Now that our release binaries are build in a glibc 2.24 and 2.27
environment, we can't use a symbol from glibc 2.28 to test our checks.
Replace renameat2() with nextup(), which was introduced in 2.24.
Note that this also means re-disabling the test for RISC-V, however
RISC-V is built in a glibc 2.27 environment, and our minimum required
glibc for that binary is 2.27.
This is important to make sure that we're not testing tools different
from the one we're building with.
Introduce determine_wellknown_cmd, which encapsulates how we
should handle well-known tools specification (IFS splitting, env
override, etc.).
e8cd3700ee devtools: Integrate ARCH_MIN_GLIBC_VER table into MAX_VERSIONS in symbol-check.py (W. J. van der Laan)
a33381acf5 devtools: Add xkb version to symbol-check (W. J. van der Laan)
19e598bab0 devtools: Fix verneed section parsing in pixie (W. J. van der Laan)
Pull request description:
I misunderstood the ELF specification for version symbols (verneed): The `vn_aux` pointer is relative to the main verneed record, not the start of the section.
This caused many symbols to not be versioned properly in the return value of `elf.dyn_symbols`. This was discovered in #21454.
Fix it by correcting the offset computation.
- xkb versions symbols (using the prefix `V`), as this library is used by bitcoin-qt, add it to the valid versions in `symbol-check.py`
This unfortunately brings to light some symbols that have been introduced since and weren't caught (from a gitian compile of master):
```
bitcoin-cli: symbol getrandom from unsupported version GLIBC_2.25
bitcoin-cli: failed IMPORTED_SYMBOLS
bitcoind: symbol getrandom from unsupported version GLIBC_2.25
bitcoind: symbol log from unsupported version GLIBC_2.29
bitcoind: symbol fcntl64 from unsupported version GLIBC_2.28
bitcoind: symbol pow from unsupported version GLIBC_2.29
bitcoind: symbol exp from unsupported version GLIBC_2.29
bitcoind: failed IMPORTED_SYMBOLS
bitcoin-qt: symbol exp from unsupported version GLIBC_2.29
bitcoin-qt: symbol fcntl64 from unsupported version GLIBC_2.28
bitcoin-qt: symbol log from unsupported version GLIBC_2.29
bitcoin-qt: symbol pow from unsupported version GLIBC_2.29
bitcoin-qt: symbol statx from unsupported version GLIBC_2.28
bitcoin-qt: symbol getrandom from unsupported version GLIBC_2.25
bitcoin-qt: symbol renameat2 from unsupported version GLIBC_2.28
bitcoin-qt: symbol getentropy from unsupported version GLIBC_2.25
bitcoin-qt: failed IMPORTED_SYMBOLS
bitcoin-wallet: symbol exp from unsupported version GLIBC_2.29
bitcoin-wallet: symbol log from unsupported version GLIBC_2.29
bitcoin-wallet: symbol fcntl64 from unsupported version GLIBC_2.28
bitcoin-wallet: failed IMPORTED_SYMBOLS
test_bitcoin: symbol getrandom from unsupported version GLIBC_2.25
test_bitcoin: symbol log from unsupported version GLIBC_2.29
test_bitcoin: symbol fcntl64 from unsupported version GLIBC_2.28
test_bitcoin: symbol pow from unsupported version GLIBC_2.29
test_bitcoin: symbol exp from unsupported version GLIBC_2.29
test_bitcoin: failed IMPORTED_SYMBOLS
```
ACKs for top commit:
hebasto:
ACK e8cd3700ee
Tree-SHA512: 8c15e3478eb642f01a1ddaadef03f80583f088f9fa8e3bf171ce16b0ec05ffb4675ec147d7ffc6a4360637ed47fca517c6ca2bac7bb30d794c03783cfb964b79
The (ancient) versions specified here were deceptive. Entries older than
MAX_VERSIONS['GLIBC'], which is 2.17, are ignored here. So reorganize
the code to avoid confusion for other people reading this code.
I misunderstood the ELF specification for version symbols (verneed):
The `vn_aux` pointer is relative to the main verneed record, not the
start of the section.
This caused many symbols to not be versioned properly in the return
value of `elf.dyn_symbols`. This was discovered in #21454.
Fix it by correcting the offset computation.
