___stack_chk_guard
```
8334ee31f868f0f9baf0920d14d20174ed889dbe is a follow up to #18295 and adds test cases to `test-security-check.py` that for some reason I didn't add at the time. I'll sort out #18434 so that we can run these tests in the CI.
ACKs for top commit:
practicalswift:
ACK 8334ee31f868f0f9baf0920d14d20174ed889dbe: Mitigations are important. Important things are worth asserting :)
jonasschnelli:
utACK 8334ee31f868f0f9baf0920d14d20174ed889dbe.
Tree-SHA512: 1aa5ded34bbd187eddb112b27278deb328bfc21ac82316b20fab6ad894f223b239a76b53dab0ac1770d194c1760fcc40d4da91ec09959ba4fc8eadedb173936a
5ca90f8b598978437340bb8467f527b9edfb2bbf scripts: add MACHO lazy bindings check to security-check.py (fanquake)
Pull request description:
This is a slightly belated follow up to #17686 and some discussion with Cory. It's not entirely clear if we should make this change due to the way the macOS dynamic loader appears to work. However I'm opening this for some discussion. Also related to #17768.
#### Issue:
[`LD64`](https://opensource.apple.com/source/ld64/) doesn't set the [MH_BINDATLOAD](https://opensource.apple.com/source/xnu/xnu-6153.11.26/EXTERNAL_HEADERS/mach-o/loader.h.auto.html) bit in the header of MACHO executables, when building with `-bind_at_load`. This is in contradiction to the [documentation](https://opensource.apple.com/source/ld64/ld64-450.3/doc/man/man1/ld.1.auto.html):
```bash
-bind_at_load
Sets a bit in the mach header of the resulting binary which tells dyld to
bind all symbols when the binary is loaded, rather than lazily.
```
The [`ld` in Apples cctools](https://opensource.apple.com/source/cctools/cctools-927.0.2/ld/layout.c.auto.html) does set the bit, however the [cctools-port](https://github.com/tpoechtrager/cctools-port/) that we use for release builds, bundles `LD64`.
However; even if the linker hasn't set that bit, the dynamic loader ([`dyld`](https://opensource.apple.com/source/dyld/)) doesn't seem to ever check for it, and from what I understand, it looks at a different part of the header when determining whether to lazily load symbols.
Note that our release binaries are currently working as expected, and no lazy loading occurs.
#### Example:
Using a small program, we can observe the behaviour of the dynamic loader.
Conducted using:
```bash
clang++ --version
Apple clang version 11.0.0 (clang-1100.0.33.17)
Target: x86_64-apple-darwin18.7.0
ld -v
@(#)PROGRAM:ld PROJECT:ld64-530
BUILD 18:57:17 Dec 13 2019
LTO support using: LLVM version 11.0.0, (clang-1100.0.33.17) (static support for 23, runtime is 23)
TAPI support using: Apple TAPI version 11.0.0 (tapi-1100.0.11)
```
```cpp
#include <iostream>
int main() {
std::cout << "Hello World!\n";
return 0;
}
```
Compile and check the MACHO header:
```bash
clang++ test.cpp -o test
otool -vh test
...
Mach header
magic cputype cpusubtype caps filetype ncmds sizeofcmds flags
MH_MAGIC_64 X86_64 ALL LIB64 EXECUTE 16 1424 NOUNDEFS DYLDLINK TWOLEVEL WEAK_DEFINES BINDS_TO_WEAK PIE
# Run and dump dynamic loader bindings:
DYLD_PRINT_BINDINGS=1 DYLD_PRINT_TO_FILE=no_bind.txt ./test
Hello World!
```
Recompile with `-bind_at_load`. Note still no `BINDATLOAD` flag:
```bash
clang++ test.cpp -o test -Wl,-bind_at_load
otool -vh test
Mach header
magic cputype cpusubtype caps filetype ncmds sizeofcmds flags
MH_MAGIC_64 X86_64 ALL LIB64 EXECUTE 16 1424 NOUNDEFS DYLDLINK TWOLEVEL WEAK_DEFINES BINDS_TO_WEAK PIE
...
DYLD_PRINT_BINDINGS=1 DYLD_PRINT_TO_FILE=bind.txt ./test
Hello World!
