Commit Graph

476 Commits

Author SHA1 Message Date
Chris Lalancette
07578d82e0
Shrink the tz_offset size to 41. (#1110)
When building with gcc TSan on, and in Debug mode, we see a warning
like:

benchmark/src/timers.cc: In function ‘std::string benchmark::LocalDateTimeString()’:
src/timers.cc:241:15: warning: ‘char* strncat(char*, const char*, size_t)’ output may be truncated copying 108 bytes from a string of length 127 [-Wstringop-truncation]
  241 |   std::strncat(storage, tz_offset, sizeof(storage) - timestamp_len - 1);
      |   ~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

While this is essentially a false positive (we never expect
the number of bytes in tz_offset to be too large), the compiler can't
actually tell that.  Shrink the size of tz_offset to a smaller, but still safe
size to eliminate this warning.

Signed-off-by: Chris Lalancette <clalancette@openrobotics.org>
2021-04-09 17:32:00 +01:00
Tobias Schmidt
5e387e7d33
Implement custom benchmark name (#1107)
* Implement custom benchmark name

The benchmark's name can be changed using the Name() function
which internally uses SetName().

* Update AUTHORS and CONTRIBUTORS

* Describe new feature in README

* Move new name function up

Fixes #1106
2021-03-30 16:43:03 +03:00
Phoenix Meadowlark
a9b9471c02
Fix typo in invalid file name error message. (#1094) 2021-02-22 09:55:07 +00:00
SSE4
d90321ff7a
- add support for Elbrus 2000 (e2k) (#1091)
Signed-off-by: SSE4 <tomskside@gmail.com>
2021-02-14 17:45:57 +03:00
Michał Janiszewski
ea5a5bbff4
Add MSVC ARM64 support for reading clocks (#1052)
Lacks CMake support, see https://github.com/google/benchmark/pull/1090
2021-02-12 13:30:26 +03:00
Scott K Logan
17a6b21ee1
Fix Range when starting at zero (#1073)
The existing behavior results in the `0` value being added twice. Since
`lo` is always added to `dst`, we never want to explicitly add `0` if
`lo` is equal to `0`.
2020-11-26 11:12:45 +00:00
Steven Wan
d9abf01763
Rename 'mftbl' to 'mftb' (#1069)
* Rename 'mftbl' to 'mftb'

* Add my name to the contributor list
2020-11-03 09:08:46 +00:00
Abhina Sree
a9704c268d
Nanosleep workaround for z/OS in sleep.cc (#1067)
* z/OS does not support nanosleep, add workaround to use sleep() and usleep() instead

* change unsigned to int, and fix while loop
2020-10-29 08:49:02 +00:00
Fanbo Meng
dce3322a54
Add support for z/OS XL compiler inline asm syntax (#1063)
On s390 architecture, z/OS XL compiler uses HLASM inline assembly, which has different syntax and needs to be distinguished to avoid compilation error.
2020-10-21 16:39:54 +01:00
Sergei Trofimovich
3d1c267768
src/benchmark_register.h: add missing <limits> inclusion (#1060)
Noticed missing header when was building llvm with gcc-11:

```
llvm-project/llvm/utils/benchmark/src/benchmark_register.h:17:30:
  error: 'numeric_limits' is not a member of 'std'
   17 |   static const T kmax = std::numeric_limits<T>::max();
      |                              ^~~~~~~~~~~~~~
```
2020-10-15 11:12:40 +03:00
Michael Neumann
af72911f2f
Add support for DragonFly BSD (#1058)
Without this commit, compilation fails on DragonFly with the following message:

```
/home/mneumann/Dev/benchmark.old/src/sysinfo.cc:446:2: error: #warning "HOST_NAME_MAX not defined. using 64" [-Werror=cpp]
^~~~~~~
```

Also note that the sysctl is actually `hw.tsc_frequency` on DragonFly:

```
$ sysctl hw.tsc_frequency
hw.tsc_frequency: 3498984022
```

