Move the implementation of programs/timefn.c into rust/src/timefn.rs. The
file provides the monotonic nanosecond clock (UTIL_getTime, span helpers,
UTIL_waitForNextTick, UTIL_support_MT_measurements) used by the CLI and by
several C test tools. timefn.c remains as a declaration-only shim so the
original source lists and header configuration keep working, and it pins the
ABI with static asserts: UTIL_time_t is returned by value and must stay a
plain 64-bit counter, which the Rust #[repr(C)] mirror also asserts.
Platform selection mirrors the C preprocessor structure: Windows uses
QueryPerformanceCounter, Apple targets use mach_absolute_time, and other
POSIX systems use libc clock_gettime(CLOCK_MONOTONIC). Only the unix path is
exercised by this environment; the Windows and Apple paths are written from
the C source and compile-checked logically but are untested here. The C90
clock() fallback is unreachable on Rust-supported targets, so multi-threaded
measurement support is always reported.
The symbols live in the program-only zstd-cli-rs package, keeping them out
of library builds. Linking that archive into C test binaries surfaced a
structural problem: rustc's local ThinLTO promotes internal symbols across
codegen units, so extracting the timefn object could drag in the zstd_cli
parser object, whose FIO_* externs test binaries cannot satisfy. The parser
is therefore gated behind a new additive `cli` cargo feature (default on).
Program archives build with cli,compression,decompression as before, while
tests/Makefile links a helpers-only archive (rust/target/cli-helpers) built
with --no-default-features, which contains no fileio references at all.
tests/Makefile gains build rules for the helpers archive and adds it as a
prerequisite of every binary that compiles the timefn shim: fullbench(32),
fullbench-lib, fullbench-dll, fuzzer(32), zstreamtest(32/asan/tsan/ubsan),
paramgrill, decodecorpus, and poolTests. Prerequisite order places the
archive after all C objects in `$^` link lines; the known-broken -dll
recipes filter to %.c, so they name the archive explicitly. Original C test
sources are untouched; only link inputs changed.
zstd-cli-rs now depends on libc (already used by the core crate) for
clock_gettime and the Mach timebase bindings.
Test Plan:
- cd rust && cargo fmt --check && cargo clippy --all-targets -- -D warnings
&& cargo test --all-targets && cargo build --release
- cd rust/cli && cargo fmt --check && cargo clippy --all-targets -- -D
warnings && cargo test --all-targets; repeat tests with
--no-default-features plus features compression / decompression / (none)
- make -C programs zstd; roundtrip echo hello | zstd | zstd -d
- make -C tests fullbench fuzzer zstreamtest paramgrill decodecorpus
poolTests; ./tests/fullbench -i0; ./tests/fuzzer -i1 --no-big-tests;
./tests/poolTests; make -C tests test-rust-lib-smoke
- verified with nm that the helpers archive member defining UTIL_getTime has
no FIO_*/ZSTD_* undefined references
Refs: rust/README.md
The timer storage type is no longer dependent on OS.
This will make it possible to re-enable posix precise timers
since the timer storage type will no longer be sensible to #include order.
See #3168 for details of pbs of previous interface.
Suggestion by @terrelln
```
for f in $(find . \( -path ./.git -o -path ./tests/fuzz/corpora \) -prune -o -type f);
do
sed -i 's/Facebook, Inc\./Meta Platforms, Inc. and affiliates./' $f;
done
```
* Switch to yearless copyright per FB policy
* Fix up SPDX-License-Identifier lines in `contrib/linux-kernel` sources
* Add zstd copyright/license header to the `contrib/linux-kernel` sources
* Update the `tests/test-license.py` to check for yearless copyright
* Improvements to `tests/test-license.py`
* Check `contrib/linux-kernel` in `tests/test-license.py`
* All copyright lines now have -2020 instead of -present
* All copyright lines include "Facebook, Inc"
* All licenses are now standardized
The copyright in `threading.{h,c}` is not changed because it comes from
zstdmt.
The copyright and license of `divsufsort.{h,c}` is not changed.
not sure why, but msan fires an "unitialized variable" error
when time gets properly initialized by timespec_get().
Maybe in some cases, not all bytes of the structure are initialized ?
Or maybe msan fails to detect the initialization ?
Anyway, pre-initializing the variable before passing it to timespec_get() works.
C11 mandates the definition of timespec_get() and TIME_UTC.
However, FreeBSD11 announce C11 compliance, but does not provifr timespec_get(),
breaking link stage for benchfn.
Since it does not provide TIME_UTC either, which is also required by C11,
test this macro: this will automatically rule out FreeBSD 11 for this code path
(it will use the backup C90 path instead, based on clock_t).
The issue seeems fixed in FreeBSD 12.
benchfn used to rely on mem.h, and util,
which in turn relied on platform.h.
Using benchfn outside of zstd required to bring all these dependencies.
Now, dependency is reduced to timefn only.
This required to create a separate timefn from util,
and rewrite benchfn and timefn to no longer need mem.h.
Separating timefn from util has a wide effect accross the code base,
as usage of time functions is widespread.
A lot of build scripts had to be updated to also include timefn.
ddidderr
Dimitri Papadopoulos