```
If we diff the outputs, you can see that `dyld` doesn't perform any lazy bindings when the binary is compiled with `-bind_at_load`, even if the `BINDATLOAD` flag is not set:
```diff
@@ -1,11 +1,27 @@
+dyld: bind: test:0x103EDF030 = libc++.1.dylib:__ZNKSt3__16locale9use_facetERNS0_2idE, *0x103EDF030 = 0x7FFF70C9FA58
+dyld: bind: test:0x103EDF038 = libc++.1.dylib:__ZNKSt3__18ios_base6getlocEv, *0x103EDF038 = 0x7FFF70CA12C2
+dyld: bind: test:0x103EDF068 = libc++.1.dylib:__ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentryC1ERS3_, *0x103EDF068 = 0x7FFF70CA12B6
+dyld: bind: test:0x103EDF070 = libc++.1.dylib:__ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentryD1Ev, *0x103EDF070 = 0x7FFF70CA1528
+dyld: bind: test:0x103EDF080 = libc++.1.dylib:__ZNSt3__16localeD1Ev, *0x103EDF080 = 0x7FFF70C9FAE6
<trim>
-dyld: lazy bind: test:0x10D4AC0C8 = libsystem_platform.dylib:_strlen, *0x10D4AC0C8 = 0x7FFF73C5C6E0
-dyld: lazy bind: test:0x10D4AC068 = libc++.1.dylib:__ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentryC1ERS3_, *0x10D4AC068 = 0x7FFF70CA12B6
-dyld: lazy bind: test:0x10D4AC038 = libc++.1.dylib:__ZNKSt3__18ios_base6getlocEv, *0x10D4AC038 = 0x7FFF70CA12C2
-dyld: lazy bind: test:0x10D4AC030 = libc++.1.dylib:__ZNKSt3__16locale9use_facetERNS0_2idE, *0x10D4AC030 = 0x7FFF70C9FA58
-dyld: lazy bind: test:0x10D4AC080 = libc++.1.dylib:__ZNSt3__16localeD1Ev, *0x10D4AC080 = 0x7FFF70C9FAE6
-dyld: lazy bind: test:0x10D4AC070 = libc++.1.dylib:__ZNSt3__113basic_ostreamIcNS_11char_traitsIcEEE6sentryD1Ev, *0x10D4AC070 = 0x7FFF70CA1528
```
Note: `dyld` also has a `DYLD_BIND_AT_LAUNCH=1` environment variable, that when set, will force any lazy bindings to be non-lazy:
```bash
dyld: forced lazy bind: test:0x10BEC8068 = libc++.1.dylib:__ZNSt3__113basic_ostream
```
#### Thoughts:
After looking at the dyld source, I can't find any checks for `MH_BINDATLOAD`. You can see the flags it does check for, such as MH_PIE or MH_BIND_TO_WEAK [here](https://opensource.apple.com/source/dyld/dyld-732.8/src/ImageLoaderMachO.cpp.auto.html).
It seems that the lazy binding of any symbols depends on whether or not [lazy_bind_size](https://opensource.apple.com/source/xnu/xnu-6153.11.26/EXTERNAL_HEADERS/mach-o/loader.h.auto.html) from the `LC_DYLD_INFO_ONLY` load command is > 0. Which was mentioned in [#17686](https://github.com/bitcoin/bitcoin/pull/17686#issue-350216254).
#### Changes:
This PR is one of [Corys commits](7b6ba26178), that I've rebased and modified to make build. I've also included an addition to the `security-check.py` script to check for the flag.
However, given the above, I'm not entirely sure this patch is the correct approach. If the linker no-longer inserts it, and the dynamic loader doesn't look for it, there might be little benefit to setting it. Or, maybe this is an oversight from Apple and needs some upstream discussion. Looking for some thoughts / Concept ACK/NACK.
One alternate approach we could take is to drop the patch and modify security-check.py to look for `lazy_bind_size` == 0 in the `LC_DYLD_INFO_ONLY` load command, using `otool -l`.