Tested on:

```
$ uname -a
DragonFly box.localnet 5.9-DEVELOPMENT DragonFly v5.9.0.742.g4b29dd-DEVELOPMENT #5: Tue Aug 18 00:21:31 CEST 2020
```
2020-10-12 23:41:49 +03:00
Min-Yih Hsu
ffe1342eb2
Add CycleTimer implementation for M68K architecture (#1050)
As per discussions in here [1], LLVM is going to get backend support on
Motorola 68000 series CPUs (a.k.a M68K or M680x0). So it's necessary to
add CycleTimer implementation here, which is simply using `gettimeofday`
same as MIPS. This fixes #1049

[1] https://reviews.llvm.org/D88389
2020-09-29 09:35:18 +03:00
Christian Wassermann
4857962394
Add CartesianProduct with associated test (#1029)
* Add CartesianProduct with associated test

* Use CartesianProduct in Ranges to avoid code duplication
* Add new cartesian_product_test to CMakeLists.txt
* Update AUTHORS & CONTRIBUTORS

* Rename CartesianProduct to ArgsProduct

* Rename test & fixture accordingly
* Add example for ArgsProduct to README
2020-08-25 13:47:44 +01:00
Dominic Hamon
5b72b6c2da
Remove "BENCHMARK_" prefix from env var version of command line flags (#997)
As noted in #995, this causes issues when the command line flag already
starts with "benchmark_", which they all do.

Not caught by tests as the test flags didn't start with "benchmark".

Fixes #995
2020-08-18 10:02:20 +01:00
Dominic Hamon
1302d2ce09
Add missing breaks for QNX cache counting (#1012) 2020-07-30 09:51:48 +01:00
Alexander Enaldiev
9901011880
JSONReporter: don't report on scaling if we didn't get it (#1005) (#1008)
* JSONReporter: don't report on scaling if we didn't get it (#1005)

* JSONReporter: fix due to review (std::pair<bool, bool> -> enum)

* JSONReporter: scaling: fix the algo (due to review discussion)

* benchmark.h: revert to old-fashioned enum's (C++03 compatibility); rreporter_output_test: let's skip scaling
2020-07-28 12:46:07 +01:00
Reid Paape
15e6dfd718
timers: silence strncat truncation warning (#984) 2020-06-17 14:58:12 +03:00
Brian Wolfe
7cc06ef80c timers: just make the buffers big enough 2020-06-15 16:16:19 -07:00
Brian Wolfe
f25ea40ae1 timers: use snprintf instead of sprintf 2020-06-15 14:16:20 -07:00
Brian Wolfe
f6ac240cd2 timers: silence format overflow warning 2020-06-15 14:02:15 -07:00
Brian Wolfe
99c52f1414
use rfc3339-formatted timestamps in output [output format change] (#965)
* timestamp: use rfc3339-formatted timestamps in output

Replace localized timestamps with machine-readable IETF RFC 3339 format
timestamps. This is an attempt to make the output timestamps easily
machine-readable. ISO8601 specifies standards for time interchange
formats. IETF RFC 3339: https://tools.ietf.org/html/rfc3339 defines a
subset of these for use in the internet. The general form for these
timestamps is:

YYYY-MM-DDTHH:mm:SS[+-]hhmm

This replaces the localized time formats that are currently being used
in the benchmark output to prioritize interchangeability and
machine-readability.

This might break existing programs that rely on the particular date-time
format. This might also may make times less human readable. RFC3339 was
intended to balance human readability and simplicity for machine
readability, but it is primarily intended as an internal representation.

* timers: remove utc string formatting

We only ever need local time printing. Remove the UTC printing
and cosnolidate the logic slightly.

* timers: manually create rfc3339 string

The C++ standard library does not output the time offset in RFC3339
format, it is missing the : between hours and minutes. VS does not
appear to support timezone information by default. To avoid adding too
much complexity to benchmark around timezone handling e.g. a full
date library like https://github.com/HowardHinnant/date, we fall back
to outputting GMT time with a -00:00 offset for those cases.

* timers: use reentrant form for localtime_r & tmtime_r

For non-windows, use the reentrant form for the time conversion
functions.

* timers: cleanup

Use strtol instead of brittle moving characters around.