Clangs Darwin driver should infer the SDK version used during compilation, and
forward that through to the linker. Add a check that this has been done, and the
expected SDK version is set.
Should help prevent issues like #21771 in future.
We use linker flags (-Wl,--major/minor-subsystem-version) to set the
minimum required version of Windows needed to run our binaries. This
adds a sanity check that the version is being set as expected.
We use a compile flag (-mmacosx-version-min) to set the minimum required
version of macOS needed to run our binaries. This adds a sanity check
that the version is being set as expected.
Because only macOS wasy mentioned, I was unsure if this would be a macOS specific tool. I guess Linux is more used than Mac, so Linux guide should be there, too.
remove fix_configure_mac.patch
Fixed upstream: https://bugreports.qt.io/browse/QTBUG-67286
remove fix_riscv64_arch.patch
Was fixed upstream in 6a39e49a6cdeb28a04a3657bb6a22f848d5dfa9d
remove fix_rcc_determinism.patch
Fixed upstream in https://bugreports.qt.io/browse/QTBUG-62511
remove freetype_back_compat.patch
By the time we ship a release with Qt 5.12, we'll certainly no-longer be
supporting Ubuntu 14.04 and Ubuntu 16.04 ships with FreeType 2.6.1,
which is new enough that using the symbol is no-longer an issue.
The renaming of FT_Get_X11_Font_Format() happened in FreeType 2.6
remove xkb-default.patch
This was removed upstream in d5abf545971da717014d316127045fc19edbcd65
Co-authored-by: Hennadii Stepanov <32963518+hebasto@users.noreply.github.com>
a0a771843f contrib: Changes to checks for PowerPC64 (Luke Dashjr)
634f6ec4eb contrib: Parse ELF directly for symbol and security checks (Wladimir J. van der Laan)
Pull request description:
Instead of the ever-messier text parsing of the output of the readelf tool (which is clearly meant for human consumption not to be machine parseable), parse the ELF binaries directly.
Add a small dependency-less ELF parser specific to the checks.
This is slightly more secure, too, because it removes potential ambiguity due to misparsing and changes in the output format of `elfread`. It also allows for stricter and more specific ELF format checks in the future.
This removes the build-time dependency for `readelf`.
It passes the test-security-check for me locally, ~~though I haven't checked on all platforms~~. I've checked that this works on the cross-compile output for all ELF platforms supported by Bitcoin Core at the moment, as well as PPC64 LE and BE.
Top commit has no ACKs.
Tree-SHA512: 7f9241fec83ee512642fecf5afd90546964561efd8c8c0f99826dcf6660604a4db2b7255e1afb1e9bb0211fd06f5dbad18a6175dfc03e39761a40025118e7bfc
6690adba08 Warn when binaries are built from a dirty branch. (Tyler Chambers)
Pull request description:
- Adjusted `--version` flag behavior in bitcoind and bitcoin-wallet to have the same behavior.
- Added `--version` flag to bitcoin-tx to match.
- Added functionality in gen-manpages.sh to error when attempting to generate man pages for binaries built from a dirty branch.
mitigates problem with issue #20412
ACKs for top commit:
laanwj:
Tested ACK 6690adba08
Tree-SHA512: b5ca509f1a57f66808c2bebc4b710ca00c6fec7b5ebd7eef58018e28e716f5f2358e36551b8a4df571bf3204baed565a297aeefb93990e7a99add502b97ee1b8
Check both failure cases:
- Use a glibc symbol from a version that is too new
- Use a symbol from a library that is not in the allowlist
And also check a conforming binary.
Adding a similar check for Windows PE can be done in a separate PR.
Adjusted version flag behavior in bitcoin-tx, bitcoin-wallet, and
bitcoind to match. Added functionality in gen-manpages.sh to warning when
attempting to generate man pages for binaries built from a dirty
branch.
Instead of the ever-messier text parsing of the output of the readelf
tool (which is clearly meant for human consumption not to be machine
parseable), parse the ELF binaries directly.
Add a small dependency-less ELF parser specific to the checks.
This is slightly more secure, too, because it removes potential
ambiguity due to misparsing and changes in the output format of `elfread`. It
also allows for stricter and more specific ELF format checks in the future.
This removes the build-time dependency for `readelf`.
It passes the test-security-check for me locally, though I haven't
checked on all platforms.