ACKs for top commit:
theuni:
ACK 5ca90f8b598978437340bb8467f527b9edfb2bbf
Tree-SHA512: 444022ea9d19ed74dd06dc2ab3857a9c23fbc2f6475364e8552d761b712d684b3a7114d144f20de42328d1a99403b48667ba96885121392affb2e05b834b6e1c
7142d50ac33e0ad7d24e49e04c1fc7e3e769ed46 scripts: rename test_64bit_PE to test_PE (fanquake)
edaca2dd123cef958699c07ab248cf0ffc71af07 scripts: add MACHO NX check to security-check.py (fanquake)
1a4e9f32efcc5f6a74290446dc58784fd85c7b31 scripts: add MACHO tests to test-security-check.py (fanquake)
Pull request description:
Adds tests for the MACHO checks in security-check.py:
ac579ada7e/contrib/devtools/security-check.py (L212-L214)
I'm planning on following up with more checks in security-check.py, and corresponding tests in test-security-check.py.
Note that you'll probably have to be on macOS to run them. You can run just this suite with `python3 test-security-check.py TestSecurityChecks.test_MACHO`.
ACKs for top commit:
laanwj:
ACK 7142d50ac33e0ad7d24e49e04c1fc7e3e769ed46
Tree-SHA512: ace3ca9f6df5d4fedd5988938fb7dc7563ec7dc587aa275f780b5f51e9b8d7d6f7768e0a1e05ce438510a07b8640aba92c76847b30c2990f46c66b78a0acf960
ea3c7e585c382998212fd7f41114462a8168a734 test: Remove libssl-dev packages from CI scripts (Wladimir J. van der Laan)
7ea55264b9d60325bc7a5c15d78e9063de145970 test: remove lsan suppression for libcrypto (Wladimir J. van der Laan)
2d7066527a456f8e1f4f603fe104b0bd9d864559 build: remove libcrypto as internal dependency in libbitcoinconsensus.pc (Wladimir J. van der Laan)
278751ea11f2cfe68b0c98f504f65586720cb5a4 doc: Remove ssl as a required dependency from build-unix (Wladimir J. van der Laan)
Pull request description:
Some doc and build cleanups following #17265.
I intentionally left the libssl-dev install in `gitian-win-signer.yml`, as it's necessary for the ossl signer.
ACKs for top commit:
MarcoFalke:
ACK ea3c7e585c382998212fd7f41114462a8168a734 🗯
jamesob:
ACK ea3c7e585c
practicalswift:
ACK ea3c7e585c382998212fd7f41114462a8168a734 - nice!
fanquake:
ACK ea3c7e585c382998212fd7f41114462a8168a734 - thanks.
Tree-SHA512: 67ea35bdd6d6e512d69e6734713534c88cae033a2ed695677ea15c3e3d5ff570374e342775c88e60877fa43a19047853e7b2a433e2c9a4349a5c423726a7457e
7c9e821c4e6cb186208ead9c8df616d1f393a49a scripts: add MACHO NOUNDEFS check to security-check.py (fanquake)
4ca92dc6d3f3e487d63286d8871d1829b3d279ff scripts: add MACHO PIE check to security-check.py (fanquake)
Pull request description:
This uses `otool -vh` to print the mach header and look for the `PIE` flag:
```bash
otool -vh src/bitcoind
Mach header
magic cputype cpusubtype caps filetype ncmds sizeofcmds flags
MH_MAGIC_64 X86_64 ALL LIB64 EXECUTE 24 2544 NOUNDEFS DYLDLINK TWOLEVEL WEAK_DEFINES BINDS_TO_WEAK PIE
```
From [`mach-o/loader.h`](https://opensource.apple.com/source/cctools/cctools-927.0.2/include/mach-o/loader.h.auto.html):
```c
#define MH_PIE 0x200000 /* When this bit is set, the OS will
load the main executable at a
random address. Only used in
MH_EXECUTE filetypes. */
```
ACKs for top commit:
laanwj:
code review ACK 7c9e821c4e6cb186208ead9c8df616d1f393a49a
Tree-SHA512: 5ba2f60440d0e31c70371a355c91ca4f723d80f7287d04e2098bf5b11892cc74216ff8f1454603c4db9675d4f7983614843b992b8dcfca0309aadf2aa7ab2e4b