* timers: only call strftime twice.

Also size buffers to known maximum necessary size and name constants
more appropriately.

* timers: fix unused variable warning
2020-06-15 17:28:17 +01:00
Kai Wolf
56898e9a92
Add missing <cerrno> header include - fixes Android build (#960)
* Add missing <cerrno> header

This commit fixes a current build error on Android where 'errno' is an unidentified
symbol due to a missing header

* Update string_util.cc

Conditionally adds <cerrno> if BENCHMARK_STL_ANDROID_GNUSTL is defined
2020-04-23 14:19:19 +03:00
Luís Marques
ecc1685340
Fix formatting issues introduced by a77d5f7 (#959) 2020-04-17 19:31:49 +03:00
Keith Moyer
8cead00783
Remove warnings for internal use of CSVReporter (#956)
In a previous commit[1], diagnostic pragmas were used to avoid this
warning. However, the incorrect warning flag was indicated, leaving the
warning in place. -Wdeprecated is for deprecated features while
-Wdeprecated-declarations for deprecated functions, variables, and
types[2].

[1] c408461983
[2] https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
2020-04-14 10:20:22 +01:00
Luís Marques
a77d5f70ef
Fix cycleclock::Now for RISC-V and PPC (#955)
Fixes the following issues with the implementation of `cycleclock::Now`:

- The RISC-V implementation wouldn't compile due to a typo;

- Both the PPC and RISC-V implementation's asm statements lacked the
  volatile keyword. This resulted in the repeated read of the counter's
  high part being optimized away, so overflow wasn't handled at all.
  Multiple counter reads could also be misoptimized, especially in LTO
  scenarios.

- Relied on the zero/sign-extension of inline asm operands, which isn't
  guaranteed to occur and differs between compilers, namely GCC and Clang.

The PowerPC64 implementation was improved to do a single 64-bit read of
the time-base counter.

The RISC-V implementation was improved to do the overflow handing in
assembly, since Clang would generate a branch, defeating the purpose
of the non-branching counter reading approach.
2020-04-10 17:02:13 +01:00
Dominic Hamon
0ab2c2906b
Fix type conversion warnings. (#951)
* Fix type conversion warnings.

Fixes #949

Tested locally (Linux/clang), but warnings are on MSVC so may differ.

* Drop the ULP so the double test passes
2020-04-06 13:52:09 +01:00
Roman Lebedev
70d89ac519
Revert "Add d postfix to Debug libraries (#923)"
This reverts commit 5ce2429af7.

Reverts https://github.com/google/benchmark/pull/923
Reopens https://github.com/google/benchmark/issues/922
Fixes https://github.com/google/benchmark/issues/928
Closes https://github.com/google/benchmark/pull/930

See discussion in https://github.com/google/benchmark/issues/928
this broke pkg-config support, since there we don't account
for the suffix, nor is it trivial to do so.
2020-03-16 14:29:27 +03:00
Paweł Bylica
c078337494
Relax CHECK condition in benchmark_runner.cc (#938)
* Add State::error_occurred()

* Relax CHECK condition in benchmark_runner.cc

If the benchmark state contains an error, do not expect any iterations has been run.
This allows using SkipWithError() and return early from the benchmark function.

* README.md: document new possible usage of SkipWithError()
2020-02-21 17:53:25 +03:00
Ben Clayton
8982e1ee6a
Fix MSVC warning. (#935)
This fixes the Visual Studio 2019 warning:

`C4244: '=': conversion from 'int' to 'char', possible loss of data`

When implicitly casting the return value of tolower() (int) to char.

Fixes: #932
2020-02-08 02:48:01 +03:00
Jordan Williams
daff5fead3 Alias CMake Targets. Fixes #921 (#926)
* add Jordan Williams to both CONTRIBUTORS and AUTHORS

* alias benchmark libraries

Provide aliased CMake targets for the benchmark and benchmark_main targets.
The alias targets are namespaced under benchmark::, which is the namespace when they are exported.
I chose not to use either the PROJECT_NAME or the namespace variable but to hard-code the namespace.
This is because the benchmark and benchmark_main targets are hard-coded by name themselves.
Hard-coding the namespace is also much cleaner and easier to read.