78c312c983 Replace current benchmarking framework with nanobench (Martin Ankerl)
Pull request description:
Replace current benchmarking framework with nanobench
This replaces the current benchmarking framework with nanobench [1], an
MIT licensed single-header benchmarking library, of which I am the
autor. This has in my opinion several advantages, especially on Linux:
* fast: Running all benchmarks takes ~6 seconds instead of 4m13s on
an Intel i7-8700 CPU @ 3.20GHz.
* accurate: I ran e.g. the benchmark for SipHash_32b 10 times and
calculate standard deviation / mean = coefficient of variation:
* 0.57% CV for old benchmarking framework
* 0.20% CV for nanobench
So the benchmark results with nanobench seem to vary less than with
the old framework.
* It automatically determines runtime based on clock precision, no need
to specify number of evaluations.
* measure instructions, cycles, branches, instructions per cycle,
branch misses (only Linux, when performance counters are available)
* output in markdown table format.
* Warn about unstable environment (frequency scaling, turbo, ...)
* For better profiling, it is possible to set the environment variable
NANOBENCH_ENDLESS to force endless running of a particular benchmark
without the need to recompile. This makes it to e.g. run "perf top"
and look at hotspots.
Here is an example copy & pasted from the terminal output:
| ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark
|--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:----------
| 2.52 | 396,529,415.94 | 0.6% | 25.42 | 8.02 | 3.169 | 0.06 | 0.0% | 0.03 | `bench/crypto_hash.cpp RIPEMD160`
| 1.87 | 535,161,444.83 | 0.3% | 21.36 | 5.95 | 3.589 | 0.06 | 0.0% | 0.02 | `bench/crypto_hash.cpp SHA1`
| 3.22 | 310,344,174.79 | 1.1% | 36.80 | 10.22 | 3.601 | 0.09 | 0.0% | 0.04 | `bench/crypto_hash.cpp SHA256`
| 2.01 | 496,375,796.23 | 0.0% | 18.72 | 6.43 | 2.911 | 0.01 | 1.0% | 0.00 | `bench/crypto_hash.cpp SHA256D64_1024`
| 7.23 | 138,263,519.35 | 0.1% | 82.66 | 23.11 | 3.577 | 1.63 | 0.1% | 0.00 | `bench/crypto_hash.cpp SHA256_32b`
| 3.04 | 328,780,166.40 | 0.3% | 35.82 | 9.69 | 3.696 | 0.03 | 0.0% | 0.03 | `bench/crypto_hash.cpp SHA512`
[1] https://github.com/martinus/nanobench
ACKs for top commit:
laanwj:
ACK 78c312c983
Tree-SHA512: 9e18770b18b6f95a7d0105a4a5497d31cf4eb5efe6574f4482f6f1b4c88d7e0946b9a4a1e9e8e6ecbf41a3f2d7571240677dcb45af29a6f0584e89b25f32e49e
Check that sections are appropriately separated in virtual memory,
based on their (expected) permissions. This checks for missing
-Wl,-z,separate-code and potentially other problems.
Co-authored-by: fanquake <fanquake@gmail.com>
The RandomOrphan function and the function ecdsa_signature_parse_der_lax
in pubkey.cpp were causing non-deterministic test coverage.
Force seed in the beginning of the test to make it deterministic.
The seed is selected carefully so that all branches of the function
ecdsa_signature_parse_der_lax are executed. Prior to this fix, the test
was exhibiting non-deterministic coverage since none of the ECDSA
signatures that were generated during the test had leading zeroes in
either R, S, or both, resulting in some branches of said function not
being executed. The seed ensures that both conditions are hit.
Removed denialofservice_tests test entry from the list of non-deterministic
tests in the coverage script.
This replaces the current benchmarking framework with nanobench [1], an
MIT licensed single-header benchmarking library, of which I am the
autor. This has in my opinion several advantages, especially on Linux:
* fast: Running all benchmarks takes ~6 seconds instead of 4m13s on
an Intel i7-8700 CPU @ 3.20GHz.
* accurate: I ran e.g. the benchmark for SipHash_32b 10 times and
calculate standard deviation / mean = coefficient of variation:
* 0.57% CV for old benchmarking framework
* 0.20% CV for nanobench
So the benchmark results with nanobench seem to vary less than with
the old framework.
* It automatically determines runtime based on clock precision, no need
to specify number of evaluations.