* link to aliased benchmark targets

It is safer to link against namespaced targets because of how CMake interprets the double colon.
Typo's will be caught by CMake at configuration-time instead of during compile / link time.

* document the provided alias targets

* add "Usage with CMake" section in documentation

This section covers linking against the alias/import CMake targets and including them using either find_package or add_subdirectory.

* format the "Usage with CMake" README section

Added a newline after the "Usage with CMake" section header.
Dropped the header level of the section by one to make it a direct subsection of the "Usage" section.
Wrapped lines to be no longer than 80 characters in length.
2020-01-14 23:21:24 +03:00
Szitár Gergő
5ce2429af7 Add d postfix to Debug libraries (#923)
* Add DEBUG_POSTFIX to libraries.

Makes it possible to install Debug and Release versions on the
same system. Without this, there were only linker errors when using
the wrong configuration.

* Update CONTRIBUTORS and AUTHORS according to guide
2020-01-05 12:32:40 +00:00
Tetsuo Kiso
0811f1d782 Fix typo in mutex.h (#917) 2019-12-15 13:38:54 +03:00
Roman Lebedev
367119482f CPU caches are binary units, not SI. (#911)
As disscussed in https://github.com/google/benchmark/issues/899,
it is all but certain that the multiplier should be 1024, not 1000.

Fixes https://github.com/google/benchmark/issues/899
2019-12-02 09:29:16 +00:00
Dominic Hamon
49aa79b635
update header guard to match style 2019-11-25 13:05:13 +00:00
Roman Lebedev
51d991f1d7
ParseCommandLineFlags(): do not dereference argc if it is null
Higher up we dereference argc only if it is not null.
But here we do no such check.
2019-11-23 00:23:11 +03:00
Roman Lebedev
c22c266eaf
JSONReporter: RoundDouble(): use std::lround() to round double to int
From clang-tidy bugprone-incorrect-roundings check:
> casting (double + 0.5) to integer leads to incorrect rounding; consider using lround (#include <cmath>) instead
2019-11-23 00:19:02 +03:00
Roman Lebedev
cc7f50e126
BenchmarkRunner: use std::lround() to round double to int
From clang-tidy bugprone-incorrect-roundings check:
> casting (double + 0.5) to integer leads to incorrect rounding; consider using lround (#include <cmath>) instead
2019-11-23 00:18:28 +03:00
Roman Lebedev
173aff82ee
src/counter.h: add header guard 2019-11-22 15:06:43 +03:00
András Leitereg
cf446a18bf Remove superfluous cache line scaling in JSON reporter. (#896)
Cache size is already stored in bytes.
2019-10-24 22:13:03 +03:00
Martin Blanchard
bc200ed8ee Read options from environment (#881) (#883)
Initialize option flags from environment variables values if they are defined, eg. `BENCHMARK_OUT=<filename>` for `--benchmark_out=<filename>`. Command line flag value always prevails.

Fixes https://github.com/google/benchmark/issues/881.
2019-10-23 11:07:08 +03:00
Geoffrey Martin-Noble
b874e72208 Guard definition of __STDC_FORMAT_MACROS in ifndef (#875)
This macro is sometimes already defined and redefining it results
in build errors.
2019-09-23 10:53:09 +01:00
Geoffrey Martin-Noble
7411874d95 Define HOST_NAME_MAX for NaCl and RTEMS (#876)
These OS's don't always have HOST_NAME_MAX defined, resulting in
build errors.

A few related changes as well:
* Only define HOST_NAME_MAX if it's not already defined. There are
  some cases where this is already defined, e.g. with NaCl if
  __USE_POSIX is set. To avoid all of these, only define it if it's
  not already defined.
* Default HOST_NAME_MAX to 64 and issue a #warning. Having the wrong
  max length is pretty harmless. The name just ends up getting
  truncated and this is only for printing debug info. Because we're
  constructing a std::string from a char[] (so defined length), we
  don't need to worry about gethostname's undefined behavior for
  whether the truncation is null-terminated when the hostname
  doesn't fit in HOST_NAME_MAX. Of course, this doesn't help people
  who have -Werror set, since they'll still get a warning.
2019-09-23 10:38:34 +01:00
Sayan Bhattacharjee
7ee72863fd Remove unused doc argument from DEFINE_ macros. (#857)
- Adresses : #856

  - The unused `doc` argument was removed from the `DEFINE_` macros in
    `commandlineflags.h`

  - Converted all the previous `doc` strings passed to the `DEFINE_`
    macros to multiline comments.
2019-08-21 14:12:03 -07:00
Roman Lebedev
7d97a057e1
Custom user counters: add invert modifier. (#850)
While current counters can e.g. answer the question
"how many items is processed per second", it is impossible to get
it to tell "how many seconds it takes to process a single item".

The solution is to add a yet another modifier `kInvert`,
that is *always* considered last, which simply inverts the answer.

Fixes #781, #830, #848.
2019-08-12 17:47:46 +03:00
Eric Fiselier
c408461983 Disable deprecated warnings when touching CSVReporter internally.
The CSVReporter is deprecated, but we still need to reference it in
a few places. To avoid breaking the build when warnings are errors,
we need to disable the warning when we do so.
2019-08-07 15:55:40 -04:00
Roman Lebedev
8e48105d46 CMake; windows: link to lowercase 'shlwapi' - consistent with headers (#840)
The filenames are consistently inconsistent in windows world, might
have something to do with default file system being case-insensitive.
While the native MinGW buils were fixed in 5261307982
that only addressed the headers, but not libraries.
The problem remains when one tries to do a MinGW cross-build from
case-sensitive filesystem.
2019-07-22 13:42:12 +01:00
Sam Elliott
4abdfbb802 Add RISC-V support in cycleclock::Now (#833)
The RISC-V implementation of `cycleclock::Now` uses the user-space
`rdcycle` instruction to query how many cycles have happened since the
core started.

The only complexity here is on 32-bit RISC-V, where `rdcycle` can only
read the lower 32 bits of the 64-bit hardware counter. In this case,
`rdcycleh` reads the higher 32 bits of the counter. We match the powerpc
implementation to detect and correct for overflow in the high bits.
2019-07-05 09:28:17 +01:00
Orgad Shaneh
04a9343fc9 Make some functions const (#832)
and ThreadManager ctor explicit.

Reported by CppCheck.
2019-06-26 09:06:24 +01:00
Roman Lebedev
090faecb45
Use IterationCount in one more place
Found in -UNDEBUG build
2019-05-13 22:42:18 +03:00
Roman Lebedev
f92903cc53
Iteration counts should be uint64_t globally. (#817)
This is a shameless rip-off of https://github.com/google/benchmark/pull/646
I did promise to look into why that proposed PR was producing
so much worse assembly, and so i finally did.

The reason is - that diff changes `size_t` (unsigned) to `int64_t` (signed).

There is this nice little `assert`:
7a1c370283/include/benchmark/benchmark.h (L744)
It ensures that we didn't magically decide to advance our iterator
when we should have finished benchmarking.

When `cached_` was unsigned, the `assert` was `cached_ UGT 0`.
But we only ever get to that `assert` if `cached_ NE 0`,
and naturally if `cached_` is not `0`, then it is bigger than `0`,
so the `assert` is tautological, and gets folded away.

But now that `cached_` became signed, the assert became `cached_ SGT 0`.
And we still only know that `cached_ NE 0`, so the assert can't be
optimized out, or at least it doesn't currently.

Regardless of whether or not that is a bug in itself,
that particular diff would have regressed the normal 64-bit systems,
by halving the maximal iteration space (since we go from unsigned counter
to signed one, of the same bit-width), which seems like a bug.
And just so it happens, fixing *this* bug, fixes the other bug.

This produces fully (bit-by-bit) identical state_assembly_test.s
The filecheck change is actually needed regardless of this patch,
else this test does not pass for me even without this diff.
2019-05-13 12:33:11 +03:00