* measure instructions, cycles, branches, instructions per cycle,
branch misses (only Linux, when performance counters are available)
* output in markdown table format.
* Warn about unstable environment (frequency scaling, turbo, ...)
* For better profiling, it is possible to set the environment variable
NANOBENCH_ENDLESS to force endless running of a particular benchmark
without the need to recompile. This makes it to e.g. run "perf top"
and look at hotspots.
Here is an example copy & pasted from the terminal output:
| ns/byte | byte/s | err% | ins/byte | cyc/byte | IPC | bra/byte | miss% | total | benchmark
|--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|--------:|----------:|:----------
| 2.52 | 396,529,415.94 | 0.6% | 25.42 | 8.02 | 3.169 | 0.06 | 0.0% | 0.03 | `bench/crypto_hash.cpp RIPEMD160`
| 1.87 | 535,161,444.83 | 0.3% | 21.36 | 5.95 | 3.589 | 0.06 | 0.0% | 0.02 | `bench/crypto_hash.cpp SHA1`
| 3.22 | 310,344,174.79 | 1.1% | 36.80 | 10.22 | 3.601 | 0.09 | 0.0% | 0.04 | `bench/crypto_hash.cpp SHA256`
| 2.01 | 496,375,796.23 | 0.0% | 18.72 | 6.43 | 2.911 | 0.01 | 1.0% | 0.00 | `bench/crypto_hash.cpp SHA256D64_1024`
| 7.23 | 138,263,519.35 | 0.1% | 82.66 | 23.11 | 3.577 | 1.63 | 0.1% | 0.00 | `bench/crypto_hash.cpp SHA256_32b`
| 3.04 | 328,780,166.40 | 0.3% | 35.82 | 9.69 | 3.696 | 0.03 | 0.0% | 0.03 | `bench/crypto_hash.cpp SHA512`
[1] https://github.com/martinus/nanobench
* Adds support for asymptotes
This adds support to calculate asymptotic complexity of a benchmark.
This is similar to #17375, but currently only one asymptote is
supported, and I have added support in the benchmark `ComplexMemPool`
as an example.
Usage is e.g. like this:
```
./bench_bitcoin -filter=ComplexMemPool -asymptote=25,50,100,200,400,600,800
```
This runs the benchmark `ComplexMemPool` several times but with
different complexityN settings. The benchmark can extract that number
and use it accordingly. Here, it's used for `childTxs`. The output is
this:
| complexityN | ns/op | op/s | err% | ins/op | cyc/op | IPC | total | benchmark
|------------:|--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|----------:|:----------
| 25 | 1,064,241.00 | 939.64 | 1.4% | 3,960,279.00 | 2,829,708.00 | 1.400 | 0.01 | `ComplexMemPool`
| 50 | 1,579,530.00 | 633.10 | 1.0% | 6,231,810.00 | 4,412,674.00 | 1.412 | 0.02 | `ComplexMemPool`
| 100 | 4,022,774.00 | 248.58 | 0.6% | 16,544,406.00 | 11,889,535.00 | 1.392 | 0.04 | `ComplexMemPool`
| 200 | 15,390,986.00 | 64.97 | 0.2% | 63,904,254.00 | 47,731,705.00 | 1.339 | 0.17 | `ComplexMemPool`
| 400 | 69,394,711.00 | 14.41 | 0.1% | 272,602,461.00 | 219,014,691.00 | 1.245 | 0.76 | `ComplexMemPool`
| 600 | 168,977,165.00 | 5.92 | 0.1% | 639,108,082.00 | 535,316,887.00 | 1.194 | 1.86 | `ComplexMemPool`
| 800 | 310,109,077.00 | 3.22 | 0.1% |1,149,134,246.00 | 984,620,812.00 | 1.167 | 3.41 | `ComplexMemPool`
| coefficient | err% | complexity
|--------------:|-------:|------------
| 4.78486e-07 | 4.5% | O(n^2)
| 6.38557e-10 | 21.7% | O(n^3)
| 3.42338e-05 | 38.0% | O(n log n)
| 0.000313914 | 46.9% | O(n)
| 0.0129823 | 114.4% | O(log n)
| 0.0815055 | 133.8% | O(1)
The best fitting curve is O(n^2), so the algorithm seems to scale
quadratic with `childTxs` in the range 25 to 800.
